train_imagen_all_u1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "UNET = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import configparser\n",
    "\n",
    "# Loading configurations\n",
    "configParser = configparser.RawConfigParser()   \n",
    "configFilePath = r'configuration.txt'\n",
    "configParser.read(configFilePath)\n",
    "\n",
    "insert_amd_env_vars =  int(configParser.get('COMMON', 'insert_amd_env_vars'))\n",
    "HSA_OVERRIDE_GFX_VERSION =  configParser.get('COMMON', 'HSA_OVERRIDE_GFX_VERSION')\n",
    "ROCM_PATH =  configParser.get('COMMON', 'ROCM_PATH')\n",
    "\n",
    "if(insert_amd_env_vars != 0):\n",
    "    os.environ[\"HSA_OVERRIDE_GFX_VERSION\"] = HSA_OVERRIDE_GFX_VERSION\n",
    "    os.environ[\"ROCM_PATH\"] = ROCM_PATH\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from textwrap import wrap\n",
    "import os\n",
    "import gc\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "#TODO document jupyter\n",
    "import pickle\n",
    "import configparser\n",
    "import sqlite3 as sl\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from matplotlib import pyplot as plt\n",
    "from pathlib import Path\n",
    "\n",
    "configParser = configparser.RawConfigParser()   \n",
    "configFilePath = r'configuration.txt'\n",
    "configParser.read(configFilePath)\n",
    "datasetPathDatabase =  configParser.get('COMMON', 'datasetPathDatabase') + '/dataset.db'\n",
    "datasetPathDatabaseAdditional =  configParser.get('COMMON', 'datasetPathDatabase') + '/datasetAdditional.db'\n",
    "datasetPathDatabaseVgg =  configParser.get('COMMON', 'datasetPathDatabase') + '/datasetFacesBlurred.db'\n",
    "image_size =  int(configParser.get('COMMON', 'resizeImageTo'))\n",
    "begin_with_image_size = int(configParser.get('COMMON', 'begin_with_image_size'))\n",
    "unet1_dim =  int(configParser.get('COMMON', 'unet1_dim'))\n",
    "unet2_dim =  int(configParser.get('COMMON', 'unet2_dim'))\n",
    "unet1_image_size =  int(configParser.get('COMMON', 'unet1_image_size'))\n",
    "audio_embs =  str(configParser.get('COMMON', 'audio_embs'))\n",
    "audio_length_used =  configParser.get('train_imagen', 'audio_length_used') \n",
    "model_filename =  configParser.get('train_imagen', 'model_filename')\n",
    "sub_epochs=  int(configParser.get('train_imagen', 'sub_epochs') )\n",
    "inner_epochs=  int(configParser.get('train_imagen', 'inner_epochs') )\n",
    "batch_size=  int(configParser.get('train_imagen', 'batch_size') )\n",
    "timesteps= int(configParser.get('COMMON', 'timesteps') )\n",
    "sample_every=  int(configParser.get('train_imagen', 'sample_every') ) - 1\n",
    "sample_probability = int(configParser.get('train_imagen', 'sample_probability') )\n",
    "save_model_every=  int(configParser.get('train_imagen', 'save_model_every') )\n",
    "sample_every_offset=  int(configParser.get('train_imagen', 'sample_every_offset') ) - 1\n",
    "save_every_offset=  int(configParser.get('train_imagen', 'save_every_offset') ) - 1\n",
    "imagen_samples_folder = configParser.get('train_imagen', 'imagen_samples_folder') \n",
    "db_chunk = int(configParser.get('train_imagen', 'db_chunk'))\n",
    "dask_chunk = int(configParser.get('train_imagen', 'dask_chunk'))\n",
    "stop_at_no_of_samples = int(configParser.get('train_imagen', 'stop_at_no_of_samples'))\n",
    "epochs = int(configParser.get('train_imagen', 'epochs'))\n",
    "ignore_speaker_embedding = bool(int(configParser.get('train_imagen', 'ignore_speaker_embedding') ))\n",
    "ignore_speech_brain = bool(int(configParser.get('train_imagen', 'ignore_speech_brain') ))\n",
    "ignore_pyannote_titanet_speakernet = bool(int(configParser.get('train_imagen', 'ignore_pyannote_titanet_speakernet') ))\n",
    "ignore_audio_features = bool(int(configParser.get('train_imagen', 'ignore_audio_features') ))\n",
    "ignore_pyAudioAnalysis = bool(int(configParser.get('train_imagen', 'ignore_pyAudioAnalysis') ))\n",
    "ignore_librosa = bool(int(configParser.get('train_imagen', 'ignore_librosa') ))\n",
    "ignore_image_guide = bool(int(configParser.get('train_imagen', 'ignore_image_guide') ))\n",
    "ignore_additional_attributes = bool(int(configParser.get('train_imagen', 'ignore_additional_attributes') ))\n",
    "ignore_age = bool(int(configParser.get('train_imagen', 'ignore_age') ))\n",
    "ignore_gender = bool(int(configParser.get('train_imagen', 'ignore_gender') ))\n",
    "ignore_ethnicity = bool(int(configParser.get('train_imagen', 'ignore_ethnicity') ))\n",
    "ignore_language_spoken = bool(int(configParser.get('train_imagen', 'ignore_language_spoken') ))\n",
    "\n",
    "\n",
    "override_image_path = bool(int(configParser.get('train_imagen', 'override_image_path') ))\n",
    "override_image_path_string =  configParser.get('train_imagen', 'override_image_path_string')\n",
    "override_leave_parent_level = int(configParser.get('train_imagen', 'override_leave_parent_level') )\n",
    "\n",
    "max_checkpoints_keep = int(configParser.get('train_imagen', 'max_checkpoints_keep'))\n",
    "\n",
    "pd.set_option('display.max_columns', None)\n",
    "pd.options.mode.chained_assignment = None\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pathlib import Path\n",
    "\n",
    "\n",
    "def get_parent_by_level(file,parent_level):\n",
    "    path = Path(file)\n",
    "    file = path.parent.absolute()\n",
    "    if(parent_level == 0):\n",
    "        return str(file)\n",
    "    else:\n",
    "        return get_parent_by_level(file,parent_level - 1)\n",
    "\n",
    "def fix_image_path(og_path):\n",
    "    if(override_image_path):\n",
    "        parent_to_replace = get_parent_by_level(og_path,override_leave_parent_level)\n",
    "        new_path = og_path.replace( parent_to_replace, override_image_path_string)\n",
    "        #print(og_path)\n",
    "        #print(new_path)\n",
    "        return new_path\n",
    "    else:\n",
    "        return og_path"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def speaker_emb_preprocess(speaker_emb2):\n",
    "\n",
    "    if(ignore_speaker_embedding or ignore_speech_brain):\n",
    "        #print('ignore speech brain')\n",
    "        return np.zeros((1,768))\n",
    "    else:\n",
    "        speaker_emb2 = pickle.loads(speaker_emb2)\n",
    "        #print(speaker_emb2.print)\n",
    "        speaker_emb2 = speaker_emb2.squeeze()\n",
    "        speaker_emb2 = np.pad(speaker_emb2, (288), 'constant', constant_values=(0))\n",
    "        speaker_emb2 = np.tile(speaker_emb2, (1, 1))\n",
    "        speaker_emb2 = speaker_emb2 / 200.0\n",
    "        #print(speaker_emb2.shape)\n",
    "        #speaker_emb2 = np.array(speaker_emb2).tolist()\n",
    "        #print(type(speaker_emb2))\n",
    "        return speaker_emb2\n",
    "\n",
    "\n",
    "def audio_emb_preprocess2(speaker_emb2):\n",
    "    \n",
    "    if(ignore_pyannote_titanet_speakernet or ignore_speaker_embedding):\n",
    "        #print('ignore speech brain')\n",
    "        return np.zeros((28,768))\n",
    "    else:\n",
    "        #print(speaker_emb2)\n",
    "        speaker_emb2 = pickle.loads(speaker_emb2)\n",
    "        a = speaker_emb2 #np.zeros(shape=(24, 512))\n",
    "        a = a[-28:,:] # Actually only last 39 are the relevant ones\n",
    "        #print('au')\n",
    "        #print(a.shape)\n",
    "        b = np.zeros(shape=(a.shape[0], 768-a.shape[1]))\n",
    "        #c = np.zeros(shape=(28-a.shape[0], 768))\n",
    "        arr = np.concatenate((a, b), axis=1)\n",
    "        #arr = np.concatenate((arr, c), axis=0)\n",
    "        arr = arr / (1 if(audio_embs == 'wav2vec') else 10 if(audio_embs == 'openl3')  else 1 if(audio_embs == 'pyannoteTitaNet') else 1)\n",
    "        #print(str(arr.max()))\n",
    "        #print(arr.shape)\n",
    "        speaker_emb2 = np.array(arr)\n",
    "        return speaker_emb2\n",
    "\n",
    "def audio_features_preprocess(video_id,cursor_passed):\n",
    "    #  79 belong to pyaudioanalysis\n",
    "    # 111 belogn to liborsa\n",
    "    if(ignore_audio_features):\n",
    "        #print('ignore speech brain')\n",
    "        return np.zeros((190,768))\n",
    "    else:\n",
    "        #print(video_id)\n",
    "        if(cursor_passed != None):\n",
    "            cur = cursor_passed\n",
    "        else:\n",
    "            conAdditional = sl.connect(datasetPathDatabaseAdditional)\n",
    "            cur = conAdditional.cursor()\n",
    "        sql = '''SELECT AUDIO_FEATURES FROM AUDIO WHERE VIDEO_ID = ?'''\n",
    "        cur.execute(sql, [video_id])\n",
    "        audio_features = cur.fetchall()\n",
    "        #print(audio_features[0])\n",
    "        audio_features = pickle.loads(audio_features[0][0])\n",
    "        #print(audio_features.shape) # 190 x 128\n",
    "\n",
    "        #print(video_id)\n",
    "        #print(audio_features)\n",
    "\n",
    "        audio_features = audio_features[0:190]\n",
    "        \n",
    "        #import sys\n",
    "        #np.set_printoptions(threshold=sys.maxsize)\n",
    "        #print(np.argwhere(audio_features == 0))\n",
    "\n",
    "        if(ignore_pyAudioAnalysis and not ignore_librosa):\n",
    "            zpa = np.zeros((190-110,128))\n",
    "            audio_features = audio_features[80:190]\n",
    "            audio_features = np.vstack((audio_features,zpa))\n",
    "            #print('ignore pyAudioAnalysis')\n",
    "            #print(audio_features.shape)\n",
    "        elif(ignore_librosa and not ignore_pyAudioAnalysis):\n",
    "            zpa = np.zeros((190-80,128))\n",
    "            audio_features = audio_features[0:80]\n",
    "            audio_features = np.vstack((audio_features,zpa))\n",
    "            #print('ignore librosa')\n",
    "            #print(audio_features.shape)\n",
    "        else:\n",
    "            audio_features = np.zeros((190,128))\n",
    "            #print('ignore pyaudioanalysis and librosa')\n",
    "\n",
    "\n",
    "        z1 = np.zeros((190,768-128))\n",
    "        audio_features = np.hstack((audio_features,z1))\n",
    "        audio_features = audio_features / 100.0\n",
    "        #print(audio_features.shape)\n",
    "        if(cursor_passed != None):\n",
    "            pass\n",
    "        else:\n",
    "            cur.close()\n",
    "            conAdditional.close()\n",
    "    return audio_features\n",
    "\n",
    "def audio_transformer_features_preprocess(video_id):\n",
    "    #print(video_id)\n",
    "    conAdditional = sl.connect(datasetPathDatabaseAdditional)\n",
    "    cur = conAdditional.cursor()\n",
    "    sql = '''SELECT AUDIO_FEATURES FROM AUDIO_TRANSFORMER WHERE VIDEO_ID = ?'''\n",
    "    cur.execute(sql, [video_id])\n",
    "    audio_features = cur.fetchall()\n",
    "    #print(audio_features[0])\n",
    "    audio_features = pickle.loads(audio_features[0][0])\n",
    "    #print(audio_features.shape) # 514 x 768\n",
    "    audio_features = audio_features.squeeze()\n",
    "    audio_features = audio_features[0::3]#172*768\n",
    "    #z1 = np.zeros((161,768-128))\n",
    "    #audio_features = np.hstack((audio_features,z1))\n",
    "    #audio_features = audio_features / 100.0\n",
    "    return audio_features\n",
    "\n",
    "import random\n",
    "def image_guide_preprocess(face_id):\n",
    "    #print(face_id)\n",
    "\n",
    "    if(random.random() > 2):\n",
    "        image_guide = np.zeros((49,768))\n",
    "        #print(image_guide.shape)\n",
    "    else:\n",
    "        conAdditional = sl.connect(datasetPathDatabaseVgg)\n",
    "        cur = conAdditional.cursor()\n",
    "        sql = '''SELECT BLURRED_FACE_EMB FROM FACES_BLURRED WHERE FACE_ID = ?'''\n",
    "        cur.execute(sql, [face_id])\n",
    "        image_guide = cur.fetchall()\n",
    "        #print(image_guide)\n",
    "        image_guide = pickle.loads(image_guide[0][0])\n",
    "        image_guide = image_guide.squeeze() #197 x 768\n",
    "        image_guide = image_guide[1::4] #49 x 768\n",
    "    return image_guide\n",
    "\n",
    "\n",
    "#boxBlurMin =  int(configParser.get('extractVggBlurred', 'boxBlurMin'))\n",
    "#boxBlurMax =  int(configParser.get('extractVggBlurred', 'boxBlurMax'))\n",
    "\n",
    "#gaussianBlurMin =  int(configParser.get('extractVggBlurred', 'gaussianBlurMin'))\n",
    "#gaussianBlurMax =  int(configParser.get('extractVggBlurred', 'gaussianBlurMax'))\n",
    "\n",
    "\n",
    "from PIL import Image,ImageFilter\n",
    "import random\n",
    "#import cv2\n",
    "import numpy as np\n",
    "#from matplotlib import pyplot as plt\n",
    "  \n",
    "\n",
    "\n",
    "def image_guide_preprocess_low_res_dummy(path):\n",
    "    #print(face_id)\n",
    "\n",
    "    image_guide = np.zeros((1, 768))\n",
    "    return image_guide\n",
    "\n",
    "def image_guide_preprocess_low_res(path):\n",
    "    #print(face_id)\n",
    "\n",
    "\n",
    "    if(ignore_image_guide):\n",
    "        return np.nan\n",
    "    else:\n",
    "        path = fix_image_path(path)\n",
    "        image = Image.open(path).convert('RGB')\n",
    "\n",
    "        #print(image.size)\n",
    "\n",
    "        w_s = image_size / (1+2 * 0.4)\n",
    "        h_s = image_size / (1+2 * 0.4)\n",
    "\n",
    "        image = image.crop((0.2*w_s, 0.0*h_s, 1.6*w_s, 1.4*h_s))\n",
    "\n",
    "        image = image.resize((image_size,image_size))\n",
    "\n",
    "        image = image.resize((begin_with_image_size,begin_with_image_size))\n",
    "        im = image\n",
    "        \n",
    "\n",
    "        #print('saving')\n",
    "        #image.save('opop.png')\n",
    "\n",
    "        #print(np.array(image,np.float32).shape)\n",
    "\n",
    "        pix = np.array(image, np.float32)\n",
    "        pix = np.moveaxis(pix, -1, 0)\n",
    "\n",
    "        pix = pix / 255\n",
    "        image.close()\n",
    "        im.close()\n",
    "        return pix\n",
    "    return image_guide\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "\n",
    "def process_age(age):\n",
    "    if(random.random() < 0.2 or ignore_age or ignore_additional_attributes):\n",
    "        x = np.zeros(768)\n",
    "        x[767] = 1\n",
    "        return x\n",
    "    else:\n",
    "        try:\n",
    "            x = np.ones(768) * (age / 100.0)\n",
    "            x[767] = 0\n",
    "        except:\n",
    "            x = np.zeros(768)\n",
    "            x[767] = 1\n",
    "\n",
    "        return x\n",
    "\n",
    "def process_gender(gender):\n",
    "    if(random.random() < 0.2 or ignore_gender or ignore_additional_attributes):\n",
    "        return np.zeros(768)\n",
    "    elif(gender == 'man'):\n",
    "        return np.ones(768)\n",
    "    elif(gender == \"woman\"):\n",
    "        return np.ones(768) * -1\n",
    "    else:\n",
    "        return np.zeros(768)\n",
    "    \n",
    "# TODO\n",
    "def process_ethnicity(eth):\n",
    "    x = np.zeros(768)\n",
    "    if(random.random() < 0.2 or ignore_ethnicity or ignore_additional_attributes):\n",
    "        x = x\n",
    "    elif(eth == \"indian\"):\n",
    "        x[0] = 1\n",
    "    elif(eth == \"asian\"):\n",
    "        x[16]=1\n",
    "    elif(eth == \"latino hispanic\"):\n",
    "        x[2]=1\n",
    "    elif(eth == \"black\"):\n",
    "        x[3]=1\n",
    "    elif(eth == \"middle eastern\"):\n",
    "        x[4]=1\n",
    "    elif(eth == \"white\"):\n",
    "        x[5]=1 \n",
    "    else:\n",
    "        x = x\n",
    "    return x     \n",
    "\n",
    "def process_language(lan):\n",
    "    x = np.zeros(768)\n",
    "    if(random.random() < 0.2 or ignore_language_spoken or ignore_additional_attributes):\n",
    "        x = x\n",
    "    elif(lan == \"Arabic\"):\n",
    "        x[0] = 1\n",
    "    elif(lan == \"Portuguese\"):\n",
    "        x[16]=1\n",
    "    elif(lan == \"Romansh_Sursilvan\"):\n",
    "        x[2]=1\n",
    "    elif(lan == \"Japanese\"):\n",
    "        x[3]=1\n",
    "    elif(lan == \"Ukranian\"):\n",
    "        x[4]=1\n",
    "    elif(lan == \"German\"):\n",
    "        x[5]=1   \n",
    "    elif(lan == \"Chinese_China\"):\n",
    "        x[6]=1   \n",
    "    elif(lan == \"Welsh\"):\n",
    "        x[7]=1  \n",
    "    elif(lan == \"English\"):\n",
    "        x[8]=1\n",
    "    elif(lan == \"Kabyle\"):\n",
    "        x[9]=1 \n",
    "    elif(lan == \"Kyrgyz\"):\n",
    "        x[10]=1\n",
    "    elif(lan == \"Georgian\"):\n",
    "        x[11]=1\n",
    "    elif(lan == \"Persian\"):\n",
    "        x[12]=1 \n",
    "    elif(lan == \"French\"):\n",
    "        x[13]=1\n",
    "    elif(lan == \"Interlingua\"):\n",
    "        x[14]=1\n",
    "    elif(lan == \"Swedish\"):\n",
    "        x[15]=1\n",
    "    elif(lan == \"Spanish\"):\n",
    "        x[16]=1 \n",
    "    elif(lan == \"Dhivehi\"):\n",
    "        x[17]=1\n",
    "    elif(lan == \"Kinyarwanda\"):\n",
    "        x[18]=1 \n",
    "    elif(lan == \"Tatar\"):\n",
    "        x[19]=1\n",
    "    elif(lan == \"Hakha_Chin\"):\n",
    "        x[20]=1 \n",
    "    elif(lan == \"Tamil\"):\n",
    "        x[21]=1 \n",
    "    elif(lan == \"Greek\"):\n",
    "        x[22]=1\n",
    "    elif(lan == \"Latvian\"):\n",
    "        x[23]=1 \n",
    "    elif(lan == \"Russian\"):\n",
    "        x[24]=1\n",
    "    elif(lan == \"Breton\"):\n",
    "        x[25]=1\n",
    "    elif(lan == \"Catalan\"):\n",
    "        x[26]=1    \n",
    "    elif(lan == \"Maltese\"):\n",
    "        x[27]=1 \n",
    "    elif(lan == \"Slovenian\"):\n",
    "        x[28]=1    \n",
    "    elif(lan == \"Indonesian\"):\n",
    "        x[29]=1    \n",
    "    elif(lan == \"Dutch\"):\n",
    "        x[30]=1\n",
    "    elif(lan == \"Chinese_Taiwan\"):\n",
    "        x[31]=1 \n",
    "    elif(lan == \"Sakha\"):\n",
    "        x[32]=1 \n",
    "    elif(lan == \"Polish\"):\n",
    "        x[33]=1 \n",
    "    elif(lan == \"Czech\"):\n",
    "        x[34]=1 \n",
    "    elif(lan == \"Romanian\"):\n",
    "        x[35]=1 \n",
    "    elif(lan == \"Mangolian\"):\n",
    "        x[36]=1 \n",
    "    elif(lan == \"Italian\"):\n",
    "        x[37]=1 \n",
    "    elif(lan == \"Chinese_Hongkong\"):\n",
    "        x[38]=1 \n",
    "    elif(lan == \"Estonian\"):\n",
    "        x[39]=1 \n",
    "    elif(lan == \"Basque\"):\n",
    "        x[40]=1 \n",
    "    elif(lan == \"Esperanto\"):\n",
    "        x[41]=1 \n",
    "    elif(lan == \"Frisian\"):\n",
    "        x[42]=1 \n",
    "    elif(lan == \"Turkish\"):\n",
    "        x[43]=1 \n",
    "    elif(lan == \"Chuvash\"):\n",
    "        x[44]=1 \n",
    "    else:\n",
    "        x = x\n",
    "    return x       "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "def process_input(row):\n",
    "    #print(row)\n",
    "    age = row[\"caption_a\"]\n",
    "    ethnicity=row[\"caption_e\"]\n",
    "    gender=row[\"caption_g\"]\n",
    "    language=row[\"caption_l\"]\n",
    "    speaker_emb=row[\"SPEAKER_EMB\"]\n",
    "    audio_emb=row[\"AUDIO_EMB\"]\n",
    "    audio_features=row[\"AUDIO_FEATURES\"]\n",
    "    image_guide=row[\"image_guide\"]\n",
    "    #print(image_guide)\n",
    "    #print(speaker_emb)\n",
    "    speaker_emb = np.asarray(speaker_emb, dtype=np.float32)\n",
    "    speaker_emb = speaker_emb.squeeze()\n",
    "    audio_emb = np.asarray(audio_emb, dtype=np.float32)\n",
    "    audio_emb = audio_emb.squeeze()\n",
    "    #print(speaker_emb.shape)\n",
    "    h = np.vstack((age, ethnicity))\n",
    "    h = np.vstack((h, gender))\n",
    "    h = np.vstack((h, language))\n",
    "    h = np.vstack((h, speaker_emb))\n",
    "    h = np.vstack((h, audio_emb))\n",
    "    h = np.vstack((h,audio_features))\n",
    "    h = np.vstack((h,image_guide))\n",
    "    #print('aaaaaaaaaaa')\n",
    "    #print(h.shape)\n",
    "    j = np.zeros(768)\n",
    "    j = np.tile(j,(256-h.shape[0],1))\n",
    "    h = np.vstack((h, j))\n",
    "    #print(h.shape)\n",
    "    \n",
    "    return h\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "def process_image_path(path):\n",
    "    \n",
    "    path = fix_image_path(path)\n",
    "    image = Image.open(path).convert('RGB')\n",
    "\n",
    "    w_s = image_size / (1+2 * 0.4)\n",
    "    h_s = image_size / (1+2 * 0.4)\n",
    "\n",
    "    #print(image.size)\n",
    "    image = image.crop((0.2*w_s,0.0*h_s,1.6*w_s,1.4*h_s))\n",
    "    image = image.resize((image_size,image_size))\n",
    "\n",
    "    #print('saving')\n",
    "    #image.save(str(random.random()) + '.png')\n",
    "\n",
    "    \n",
    "\n",
    "    #print(np.array(image,np.float32).shape)\n",
    "    pix = np.array(image,np.float32)\n",
    "    pix = np.moveaxis(pix, -1, 0)\n",
    "    \n",
    "    pix = pix / 255\n",
    "    image.close()\n",
    "    return pix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.signal import savgol_filter\n",
    "from torch.utils.data import TensorDataset, DataLoader\n",
    "import time\n",
    "import torch\n",
    "from torch.utils.data import Dataset\n",
    "from torch.utils.data import DataLoader\n",
    "from torchvision.transforms import ToTensor,Compose\n",
    "import dask.dataframe as dd\n",
    "from multiprocessing import Manager,RLock,Value,Queue\n",
    "import ctypes\n",
    "from random import randrange\n",
    "import random\n",
    "\n",
    "class CustomDataset(Dataset):\n",
    "\n",
    "    total_rows = None\n",
    "    con = None\n",
    "    con_additional = None\n",
    "    lock = None\n",
    "    samples_seen =  Value('i', 0,lock=False)\n",
    "    c = None\n",
    "    c_additional = None\n",
    "    \n",
    "    def no_of_samples_seen(self):\n",
    "        return self.samples_seen.value\n",
    "\n",
    "    def samples_seen_multiple_times(self,no_of_samples_seen_value,multiple_times):\n",
    "        self.lock.acquire(block=True)\n",
    "        self.samples_seen.value = self.samples_seen.value + (no_of_samples_seen_value * (multiple_times - 1))\n",
    "        self.lock.release()\n",
    "\n",
    "    def __init__(self,datasetPathDatabase_a,lock,last_sample_no,datasetPathDatabaseAdditional_a):\n",
    "        self.lock = lock\n",
    "        self.samples_seen.value = last_sample_no\n",
    "        self.con = sl.connect(datasetPathDatabase_a)\n",
    "        self.con.isolation_level = None\n",
    "        self.c = self.con.cursor()\n",
    "        self.c.execute(\"PRAGMA journal_mode=OFF\")\n",
    "        self.c.execute(\"begin\")\n",
    "\n",
    "        print(datasetPathDatabase_a)\n",
    "        print(datasetPathDatabaseAdditional_a)\n",
    "\n",
    "        \n",
    "        self.con_additional = sl.connect(datasetPathDatabaseAdditional_a,check_same_thread=False)\n",
    "        self.con_additional.isolation_level = None\n",
    "        self.c_additional = self.con_additional.cursor()\n",
    "        self.c_additional.execute(\"PRAGMA journal_mode=OFF\")\n",
    "        self.c_additional.execute(\"begin\")\n",
    "        pass\n",
    "\n",
    "    def destroy(self):\n",
    "        self.lock.acquire(block=True)\n",
    "        self.c.execute(\"commit\")\n",
    "        self.c.close()\n",
    "        self.con.close()\n",
    "\n",
    "        self.c_additional.execute(\"commit\")\n",
    "        self.c_additional.close()\n",
    "        self.con_additional.close()\n",
    "\n",
    "        self.lock.release()\n",
    "\n",
    "\n",
    "\n",
    "    def __len__(self):\n",
    "        \n",
    "        if(self.total_rows == None):\n",
    "            self.lock.acquire(block=True)\n",
    "            data = self.con.execute(\"select count(V2.ID) from VIDEO v2 WHERE  AUDIO_PRE IN (3,4) AND FACES_PRE = 2 \")\n",
    "            count = data.fetchall()\n",
    "            self.total_rows = count[0][0]\n",
    "            data.close()\n",
    "            print(\"Found \" + str(self.total_rows) + \" data points.\")\n",
    "\n",
    "            #data = self.con.execute(\"SELECT COUNT(*) FROM VIDEO v WHERE V.TRAINED = 1\")\n",
    "            #count = data.fetchall()\n",
    "            #data.close()\n",
    "            #self.samples_seen.value = count[0][0]\n",
    "            #print(\"Starting from \" + str(self.samples_seen.value) + \" data points.\")\n",
    "            self.lock.release()\n",
    "\n",
    "        return self.total_rows\n",
    "    def get_row_data(self,idx):\n",
    "\n",
    "        sql = (\"SELECT V.ID,F.ID,V.VIDEO_PATH, V.AGE CAPTION_A, \" + \n",
    "                            \"V.ETHNICITY CAPTION_E, \" +\n",
    "                            \"lower(V.GENDER) CAPTION_G, \" +\n",
    "                                \"A.SPEAKER_EMB, \"+ (\"A.WAV_TO_VEC, \" if(audio_embs == 'wav2vec') else \"A.AUDIO_EMB2, \" if(audio_embs == 'openl3')  else \"A.PYANNOTE_TITANET, \" if(audio_embs == 'pyannoteTitaNet') else ', ') +\n",
    "                            \"A.AUDIO_FEATURES, \" +\n",
    "                            \"A.LANG CAPTION_L, \"+\n",
    "                            \"F.FACE_PATH \"+\n",
    "                            \"FROM VIDEO V \"+\n",
    "                            \"INNER JOIN AUDIO A ON V.ID = A.VIDEO_ID \" +\n",
    "                            \"INNER JOIN FACE F ON F.ID = (select ID from FACE f2 where f2.video_id = v.ID ORDER By ID limit 1 ) \" + \n",
    "                            \"WHERE AUDIO_LENGTH = \" + audio_length_used + ' ' +\n",
    "                            \"AND V.ID in (select V2.ID from VIDEO v2 WHERE  AUDIO_PRE IN (3,4) AND FACES_PRE = 2 LIMIT  1 OFFSET \" +str(idx) + \" )\" )\n",
    "        data = self.con.execute(sql)\n",
    "        dataGotten = data.fetchall()\n",
    "        df = pd.DataFrame(dataGotten,columns = ['ID','FACE_ID','VIDEO_PATH','caption_a','caption_e','caption_g','SPEAKER_EMB','AUDIO_EMB','AUDIO_FEATURES','caption_l','image_path'])\n",
    "        data.close()\n",
    "        return df\n",
    "\n",
    "    def __getitem__(self, idx):\n",
    "        self.lock.acquire(block=True)\n",
    "        #print(str(idx) + \"/n\")\n",
    "        while True:\n",
    "            df = self.get_row_data(idx)\n",
    "            if(len(df) != 0):\n",
    "                break\n",
    "            print(\"retrying for data\")\n",
    "            idx = random.randint(1, self.total_rows)\n",
    "        #print(df)\n",
    "        \n",
    "        self.samples_seen.value +=  1\n",
    "        if not idx%100:\n",
    "            print (self.samples_seen.value,end='\\r')\n",
    "\n",
    "\n",
    "        self.lock.release()\n",
    "\n",
    "        \n",
    "        \n",
    "        df = df.sample(frac=1).reset_index(drop=True)\n",
    "        df = dd.from_pandas(df, npartitions=1)\n",
    "        df[\"image_guide\"] = np.nan\n",
    "\n",
    "        #print(df.head(10))\n",
    "        data_frame = df[[\"ID\",\"FACE_ID\",\"image_path\",\"caption_a\",\"caption_e\",\"caption_g\",\"caption_l\"]]\n",
    "        data_frame['SPEAKER_EMB'] = df['SPEAKER_EMB']\n",
    "        data_frame['AUDIO_EMB'] = df['AUDIO_EMB']\n",
    "        data_frame['image_guide'] = df['image_guide']\n",
    "\n",
    "        \n",
    "        data_frame['SPEAKER_EMB'] = data_frame['SPEAKER_EMB'].apply(lambda x: speaker_emb_preprocess(x),meta=('1', 'object'))\n",
    "        data_frame['AUDIO_EMB'] = data_frame['AUDIO_EMB'].apply(lambda x: audio_emb_preprocess2(x),meta=('1', 'object'))\n",
    "        data_frame['AUDIO_FEATURES'] = data_frame['ID'].apply(lambda x: audio_features_preprocess(x,self.c_additional),meta=('1', 'object'))\n",
    "        data_frame = data_frame.drop(['ID'], axis=1)\n",
    "        data_frame['image_guide'] = data_frame['image_path'].apply(lambda x: image_guide_preprocess_low_res_dummy(x),meta=('1', 'str'))\n",
    "        data_frame = data_frame.drop(['FACE_ID'], axis=1)\n",
    "        data_frame['caption_a'] = data_frame['caption_a'].apply(lambda x: process_age(x),meta=('1', 'object'))\n",
    "        data_frame['caption_g'] = data_frame['caption_g'].apply(lambda x: process_gender(x),meta=('1', 'object'))\n",
    "        data_frame['caption_l'] = data_frame['caption_l'].apply(lambda x: process_language(x),meta=('1', 'object'))\n",
    "        data_frame['caption_e'] = data_frame['caption_e'].apply(lambda x: process_ethnicity(x),meta=('1', 'object'))\n",
    "        data_frame['low_res_image'] = data_frame['image_path'].apply(lambda x: image_guide_preprocess_low_res(x),meta=('1', 'object'))\n",
    "\n",
    "        data_frame['INPUT'] = data_frame['SPEAKER_EMB']\n",
    "        #print(data_frame)\n",
    "\n",
    "        data_frame['INPUT'] = data_frame.apply(process_input,args=(),axis=1,meta=('1', 'object'))\n",
    "\n",
    "        #for index, row in data_frame.iterrows():\n",
    "        #    x =946922              data_frame.loc[index,\"caption_l\"],data_frame.loc[index,\"SPEAKER_EMB\"],\n",
    "        #                    data_frame.loc[index,\"AUDIO_EMB\"],data_frame.loc[index,\"AUDIO_FEATURES\"],\n",
    "        #                    data_frame.loc[index,\"image_guide\"])\n",
    "        #    x = [x]\n",
    "        #    #AADFS = AADFS\n",
    "        #    data_frame.loc[index,\"INPUT\"] = x\n",
    "\n",
    "        data_frame = data_frame.drop(['caption_e', 'caption_g','caption_l','SPEAKER_EMB','AUDIO_EMB'], axis=1)\n",
    "\n",
    "\n",
    "        data_frame['image_path'] = data_frame['image_path'].apply(lambda x: process_image_path(x),meta=('1', 'str'))\n",
    "        data_frame.compute()\n",
    "\n",
    "        part = data_frame[['INPUT','image_path','low_res_image']].partitions[0]\n",
    "\n",
    "                \n",
    "\n",
    "        input0 = np.array(part['INPUT'])\n",
    "        input0 = np.array([np.array(xi) for xi in input0])\n",
    "        input0[np.isnan(input0)] = 0\n",
    "        input0[input0 > 10] = 10\n",
    "        input0[input0 < -10] = -10\n",
    "        #print(sys.getsizeof(input))\n",
    "        \n",
    "        output = np.array(part['image_path'])\n",
    "        output = np.array([np.array(xi) for xi in output])\n",
    "        output.squeeze().shape\n",
    "        #print(sys.getsizeof(output))\n",
    "\n",
    "        input2 = np.array(part['low_res_image'])\n",
    "        input2 = np.array([np.array(xi) for xi in input2])\n",
    "        input2.squeeze()\n",
    "        #print(sys.getsizeof(input2))\n",
    "        \n",
    "\n",
    "\n",
    "    \n",
    "        \n",
    "        #print(len(procs))\n",
    "        #if(len(procs) > 0):\n",
    "        #    procs[0].join() # Wait for previous process to finish\n",
    "        #    print(\"Model trained using a batch of data...\")\n",
    "        #    prProcess,\n",
    "        input0 = torch.from_numpy(input0)\n",
    "        input0 = input0.to(torch.float)\n",
    "\n",
    "\n",
    "        input2 = torch.from_numpy(input2)\n",
    "        input2 = input2.to(torch.float)\n",
    "\n",
    "        output = torch.from_numpy(output)\n",
    "        output = output.to(torch.float)\n",
    "\n",
    "        input0 = input0.squeeze()\n",
    "        input2 = input2.squeeze()\n",
    "        output = output.squeeze()\n",
    "\n",
    "        return output,input0\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "def plot_loss(sample_no,loss_value,plot_now,loss_total):\n",
    "\n",
    "\n",
    "    if(loss_total == None):\n",
    "        if(UNET == 1):\n",
    "            my_file = Path(model_filename + '/loss_total_1.picke')\n",
    "            if my_file.is_file():\n",
    "                with open(model_filename + '/loss_total_1.picke', 'rb') as handle:\n",
    "                    loss_total = pickle.load(handle)\n",
    "            else:\n",
    "                loss_total = []\n",
    "        else:\n",
    "            my_file = Path(model_filename + '/loss_total_2.picke')\n",
    "            if my_file.is_file():\n",
    "                with open(model_filename + '/loss_total_2.picke', 'rb') as handle:\n",
    "                    loss_total = pickle.load(handle)\n",
    "            else:\n",
    "                loss_total = []\n",
    "    \n",
    "    #print(loss_list)\n",
    "    #print(loss_total)\n",
    "    loss_total.append([sample_no,loss_value])\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "    if(plot_now):\n",
    "\n",
    "\n",
    "        if(UNET == 1):\n",
    "            with open(model_filename +'/loss_total_1.picke', 'wb') as handle:\n",
    "                pickle.dump(loss_total, handle)\n",
    "        else:\n",
    "            with open(model_filename +'/loss_total_2.picke', 'wb') as handle:\n",
    "                pickle.dump(loss_total, handle)\n",
    "        #print(loss_total)\n",
    "\n",
    "        fig = plt.figure()\n",
    "        x_val = [x[0] for x in loss_total]\n",
    "        y_val = [x[1] for x in loss_total]\n",
    "        plt.plot(x_val,y_val)\n",
    "        plt.title(\"Training Loss\")\n",
    "        plt.xlabel(\"Training Sample\")\n",
    "        plt.ylabel(\"MSE Loss\")\n",
    "        \n",
    "\n",
    "\n",
    "\n",
    "        if(UNET == 1):\n",
    "            if(os.path.isfile(model_filename + '/loss_1_plot.png')):\n",
    "                os.remove(model_filename + '/loss_1_plot.png')\n",
    "            fig.savefig(model_filename + '/loss_1_plot.png')\n",
    "        else:\n",
    "            if(os.path.isfile(model_filename + '/loss_2_plot.png')):\n",
    "                os.remove(model_filename + '/loss_2_plot.png')\n",
    "            fig.savefig(model_filename + '/loss_2_plot.png')\n",
    "        plt.close()\n",
    "\n",
    "        try:\n",
    "            fig = plt.figure()\n",
    "            plt.plot(x_val[1000::],y_val[1000::]\n",
    "                    ,'.',zorder=-100)\n",
    "\n",
    "            #plt.axvline(x=1000,linestyle='--',color='green',label='1000 inner epochs')\n",
    "            #plt.axvline(x=17000,linestyle='--',color='purple',label='100 inner epochs - 100000 unique samples seen' )\n",
    "            #plt.axvline(x=23100,linestyle='-.',color='black',label='1 inner epoch - end of 1st epoch ')\n",
    "            #plt.legend(bbox_to_anchor = (1.0, 1), loc = 'upper right')\n",
    "            plt.grid()\n",
    "            yhat = savgol_filter(y_val, 1000, 3)\n",
    "            plt.plot(x_val[1000::],yhat[1000::],'r')\n",
    "            plt.title(\"Training Loss\")\n",
    "            plt.xlabel(\"Training Sample\")\n",
    "            plt.ylabel(\"MSE Loss\")\n",
    "            \n",
    "\n",
    "\n",
    "\n",
    "            if(UNET == 1):\n",
    "                if(os.path.isfile(model_filename + '/loss_zoomed_1_plot.png')):\n",
    "                    os.remove(model_filename + '/loss_zoomed_1_plot.png')\n",
    "                fig.savefig(model_filename + '/loss_zoomed_1_plot.png')\n",
    "            else:\n",
    "                if(os.path.isfile(model_filename + '/loss_zoomed_2_plot.png')):\n",
    "                    os.remove(model_filename + '/loss_zoomed_2_plot.png')\n",
    "                fig.savefig(model_filename + '/loss_zoomed_2_plot.png')\n",
    "            plt.close()\n",
    "\n",
    "\n",
    "        except:\n",
    "            try:\n",
    "                plt.close()\n",
    "            except:\n",
    "                pass\n",
    "\n",
    "\n",
    "    return loss_total\n",
    "\n",
    "\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "last_sample_no = 0\n",
    "if(UNET == 1):\n",
    "    my_file = Path(model_filename + '/last_sample_no1.picke')\n",
    "    if my_file.is_file():\n",
    "        with open(model_filename + '/last_sample_no1.picke', 'rb') as handle:\n",
    "            last_sample_no = pickle.load(handle)\n",
    "else:\n",
    "    my_file = Path(model_filename + '/last_sample_no2.picke')\n",
    "    if my_file.is_file():\n",
    "        with open(model_filename + '/last_sample_no2.picke', 'rb') as handle:\n",
    "            last_sample_no = pickle.load(handle)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/data_store/databases/dataset.db\n",
      "/data_store/databases/datasetAdditional.db\n",
      "Found 402440 data points.\n",
      "402440\n"
     ]
    }
   ],
   "source": [
    "my_dataset = CustomDataset(datasetPathDatabase,RLock(),last_sample_no,datasetPathDatabaseAdditional)\n",
    "print(len(my_dataset))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "my_dataloader = DataLoader(my_dataset,batch_size = batch_size * 16 * 4,shuffle=True,num_workers=16) # create your dataloader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "#for test_images, test_labels in my_dataloader:  \n",
    "#    pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "checkpoint loaded from /data_store/checkpoint/checkpoint.211566.pt\n",
      "epoch:1 from 100000\n",
      "loss: 0.0030716538535671134\n",
      "epoch:2 from 100000\n",
      "loss: 0.0028201217017738145\n",
      "epoch:3 from 100000\n",
      "loss: 0.0023626026441547765\n",
      "epoch:4 from 100000\n",
      "loss: 0.0021428935996823384\n",
      "epoch:5 from 100000\n",
      "loss: 0.001929764614231999\n",
      "epoch:6 from 100000\n",
      "loss: 0.0026440658397461902\n",
      "epoch:7 from 100000\n",
      "loss: 0.0021903172148256544\n",
      "epoch:8 from 100000\n",
      "loss: 0.0021356378917971597\n",
      "epoch:9 from 100000\n",
      "loss: 0.002322255689193753\n",
      "epoch:10 from 100000\n",
      "loss: 0.002570857139119198\n",
      "epoch:11 from 100000\n",
      "loss: 0.0021705554493394175\n",
      "epoch:12 from 100000\n",
      "loss: 0.0024437696019390387\n",
      "epoch:13 from 100000\n",
      "loss: 0.002340624006464509\n",
      "epoch:14 from 100000\n",
      "loss: 0.002411892741747579\n",
      "epoch:15 from 100000\n",
      "loss: 0.0020317272061163294\n",
      "epoch:16 from 100000\n",
      "loss: 0.002928948836768086\n",
      "epoch:17 from 100000\n",
      "loss: 0.0023213974389761916\n",
      "epoch:18 from 100000\n",
      "loss: 0.0026455830617635456\n",
      "epoch:19 from 100000\n",
      "loss: 0.0020918298231435983\n",
      "epoch:20 from 100000\n",
      "loss: 0.002395878139736851\n",
      "epoch:21 from 100000\n",
      "loss: 0.002459607023638455\n",
      "epoch:22 from 100000\n",
      "loss: 0.0024817120290947514\n",
      "epoch:23 from 100000\n",
      "loss: 0.002576664724301736\n",
      "epoch:24 from 100000\n",
      "loss: 0.00272593299143864\n",
      "epoch:25 from 100000\n",
      "loss: 0.002594339476672758\n",
      "epoch:26 from 100000\n",
      "loss: 0.0024736074584836842\n",
      "epoch:27 from 100000\n",
      "loss: 0.00219323273131522\n",
      "epoch:28 from 100000\n",
      "loss: 0.0024135130186664355\n",
      "epoch:29 from 100000\n",
      "loss: 0.0026876575322205554\n",
      "epoch:30 from 100000\n",
      "loss: 0.0024212136762997716\n",
      "epoch:31 from 100000\n",
      "loss: 0.0021760346108408157\n",
      "epoch:32 from 100000\n",
      "loss: 0.002065652336710855\n",
      "epoch:33 from 100000\n",
      "loss: 0.002469026726373613\n",
      "epoch:34 from 100000\n",
      "loss: 0.002641555206423618\n",
      "epoch:35 from 100000\n",
      "loss: 0.0031880444067837743\n",
      "epoch:36 from 100000\n",
      "loss: 0.0029746381156314783\n",
      "epoch:37 from 100000\n",
      "loss: 0.002968420272821959\n",
      "epoch:38 from 100000\n",
      "loss: 0.002725503750752978\n",
      "epoch:39 from 100000\n",
      "loss: 0.0026035005308813197\n",
      "epoch:40 from 100000\n",
      "loss: 0.002022465465302048\n",
      "epoch:41 from 100000\n",
      "loss: 0.0028119312179057943\n",
      "epoch:42 from 100000\n",
      "loss: 0.0026347750715558504\n",
      "epoch:43 from 100000\n",
      "loss: 0.0026488275725000676\n",
      "epoch:44 from 100000\n",
      "loss: 0.0019608762314931028\n",
      "epoch:45 from 100000\n",
      "loss: 0.0026644643611462016\n",
      "epoch:46 from 100000\n",
      "loss: 0.0027504865732908\n",
      "epoch:47 from 100000\n",
      "loss: 0.0018373755078755494\n",
      "epoch:48 from 100000\n",
      "loss: 0.0027811693327350895\n",
      "epoch:49 from 100000\n",
      "loss: 0.001883651886316784\n",
      "epoch:50 from 100000\n",
      "loss: 0.0026353485748700223\n",
      "epoch:51 from 100000\n",
      "loss: 0.002262458970079706\n",
      "epoch:52 from 100000\n",
      "loss: 0.0026402156421454492\n",
      "epoch:53 from 100000\n",
      "loss: 0.0024495384435567757\n",
      "epoch:54 from 100000\n",
      "loss: 0.0022027205625079205\n",
      "epoch:55 from 100000\n",
      "loss: 0.0021553822923010557\n",
      "epoch:56 from 100000\n",
      "loss: 0.0025497311894184804\n",
      "epoch:57 from 100000\n",
      "loss: 0.002922733926880028\n",
      "epoch:58 from 100000\n",
      "loss: 0.002482984495316032\n",
      "epoch:59 from 100000\n",
      "loss: 0.002575984756987282\n",
      "epoch:60 from 100000\n",
      "loss: 0.0024585114686885845\n",
      "epoch:61 from 100000\n",
      "loss: 0.002295532949688095\n",
      "epoch:62 from 100000\n",
      "loss: 0.0019854019691223357\n",
      "epoch:63 from 100000\n",
      "loss: 0.002386111009883507\n",
      "epoch:64 from 100000\n",
      "loss: 0.002508103117122573\n",
      "epoch:65 from 100000\n",
      "loss: 0.002271407433454442\n",
      "epoch:66 from 100000\n",
      "loss: 0.0019062007571051254\n",
      "epoch:67 from 100000\n",
      "loss: 0.002877023623796049\n",
      "epoch:68 from 100000\n",
      "loss: 0.00301484191209056\n",
      "epoch:69 from 100000\n",
      "loss: 0.0028548560231058673\n",
      "epoch:70 from 100000\n",
      "loss: 0.001826000666113714\n",
      "epoch:71 from 100000\n",
      "loss: 0.0025866563325394054\n",
      "epoch:72 from 100000\n",
      "loss: 0.0017888507114776075\n",
      "epoch:73 from 100000\n",
      "loss: 0.0023365814567455345\n",
      "epoch:74 from 100000\n",
      "loss: 0.002289691668266869\n",
      "epoch:75 from 100000\n",
      "loss: 0.002960622862758555\n",
      "epoch:76 from 100000\n",
      "loss: 0.0025482241052827703\n",
      "epoch:77 from 100000\n",
      "loss: 0.0028839760942389603\n",
      "epoch:78 from 100000\n",
      "loss: 0.0030441185535039494\n",
      "epoch:79 from 100000\n",
      "loss: 0.0025770736521337767\n",
      "epoch:80 from 100000\n",
      "loss: 0.0024410172019884158\n",
      "epoch:81 from 100000\n",
      "loss: 0.003154123874111292\n",
      "epoch:82 from 100000\n",
      "loss: 0.0027953286849822234\n",
      "epoch:83 from 100000\n",
      "loss: 0.002665250580321299\n",
      "epoch:84 from 100000\n",
      "loss: 0.002687024885460687\n",
      "epoch:85 from 100000\n",
      "loss: 0.0025896073129274555\n",
      "epoch:86 from 100000\n",
      "loss: 0.0024350257842513656\n",
      "epoch:87 from 100000\n",
      "loss: 0.0028891833753164065\n",
      "epoch:88 from 100000\n",
      "loss: 0.0019643520892316246\n",
      "epoch:89 from 100000\n",
      "loss: 0.002086823445011987\n",
      "epoch:90 from 100000\n",
      "loss: 0.0024977168712041475\n",
      "epoch:91 from 100000\n",
      "loss: 0.0021640063202426063\n",
      "epoch:92 from 100000\n",
      "loss: 0.002527796308498864\n",
      "epoch:93 from 100000\n",
      "loss: 0.002553364957435633\n",
      "epoch:94 from 100000\n",
      "loss: 0.002461032612359304\n",
      "epoch:95 from 100000\n",
      "loss: 0.002435919417052901\n",
      "epoch:96 from 100000\n",
      "loss: 0.002104467544711497\n",
      "epoch:97 from 100000\n",
      "loss: 0.0025827055953584477\n",
      "epoch:98 from 100000\n",
      "loss: 0.002321065550347612\n",
      "epoch:99 from 100000\n",
      "loss: 0.0029724255161340807\n",
      "epoch:100 from 100000\n",
      "loss: 0.0028766603411440883\n",
      "epoch:101 from 100000\n",
      "loss: 0.002470585391080249\n",
      "epoch:102 from 100000\n",
      "loss: 0.002609309151864636\n",
      "epoch:103 from 100000\n",
      "loss: 0.002225229408281848\n",
      "epoch:104 from 100000\n",
      "loss: 0.0021761137398357278\n",
      "epoch:105 from 100000\n",
      "loss: 0.0021987183531280152\n",
      "epoch:106 from 100000\n",
      "loss: 0.002907946399091088\n",
      "epoch:107 from 100000\n",
      "loss: 0.0024659287287960296\n",
      "epoch:108 from 100000\n",
      "loss: 0.0024706753407883753\n",
      "epoch:109 from 100000\n",
      "loss: 0.002428801551872084\n",
      "epoch:110 from 100000\n",
      "loss: 0.002357679271113966\n",
      "epoch:111 from 100000\n",
      "loss: 0.0025864677111826495\n",
      "epoch:112 from 100000\n",
      "loss: 0.0019617618370375567\n",
      "epoch:113 from 100000\n",
      "loss: 0.002039836572294007\n",
      "epoch:114 from 100000\n",
      "loss: 0.002430975744374564\n",
      "epoch:115 from 100000\n",
      "loss: 0.0022189760021262828\n",
      "epoch:116 from 100000\n",
      "loss: 0.0028534020061528054\n",
      "epoch:117 from 100000\n",
      "loss: 0.0023667769638677782\n",
      "epoch:118 from 100000\n",
      "loss: 0.002448892014086823\n",
      "epoch:119 from 100000\n",
      "loss: 0.0028178107454550627\n",
      "epoch:120 from 100000\n",
      "loss: 0.002915111824336236\n",
      "epoch:121 from 100000\n",
      "loss: 0.002640897142129006\n",
      "epoch:122 from 100000\n",
      "loss: 0.0021406431104651347\n",
      "epoch:123 from 100000\n",
      "loss: 0.0024864016593113547\n",
      "epoch:124 from 100000\n",
      "loss: 0.0023974817486589473\n",
      "epoch:125 from 100000\n",
      "loss: 0.002398299817116367\n",
      "epoch:126 from 100000\n",
      "loss: 0.0028024740915668644\n",
      "epoch:127 from 100000\n",
      "loss: 0.0025950290163052614\n",
      "epoch:128 from 100000\n",
      "loss: 0.002783531604137579\n",
      "epoch:129 from 100000\n",
      "loss: 0.002327413964025027\n",
      "epoch:130 from 100000\n",
      "loss: 0.002114291653157352\n",
      "epoch:131 from 100000\n",
      "loss: 0.002640537630043127\n",
      "epoch:132 from 100000\n",
      "loss: 0.002605834613348179\n",
      "epoch:133 from 100000\n",
      "loss: 0.0025421353921615264\n",
      "epoch:134 from 100000\n",
      "loss: 0.0022566832026882366\n",
      "epoch:135 from 100000\n",
      "loss: 0.002228644307294303\n",
      "epoch:136 from 100000\n",
      "loss: 0.0029615865098548966\n",
      "epoch:137 from 100000\n",
      "loss: 0.0029570975329420435\n",
      "epoch:138 from 100000\n",
      "loss: 0.0024699482596872713\n",
      "epoch:139 from 100000\n",
      "loss: 0.0026696246008555136\n",
      "epoch:140 from 100000\n",
      "loss: 0.0025601444425547015\n",
      "epoch:141 from 100000\n",
      "loss: 0.0026753027684094377\n",
      "epoch:142 from 100000\n",
      "loss: 0.002815345144682624\n",
      "epoch:143 from 100000\n",
      "loss: 0.0022391285239962144\n",
      "epoch:144 from 100000\n",
      "loss: 0.00273924331128228\n",
      "epoch:145 from 100000\n",
      "loss: 0.0029620367674851877\n",
      "epoch:146 from 100000\n",
      "loss: 0.003232547024708765\n",
      "epoch:147 from 100000\n",
      "loss: 0.002580790269149702\n",
      "epoch:148 from 100000\n",
      "loss: 0.002723483788258818\n",
      "epoch:149 from 100000\n",
      "loss: 0.003006927680814897\n",
      "epoch:150 from 100000\n",
      "loss: 0.001984415838961695\n",
      "epoch:151 from 100000\n",
      "loss: 0.0028252031363535934\n",
      "epoch:152 from 100000\n",
      "loss: 0.002380081661044642\n",
      "epoch:153 from 100000\n",
      "loss: 0.00247482438024349\n",
      "epoch:154 from 100000\n",
      "loss: 0.002051693112321118\n",
      "epoch:155 from 100000\n",
      "loss: 0.0023231075778653576\n",
      "epoch:156 from 100000\n",
      "loss: 0.0025226746729458682\n",
      "epoch:157 from 100000\n",
      "loss: 0.0026073010244260786\n",
      "epoch:158 from 100000\n",
      "loss: 0.0019193777027908254\n",
      "epoch:159 from 100000\n",
      "loss: 0.0031169730846158927\n",
      "epoch:160 from 100000\n",
      "loss: 0.002706716863087877\n",
      "epoch:161 from 100000\n",
      "loss: 0.0025614502112603077\n",
      "epoch:162 from 100000\n",
      "loss: 0.002446128829973304\n",
      "epoch:163 from 100000\n",
      "loss: 0.0024515227647725624\n",
      "epoch:164 from 100000\n",
      "loss: 0.0026415459973918587\n",
      "epoch:165 from 100000\n",
      "loss: 0.002414985568066186\n",
      "epoch:166 from 100000\n",
      "loss: 0.0033756333969563457\n",
      "epoch:167 from 100000\n",
      "loss: 0.0022597669727417724\n",
      "epoch:168 from 100000\n",
      "loss: 0.0025297427396253624\n",
      "epoch:169 from 100000\n",
      "loss: 0.002583504341430398\n",
      "epoch:170 from 100000\n",
      "loss: 0.002284320958153785\n",
      "epoch:171 from 100000\n",
      "loss: 0.00239260582964107\n",
      "epoch:172 from 100000\n",
      "loss: 0.002189532873671851\n",
      "epoch:173 from 100000\n",
      "loss: 0.0024584315937374868\n",
      "epoch:174 from 100000\n",
      "loss: 0.002071453190922057\n",
      "epoch:175 from 100000\n",
      "loss: 0.0024831106504734635\n",
      "epoch:176 from 100000\n",
      "loss: 0.0019440720159593639\n",
      "epoch:177 from 100000\n",
      "loss: 0.002700721218786839\n",
      "epoch:178 from 100000\n",
      "loss: 0.0030338467705917083\n",
      "epoch:179 from 100000\n",
      "loss: 0.0023777118042573875\n",
      "epoch:180 from 100000\n",
      "loss: 0.0020431918474059785\n",
      "epoch:181 from 100000\n",
      "loss: 0.002414546852406829\n",
      "epoch:182 from 100000\n",
      "loss: 0.002351349196146657\n",
      "epoch:183 from 100000\n",
      "loss: 0.0024986540536815482\n",
      "epoch:184 from 100000\n",
      "loss: 0.0030030274134560386\n",
      "epoch:185 from 100000\n",
      "loss: 0.002466389216465359\n",
      "epoch:186 from 100000\n",
      "loss: 0.00253684902318696\n",
      "epoch:187 from 100000\n",
      "loss: 0.002174956474021883\n",
      "epoch:188 from 100000\n",
      "loss: 0.0030170734650880604\n",
      "epoch:189 from 100000\n",
      "loss: 0.0021897625584017533\n",
      "epoch:190 from 100000\n",
      "loss: 0.0020895669312608334\n",
      "epoch:191 from 100000\n",
      "loss: 0.0027634196208907724\n",
      "epoch:192 from 100000\n",
      "loss: 0.0023696742731544873\n",
      "epoch:193 from 100000\n",
      "loss: 0.0027164277385054447\n",
      "epoch:194 from 100000\n",
      "loss: 0.0018728007116379786\n",
      "epoch:195 from 100000\n",
      "loss: 0.0026061425950274497\n",
      "epoch:196 from 100000\n",
      "loss: 0.0028262993404268855\n",
      "epoch:197 from 100000\n",
      "loss: 0.0028591315518724514\n",
      "epoch:198 from 100000\n",
      "loss: 0.002647553470154662\n",
      "epoch:199 from 100000\n",
      "loss: 0.0023541707277843216\n",
      "epoch:200 from 100000\n",
      "loss: 0.002823471795096566\n",
      "epoch:201 from 100000\n",
      "loss: 0.0023863813292734903\n",
      "epoch:202 from 100000\n",
      "loss: 0.003135503214991786\n",
      "epoch:203 from 100000\n",
      "loss: 0.0026294869609557736\n",
      "epoch:204 from 100000\n",
      "loss: 0.0026556846458731798\n",
      "epoch:205 from 100000\n",
      "loss: 0.0025577505610954177\n",
      "epoch:206 from 100000\n",
      "loss: 0.0031398742949022562\n",
      "epoch:207 from 100000\n",
      "loss: 0.0023253352517258463\n",
      "epoch:208 from 100000\n",
      "loss: 0.002169237820567105\n",
      "epoch:209 from 100000\n",
      "loss: 0.002378748971793243\n",
      "epoch:210 from 100000\n",
      "loss: 0.002154775827477806\n",
      "epoch:211 from 100000\n",
      "loss: 0.002464704523134742\n",
      "epoch:212 from 100000\n",
      "loss: 0.0024736298075254126\n",
      "epoch:213 from 100000\n",
      "loss: 0.003112922755690306\n",
      "epoch:214 from 100000\n",
      "loss: 0.002170149524573617\n",
      "epoch:215 from 100000\n",
      "loss: 0.002792320513606228\n",
      "epoch:216 from 100000\n",
      "loss: 0.0028291992227309493\n",
      "epoch:217 from 100000\n",
      "loss: 0.002561697407656993\n",
      "epoch:218 from 100000\n",
      "loss: 0.0020926713543758524\n",
      "epoch:219 from 100000\n",
      "loss: 0.0025900609514337702\n",
      "epoch:220 from 100000\n",
      "loss: 0.001952959076049865\n",
      "epoch:221 from 100000\n",
      "loss: 0.0016495617962561937\n",
      "epoch:222 from 100000\n",
      "loss: 0.002624317885903338\n",
      "epoch:223 from 100000\n",
      "loss: 0.002634621370447121\n",
      "epoch:224 from 100000\n",
      "loss: 0.0030611358953649415\n",
      "epoch:225 from 100000\n",
      "loss: 0.002541836568354938\n",
      "epoch:226 from 100000\n",
      "loss: 0.002640265425668531\n",
      "epoch:227 from 100000\n",
      "loss: 0.002543091509323858\n",
      "epoch:228 from 100000\n",
      "loss: 0.002384704375202773\n",
      "epoch:229 from 100000\n",
      "loss: 0.003126060613340087\n",
      "epoch:230 from 100000\n",
      "loss: 0.003113912988510492\n",
      "epoch:231 from 100000\n",
      "loss: 0.0030019469209037197\n",
      "epoch:232 from 100000\n",
      "loss: 0.0025688315560330466\n",
      "epoch:233 from 100000\n",
      "loss: 0.0019907629172593033\n",
      "epoch:234 from 100000\n",
      "loss: 0.0026650873357425553\n",
      "epoch:235 from 100000\n",
      "loss: 0.002242199091369912\n",
      "epoch:236 from 100000\n",
      "loss: 0.002754578446626965\n",
      "epoch:237 from 100000\n",
      "loss: 0.0023850556243303345\n",
      "epoch:238 from 100000\n",
      "loss: 0.0024996230041836043\n",
      "epoch:239 from 100000\n",
      "loss: 0.0018593522272567498\n",
      "epoch:240 from 100000\n",
      "loss: 0.002267126907469219\n",
      "epoch:241 from 100000\n",
      "loss: 0.0025128476786449028\n",
      "epoch:242 from 100000\n",
      "loss: 0.0024161696013038636\n",
      "epoch:243 from 100000\n",
      "loss: 0.0025901243625980896\n",
      "epoch:244 from 100000\n",
      "loss: 0.002926370352639651\n",
      "epoch:245 from 100000\n",
      "loss: 0.00229750903864101\n",
      "epoch:246 from 100000\n",
      "loss: 0.002555869654656817\n",
      "epoch:247 from 100000\n",
      "loss: 0.0025376678755719695\n",
      "epoch:248 from 100000\n",
      "loss: 0.002495938460981506\n",
      "epoch:249 from 100000\n",
      "loss: 0.0027033723441718394\n",
      "epoch:250 from 100000\n",
      "loss: 0.003095335257057741\n",
      "epoch:251 from 100000\n",
      "loss: 0.0027675140682958954\n",
      "epoch:252 from 100000\n",
      "loss: 0.0025442279055312156\n",
      "epoch:253 from 100000\n",
      "loss: 0.0022267298000713254\n",
      "epoch:254 from 100000\n",
      "loss: 0.002350901614875056\n",
      "epoch:255 from 100000\n",
      "loss: 0.0025802024530889867\n",
      "epoch:256 from 100000\n",
      "loss: 0.0028089529732824303\n",
      "epoch:257 from 100000\n",
      "loss: 0.0031847065985175504\n",
      "epoch:258 from 100000\n",
      "loss: 0.0023059688424780234\n",
      "epoch:259 from 100000\n",
      "loss: 0.002380966322476752\n",
      "epoch:260 from 100000\n",
      "loss: 0.0020673508665822737\n",
      "epoch:261 from 100000\n",
      "loss: 0.0025574981145268794\n",
      "epoch:262 from 100000\n",
      "loss: 0.0021410561532206884\n",
      "epoch:263 from 100000\n",
      "loss: 0.0023911267052483254\n",
      "epoch:264 from 100000\n",
      "loss: 0.002050454226719012\n",
      "epoch:265 from 100000\n",
      "loss: 0.0027735879058923274\n",
      "epoch:266 from 100000\n",
      "loss: 0.001974209119111947\n",
      "epoch:267 from 100000\n",
      "loss: 0.0023206285585501973\n",
      "epoch:268 from 100000\n",
      "loss: 0.0020602249985017806\n",
      "epoch:269 from 100000\n",
      "loss: 0.0030567368729919053\n",
      "epoch:270 from 100000\n",
      "loss: 0.002689180307584138\n",
      "epoch:271 from 100000\n",
      "loss: 0.0025780760358316\n",
      "epoch:272 from 100000\n",
      "loss: 0.002358507516390773\n",
      "epoch:273 from 100000\n",
      "loss: 0.0021115163079912236\n",
      "epoch:274 from 100000\n",
      "loss: 0.0029698025683444484\n",
      "epoch:275 from 100000\n",
      "loss: 0.002787449979990697\n",
      "epoch:276 from 100000\n",
      "loss: 0.0025900767023472326\n",
      "epoch:277 from 100000\n",
      "loss: 0.002033751424107777\n",
      "epoch:278 from 100000\n",
      "loss: 0.002459553704568407\n",
      "epoch:279 from 100000\n",
      "loss: 0.0022080412821310347\n",
      "epoch:280 from 100000\n",
      "loss: 0.003270936673430924\n",
      "epoch:281 from 100000\n",
      "loss: 0.0023944602674816906\n",
      "epoch:282 from 100000\n",
      "loss: 0.0022049645842230348\n",
      "epoch:283 from 100000\n",
      "loss: 0.0026962236833014686\n",
      "epoch:284 from 100000\n",
      "loss: 0.002880508465693765\n",
      "epoch:285 from 100000\n",
      "loss: 0.002672300680075068\n",
      "epoch:286 from 100000\n",
      "loss: 0.0022037383969291113\n",
      "epoch:287 from 100000\n",
      "loss: 0.0026749777505443717\n",
      "epoch:288 from 100000\n",
      "loss: 0.002176424407338118\n",
      "epoch:289 from 100000\n",
      "loss: 0.0023117590272647703\n",
      "epoch:290 from 100000\n",
      "loss: 0.0026728956050305896\n",
      "epoch:291 from 100000\n",
      "loss: 0.0023426571825382325\n",
      "epoch:292 from 100000\n",
      "loss: 0.00313195446761938\n",
      "epoch:293 from 100000\n",
      "loss: 0.0025595565467710912\n",
      "epoch:294 from 100000\n",
      "loss: 0.0030479401998491085\n",
      "epoch:295 from 100000\n",
      "loss: 0.0025504639665996365\n",
      "epoch:296 from 100000\n",
      "loss: 0.0028151616643583566\n",
      "epoch:297 from 100000\n",
      "loss: 0.002161643289269932\n",
      "epoch:298 from 100000\n",
      "loss: 0.002377294710015576\n",
      "epoch:299 from 100000\n",
      "loss: 0.0023423243612796796\n",
      "epoch:300 from 100000\n",
      "loss: 0.0025887908273034554\n",
      "epoch:301 from 100000\n",
      "loss: 0.0024104637803361584\n",
      "epoch:302 from 100000\n",
      "loss: 0.0025582930755376765\n",
      "epoch:303 from 100000\n",
      "loss: 0.0030192786151417295\n",
      "epoch:304 from 100000\n",
      "loss: 0.003155835605213042\n",
      "epoch:305 from 100000\n",
      "loss: 0.0026994120189272053\n",
      "epoch:306 from 100000\n",
      "loss: 0.0026989314690979427\n",
      "epoch:307 from 100000\n",
      "loss: 0.002938522205653271\n",
      "retrying for data\n",
      "epoch:308 from 100000\n",
      "loss: 0.0025190687272811374\n",
      "epoch:309 from 100000\n",
      "loss: 0.0026364146407331646\n",
      "epoch:310 from 100000\n",
      "loss: 0.0023390719068174803\n",
      "epoch:311 from 100000\n",
      "loss: 0.0026579473018273347\n",
      "epoch:312 from 100000\n",
      "loss: 0.002200554489490969\n",
      "epoch:313 from 100000\n",
      "loss: 0.002531512911758682\n",
      "epoch:314 from 100000\n",
      "loss: 0.001944128851917526\n",
      "epoch:315 from 100000\n",
      "loss: 0.0022123723621234603\n",
      "epoch:316 from 100000\n",
      "loss: 0.0024462558846494176\n",
      "epoch:317 from 100000\n",
      "loss: 0.002277999601801639\n",
      "epoch:318 from 100000\n",
      "loss: 0.0019361161241704394\n",
      "epoch:319 from 100000\n",
      "loss: 0.0024299848342081987\n",
      "epoch:320 from 100000\n",
      "loss: 0.0022256376748401863\n",
      "epoch:321 from 100000\n",
      "loss: 0.00234674166480886\n",
      "epoch:322 from 100000\n",
      "loss: 0.0023946483461827484\n",
      "epoch:323 from 100000\n",
      "loss: 0.0023565702782377684\n",
      "epoch:324 from 100000\n",
      "loss: 0.002521753414697514\n",
      "epoch:325 from 100000\n",
      "loss: 0.0026962616992136645\n",
      "epoch:326 from 100000\n",
      "loss: 0.002203647380468965\n",
      "epoch:327 from 100000\n",
      "loss: 0.0022765830559023925\n",
      "epoch:328 from 100000\n",
      "loss: 0.002488203814834833\n",
      "epoch:329 from 100000\n",
      "loss: 0.003254052152811937\n",
      "epoch:330 from 100000\n",
      "loss: 0.002423110132582451\n",
      "epoch:331 from 100000\n",
      "loss: 0.0023378603851540447\n",
      "epoch:332 from 100000\n",
      "loss: 0.002607274709532703\n",
      "epoch:333 from 100000\n",
      "loss: 0.002461589243665685\n",
      "epoch:334 from 100000\n",
      "loss: 0.0024392433533790836\n",
      "epoch:335 from 100000\n",
      "loss: 0.0031366838857138646\n",
      "epoch:336 from 100000\n",
      "loss: 0.0030110910807081837\n",
      "epoch:337 from 100000\n",
      "loss: 0.002403580951437334\n",
      "epoch:338 from 100000\n",
      "loss: 0.002435647249910744\n",
      "epoch:339 from 100000\n",
      "loss: 0.0027517583457097317\n",
      "epoch:340 from 100000\n",
      "loss: 0.002506808455848386\n",
      "epoch:341 from 100000\n",
      "loss: 0.0021685113290175195\n",
      "epoch:342 from 100000\n",
      "loss: 0.0026658601896656364\n",
      "epoch:343 from 100000\n",
      "loss: 0.0032058952703835075\n",
      "epoch:344 from 100000\n",
      "loss: 0.002476926430389881\n",
      "epoch:345 from 100000\n",
      "loss: 0.0023549974887373537\n",
      "epoch:346 from 100000\n",
      "loss: 0.0025942145377939596\n",
      "epoch:347 from 100000\n",
      "loss: 0.00331435314222972\n",
      "epoch:348 from 100000\n",
      "loss: 0.003019760774122915\n",
      "epoch:349 from 100000\n",
      "loss: 0.002131648600141034\n",
      "epoch:350 from 100000\n",
      "loss: 0.002639502937825\n",
      "epoch:351 from 100000\n",
      "loss: 0.00241416581974363\n",
      "epoch:352 from 100000\n",
      "loss: 0.002734000095756528\n",
      "epoch:353 from 100000\n",
      "loss: 0.0025125837796622363\n",
      "epoch:354 from 100000\n",
      "loss: 0.0022931165939041875\n",
      "epoch:355 from 100000\n",
      "loss: 0.002012309315972516\n",
      "epoch:356 from 100000\n",
      "loss: 0.002455277990321747\n",
      "epoch:357 from 100000\n",
      "loss: 0.002207756483400658\n",
      "epoch:358 from 100000\n",
      "loss: 0.0022989259310293164\n",
      "epoch:359 from 100000\n",
      "loss: 0.0025563995727679867\n",
      "epoch:360 from 100000\n",
      "loss: 0.002855611376901379\n",
      "epoch:361 from 100000\n",
      "loss: 0.002981793926579712\n",
      "epoch:362 from 100000\n",
      "loss: 0.0026560980378178556\n",
      "epoch:363 from 100000\n",
      "loss: 0.002661303982165464\n",
      "epoch:364 from 100000\n",
      "loss: 0.0023676610627489936\n",
      "epoch:365 from 100000\n",
      "loss: 0.003064874299312237\n",
      "epoch:366 from 100000\n",
      "loss: 0.002482001203617301\n",
      "epoch:367 from 100000\n",
      "loss: 0.0022079990048951004\n",
      "epoch:368 from 100000\n",
      "loss: 0.0028187641459282986\n",
      "epoch:369 from 100000\n",
      "loss: 0.0025974951836076343\n",
      "epoch:370 from 100000\n",
      "loss: 0.0032883787434911937\n",
      "epoch:371 from 100000\n",
      "loss: 0.0025648114442731185\n",
      "epoch:372 from 100000\n",
      "loss: 0.0025138121176269124\n",
      "epoch:373 from 100000\n",
      "loss: 0.002278964557774543\n",
      "epoch:374 from 100000\n",
      "loss: 0.002329737862339698\n",
      "epoch:375 from 100000\n",
      "loss: 0.0023866191326362696\n",
      "epoch:376 from 100000\n",
      "loss: 0.0023203036098493612\n",
      "epoch:377 from 100000\n",
      "loss: 0.00254014681542003\n",
      "epoch:378 from 100000\n",
      "loss: 0.0024220076212202457\n",
      "epoch:379 from 100000\n",
      "loss: 0.002294418020625244\n",
      "epoch:380 from 100000\n",
      "loss: 0.0027066127165085163\n",
      "epoch:381 from 100000\n",
      "loss: 0.0028471745736169396\n",
      "epoch:382 from 100000\n",
      "loss: 0.002298209378437832\n",
      "epoch:383 from 100000\n",
      "loss: 0.0028084281904057207\n",
      "epoch:384 from 100000\n",
      "loss: 0.002631947408247015\n",
      "epoch:385 from 100000\n",
      "loss: 0.0018932097526658254\n",
      "epoch:386 from 100000\n",
      "loss: 0.0026705328611456025\n",
      "epoch:387 from 100000\n",
      "loss: 0.0020024772941269475\n",
      "epoch:388 from 100000\n",
      "loss: 0.002736124865180045\n",
      "epoch:389 from 100000\n",
      "loss: 0.002644126292693727\n",
      "epoch:390 from 100000\n",
      "loss: 0.00297387716926778\n",
      "epoch:391 from 100000\n",
      "loss: 0.0024740238199001396\n",
      "epoch:392 from 100000\n",
      "loss: 0.0029075230884245684\n",
      "epoch:393 from 100000\n",
      "loss: 0.002093047826349448\n",
      "epoch:394 from 100000\n",
      "loss: 0.0033459745889103942\n",
      "epoch:395 from 100000\n",
      "loss: 0.002451029490714518\n",
      "epoch:396 from 100000\n",
      "loss: 0.002113006414759866\n",
      "epoch:397 from 100000\n",
      "loss: 0.002326204973030599\n",
      "epoch:398 from 100000\n",
      "loss: 0.0026196286251263246\n",
      "epoch:399 from 100000\n",
      "loss: 0.002164556361094583\n",
      "epoch:400 from 100000\n",
      "loss: 0.002285247468819307\n",
      "epoch:401 from 100000\n",
      "loss: 0.0029017621392881665\n",
      "epoch:402 from 100000\n",
      "loss: 0.002508447757236354\n",
      "epoch:403 from 100000\n",
      "loss: 0.002384805485775132\n",
      "epoch:404 from 100000\n",
      "loss: 0.0022148414934406446\n",
      "epoch:405 from 100000\n",
      "loss: 0.0023942644447032535\n",
      "epoch:406 from 100000\n",
      "loss: 0.0026875918999280657\n",
      "epoch:407 from 100000\n",
      "loss: 0.002989686760315635\n",
      "epoch:408 from 100000\n",
      "loss: 0.002991766846662358\n",
      "epoch:409 from 100000\n",
      "loss: 0.002278856777593319\n",
      "epoch:410 from 100000\n",
      "loss: 0.0024805800811975587\n",
      "epoch:411 from 100000\n",
      "loss: 0.0025300683591353845\n",
      "epoch:412 from 100000\n",
      "loss: 0.0022568076992683928\n",
      "epoch:413 from 100000\n",
      "loss: 0.002428801594618335\n",
      "epoch:414 from 100000\n",
      "loss: 0.0021237064925685445\n",
      "epoch:415 from 100000\n",
      "loss: 0.0022404965895646\n",
      "epoch:416 from 100000\n",
      "loss: 0.002078869664330796\n",
      "epoch:417 from 100000\n",
      "loss: 0.0024681872626786117\n",
      "epoch:418 from 100000\n",
      "loss: 0.0033215190291002727\n",
      "epoch:419 from 100000\n",
      "loss: 0.002105320119795806\n",
      "epoch:420 from 100000\n",
      "loss: 0.0027331367837746257\n",
      "epoch:421 from 100000\n",
      "loss: 0.002267081366341017\n",
      "epoch:422 from 100000\n",
      "loss: 0.0027485994060896246\n",
      "epoch:423 from 100000\n",
      "loss: 0.0024042137785613704\n",
      "epoch:424 from 100000\n",
      "loss: 0.002275345088321501\n",
      "epoch:425 from 100000\n",
      "loss: 0.002633304123094149\n",
      "epoch:426 from 100000\n",
      "loss: 0.0020848280850032097\n",
      "epoch:427 from 100000\n",
      "loss: 0.002150801209779729\n",
      "epoch:428 from 100000\n",
      "loss: 0.0027214331955462967\n",
      "epoch:429 from 100000\n",
      "loss: 0.002289625981433119\n",
      "epoch:430 from 100000\n",
      "loss: 0.002512503888453921\n",
      "epoch:431 from 100000\n",
      "loss: 0.0018264726861048075\n",
      "epoch:432 from 100000\n",
      "loss: 0.002579181499754668\n",
      "epoch:433 from 100000\n",
      "loss: 0.0022360332989705967\n",
      "epoch:434 from 100000\n",
      "loss: 0.0019076777786608545\n",
      "epoch:435 from 100000\n",
      "loss: 0.0021448146023885784\n",
      "epoch:436 from 100000\n",
      "loss: 0.002813542369779043\n",
      "epoch:437 from 100000\n",
      "loss: 0.002519395361275656\n",
      "epoch:438 from 100000\n",
      "loss: 0.0028726880590284054\n",
      "epoch:439 from 100000\n",
      "loss: 0.0029104595932381017\n",
      "epoch:440 from 100000\n",
      "loss: 0.0030244119723192853\n",
      "epoch:441 from 100000\n",
      "loss: 0.0026708609756269652\n",
      "epoch:442 from 100000\n",
      "loss: 0.0023220727217108106\n",
      "epoch:443 from 100000\n",
      "loss: 0.0027750017360403945\n",
      "epoch:444 from 100000\n",
      "loss: 0.0025021255249271235\n",
      "epoch:445 from 100000\n",
      "loss: 0.002497742084813126\n",
      "epoch:446 from 100000\n",
      "loss: 0.0024200693450211475\n",
      "epoch:447 from 100000\n",
      "loss: 0.0028365248018644706\n",
      "epoch:448 from 100000\n",
      "loss: 0.0024431549240375716\n",
      "epoch:449 from 100000\n",
      "loss: 0.0029974458860806408\n",
      "epoch:450 from 100000\n",
      "loss: 0.00194627919222512\n",
      "epoch:451 from 100000\n",
      "loss: 0.0022787151376917336\n",
      "epoch:452 from 100000\n",
      "loss: 0.002696902790717104\n",
      "epoch:453 from 100000\n",
      "loss: 0.0022131939282701296\n",
      "epoch:454 from 100000\n",
      "loss: 0.002229551022026044\n",
      "epoch:455 from 100000\n",
      "loss: 0.002496540781123713\n",
      "epoch:456 from 100000\n",
      "loss: 0.002206714929542386\n",
      "epoch:457 from 100000\n",
      "loss: 0.0024055352365053295\n",
      "epoch:458 from 100000\n",
      "loss: 0.002278848987685933\n",
      "epoch:459 from 100000\n",
      "loss: 0.002694639948316535\n",
      "epoch:460 from 100000\n",
      "loss: 0.0023802826519272458\n",
      "epoch:461 from 100000\n",
      "loss: 0.002866039315534863\n",
      "epoch:462 from 100000\n",
      "loss: 0.0025671269876568203\n",
      "epoch:463 from 100000\n",
      "loss: 0.0027320475516603437\n",
      "epoch:464 from 100000\n",
      "loss: 0.002361008569366163\n",
      "epoch:465 from 100000\n",
      "loss: 0.002975976171029515\n",
      "epoch:466 from 100000\n",
      "loss: 0.0024412243950564516\n",
      "epoch:467 from 100000\n",
      "loss: 0.0026082399474773865\n",
      "epoch:468 from 100000\n",
      "loss: 0.002799441305825212\n",
      "epoch:469 from 100000\n",
      "loss: 0.0022694138260703767\n",
      "epoch:470 from 100000\n",
      "loss: 0.002442812269521255\n",
      "epoch:471 from 100000\n",
      "loss: 0.0022926970979142425\n",
      "epoch:472 from 100000\n",
      "loss: 0.0022206149027113042\n",
      "epoch:473 from 100000\n",
      "loss: 0.002723320275208607\n",
      "epoch:474 from 100000\n",
      "loss: 0.0023149826593282796\n",
      "epoch:475 from 100000\n",
      "loss: 0.002913293547919693\n",
      "epoch:476 from 100000\n",
      "loss: 0.002263785165155241\n",
      "epoch:477 from 100000\n",
      "loss: 0.0022045858294745813\n",
      "epoch:478 from 100000\n",
      "loss: 0.002487466031226404\n",
      "epoch:479 from 100000\n",
      "loss: 0.002814013440058716\n",
      "epoch:480 from 100000\n",
      "loss: 0.002709027282094212\n",
      "epoch:481 from 100000\n",
      "loss: 0.0021677156549344545\n",
      "epoch:482 from 100000\n",
      "loss: 0.002740825247911971\n",
      "epoch:483 from 100000\n",
      "loss: 0.002545517526243657\n",
      "epoch:484 from 100000\n",
      "loss: 0.001809447364934158\n",
      "epoch:485 from 100000\n",
      "loss: 0.002622257830836361\n",
      "epoch:486 from 100000\n",
      "loss: 0.002026298684597805\n",
      "epoch:487 from 100000\n",
      "loss: 0.0020239355791886737\n",
      "epoch:488 from 100000\n",
      "loss: 0.0026673997386978954\n",
      "epoch:489 from 100000\n",
      "loss: 0.0022312598351845736\n",
      "epoch:490 from 100000\n",
      "loss: 0.0027407709312683437\n",
      "epoch:491 from 100000\n",
      "loss: 0.002709898808006983\n",
      "epoch:492 from 100000\n",
      "loss: 0.003166357397049069\n",
      "epoch:493 from 100000\n",
      "loss: 0.0024341070945865795\n",
      "epoch:494 from 100000\n",
      "loss: 0.002813386183589728\n",
      "epoch:495 from 100000\n",
      "loss: 0.0026918425034239135\n",
      "epoch:496 from 100000\n",
      "loss: 0.002572805392844657\n",
      "epoch:497 from 100000\n",
      "loss: 0.0018482997221553887\n",
      "epoch:498 from 100000\n",
      "loss: 0.002379436073653096\n",
      "epoch:499 from 100000\n",
      "loss: 0.0028432734728767173\n",
      "epoch:500 from 100000\n",
      "loss: 0.0029406331839254563\n",
      "epoch:501 from 100000\n",
      "loss: 0.002826288768289942\n",
      "epoch:502 from 100000\n",
      "loss: 0.0026308688285325843\n",
      "epoch:503 from 100000\n",
      "loss: 0.002669083283450391\n",
      "epoch:504 from 100000\n",
      "loss: 0.003034796509325588\n",
      "epoch:505 from 100000\n",
      "loss: 0.0028728395971029386\n",
      "epoch:506 from 100000\n",
      "loss: 0.0025797314108899627\n",
      "epoch:507 from 100000\n",
      "loss: 0.00255785370463002\n",
      "epoch:508 from 100000\n",
      "loss: 0.002515551732074073\n",
      "epoch:509 from 100000\n",
      "loss: 0.002186846464795167\n",
      "epoch:510 from 100000\n",
      "loss: 0.002310141243299313\n",
      "epoch:511 from 100000\n",
      "loss: 0.002476522001700232\n",
      "retrying for data\n",
      "epoch:512 from 100000\n",
      "loss: 0.0025243908201275644\n",
      "epoch:513 from 100000\n",
      "loss: 0.002972734076820416\n",
      "epoch:514 from 100000\n",
      "loss: 0.0024449587519939087\n",
      "epoch:515 from 100000\n",
      "loss: 0.0027910593657907157\n",
      "epoch:516 from 100000\n",
      "loss: 0.002302083126494381\n",
      "epoch:517 from 100000\n",
      "loss: 0.002479821314267383\n",
      "epoch:518 from 100000\n",
      "loss: 0.0025753472891096862\n",
      "epoch:519 from 100000\n",
      "loss: 0.0024981998012094664\n",
      "epoch:520 from 100000\n",
      "loss: 0.00245036349335237\n",
      "epoch:521 from 100000\n",
      "loss: 0.002310593002277983\n",
      "epoch:522 from 100000\n",
      "loss: 0.0020930315573082225\n",
      "epoch:523 from 100000\n",
      "loss: 0.0023640803813407274\n",
      "epoch:524 from 100000\n",
      "loss: 0.002843379198651519\n",
      "epoch:525 from 100000\n",
      "loss: 0.0029844219283177154\n",
      "epoch:526 from 100000\n",
      "loss: 0.002469787517526356\n",
      "epoch:527 from 100000\n",
      "loss: 0.0018562094395235818\n",
      "epoch:528 from 100000\n",
      "loss: 0.00253360373719147\n",
      "epoch:529 from 100000\n",
      "loss: 0.002576510156217182\n",
      "epoch:530 from 100000\n",
      "loss: 0.0021251720240513805\n",
      "epoch:531 from 100000\n",
      "loss: 0.00286316151363053\n",
      "epoch:532 from 100000\n",
      "loss: 0.002226493024551246\n",
      "epoch:533 from 100000\n",
      "loss: 0.0029769974474902483\n",
      "epoch:534 from 100000\n",
      "loss: 0.0019462496898938753\n",
      "epoch:535 from 100000\n",
      "loss: 0.002415736264993029\n",
      "epoch:536 from 100000\n",
      "loss: 0.003037042314588234\n",
      "epoch:537 from 100000\n",
      "loss: 0.0025169950137637898\n",
      "epoch:538 from 100000\n",
      "loss: 0.0028791760288697787\n",
      "epoch:539 from 100000\n",
      "loss: 0.0024429379153616537\n",
      "epoch:540 from 100000\n",
      "loss: 0.002767835224204873\n",
      "epoch:541 from 100000\n",
      "loss: 0.0022420166446295298\n",
      "epoch:542 from 100000\n",
      "loss: 0.0023430332972225187\n",
      "epoch:543 from 100000\n",
      "loss: 0.002558780934606375\n",
      "epoch:544 from 100000\n",
      "loss: 0.0023907658173243362\n",
      "epoch:545 from 100000\n",
      "loss: 0.002379910044908229\n",
      "epoch:546 from 100000\n",
      "loss: 0.0019420462731147836\n",
      "epoch:547 from 100000\n",
      "loss: 0.002244849984975872\n",
      "epoch:548 from 100000\n",
      "loss: 0.002406060450752534\n",
      "epoch:549 from 100000\n",
      "loss: 0.002555881797945858\n",
      "epoch:550 from 100000\n",
      "loss: 0.002576612706036485\n",
      "epoch:551 from 100000\n",
      "loss: 0.0025946857842313875\n",
      "epoch:552 from 100000\n",
      "loss: 0.002245754409472056\n",
      "epoch:553 from 100000\n",
      "loss: 0.002478310187768784\n",
      "epoch:554 from 100000\n",
      "loss: 0.002509544868246394\n",
      "epoch:555 from 100000\n",
      "loss: 0.0025979763230168373\n",
      "epoch:556 from 100000\n",
      "loss: 0.002010947063808999\n",
      "epoch:557 from 100000\n",
      "loss: 0.0023302434284460105\n",
      "epoch:558 from 100000\n",
      "loss: 0.0021776160623403484\n",
      "epoch:559 from 100000\n",
      "loss: 0.002931316439259035\n",
      "epoch:560 from 100000\n",
      "loss: 0.002804975782794372\n",
      "epoch:561 from 100000\n",
      "loss: 0.0022476565375768587\n",
      "epoch:562 from 100000\n",
      "loss: 0.003088897846907912\n",
      "epoch:563 from 100000\n",
      "loss: 0.002446963270784863\n",
      "epoch:564 from 100000\n",
      "loss: 0.0023602184409128313\n",
      "epoch:565 from 100000\n",
      "loss: 0.002111599538778819\n",
      "epoch:566 from 100000\n",
      "loss: 0.0024647862095150685\n",
      "epoch:567 from 100000\n",
      "loss: 0.002310399189951795\n",
      "epoch:568 from 100000\n",
      "loss: 0.002259138034986563\n",
      "epoch:569 from 100000\n",
      "loss: 0.002542793217344297\n",
      "epoch:570 from 100000\n",
      "loss: 0.0024480092145893195\n",
      "epoch:571 from 100000\n",
      "loss: 0.002503145869070522\n",
      "epoch:572 from 100000\n",
      "loss: 0.0021531936527310336\n",
      "epoch:573 from 100000\n",
      "loss: 0.002803041554301444\n",
      "epoch:574 from 100000\n",
      "loss: 0.00254202727160191\n",
      "epoch:575 from 100000\n",
      "loss: 0.0026871088489599515\n",
      "epoch:576 from 100000\n",
      "loss: 0.0022380425396022474\n",
      "epoch:577 from 100000\n",
      "loss: 0.002417605975750803\n",
      "epoch:578 from 100000\n",
      "loss: 0.0027666611512131567\n",
      "epoch:579 from 100000\n",
      "loss: 0.002340447113283517\n",
      "epoch:580 from 100000\n",
      "loss: 0.0025429720488148178\n",
      "epoch:581 from 100000\n",
      "loss: 0.0021886867765203988\n",
      "epoch:582 from 100000\n",
      "loss: 0.0023192563729139692\n",
      "epoch:583 from 100000\n",
      "loss: 0.0023405906797790976\n",
      "epoch:584 from 100000\n",
      "loss: 0.0017308597662122338\n",
      "epoch:585 from 100000\n",
      "loss: 0.002160898758120311\n",
      "epoch:586 from 100000\n",
      "loss: 0.002683787229443624\n",
      "epoch:587 from 100000\n",
      "loss: 0.0023984989380494426\n",
      "epoch:588 from 100000\n",
      "loss: 0.003367114548268546\n",
      "epoch:589 from 100000\n",
      "loss: 0.0021104285397655076\n",
      "epoch:590 from 100000\n",
      "loss: 0.0023799904058989796\n",
      "epoch:591 from 100000\n",
      "loss: 0.002876257502464341\n",
      "epoch:592 from 100000\n",
      "loss: 0.0021146362310844324\n",
      "epoch:593 from 100000\n",
      "loss: 0.0021162508206771236\n",
      "epoch:594 from 100000\n",
      "loss: 0.002852642137810335\n",
      "epoch:595 from 100000\n",
      "loss: 0.002347456708264417\n",
      "epoch:596 from 100000\n",
      "loss: 0.0028451563513840483\n",
      "epoch:597 from 100000\n",
      "loss: 0.0022556727527671683\n",
      "epoch:598 from 100000\n",
      "loss: 0.0022741318536532162\n",
      "epoch:599 from 100000\n",
      "loss: 0.0024062665144128914\n",
      "epoch:600 from 100000\n",
      "loss: 0.0024372883772230125\n",
      "epoch:601 from 100000\n",
      "loss: 0.0025225791292200483\n",
      "epoch:602 from 100000\n",
      "loss: 0.0023057962572750057\n",
      "epoch:603 from 100000\n",
      "loss: 0.002168735469524563\n",
      "epoch:604 from 100000\n",
      "loss: 0.00292304199271598\n",
      "epoch:605 from 100000\n",
      "loss: 0.002355623851542532\n",
      "epoch:606 from 100000\n",
      "loss: 0.0027223567051919417\n",
      "epoch:607 from 100000\n",
      "loss: 0.0022210501473125532\n",
      "epoch:608 from 100000\n",
      "loss: 0.0029337554509538677\n",
      "epoch:609 from 100000\n",
      "loss: 0.003200609931326426\n",
      "epoch:610 from 100000\n",
      "loss: 0.0026322460208518805\n",
      "epoch:611 from 100000\n",
      "loss: 0.0021795634493173566\n",
      "epoch:612 from 100000\n",
      "loss: 0.0023894131408042085\n",
      "epoch:613 from 100000\n",
      "loss: 0.002504691958222338\n",
      "epoch:614 from 100000\n",
      "loss: 0.002742708319431131\n",
      "epoch:615 from 100000\n",
      "loss: 0.0025310520312018525\n",
      "epoch:616 from 100000\n",
      "loss: 0.002136379170224245\n",
      "epoch:617 from 100000\n",
      "loss: 0.0026878008094115557\n",
      "epoch:618 from 100000\n",
      "loss: 0.0024162395441464923\n",
      "epoch:619 from 100000\n",
      "loss: 0.0029090221341334654\n",
      "epoch:620 from 100000\n",
      "loss: 0.0022461752984099803\n",
      "epoch:621 from 100000\n",
      "loss: 0.002646984347165926\n",
      "epoch:622 from 100000\n",
      "loss: 0.002616586856390768\n",
      "epoch:623 from 100000\n",
      "loss: 0.0033857247288437975\n",
      "epoch:624 from 100000\n",
      "loss: 0.0033327775794305126\n",
      "epoch:625 from 100000\n",
      "loss: 0.002589715461454034\n",
      "epoch:626 from 100000\n",
      "loss: 0.0031289841121520112\n",
      "epoch:627 from 100000\n",
      "loss: 0.002001007312543379\n",
      "epoch:628 from 100000\n",
      "loss: 0.0022442536188123086\n",
      "epoch:629 from 100000\n",
      "loss: 0.002335547746923794\n",
      "epoch:630 from 100000\n",
      "loss: 0.0024500574499199956\n",
      "epoch:631 from 100000\n",
      "loss: 0.0024502292936574577\n",
      "epoch:632 from 100000\n",
      "loss: 0.0025953891198469137\n",
      "epoch:633 from 100000\n",
      "loss: 0.00205091401639379\n",
      "epoch:634 from 100000\n",
      "loss: 0.001978892804629595\n",
      "epoch:635 from 100000\n",
      "loss: 0.0019160790585885934\n",
      "epoch:636 from 100000\n",
      "loss: 0.002357144420884083\n",
      "epoch:637 from 100000\n",
      "loss: 0.0019343625804140174\n",
      "epoch:638 from 100000\n",
      "loss: 0.0023311257919544914\n",
      "epoch:639 from 100000\n",
      "loss: 0.0021569839868504914\n",
      "epoch:640 from 100000\n",
      "loss: 0.0023601645091275714\n",
      "epoch:641 from 100000\n",
      "loss: 0.0022120328411006085\n",
      "epoch:642 from 100000\n",
      "loss: 0.0021469970521081905\n",
      "epoch:643 from 100000\n",
      "loss: 0.002297730459304148\n",
      "epoch:644 from 100000\n",
      "loss: 0.0020478470179341457\n",
      "epoch:645 from 100000\n",
      "loss: 0.0022113688548870414\n",
      "epoch:646 from 100000\n",
      "loss: 0.003306943135896745\n",
      "epoch:647 from 100000\n",
      "loss: 0.0019509167645423986\n",
      "epoch:648 from 100000\n",
      "loss: 0.0023439843534163174\n",
      "epoch:649 from 100000\n",
      "loss: 0.0019608250316025533\n",
      "epoch:650 from 100000\n",
      "loss: 0.002527192512957299\n",
      "epoch:651 from 100000\n",
      "loss: 0.0023542826913285353\n",
      "epoch:652 from 100000\n",
      "loss: 0.0028465547671316926\n",
      "epoch:653 from 100000\n",
      "loss: 0.002224321578097488\n",
      "epoch:654 from 100000\n",
      "loss: 0.0022955936620405737\n",
      "epoch:655 from 100000\n",
      "loss: 0.0024384574127509495\n",
      "epoch:656 from 100000\n",
      "loss: 0.0028686434353346613\n",
      "epoch:657 from 100000\n",
      "loss: 0.002992013320181286\n",
      "epoch:658 from 100000\n",
      "loss: 0.0024074529331130634\n",
      "epoch:659 from 100000\n",
      "loss: 0.0020918388307791247\n",
      "epoch:660 from 100000\n",
      "loss: 0.002530257553033266\n",
      "epoch:661 from 100000\n",
      "loss: 0.0022766448592506094\n",
      "epoch:662 from 100000\n",
      "loss: 0.0028488527607635206\n",
      "epoch:663 from 100000\n",
      "loss: 0.0025235582723155403\n",
      "epoch:664 from 100000\n",
      "loss: 0.002669872716012378\n",
      "epoch:665 from 100000\n",
      "loss: 0.0021420165833774263\n",
      "epoch:666 from 100000\n",
      "loss: 0.0026343356534539453\n",
      "epoch:667 from 100000\n",
      "loss: 0.00284793359418245\n",
      "epoch:668 from 100000\n",
      "loss: 0.0029006873190837723\n",
      "epoch:669 from 100000\n",
      "loss: 0.0023797070961109057\n",
      "epoch:670 from 100000\n",
      "loss: 0.0025689865232152442\n",
      "epoch:671 from 100000\n",
      "loss: 0.0024635872027829464\n",
      "epoch:672 from 100000\n",
      "loss: 0.0026307389742612486\n",
      "epoch:673 from 100000\n",
      "loss: 0.0028460077717795684\n",
      "epoch:674 from 100000\n",
      "loss: 0.002736140974718637\n",
      "epoch:675 from 100000\n",
      "loss: 0.002317057559054092\n",
      "epoch:676 from 100000\n",
      "loss: 0.002402922425659426\n",
      "epoch:677 from 100000\n",
      "loss: 0.0023192144332000453\n",
      "epoch:678 from 100000\n",
      "loss: 0.0020735160510980677\n",
      "epoch:679 from 100000\n",
      "loss: 0.002770274869590139\n",
      "epoch:680 from 100000\n",
      "loss: 0.002780628541131591\n",
      "epoch:681 from 100000\n",
      "loss: 0.002384256485612468\n",
      "epoch:682 from 100000\n",
      "loss: 0.002952571190434128\n",
      "epoch:683 from 100000\n",
      "loss: 0.0021916051248069834\n",
      "epoch:684 from 100000\n",
      "loss: 0.0027476229021488052\n",
      "epoch:685 from 100000\n",
      "loss: 0.002678556053780312\n",
      "epoch:686 from 100000\n",
      "loss: 0.002617662705603152\n",
      "epoch:687 from 100000\n",
      "loss: 0.0021881542118364905\n",
      "epoch:688 from 100000\n",
      "loss: 0.0026958787474313795\n",
      "epoch:689 from 100000\n",
      "loss: 0.002096010421354322\n",
      "epoch:690 from 100000\n",
      "loss: 0.002141460135817397\n",
      "epoch:691 from 100000\n",
      "loss: 0.002601593876931929\n",
      "epoch:692 from 100000\n",
      "loss: 0.002347456370898726\n",
      "epoch:693 from 100000\n",
      "loss: 0.003104182194107352\n",
      "epoch:694 from 100000\n",
      "loss: 0.002444315585790946\n",
      "epoch:695 from 100000\n",
      "loss: 0.0023542713290396478\n",
      "epoch:696 from 100000\n",
      "loss: 0.002469607668643903\n",
      "epoch:697 from 100000\n",
      "loss: 0.002473767115816372\n",
      "epoch:698 from 100000\n",
      "loss: 0.0026609058913891204\n",
      "epoch:699 from 100000\n",
      "loss: 0.002283958186893642\n",
      "epoch:700 from 100000\n",
      "loss: 0.00249364200499258\n",
      "epoch:701 from 100000\n",
      "loss: 0.002446173490568526\n",
      "epoch:702 from 100000\n",
      "loss: 0.0028069183275931664\n",
      "epoch:703 from 100000\n",
      "loss: 0.0027560563096358237\n",
      "epoch:704 from 100000\n",
      "loss: 0.002619028424192038\n",
      "epoch:705 from 100000\n",
      "loss: 0.0021665396199352926\n",
      "epoch:706 from 100000\n",
      "loss: 0.002654281664234759\n",
      "epoch:707 from 100000\n",
      "loss: 0.0027723719078665\n",
      "epoch:708 from 100000\n",
      "loss: 0.002725261444112448\n",
      "epoch:709 from 100000\n",
      "loss: 0.002057691338706036\n",
      "epoch:710 from 100000\n",
      "loss: 0.0026542245000200637\n",
      "epoch:711 from 100000\n",
      "loss: 0.0024290378838998095\n",
      "epoch:712 from 100000\n",
      "loss: 0.002519797897150511\n",
      "epoch:713 from 100000\n",
      "loss: 0.002383087157852515\n",
      "epoch:714 from 100000\n",
      "loss: 0.002550914666102244\n",
      "epoch:715 from 100000\n",
      "loss: 0.002559778303236726\n",
      "epoch:716 from 100000\n",
      "loss: 0.002255320320642795\n",
      "epoch:717 from 100000\n",
      "loss: 0.002481731395675979\n",
      "epoch:718 from 100000\n",
      "loss: 0.002303231417158713\n",
      "epoch:719 from 100000\n",
      "loss: 0.0022223115952897388\n",
      "epoch:720 from 100000\n",
      "loss: 0.0024468647146562716\n",
      "epoch:721 from 100000\n",
      "loss: 0.002729557400101612\n",
      "epoch:722 from 100000\n",
      "loss: 0.00246015248887943\n",
      "epoch:723 from 100000\n",
      "loss: 0.0027038803356163044\n",
      "epoch:724 from 100000\n",
      "loss: 0.0028845251052302956\n",
      "epoch:725 from 100000\n",
      "loss: 0.0025196057650873627\n",
      "epoch:726 from 100000\n",
      "loss: 0.0024337682485793266\n",
      "epoch:727 from 100000\n",
      "loss: 0.002969479242779016\n",
      "epoch:728 from 100000\n",
      "loss: 0.002141949450987113\n",
      "epoch:729 from 100000\n",
      "loss: 0.0020234908498935056\n",
      "epoch:730 from 100000\n",
      "loss: 0.002484619099050178\n",
      "epoch:731 from 100000\n",
      "loss: 0.0029552797240626205\n",
      "epoch:732 from 100000\n",
      "loss: 0.0029206884037762393\n",
      "epoch:733 from 100000\n",
      "loss: 0.002659455782350051\n",
      "epoch:734 from 100000\n",
      "loss: 0.002502740011379956\n",
      "epoch:735 from 100000\n",
      "loss: 0.0030864667646710586\n",
      "epoch:736 from 100000\n",
      "loss: 0.002244572737964745\n",
      "epoch:737 from 100000\n",
      "loss: 0.002327780427037851\n",
      "epoch:738 from 100000\n",
      "loss: 0.002245965783686188\n",
      "retrying for data\n",
      "epoch:739 from 100000\n",
      "loss: 0.001754148038855874\n",
      "epoch:740 from 100000\n",
      "loss: 0.0020705865104133636\n",
      "epoch:741 from 100000\n",
      "loss: 0.0024367770395485877\n",
      "epoch:742 from 100000\n",
      "loss: 0.0021649533898653317\n",
      "epoch:743 from 100000\n",
      "loss: 0.0023646785722348795\n",
      "epoch:744 from 100000\n",
      "loss: 0.0024612850203595826\n",
      "epoch:745 from 100000\n",
      "loss: 0.0023431761195524814\n",
      "epoch:746 from 100000\n",
      "loss: 0.002537136091348202\n",
      "epoch:747 from 100000\n",
      "loss: 0.001957618131569916\n",
      "epoch:748 from 100000\n",
      "loss: 0.0025556276206089024\n",
      "epoch:749 from 100000\n",
      "loss: 0.002143328317799842\n",
      "epoch:750 from 100000\n",
      "loss: 0.002741927537357469\n",
      "epoch:751 from 100000\n",
      "loss: 0.0032367065841754084\n",
      "epoch:752 from 100000\n",
      "loss: 0.002560656535109729\n",
      "epoch:753 from 100000\n",
      "loss: 0.0018762666039719988\n",
      "epoch:754 from 100000\n",
      "loss: 0.002264633014789297\n",
      "epoch:755 from 100000\n",
      "loss: 0.002477815635927527\n",
      "epoch:756 from 100000\n",
      "loss: 0.0026973373244914\n",
      "epoch:757 from 100000\n",
      "loss: 0.0018836275428668614\n",
      "epoch:758 from 100000\n",
      "loss: 0.002788482827838834\n",
      "epoch:759 from 100000\n",
      "loss: 0.0022567365981558396\n",
      "epoch:760 from 100000\n",
      "loss: 0.0026592770632873908\n",
      "epoch:761 from 100000\n",
      "loss: 0.0025546564402247895\n",
      "epoch:762 from 100000\n",
      "loss: 0.0025584785097976237\n",
      "epoch:763 from 100000\n",
      "loss: 0.003145559975166634\n",
      "epoch:764 from 100000\n",
      "loss: 0.0026867437466080446\n",
      "epoch:765 from 100000\n",
      "loss: 0.0025100812114260407\n",
      "epoch:766 from 100000\n",
      "loss: 0.0028393363685097484\n",
      "epoch:767 from 100000\n",
      "loss: 0.002317605935274969\n",
      "epoch:768 from 100000\n",
      "loss: 0.002846314787063875\n",
      "epoch:769 from 100000\n",
      "loss: 0.002859014292653228\n",
      "epoch:770 from 100000\n",
      "loss: 0.0027085592196272046\n",
      "epoch:771 from 100000\n",
      "loss: 0.002835876479423405\n",
      "epoch:772 from 100000\n",
      "loss: 0.0020233921301837654\n",
      "epoch:773 from 100000\n",
      "loss: 0.0025204578724355997\n",
      "epoch:774 from 100000\n",
      "loss: 0.002422316780013034\n",
      "epoch:775 from 100000\n",
      "loss: 0.002533106866479784\n",
      "epoch:776 from 100000\n",
      "loss: 0.0023927668113117306\n",
      "epoch:777 from 100000\n",
      "loss: 0.00242177281981526\n",
      "epoch:778 from 100000\n",
      "loss: 0.002551519736869068\n",
      "epoch:779 from 100000\n",
      "loss: 0.002119110368425936\n",
      "epoch:780 from 100000\n",
      "loss: 0.002473711774342746\n",
      "epoch:781 from 100000\n",
      "loss: 0.0026729298027134973\n",
      "epoch:782 from 100000\n",
      "loss: 0.003023622499455314\n",
      "epoch:783 from 100000\n",
      "loss: 0.0026793737585393274\n",
      "epoch:784 from 100000\n",
      "loss: 0.0029964905507995354\n",
      "epoch:785 from 100000\n",
      "loss: 0.0025288133716117045\n",
      "epoch:786 from 100000\n",
      "loss: 0.0024186720058594346\n",
      "epoch:787 from 100000\n",
      "loss: 0.002449024378222475\n",
      "epoch:788 from 100000\n",
      "loss: 0.0027480735660674327\n",
      "epoch:789 from 100000\n",
      "loss: 0.0023843957872600186\n",
      "epoch:790 from 100000\n",
      "loss: 0.0031151070635360156\n",
      "epoch:791 from 100000\n",
      "loss: 0.002619930460582509\n",
      "epoch:792 from 100000\n",
      "loss: 0.0025579599088985105\n",
      "epoch:793 from 100000\n",
      "loss: 0.0018110277735559066\n",
      "epoch:794 from 100000\n",
      "loss: 0.0021894505367470174\n",
      "epoch:795 from 100000\n",
      "loss: 0.0025123821901757992\n",
      "epoch:796 from 100000\n",
      "loss: 0.002685924803841999\n",
      "epoch:797 from 100000\n",
      "loss: 0.0020561377477577025\n",
      "epoch:798 from 100000\n",
      "loss: 0.0022229763098380317\n",
      "epoch:799 from 100000\n",
      "loss: 0.0021470172410431587\n",
      "epoch:800 from 100000\n",
      "loss: 0.0027360472585939988\n",
      "epoch:801 from 100000\n",
      "loss: 0.0026706134921710145\n",
      "epoch:802 from 100000\n",
      "loss: 0.002169620024943697\n",
      "epoch:803 from 100000\n",
      "loss: 0.00270445447961265\n",
      "epoch:804 from 100000\n",
      "loss: 0.002183297549265717\n",
      "epoch:805 from 100000\n",
      "loss: 0.0023274051249018157\n",
      "epoch:806 from 100000\n",
      "loss: 0.0018286307729340479\n",
      "epoch:807 from 100000\n",
      "loss: 0.002530042214743844\n",
      "epoch:808 from 100000\n",
      "loss: 0.002200170180060468\n",
      "epoch:809 from 100000\n",
      "loss: 0.0021950361024209997\n",
      "epoch:810 from 100000\n",
      "loss: 0.002607136275656785\n",
      "epoch:811 from 100000\n",
      "loss: 0.002949469365916002\n",
      "epoch:812 from 100000\n",
      "loss: 0.0025815766755385994\n",
      "epoch:813 from 100000\n",
      "loss: 0.0022001581016297678\n",
      "epoch:814 from 100000\n",
      "loss: 0.0026598047476795728\n",
      "epoch:815 from 100000\n",
      "loss: 0.002486071601850881\n",
      "epoch:816 from 100000\n",
      "loss: 0.00251257061825072\n",
      "epoch:817 from 100000\n",
      "loss: 0.0029925470567206958\n",
      "epoch:818 from 100000\n",
      "loss: 0.002502804516694823\n",
      "epoch:819 from 100000\n",
      "loss: 0.0023600184879342123\n",
      "epoch:820 from 100000\n",
      "loss: 0.002127756736882702\n",
      "epoch:821 from 100000\n",
      "loss: 0.0023235882866003976\n",
      "epoch:822 from 100000\n",
      "loss: 0.0024435667684485907\n",
      "epoch:823 from 100000\n",
      "loss: 0.0023931216388461962\n",
      "epoch:824 from 100000\n",
      "loss: 0.002349671966214828\n",
      "epoch:825 from 100000\n",
      "loss: 0.0024353124981644214\n",
      "epoch:826 from 100000\n",
      "loss: 0.00257453867831714\n",
      "epoch:827 from 100000\n",
      "loss: 0.0021915224949395906\n",
      "epoch:828 from 100000\n",
      "loss: 0.0024441482327404174\n",
      "epoch:829 from 100000\n",
      "loss: 0.0030070668451287474\n",
      "epoch:830 from 100000\n",
      "loss: 0.0026420162454598994\n",
      "epoch:831 from 100000\n",
      "loss: 0.002788641278414161\n",
      "epoch:832 from 100000\n",
      "loss: 0.0020617005655196863\n",
      "epoch:833 from 100000\n",
      "loss: 0.0024918685317061318\n",
      "epoch:834 from 100000\n",
      "loss: 0.001985666960507615\n",
      "epoch:835 from 100000\n",
      "loss: 0.0027840261313656356\n",
      "epoch:836 from 100000\n",
      "loss: 0.0028114087145070243\n",
      "epoch:837 from 100000\n",
      "loss: 0.002702815467102937\n",
      "epoch:838 from 100000\n",
      "loss: 0.002274454688745209\n",
      "epoch:839 from 100000\n",
      "loss: 0.001937609258646944\n",
      "epoch:840 from 100000\n",
      "loss: 0.0024512284250022276\n",
      "epoch:841 from 100000\n",
      "loss: 0.002763198165155245\n",
      "epoch:842 from 100000\n",
      "loss: 0.0026680389862576703\n",
      "epoch:843 from 100000\n",
      "loss: 0.002282729679336626\n",
      "epoch:844 from 100000\n",
      "loss: 0.0021598413296146646\n",
      "epoch:845 from 100000\n",
      "loss: 0.002557494222372725\n",
      "epoch:846 from 100000\n",
      "loss: 0.002202583264988789\n",
      "epoch:847 from 100000\n",
      "loss: 0.0021125411473406075\n",
      "epoch:848 from 100000\n",
      "loss: 0.002780396981506783\n",
      "epoch:849 from 100000\n",
      "loss: 0.002481889741716259\n",
      "epoch:850 from 100000\n",
      "loss: 0.002876740945112033\n",
      "epoch:851 from 100000\n",
      "loss: 0.0025982658615362197\n",
      "epoch:852 from 100000\n",
      "loss: 0.0028882651945423277\n",
      "epoch:853 from 100000\n",
      "loss: 0.0021229495613397376\n",
      "epoch:854 from 100000\n",
      "loss: 0.0021037396236351924\n",
      "epoch:855 from 100000\n",
      "loss: 0.0026394281782415874\n",
      "epoch:856 from 100000\n",
      "loss: 0.0027411081558170736\n",
      "epoch:857 from 100000\n",
      "loss: 0.002503034505735968\n",
      "epoch:858 from 100000\n",
      "loss: 0.0024857428209656973\n",
      "epoch:859 from 100000\n",
      "loss: 0.00292301669927042\n",
      "epoch:860 from 100000\n",
      "loss: 0.002873977166302666\n",
      "epoch:861 from 100000\n",
      "loss: 0.002576427581743701\n",
      "epoch:862 from 100000\n",
      "loss: 0.002653212364208457\n",
      "epoch:863 from 100000\n",
      "loss: 0.0023587397874962335\n",
      "epoch:864 from 100000\n",
      "loss: 0.002263782500790512\n",
      "epoch:865 from 100000\n",
      "loss: 0.0023964825172129167\n",
      "epoch:866 from 100000\n",
      "loss: 0.0026012003443725007\n",
      "epoch:867 from 100000\n",
      "loss: 0.002559242216335633\n",
      "epoch:868 from 100000\n",
      "loss: 0.0030511127521890558\n",
      "epoch:869 from 100000\n",
      "loss: 0.0021775929493514923\n",
      "epoch:870 from 100000\n",
      "loss: 0.002675966024227705\n",
      "epoch:871 from 100000\n",
      "loss: 0.0023410454068653053\n",
      "epoch:872 from 100000\n",
      "loss: 0.0026491813010807164\n",
      "epoch:873 from 100000\n",
      "loss: 0.003034808971762004\n",
      "epoch:874 from 100000\n",
      "loss: 0.002534827078136459\n",
      "epoch:875 from 100000\n",
      "loss: 0.002606201273977149\n",
      "epoch:876 from 100000\n",
      "loss: 0.002751201730717412\n",
      "epoch:877 from 100000\n",
      "loss: 0.002334850370203867\n",
      "epoch:878 from 100000\n",
      "loss: 0.00226399032536051\n",
      "epoch:879 from 100000\n",
      "loss: 0.002374277565451166\n",
      "epoch:880 from 100000\n",
      "loss: 0.0024078801055225085\n",
      "epoch:881 from 100000\n",
      "loss: 0.0029219505513538024\n",
      "retrying for data\n",
      "epoch:882 from 100000\n",
      "loss: 0.0024466082122387434\n",
      "epoch:883 from 100000\n",
      "loss: 0.0022805598723891762\n",
      "epoch:884 from 100000\n",
      "loss: 0.002498676962659374\n",
      "epoch:885 from 100000\n",
      "loss: 0.0027518981513310337\n",
      "epoch:886 from 100000\n",
      "loss: 0.0027616788233615353\n",
      "epoch:887 from 100000\n",
      "loss: 0.002646421061001547\n",
      "epoch:888 from 100000\n",
      "loss: 0.0023269082957710907\n",
      "epoch:889 from 100000\n",
      "loss: 0.0022681353365783252\n",
      "epoch:890 from 100000\n",
      "loss: 0.0021733908357646214\n",
      "epoch:891 from 100000\n",
      "loss: 0.0025584801553293346\n",
      "epoch:892 from 100000\n",
      "loss: 0.0023132701220447416\n",
      "epoch:893 from 100000\n",
      "loss: 0.0026835595653835753\n",
      "epoch:894 from 100000\n",
      "loss: 0.0027825341225025113\n",
      "epoch:895 from 100000\n",
      "loss: 0.0025426330105347006\n",
      "epoch:896 from 100000\n",
      "loss: 0.0024177929204824977\n",
      "epoch:897 from 100000\n",
      "loss: 0.0024060343059773004\n",
      "epoch:898 from 100000\n",
      "loss: 0.002539926028134687\n",
      "epoch:899 from 100000\n",
      "loss: 0.002335503857892718\n",
      "epoch:900 from 100000\n",
      "loss: 0.0029162357569134656\n",
      "epoch:901 from 100000\n",
      "loss: 0.003034790472241866\n",
      "epoch:902 from 100000\n",
      "loss: 0.002250869117375487\n",
      "epoch:903 from 100000\n",
      "loss: 0.00207421993366097\n",
      "epoch:904 from 100000\n",
      "loss: 0.0023339258049190903\n",
      "epoch:905 from 100000\n",
      "loss: 0.0026074268823776947\n",
      "epoch:906 from 100000\n",
      "loss: 0.0022920289003991456\n",
      "epoch:907 from 100000\n",
      "loss: 0.0027238921071557343\n",
      "epoch:908 from 100000\n",
      "loss: 0.0028992708286779134\n",
      "epoch:909 from 100000\n",
      "loss: 0.0023609887751661063\n",
      "epoch:910 from 100000\n",
      "loss: 0.0031045893521479684\n",
      "epoch:911 from 100000\n",
      "loss: 0.0027703129119913683\n",
      "epoch:912 from 100000\n",
      "loss: 0.002765147474690366\n",
      "epoch:913 from 100000\n",
      "loss: 0.002222692395747572\n",
      "epoch:914 from 100000\n",
      "loss: 0.0021385306191064046\n",
      "epoch:915 from 100000\n",
      "loss: 0.002542956247850725\n",
      "epoch:916 from 100000\n",
      "loss: 0.002802324610229334\n",
      "epoch:917 from 100000\n",
      "loss: 0.003623643923504005\n",
      "epoch:918 from 100000\n",
      "loss: 0.0018765058173926263\n",
      "epoch:919 from 100000\n",
      "loss: 0.002406991420059512\n",
      "epoch:920 from 100000\n",
      "loss: 0.0028354866531117295\n",
      "epoch:921 from 100000\n",
      "loss: 0.002621001035691961\n",
      "epoch:922 from 100000\n",
      "loss: 0.0022114262860384315\n",
      "epoch:923 from 100000\n",
      "loss: 0.00200879653138486\n",
      "epoch:924 from 100000\n",
      "loss: 0.002345854951357751\n",
      "epoch:925 from 100000\n",
      "loss: 0.002368185830306402\n",
      "epoch:926 from 100000\n",
      "loss: 0.002346874048924974\n",
      "epoch:927 from 100000\n",
      "loss: 0.0024248329671934243\n",
      "epoch:928 from 100000\n",
      "loss: 0.0028525388655680217\n",
      "epoch:929 from 100000\n",
      "loss: 0.00290088848028347\n",
      "epoch:930 from 100000\n",
      "loss: 0.002610233035454712\n",
      "epoch:931 from 100000\n",
      "loss: 0.002328494129841374\n",
      "epoch:932 from 100000\n",
      "loss: 0.0028822177083611678\n",
      "epoch:933 from 100000\n",
      "loss: 0.0024641166931758107\n",
      "epoch:934 from 100000\n",
      "loss: 0.0023270749582593453\n",
      "epoch:935 from 100000\n",
      "loss: 0.0027474451527496058\n",
      "epoch:936 from 100000\n",
      "loss: 0.0026128018475901627\n",
      "epoch:937 from 100000\n",
      "loss: 0.0023619523173294965\n",
      "epoch:938 from 100000\n",
      "loss: 0.002405269166274593\n",
      "epoch:939 from 100000\n",
      "loss: 0.002543558596443063\n",
      "epoch:940 from 100000\n",
      "loss: 0.0023543491311102116\n",
      "epoch:941 from 100000\n",
      "loss: 0.0025473139980647375\n",
      "epoch:942 from 100000\n",
      "loss: 0.0030341658921031467\n",
      "epoch:943 from 100000\n",
      "loss: 0.0023440569214301377\n",
      "epoch:944 from 100000\n",
      "loss: 0.0030729820580290834\n",
      "epoch:945 from 100000\n",
      "loss: 0.002661262143959675\n",
      "epoch:946 from 100000\n",
      "loss: 0.002165678809774363\n",
      "epoch:947 from 100000\n",
      "loss: 0.002739581161335991\n",
      "epoch:948 from 100000\n",
      "loss: 0.002494223969563336\n",
      "epoch:949 from 100000\n",
      "loss: 0.002249315667711471\n",
      "epoch:950 from 100000\n",
      "loss: 0.0022241697172660224\n",
      "epoch:951 from 100000\n",
      "loss: 0.0018442549515782503\n",
      "epoch:952 from 100000\n",
      "loss: 0.001881044710771107\n",
      "epoch:953 from 100000\n",
      "loss: 0.0020256726326408625\n",
      "epoch:954 from 100000\n",
      "loss: 0.0022278898773038236\n",
      "epoch:955 from 100000\n",
      "loss: 0.002538227550388683\n",
      "epoch:956 from 100000\n",
      "loss: 0.002652853738894123\n",
      "epoch:957 from 100000\n",
      "loss: 0.002404799477602637\n",
      "epoch:958 from 100000\n",
      "loss: 0.0023662366373002897\n",
      "epoch:959 from 100000\n",
      "loss: 0.0025691528686024867\n",
      "epoch:960 from 100000\n",
      "loss: 0.002464732195818442\n",
      "epoch:961 from 100000\n",
      "loss: 0.002268630792798376\n",
      "epoch:962 from 100000\n",
      "loss: 0.0024536851881009625\n",
      "epoch:963 from 100000\n",
      "loss: 0.002691108532872022\n",
      "retrying for data\n",
      "epoch:964 from 100000\n",
      "loss: 0.0026762576823671225\n",
      "epoch:965 from 100000\n",
      "loss: 0.0026908621690324708\n",
      "epoch:966 from 100000\n",
      "loss: 0.002556713457636306\n",
      "epoch:967 from 100000\n",
      "loss: 0.0022032780395306872\n",
      "epoch:968 from 100000\n",
      "loss: 0.0024809035181192485\n",
      "epoch:969 from 100000\n",
      "loss: 0.0024244956417760477\n",
      "epoch:970 from 100000\n",
      "loss: 0.002488464049946515\n",
      "epoch:971 from 100000\n",
      "loss: 0.0020944274133967156\n",
      "epoch:972 from 100000\n",
      "loss: 0.0029272742899024706\n",
      "epoch:973 from 100000\n",
      "loss: 0.0027356972022118953\n",
      "epoch:974 from 100000\n",
      "loss: 0.0029387695320792773\n",
      "epoch:975 from 100000\n",
      "loss: 0.0026519837784633182\n",
      "epoch:976 from 100000\n",
      "loss: 0.002220910376649954\n",
      "epoch:977 from 100000\n",
      "loss: 0.0023602925096781746\n",
      "epoch:978 from 100000\n",
      "loss: 0.0027129957035185726\n",
      "epoch:979 from 100000\n",
      "loss: 0.0025935714600962\n",
      "epoch:980 from 100000\n",
      "loss: 0.0020196459409191903\n",
      "epoch:981 from 100000\n",
      "loss: 0.0028561620144387234\n",
      "epoch:982 from 100000\n",
      "loss: 0.00284632950069863\n",
      "epoch:983 from 100000\n",
      "loss: 0.0024887286500643313\n",
      "epoch:984 from 100000\n",
      "loss: 0.0022502090809837227\n",
      "epoch:985 from 100000\n",
      "loss: 0.002113394649285283\n",
      "epoch:986 from 100000\n",
      "loss: 0.0030037415577339743\n",
      "epoch:987 from 100000\n",
      "loss: 0.0025504879888842424\n",
      "epoch:988 from 100000\n",
      "loss: 0.0017827685764757462\n",
      "epoch:989 from 100000\n",
      "loss: 0.002436990129098149\n",
      "epoch:990 from 100000\n",
      "loss: 0.0028085747725867805\n",
      "epoch:991 from 100000\n",
      "loss: 0.002438741393774535\n",
      "epoch:992 from 100000\n",
      "loss: 0.002422580776880068\n",
      "epoch:993 from 100000\n",
      "loss: 0.0019463920280600178\n",
      "epoch:994 from 100000\n",
      "loss: 0.0027764262640062043\n",
      "epoch:995 from 100000\n",
      "loss: 0.00318745188246794\n",
      "epoch:996 from 100000\n",
      "loss: 0.0029174503069384627\n",
      "epoch:997 from 100000\n",
      "loss: 0.002110760374506526\n",
      "epoch:998 from 100000\n",
      "loss: 0.0032114192731000912\n",
      "epoch:999 from 100000\n",
      "loss: 0.0019041008862501485\n",
      "epoch:1000 from 100000\n",
      "loss: 0.00206522001809617\n",
      "epoch:1001 from 100000\n",
      "loss: 0.0020167696503108345\n",
      "epoch:1002 from 100000\n",
      "loss: 0.0022026273663300344\n",
      "epoch:1003 from 100000\n",
      "loss: 0.0022875767490972976\n",
      "epoch:1004 from 100000\n",
      "loss: 0.002652502181604177\n",
      "epoch:1005 from 100000\n",
      "loss: 0.0026552436651172684\n",
      "epoch:1006 from 100000\n",
      "loss: 0.0024129486711217396\n",
      "epoch:1007 from 100000\n",
      "loss: 0.002283091704356366\n",
      "epoch:1008 from 100000\n",
      "loss: 0.0025038971398885224\n",
      "epoch:1009 from 100000\n",
      "loss: 0.002601729323401969\n",
      "epoch:1010 from 100000\n",
      "loss: 0.0022601014214274073\n",
      "epoch:1011 from 100000\n",
      "loss: 0.003039508526228474\n",
      "epoch:1012 from 100000\n",
      "loss: 0.0024031099240460208\n",
      "epoch:1013 from 100000\n",
      "loss: 0.002993445122626781\n",
      "epoch:1014 from 100000\n",
      "loss: 0.002826941886212353\n",
      "epoch:1015 from 100000\n",
      "loss: 0.0024606897878243217\n",
      "epoch:1016 from 100000\n",
      "loss: 0.002540118298242078\n",
      "epoch:1017 from 100000\n",
      "loss: 0.0028974975265043668\n",
      "epoch:1018 from 100000\n",
      "loss: 0.002617708982285194\n",
      "epoch:1019 from 100000\n",
      "loss: 0.0029124722859990015\n",
      "epoch:1020 from 100000\n",
      "loss: 0.0028294137273405795\n",
      "epoch:1021 from 100000\n",
      "loss: 0.002635487256554825\n",
      "epoch:1022 from 100000\n",
      "loss: 0.002487636896347567\n",
      "epoch:1023 from 100000\n",
      "loss: 0.002639180625408244\n",
      "epoch:1024 from 100000\n",
      "loss: 0.002929408430929925\n",
      "epoch:1025 from 100000\n",
      "loss: 0.0021601959956001338\n",
      "epoch:1026 from 100000\n",
      "loss: 0.0024543866277326742\n",
      "epoch:1027 from 100000\n",
      "loss: 0.001972453430596488\n",
      "epoch:1028 from 100000\n",
      "loss: 0.002524689306142136\n",
      "epoch:1029 from 100000\n",
      "loss: 0.0026206089161746604\n",
      "epoch:1030 from 100000\n",
      "loss: 0.002653559145926465\n",
      "epoch:1031 from 100000\n",
      "loss: 0.0021263228944974344\n",
      "epoch:1032 from 100000\n",
      "loss: 0.002652865643625546\n",
      "epoch:1033 from 100000\n",
      "loss: 0.0029097407530684904\n",
      "epoch:1034 from 100000\n",
      "loss: 0.002439729892557807\n",
      "epoch:1035 from 100000\n",
      "loss: 0.0023769694443274147\n",
      "epoch:1036 from 100000\n",
      "loss: 0.0029963757552593506\n",
      "epoch:1037 from 100000\n",
      "loss: 0.002282937853436806\n",
      "epoch:1038 from 100000\n",
      "loss: 0.0023089997823149133\n",
      "epoch:1039 from 100000\n",
      "loss: 0.0020331685020948953\n",
      "epoch:1040 from 100000\n",
      "loss: 0.002018474909050383\n",
      "epoch:1041 from 100000\n",
      "loss: 0.0022230294643179604\n",
      "epoch:1042 from 100000\n",
      "loss: 0.0019838680730543956\n",
      "epoch:1043 from 100000\n",
      "loss: 0.002281955743768549\n",
      "epoch:1044 from 100000\n",
      "loss: 0.0024927588515879506\n",
      "epoch:1045 from 100000\n",
      "loss: 0.002683950804851065\n",
      "epoch:1046 from 100000\n",
      "loss: 0.0025262906351599668\n",
      "epoch:1047 from 100000\n",
      "loss: 0.002084706875990605\n",
      "epoch:1048 from 100000\n",
      "loss: 0.003026060302431688\n",
      "epoch:1049 from 100000\n",
      "loss: 0.0016752438360185806\n",
      "epoch:1050 from 100000\n",
      "loss: 0.002068626289059239\n",
      "epoch:1051 from 100000\n",
      "loss: 0.0024549176157790953\n",
      "epoch:1052 from 100000\n",
      "loss: 0.0025668053525294\n",
      "epoch:1053 from 100000\n",
      "loss: 0.002030913636247078\n",
      "epoch:1054 from 100000\n",
      "loss: 0.001987615124704689\n",
      "epoch:1055 from 100000\n",
      "loss: 0.002186387691267555\n",
      "epoch:1056 from 100000\n",
      "loss: 0.002557471460306715\n",
      "epoch:1057 from 100000\n",
      "loss: 0.002700087469918344\n",
      "epoch:1058 from 100000\n",
      "loss: 0.002400602459857737\n",
      "epoch:1059 from 100000\n",
      "loss: 0.002118871283187218\n",
      "epoch:1060 from 100000\n",
      "loss: 0.0024100093762911\n",
      "epoch:1061 from 100000\n",
      "loss: 0.0020634227403064642\n",
      "epoch:1062 from 100000\n",
      "loss: 0.002529439685602597\n",
      "epoch:1063 from 100000\n",
      "loss: 0.002133522759550033\n",
      "epoch:1064 from 100000\n",
      "loss: 0.002842141749709981\n",
      "epoch:1065 from 100000\n",
      "loss: 0.002444349296837345\n",
      "epoch:1066 from 100000\n",
      "loss: 0.0022278334512861875\n",
      "epoch:1067 from 100000\n",
      "loss: 0.0021927449578242886\n",
      "epoch:1068 from 100000\n",
      "loss: 0.002438130854670817\n",
      "epoch:1069 from 100000\n",
      "loss: 0.0026599205017987515\n",
      "epoch:1070 from 100000\n",
      "loss: 0.0022378598917498493\n",
      "epoch:1071 from 100000\n",
      "loss: 0.002574870072464819\n",
      "epoch:1072 from 100000\n",
      "loss: 0.0029150183529509377\n",
      "epoch:1073 from 100000\n",
      "loss: 0.0025327342788159513\n",
      "epoch:1074 from 100000\n",
      "loss: 0.002333349255138728\n",
      "epoch:1075 from 100000\n",
      "loss: 0.0022021672859295904\n",
      "epoch:1076 from 100000\n",
      "loss: 0.0020425081877135653\n",
      "epoch:1077 from 100000\n",
      "loss: 0.0023119185568418743\n",
      "epoch:1078 from 100000\n",
      "loss: 0.00273659315422492\n",
      "epoch:1079 from 100000\n",
      "loss: 0.002793277680780193\n",
      "epoch:1080 from 100000\n",
      "loss: 0.0024607406854499914\n",
      "epoch:1081 from 100000\n",
      "loss: 0.0026856366748404525\n",
      "epoch:1082 from 100000\n",
      "loss: 0.0025583484720925753\n",
      "epoch:1083 from 100000\n",
      "loss: 0.002653859856877716\n",
      "epoch:1084 from 100000\n",
      "loss: 0.0022301439961864844\n",
      "epoch:1085 from 100000\n",
      "loss: 0.003201138168037687\n",
      "epoch:1086 from 100000\n",
      "loss: 0.002698643861663186\n",
      "epoch:1087 from 100000\n",
      "loss: 0.0023821291960643975\n",
      "epoch:1088 from 100000\n",
      "loss: 0.0022654273552120685\n",
      "epoch:1089 from 100000\n",
      "loss: 0.002986473403694845\n",
      "retrying for data\n",
      "epoch:1090 from 100000\n",
      "loss: 0.0024417776092491295\n",
      "epoch:1091 from 100000\n",
      "loss: 0.0026569516516019576\n",
      "epoch:1092 from 100000\n",
      "loss: 0.0029107883374308585\n",
      "epoch:1093 from 100000\n",
      "loss: 0.0025650480419869837\n",
      "epoch:1094 from 100000\n",
      "loss: 0.0023462930022617456\n",
      "retrying for data\n",
      "epoch:1095 from 100000\n",
      "loss: 0.0024830437132550287\n",
      "epoch:1096 from 100000\n",
      "loss: 0.0027965548052861777\n",
      "epoch:1097 from 100000\n",
      "loss: 0.0020378302963877104\n",
      "epoch:1098 from 100000\n",
      "loss: 0.0025149063007461336\n",
      "epoch:1099 from 100000\n",
      "loss: 0.002881345197209839\n",
      "epoch:1100 from 100000\n",
      "loss: 0.0018384270358069443\n",
      "epoch:1101 from 100000\n",
      "loss: 0.0028405245409999225\n",
      "epoch:1102 from 100000\n",
      "loss: 0.0026273315233993344\n",
      "epoch:1103 from 100000\n",
      "loss: 0.002311398120468766\n",
      "epoch:1104 from 100000\n",
      "loss: 0.0022992751218851026\n",
      "epoch:1105 from 100000\n",
      "loss: 0.0023392559299111326\n",
      "epoch:1106 from 100000\n",
      "loss: 0.0025080194905058306\n",
      "epoch:1107 from 100000\n",
      "loss: 0.0018121929980452478\n",
      "epoch:1108 from 100000\n",
      "loss: 0.002089175287466105\n",
      "epoch:1109 from 100000\n",
      "loss: 0.00278796696346717\n",
      "epoch:1110 from 100000\n",
      "loss: 0.0021246468505182747\n",
      "epoch:1111 from 100000\n",
      "loss: 0.0019638309948533106\n",
      "epoch:1112 from 100000\n",
      "loss: 0.0025319835483514908\n",
      "epoch:1113 from 100000\n",
      "loss: 0.0027506959051777358\n",
      "epoch:1114 from 100000\n",
      "loss: 0.0026714169895285522\n",
      "epoch:1115 from 100000\n",
      "loss: 0.0027970898932210275\n",
      "epoch:1116 from 100000\n",
      "loss: 0.0025585496608186986\n",
      "epoch:1117 from 100000\n",
      "loss: 0.0019965896369456004\n",
      "epoch:1118 from 100000\n",
      "loss: 0.00196035571440234\n",
      "epoch:1119 from 100000\n",
      "loss: 0.0021234821229256795\n",
      "epoch:1120 from 100000\n",
      "loss: 0.002434257492126335\n",
      "epoch:1121 from 100000\n",
      "loss: 0.0026749334019200433\n",
      "epoch:1122 from 100000\n",
      "loss: 0.002380653620051021\n",
      "epoch:1123 from 100000\n",
      "loss: 0.0024318449799238806\n",
      "epoch:1124 from 100000\n",
      "loss: 0.0026607186808007555\n",
      "epoch:1125 from 100000\n",
      "loss: 0.0026507146513381485\n",
      "epoch:1126 from 100000\n",
      "loss: 0.0024780082646600476\n",
      "epoch:1127 from 100000\n",
      "loss: 0.002639270445229158\n",
      "epoch:1128 from 100000\n",
      "loss: 0.0027988775087521844\n",
      "epoch:1129 from 100000\n",
      "loss: 0.002930852512349702\n",
      "epoch:1130 from 100000\n",
      "loss: 0.002400613177286459\n",
      "epoch:1131 from 100000\n",
      "loss: 0.002975371345712574\n",
      "epoch:1132 from 100000\n",
      "loss: 0.0025394525964941295\n",
      "epoch:1133 from 100000\n",
      "loss: 0.0022130018469681545\n",
      "epoch:1134 from 100000\n",
      "loss: 0.0019504401170706842\n",
      "epoch:1135 from 100000\n",
      "loss: 0.0020838509071836597\n",
      "epoch:1136 from 100000\n",
      "loss: 0.002638301242853913\n",
      "epoch:1137 from 100000\n",
      "loss: 0.0027732383106240377\n",
      "epoch:1138 from 100000\n",
      "loss: 0.0026118042833331856\n",
      "epoch:1139 from 100000\n",
      "loss: 0.0023022365434144376\n",
      "epoch:1140 from 100000\n",
      "loss: 0.0026533746749066722\n",
      "epoch:1141 from 100000\n",
      "loss: 0.0025194128647854086\n",
      "epoch:1142 from 100000\n",
      "loss: 0.002405292181464347\n",
      "epoch:1143 from 100000\n",
      "loss: 0.0022563558397337147\n",
      "epoch:1144 from 100000\n",
      "loss: 0.002388328504906667\n",
      "epoch:1145 from 100000\n",
      "loss: 0.001983770898249304\n",
      "epoch:1146 from 100000\n",
      "loss: 0.0027001733512861392\n",
      "epoch:1147 from 100000\n",
      "loss: 0.0023857536537548185\n",
      "epoch:1148 from 100000\n",
      "loss: 0.002881902594396024\n",
      "epoch:1149 from 100000\n",
      "loss: 0.0024088195658578115\n",
      "epoch:1150 from 100000\n",
      "loss: 0.002584608963360324\n",
      "epoch:1151 from 100000\n",
      "loss: 0.002358175609629143\n",
      "epoch:1152 from 100000\n",
      "loss: 0.002273296627691934\n",
      "epoch:1153 from 100000\n",
      "loss: 0.00275747935745585\n",
      "epoch:1154 from 100000\n",
      "loss: 0.0025091105532908387\n",
      "epoch:1155 from 100000\n",
      "loss: 0.0030209142897490437\n",
      "epoch:1156 from 100000\n",
      "loss: 0.002797440194513001\n",
      "epoch:1157 from 100000\n",
      "loss: 0.001972995431032132\n",
      "epoch:1158 from 100000\n",
      "loss: 0.0021204366039739853\n",
      "epoch:1159 from 100000\n",
      "loss: 0.0019387388367420044\n",
      "epoch:1160 from 100000\n",
      "loss: 0.0028746923133979863\n",
      "epoch:1161 from 100000\n",
      "loss: 0.0024971842729542004\n",
      "epoch:1162 from 100000\n",
      "loss: 0.0031395091743320336\n",
      "epoch:1163 from 100000\n",
      "loss: 0.0025481300993988043\n",
      "epoch:1164 from 100000\n",
      "loss: 0.0023326781634693816\n",
      "epoch:1165 from 100000\n",
      "loss: 0.0021249721492324625\n",
      "epoch:1166 from 100000\n",
      "loss: 0.0020838774956075667\n",
      "epoch:1167 from 100000\n",
      "loss: 0.002091664368435886\n",
      "epoch:1168 from 100000\n",
      "loss: 0.0022940631698133984\n",
      "epoch:1169 from 100000\n",
      "loss: 0.002122814566661191\n",
      "epoch:1170 from 100000\n",
      "loss: 0.003182496541882074\n",
      "epoch:1171 from 100000\n",
      "loss: 0.002988937812972381\n",
      "epoch:1172 from 100000\n",
      "loss: 0.0029451223904573\n",
      "epoch:1173 from 100000\n",
      "loss: 0.0029381335974392186\n",
      "epoch:1174 from 100000\n",
      "loss: 0.0029808409143043946\n",
      "epoch:1175 from 100000\n",
      "loss: 0.0025458710558154962\n",
      "epoch:1176 from 100000\n",
      "loss: 0.0025696967998669606\n",
      "epoch:1177 from 100000\n",
      "loss: 0.0026804991304629766\n",
      "epoch:1178 from 100000\n",
      "loss: 0.002274910014477882\n",
      "epoch:1179 from 100000\n",
      "loss: 0.002470530341867061\n",
      "epoch:1180 from 100000\n",
      "loss: 0.002586832408326245\n",
      "epoch:1181 from 100000\n",
      "loss: 0.003356024285892545\n",
      "epoch:1182 from 100000\n",
      "loss: 0.0021818823005901322\n",
      "epoch:1183 from 100000\n",
      "loss: 0.0022074331213843834\n",
      "epoch:1184 from 100000\n",
      "loss: 0.0026137848117855356\n",
      "epoch:1185 from 100000\n",
      "loss: 0.002285073528611292\n",
      "epoch:1186 from 100000\n",
      "loss: 0.0026902888755842014\n",
      "epoch:1187 from 100000\n",
      "loss: 0.002500523519302078\n",
      "epoch:1188 from 100000\n",
      "loss: 0.002592799359717901\n",
      "epoch:1189 from 100000\n",
      "loss: 0.0025854941960830047\n",
      "epoch:1190 from 100000\n",
      "loss: 0.002753032107392528\n",
      "epoch:1191 from 100000\n",
      "loss: 0.002141429849729093\n",
      "epoch:1192 from 100000\n",
      "loss: 0.0027695536000749144\n",
      "epoch:1193 from 100000\n",
      "loss: 0.0026186091354247765\n",
      "epoch:1194 from 100000\n",
      "loss: 0.0025086587878604405\n",
      "epoch:1195 from 100000\n",
      "loss: 0.0029299757861167564\n",
      "epoch:1196 from 100000\n",
      "loss: 0.0024864866101097505\n",
      "epoch:1197 from 100000\n",
      "loss: 0.0029277794031941085\n",
      "epoch:1198 from 100000\n",
      "loss: 0.002373838365542724\n",
      "epoch:1199 from 100000\n",
      "loss: 0.0021853840241874423\n",
      "epoch:1200 from 100000\n",
      "loss: 0.0021209803518900117\n",
      "epoch:1201 from 100000\n",
      "loss: 0.0024226074262969632\n",
      "epoch:1202 from 100000\n",
      "loss: 0.0021446274615470884\n",
      "epoch:1203 from 100000\n",
      "loss: 0.002279082780603403\n",
      "epoch:1204 from 100000\n",
      "loss: 0.0023194726921360598\n",
      "epoch:1205 from 100000\n",
      "loss: 0.0030631839317720733\n",
      "epoch:1206 from 100000\n",
      "loss: 0.002270981314808296\n",
      "epoch:1207 from 100000\n",
      "loss: 0.002784643124968511\n",
      "epoch:1208 from 100000\n",
      "loss: 0.0020799672933833335\n",
      "epoch:1209 from 100000\n",
      "loss: 0.0023412937417219837\n",
      "epoch:1210 from 100000\n",
      "loss: 0.0022998543747689837\n",
      "epoch:1211 from 100000\n",
      "loss: 0.002851631670921506\n",
      "epoch:1212 from 100000\n",
      "loss: 0.0020713292747132073\n",
      "epoch:1213 from 100000\n",
      "loss: 0.0024091709328786237\n",
      "epoch:1214 from 100000\n",
      "loss: 0.0028683580839867773\n",
      "epoch:1215 from 100000\n",
      "loss: 0.0023590656846295133\n",
      "epoch:1216 from 100000\n",
      "loss: 0.0021072737519602924\n",
      "epoch:1217 from 100000\n",
      "loss: 0.002295407561604179\n",
      "epoch:1218 from 100000\n",
      "loss: 0.0021281800256645056\n",
      "epoch:1219 from 100000\n",
      "loss: 0.0028878747574765384\n",
      "epoch:1220 from 100000\n",
      "loss: 0.001952715286165585\n",
      "epoch:1221 from 100000\n",
      "loss: 0.0025924829014911666\n",
      "epoch:1222 from 100000\n",
      "loss: 0.002387253342362783\n",
      "epoch:1223 from 100000\n",
      "loss: 0.0023275070271324694\n",
      "epoch:1224 from 100000\n",
      "loss: 0.0026866999881463016\n",
      "epoch:1225 from 100000\n",
      "loss: 0.0021665011236677856\n",
      "epoch:1226 from 100000\n",
      "loss: 0.002669730836345252\n",
      "epoch:1227 from 100000\n",
      "loss: 0.0028375386258403523\n",
      "epoch:1228 from 100000\n",
      "loss: 0.002300027658577619\n",
      "epoch:1229 from 100000\n",
      "loss: 0.0020177337340214763\n",
      "epoch:1230 from 100000\n",
      "loss: 0.0024561542736591946\n",
      "epoch:1231 from 100000\n",
      "loss: 0.002554179759300723\n",
      "epoch:1232 from 100000\n",
      "loss: 0.0029365348400745006\n",
      "epoch:1233 from 100000\n",
      "loss: 0.0024544564486745912\n",
      "epoch:1234 from 100000\n",
      "loss: 0.0022632316515966977\n",
      "epoch:1235 from 100000\n",
      "loss: 0.003043114109658518\n",
      "epoch:1236 from 100000\n",
      "loss: 0.0021528955692815543\n",
      "epoch:1237 from 100000\n",
      "loss: 0.002714935766675808\n",
      "epoch:1238 from 100000\n",
      "loss: 0.0018591738895565868\n",
      "epoch:1239 from 100000\n",
      "loss: 0.0027418143266686457\n",
      "epoch:1240 from 100000\n",
      "loss: 0.002396083285361783\n",
      "epoch:1241 from 100000\n",
      "loss: 0.002814965061190833\n",
      "epoch:1242 from 100000\n",
      "loss: 0.002096121461391931\n",
      "epoch:1243 from 100000\n",
      "loss: 0.0026822568727880025\n",
      "epoch:1244 from 100000\n",
      "loss: 0.0029281468187321025\n",
      "epoch:1245 from 100000\n",
      "loss: 0.002415017621615334\n",
      "epoch:1246 from 100000\n",
      "loss: 0.002397244663455922\n",
      "epoch:1247 from 100000\n",
      "loss: 0.002245252398665798\n",
      "epoch:1248 from 100000\n",
      "loss: 0.0024300720303926937\n",
      "epoch:1249 from 100000\n",
      "loss: 0.0026003058023889025\n",
      "epoch:1250 from 100000\n",
      "loss: 0.002508663385469845\n",
      "epoch:1251 from 100000\n",
      "loss: 0.0023501563040611018\n",
      "epoch:1252 from 100000\n",
      "loss: 0.002555765483293726\n",
      "epoch:1253 from 100000\n",
      "loss: 0.002443196088449895\n",
      "epoch:1254 from 100000\n",
      "loss: 0.002936626141092802\n",
      "epoch:1255 from 100000\n",
      "loss: 0.0023436285145237434\n",
      "epoch:1256 from 100000\n",
      "loss: 0.0029946801337814577\n",
      "epoch:1257 from 100000\n",
      "loss: 0.0022683197096284857\n",
      "epoch:1258 from 100000\n",
      "loss: 0.0020588616006591565\n",
      "epoch:1259 from 100000\n",
      "loss: 0.002955913680523281\n",
      "epoch:1260 from 100000\n",
      "loss: 0.0022249091886124006\n",
      "epoch:1261 from 100000\n",
      "loss: 0.002350345961133371\n",
      "epoch:1262 from 100000\n",
      "loss: 0.0021396271379785503\n",
      "epoch:1263 from 100000\n",
      "loss: 0.002336672639387416\n",
      "epoch:1264 from 100000\n",
      "loss: 0.0025607282381940877\n",
      "epoch:1265 from 100000\n",
      "loss: 0.0024061587339190282\n",
      "epoch:1266 from 100000\n",
      "loss: 0.0026221297233917085\n",
      "epoch:1267 from 100000\n",
      "loss: 0.002756464760267363\n",
      "epoch:1268 from 100000\n",
      "loss: 0.002512313273257405\n",
      "epoch:1269 from 100000\n",
      "loss: 0.0025447540699019555\n",
      "epoch:1270 from 100000\n",
      "loss: 0.0025509089277306884\n",
      "epoch:1271 from 100000\n",
      "loss: 0.0020436241082961715\n",
      "epoch:1272 from 100000\n",
      "loss: 0.0019194154393744611\n",
      "epoch:1273 from 100000\n",
      "loss: 0.002293324153924914\n",
      "epoch:1274 from 100000\n",
      "loss: 0.0021730728786906184\n",
      "epoch:1275 from 100000\n",
      "loss: 0.0025077739351786477\n",
      "epoch:1276 from 100000\n",
      "loss: 0.002925506550184309\n",
      "epoch:1277 from 100000\n",
      "loss: 0.0024085574613366134\n",
      "epoch:1278 from 100000\n",
      "loss: 0.002466918319356637\n",
      "epoch:1279 from 100000\n",
      "loss: 0.0027590406216404517\n",
      "epoch:1280 from 100000\n",
      "loss: 0.002818765028365533\n",
      "epoch:1281 from 100000\n",
      "loss: 0.002763803031456291\n",
      "epoch:1282 from 100000\n",
      "loss: 0.0026564957393304667\n",
      "epoch:1283 from 100000\n",
      "loss: 0.002976385056683739\n",
      "epoch:1284 from 100000\n",
      "loss: 0.00298966477345175\n",
      "epoch:1285 from 100000\n",
      "loss: 0.0026317964191662213\n",
      "epoch:1286 from 100000\n",
      "loss: 0.0026298118695535777\n",
      "epoch:1287 from 100000\n",
      "loss: 0.0027043225828151662\n",
      "epoch:1288 from 100000\n",
      "loss: 0.002700985917527987\n",
      "epoch:1289 from 100000\n",
      "loss: 0.002409492889427156\n",
      "epoch:1290 from 100000\n",
      "loss: 0.002668752348341741\n",
      "epoch:1291 from 100000\n",
      "loss: 0.002333090774300217\n",
      "epoch:1292 from 100000\n",
      "loss: 0.002564910451440028\n",
      "epoch:1293 from 100000\n",
      "loss: 0.0021120882668981267\n",
      "epoch:1294 from 100000\n",
      "loss: 0.0030617708679869793\n",
      "epoch:1295 from 100000\n",
      "loss: 0.002570538876966566\n",
      "epoch:1296 from 100000\n",
      "loss: 0.0022066036183900906\n",
      "epoch:1297 from 100000\n",
      "loss: 0.002547646256914504\n",
      "epoch:1298 from 100000\n",
      "loss: 0.0024034997857143026\n",
      "epoch:1299 from 100000\n",
      "loss: 0.0029946617856921876\n",
      "epoch:1300 from 100000\n",
      "loss: 0.0026488568047398076\n",
      "epoch:1301 from 100000\n",
      "loss: 0.003112517427325656\n",
      "epoch:1302 from 100000\n",
      "loss: 0.002461486528147816\n",
      "epoch:1303 from 100000\n",
      "loss: 0.002588477133855349\n",
      "epoch:1304 from 100000\n",
      "loss: 0.002472440887117955\n",
      "epoch:1305 from 100000\n",
      "loss: 0.0027980641310705323\n",
      "epoch:1306 from 100000\n",
      "loss: 0.00210731133694253\n",
      "epoch:1307 from 100000\n",
      "loss: 0.0026715846236555763\n",
      "epoch:1308 from 100000\n",
      "loss: 0.002855427350567652\n",
      "epoch:1309 from 100000\n",
      "loss: 0.0023479925257419154\n",
      "epoch:1310 from 100000\n",
      "loss: 0.0030078891844027567\n",
      "epoch:1311 from 100000\n",
      "loss: 0.0022095424820349763\n",
      "epoch:1312 from 100000\n",
      "loss: 0.0023293759369664713\n",
      "epoch:1313 from 100000\n",
      "loss: 0.0023159831843031498\n",
      "epoch:1314 from 100000\n",
      "loss: 0.002608579467732852\n",
      "epoch:1315 from 100000\n",
      "loss: 0.002318490122135586\n",
      "epoch:1316 from 100000\n",
      "loss: 0.0019356037892066524\n",
      "epoch:1317 from 100000\n",
      "loss: 0.0022826982121841866\n",
      "epoch:1318 from 100000\n",
      "loss: 0.0027945559070587933\n",
      "epoch:1319 from 100000\n",
      "loss: 0.002807878754367721\n",
      "epoch:1320 from 100000\n",
      "loss: 0.002755864480633363\n",
      "epoch:1321 from 100000\n",
      "loss: 0.0027411240005790205\n",
      "epoch:1322 from 100000\n",
      "loss: 0.0025915308092976375\n",
      "epoch:1323 from 100000\n",
      "loss: 0.0025637461253609217\n",
      "epoch:1324 from 100000\n",
      "loss: 0.002648698875759692\n",
      "epoch:1325 from 100000\n",
      "loss: 0.0023592404807573075\n",
      "retrying for data\n",
      "epoch:1326 from 100000\n",
      "loss: 0.002481077423624356\n",
      "epoch:1327 from 100000\n",
      "loss: 0.002312247661507172\n",
      "epoch:1328 from 100000\n",
      "loss: 0.002374761171950013\n",
      "epoch:1329 from 100000\n",
      "loss: 0.0022961808817854035\n",
      "epoch:1330 from 100000\n",
      "loss: 0.0023538301558687635\n",
      "epoch:1331 from 100000\n",
      "loss: 0.0026498845543585503\n",
      "epoch:1332 from 100000\n",
      "loss: 0.0023037489955726187\n",
      "epoch:1333 from 100000\n",
      "loss: 0.0024014589941145914\n",
      "epoch:1334 from 100000\n",
      "loss: 0.0023338478283676523\n",
      "epoch:1335 from 100000\n",
      "loss: 0.003009011650078719\n",
      "epoch:1336 from 100000\n",
      "loss: 0.0025166914553835795\n",
      "epoch:1337 from 100000\n",
      "loss: 0.0028808428187971913\n",
      "epoch:1338 from 100000\n",
      "loss: 0.002363908805563142\n",
      "epoch:1339 from 100000\n",
      "loss: 0.0022003106852537258\n",
      "epoch:1340 from 100000\n",
      "loss: 0.0016949260995744453\n",
      "epoch:1341 from 100000\n",
      "loss: 0.002955032075476538\n",
      "epoch:1342 from 100000\n",
      "loss: 0.00230053628541782\n",
      "epoch:1343 from 100000\n",
      "loss: 0.002491238463164791\n",
      "epoch:1344 from 100000\n",
      "loss: 0.002511270014508682\n",
      "epoch:1345 from 100000\n",
      "loss: 0.0025482378575674147\n",
      "epoch:1346 from 100000\n",
      "loss: 0.0027961008242982643\n",
      "epoch:1347 from 100000\n",
      "loss: 0.0023209095786853595\n",
      "epoch:1348 from 100000\n",
      "loss: 0.0026989906890264592\n",
      "epoch:1349 from 100000\n",
      "loss: 0.002653011177926601\n",
      "epoch:1350 from 100000\n",
      "loss: 0.001911325278001641\n",
      "epoch:1351 from 100000\n",
      "loss: 0.0024821394556227006\n",
      "epoch:1352 from 100000\n",
      "loss: 0.0022335067452559088\n",
      "epoch:1353 from 100000\n",
      "loss: 0.002276531781859603\n",
      "epoch:1354 from 100000\n",
      "loss: 0.0032692384875474545\n",
      "epoch:1355 from 100000\n",
      "loss: 0.0020181459998127593\n",
      "epoch:1356 from 100000\n",
      "loss: 0.002163775857695782\n",
      "epoch:1357 from 100000\n",
      "loss: 0.0024342567301118834\n",
      "epoch:1358 from 100000\n",
      "loss: 0.0031028648328685904\n",
      "epoch:1359 from 100000\n",
      "loss: 0.0029406165072316526\n",
      "epoch:1360 from 100000\n",
      "loss: 0.0020877748476664237\n",
      "epoch:1361 from 100000\n",
      "loss: 0.0022448245270254574\n",
      "epoch:1362 from 100000\n",
      "loss: 0.0021895976946098017\n",
      "epoch:1363 from 100000\n",
      "loss: 0.003029382865605612\n",
      "epoch:1364 from 100000\n",
      "loss: 0.0022969116623698937\n",
      "epoch:1365 from 100000\n",
      "loss: 0.0024773210840578486\n",
      "epoch:1366 from 100000\n",
      "loss: 0.002364884264380862\n",
      "epoch:1367 from 100000\n",
      "loss: 0.002947281313709027\n",
      "epoch:1368 from 100000\n",
      "loss: 0.0020574403004616215\n",
      "epoch:1369 from 100000\n",
      "loss: 0.002381569375188519\n",
      "epoch:1370 from 100000\n",
      "loss: 0.0026447653216905564\n",
      "epoch:1371 from 100000\n",
      "loss: 0.002849912395447518\n",
      "epoch:1372 from 100000\n",
      "loss: 0.002480393262715097\n",
      "epoch:1373 from 100000\n",
      "loss: 0.0024501994936372284\n",
      "epoch:1374 from 100000\n",
      "loss: 0.0027930838510741296\n",
      "epoch:1375 from 100000\n",
      "loss: 0.002715847928129733\n",
      "epoch:1376 from 100000\n",
      "loss: 0.002117992974092431\n",
      "epoch:1377 from 100000\n",
      "loss: 0.0026133921891755563\n",
      "epoch:1378 from 100000\n",
      "loss: 0.002583787573769314\n",
      "epoch:1379 from 100000\n",
      "loss: 0.00251621280301606\n",
      "epoch:1380 from 100000\n",
      "loss: 0.002217323878625166\n",
      "retrying for data\n",
      "epoch:1381 from 100000\n",
      "loss: 0.0029081665370540577\n",
      "epoch:1382 from 100000\n",
      "loss: 0.0021793159078811186\n",
      "epoch:1383 from 100000\n",
      "loss: 0.002420254422332846\n",
      "epoch:1384 from 100000\n",
      "loss: 0.0029148716328677438\n",
      "epoch:1385 from 100000\n",
      "loss: 0.0023611049030023423\n",
      "epoch:1386 from 100000\n",
      "loss: 0.0021972970648675982\n",
      "epoch:1387 from 100000\n",
      "loss: 0.002633821234383049\n",
      "epoch:1388 from 100000\n",
      "loss: 0.002293881266254516\n",
      "epoch:1389 from 100000\n",
      "loss: 0.0026987812684069468\n",
      "epoch:1390 from 100000\n",
      "loss: 0.002363959817500927\n",
      "epoch:1391 from 100000\n",
      "loss: 0.002484027654361398\n",
      "epoch:1392 from 100000\n",
      "loss: 0.002666554211799621\n",
      "epoch:1393 from 100000\n",
      "loss: 0.0021875218945694996\n",
      "epoch:1394 from 100000\n",
      "loss: 0.0027120322906739602\n",
      "epoch:1395 from 100000\n",
      "loss: 0.0027082150649846426\n",
      "epoch:1396 from 100000\n",
      "loss: 0.0024519181922073585\n",
      "epoch:1397 from 100000\n",
      "loss: 0.002430203357079108\n",
      "epoch:1398 from 100000\n",
      "loss: 0.0025418567084614097\n",
      "epoch:1399 from 100000\n",
      "loss: 0.0027389813969449506\n",
      "epoch:1400 from 100000\n",
      "loss: 0.002593616883046934\n",
      "epoch:1401 from 100000\n",
      "loss: 0.0029277601933586084\n",
      "epoch:1402 from 100000\n",
      "loss: 0.002576884651944056\n",
      "epoch:1403 from 100000\n",
      "loss: 0.0019530462809029814\n",
      "epoch:1404 from 100000\n",
      "loss: 0.002634710940384366\n",
      "epoch:1405 from 100000\n",
      "loss: 0.0033808140619839833\n",
      "epoch:1406 from 100000\n",
      "loss: 0.00228525874624097\n",
      "epoch:1407 from 100000\n",
      "loss: 0.002509605069576537\n",
      "epoch:1408 from 100000\n",
      "loss: 0.0027082543490450917\n",
      "epoch:1409 from 100000\n",
      "loss: 0.00271823230531254\n",
      "epoch:1410 from 100000\n",
      "loss: 0.0027782852316420303\n",
      "epoch:1411 from 100000\n",
      "loss: 0.002232323426540006\n",
      "epoch:1412 from 100000\n",
      "loss: 0.0023340621482788038\n",
      "epoch:1413 from 100000\n",
      "loss: 0.0025690818377199776\n",
      "epoch:1414 from 100000\n",
      "loss: 0.0027631961703775687\n",
      "epoch:1415 from 100000\n",
      "loss: 0.0029644052154935707\n",
      "epoch:1416 from 100000\n",
      "loss: 0.0017361520693270904\n",
      "epoch:1417 from 100000\n",
      "loss: 0.0019809569008373273\n",
      "epoch:1418 from 100000\n",
      "loss: 0.0024666689618300097\n",
      "epoch:1419 from 100000\n",
      "loss: 0.002350061752537158\n",
      "epoch:1420 from 100000\n",
      "loss: 0.002436535731447975\n",
      "epoch:1421 from 100000\n",
      "loss: 0.002497279034770372\n",
      "epoch:1422 from 100000\n",
      "loss: 0.002530535646513954\n",
      "epoch:1423 from 100000\n",
      "loss: 0.002213327271022081\n",
      "epoch:1424 from 100000\n",
      "loss: 0.0024747267277120955\n",
      "epoch:1425 from 100000\n",
      "loss: 0.002612087060470003\n",
      "epoch:1426 from 100000\n",
      "loss: 0.0022741794717831\n",
      "epoch:1427 from 100000\n",
      "loss: 0.002306892464275734\n",
      "epoch:1428 from 100000\n",
      "loss: 0.0026408986869910223\n",
      "epoch:1429 from 100000\n",
      "loss: 0.0026478693360161287\n",
      "epoch:1430 from 100000\n",
      "loss: 0.0023857802506199732\n",
      "epoch:1431 from 100000\n",
      "loss: 0.0023478008115205284\n",
      "epoch:1432 from 100000\n",
      "loss: 0.0026534499298804803\n",
      "epoch:1433 from 100000\n",
      "loss: 0.002461290228438884\n",
      "epoch:1434 from 100000\n",
      "loss: 0.0023308282258938107\n",
      "epoch:1435 from 100000\n",
      "loss: 0.002513483639262404\n",
      "epoch:1436 from 100000\n",
      "loss: 0.0023167482067947276\n",
      "epoch:1437 from 100000\n",
      "loss: 0.002737793561351509\n",
      "epoch:1438 from 100000\n",
      "loss: 0.0023830545009957405\n",
      "epoch:1439 from 100000\n",
      "loss: 0.002168166985171638\n",
      "epoch:1440 from 100000\n",
      "loss: 0.002376596585918378\n",
      "epoch:1441 from 100000\n",
      "loss: 0.002262548647195217\n",
      "epoch:1442 from 100000\n",
      "loss: 0.002244353600275417\n",
      "epoch:1443 from 100000\n",
      "loss: 0.0027349430812648734\n",
      "epoch:1444 from 100000\n",
      "loss: 0.0025480252740237574\n",
      "epoch:1445 from 100000\n",
      "loss: 0.002032821732541379\n",
      "epoch:1446 from 100000\n",
      "loss: 0.0026135244544889247\n",
      "epoch:1447 from 100000\n",
      "loss: 0.002783327754457332\n",
      "epoch:1448 from 100000\n",
      "loss: 0.0024979659633856954\n",
      "epoch:1449 from 100000\n",
      "loss: 0.002241831454256271\n",
      "epoch:1450 from 100000\n",
      "loss: 0.0028484441379532655\n",
      "epoch:1451 from 100000\n",
      "loss: 0.0017093500626401692\n",
      "epoch:1452 from 100000\n",
      "loss: 0.002559165487980408\n",
      "epoch:1453 from 100000\n",
      "loss: 0.001949171073619027\n",
      "epoch:1454 from 100000\n",
      "loss: 0.0029049581605136154\n",
      "epoch:1455 from 100000\n",
      "loss: 0.0034589721546467445\n",
      "epoch:1456 from 100000\n",
      "loss: 0.0029999068313202315\n",
      "epoch:1457 from 100000\n",
      "loss: 0.002581856940025773\n",
      "epoch:1458 from 100000\n",
      "loss: 0.0028069005438737804\n",
      "epoch:1459 from 100000\n",
      "loss: 0.002917919772443156\n",
      "epoch:1460 from 100000\n",
      "loss: 0.0028551398936258465\n",
      "epoch:1461 from 100000\n",
      "loss: 0.002550175172075342\n",
      "epoch:1462 from 100000\n",
      "loss: 0.0025779434860737638\n",
      "epoch:1463 from 100000\n",
      "loss: 0.002700761356749126\n",
      "epoch:1464 from 100000\n",
      "loss: 0.00276481362902814\n",
      "epoch:1465 from 100000\n",
      "loss: 0.002168954827681091\n",
      "epoch:1466 from 100000\n",
      "loss: 0.0025262293155208226\n",
      "epoch:1467 from 100000\n",
      "loss: 0.0024639126285421753\n",
      "retrying for data\n",
      "epoch:1468 from 100000\n",
      "loss: 0.00280859411043366\n",
      "epoch:1469 from 100000\n",
      "loss: 0.0020821151773873225\n",
      "epoch:1470 from 100000\n",
      "loss: 0.002986402320743764\n",
      "epoch:1471 from 100000\n",
      "loss: 0.0022872081713671832\n",
      "epoch:1472 from 100000\n",
      "loss: 0.002588128284344293\n",
      "epoch:1473 from 100000\n",
      "loss: 0.002111232563009935\n",
      "epoch:1474 from 100000\n",
      "loss: 0.0022289898871008518\n",
      "epoch:1475 from 100000\n",
      "loss: 0.0021901174693823577\n",
      "epoch:1476 from 100000\n",
      "loss: 0.002224208289874241\n",
      "epoch:1477 from 100000\n",
      "loss: 0.002294692966728462\n",
      "epoch:1478 from 100000\n",
      "loss: 0.0024658702681676914\n",
      "epoch:1479 from 100000\n",
      "loss: 0.0020684519294036363\n",
      "epoch:1480 from 100000\n",
      "loss: 0.002218436630926135\n",
      "epoch:1481 from 100000\n",
      "loss: 0.002564935794111989\n",
      "epoch:1482 from 100000\n",
      "loss: 0.0027838779890601018\n",
      "epoch:1483 from 100000\n",
      "loss: 0.0022914832574087995\n",
      "epoch:1484 from 100000\n",
      "loss: 0.0026900826803171185\n",
      "epoch:1485 from 100000\n",
      "loss: 0.002498368155954722\n",
      "epoch:1486 from 100000\n",
      "loss: 0.002994445582885419\n",
      "epoch:1487 from 100000\n",
      "loss: 0.0027427558922568096\n",
      "epoch:1488 from 100000\n",
      "loss: 0.002962536351219569\n",
      "epoch:1489 from 100000\n",
      "loss: 0.0024282474263941367\n",
      "epoch:1490 from 100000\n",
      "loss: 0.0024733005655832585\n",
      "epoch:1491 from 100000\n",
      "loss: 0.002208263929730947\n",
      "epoch:1492 from 100000\n",
      "loss: 0.0022280417152131804\n",
      "epoch:1493 from 100000\n",
      "loss: 0.002054100803121628\n",
      "epoch:1494 from 100000\n",
      "loss: 0.002358838811687747\n",
      "epoch:1495 from 100000\n",
      "loss: 0.0028962960768126322\n",
      "epoch:1496 from 100000\n",
      "loss: 0.0029032472421306466\n",
      "epoch:1497 from 100000\n",
      "loss: 0.0029015887832315457\n",
      "epoch:1498 from 100000\n",
      "loss: 0.0025517616334127524\n",
      "epoch:1499 from 100000\n",
      "loss: 0.0022523847968329846\n",
      "epoch:1500 from 100000\n",
      "loss: 0.0026100967843376566\n",
      "epoch:1501 from 100000\n",
      "loss: 0.002886042997658933\n",
      "epoch:1502 from 100000\n",
      "loss: 0.0022693068744388256\n",
      "epoch:1503 from 100000\n",
      "loss: 0.0024210255521666113\n",
      "epoch:1504 from 100000\n",
      "loss: 0.002685548877337851\n",
      "epoch:1505 from 100000\n",
      "loss: 0.002297900248791507\n",
      "epoch:1506 from 100000\n",
      "loss: 0.0023214135355829058\n",
      "epoch:1507 from 100000\n",
      "loss: 0.0026765405072524118\n",
      "epoch:1508 from 100000\n",
      "loss: 0.002465574844990215\n",
      "epoch:1509 from 100000\n",
      "loss: 0.00235821229927069\n",
      "epoch:1510 from 100000\n",
      "loss: 0.0024270092145570743\n",
      "retrying for data\n",
      "epoch:1511 from 100000\n",
      "loss: 0.0023878518511963875\n",
      "epoch:1512 from 100000\n",
      "loss: 0.0027098344675238195\n",
      "epoch:1513 from 100000\n",
      "loss: 0.0021585487133393144\n",
      "epoch:1514 from 100000\n",
      "loss: 0.002329561964614868\n",
      "epoch:1515 from 100000\n",
      "loss: 0.0021203891593017943\n",
      "epoch:1516 from 100000\n",
      "loss: 0.002257657804932478\n",
      "epoch:1517 from 100000\n",
      "loss: 0.0027720255010308392\n",
      "epoch:1518 from 100000\n",
      "loss: 0.0028294101872461397\n",
      "epoch:1519 from 100000\n",
      "loss: 0.0029279450268404617\n",
      "epoch:1520 from 100000\n",
      "loss: 0.0024941527645125916\n",
      "epoch:1521 from 100000\n",
      "loss: 0.0028205195770851788\n",
      "epoch:1522 from 100000\n",
      "loss: 0.002990250789480342\n",
      "epoch:1523 from 100000\n",
      "loss: 0.0025320426540247354\n",
      "epoch:1524 from 100000\n",
      "loss: 0.002589534804911864\n",
      "epoch:1525 from 100000\n",
      "loss: 0.0021565945065731285\n",
      "epoch:1526 from 100000\n",
      "loss: 0.002019903704706394\n",
      "epoch:1527 from 100000\n",
      "loss: 0.0021800273114962465\n",
      "epoch:1528 from 100000\n",
      "loss: 0.0023528156812915313\n",
      "epoch:1529 from 100000\n",
      "loss: 0.0025029489216876755\n",
      "epoch:1530 from 100000\n",
      "loss: 0.0023011167683080203\n",
      "epoch:1531 from 100000\n",
      "loss: 0.00264569061482689\n",
      "epoch:1532 from 100000\n",
      "loss: 0.0024431739325905255\n",
      "epoch:1533 from 100000\n",
      "loss: 0.0025497782641394906\n",
      "epoch:1534 from 100000\n",
      "loss: 0.002429991739319348\n",
      "epoch:1535 from 100000\n",
      "loss: 0.0019420210848863917\n",
      "epoch:1536 from 100000\n",
      "loss: 0.0024582516186626435\n",
      "epoch:1537 from 100000\n",
      "loss: 0.002316539767747372\n",
      "epoch:1538 from 100000\n",
      "loss: 0.002155920672493039\n",
      "epoch:1539 from 100000\n",
      "loss: 0.0023385406234410766\n",
      "epoch:1540 from 100000\n",
      "loss: 0.002278358583168938\n",
      "epoch:1541 from 100000\n",
      "loss: 0.002795396580694387\n",
      "epoch:1542 from 100000\n",
      "loss: 0.002267011852637779\n",
      "epoch:1543 from 100000\n",
      "loss: 0.0027545632353849214\n",
      "epoch:1544 from 100000\n",
      "loss: 0.0017218682685040676\n",
      "epoch:1545 from 100000\n",
      "loss: 0.0024559596321864774\n",
      "epoch:1546 from 100000\n",
      "loss: 0.0024909570863229646\n",
      "epoch:1547 from 100000\n",
      "loss: 0.0024545715618558006\n",
      "epoch:1548 from 100000\n",
      "loss: 0.002218845977267847\n",
      "epoch:1549 from 100000\n",
      "loss: 0.002300544732946719\n",
      "epoch:1550 from 100000\n",
      "loss: 0.002584426979666432\n",
      "epoch:1551 from 100000\n",
      "loss: 0.0029032550156671277\n",
      "epoch:1552 from 100000\n",
      "loss: 0.0029104758828566446\n",
      "epoch:1553 from 100000\n",
      "loss: 0.002058471128975725\n",
      "epoch:1554 from 100000\n",
      "loss: 0.0022198723402198084\n",
      "epoch:1555 from 100000\n",
      "loss: 0.0026569885399396753\n",
      "epoch:1556 from 100000\n",
      "loss: 0.002259538377956005\n",
      "epoch:1557 from 100000\n",
      "loss: 0.002463878752820392\n",
      "epoch:1558 from 100000\n",
      "loss: 0.0025821869928108754\n",
      "epoch:1559 from 100000\n",
      "loss: 0.0025329389910666578\n",
      "epoch:1560 from 100000\n",
      "loss: 0.00242531943209201\n",
      "epoch:1561 from 100000\n",
      "loss: 0.0027698560820397233\n",
      "epoch:1562 from 100000\n",
      "loss: 0.0025528627722763986\n",
      "epoch:1563 from 100000\n",
      "loss: 0.00220049537210798\n",
      "epoch:1564 from 100000\n",
      "loss: 0.0030038301934496303\n",
      "epoch:1565 from 100000\n",
      "loss: 0.0025728774610342953\n",
      "epoch:1566 from 100000\n",
      "loss: 0.002899642768625199\n",
      "epoch:1567 from 100000\n",
      "loss: 0.0026912098142588547\n",
      "epoch:1568 from 100000\n",
      "loss: 0.0024298777808411387\n",
      "epoch:1569 from 100000\n",
      "loss: 0.002837864241968191\n",
      "epoch:1570 from 100000\n",
      "loss: 0.0026957555940043676\n",
      "epoch:1571 from 100000\n",
      "loss: 0.002519993952802224\n",
      "epoch:1572 from 100000\n",
      "loss: 0.002465744650663737\n",
      "epoch:1573 from 100000\n",
      "loss: 0.0023767218946204594\n",
      "epoch:1574 from 100000\n",
      "loss: 0.002718590456822767\n",
      "epoch:1575 from 100000\n",
      "loss: 0.0023681736731759884\n",
      "epoch:1576 from 100000\n",
      "loss: 0.002705554711582181\n",
      "epoch:1577 from 100000\n",
      "loss: 0.0027443230700896493\n",
      "epoch:1578 from 100000\n",
      "loss: 0.0025904867367216866\n",
      "epoch:1579 from 100000\n",
      "loss: 0.0027606134582782715\n",
      "epoch:1580 from 100000\n",
      "loss: 0.0022265384855302273\n",
      "epoch:1581 from 100000\n",
      "loss: 0.0027653360498050006\n",
      "epoch:1582 from 100000\n",
      "loss: 0.0028533695917474233\n",
      "epoch:1583 from 100000\n",
      "loss: 0.002260467957171386\n",
      "epoch:1584 from 100000\n",
      "loss: 0.002043336716781141\n",
      "epoch:1585 from 100000\n",
      "loss: 0.002714128623111378\n",
      "epoch:1586 from 100000\n",
      "loss: 0.0032153841769968494\n",
      "epoch:1587 from 100000\n",
      "loss: 0.0024378891012304393\n",
      "epoch:1588 from 100000\n",
      "loss: 0.0025811795073593657\n",
      "epoch:1589 from 100000\n",
      "loss: 0.00245197112593587\n",
      "epoch:1590 from 100000\n",
      "loss: 0.002592998758501608\n",
      "epoch:1591 from 100000\n",
      "loss: 0.002467999952301625\n",
      "epoch:1592 from 100000\n",
      "loss: 0.0019910132230620547\n",
      "epoch:1593 from 100000\n",
      "loss: 0.0022293232199501745\n",
      "epoch:1594 from 100000\n",
      "loss: 0.0024290474726740285\n",
      "epoch:1595 from 100000\n",
      "loss: 0.002440900585355621\n",
      "epoch:1596 from 100000\n",
      "loss: 0.0026974141502478233\n",
      "epoch:1597 from 100000\n",
      "loss: 0.0025330359573842998\n",
      "epoch:1598 from 100000\n",
      "loss: 0.00213587068580523\n",
      "epoch:1599 from 100000\n",
      "loss: 0.002230703645977883\n",
      "epoch:1600 from 100000\n",
      "loss: 0.0028282269492478918\n",
      "epoch:1601 from 100000\n",
      "loss: 0.0024594055717841456\n",
      "epoch:1602 from 100000\n",
      "loss: 0.002330765147519287\n",
      "epoch:1603 from 100000\n",
      "loss: 0.0022081512793192815\n",
      "epoch:1604 from 100000\n",
      "loss: 0.0023850784925230073\n",
      "epoch:1605 from 100000\n",
      "loss: 0.0029306869825518334\n",
      "epoch:1606 from 100000\n",
      "loss: 0.002397814165277623\n",
      "epoch:1607 from 100000\n",
      "loss: 0.00275018140663974\n",
      "epoch:1608 from 100000\n",
      "loss: 0.002342228867803442\n",
      "epoch:1609 from 100000\n",
      "loss: 0.0025752808039953834\n",
      "epoch:1610 from 100000\n",
      "loss: 0.0023702438082011668\n",
      "epoch:1611 from 100000\n",
      "loss: 0.0024863683208877774\n",
      "epoch:1612 from 100000\n",
      "loss: 0.00224079300522817\n",
      "epoch:1613 from 100000\n",
      "loss: 0.002878396392787863\n",
      "epoch:1614 from 100000\n",
      "loss: 0.0028284428919391758\n",
      "epoch:1615 from 100000\n",
      "loss: 0.0020367360886552888\n",
      "epoch:1616 from 100000\n",
      "loss: 0.002580500387068696\n",
      "epoch:1617 from 100000\n",
      "loss: 0.002821273024949278\n",
      "epoch:1618 from 100000\n",
      "loss: 0.0029554879809552403\n",
      "epoch:1619 from 100000\n",
      "loss: 0.0028224510446079876\n",
      "epoch:1620 from 100000\n",
      "loss: 0.0023307634838687363\n",
      "epoch:1621 from 100000\n",
      "loss: 0.0023772147769705043\n",
      "epoch:1622 from 100000\n",
      "loss: 0.002442496541050332\n",
      "epoch:1623 from 100000\n",
      "loss: 0.0026766404960483214\n",
      "epoch:1624 from 100000\n",
      "loss: 0.0022200937140155474\n",
      "epoch:1625 from 100000\n",
      "loss: 0.0027622442939616576\n",
      "epoch:1626 from 100000\n",
      "loss: 0.00263762877324325\n",
      "epoch:1627 from 100000\n",
      "loss: 0.0025807311703829328\n",
      "epoch:1628 from 100000\n",
      "loss: 0.002957603262984776\n",
      "epoch:1629 from 100000\n",
      "loss: 0.0027018817160069375\n",
      "epoch:1630 from 100000\n",
      "loss: 0.0024684637687357736\n",
      "epoch:1631 from 100000\n",
      "loss: 0.0022566069386584786\n",
      "epoch:1632 from 100000\n",
      "loss: 0.0021361720502000026\n",
      "epoch:1633 from 100000\n",
      "loss: 0.0026951691087901963\n",
      "epoch:1634 from 100000\n",
      "loss: 0.002279319874361363\n",
      "epoch:1635 from 100000\n",
      "loss: 0.002371280892134564\n",
      "epoch:1636 from 100000\n",
      "loss: 0.0030992430601486376\n",
      "epoch:1637 from 100000\n",
      "loss: 0.0025596620401984183\n",
      "epoch:1638 from 100000\n",
      "loss: 0.0026763948316670394\n",
      "epoch:1639 from 100000\n",
      "loss: 0.002244258948451261\n",
      "epoch:1640 from 100000\n",
      "loss: 0.002309797819918913\n",
      "epoch:1641 from 100000\n",
      "loss: 0.002510069560713646\n",
      "epoch:1642 from 100000\n",
      "loss: 0.0027221624313256143\n",
      "epoch:1643 from 100000\n",
      "loss: 0.0028230407083924547\n",
      "epoch:1644 from 100000\n",
      "loss: 0.002512330740927382\n",
      "epoch:1645 from 100000\n",
      "loss: 0.002240921340217028\n",
      "epoch:1646 from 100000\n",
      "loss: 0.0026121012053579307\n",
      "epoch:1647 from 100000\n",
      "loss: 0.0026125575892876896\n",
      "epoch:1648 from 100000\n",
      "loss: 0.0023223092348416685\n",
      "epoch:1649 from 100000\n",
      "loss: 0.0027330950052828484\n",
      "epoch:1650 from 100000\n",
      "loss: 0.0025905471956377824\n",
      "epoch:1651 from 100000\n",
      "loss: 0.0022038283169365513\n",
      "epoch:1652 from 100000\n",
      "loss: 0.0032341070918846526\n",
      "epoch:1653 from 100000\n",
      "loss: 0.0025771049245122413\n",
      "epoch:1654 from 100000\n",
      "loss: 0.002599057534126814\n",
      "epoch:1655 from 100000\n",
      "loss: 0.0024873331883839\n",
      "epoch:1656 from 100000\n",
      "loss: 0.002197878855270119\n",
      "epoch:1657 from 100000\n",
      "loss: 0.00243702972244364\n",
      "epoch:1658 from 100000\n",
      "loss: 0.0022393354442442615\n",
      "epoch:1659 from 100000\n",
      "loss: 0.002268346728214965\n",
      "epoch:1660 from 100000\n",
      "loss: 0.0025473498525059313\n",
      "epoch:1661 from 100000\n",
      "loss: 0.003442268246431013\n",
      "epoch:1662 from 100000\n",
      "loss: 0.002475595359953786\n",
      "epoch:1663 from 100000\n",
      "loss: 0.0024901979901983395\n",
      "epoch:1664 from 100000\n",
      "loss: 0.002468489969658094\n",
      "epoch:1665 from 100000\n",
      "loss: 0.0018235778560722338\n",
      "epoch:1666 from 100000\n",
      "loss: 0.002016880624211126\n",
      "epoch:1667 from 100000\n",
      "loss: 0.003377883795621983\n",
      "epoch:1668 from 100000\n",
      "loss: 0.002625400382811449\n",
      "epoch:1669 from 100000\n",
      "loss: 0.0026733717300544413\n",
      "retrying for data\n",
      "epoch:1670 from 100000\n",
      "loss: 0.002563938800165033\n",
      "epoch:1671 from 100000\n",
      "loss: 0.0020659662917807964\n",
      "epoch:1672 from 100000\n",
      "loss: 0.002839142018430607\n",
      "epoch:1673 from 100000\n",
      "loss: 0.00279641047652035\n",
      "epoch:1674 from 100000\n",
      "loss: 0.0022566293055774622\n",
      "epoch:1675 from 100000\n",
      "loss: 0.0022532874304204142\n",
      "epoch:1676 from 100000\n",
      "loss: 0.0022406849630840497\n",
      "epoch:1677 from 100000\n",
      "loss: 0.0017451582504008911\n",
      "epoch:1678 from 100000\n",
      "loss: 0.0021364161432302353\n",
      "epoch:1679 from 100000\n",
      "loss: 0.0023414520912865555\n",
      "epoch:1680 from 100000\n",
      "loss: 0.0026222757391138884\n",
      "epoch:1681 from 100000\n",
      "loss: 0.0023113806371100054\n",
      "epoch:1682 from 100000\n",
      "loss: 0.003119705573283227\n",
      "epoch:1683 from 100000\n",
      "loss: 0.0022286314483324077\n",
      "epoch:1684 from 100000\n",
      "loss: 0.0021267115647845003\n",
      "epoch:1685 from 100000\n",
      "loss: 0.002978554319270188\n",
      "epoch:1686 from 100000\n",
      "loss: 0.0024702545182151425\n",
      "epoch:1687 from 100000\n",
      "loss: 0.0026370366153400937\n",
      "epoch:1688 from 100000\n",
      "loss: 0.002358402533559456\n",
      "epoch:1689 from 100000\n",
      "loss: 0.0019858552582263655\n",
      "epoch:1690 from 100000\n",
      "loss: 0.002380976519560818\n",
      "epoch:1691 from 100000\n",
      "loss: 0.0023917840082390285\n",
      "epoch:1692 from 100000\n",
      "loss: 0.0020735913063845146\n",
      "epoch:1693 from 100000\n",
      "loss: 0.0024362380442823905\n",
      "epoch:1694 from 100000\n",
      "loss: 0.0022605116072895726\n",
      "epoch:1695 from 100000\n",
      "loss: 0.002480053285026429\n",
      "epoch:1696 from 100000\n",
      "loss: 0.00288475016236589\n",
      "epoch:1697 from 100000\n",
      "loss: 0.002429704664024257\n",
      "epoch:1698 from 100000\n",
      "loss: 0.002822203508799248\n",
      "epoch:1699 from 100000\n",
      "loss: 0.0023851647162018708\n",
      "epoch:1700 from 100000\n",
      "loss: 0.0025749461418058672\n",
      "epoch:1701 from 100000\n",
      "loss: 0.002352945613040447\n",
      "epoch:1702 from 100000\n",
      "loss: 0.00285712957249018\n",
      "epoch:1703 from 100000\n",
      "loss: 0.002040933205307738\n",
      "epoch:1704 from 100000\n",
      "loss: 0.0025417877590712123\n",
      "epoch:1705 from 100000\n",
      "loss: 0.002409768113039945\n",
      "epoch:1706 from 100000\n",
      "loss: 0.0030130038854849772\n",
      "epoch:1707 from 100000\n",
      "loss: 0.002186611104775693\n",
      "epoch:1708 from 100000\n",
      "loss: 0.0028572181589794354\n",
      "epoch:1709 from 100000\n",
      "loss: 0.0027430039967271114\n",
      "epoch:1710 from 100000\n",
      "loss: 0.002270533865100788\n",
      "epoch:1711 from 100000\n",
      "loss: 0.0024322224943773563\n",
      "epoch:1712 from 100000\n",
      "loss: 0.002472643732346569\n",
      "epoch:1713 from 100000\n",
      "loss: 0.002944643075437625\n",
      "epoch:1714 from 100000\n",
      "loss: 0.0018930858812211682\n",
      "epoch:1715 from 100000\n",
      "loss: 0.0027514053112440706\n",
      "epoch:1716 from 100000\n",
      "loss: 0.002905796211962297\n",
      "epoch:1717 from 100000\n",
      "loss: 0.002160613693064306\n",
      "epoch:1718 from 100000\n",
      "loss: 0.00240581033693843\n",
      "epoch:1719 from 100000\n",
      "loss: 0.0027335559444736646\n",
      "epoch:1720 from 100000\n",
      "loss: 0.0022016915285121286\n",
      "epoch:1721 from 100000\n",
      "loss: 0.0022844119783940187\n",
      "epoch:1722 from 100000\n",
      "loss: 0.0016926603880733637\n",
      "epoch:1723 from 100000\n",
      "loss: 0.0024616850143672764\n",
      "epoch:1724 from 100000\n",
      "loss: 0.0026624208123848803\n",
      "epoch:1725 from 100000\n",
      "loss: 0.002551357883916694\n",
      "epoch:1726 from 100000\n",
      "loss: 0.002283266385660454\n",
      "epoch:1727 from 100000\n",
      "loss: 0.0031079160435751874\n",
      "epoch:1728 from 100000\n",
      "loss: 0.0024910735714911425\n",
      "epoch:1729 from 100000\n",
      "loss: 0.002361679493247948\n",
      "epoch:1730 from 100000\n",
      "loss: 0.002455074249979816\n",
      "epoch:1731 from 100000\n",
      "loss: 0.002055591718658434\n",
      "epoch:1732 from 100000\n",
      "loss: 0.0032164334826632057\n",
      "epoch:1733 from 100000\n",
      "loss: 0.0020595181750309166\n",
      "epoch:1734 from 100000\n",
      "loss: 0.002397864707631925\n",
      "epoch:1735 from 100000\n",
      "loss: 0.0019392593060274521\n",
      "epoch:1736 from 100000\n",
      "loss: 0.0023408763623251616\n",
      "epoch:1737 from 100000\n",
      "loss: 0.0020307400220644922\n",
      "epoch:1738 from 100000\n",
      "loss: 0.002531422754174173\n",
      "epoch:1739 from 100000\n",
      "loss: 0.0022650465649292073\n",
      "epoch:1740 from 100000\n",
      "loss: 0.0028018364076842772\n",
      "epoch:1741 from 100000\n",
      "loss: 0.0031358032425430338\n",
      "epoch:1742 from 100000\n",
      "loss: 0.0020115180404332023\n",
      "epoch:1743 from 100000\n",
      "loss: 0.0022538982849482636\n",
      "epoch:1744 from 100000\n",
      "loss: 0.003012527834755474\n",
      "epoch:1745 from 100000\n",
      "loss: 0.0026865565343996423\n",
      "epoch:1746 from 100000\n",
      "loss: 0.0025821268432082434\n",
      "epoch:1747 from 100000\n",
      "loss: 0.002067483996398778\n",
      "epoch:1748 from 100000\n",
      "loss: 0.0027406785078483153\n",
      "epoch:1749 from 100000\n",
      "loss: 0.0025709368557329526\n",
      "epoch:1750 from 100000\n",
      "loss: 0.0023449468163221354\n",
      "epoch:1751 from 100000\n",
      "loss: 0.00236284587145974\n",
      "epoch:1752 from 100000\n",
      "loss: 0.002481795903293005\n",
      "epoch:1753 from 100000\n",
      "loss: 0.0023335704060514217\n",
      "epoch:1754 from 100000\n",
      "loss: 0.00217619411847636\n",
      "epoch:1755 from 100000\n",
      "loss: 0.0028213497936349086\n",
      "epoch:1756 from 100000\n",
      "loss: 0.00211324738270946\n",
      "epoch:1757 from 100000\n",
      "loss: 0.002516838375356656\n",
      "epoch:1758 from 100000\n",
      "loss: 0.0022573603222610927\n",
      "epoch:1759 from 100000\n",
      "loss: 0.0017393100220033375\n",
      "epoch:1760 from 100000\n",
      "loss: 0.0027500696559172866\n",
      "epoch:1761 from 100000\n",
      "loss: 0.002420214639812457\n",
      "epoch:1762 from 100000\n",
      "loss: 0.0023965726651340447\n",
      "epoch:1763 from 100000\n",
      "loss: 0.0020627202414402745\n",
      "epoch:1764 from 100000\n",
      "loss: 0.002247211301181551\n",
      "epoch:1765 from 100000\n",
      "loss: 0.0024059037968413577\n",
      "epoch:1766 from 100000\n",
      "loss: 0.0028304842817874487\n",
      "epoch:1767 from 100000\n",
      "loss: 0.002475008683319402\n",
      "epoch:1768 from 100000\n",
      "loss: 0.0023741677269697448\n",
      "epoch:1769 from 100000\n",
      "loss: 0.0021554939797283623\n",
      "epoch:1770 from 100000\n",
      "loss: 0.002565996255725622\n",
      "epoch:1771 from 100000\n",
      "loss: 0.0021503587670395063\n",
      "epoch:1772 from 100000\n",
      "loss: 0.002775589266548195\n",
      "epoch:1773 from 100000\n",
      "loss: 0.002622025677737838\n",
      "epoch:1774 from 100000\n",
      "loss: 0.002777909641537235\n",
      "epoch:1775 from 100000\n",
      "loss: 0.002439558191326796\n",
      "epoch:1776 from 100000\n",
      "loss: 0.0026003086531716235\n",
      "epoch:1777 from 100000\n",
      "loss: 0.0022545131687650155\n",
      "epoch:1778 from 100000\n",
      "loss: 0.002563861199234907\n",
      "epoch:1779 from 100000\n",
      "loss: 0.002526092503956079\n",
      "epoch:1780 from 100000\n",
      "loss: 0.002916709287632102\n",
      "epoch:1781 from 100000\n",
      "loss: 0.002878287204453045\n",
      "epoch:1782 from 100000\n",
      "loss: 0.002379589757225631\n",
      "epoch:1783 from 100000\n",
      "loss: 0.0025431287536719083\n",
      "epoch:1784 from 100000\n",
      "loss: 0.0025339513161384275\n",
      "epoch:1785 from 100000\n",
      "loss: 0.002567600424953298\n",
      "epoch:1786 from 100000\n",
      "loss: 0.0028341194878578335\n",
      "epoch:1787 from 100000\n",
      "loss: 0.0023959644728677176\n",
      "epoch:1788 from 100000\n",
      "loss: 0.002562184606944129\n",
      "epoch:1789 from 100000\n",
      "loss: 0.0028287406874767385\n",
      "epoch:1790 from 100000\n",
      "loss: 0.0030071354014182816\n",
      "epoch:1791 from 100000\n",
      "loss: 0.002952877459136971\n",
      "epoch:1792 from 100000\n",
      "loss: 0.0024655199806886685\n",
      "epoch:1793 from 100000\n",
      "loss: 0.0028489380310929846\n",
      "epoch:1794 from 100000\n",
      "loss: 0.0023554634763343074\n",
      "epoch:1795 from 100000\n",
      "loss: 0.001978487678371721\n",
      "epoch:1796 from 100000\n",
      "loss: 0.0023650379947071087\n",
      "epoch:1797 from 100000\n",
      "loss: 0.0022812227395263562\n",
      "epoch:1798 from 100000\n",
      "loss: 0.0024732823737139142\n",
      "epoch:1799 from 100000\n",
      "loss: 0.00302847401951567\n",
      "epoch:1800 from 100000\n",
      "loss: 0.0020459783150954536\n",
      "epoch:1801 from 100000\n",
      "loss: 0.002551550201133068\n",
      "epoch:1802 from 100000\n",
      "loss: 0.0023760897608440246\n",
      "epoch:1803 from 100000\n",
      "loss: 0.002897459549217274\n",
      "epoch:1804 from 100000\n",
      "loss: 0.0023995501200033686\n",
      "epoch:1805 from 100000\n",
      "loss: 0.002416366324041519\n",
      "epoch:1806 from 100000\n",
      "loss: 0.0031210744156737746\n",
      "epoch:1807 from 100000\n",
      "loss: 0.0020833268391982074\n",
      "epoch:1808 from 100000\n",
      "loss: 0.0028210376778190493\n",
      "epoch:1809 from 100000\n",
      "loss: 0.001928140189562555\n",
      "epoch:1810 from 100000\n",
      "loss: 0.003049905193449831\n",
      "epoch:1811 from 100000\n",
      "loss: 0.0023055882489870783\n",
      "epoch:1812 from 100000\n",
      "loss: 0.0021373226643959242\n",
      "epoch:1813 from 100000\n",
      "loss: 0.0026234505149602683\n",
      "epoch:1814 from 100000\n",
      "loss: 0.0027568543669644896\n",
      "epoch:1815 from 100000\n",
      "loss: 0.002407955509227122\n",
      "epoch:1816 from 100000\n",
      "loss: 0.0025863301766975155\n",
      "epoch:1817 from 100000\n",
      "loss: 0.00266317157226581\n",
      "epoch:1818 from 100000\n",
      "loss: 0.0023871618780333392\n",
      "epoch:1819 from 100000\n",
      "loss: 0.002656924274958783\n",
      "epoch:1820 from 100000\n",
      "loss: 0.003219146962578634\n",
      "epoch:1821 from 100000\n",
      "loss: 0.0024990432412721475\n",
      "epoch:1822 from 100000\n",
      "loss: 0.002166060363009592\n",
      "epoch:1823 from 100000\n",
      "loss: 0.0023940961962978236\n",
      "epoch:1824 from 100000\n",
      "loss: 0.002801918402838055\n",
      "epoch:1825 from 100000\n",
      "loss: 0.0022338989804779885\n",
      "retrying for data\n",
      "epoch:1826 from 100000\n",
      "loss: 0.002510755475583437\n",
      "epoch:1827 from 100000\n",
      "loss: 0.002363881323788064\n",
      "epoch:1828 from 100000\n",
      "loss: 0.0027481889483169653\n",
      "epoch:1829 from 100000\n",
      "loss: 0.0025941407886591605\n",
      "epoch:1830 from 100000\n",
      "loss: 0.0024874931057468075\n",
      "epoch:1831 from 100000\n",
      "loss: 0.002539515146281701\n",
      "epoch:1832 from 100000\n",
      "loss: 0.0023956463783747495\n",
      "epoch:1833 from 100000\n",
      "loss: 0.0023746575006384774\n",
      "epoch:1834 from 100000\n",
      "loss: 0.0020595322216081513\n",
      "epoch:1835 from 100000\n",
      "loss: 0.003014993901899743\n",
      "epoch:1836 from 100000\n",
      "loss: 0.0025330952291255926\n",
      "epoch:1837 from 100000\n",
      "loss: 0.0024799208457011446\n",
      "epoch:1838 from 100000\n",
      "loss: 0.0029366974756328545\n",
      "epoch:1839 from 100000\n",
      "loss: 0.0028220154426037425\n",
      "epoch:1840 from 100000\n",
      "loss: 0.002595384765641029\n",
      "epoch:1841 from 100000\n",
      "loss: 0.0023840028220547538\n",
      "epoch:1842 from 100000\n",
      "loss: 0.0028160265707697363\n",
      "epoch:1843 from 100000\n",
      "loss: 0.002715847639308322\n",
      "epoch:1844 from 100000\n",
      "loss: 0.00224723236468094\n",
      "epoch:1845 from 100000\n",
      "loss: 0.0019004753668241392\n",
      "epoch:1846 from 100000\n",
      "loss: 0.0024235044878651024\n",
      "epoch:1847 from 100000\n",
      "loss: 0.0029869547194891766\n",
      "epoch:1848 from 100000\n",
      "loss: 0.002281945305853128\n",
      "epoch:1849 from 100000\n",
      "loss: 0.0026989497706608745\n",
      "epoch:1850 from 100000\n",
      "loss: 0.0027294090261307247\n",
      "epoch:1851 from 100000\n",
      "loss: 0.0019429610043175671\n",
      "epoch:1852 from 100000\n",
      "loss: 0.002549989552505849\n",
      "epoch:1853 from 100000\n",
      "loss: 0.003024689141454928\n",
      "epoch:1854 from 100000\n",
      "loss: 0.0025887268700159893\n",
      "epoch:1855 from 100000\n",
      "loss: 0.002781855106718467\n",
      "epoch:1856 from 100000\n",
      "loss: 0.002510543231949214\n",
      "epoch:1857 from 100000\n",
      "loss: 0.0022382848018196455\n",
      "retrying for data\n",
      "epoch:1858 from 100000\n",
      "loss: 0.0021044452422529503\n",
      "epoch:1859 from 100000\n",
      "loss: 0.0018306351497017204\n",
      "epoch:1860 from 100000\n",
      "loss: 0.002476007002314873\n",
      "epoch:1861 from 100000\n",
      "loss: 0.0028143997932659204\n",
      "epoch:1862 from 100000\n",
      "loss: 0.002535270924198585\n",
      "epoch:1863 from 100000\n",
      "loss: 0.003100797708668779\n",
      "epoch:1864 from 100000\n",
      "loss: 0.002845015440982479\n",
      "epoch:1865 from 100000\n",
      "loss: 0.0030850405613023213\n",
      "epoch:1866 from 100000\n",
      "loss: 0.0023004507106634264\n",
      "epoch:1867 from 100000\n",
      "loss: 0.0021225229689321168\n",
      "epoch:1868 from 100000\n",
      "loss: 0.002640260040919884\n",
      "epoch:1869 from 100000\n",
      "loss: 0.0022086568237682513\n",
      "epoch:1870 from 100000\n",
      "loss: 0.002461391200910157\n",
      "epoch:1871 from 100000\n",
      "loss: 0.002571143758245853\n",
      "epoch:1872 from 100000\n",
      "loss: 0.002060905624659881\n",
      "epoch:1873 from 100000\n",
      "loss: 0.002985055280191773\n",
      "epoch:1874 from 100000\n",
      "loss: 0.002539743121644733\n",
      "epoch:1875 from 100000\n",
      "loss: 0.0020379964982453203\n",
      "epoch:1876 from 100000\n",
      "loss: 0.002522690908790537\n",
      "retrying for data\n",
      "epoch:1877 from 100000\n",
      "loss: 0.002457123577613629\n",
      "epoch:1878 from 100000\n",
      "loss: 0.001961753150766299\n",
      "epoch:1879 from 100000\n",
      "loss: 0.0027494593375934073\n",
      "epoch:1880 from 100000\n",
      "loss: 0.002678299556549746\n",
      "epoch:1881 from 100000\n",
      "loss: 0.0027452709193767078\n",
      "epoch:1882 from 100000\n",
      "loss: 0.0023108948275876173\n",
      "epoch:1883 from 100000\n",
      "loss: 0.0027981523137867725\n",
      "epoch:1884 from 100000\n",
      "loss: 0.0022523104475453692\n",
      "epoch:1885 from 100000\n",
      "loss: 0.0022001707601475573\n",
      "epoch:1886 from 100000\n",
      "loss: 0.0024043714601305055\n",
      "epoch:1887 from 100000\n",
      "loss: 0.0023569203698485808\n",
      "epoch:1888 from 100000\n",
      "loss: 0.0027576278808396637\n",
      "epoch:1889 from 100000\n",
      "loss: 0.0022759133820784427\n",
      "epoch:1890 from 100000\n",
      "loss: 0.0029103955908453827\n",
      "epoch:1891 from 100000\n",
      "loss: 0.0025574745203584826\n",
      "epoch:1892 from 100000\n",
      "loss: 0.0022829283111889254\n",
      "epoch:1893 from 100000\n",
      "loss: 0.002262031562594302\n",
      "epoch:1894 from 100000\n",
      "loss: 0.0022710519703252885\n",
      "epoch:1895 from 100000\n",
      "loss: 0.00235071302327583\n",
      "epoch:1896 from 100000\n",
      "loss: 0.0028367809818519163\n",
      "epoch:1897 from 100000\n",
      "loss: 0.00319456799311979\n",
      "epoch:1898 from 100000\n",
      "loss: 0.0022934425518599255\n",
      "epoch:1899 from 100000\n",
      "loss: 0.00254435460857394\n",
      "epoch:1900 from 100000\n",
      "loss: 0.0023620618359814216\n",
      "epoch:1901 from 100000\n",
      "loss: 0.0020407858424391634\n",
      "epoch:1902 from 100000\n",
      "loss: 0.0021916594134836487\n",
      "epoch:1903 from 100000\n",
      "loss: 0.002492333957917481\n",
      "epoch:1904 from 100000\n",
      "loss: 0.002115302505899308\n",
      "epoch:1905 from 100000\n",
      "loss: 0.0023568978532466645\n",
      "epoch:1906 from 100000\n",
      "loss: 0.0025264665273425635\n",
      "epoch:1907 from 100000\n",
      "loss: 0.0024388447912997435\n",
      "epoch:1908 from 100000\n",
      "loss: 0.0017816419655218851\n",
      "epoch:1909 from 100000\n",
      "loss: 0.0021874641969645836\n",
      "epoch:1910 from 100000\n",
      "loss: 0.0027763284097375163\n",
      "epoch:1911 from 100000\n",
      "loss: 0.0021833500588854804\n",
      "epoch:1912 from 100000\n",
      "loss: 0.0025546832484337756\n",
      "epoch:1913 from 100000\n",
      "loss: 0.0029551493910986437\n",
      "epoch:1914 from 100000\n",
      "loss: 0.0024024684087180503\n",
      "epoch:1915 from 100000\n",
      "loss: 0.0027384186281551592\n",
      "epoch:1916 from 100000\n",
      "loss: 0.0021745229299483526\n",
      "epoch:1917 from 100000\n",
      "loss: 0.002666279160422391\n",
      "epoch:1918 from 100000\n",
      "loss: 0.0022646111287087933\n",
      "epoch:1919 from 100000\n",
      "loss: 0.0031922573864449078\n",
      "epoch:1920 from 100000\n",
      "loss: 0.0025144885981660536\n",
      "epoch:1921 from 100000\n",
      "loss: 0.0025798040092865904\n",
      "epoch:1922 from 100000\n",
      "loss: 0.0019657141891116225\n",
      "epoch:1923 from 100000\n",
      "loss: 0.0026881585262401586\n",
      "epoch:1924 from 100000\n",
      "loss: 0.002095708300203114\n",
      "epoch:1925 from 100000\n",
      "loss: 0.0021719637983892426\n",
      "epoch:1926 from 100000\n",
      "loss: 0.0021437026388468894\n",
      "epoch:1927 from 100000\n",
      "loss: 0.003076328419751917\n",
      "epoch:1928 from 100000\n",
      "loss: 0.002777273248085521\n",
      "epoch:1929 from 100000\n",
      "loss: 0.002949243629046805\n",
      "epoch:1930 from 100000\n",
      "loss: 0.0022912688400253955\n",
      "epoch:1931 from 100000\n",
      "loss: 0.002689077683811547\n",
      "epoch:1932 from 100000\n",
      "loss: 0.0026884437610021905\n",
      "epoch:1933 from 100000\n",
      "loss: 0.002563849842857735\n",
      "epoch:1934 from 100000\n",
      "loss: 0.002101849047789983\n",
      "epoch:1935 from 100000\n",
      "loss: 0.0028389207874681688\n",
      "epoch:1936 from 100000\n",
      "loss: 0.002663913460708045\n",
      "epoch:1937 from 100000\n",
      "loss: 0.0027368629186241833\n",
      "epoch:1938 from 100000\n",
      "loss: 0.0021351912378122506\n",
      "epoch:1939 from 100000\n",
      "loss: 0.002569460976900473\n",
      "epoch:1940 from 100000\n",
      "loss: 0.0026094601163322295\n",
      "epoch:1941 from 100000\n",
      "loss: 0.0021893374870813886\n",
      "epoch:1942 from 100000\n",
      "loss: 0.00202696484019782\n",
      "epoch:1943 from 100000\n",
      "loss: 0.002371210288771408\n",
      "epoch:1944 from 100000\n",
      "loss: 0.0027534556083423922\n",
      "epoch:1945 from 100000\n",
      "loss: 0.0021550213297700793\n",
      "epoch:1946 from 100000\n",
      "loss: 0.00229749375895949\n",
      "epoch:1947 from 100000\n",
      "loss: 0.002412335251733566\n",
      "epoch:1948 from 100000\n",
      "loss: 0.002424355502171238\n",
      "epoch:1949 from 100000\n",
      "loss: 0.002054628656821933\n",
      "epoch:1950 from 100000\n",
      "loss: 0.002111551129246436\n",
      "epoch:1951 from 100000\n",
      "loss: 0.002611981781939221\n",
      "epoch:1952 from 100000\n",
      "loss: 0.0023915401701231076\n",
      "epoch:1953 from 100000\n",
      "loss: 0.002123744918861803\n",
      "epoch:1954 from 100000\n",
      "loss: 0.002662520906142163\n",
      "epoch:1955 from 100000\n",
      "loss: 0.002559197256061907\n",
      "epoch:1956 from 100000\n",
      "loss: 0.0021275999659025047\n",
      "epoch:1957 from 100000\n",
      "loss: 0.0026136144527697525\n",
      "epoch:1958 from 100000\n",
      "loss: 0.0023822662062684685\n",
      "epoch:1959 from 100000\n",
      "loss: 0.0022478264154983663\n",
      "epoch:1960 from 100000\n",
      "loss: 0.002776362834936208\n",
      "epoch:1961 from 100000\n",
      "loss: 0.0022655956947801315\n",
      "retrying for data\n",
      "epoch:1962 from 100000\n",
      "loss: 0.0021439432690328886\n",
      "epoch:1963 from 100000\n",
      "loss: 0.002390159966623173\n",
      "epoch:1964 from 100000\n",
      "loss: 0.002652742677213382\n",
      "epoch:1965 from 100000\n",
      "loss: 0.0024934608877913433\n",
      "epoch:1966 from 100000\n",
      "loss: 0.0027283738875780728\n",
      "epoch:1967 from 100000\n",
      "loss: 0.002514845131287302\n",
      "epoch:1968 from 100000\n",
      "loss: 0.002646737438482205\n",
      "epoch:1969 from 100000\n",
      "loss: 0.003100642022076272\n",
      "epoch:1970 from 100000\n",
      "loss: 0.002603599446501903\n",
      "epoch:1971 from 100000\n",
      "loss: 0.0024906814410314837\n",
      "epoch:1972 from 100000\n",
      "loss: 0.0024599625846519757\n",
      "retrying for data\n",
      "epoch:1973 from 100000\n",
      "loss: 0.0028503076703714214\n",
      "epoch:1974 from 100000\n",
      "loss: 0.002406594115598182\n",
      "epoch:1975 from 100000\n",
      "loss: 0.0018272712260625212\n",
      "epoch:1976 from 100000\n",
      "loss: 0.002663046153998039\n",
      "epoch:1977 from 100000\n",
      "loss: 0.0026025609291195906\n",
      "epoch:1978 from 100000\n",
      "loss: 0.0024964033293031207\n",
      "epoch:1979 from 100000\n",
      "loss: 0.0026439234590611704\n",
      "epoch:1980 from 100000\n",
      "loss: 0.002487070444260553\n",
      "epoch:1981 from 100000\n",
      "loss: 0.002844519808007817\n",
      "epoch:1982 from 100000\n",
      "loss: 0.0026117498415487717\n",
      "epoch:1983 from 100000\n",
      "loss: 0.0023713832489420383\n",
      "epoch:1984 from 100000\n",
      "loss: 0.0025942464536683474\n",
      "epoch:1985 from 100000\n",
      "loss: 0.0026778752708480624\n",
      "epoch:1986 from 100000\n",
      "loss: 0.0023087098579708254\n",
      "epoch:1987 from 100000\n",
      "loss: 0.0026224959808018866\n",
      "epoch:1988 from 100000\n",
      "loss: 0.0023134133114695032\n",
      "epoch:1989 from 100000\n",
      "loss: 0.0019448759938995863\n",
      "epoch:1990 from 100000\n",
      "loss: 0.00240261864388458\n",
      "epoch:1991 from 100000\n",
      "loss: 0.0023663172922852027\n",
      "epoch:1992 from 100000\n",
      "loss: 0.0025095520452680375\n",
      "retrying for data\n",
      "epoch:1993 from 100000\n",
      "loss: 0.0023569333104518364\n",
      "epoch:1994 from 100000\n",
      "loss: 0.0021219470602034107\n",
      "epoch:1995 from 100000\n",
      "loss: 0.0024189977897890458\n",
      "epoch:1996 from 100000\n",
      "loss: 0.0023503707495251547\n",
      "epoch:1997 from 100000\n",
      "loss: 0.002723001576413253\n",
      "epoch:1998 from 100000\n",
      "loss: 0.0026813618996470723\n",
      "epoch:1999 from 100000\n",
      "loss: 0.0027821082960883814\n",
      "epoch:2000 from 100000\n",
      "loss: 0.002715672893714327\n",
      "epoch:2001 from 100000\n",
      "loss: 0.0025162550921038473\n",
      "epoch:2002 from 100000\n",
      "loss: 0.0024644113729266337\n",
      "epoch:2003 from 100000\n",
      "loss: 0.0021940280653325317\n",
      "epoch:2004 from 100000\n",
      "loss: 0.002091642457855869\n",
      "retrying for data\n",
      "epoch:2005 from 100000\n",
      "loss: 0.0029522912925301625\n",
      "epoch:2006 from 100000\n",
      "loss: 0.0022384742638905664\n",
      "epoch:2007 from 100000\n",
      "loss: 0.0028225713242591155\n",
      "epoch:2008 from 100000\n",
      "loss: 0.0027231246974110945\n",
      "epoch:2009 from 100000\n",
      "loss: 0.0028707950871762478\n",
      "epoch:2010 from 100000\n",
      "loss: 0.0029050784924606887\n",
      "epoch:2011 from 100000\n",
      "loss: 0.0026221644772874697\n",
      "epoch:2012 from 100000\n",
      "loss: 0.002408662311694343\n",
      "epoch:2013 from 100000\n",
      "loss: 0.0027430637071574893\n",
      "epoch:2014 from 100000\n",
      "loss: 0.0026892147923831544\n",
      "epoch:2015 from 100000\n",
      "loss: 0.0024110983207776826\n",
      "epoch:2016 from 100000\n",
      "loss: 0.002382344914053647\n",
      "epoch:2017 from 100000\n",
      "loss: 0.002575316238363712\n",
      "epoch:2018 from 100000\n",
      "loss: 0.0022955917838487494\n",
      "epoch:2019 from 100000\n",
      "loss: 0.0026270977398326067\n",
      "epoch:2020 from 100000\n",
      "loss: 0.002238383750864159\n",
      "epoch:2021 from 100000\n",
      "loss: 0.002669728283706263\n",
      "epoch:2022 from 100000\n",
      "loss: 0.002505696625547671\n",
      "epoch:2023 from 100000\n",
      "loss: 0.002923808827006269\n",
      "epoch:2024 from 100000\n",
      "loss: 0.0020927161342001455\n",
      "epoch:2025 from 100000\n",
      "loss: 0.002440665373626416\n",
      "epoch:2026 from 100000\n",
      "loss: 0.0024995568311965144\n",
      "epoch:2027 from 100000\n",
      "loss: 0.0021354348071440654\n",
      "epoch:2028 from 100000\n",
      "loss: 0.002800948359947597\n",
      "epoch:2029 from 100000\n",
      "loss: 0.003346412379244157\n",
      "epoch:2030 from 100000\n",
      "loss: 0.0021728413401831403\n",
      "epoch:2031 from 100000\n",
      "loss: 0.0029214667110863957\n",
      "epoch:2032 from 100000\n",
      "loss: 0.0022627782506106087\n",
      "epoch:2033 from 100000\n",
      "loss: 0.002233299636486663\n",
      "epoch:2034 from 100000\n",
      "loss: 0.0033981011263790606\n",
      "epoch:2035 from 100000\n",
      "loss: 0.00232633213789768\n",
      "epoch:2036 from 100000\n",
      "loss: 0.002277428128735437\n",
      "epoch:2037 from 100000\n",
      "loss: 0.002606727179511381\n",
      "epoch:2038 from 100000\n",
      "loss: 0.0027383243625536124\n",
      "epoch:2039 from 100000\n",
      "loss: 0.003061431689758365\n",
      "epoch:2040 from 100000\n",
      "loss: 0.002080226378836869\n",
      "epoch:2041 from 100000\n",
      "loss: 0.0023505814138218284\n",
      "epoch:2042 from 100000\n",
      "loss: 0.0024537897423044797\n",
      "epoch:2043 from 100000\n",
      "loss: 0.0025801806269214467\n",
      "epoch:2044 from 100000\n",
      "loss: 0.0024126387440901453\n",
      "epoch:2045 from 100000\n",
      "loss: 0.0022862758495421076\n",
      "epoch:2046 from 100000\n",
      "loss: 0.0022811413821841597\n",
      "epoch:2047 from 100000\n",
      "loss: 0.0025330435643695637\n",
      "epoch:2048 from 100000\n",
      "loss: 0.0024127221039123015\n",
      "epoch:2049 from 100000\n",
      "loss: 0.002775506090813451\n",
      "epoch:2050 from 100000\n",
      "loss: 0.0030244982597196213\n",
      "epoch:2051 from 100000\n",
      "loss: 0.002507227250447386\n",
      "epoch:2052 from 100000\n",
      "loss: 0.002265738482606139\n",
      "epoch:2053 from 100000\n",
      "loss: 0.0022230287486308953\n",
      "epoch:2054 from 100000\n",
      "loss: 0.0029157583199435066\n",
      "epoch:2055 from 100000\n",
      "loss: 0.002497810498880426\n",
      "epoch:2056 from 100000\n",
      "loss: 0.0027968080248115257\n",
      "epoch:2057 from 100000\n",
      "loss: 0.0023928591808726196\n",
      "epoch:2058 from 100000\n",
      "loss: 0.002863922545998321\n",
      "epoch:2059 from 100000\n",
      "loss: 0.003019266412337629\n",
      "epoch:2060 from 100000\n",
      "loss: 0.0019477147374402648\n",
      "epoch:2061 from 100000\n",
      "loss: 0.0026843915276799635\n",
      "epoch:2062 from 100000\n",
      "loss: 0.0025406558395388856\n",
      "epoch:2063 from 100000\n",
      "loss: 0.002812644501716477\n",
      "epoch:2064 from 100000\n",
      "loss: 0.0026029896975501288\n",
      "epoch:2065 from 100000\n",
      "loss: 0.001909685399141381\n",
      "epoch:2066 from 100000\n",
      "loss: 0.002333857727421673\n",
      "epoch:2067 from 100000\n",
      "loss: 0.0023765233962933507\n",
      "epoch:2068 from 100000\n",
      "loss: 0.00264290665498379\n",
      "epoch:2069 from 100000\n",
      "loss: 0.0021738182684742924\n",
      "epoch:2070 from 100000\n",
      "loss: 0.002734944598500988\n",
      "epoch:2071 from 100000\n",
      "loss: 0.0023166752336010177\n",
      "epoch:2072 from 100000\n",
      "loss: 0.00221790278240519\n",
      "epoch:2073 from 100000\n",
      "loss: 0.0018619730526694411\n",
      "epoch:2074 from 100000\n",
      "loss: 0.002868895970777885\n",
      "epoch:2075 from 100000\n",
      "loss: 0.0025277952423437\n",
      "epoch:2076 from 100000\n",
      "loss: 0.0028568318529664793\n",
      "epoch:2077 from 100000\n",
      "loss: 0.0029782666456981133\n",
      "epoch:2078 from 100000\n",
      "loss: 0.0028123025921615863\n",
      "epoch:2079 from 100000\n",
      "loss: 0.0016565115380444695\n",
      "epoch:2080 from 100000\n",
      "loss: 0.0026928499943039697\n",
      "epoch:2081 from 100000\n",
      "loss: 0.002461264693749854\n",
      "epoch:2082 from 100000\n",
      "loss: 0.002157843937936832\n",
      "epoch:2083 from 100000\n",
      "loss: 0.0024183551822716254\n",
      "epoch:2084 from 100000\n",
      "loss: 0.0022379234086997712\n",
      "epoch:2085 from 100000\n",
      "loss: 0.002420230146611857\n",
      "epoch:2086 from 100000\n",
      "loss: 0.002507879887502895\n",
      "epoch:2087 from 100000\n",
      "loss: 0.001728402374268967\n",
      "epoch:2088 from 100000\n",
      "loss: 0.0023360956287206136\n",
      "epoch:2089 from 100000\n",
      "loss: 0.002927024271485834\n",
      "epoch:2090 from 100000\n",
      "loss: 0.003164767970815774\n",
      "epoch:2091 from 100000\n",
      "loss: 0.0025484235211479245\n",
      "epoch:2092 from 100000\n",
      "loss: 0.002324134799323474\n",
      "retrying for data\n",
      "retrying for data\n",
      "epoch:2093 from 100000\n",
      "loss: 0.0026876859236324435\n",
      "epoch:2094 from 100000\n",
      "loss: 0.0026887347138426776\n",
      "epoch:2095 from 100000\n",
      "loss: 0.002567625387598582\n",
      "epoch:2096 from 100000\n",
      "loss: 0.0031841620536852133\n",
      "epoch:2097 from 100000\n",
      "loss: 0.002622951428918441\n",
      "epoch:2098 from 100000\n",
      "loss: 0.0027636203526526515\n",
      "epoch:2099 from 100000\n",
      "loss: 0.002160579632942472\n",
      "epoch:2100 from 100000\n",
      "loss: 0.0023706236307816653\n",
      "epoch:2101 from 100000\n",
      "loss: 0.0029747042072756358\n",
      "epoch:2102 from 100000\n",
      "loss: 0.002424432330428772\n",
      "epoch:2103 from 100000\n",
      "loss: 0.002559393326237114\n",
      "epoch:2104 from 100000\n",
      "loss: 0.002484195057547822\n",
      "epoch:2105 from 100000\n",
      "loss: 0.0022395414526954482\n",
      "epoch:2106 from 100000\n",
      "loss: 0.0029249804978803695\n",
      "epoch:2107 from 100000\n",
      "loss: 0.0023137069430276824\n",
      "epoch:2108 from 100000\n",
      "loss: 0.002602747426777796\n",
      "epoch:2109 from 100000\n",
      "loss: 0.0024142943874920775\n",
      "epoch:2110 from 100000\n",
      "loss: 0.002288136953808362\n",
      "epoch:2111 from 100000\n",
      "loss: 0.0032511480315093877\n",
      "epoch:2112 from 100000\n",
      "loss: 0.0023465369928885593\n",
      "epoch:2113 from 100000\n",
      "loss: 0.0023643468790908173\n",
      "epoch:2114 from 100000\n",
      "loss: 0.002689200237739442\n",
      "epoch:2115 from 100000\n",
      "loss: 0.0028625107632933577\n",
      "epoch:2116 from 100000\n",
      "loss: 0.002396396714061666\n",
      "epoch:2117 from 100000\n",
      "loss: 0.002717420086611355\n",
      "epoch:2118 from 100000\n",
      "loss: 0.002930522843030303\n",
      "epoch:2119 from 100000\n",
      "loss: 0.002534518667573593\n",
      "epoch:2120 from 100000\n",
      "loss: 0.00222921482276206\n",
      "epoch:2121 from 100000\n",
      "loss: 0.0023822234117574226\n",
      "epoch:2122 from 100000\n",
      "loss: 0.002151261984863595\n",
      "epoch:2123 from 100000\n",
      "loss: 0.002195150799309431\n",
      "epoch:2124 from 100000\n",
      "loss: 0.0025114690404279827\n",
      "epoch:2125 from 100000\n",
      "loss: 0.002558414710733814\n",
      "epoch:2126 from 100000\n",
      "loss: 0.0023907433580916404\n",
      "epoch:2127 from 100000\n",
      "loss: 0.002271426635587659\n",
      "epoch:2128 from 100000\n",
      "loss: 0.00271490472036362\n",
      "epoch:2129 from 100000\n",
      "loss: 0.0023554277832147363\n",
      "epoch:2130 from 100000\n",
      "loss: 0.0027798754790921976\n",
      "epoch:2131 from 100000\n",
      "loss: 0.0016410183331743156\n",
      "epoch:2132 from 100000\n",
      "loss: 0.002243350037119285\n",
      "epoch:2133 from 100000\n",
      "loss: 0.002626668122161391\n",
      "epoch:2134 from 100000\n",
      "loss: 0.0023720752381564125\n",
      "epoch:2135 from 100000\n",
      "loss: 0.0028351117907163825\n",
      "epoch:2136 from 100000\n",
      "loss: 0.0025711750000994016\n",
      "epoch:2137 from 100000\n",
      "loss: 0.002978180883673076\n",
      "epoch:2138 from 100000\n",
      "loss: 0.0024058496710210875\n",
      "epoch:2139 from 100000\n",
      "loss: 0.002920687681466916\n",
      "epoch:2140 from 100000\n",
      "loss: 0.002758538429247892\n",
      "epoch:2141 from 100000\n",
      "loss: 0.0026845941754345404\n",
      "epoch:2142 from 100000\n",
      "loss: 0.0027264654288501333\n",
      "epoch:2143 from 100000\n",
      "loss: 0.002346093921460124\n",
      "epoch:2144 from 100000\n",
      "loss: 0.0026029770474451652\n",
      "epoch:2145 from 100000\n",
      "loss: 0.002449614946215206\n",
      "epoch:2146 from 100000\n",
      "loss: 0.002413770459867237\n",
      "epoch:2147 from 100000\n",
      "loss: 0.0025605321187640584\n",
      "epoch:2148 from 100000\n",
      "loss: 0.0025099971250597264\n",
      "epoch:2149 from 100000\n",
      "loss: 0.002485933399299256\n",
      "epoch:2150 from 100000\n",
      "loss: 0.002759675874074219\n",
      "epoch:2151 from 100000\n",
      "loss: 0.0024425757850963237\n",
      "epoch:2152 from 100000\n",
      "loss: 0.002624780218070555\n",
      "epoch:2153 from 100000\n",
      "loss: 0.0023713294000060614\n",
      "epoch:2154 from 100000\n",
      "loss: 0.002661921304138559\n",
      "epoch:2155 from 100000\n",
      "loss: 0.0017403601626426735\n",
      "epoch:2156 from 100000\n",
      "loss: 0.0023241529568451824\n",
      "epoch:2157 from 100000\n",
      "loss: 0.002862131972165116\n",
      "epoch:2158 from 100000\n",
      "loss: 0.002474588886229867\n",
      "epoch:2159 from 100000\n",
      "loss: 0.0019996956335148752\n",
      "epoch:2160 from 100000\n",
      "loss: 0.002689392957165637\n",
      "epoch:2161 from 100000\n",
      "loss: 0.002576970458761707\n",
      "epoch:2162 from 100000\n",
      "loss: 0.0021888115417141307\n",
      "epoch:2163 from 100000\n",
      "loss: 0.0022004040482386245\n",
      "epoch:2164 from 100000\n",
      "loss: 0.0027020364314580547\n",
      "epoch:2165 from 100000\n",
      "loss: 0.003079372627041721\n",
      "epoch:2166 from 100000\n",
      "loss: 0.0024299998750336727\n",
      "epoch:2167 from 100000\n",
      "loss: 0.00230247345615453\n",
      "epoch:2168 from 100000\n",
      "loss: 0.0023124168116623878\n",
      "epoch:2169 from 100000\n",
      "loss: 0.0024901070119369706\n",
      "epoch:2170 from 100000\n",
      "loss: 0.0026503651915561477\n",
      "epoch:2171 from 100000\n",
      "loss: 0.003245492668725092\n",
      "epoch:2172 from 100000\n",
      "loss: 0.002672857667448625\n",
      "epoch:2173 from 100000\n",
      "loss: 0.0023811537040785424\n",
      "epoch:2174 from 100000\n",
      "loss: 0.0025699939464232102\n",
      "epoch:2175 from 100000\n",
      "loss: 0.0022784103182118542\n",
      "epoch:2176 from 100000\n",
      "loss: 0.0028304579480789016\n",
      "epoch:2177 from 100000\n",
      "loss: 0.001965207459733165\n",
      "epoch:2178 from 100000\n",
      "loss: 0.0025285919577129334\n",
      "epoch:2179 from 100000\n",
      "loss: 0.002815978544816744\n",
      "epoch:2180 from 100000\n",
      "loss: 0.002901851601762928\n",
      "epoch:2181 from 100000\n",
      "loss: 0.0031931153176287808\n",
      "epoch:2182 from 100000\n",
      "loss: 0.0022614576932369346\n",
      "epoch:2183 from 100000\n",
      "loss: 0.002815377615235093\n",
      "epoch:2184 from 100000\n",
      "loss: 0.002429484501760726\n",
      "epoch:2185 from 100000\n",
      "loss: 0.0025383859021417265\n",
      "epoch:2186 from 100000\n",
      "loss: 0.0032768535152740697\n",
      "epoch:2187 from 100000\n",
      "loss: 0.0025907905703661527\n",
      "epoch:2188 from 100000\n",
      "loss: 0.002862447975928717\n",
      "epoch:2189 from 100000\n",
      "loss: 0.0030309296017492215\n",
      "epoch:2190 from 100000\n",
      "loss: 0.0021707552580210177\n",
      "epoch:2191 from 100000\n",
      "loss: 0.002874140519509183\n",
      "epoch:2192 from 100000\n",
      "loss: 0.00234485431298026\n",
      "epoch:2193 from 100000\n",
      "loss: 0.0030929527279113245\n",
      "epoch:2194 from 100000\n",
      "loss: 0.002382527244179755\n",
      "epoch:2195 from 100000\n",
      "loss: 0.0024789159089095847\n",
      "epoch:2196 from 100000\n",
      "loss: 0.002701950302395062\n",
      "epoch:2197 from 100000\n",
      "loss: 0.0024883963127706465\n",
      "epoch:2198 from 100000\n",
      "loss: 0.003005686276082997\n",
      "epoch:2199 from 100000\n",
      "loss: 0.0019685276523375705\n",
      "epoch:2200 from 100000\n",
      "loss: 0.001895691884669759\n",
      "epoch:2201 from 100000\n",
      "loss: 0.002535912606902002\n",
      "epoch:2202 from 100000\n",
      "loss: 0.00198568985800307\n",
      "epoch:2203 from 100000\n",
      "loss: 0.0025367005377177065\n",
      "epoch:2204 from 100000\n",
      "loss: 0.002439461394146747\n",
      "epoch:2205 from 100000\n",
      "loss: 0.0026590283756036115\n",
      "epoch:2206 from 100000\n",
      "loss: 0.0024311098287341792\n",
      "epoch:2207 from 100000\n",
      "loss: 0.0023890146120493227\n",
      "epoch:2208 from 100000\n",
      "loss: 0.0023870683882591948\n",
      "epoch:2209 from 100000\n",
      "loss: 0.002260668207384242\n",
      "epoch:2210 from 100000\n",
      "loss: 0.002599522905853746\n",
      "epoch:2211 from 100000\n",
      "loss: 0.0021870050538552732\n",
      "epoch:2212 from 100000\n",
      "loss: 0.002498885871176526\n",
      "epoch:2213 from 100000\n",
      "loss: 0.002572514325379416\n",
      "epoch:2214 from 100000\n",
      "loss: 0.0023282890354607844\n",
      "epoch:2215 from 100000\n",
      "loss: 0.0025131643632789746\n",
      "epoch:2216 from 100000\n",
      "loss: 0.002514006907460953\n",
      "epoch:2217 from 100000\n",
      "loss: 0.002453142745991954\n",
      "epoch:2218 from 100000\n",
      "loss: 0.0023251839058957557\n",
      "epoch:2219 from 100000\n",
      "loss: 0.002949436583548959\n",
      "epoch:2220 from 100000\n",
      "loss: 0.0033820414442686797\n",
      "epoch:2221 from 100000\n",
      "loss: 0.002723155043611314\n",
      "epoch:2222 from 100000\n",
      "loss: 0.0028123679771567822\n",
      "epoch:2223 from 100000\n",
      "loss: 0.0025441836105812854\n",
      "epoch:2224 from 100000\n",
      "loss: 0.002328481886223699\n",
      "epoch:2225 from 100000\n",
      "loss: 0.0022063211146985395\n",
      "epoch:2226 from 100000\n",
      "loss: 0.0024253829275835415\n",
      "epoch:2227 from 100000\n",
      "loss: 0.00286048356434776\n",
      "epoch:2228 from 100000\n",
      "loss: 0.002687261458845569\n",
      "epoch:2229 from 100000\n",
      "loss: 0.002837534033318434\n",
      "epoch:2230 from 100000\n",
      "loss: 0.002460140496623353\n",
      "epoch:2231 from 100000\n",
      "loss: 0.0024617312961368043\n",
      "epoch:2232 from 100000\n",
      "loss: 0.002606774214370944\n",
      "epoch:2233 from 100000\n",
      "loss: 0.00301340562575092\n",
      "epoch:2234 from 100000\n",
      "loss: 0.0020292055344839355\n",
      "epoch:2235 from 100000\n",
      "loss: 0.002160373681135752\n",
      "epoch:2236 from 100000\n",
      "loss: 0.002135937994808046\n",
      "epoch:2237 from 100000\n",
      "loss: 0.0027631284611686624\n",
      "epoch:2238 from 100000\n",
      "loss: 0.0024214472296648637\n",
      "epoch:2239 from 100000\n",
      "loss: 0.002470304510097776\n",
      "epoch:2240 from 100000\n",
      "loss: 0.0030143359220460297\n",
      "epoch:2241 from 100000\n",
      "loss: 0.002666585802046484\n",
      "epoch:2242 from 100000\n",
      "loss: 0.0025004988960972696\n",
      "epoch:2243 from 100000\n",
      "loss: 0.002514296341416866\n",
      "epoch:2244 from 100000\n",
      "loss: 0.002531430037009841\n",
      "epoch:2245 from 100000\n",
      "loss: 0.0020599504389622325\n",
      "epoch:2246 from 100000\n",
      "loss: 0.0025889019646001543\n",
      "epoch:2247 from 100000\n",
      "loss: 0.002723740801684471\n",
      "epoch:2248 from 100000\n",
      "loss: 0.002527577988857388\n",
      "epoch:2249 from 100000\n",
      "loss: 0.0024305172183574086\n",
      "epoch:2250 from 100000\n",
      "loss: 0.0023823992609095512\n",
      "epoch:2251 from 100000\n",
      "loss: 0.002051522692852359\n",
      "epoch:2252 from 100000\n",
      "loss: 0.0030269027831764106\n",
      "epoch:2253 from 100000\n",
      "loss: 0.0023613663144033126\n",
      "epoch:2254 from 100000\n",
      "loss: 0.0027566514993537794\n",
      "epoch:2255 from 100000\n",
      "loss: 0.0020601495710934614\n",
      "epoch:2256 from 100000\n",
      "loss: 0.002087368077411611\n",
      "epoch:2257 from 100000\n",
      "loss: 0.002778715873375859\n",
      "epoch:2258 from 100000\n",
      "loss: 0.002554324432651356\n",
      "epoch:2259 from 100000\n",
      "loss: 0.002158733271755864\n",
      "epoch:2260 from 100000\n",
      "loss: 0.002153278679472237\n",
      "epoch:2261 from 100000\n",
      "loss: 0.002381160521451875\n",
      "epoch:2262 from 100000\n",
      "loss: 0.002307017831981284\n",
      "epoch:2263 from 100000\n",
      "loss: 0.0024699972867665565\n",
      "epoch:2264 from 100000\n",
      "loss: 0.002604691589624508\n",
      "epoch:2265 from 100000\n",
      "loss: 0.0028062793865899494\n",
      "epoch:2266 from 100000\n",
      "loss: 0.003083086098286003\n",
      "epoch:2267 from 100000\n",
      "loss: 0.0023814187711934665\n",
      "epoch:2268 from 100000\n",
      "loss: 0.0023859185058938692\n",
      "epoch:2269 from 100000\n",
      "loss: 0.0026628882047248226\n",
      "epoch:2270 from 100000\n",
      "loss: 0.002375717994311799\n",
      "epoch:2271 from 100000\n",
      "loss: 0.001886745851379601\n",
      "epoch:2272 from 100000\n",
      "loss: 0.0035684419596435646\n",
      "epoch:2273 from 100000\n",
      "loss: 0.002420707191447491\n",
      "epoch:2274 from 100000\n",
      "loss: 0.002525224422328165\n",
      "epoch:2275 from 100000\n",
      "loss: 0.003058126213915102\n",
      "epoch:2276 from 100000\n",
      "loss: 0.002256991242575168\n",
      "epoch:2277 from 100000\n",
      "loss: 0.002463791595772591\n",
      "epoch:2278 from 100000\n",
      "loss: 0.0023650087731823533\n",
      "epoch:2279 from 100000\n",
      "loss: 0.002506433842512479\n",
      "epoch:2280 from 100000\n",
      "loss: 0.002402093521311599\n",
      "epoch:2281 from 100000\n",
      "loss: 0.002799409082442139\n",
      "epoch:2282 from 100000\n",
      "loss: 0.0024154925665129667\n",
      "epoch:2283 from 100000\n",
      "loss: 0.0027002007828969\n",
      "epoch:2284 from 100000\n",
      "loss: 0.0028192416988304103\n",
      "epoch:2285 from 100000\n",
      "loss: 0.0026292018612821266\n",
      "epoch:2286 from 100000\n",
      "loss: 0.002875090200092245\n",
      "epoch:2287 from 100000\n",
      "loss: 0.0025951586925430092\n",
      "epoch:2288 from 100000\n",
      "loss: 0.002754176710126899\n",
      "epoch:2289 from 100000\n",
      "loss: 0.002229660807117284\n",
      "epoch:2290 from 100000\n",
      "loss: 0.0025275795238144383\n",
      "epoch:2291 from 100000\n",
      "loss: 0.00278698585117354\n",
      "epoch:2292 from 100000\n",
      "loss: 0.002522946866506004\n",
      "epoch:2293 from 100000\n",
      "loss: 0.0028702422292781193\n",
      "epoch:2294 from 100000\n",
      "loss: 0.002569220995667365\n",
      "epoch:2295 from 100000\n",
      "loss: 0.002619457689718274\n",
      "epoch:2296 from 100000\n",
      "loss: 0.00246476001993301\n",
      "retrying for data\n",
      "epoch:2297 from 100000\n",
      "loss: 0.002984711466638146\n",
      "epoch:2298 from 100000\n",
      "loss: 0.0025576336053916293\n",
      "epoch:2299 from 100000\n",
      "loss: 0.0026459404256797825\n",
      "epoch:2300 from 100000\n",
      "loss: 0.0028297685726101918\n",
      "epoch:2301 from 100000\n",
      "loss: 0.002585362627826271\n",
      "epoch:2302 from 100000\n",
      "loss: 0.0027265257873523296\n",
      "epoch:2303 from 100000\n",
      "loss: 0.0029263469099589656\n",
      "epoch:2304 from 100000\n",
      "loss: 0.002383206058709675\n",
      "epoch:2305 from 100000\n",
      "loss: 0.0021549033182282074\n",
      "epoch:2306 from 100000\n",
      "loss: 0.0027445375041565967\n",
      "epoch:2307 from 100000\n",
      "loss: 0.0022788259499293417\n",
      "epoch:2308 from 100000\n",
      "loss: 0.002106280164269947\n",
      "epoch:2309 from 100000\n",
      "loss: 0.002738782427741171\n",
      "epoch:2310 from 100000\n",
      "loss: 0.0025057278466817934\n",
      "epoch:2311 from 100000\n",
      "loss: 0.0022778404655809936\n",
      "epoch:2312 from 100000\n",
      "loss: 0.002647607577500821\n",
      "epoch:2313 from 100000\n",
      "loss: 0.0027556105044084234\n",
      "epoch:2314 from 100000\n",
      "loss: 0.0026897533590499734\n",
      "epoch:2315 from 100000\n",
      "loss: 0.00209758798735038\n",
      "epoch:2316 from 100000\n",
      "loss: 0.002579870917742255\n",
      "epoch:2317 from 100000\n",
      "loss: 0.002892658343853327\n",
      "epoch:2318 from 100000\n",
      "loss: 0.0024039971006857286\n",
      "epoch:2319 from 100000\n",
      "loss: 0.0027297842710822806\n",
      "epoch:2320 from 100000\n",
      "loss: 0.0026031164330220236\n",
      "epoch:2321 from 100000\n",
      "loss: 0.0021850405983343535\n",
      "epoch:2322 from 100000\n",
      "loss: 0.0027326009108321614\n",
      "epoch:2323 from 100000\n",
      "loss: 0.002433957813707366\n",
      "epoch:2324 from 100000\n",
      "loss: 0.0027484416672791667\n",
      "epoch:2325 from 100000\n",
      "loss: 0.0025925462689713186\n",
      "epoch:2326 from 100000\n",
      "loss: 0.0025899610163548914\n",
      "epoch:2327 from 100000\n",
      "loss: 0.0025553996536586965\n",
      "epoch:2328 from 100000\n",
      "loss: 0.0025834223341121287\n",
      "epoch:2329 from 100000\n",
      "loss: 0.003126369506304627\n",
      "epoch:2330 from 100000\n",
      "loss: 0.002486432999006638\n",
      "epoch:2331 from 100000\n",
      "loss: 0.0026428088051488885\n",
      "epoch:2332 from 100000\n",
      "loss: 0.0027800373959223634\n",
      "epoch:2333 from 100000\n",
      "loss: 0.0029754035769542497\n",
      "epoch:2334 from 100000\n",
      "loss: 0.002423539289367227\n",
      "epoch:2335 from 100000\n",
      "loss: 0.002234820007458893\n",
      "epoch:2336 from 100000\n",
      "loss: 0.0019230945372612496\n",
      "epoch:2337 from 100000\n",
      "loss: 0.0027733630994077885\n",
      "epoch:2338 from 100000\n",
      "loss: 0.002419303835949904\n",
      "epoch:2339 from 100000\n",
      "loss: 0.0023531075551375125\n",
      "epoch:2340 from 100000\n",
      "loss: 0.0025576157578939274\n",
      "epoch:2341 from 100000\n",
      "loss: 0.00199019117914645\n",
      "epoch:2342 from 100000\n",
      "loss: 0.002729205255548095\n",
      "epoch:2343 from 100000\n",
      "loss: 0.002819241171152953\n",
      "epoch:2344 from 100000\n",
      "loss: 0.001911774317278514\n",
      "epoch:2345 from 100000\n",
      "loss: 0.002326265208807854\n",
      "epoch:2346 from 100000\n",
      "loss: 0.0025485146159098804\n",
      "epoch:2347 from 100000\n",
      "loss: 0.001945627589407195\n",
      "epoch:2348 from 100000\n",
      "loss: 0.002543223609677625\n",
      "epoch:2349 from 100000\n",
      "loss: 0.0024644627464169844\n",
      "epoch:2350 from 100000\n",
      "loss: 0.002455443480386066\n",
      "epoch:2351 from 100000\n",
      "loss: 0.0019825716498473867\n",
      "epoch:2352 from 100000\n",
      "loss: 0.002112739823019183\n",
      "epoch:2353 from 100000\n",
      "loss: 0.0028450280590277544\n",
      "epoch:2354 from 100000\n",
      "loss: 0.002309434768733354\n",
      "epoch:2355 from 100000\n",
      "loss: 0.0025851783915982196\n",
      "epoch:2356 from 100000\n",
      "loss: 0.0023049216540584894\n",
      "epoch:2357 from 100000\n",
      "loss: 0.002561033218682951\n",
      "epoch:2358 from 100000\n",
      "loss: 0.002528883034926821\n",
      "epoch:2359 from 100000\n",
      "loss: 0.0031407757732324626\n",
      "epoch:2360 from 100000\n",
      "loss: 0.0025009874111674435\n",
      "epoch:2361 from 100000\n",
      "loss: 0.002462424306173716\n",
      "epoch:2362 from 100000\n",
      "loss: 0.00218981715485711\n",
      "epoch:2363 from 100000\n",
      "loss: 0.0023043459626705953\n",
      "epoch:2364 from 100000\n",
      "loss: 0.0021133919200195805\n",
      "epoch:2365 from 100000\n",
      "loss: 0.0021901945525257815\n",
      "epoch:2366 from 100000\n",
      "loss: 0.002216966884745375\n",
      "epoch:2367 from 100000\n",
      "loss: 0.002165449384705198\n",
      "epoch:2368 from 100000\n",
      "loss: 0.0024510695241133362\n",
      "epoch:2369 from 100000\n",
      "loss: 0.002622037624433915\n",
      "epoch:2370 from 100000\n",
      "loss: 0.0030681065758244586\n",
      "epoch:2371 from 100000\n",
      "loss: 0.002162538185757512\n",
      "epoch:2372 from 100000\n",
      "loss: 0.00237122715697069\n",
      "epoch:2373 from 100000\n",
      "loss: 0.0024443310435344756\n",
      "epoch:2374 from 100000\n",
      "loss: 0.002313761836830963\n",
      "epoch:2375 from 100000\n",
      "loss: 0.002494203948458562\n",
      "epoch:2376 from 100000\n",
      "loss: 0.0024928910837758167\n",
      "epoch:2377 from 100000\n",
      "loss: 0.0024064558303109607\n",
      "epoch:2378 from 100000\n",
      "loss: 0.0029202162655508346\n",
      "epoch:2379 from 100000\n",
      "loss: 0.0022927133065309135\n",
      "epoch:2380 from 100000\n",
      "loss: 0.002848643325194189\n",
      "epoch:2381 from 100000\n",
      "loss: 0.003090650913634363\n",
      "epoch:2382 from 100000\n",
      "loss: 0.002417909396399409\n",
      "epoch:2383 from 100000\n",
      "loss: 0.0020407971305189676\n",
      "epoch:2384 from 100000\n",
      "loss: 0.002953055365196633\n",
      "epoch:2385 from 100000\n",
      "loss: 0.0028614839137901527\n",
      "epoch:2386 from 100000\n",
      "loss: 0.0026470414229038397\n",
      "epoch:2387 from 100000\n",
      "loss: 0.0028280048315991735\n",
      "epoch:2388 from 100000\n",
      "loss: 0.002184582013967429\n",
      "epoch:2389 from 100000\n",
      "loss: 0.002091638756894554\n",
      "epoch:2390 from 100000\n",
      "loss: 0.0024550236927325386\n",
      "epoch:2391 from 100000\n",
      "loss: 0.00267134717631734\n",
      "epoch:2392 from 100000\n",
      "loss: 0.002796381034102069\n",
      "epoch:2393 from 100000\n",
      "loss: 0.0026026473158253793\n",
      "epoch:2394 from 100000\n",
      "loss: 0.002252029365706676\n",
      "epoch:2395 from 100000\n",
      "loss: 0.002669406378458916\n",
      "epoch:2396 from 100000\n",
      "loss: 0.002717904085642431\n",
      "epoch:2397 from 100000\n",
      "loss: 0.002595052483627569\n",
      "epoch:2398 from 100000\n",
      "loss: 0.002740016488218089\n",
      "epoch:2399 from 100000\n",
      "loss: 0.0022384232991896624\n",
      "epoch:2400 from 100000\n",
      "loss: 0.003260147327296181\n",
      "epoch:2401 from 100000\n",
      "loss: 0.0019386504861529374\n",
      "epoch:2402 from 100000\n",
      "loss: 0.002728359976288175\n",
      "epoch:2403 from 100000\n",
      "loss: 0.0027045936378442548\n",
      "epoch:2404 from 100000\n",
      "loss: 0.00277016647049777\n",
      "epoch:2405 from 100000\n",
      "loss: 0.002197872125094591\n",
      "epoch:2406 from 100000\n",
      "loss: 0.0021767285297755734\n",
      "epoch:2407 from 100000\n",
      "loss: 0.002170872593126205\n",
      "epoch:2408 from 100000\n",
      "loss: 0.0027472126682255293\n",
      "epoch:2409 from 100000\n",
      "loss: 0.002618121578308319\n",
      "epoch:2410 from 100000\n",
      "loss: 0.00266717184291565\n",
      "epoch:2411 from 100000\n",
      "loss: 0.002627026019240475\n",
      "epoch:2412 from 100000\n",
      "loss: 0.002821765725826708\n",
      "epoch:2413 from 100000\n",
      "loss: 0.002841202908228979\n",
      "epoch:2414 from 100000\n",
      "loss: 0.0027430062056623683\n",
      "epoch:2415 from 100000\n",
      "loss: 0.0028677001951393777\n",
      "epoch:2416 from 100000\n",
      "loss: 0.002130578920400694\n",
      "epoch:2417 from 100000\n",
      "loss: 0.0031740394875328093\n",
      "epoch:2418 from 100000\n",
      "loss: 0.0025082526505997293\n",
      "epoch:2419 from 100000\n",
      "loss: 0.002711082342983673\n",
      "epoch:2420 from 100000\n",
      "loss: 0.0026735613716653006\n",
      "epoch:2421 from 100000\n",
      "loss: 0.002774991966731477\n",
      "epoch:2422 from 100000\n",
      "loss: 0.0020735458303988707\n",
      "epoch:2423 from 100000\n",
      "loss: 0.0024619833097005994\n",
      "epoch:2424 from 100000\n",
      "loss: 0.0024152041015099712\n",
      "epoch:2425 from 100000\n",
      "loss: 0.002874689682897724\n",
      "epoch:2426 from 100000\n",
      "loss: 0.0025324222044105227\n",
      "epoch:2427 from 100000\n",
      "loss: 0.001832426769368567\n",
      "epoch:2428 from 100000\n",
      "loss: 0.0021816474435354394\n",
      "epoch:2429 from 100000\n",
      "loss: 0.0026998890794800445\n",
      "epoch:2430 from 100000\n",
      "loss: 0.002602145161660019\n",
      "epoch:2431 from 100000\n",
      "loss: 0.0020238442057802786\n",
      "epoch:2432 from 100000\n",
      "loss: 0.0020252961179494378\n",
      "epoch:2433 from 100000\n",
      "loss: 0.00208810968393891\n",
      "epoch:2434 from 100000\n",
      "loss: 0.0023768741110075098\n",
      "epoch:2435 from 100000\n",
      "loss: 0.002570130538003923\n",
      "epoch:2436 from 100000\n",
      "loss: 0.002605741617401236\n",
      "epoch:2437 from 100000\n",
      "loss: 0.0028127198663128183\n",
      "epoch:2438 from 100000\n",
      "loss: 0.0026444359592403544\n",
      "epoch:2439 from 100000\n",
      "loss: 0.0029476459879447248\n",
      "epoch:2440 from 100000\n",
      "loss: 0.00214176443495262\n",
      "epoch:2441 from 100000\n",
      "loss: 0.0032663977792708465\n",
      "epoch:2442 from 100000\n",
      "loss: 0.002698513000808589\n",
      "epoch:2443 from 100000\n",
      "loss: 0.0028122099649579013\n",
      "epoch:2444 from 100000\n",
      "loss: 0.0030111297688222294\n",
      "epoch:2445 from 100000\n",
      "loss: 0.0024163231998386436\n",
      "epoch:2446 from 100000\n",
      "loss: 0.0018433120993393004\n",
      "epoch:2447 from 100000\n",
      "loss: 0.0023727822187140646\n",
      "epoch:2448 from 100000\n",
      "loss: 0.002318388086905543\n",
      "epoch:2449 from 100000\n",
      "loss: 0.00242178627524936\n",
      "epoch:2450 from 100000\n",
      "loss: 0.0028279686214034427\n",
      "epoch:2451 from 100000\n",
      "loss: 0.0023802217417738802\n",
      "epoch:2452 from 100000\n",
      "loss: 0.002719423637984164\n",
      "epoch:2453 from 100000\n",
      "loss: 0.0023770914046536973\n",
      "epoch:2454 from 100000\n",
      "loss: 0.0022963948759411323\n",
      "epoch:2455 from 100000\n",
      "loss: 0.0021814599736273976\n",
      "epoch:2456 from 100000\n",
      "loss: 0.0025737707288726597\n",
      "epoch:2457 from 100000\n",
      "loss: 0.0021080542002636093\n",
      "epoch:2458 from 100000\n",
      "loss: 0.002496526284232914\n",
      "epoch:2459 from 100000\n",
      "loss: 0.0021576857601957045\n",
      "epoch:2460 from 100000\n",
      "loss: 0.002544248283641082\n",
      "epoch:2461 from 100000\n",
      "loss: 0.002314824960009787\n",
      "epoch:2462 from 100000\n",
      "loss: 0.0023581605210267753\n",
      "epoch:2463 from 100000\n",
      "loss: 0.002425380039966285\n",
      "epoch:2464 from 100000\n",
      "loss: 0.0028149330082385404\n",
      "epoch:2465 from 100000\n",
      "loss: 0.0021553195806802705\n",
      "epoch:2466 from 100000\n",
      "loss: 0.002852294694946522\n",
      "epoch:2467 from 100000\n",
      "loss: 0.0022675526004860558\n",
      "epoch:2468 from 100000\n",
      "loss: 0.0028062188830659807\n",
      "epoch:2469 from 100000\n",
      "loss: 0.0020547633482976835\n",
      "epoch:2470 from 100000\n",
      "loss: 0.0023928727442523723\n",
      "epoch:2471 from 100000\n",
      "loss: 0.002527230570763095\n",
      "epoch:2472 from 100000\n",
      "loss: 0.0023417261767235686\n",
      "epoch:2473 from 100000\n",
      "loss: 0.0027348174789665336\n",
      "epoch:2474 from 100000\n",
      "loss: 0.0025857636138937323\n",
      "epoch:2475 from 100000\n",
      "loss: 0.0023452550122442517\n",
      "epoch:2476 from 100000\n",
      "loss: 0.0022193569020316772\n",
      "epoch:2477 from 100000\n",
      "loss: 0.0027559966487160636\n",
      "epoch:2478 from 100000\n",
      "loss: 0.0022711805986546096\n",
      "epoch:2479 from 100000\n",
      "loss: 0.002285013556161175\n",
      "epoch:2480 from 100000\n",
      "loss: 0.0023802719247214554\n",
      "epoch:2481 from 100000\n",
      "loss: 0.003496678451625712\n",
      "epoch:2482 from 100000\n",
      "loss: 0.002617607209771222\n",
      "epoch:2483 from 100000\n",
      "loss: 0.0024965937454339837\n",
      "epoch:2484 from 100000\n",
      "loss: 0.002380697754119865\n",
      "epoch:2485 from 100000\n",
      "loss: 0.0023971464522958286\n",
      "epoch:2486 from 100000\n",
      "loss: 0.0025505918900137203\n",
      "epoch:2487 from 100000\n",
      "loss: 0.00267236604005916\n",
      "epoch:2488 from 100000\n",
      "loss: 0.0024345834471830585\n",
      "epoch:2489 from 100000\n",
      "loss: 0.0027592067873172255\n",
      "epoch:2490 from 100000\n",
      "loss: 0.0023528426365828636\n",
      "epoch:2491 from 100000\n",
      "loss: 0.001976690887914856\n",
      "epoch:2492 from 100000\n",
      "loss: 0.002712994716347339\n",
      "epoch:2493 from 100000\n",
      "loss: 0.0027615248235406398\n",
      "epoch:2494 from 100000\n",
      "loss: 0.0029058543135818127\n",
      "epoch:2495 from 100000\n",
      "loss: 0.0022150282833592883\n",
      "epoch:2496 from 100000\n",
      "loss: 0.0021998154649622848\n",
      "epoch:2497 from 100000\n",
      "loss: 0.0030100516065658667\n",
      "epoch:2498 from 100000\n",
      "loss: 0.0023441173341609556\n",
      "epoch:2499 from 100000\n",
      "loss: 0.002843379979083238\n",
      "epoch:2500 from 100000\n",
      "loss: 0.00228863194627138\n",
      "epoch:2501 from 100000\n",
      "loss: 0.002167522726807647\n",
      "epoch:2502 from 100000\n",
      "loss: 0.0024519653317724988\n",
      "epoch:2503 from 100000\n",
      "loss: 0.002215229651227446\n",
      "epoch:2504 from 100000\n",
      "loss: 0.0025284250720289947\n",
      "epoch:2505 from 100000\n",
      "loss: 0.002764156083259195\n",
      "epoch:2506 from 100000\n",
      "loss: 0.0024330150568800946\n",
      "epoch:2507 from 100000\n",
      "loss: 0.002070450148863756\n",
      "epoch:2508 from 100000\n",
      "loss: 0.002297538619814077\n",
      "epoch:2509 from 100000\n",
      "loss: 0.0020454352466572345\n",
      "epoch:2510 from 100000\n",
      "loss: 0.0024148771739191943\n",
      "epoch:2511 from 100000\n",
      "loss: 0.002390901476857721\n",
      "epoch:2512 from 100000\n",
      "loss: 0.0025399635577798563\n",
      "epoch:2513 from 100000\n",
      "loss: 0.002474438242074939\n",
      "epoch:2514 from 100000\n",
      "loss: 0.002186491474247987\n",
      "epoch:2515 from 100000\n",
      "loss: 0.002656162362086434\n",
      "epoch:2516 from 100000\n",
      "loss: 0.0026583535881741227\n",
      "epoch:2517 from 100000\n",
      "loss: 0.0023525206893850736\n",
      "epoch:2518 from 100000\n",
      "loss: 0.0018931833629665107\n",
      "epoch:2519 from 100000\n",
      "loss: 0.002109629780505884\n",
      "epoch:2520 from 100000\n",
      "loss: 0.0021718682661457933\n",
      "epoch:2521 from 100000\n",
      "loss: 0.002012412215549375\n",
      "epoch:2522 from 100000\n",
      "loss: 0.0021423463813903254\n",
      "epoch:2523 from 100000\n",
      "loss: 0.0025154274580359015\n",
      "epoch:2524 from 100000\n",
      "loss: 0.0019897521798242224\n",
      "epoch:2525 from 100000\n",
      "loss: 0.002191871290435188\n",
      "epoch:2526 from 100000\n",
      "loss: 0.0022988099747465185\n",
      "epoch:2527 from 100000\n",
      "loss: 0.0025926697417446576\n",
      "epoch:2528 from 100000\n",
      "loss: 0.0024463645693515446\n",
      "epoch:2529 from 100000\n",
      "loss: 0.002074652543939237\n",
      "epoch:2530 from 100000\n",
      "loss: 0.0021364321385988205\n",
      "epoch:2531 from 100000\n",
      "loss: 0.0022558143057835878\n",
      "epoch:2532 from 100000\n",
      "loss: 0.0024759807206891082\n",
      "epoch:2533 from 100000\n",
      "loss: 0.00187525584462378\n",
      "epoch:2534 from 100000\n",
      "loss: 0.0024652073219897375\n",
      "epoch:2535 from 100000\n",
      "loss: 0.0023386537075111846\n",
      "epoch:2536 from 100000\n",
      "loss: 0.0022175276403118005\n",
      "epoch:2537 from 100000\n",
      "loss: 0.0024685086310540783\n",
      "epoch:2538 from 100000\n",
      "loss: 0.002534223772471478\n",
      "epoch:2539 from 100000\n",
      "loss: 0.0025258629254949483\n",
      "epoch:2540 from 100000\n",
      "loss: 0.002560332842975299\n",
      "epoch:2541 from 100000\n",
      "loss: 0.0026750339289236535\n",
      "epoch:2542 from 100000\n",
      "loss: 0.002214320054036989\n",
      "epoch:2543 from 100000\n",
      "loss: 0.002381497060611082\n",
      "epoch:2544 from 100000\n",
      "loss: 0.003229027218566216\n",
      "epoch:2545 from 100000\n",
      "loss: 0.0026582932850089946\n",
      "epoch:2546 from 100000\n",
      "loss: 0.0030573676101539604\n",
      "epoch:2547 from 100000\n",
      "loss: 0.0029322779194984605\n",
      "epoch:2548 from 100000\n",
      "loss: 0.003022609502920659\n",
      "epoch:2549 from 100000\n",
      "loss: 0.0031456576313075857\n",
      "epoch:2550 from 100000\n",
      "loss: 0.0023262594817765603\n",
      "epoch:2551 from 100000\n",
      "loss: 0.0026500503466309056\n",
      "epoch:2552 from 100000\n",
      "loss: 0.0022526977148800142\n",
      "epoch:2553 from 100000\n",
      "loss: 0.0021634483020989137\n",
      "epoch:2554 from 100000\n",
      "loss: 0.002422859011161904\n",
      "epoch:2555 from 100000\n",
      "loss: 0.0026375375837659476\n",
      "epoch:2556 from 100000\n",
      "loss: 0.0023043855891842213\n",
      "epoch:2557 from 100000\n",
      "loss: 0.003074835787998609\n",
      "epoch:2558 from 100000\n",
      "loss: 0.002746698282123816\n",
      "epoch:2559 from 100000\n",
      "loss: 0.002673771109698464\n",
      "epoch:2560 from 100000\n",
      "loss: 0.0022701716542883332\n",
      "epoch:2561 from 100000\n",
      "loss: 0.0022876370194921947\n",
      "epoch:2562 from 100000\n",
      "loss: 0.0027262662248404013\n",
      "epoch:2563 from 100000\n",
      "loss: 0.0023322217273289425\n",
      "epoch:2564 from 100000\n",
      "loss: 0.0025290622955367326\n",
      "epoch:2565 from 100000\n",
      "loss: 0.002188762304172087\n",
      "epoch:2566 from 100000\n",
      "loss: 0.0021398633692797375\n",
      "epoch:2567 from 100000\n",
      "loss: 0.002061050029027456\n",
      "epoch:2568 from 100000\n",
      "loss: 0.002667714774617025\n",
      "epoch:2569 from 100000\n",
      "loss: 0.002176381124826321\n",
      "epoch:2570 from 100000\n",
      "loss: 0.002724505859262649\n",
      "epoch:2571 from 100000\n",
      "loss: 0.002557706785694336\n",
      "epoch:2572 from 100000\n",
      "loss: 0.002293929959733987\n",
      "epoch:2573 from 100000\n",
      "loss: 0.0027786321045937257\n",
      "epoch:2574 from 100000\n",
      "loss: 0.0022849036110699217\n",
      "epoch:2575 from 100000\n",
      "loss: 0.002678571108233996\n",
      "epoch:2576 from 100000\n",
      "loss: 0.0028196765969425996\n",
      "epoch:2577 from 100000\n",
      "loss: 0.0026035027286752666\n",
      "epoch:2578 from 100000\n",
      "loss: 0.002561112177772884\n",
      "epoch:2579 from 100000\n",
      "loss: 0.0025348892030194747\n",
      "epoch:2580 from 100000\n",
      "loss: 0.0031563555583602465\n",
      "epoch:2581 from 100000\n",
      "loss: 0.002469540891922861\n",
      "epoch:2582 from 100000\n",
      "loss: 0.002781575956987581\n",
      "epoch:2583 from 100000\n",
      "loss: 0.0021243942250634973\n",
      "epoch:2584 from 100000\n",
      "loss: 0.0023578035660420937\n",
      "epoch:2585 from 100000\n",
      "loss: 0.002412648700669706\n",
      "epoch:2586 from 100000\n",
      "loss: 0.002767804220184189\n",
      "epoch:2587 from 100000\n",
      "loss: 0.0021791382212654753\n",
      "epoch:2588 from 100000\n",
      "loss: 0.0027129721170808807\n",
      "epoch:2589 from 100000\n",
      "loss: 0.0026422277349524848\n",
      "epoch:2590 from 100000\n",
      "loss: 0.0024042483315440677\n",
      "epoch:2591 from 100000\n",
      "loss: 0.0023463639241896317\n",
      "epoch:2592 from 100000\n",
      "loss: 0.003193287395902189\n",
      "epoch:2593 from 100000\n",
      "loss: 0.002654018208374964\n",
      "epoch:2594 from 100000\n",
      "loss: 0.0027896407990368743\n",
      "epoch:2595 from 100000\n",
      "loss: 0.0025160946989046806\n",
      "epoch:2596 from 100000\n",
      "loss: 0.0021969668886470117\n",
      "epoch:2597 from 100000\n",
      "loss: 0.002693530643540498\n",
      "epoch:2598 from 100000\n",
      "loss: 0.0026215727714316017\n",
      "epoch:2599 from 100000\n",
      "loss: 0.002383901925853138\n",
      "epoch:2600 from 100000\n",
      "loss: 0.0023005401320830288\n",
      "epoch:2601 from 100000\n",
      "loss: 0.0023248227460328508\n",
      "epoch:2602 from 100000\n",
      "loss: 0.0020028207550808474\n",
      "epoch:2603 from 100000\n",
      "loss: 0.002062865855009477\n",
      "epoch:2604 from 100000\n",
      "loss: 0.002502786040679439\n",
      "epoch:2605 from 100000\n",
      "loss: 0.0024451136025618325\n",
      "epoch:2606 from 100000\n",
      "loss: 0.0022227888289023667\n",
      "epoch:2607 from 100000\n",
      "loss: 0.0024282928909542534\n",
      "epoch:2608 from 100000\n",
      "loss: 0.0030173602985996695\n",
      "epoch:2609 from 100000\n",
      "loss: 0.0027903399413844454\n",
      "epoch:2610 from 100000\n",
      "loss: 0.002707368545088684\n",
      "epoch:2611 from 100000\n",
      "loss: 0.0022857682077983554\n",
      "epoch:2612 from 100000\n",
      "loss: 0.0025052973877848217\n",
      "epoch:2613 from 100000\n",
      "loss: 0.0020221277306404772\n",
      "epoch:2614 from 100000\n",
      "loss: 0.0021728120402713103\n",
      "epoch:2615 from 100000\n",
      "loss: 0.0029100704380198295\n",
      "epoch:2616 from 100000\n",
      "loss: 0.0022188632853783474\n",
      "epoch:2617 from 100000\n",
      "loss: 0.0024602046129018618\n",
      "epoch:2618 from 100000\n",
      "loss: 0.002249331740586058\n",
      "epoch:2619 from 100000\n",
      "loss: 0.002193769226266795\n",
      "epoch:2620 from 100000\n",
      "loss: 0.0022952509797846687\n",
      "epoch:2621 from 100000\n",
      "loss: 0.0020137300283096238\n",
      "epoch:2622 from 100000\n",
      "loss: 0.002087359823804036\n",
      "epoch:2623 from 100000\n",
      "loss: 0.0024899632598476273\n",
      "epoch:2624 from 100000\n",
      "loss: 0.0027714673587979632\n",
      "epoch:2625 from 100000\n",
      "loss: 0.0025602672249647185\n",
      "epoch:2626 from 100000\n",
      "loss: 0.002765838144000554\n",
      "epoch:2627 from 100000\n",
      "loss: 0.002347552708073408\n",
      "epoch:2628 from 100000\n",
      "loss: 0.0018834333668564796\n",
      "epoch:2629 from 100000\n",
      "loss: 0.002509465067078054\n",
      "epoch:2630 from 100000\n",
      "loss: 0.0024145552230834255\n",
      "epoch:2631 from 100000\n",
      "loss: 0.002139164085917855\n",
      "epoch:2632 from 100000\n",
      "loss: 0.002906643898541006\n",
      "epoch:2633 from 100000\n",
      "loss: 0.0029370161715291943\n",
      "epoch:2634 from 100000\n",
      "loss: 0.0020417814033635295\n",
      "epoch:2635 from 100000\n",
      "loss: 0.00256650825883753\n",
      "epoch:2636 from 100000\n",
      "loss: 0.002295609317314984\n",
      "epoch:2637 from 100000\n",
      "loss: 0.0028795775488958952\n",
      "epoch:2638 from 100000\n",
      "loss: 0.002704000422710351\n",
      "epoch:2639 from 100000\n",
      "loss: 0.00283556009418362\n",
      "epoch:2640 from 100000\n",
      "loss: 0.002086003529967684\n",
      "epoch:2641 from 100000\n",
      "loss: 0.002616633099023602\n",
      "epoch:2642 from 100000\n",
      "loss: 0.0029892417060182197\n",
      "epoch:2643 from 100000\n",
      "loss: 0.0022897296862822714\n",
      "epoch:2644 from 100000\n",
      "loss: 0.0025102491435120555\n",
      "epoch:2645 from 100000\n",
      "loss: 0.002744158019169163\n",
      "epoch:2646 from 100000\n",
      "loss: 0.0020743842104025134\n",
      "epoch:2647 from 100000\n",
      "loss: 0.0023391445492961793\n",
      "epoch:2648 from 100000\n",
      "loss: 0.0020914837676855313\n",
      "epoch:2649 from 100000\n",
      "loss: 0.003122845871928348\n",
      "epoch:2650 from 100000\n",
      "loss: 0.002817194485558616\n",
      "epoch:2651 from 100000\n",
      "loss: 0.003170089923173691\n",
      "epoch:2652 from 100000\n",
      "loss: 0.0025797472776503128\n",
      "epoch:2653 from 100000\n",
      "loss: 0.0027999427673535138\n",
      "epoch:2654 from 100000\n",
      "loss: 0.001993113808794078\n",
      "epoch:2655 from 100000\n",
      "loss: 0.0026431758795126825\n",
      "epoch:2656 from 100000\n",
      "loss: 0.0022389921618071185\n",
      "epoch:2657 from 100000\n",
      "loss: 0.0031494372564111472\n",
      "epoch:2658 from 100000\n",
      "loss: 0.002227027293088213\n",
      "epoch:2659 from 100000\n",
      "loss: 0.002967794998426143\n",
      "epoch:2660 from 100000\n",
      "loss: 0.002673798390333104\n",
      "epoch:2661 from 100000\n",
      "loss: 0.002795257781656346\n",
      "epoch:2662 from 100000\n",
      "loss: 0.002970743869298076\n",
      "epoch:2663 from 100000\n",
      "loss: 0.0022107472252628213\n",
      "epoch:2664 from 100000\n",
      "loss: 0.0022021046298164038\n",
      "epoch:2665 from 100000\n",
      "loss: 0.0028489348959510608\n",
      "epoch:2666 from 100000\n",
      "loss: 0.002369863837259345\n",
      "epoch:2667 from 100000\n",
      "loss: 0.00263811269419989\n",
      "epoch:2668 from 100000\n",
      "loss: 0.0027312178220881833\n",
      "epoch:2669 from 100000\n",
      "loss: 0.0021633801212033177\n",
      "epoch:2670 from 100000\n",
      "loss: 0.002175925345085261\n",
      "epoch:2671 from 100000\n",
      "loss: 0.003014409049569622\n",
      "epoch:2672 from 100000\n",
      "loss: 0.002644964879550571\n",
      "epoch:2673 from 100000\n",
      "loss: 0.002785236340088204\n",
      "epoch:2674 from 100000\n",
      "loss: 0.002430286521871494\n",
      "epoch:2675 from 100000\n",
      "loss: 0.0028511226402940792\n",
      "epoch:2676 from 100000\n",
      "loss: 0.002188946008260473\n",
      "epoch:2677 from 100000\n",
      "loss: 0.003086900502239587\n",
      "epoch:2678 from 100000\n",
      "loss: 0.0028604446631561586\n",
      "epoch:2679 from 100000\n",
      "loss: 0.00217695257040873\n",
      "epoch:2680 from 100000\n",
      "loss: 0.002366409210679876\n",
      "epoch:2681 from 100000\n",
      "loss: 0.002730827928075996\n",
      "epoch:2682 from 100000\n",
      "loss: 0.0026435710056205153\n",
      "epoch:2683 from 100000\n",
      "loss: 0.003052212285922451\n",
      "epoch:2684 from 100000\n",
      "loss: 0.0027484779489839184\n",
      "epoch:2685 from 100000\n",
      "loss: 0.002287231024894254\n",
      "epoch:2686 from 100000\n",
      "loss: 0.0021708844009822315\n",
      "epoch:2687 from 100000\n",
      "loss: 0.0027868250983260623\n",
      "epoch:2688 from 100000\n",
      "loss: 0.002213603572897682\n",
      "epoch:2689 from 100000\n",
      "loss: 0.0021879153440522714\n",
      "epoch:2690 from 100000\n",
      "loss: 0.0022812122261939294\n",
      "epoch:2691 from 100000\n",
      "loss: 0.002658199623084556\n",
      "epoch:2692 from 100000\n",
      "loss: 0.0024340261076076786\n",
      "epoch:2693 from 100000\n",
      "loss: 0.0022940905829216263\n",
      "epoch:2694 from 100000\n",
      "loss: 0.0020960239912994894\n",
      "epoch:2695 from 100000\n",
      "loss: 0.0026485735362484775\n",
      "epoch:2696 from 100000\n",
      "loss: 0.002426041341294649\n",
      "epoch:2697 from 100000\n",
      "loss: 0.002346486737963005\n",
      "epoch:2698 from 100000\n",
      "loss: 0.0021812420258413567\n",
      "epoch:2699 from 100000\n",
      "loss: 0.0019096713038209145\n",
      "epoch:2700 from 100000\n",
      "loss: 0.0021691083903760955\n",
      "epoch:2701 from 100000\n",
      "loss: 0.0023697314906598876\n",
      "epoch:2702 from 100000\n",
      "loss: 0.0021190392785115364\n",
      "epoch:2703 from 100000\n",
      "loss: 0.0025579352860631843\n",
      "epoch:2704 from 100000\n",
      "loss: 0.002371960120285621\n",
      "epoch:2705 from 100000\n",
      "loss: 0.0023394768950311118\n",
      "epoch:2706 from 100000\n",
      "loss: 0.0018813405376079118\n",
      "epoch:2707 from 100000\n",
      "loss: 0.0018036132779570835\n",
      "epoch:2708 from 100000\n",
      "loss: 0.002171476390145699\n",
      "epoch:2709 from 100000\n",
      "loss: 0.0027041585937013224\n",
      "epoch:2710 from 100000\n",
      "loss: 0.0029154227091794382\n",
      "epoch:2711 from 100000\n",
      "loss: 0.0023437194369364533\n",
      "epoch:2712 from 100000\n",
      "loss: 0.0024311975914201867\n",
      "epoch:2713 from 100000\n",
      "loss: 0.002988541738488948\n",
      "epoch:2714 from 100000\n",
      "loss: 0.003127055311381355\n",
      "epoch:2715 from 100000\n",
      "loss: 0.002582114158741433\n",
      "epoch:2716 from 100000\n",
      "loss: 0.0019404607756001724\n",
      "epoch:2717 from 100000\n",
      "loss: 0.002207905837110502\n",
      "epoch:2718 from 100000\n",
      "loss: 0.002422082944846693\n",
      "epoch:2719 from 100000\n",
      "loss: 0.002257822876714499\n",
      "epoch:2720 from 100000\n",
      "loss: 0.002360817900864731\n",
      "epoch:2721 from 100000\n",
      "loss: 0.002746249216045271\n",
      "epoch:2722 from 100000\n",
      "loss: 0.0027218730635354405\n",
      "epoch:2723 from 100000\n",
      "loss: 0.0023800496758781264\n",
      "epoch:2724 from 100000\n",
      "loss: 0.0023937364774440084\n",
      "epoch:2725 from 100000\n",
      "loss: 0.0029114611545537628\n",
      "epoch:2726 from 100000\n",
      "loss: 0.0025546809017100713\n",
      "epoch:2727 from 100000\n",
      "loss: 0.00220517820272903\n",
      "epoch:2728 from 100000\n",
      "loss: 0.002718901651746819\n",
      "epoch:2729 from 100000\n",
      "loss: 0.0022600805838521865\n",
      "epoch:2730 from 100000\n",
      "loss: 0.002483607584025549\n",
      "epoch:2731 from 100000\n",
      "loss: 0.003030008667082029\n",
      "epoch:2732 from 100000\n",
      "loss: 0.001971537340239138\n",
      "epoch:2733 from 100000\n",
      "loss: 0.0025675571542649323\n",
      "epoch:2734 from 100000\n",
      "loss: 0.002283286751946889\n",
      "epoch:2735 from 100000\n",
      "loss: 0.0027175574020645854\n",
      "epoch:2736 from 100000\n",
      "loss: 0.002687802906990555\n",
      "epoch:2737 from 100000\n",
      "loss: 0.0031208699394369432\n",
      "epoch:2738 from 100000\n",
      "loss: 0.002939088469560147\n",
      "epoch:2739 from 100000\n",
      "loss: 0.00256622954765362\n",
      "epoch:2740 from 100000\n",
      "loss: 0.0023157820718893163\n",
      "epoch:2741 from 100000\n",
      "loss: 0.0026838050912658673\n",
      "epoch:2742 from 100000\n",
      "loss: 0.002283060907984691\n",
      "epoch:2743 from 100000\n",
      "loss: 0.002380108367901812\n",
      "epoch:2744 from 100000\n",
      "loss: 0.002775348425871016\n",
      "epoch:2745 from 100000\n",
      "loss: 0.002411196132356963\n",
      "epoch:2746 from 100000\n",
      "loss: 0.0024256067254100344\n",
      "epoch:2747 from 100000\n",
      "loss: 0.002221404862751797\n",
      "epoch:2748 from 100000\n",
      "loss: 0.002250873036075518\n",
      "epoch:2749 from 100000\n",
      "loss: 0.002655853057603963\n",
      "epoch:2750 from 100000\n",
      "loss: 0.0030156145881505836\n",
      "epoch:2751 from 100000\n",
      "loss: 0.0021811257646788818\n",
      "epoch:2752 from 100000\n",
      "loss: 0.0022457477336388365\n",
      "epoch:2753 from 100000\n",
      "loss: 0.0024593795719169975\n",
      "epoch:2754 from 100000\n",
      "loss: 0.0025001537323419143\n",
      "epoch:2755 from 100000\n",
      "loss: 0.0028154131257736026\n",
      "epoch:2756 from 100000\n",
      "loss: 0.0023624745100505606\n",
      "epoch:2757 from 100000\n",
      "loss: 0.0020079961036003624\n",
      "epoch:2758 from 100000\n",
      "loss: 0.0027869102380577715\n",
      "epoch:2759 from 100000\n",
      "loss: 0.00251099516228237\n",
      "epoch:2760 from 100000\n",
      "loss: 0.003190334501368852\n",
      "epoch:2761 from 100000\n",
      "loss: 0.0023629589697691245\n",
      "epoch:2762 from 100000\n",
      "loss: 0.0026298881668651575\n",
      "epoch:2763 from 100000\n",
      "loss: 0.0025372368490650388\n",
      "epoch:2764 from 100000\n",
      "loss: 0.002379131555812819\n",
      "epoch:2765 from 100000\n",
      "loss: 0.0022514706930678585\n",
      "epoch:2766 from 100000\n",
      "loss: 0.0026447092366197467\n",
      "epoch:2767 from 100000\n",
      "loss: 0.002598333195436453\n",
      "epoch:2768 from 100000\n",
      "loss: 0.0024198308888401243\n",
      "epoch:2769 from 100000\n",
      "loss: 0.002135627703978571\n",
      "epoch:2770 from 100000\n",
      "loss: 0.002049269320679059\n",
      "epoch:2771 from 100000\n",
      "loss: 0.002267044184407041\n",
      "epoch:2772 from 100000\n",
      "loss: 0.002316686584549643\n",
      "epoch:2773 from 100000\n",
      "loss: 0.0023138853345017196\n",
      "epoch:2774 from 100000\n",
      "loss: 0.002574731593568913\n",
      "epoch:2775 from 100000\n",
      "loss: 0.002544044583714822\n",
      "epoch:2776 from 100000\n",
      "loss: 0.0032000272472600955\n",
      "epoch:2777 from 100000\n",
      "loss: 0.0026531341550821708\n",
      "epoch:2778 from 100000\n",
      "loss: 0.002470789818062258\n",
      "epoch:2779 from 100000\n",
      "loss: 0.002175266297314238\n",
      "epoch:2780 from 100000\n",
      "loss: 0.002427002221196517\n",
      "epoch:2781 from 100000\n",
      "loss: 0.0029577529866458008\n",
      "epoch:2782 from 100000\n",
      "loss: 0.0025999281418478404\n",
      "epoch:2783 from 100000\n",
      "loss: 0.0025820463697527885\n",
      "epoch:2784 from 100000\n",
      "loss: 0.002269946727437855\n",
      "epoch:2785 from 100000\n",
      "loss: 0.001969928152391276\n",
      "epoch:2786 from 100000\n",
      "loss: 0.0019132506497783197\n",
      "epoch:2787 from 100000\n",
      "loss: 0.0020743693528828544\n",
      "epoch:2788 from 100000\n",
      "loss: 0.002461836720755173\n",
      "epoch:2789 from 100000\n",
      "loss: 0.0024397547658452368\n",
      "epoch:2790 from 100000\n",
      "loss: 0.002422851409107807\n",
      "epoch:2791 from 100000\n",
      "loss: 0.0017563762361021418\n",
      "epoch:2792 from 100000\n",
      "loss: 0.002030167850534781\n",
      "epoch:2793 from 100000\n",
      "loss: 0.0020447236598784\n",
      "epoch:2794 from 100000\n",
      "loss: 0.0023859213811761038\n",
      "epoch:2795 from 100000\n",
      "loss: 0.0029519323266526953\n",
      "epoch:2796 from 100000\n",
      "loss: 0.0025152346627947963\n",
      "epoch:2797 from 100000\n",
      "loss: 0.0026074454503088873\n",
      "epoch:2798 from 100000\n",
      "loss: 0.0022233385529375482\n",
      "epoch:2799 from 100000\n",
      "loss: 0.002670038352221127\n",
      "epoch:2800 from 100000\n",
      "loss: 0.00249574615753545\n",
      "epoch:2801 from 100000\n",
      "loss: 0.0025415594751052595\n",
      "epoch:2802 from 100000\n",
      "loss: 0.002700113619368949\n",
      "epoch:2803 from 100000\n",
      "loss: 0.0025737093612860917\n",
      "epoch:2804 from 100000\n",
      "loss: 0.002435420015700629\n",
      "epoch:2805 from 100000\n",
      "loss: 0.002297752489823779\n",
      "epoch:2806 from 100000\n",
      "loss: 0.0024135770131294976\n",
      "epoch:2807 from 100000\n",
      "loss: 0.0024847699753252073\n",
      "epoch:2808 from 100000\n",
      "loss: 0.0030183278775552935\n",
      "epoch:2809 from 100000\n",
      "loss: 0.002352108301465705\n",
      "epoch:2810 from 100000\n",
      "loss: 0.002786748310910525\n",
      "epoch:2811 from 100000\n",
      "loss: 0.002574059983260213\n",
      "epoch:2812 from 100000\n",
      "loss: 0.0022461787869190175\n",
      "epoch:2813 from 100000\n",
      "loss: 0.0023685692509616274\n",
      "epoch:2814 from 100000\n",
      "loss: 0.002547452110590598\n",
      "epoch:2815 from 100000\n",
      "loss: 0.0028379755802916407\n",
      "epoch:2816 from 100000\n",
      "loss: 0.0024484155447481726\n",
      "epoch:2817 from 100000\n",
      "loss: 0.002043092942557223\n",
      "epoch:2818 from 100000\n",
      "loss: 0.0023850375416998304\n",
      "epoch:2819 from 100000\n",
      "loss: 0.0025835193078052043\n",
      "epoch:2820 from 100000\n",
      "loss: 0.002082178657417444\n",
      "epoch:2821 from 100000\n",
      "loss: 0.002935911560697946\n",
      "epoch:2822 from 100000\n",
      "loss: 0.0020351499446746857\n",
      "epoch:2823 from 100000\n",
      "loss: 0.0019503879832143411\n",
      "epoch:2824 from 100000\n",
      "loss: 0.0023363817425092748\n",
      "epoch:2825 from 100000\n",
      "loss: 0.0018437871531631345\n",
      "epoch:2826 from 100000\n",
      "loss: 0.002503862311840521\n",
      "epoch:2827 from 100000\n",
      "loss: 0.002768544438652043\n",
      "epoch:2828 from 100000\n",
      "loss: 0.002098288273828075\n",
      "epoch:2829 from 100000\n",
      "loss: 0.0022512161136489794\n",
      "epoch:2830 from 100000\n",
      "loss: 0.003254770022493858\n",
      "epoch:2831 from 100000\n",
      "loss: 0.00253435082845499\n",
      "epoch:2832 from 100000\n",
      "loss: 0.002403998999511714\n",
      "epoch:2833 from 100000\n",
      "loss: 0.0025625036772680687\n",
      "epoch:2834 from 100000\n",
      "loss: 0.0021274931006587394\n",
      "epoch:2835 from 100000\n",
      "loss: 0.0022103303565472743\n",
      "epoch:2836 from 100000\n",
      "loss: 0.00222892628602267\n",
      "epoch:2837 from 100000\n",
      "loss: 0.0019569389514799695\n",
      "epoch:2838 from 100000\n",
      "loss: 0.002842606620333754\n",
      "epoch:2839 from 100000\n",
      "loss: 0.002445305449668922\n",
      "epoch:2840 from 100000\n",
      "loss: 0.00251045740014888\n",
      "epoch:2841 from 100000\n",
      "loss: 0.0027074297366880273\n",
      "epoch:2842 from 100000\n",
      "loss: 0.0025771119729540715\n",
      "epoch:2843 from 100000\n",
      "loss: 0.002067489786270471\n",
      "epoch:2844 from 100000\n",
      "loss: 0.002409500456906244\n",
      "epoch:2845 from 100000\n",
      "loss: 0.0027120541916474394\n",
      "epoch:2846 from 100000\n",
      "loss: 0.0026215423783639835\n",
      "epoch:2847 from 100000\n",
      "loss: 0.0021951878681250037\n",
      "epoch:2848 from 100000\n",
      "loss: 0.002396770498762635\n",
      "epoch:2849 from 100000\n",
      "loss: 0.002420599533763834\n",
      "epoch:2850 from 100000\n",
      "loss: 0.0029381052810322217\n",
      "epoch:2851 from 100000\n",
      "loss: 0.002595929850713219\n",
      "epoch:2852 from 100000\n",
      "loss: 0.0031222942473618787\n",
      "epoch:2853 from 100000\n",
      "loss: 0.0034867130187876683\n",
      "epoch:2854 from 100000\n",
      "loss: 0.002608236199023395\n",
      "epoch:2855 from 100000\n",
      "loss: 0.0028631100390725805\n",
      "epoch:2856 from 100000\n",
      "loss: 0.0020791273861391346\n",
      "epoch:2857 from 100000\n",
      "loss: 0.002592416348193183\n",
      "epoch:2858 from 100000\n",
      "loss: 0.002509971338241712\n",
      "epoch:2859 from 100000\n",
      "loss: 0.002694322076138178\n",
      "epoch:2860 from 100000\n",
      "loss: 0.002788554509209007\n",
      "epoch:2861 from 100000\n",
      "loss: 0.0027064932095868244\n",
      "epoch:2862 from 100000\n",
      "loss: 0.002991756105643617\n",
      "epoch:2863 from 100000\n",
      "loss: 0.002248380608719458\n",
      "epoch:2864 from 100000\n",
      "loss: 0.002436695516706777\n",
      "epoch:2865 from 100000\n",
      "loss: 0.003045214620868819\n",
      "epoch:2866 from 100000\n",
      "loss: 0.0025460223571656115\n",
      "epoch:2867 from 100000\n",
      "loss: 0.0022150894069454807\n",
      "epoch:2868 from 100000\n",
      "loss: 0.0024190334905824784\n",
      "epoch:2869 from 100000\n",
      "loss: 0.0024794183237304424\n",
      "epoch:2870 from 100000\n",
      "loss: 0.0028128252887142935\n",
      "epoch:2871 from 100000\n",
      "loss: 0.003104230989208645\n",
      "epoch:2872 from 100000\n",
      "loss: 0.002911282000184201\n",
      "epoch:2873 from 100000\n",
      "loss: 0.0022937841142578463\n",
      "epoch:2874 from 100000\n",
      "loss: 0.002257103090101964\n",
      "epoch:2875 from 100000\n",
      "loss: 0.0027326153673215003\n",
      "epoch:2876 from 100000\n",
      "loss: 0.0023577738342055454\n",
      "epoch:2877 from 100000\n",
      "loss: 0.001957489993458239\n",
      "epoch:2878 from 100000\n",
      "loss: 0.002450954745938816\n",
      "epoch:2879 from 100000\n",
      "loss: 0.0028949707045455852\n",
      "epoch:2880 from 100000\n",
      "loss: 0.0023649542187911266\n",
      "epoch:2881 from 100000\n",
      "loss: 0.002145526922163299\n",
      "epoch:2882 from 100000\n",
      "loss: 0.0023839303530053257\n",
      "epoch:2883 from 100000\n",
      "loss: 0.0022556265761295435\n",
      "epoch:2884 from 100000\n",
      "loss: 0.0024330709936748462\n",
      "epoch:2885 from 100000\n",
      "loss: 0.0024564157089344008\n",
      "epoch:2886 from 100000\n",
      "loss: 0.0033408452065941674\n",
      "epoch:2887 from 100000\n",
      "loss: 0.0027548061111133393\n",
      "epoch:2888 from 100000\n",
      "loss: 0.0022466468159336728\n",
      "epoch:2889 from 100000\n",
      "loss: 0.0020829410837848172\n",
      "epoch:2890 from 100000\n",
      "loss: 0.002818690279468683\n",
      "epoch:2891 from 100000\n",
      "loss: 0.0028028361867313833\n",
      "epoch:2892 from 100000\n",
      "loss: 0.0025563068399492295\n",
      "epoch:2893 from 100000\n",
      "loss: 0.002576441011370889\n",
      "epoch:2894 from 100000\n",
      "loss: 0.0027190148301485806\n",
      "epoch:2895 from 100000\n",
      "loss: 0.0017931813240466\n",
      "epoch:2896 from 100000\n",
      "loss: 0.002767948246457763\n",
      "epoch:2897 from 100000\n",
      "loss: 0.0027790251104988783\n",
      "epoch:2898 from 100000\n",
      "loss: 0.0023956282485073643\n",
      "epoch:2899 from 100000\n",
      "loss: 0.002529090693656144\n",
      "epoch:2900 from 100000\n",
      "loss: 0.002351920350264436\n",
      "epoch:2901 from 100000\n",
      "loss: 0.0025676530456735236\n",
      "epoch:2902 from 100000\n",
      "loss: 0.002220749939453981\n",
      "epoch:2903 from 100000\n",
      "loss: 0.0025092441536003207\n",
      "epoch:2904 from 100000\n",
      "loss: 0.0025030215635695185\n",
      "epoch:2905 from 100000\n",
      "loss: 0.002632673745893044\n",
      "epoch:2906 from 100000\n",
      "loss: 0.0025629071791541946\n",
      "epoch:2907 from 100000\n",
      "loss: 0.0022093142030286117\n",
      "epoch:2908 from 100000\n",
      "loss: 0.002558874953706436\n",
      "epoch:2909 from 100000\n",
      "loss: 0.0024164847295082836\n",
      "epoch:2910 from 100000\n",
      "loss: 0.0035739752231620514\n",
      "epoch:2911 from 100000\n",
      "loss: 0.002433595202774086\n",
      "epoch:2912 from 100000\n",
      "loss: 0.0027370944567053357\n",
      "epoch:2913 from 100000\n",
      "loss: 0.0023760344496110974\n",
      "epoch:2914 from 100000\n",
      "loss: 0.002144889642011094\n",
      "epoch:2915 from 100000\n",
      "loss: 0.0028724097366819024\n",
      "epoch:2916 from 100000\n",
      "loss: 0.0018720048286979818\n",
      "epoch:2917 from 100000\n",
      "loss: 0.0025057950519453698\n",
      "epoch:2918 from 100000\n",
      "loss: 0.0021683425293304026\n",
      "epoch:2919 from 100000\n",
      "loss: 0.002252749072141569\n",
      "epoch:2920 from 100000\n",
      "loss: 0.0025474479646732107\n",
      "epoch:2921 from 100000\n",
      "loss: 0.0028907519212282295\n",
      "epoch:2922 from 100000\n",
      "loss: 0.002456650572810304\n",
      "epoch:2923 from 100000\n",
      "loss: 0.0028023629325844013\n",
      "epoch:2924 from 100000\n",
      "loss: 0.003055162728941241\n",
      "epoch:2925 from 100000\n",
      "loss: 0.002430174225224846\n",
      "epoch:2926 from 100000\n",
      "loss: 0.0023411052051756087\n",
      "epoch:2927 from 100000\n",
      "loss: 0.0024168788646647954\n",
      "epoch:2928 from 100000\n",
      "loss: 0.0023426053028003935\n",
      "epoch:2929 from 100000\n",
      "loss: 0.002399762039487996\n",
      "epoch:2930 from 100000\n",
      "loss: 0.0025133830287416004\n",
      "epoch:2931 from 100000\n",
      "loss: 0.0029486225698178714\n",
      "epoch:2932 from 100000\n",
      "loss: 0.002197280586699435\n",
      "epoch:2933 from 100000\n",
      "loss: 0.0028875650178292744\n",
      "epoch:2934 from 100000\n",
      "loss: 0.0023442991825959325\n",
      "epoch:2935 from 100000\n",
      "loss: 0.0022526746956543775\n",
      "epoch:2936 from 100000\n",
      "loss: 0.002276842819810554\n",
      "epoch:2937 from 100000\n",
      "loss: 0.002168350552665288\n",
      "epoch:2938 from 100000\n",
      "loss: 0.002953812926932642\n",
      "epoch:2939 from 100000\n",
      "loss: 0.00261368526099659\n",
      "epoch:2940 from 100000\n",
      "loss: 0.002298421253840388\n",
      "epoch:2941 from 100000\n",
      "loss: 0.0026888413912899978\n",
      "epoch:2942 from 100000\n",
      "loss: 0.002625505440221332\n",
      "epoch:2943 from 100000\n",
      "loss: 0.002711012828882531\n",
      "epoch:2944 from 100000\n",
      "loss: 0.002392073763701319\n",
      "epoch:2945 from 100000\n",
      "loss: 0.0019501422980283678\n",
      "epoch:2946 from 100000\n",
      "loss: 0.002258563603732\n",
      "epoch:2947 from 100000\n",
      "loss: 0.002306510803805395\n",
      "epoch:2948 from 100000\n",
      "loss: 0.00268254397445844\n",
      "epoch:2949 from 100000\n",
      "loss: 0.002457286480108678\n",
      "epoch:2950 from 100000\n",
      "loss: 0.002352150937042552\n",
      "epoch:2951 from 100000\n",
      "loss: 0.002390064754507648\n",
      "epoch:2952 from 100000\n",
      "loss: 0.0023196100262339314\n",
      "epoch:2953 from 100000\n",
      "loss: 0.002594460349541805\n",
      "epoch:2954 from 100000\n",
      "loss: 0.0026284151757067775\n",
      "epoch:2955 from 100000\n",
      "loss: 0.002488280474921112\n",
      "epoch:2956 from 100000\n",
      "loss: 0.002621313866058017\n",
      "epoch:2957 from 100000\n",
      "loss: 0.00274538716769257\n",
      "epoch:2958 from 100000\n",
      "loss: 0.002122554383163333\n",
      "epoch:2959 from 100000\n",
      "loss: 0.0028900269151677094\n",
      "epoch:2960 from 100000\n",
      "loss: 0.002253790828063984\n",
      "epoch:2961 from 100000\n",
      "loss: 0.00257619816255783\n",
      "epoch:2962 from 100000\n",
      "loss: 0.002217290037094699\n",
      "epoch:2963 from 100000\n",
      "loss: 0.0025560023498769624\n",
      "epoch:2964 from 100000\n",
      "loss: 0.0024023647973194784\n",
      "epoch:2965 from 100000\n",
      "loss: 0.002697575270900643\n",
      "epoch:2966 from 100000\n",
      "loss: 0.002959951396718452\n",
      "epoch:2967 from 100000\n",
      "loss: 0.0022402799912697446\n",
      "epoch:2968 from 100000\n",
      "loss: 0.002567968551318245\n",
      "epoch:2969 from 100000\n",
      "loss: 0.0028544387111537617\n",
      "epoch:2970 from 100000\n",
      "loss: 0.002851174300872117\n",
      "epoch:2971 from 100000\n",
      "loss: 0.0021464472654031397\n",
      "epoch:2972 from 100000\n",
      "loss: 0.0022492964465925525\n",
      "epoch:2973 from 100000\n",
      "loss: 0.002495582745154934\n",
      "retrying for data\n",
      "epoch:2974 from 100000\n",
      "loss: 0.0023584309611237586\n",
      "epoch:2975 from 100000\n",
      "loss: 0.0027130592654316388\n",
      "epoch:2976 from 100000\n",
      "loss: 0.002301461445767927\n",
      "epoch:2977 from 100000\n",
      "loss: 0.0024776511676236623\n",
      "epoch:2978 from 100000\n",
      "loss: 0.0023052567592287687\n",
      "epoch:2979 from 100000\n",
      "loss: 0.0025513855989061085\n",
      "epoch:2980 from 100000\n",
      "loss: 0.002152008130650529\n",
      "epoch:2981 from 100000\n",
      "loss: 0.0028291210466875327\n",
      "epoch:2982 from 100000\n",
      "loss: 0.002647797182675049\n",
      "epoch:2983 from 100000\n",
      "loss: 0.0020609268748614795\n",
      "epoch:2984 from 100000\n",
      "loss: 0.0027288011303312487\n",
      "epoch:2985 from 100000\n",
      "loss: 0.002409787913961736\n",
      "epoch:2986 from 100000\n",
      "loss: 0.002388318001692369\n",
      "epoch:2987 from 100000\n",
      "loss: 0.0022978741581596296\n",
      "epoch:2988 from 100000\n",
      "loss: 0.0021504538184444755\n",
      "epoch:2989 from 100000\n",
      "loss: 0.002476485596588418\n",
      "epoch:2990 from 100000\n",
      "loss: 0.002241102303315756\n",
      "epoch:2991 from 100000\n",
      "loss: 0.0026174759557022753\n",
      "epoch:2992 from 100000\n",
      "loss: 0.0025484100708581536\n",
      "epoch:2993 from 100000\n",
      "loss: 0.002800586006060257\n",
      "epoch:2994 from 100000\n",
      "loss: 0.0024311539087022993\n",
      "epoch:2995 from 100000\n",
      "loss: 0.0026742541199382686\n",
      "epoch:2996 from 100000\n",
      "loss: 0.002985310304580935\n",
      "epoch:2997 from 100000\n",
      "loss: 0.002021762597223642\n",
      "epoch:2998 from 100000\n",
      "loss: 0.0022710557519758368\n",
      "epoch:2999 from 100000\n",
      "loss: 0.0025157811217582093\n",
      "epoch:3000 from 100000\n",
      "loss: 0.0025755843478236784\n",
      "epoch:3001 from 100000\n",
      "loss: 0.0022289470291525504\n",
      "epoch:3002 from 100000\n",
      "loss: 0.0022079486503372436\n",
      "epoch:3003 from 100000\n",
      "loss: 0.0021414145439848653\n",
      "epoch:3004 from 100000\n",
      "loss: 0.0024910003824061278\n",
      "epoch:3005 from 100000\n",
      "loss: 0.0024467240536694135\n",
      "retrying for data\n",
      "epoch:3006 from 100000\n",
      "loss: 0.0027785154460389094\n",
      "epoch:3007 from 100000\n",
      "loss: 0.001955468787372183\n",
      "epoch:3008 from 100000\n",
      "loss: 0.0025921031256359583\n",
      "epoch:3009 from 100000\n",
      "loss: 0.002588835550341173\n",
      "epoch:3010 from 100000\n",
      "loss: 0.0025961943002812404\n",
      "epoch:3011 from 100000\n",
      "loss: 0.0023925840633012285\n",
      "epoch:3012 from 100000\n",
      "loss: 0.0022197899053253423\n",
      "epoch:3013 from 100000\n",
      "loss: 0.002735106765896944\n",
      "epoch:3014 from 100000\n",
      "loss: 0.0028213163042778433\n",
      "epoch:3015 from 100000\n",
      "loss: 0.0020221089133798387\n",
      "epoch:3016 from 100000\n",
      "loss: 0.0028282521427911433\n",
      "epoch:3017 from 100000\n",
      "loss: 0.0032130110978982884\n",
      "epoch:3018 from 100000\n",
      "loss: 0.002491271663416228\n",
      "epoch:3019 from 100000\n",
      "loss: 0.0024494479291661264\n",
      "epoch:3020 from 100000\n",
      "loss: 0.002281570665786603\n",
      "epoch:3021 from 100000\n",
      "loss: 0.002256165491274942\n",
      "epoch:3022 from 100000\n",
      "loss: 0.0026347776314423754\n",
      "epoch:3023 from 100000\n",
      "loss: 0.0027510620826376453\n",
      "epoch:3024 from 100000\n",
      "loss: 0.0029671577534600146\n",
      "epoch:3025 from 100000\n",
      "loss: 0.002614571243810815\n",
      "epoch:3026 from 100000\n",
      "loss: 0.0018595409880930447\n",
      "epoch:3027 from 100000\n",
      "loss: 0.0022810171163314408\n",
      "epoch:3028 from 100000\n",
      "loss: 0.002501271501074598\n",
      "epoch:3029 from 100000\n",
      "loss: 0.002402854757320938\n",
      "epoch:3030 from 100000\n",
      "loss: 0.0024455950770061463\n",
      "epoch:3031 from 100000\n",
      "loss: 0.0025801087385275423\n",
      "epoch:3032 from 100000\n",
      "loss: 0.002442034228607781\n",
      "epoch:3033 from 100000\n",
      "loss: 0.002474735505359149\n",
      "epoch:3034 from 100000\n",
      "loss: 0.0025268669670595045\n",
      "epoch:3035 from 100000\n",
      "loss: 0.002626755113212198\n",
      "epoch:3036 from 100000\n",
      "loss: 0.003107849814114161\n",
      "epoch:3037 from 100000\n",
      "loss: 0.002576634458563376\n",
      "epoch:3038 from 100000\n",
      "loss: 0.002432747203584995\n",
      "epoch:3039 from 100000\n",
      "loss: 0.00207923352763828\n",
      "epoch:3040 from 100000\n",
      "loss: 0.0024110899738616354\n",
      "epoch:3041 from 100000\n",
      "loss: 0.002364190383531195\n",
      "epoch:3042 from 100000\n",
      "loss: 0.0023323751281054683\n",
      "epoch:3043 from 100000\n",
      "loss: 0.0026544719546279794\n",
      "epoch:3044 from 100000\n",
      "loss: 0.00227606676060077\n",
      "epoch:3045 from 100000\n",
      "loss: 0.002460514670417524\n",
      "epoch:3046 from 100000\n",
      "loss: 0.002677648376788966\n",
      "epoch:3047 from 100000\n",
      "loss: 0.002527499811492362\n",
      "epoch:3048 from 100000\n",
      "loss: 0.002253773893187372\n",
      "epoch:3049 from 100000\n",
      "loss: 0.0027142726564193254\n",
      "epoch:3050 from 100000\n",
      "loss: 0.001947587776498949\n",
      "epoch:3051 from 100000\n",
      "loss: 0.002522802200999763\n",
      "epoch:3052 from 100000\n",
      "loss: 0.0023569167809966984\n",
      "epoch:3053 from 100000\n",
      "loss: 0.0020533222154881514\n",
      "epoch:3054 from 100000\n",
      "loss: 0.002699353062268983\n",
      "epoch:3055 from 100000\n",
      "loss: 0.002695299445662158\n",
      "epoch:3056 from 100000\n",
      "loss: 0.0023487324185964553\n",
      "epoch:3057 from 100000\n",
      "loss: 0.002580606864000856\n",
      "epoch:3058 from 100000\n",
      "loss: 0.002608695092362723\n",
      "epoch:3059 from 100000\n",
      "loss: 0.0025141115113740398\n",
      "epoch:3060 from 100000\n",
      "loss: 0.002861595940771622\n",
      "epoch:3061 from 100000\n",
      "loss: 0.0027073251787328445\n",
      "epoch:3062 from 100000\n",
      "loss: 0.0028307853609987887\n",
      "epoch:3063 from 100000\n",
      "loss: 0.0025437621844162095\n",
      "epoch:3064 from 100000\n",
      "loss: 0.0026156085890818304\n",
      "epoch:3065 from 100000\n",
      "loss: 0.0022568533483138253\n",
      "epoch:3066 from 100000\n",
      "loss: 0.002834674743354526\n",
      "epoch:3067 from 100000\n",
      "loss: 0.0023850849562450094\n",
      "epoch:3068 from 100000\n",
      "loss: 0.0023294380279565985\n",
      "epoch:3069 from 100000\n",
      "loss: 0.0022234170657498\n",
      "epoch:3070 from 100000\n",
      "loss: 0.00283599433406323\n",
      "epoch:3071 from 100000\n",
      "loss: 0.0021967545011989387\n",
      "epoch:3072 from 100000\n",
      "loss: 0.0020615477536978233\n",
      "epoch:3073 from 100000\n",
      "loss: 0.0024855038836051335\n",
      "epoch:3074 from 100000\n",
      "loss: 0.002714960409463174\n",
      "epoch:3075 from 100000\n",
      "loss: 0.0024541949657930218\n",
      "epoch:3076 from 100000\n",
      "loss: 0.002371959867915052\n",
      "epoch:3077 from 100000\n",
      "loss: 0.0026780139028801386\n",
      "epoch:3078 from 100000\n",
      "loss: 0.002110932043848379\n",
      "epoch:3079 from 100000\n",
      "loss: 0.002375114193966965\n",
      "epoch:3080 from 100000\n",
      "loss: 0.002355588357374927\n",
      "epoch:3081 from 100000\n",
      "loss: 0.002484989129641235\n",
      "epoch:3082 from 100000\n",
      "loss: 0.001909058545066955\n",
      "epoch:3083 from 100000\n",
      "loss: 0.002198505812117446\n",
      "epoch:3084 from 100000\n",
      "loss: 0.0025072656771953916\n",
      "epoch:3085 from 100000\n",
      "loss: 0.0024757102027592737\n",
      "epoch:3086 from 100000\n",
      "loss: 0.003244206370027314\n",
      "epoch:3087 from 100000\n",
      "loss: 0.0025906354620985894\n",
      "epoch:3088 from 100000\n",
      "loss: 0.0028197029807586205\n",
      "epoch:3089 from 100000\n",
      "loss: 0.0027549000631665876\n",
      "epoch:3090 from 100000\n",
      "loss: 0.0026209380046253727\n",
      "epoch:3091 from 100000\n",
      "loss: 0.0020099907666235595\n",
      "epoch:3092 from 100000\n",
      "loss: 0.0028802594471244447\n",
      "epoch:3093 from 100000\n",
      "loss: 0.0023390367225033515\n",
      "epoch:3094 from 100000\n",
      "loss: 0.0025892524552091345\n",
      "epoch:3095 from 100000\n",
      "loss: 0.0025318431901268923\n",
      "retrying for data\n",
      "epoch:3096 from 100000\n",
      "loss: 0.0025309852426858015\n",
      "epoch:3097 from 100000\n",
      "loss: 0.002492656921560865\n",
      "epoch:3098 from 100000\n",
      "loss: 0.0025334492318620505\n",
      "epoch:3099 from 100000\n",
      "loss: 0.0024627768338518763\n",
      "epoch:3100 from 100000\n",
      "loss: 0.0022180505942515083\n",
      "epoch:3101 from 100000\n",
      "loss: 0.0021676692745842274\n",
      "epoch:3102 from 100000\n",
      "loss: 0.0023345360787629943\n",
      "epoch:3103 from 100000\n",
      "loss: 0.0024701878886617123\n",
      "epoch:3104 from 100000\n",
      "loss: 0.002000819409175847\n",
      "epoch:3105 from 100000\n",
      "loss: 0.0023215510151999297\n",
      "epoch:3106 from 100000\n",
      "loss: 0.0027094872297936945\n",
      "epoch:3107 from 100000\n",
      "loss: 0.002104081513067513\n",
      "epoch:3108 from 100000\n",
      "loss: 0.0027346555372389503\n",
      "epoch:3109 from 100000\n",
      "loss: 0.002548138345730422\n",
      "epoch:3110 from 100000\n",
      "loss: 0.002955175454914638\n",
      "epoch:3111 from 100000\n",
      "loss: 0.002724270444105059\n",
      "epoch:3112 from 100000\n",
      "loss: 0.00218475325681311\n",
      "epoch:3113 from 100000\n",
      "loss: 0.0023841425418140716\n",
      "epoch:3114 from 100000\n",
      "loss: 0.0019145124829549331\n",
      "epoch:3115 from 100000\n",
      "loss: 0.0031108208284535976\n",
      "epoch:3116 from 100000\n",
      "loss: 0.002757131928291301\n",
      "epoch:3117 from 100000\n",
      "loss: 0.002506797350093848\n",
      "epoch:3118 from 100000\n",
      "loss: 0.0023224828991885715\n",
      "epoch:3119 from 100000\n",
      "loss: 0.0024341952798749844\n",
      "epoch:3120 from 100000\n",
      "loss: 0.002604121356057476\n",
      "epoch:3121 from 100000\n",
      "loss: 0.002819065731955561\n",
      "epoch:3122 from 100000\n",
      "loss: 0.0026309024007673543\n",
      "epoch:3123 from 100000\n",
      "loss: 0.0027498822244922394\n",
      "epoch:3124 from 100000\n",
      "loss: 0.002127349853708438\n",
      "epoch:3125 from 100000\n",
      "loss: 0.0025567181369865466\n",
      "epoch:3126 from 100000\n",
      "loss: 0.002160521368097079\n",
      "epoch:3127 from 100000\n",
      "loss: 0.002997499729303854\n",
      "epoch:3128 from 100000\n",
      "loss: 0.00227408675604579\n",
      "epoch:3129 from 100000\n",
      "loss: 0.002233065475010676\n",
      "epoch:3130 from 100000\n",
      "loss: 0.0020362335642403195\n",
      "epoch:3131 from 100000\n",
      "loss: 0.002283225160681468\n",
      "epoch:3132 from 100000\n",
      "loss: 0.0027369596779749372\n",
      "epoch:3133 from 100000\n",
      "loss: 0.0025654327213544548\n",
      "epoch:3134 from 100000\n",
      "loss: 0.002410675854235933\n",
      "epoch:3135 from 100000\n",
      "loss: 0.0021230653679822353\n",
      "epoch:3136 from 100000\n",
      "loss: 0.0027827196397964826\n",
      "epoch:3137 from 100000\n",
      "loss: 0.0024521338972363083\n",
      "epoch:3138 from 100000\n",
      "loss: 0.002930033103950791\n",
      "epoch:3139 from 100000\n",
      "loss: 0.0026877764869368548\n",
      "epoch:3140 from 100000\n",
      "loss: 0.002138467980898895\n",
      "epoch:3141 from 100000\n",
      "loss: 0.0020399780983098026\n",
      "epoch:3142 from 100000\n",
      "loss: 0.0020075206531373624\n",
      "epoch:3143 from 100000\n",
      "loss: 0.0018430701899205815\n",
      "epoch:3144 from 100000\n",
      "loss: 0.0025786801825091743\n",
      "epoch:3145 from 100000\n",
      "loss: 0.003333079417643603\n",
      "epoch:3146 from 100000\n",
      "loss: 0.0025447482196341298\n",
      "epoch:3147 from 100000\n",
      "loss: 0.0026470816147252663\n",
      "epoch:3148 from 100000\n",
      "loss: 0.0020707716813035404\n",
      "epoch:3149 from 100000\n",
      "loss: 0.00300413155497381\n",
      "epoch:3150 from 100000\n",
      "loss: 0.0029731893224038686\n",
      "epoch:3151 from 100000\n",
      "loss: 0.002813560006586613\n",
      "retrying for data\n",
      "epoch:3152 from 100000\n",
      "loss: 0.00255337963568536\n",
      "epoch:3153 from 100000\n",
      "loss: 0.0024465090071430495\n",
      "epoch:3154 from 100000\n",
      "loss: 0.0026059438129095724\n",
      "epoch:3155 from 100000\n",
      "loss: 0.0033451481027952923\n",
      "epoch:3156 from 100000\n",
      "loss: 0.002348181523700532\n",
      "epoch:3157 from 100000\n",
      "loss: 0.0023081256541388484\n",
      "epoch:3158 from 100000\n",
      "loss: 0.0027287795859933794\n",
      "epoch:3159 from 100000\n",
      "loss: 0.002386960171264718\n",
      "epoch:3160 from 100000\n",
      "loss: 0.002196674954703326\n",
      "epoch:3161 from 100000\n",
      "loss: 0.0020979717886007165\n",
      "epoch:3162 from 100000\n",
      "loss: 0.0026729476385298767\n",
      "epoch:3163 from 100000\n",
      "loss: 0.002582372910609365\n",
      "epoch:3164 from 100000\n",
      "loss: 0.002726734384111751\n",
      "epoch:3165 from 100000\n",
      "loss: 0.0025888632416695145\n",
      "epoch:3166 from 100000\n",
      "loss: 0.00218016796259235\n",
      "epoch:3167 from 100000\n",
      "loss: 0.0027109376300700205\n",
      "epoch:3168 from 100000\n",
      "loss: 0.002847688602969356\n",
      "epoch:3169 from 100000\n",
      "loss: 0.0019986444241339996\n",
      "epoch:3170 from 100000\n",
      "loss: 0.0028946097804691817\n",
      "epoch:3171 from 100000\n",
      "loss: 0.002285244456061264\n",
      "epoch:3172 from 100000\n",
      "loss: 0.002650431672890363\n",
      "epoch:3173 from 100000\n",
      "loss: 0.0021833882205442023\n",
      "epoch:3174 from 100000\n",
      "loss: 0.002838083895824184\n",
      "epoch:3175 from 100000\n",
      "loss: 0.0021199998426482125\n",
      "epoch:3176 from 100000\n",
      "loss: 0.002073381568749255\n",
      "epoch:3177 from 100000\n",
      "loss: 0.003093981185827488\n",
      "epoch:3178 from 100000\n",
      "loss: 0.00266719142767613\n",
      "epoch:3179 from 100000\n",
      "loss: 0.0021869616218737065\n",
      "epoch:3180 from 100000\n",
      "loss: 0.0021508929548303968\n",
      "epoch:3181 from 100000\n",
      "loss: 0.0025464991536523485\n",
      "epoch:3182 from 100000\n",
      "loss: 0.0022374691561140025\n",
      "epoch:3183 from 100000\n",
      "loss: 0.002203102113838895\n",
      "epoch:3184 from 100000\n",
      "loss: 0.0029618663947985624\n",
      "epoch:3185 from 100000\n",
      "loss: 0.0023430738125682637\n",
      "epoch:3186 from 100000\n",
      "loss: 0.00245362848215791\n",
      "epoch:3187 from 100000\n",
      "loss: 0.0028454316633883536\n",
      "epoch:3188 from 100000\n",
      "loss: 0.00243803684993793\n",
      "epoch:3189 from 100000\n",
      "loss: 0.0025041356450685726\n",
      "epoch:3190 from 100000\n",
      "loss: 0.002387354092121541\n",
      "epoch:3191 from 100000\n",
      "loss: 0.0022509755644080087\n",
      "epoch:3192 from 100000\n",
      "loss: 0.0021475374672377257\n",
      "epoch:3193 from 100000\n",
      "loss: 0.0033069141994133133\n",
      "epoch:3194 from 100000\n",
      "loss: 0.0033212734058452043\n",
      "epoch:3195 from 100000\n",
      "loss: 0.0020940117173040562\n",
      "epoch:3196 from 100000\n",
      "loss: 0.0027879294776766983\n",
      "epoch:3197 from 100000\n",
      "loss: 0.002498203584593739\n",
      "epoch:3198 from 100000\n",
      "loss: 0.0027488155342609844\n",
      "epoch:3199 from 100000\n",
      "loss: 0.0029218669746455816\n",
      "epoch:3200 from 100000\n",
      "loss: 0.00265758437325303\n",
      "epoch:3201 from 100000\n",
      "loss: 0.0024659768622967704\n",
      "epoch:3202 from 100000\n",
      "loss: 0.0025874251689401717\n",
      "epoch:3203 from 100000\n",
      "loss: 0.0024143141388321965\n",
      "retrying for data\n",
      "epoch:3204 from 100000\n",
      "loss: 0.0023209585553161105\n",
      "epoch:3205 from 100000\n",
      "loss: 0.002583065369265114\n",
      "epoch:3206 from 100000\n",
      "loss: 0.0022222541648488914\n",
      "epoch:3207 from 100000\n",
      "loss: 0.0024794888176415952\n",
      "epoch:3208 from 100000\n",
      "loss: 0.0028813873057060846\n",
      "epoch:3209 from 100000\n",
      "loss: 0.0019247702482800833\n",
      "epoch:3210 from 100000\n",
      "loss: 0.0023551743837799677\n",
      "epoch:3211 from 100000\n",
      "loss: 0.0023583213699680527\n",
      "epoch:3212 from 100000\n",
      "loss: 0.002485573101751015\n",
      "epoch:3213 from 100000\n",
      "loss: 0.0023563822250878275\n",
      "epoch:3214 from 100000\n",
      "loss: 0.0021090343556977587\n",
      "epoch:3215 from 100000\n",
      "loss: 0.0025825682720892473\n",
      "epoch:3216 from 100000\n",
      "loss: 0.002395260464339799\n",
      "epoch:3217 from 100000\n",
      "loss: 0.002197715101885933\n",
      "epoch:3218 from 100000\n",
      "loss: 0.0026774727497240747\n",
      "epoch:3219 from 100000\n",
      "loss: 0.0023830914229563405\n",
      "epoch:3220 from 100000\n",
      "loss: 0.0031102177772766026\n",
      "epoch:3221 from 100000\n",
      "loss: 0.002429884202882704\n",
      "epoch:3222 from 100000\n",
      "loss: 0.0022816416338287127\n",
      "epoch:3223 from 100000\n",
      "loss: 0.002286902877074226\n",
      "epoch:3224 from 100000\n",
      "loss: 0.0027310358760246345\n",
      "epoch:3225 from 100000\n",
      "loss: 0.002727220413305531\n",
      "epoch:3226 from 100000\n",
      "loss: 0.002116054606986495\n",
      "epoch:3227 from 100000\n",
      "loss: 0.0025487773806815994\n",
      "epoch:3228 from 100000\n",
      "loss: 0.002477292489345473\n",
      "epoch:3229 from 100000\n",
      "loss: 0.00198849369323284\n",
      "epoch:3230 from 100000\n",
      "loss: 0.0018898601655479297\n",
      "epoch:3231 from 100000\n",
      "loss: 0.002008201063489423\n",
      "epoch:3232 from 100000\n",
      "loss: 0.0024515986389133104\n",
      "epoch:3233 from 100000\n",
      "loss: 0.0025560952067564813\n",
      "epoch:3234 from 100000\n",
      "loss: 0.0025435437551948326\n",
      "epoch:3235 from 100000\n",
      "loss: 0.0027453049998769075\n",
      "epoch:3236 from 100000\n",
      "loss: 0.0023221124753831646\n",
      "epoch:3237 from 100000\n",
      "loss: 0.0022541205673292097\n",
      "epoch:3238 from 100000\n",
      "loss: 0.0022329380799419596\n",
      "epoch:3239 from 100000\n",
      "loss: 0.0020872010969412713\n",
      "epoch:3240 from 100000\n",
      "loss: 0.0019040850977489754\n",
      "epoch:3241 from 100000\n",
      "loss: 0.0026019491393469707\n",
      "epoch:3242 from 100000\n",
      "loss: 0.002448398358353643\n",
      "epoch:3243 from 100000\n",
      "loss: 0.002476488916300923\n",
      "epoch:3244 from 100000\n",
      "loss: 0.002302560493404826\n",
      "epoch:3245 from 100000\n",
      "loss: 0.0025103123738574595\n",
      "epoch:3246 from 100000\n",
      "loss: 0.0023032241556393274\n",
      "epoch:3247 from 100000\n",
      "loss: 0.0021815048546045546\n",
      "epoch:3248 from 100000\n",
      "loss: 0.002709911708763002\n",
      "epoch:3249 from 100000\n",
      "loss: 0.00248008756955187\n",
      "epoch:3250 from 100000\n",
      "loss: 0.0030793932023271964\n",
      "epoch:3251 from 100000\n",
      "loss: 0.002673432823087296\n",
      "epoch:3252 from 100000\n",
      "loss: 0.002576915224125287\n",
      "epoch:3253 from 100000\n",
      "loss: 0.002701446087911563\n",
      "epoch:3254 from 100000\n",
      "loss: 0.0028727764768916586\n",
      "epoch:3255 from 100000\n",
      "loss: 0.0024264903035486896\n",
      "epoch:3256 from 100000\n",
      "loss: 0.0024275759632246263\n",
      "epoch:3257 from 100000\n",
      "loss: 0.0025857403170732596\n",
      "epoch:3258 from 100000\n",
      "loss: 0.002059007920422573\n",
      "epoch:3259 from 100000\n",
      "loss: 0.002673279144630669\n",
      "epoch:3260 from 100000\n",
      "loss: 0.002492633644266107\n",
      "retrying for data\n",
      "epoch:3261 from 100000\n",
      "loss: 0.002731000003961981\n",
      "epoch:3262 from 100000\n",
      "loss: 0.0022359642422884463\n",
      "epoch:3263 from 100000\n",
      "loss: 0.0023378309984707357\n",
      "epoch:3264 from 100000\n",
      "loss: 0.0020474739249891627\n",
      "epoch:3265 from 100000\n",
      "loss: 0.0022763613864356103\n",
      "epoch:3266 from 100000\n",
      "loss: 0.00259403763467958\n",
      "epoch:3267 from 100000\n",
      "loss: 0.002292820418915653\n",
      "epoch:3268 from 100000\n",
      "loss: 0.0026131542188068124\n",
      "epoch:3269 from 100000\n",
      "loss: 0.002779534678154505\n",
      "epoch:3270 from 100000\n",
      "loss: 0.0024019153325269826\n",
      "epoch:3271 from 100000\n",
      "loss: 0.0021483976998979415\n",
      "epoch:3272 from 100000\n",
      "loss: 0.0026993061195526025\n",
      "epoch:3273 from 100000\n",
      "loss: 0.002737160519501458\n",
      "epoch:3274 from 100000\n",
      "loss: 0.0020735117223580346\n",
      "epoch:3275 from 100000\n",
      "loss: 0.0022476630998937708\n",
      "epoch:3276 from 100000\n",
      "loss: 0.0024496616702833762\n",
      "epoch:3277 from 100000\n",
      "loss: 0.002364419726035294\n",
      "epoch:3278 from 100000\n",
      "loss: 0.002612924593847765\n",
      "epoch:3279 from 100000\n",
      "loss: 0.0025433817829139116\n",
      "epoch:3280 from 100000\n",
      "loss: 0.002147238527470563\n",
      "epoch:3281 from 100000\n",
      "loss: 0.0026115188019275593\n",
      "epoch:3282 from 100000\n",
      "loss: 0.002525071018851577\n",
      "epoch:3283 from 100000\n",
      "loss: 0.0027844092044517765\n",
      "epoch:3284 from 100000\n",
      "loss: 0.002665162734331261\n",
      "epoch:3285 from 100000\n",
      "loss: 0.002089624982488658\n",
      "epoch:3286 from 100000\n",
      "loss: 0.0026371693872988544\n",
      "epoch:3287 from 100000\n",
      "loss: 0.0020245634810294177\n",
      "epoch:3288 from 100000\n",
      "loss: 0.0030923505115509897\n",
      "epoch:3289 from 100000\n",
      "loss: 0.002575985045439211\n",
      "epoch:3290 from 100000\n",
      "loss: 0.0025832892877417635\n",
      "epoch:3291 from 100000\n",
      "loss: 0.0023507860390452606\n",
      "epoch:3292 from 100000\n",
      "loss: 0.0026401360591705725\n",
      "epoch:3293 from 100000\n",
      "loss: 0.0023582216674071788\n",
      "epoch:3294 from 100000\n",
      "loss: 0.0026590759497082672\n",
      "epoch:3295 from 100000\n",
      "loss: 0.0021330631475962036\n",
      "epoch:3296 from 100000\n",
      "loss: 0.0029600031250822667\n",
      "epoch:3297 from 100000\n",
      "loss: 0.002857556028345698\n",
      "epoch:3298 from 100000\n",
      "loss: 0.002340376836230007\n",
      "epoch:3299 from 100000\n",
      "loss: 0.002166026897214124\n",
      "epoch:3300 from 100000\n",
      "loss: 0.002803778803922796\n",
      "epoch:3301 from 100000\n",
      "loss: 0.002552643184600356\n",
      "epoch:3302 from 100000\n",
      "loss: 0.0025145255845870906\n",
      "epoch:3303 from 100000\n",
      "loss: 0.00213699779237686\n",
      "epoch:3304 from 100000\n",
      "loss: 0.0028046800691043927\n",
      "epoch:3305 from 100000\n",
      "loss: 0.002404481184143492\n",
      "epoch:3306 from 100000\n",
      "loss: 0.0023307166278527802\n",
      "epoch:3307 from 100000\n",
      "loss: 0.003095604381655903\n",
      "epoch:3308 from 100000\n",
      "loss: 0.0020855153122170123\n",
      "epoch:3309 from 100000\n",
      "loss: 0.0024235792580498128\n",
      "epoch:3310 from 100000\n",
      "loss: 0.002505543056543047\n",
      "epoch:3311 from 100000\n",
      "loss: 0.0022934249662966977\n",
      "epoch:3312 from 100000\n",
      "loss: 0.0020180004730718792\n",
      "epoch:3313 from 100000\n",
      "loss: 0.0024636090577132563\n",
      "epoch:3314 from 100000\n",
      "loss: 0.0025909687728074005\n",
      "epoch:3315 from 100000\n",
      "loss: 0.0018548967049838438\n",
      "epoch:3316 from 100000\n",
      "loss: 0.0021848546915066436\n",
      "epoch:3317 from 100000\n",
      "loss: 0.002683200196031521\n",
      "epoch:3318 from 100000\n",
      "loss: 0.002451663528518111\n",
      "epoch:3319 from 100000\n",
      "loss: 0.0024470665924525292\n",
      "epoch:3320 from 100000\n",
      "loss: 0.001853056420543453\n",
      "epoch:3321 from 100000\n",
      "loss: 0.002056031313600215\n",
      "epoch:3322 from 100000\n",
      "loss: 0.0025621157172110998\n",
      "epoch:3323 from 100000\n",
      "loss: 0.002726551576898828\n",
      "epoch:3324 from 100000\n",
      "loss: 0.0021389071718118657\n",
      "epoch:3325 from 100000\n",
      "loss: 0.002316584287598289\n",
      "epoch:3326 from 100000\n",
      "loss: 0.0022210376142624\n",
      "epoch:3327 from 100000\n",
      "loss: 0.0025843586670930563\n",
      "epoch:3328 from 100000\n",
      "loss: 0.0024999075948812788\n",
      "epoch:3329 from 100000\n",
      "loss: 0.0024673893439057792\n",
      "epoch:3330 from 100000\n",
      "loss: 0.002081700550618848\n",
      "epoch:3331 from 100000\n",
      "loss: 0.0023321954061543693\n",
      "epoch:3332 from 100000\n",
      "loss: 0.002498312251191237\n",
      "epoch:3333 from 100000\n",
      "loss: 0.0025178823461544653\n",
      "epoch:3334 from 100000\n",
      "loss: 0.0026588396775935053\n",
      "epoch:3335 from 100000\n",
      "loss: 0.0031063760187635125\n",
      "epoch:3336 from 100000\n",
      "loss: 0.002562092697786511\n",
      "epoch:3337 from 100000\n",
      "loss: 0.0022085003890879307\n",
      "epoch:3338 from 100000\n",
      "loss: 0.002563549358541195\n",
      "epoch:3339 from 100000\n",
      "loss: 0.0026221337904814845\n",
      "epoch:3340 from 100000\n",
      "loss: 0.002334119312081384\n",
      "epoch:3341 from 100000\n",
      "loss: 0.0024681737564833384\n",
      "epoch:3342 from 100000\n",
      "loss: 0.0027150049428854572\n",
      "epoch:3343 from 100000\n",
      "loss: 0.002880905500035169\n",
      "epoch:3344 from 100000\n",
      "loss: 0.0022279893638028625\n",
      "epoch:3345 from 100000\n",
      "loss: 0.0027350924372342433\n",
      "epoch:3346 from 100000\n",
      "loss: 0.002310995316975095\n",
      "epoch:3347 from 100000\n",
      "loss: 0.002225768106711712\n",
      "epoch:3348 from 100000\n",
      "loss: 0.0021579784465757257\n",
      "epoch:3349 from 100000\n",
      "loss: 0.002410793315846149\n",
      "epoch:3350 from 100000\n",
      "loss: 0.0024368028240076\n",
      "epoch:3351 from 100000\n",
      "loss: 0.0025289758462463396\n",
      "epoch:3352 from 100000\n",
      "loss: 0.002378393980279725\n",
      "epoch:3353 from 100000\n",
      "loss: 0.0032706191354350267\n",
      "epoch:3354 from 100000\n",
      "loss: 0.0025891146665912856\n",
      "retrying for data\n",
      "epoch:3355 from 100000\n",
      "loss: 0.0027170001777960806\n",
      "epoch:3356 from 100000\n",
      "loss: 0.0023664887935979095\n",
      "epoch:3357 from 100000\n",
      "loss: 0.002602304229554875\n",
      "epoch:3358 from 100000\n",
      "loss: 0.002239517379280187\n",
      "epoch:3359 from 100000\n",
      "loss: 0.00215389951191014\n",
      "epoch:3360 from 100000\n",
      "loss: 0.0023372778982775344\n",
      "epoch:3361 from 100000\n",
      "loss: 0.0024002226668073945\n",
      "epoch:3362 from 100000\n",
      "loss: 0.0027836452754286256\n",
      "epoch:3363 from 100000\n",
      "loss: 0.0021569969151613577\n",
      "epoch:3364 from 100000\n",
      "loss: 0.0025774734250774145\n",
      "epoch:3365 from 100000\n",
      "loss: 0.0027882941873826894\n",
      "epoch:3366 from 100000\n",
      "loss: 0.0020980979751357154\n",
      "epoch:3367 from 100000\n",
      "loss: 0.001783565627590633\n",
      "epoch:3368 from 100000\n",
      "loss: 0.0026280186112899173\n",
      "epoch:3369 from 100000\n",
      "loss: 0.0021855050622718863\n",
      "epoch:3370 from 100000\n",
      "loss: 0.0024076084579860435\n",
      "epoch:3371 from 100000\n",
      "loss: 0.0026880344023254565\n",
      "epoch:3372 from 100000\n",
      "loss: 0.0022850533903806536\n",
      "epoch:3373 from 100000\n",
      "loss: 0.002679528281532839\n",
      "epoch:3374 from 100000\n",
      "loss: 0.0020694707239101717\n",
      "epoch:3375 from 100000\n",
      "loss: 0.0022849915478104776\n",
      "retrying for data\n",
      "epoch:3376 from 100000\n",
      "loss: 0.002311696289552856\n",
      "epoch:3377 from 100000\n",
      "loss: 0.0026999239325107283\n",
      "epoch:3378 from 100000\n",
      "loss: 0.0021579620802185673\n",
      "epoch:3379 from 100000\n",
      "loss: 0.002690438048261967\n",
      "epoch:3380 from 100000\n",
      "loss: 0.0028073466422711135\n",
      "epoch:3381 from 100000\n",
      "loss: 0.0028378438628067215\n",
      "epoch:3382 from 100000\n",
      "loss: 0.0025171195370603527\n",
      "epoch:3383 from 100000\n",
      "loss: 0.0019911551303124497\n",
      "epoch:3384 from 100000\n",
      "loss: 0.002792998647919376\n",
      "epoch:3385 from 100000\n",
      "loss: 0.0025953980807003063\n",
      "retrying for data\n",
      "epoch:3386 from 100000\n",
      "loss: 0.002408119472420367\n",
      "epoch:3387 from 100000\n",
      "loss: 0.0029229603979956664\n",
      "epoch:3388 from 100000\n",
      "loss: 0.0025285783131607786\n",
      "epoch:3389 from 100000\n",
      "loss: 0.0026881865917403047\n",
      "epoch:3390 from 100000\n",
      "loss: 0.002789867234554322\n",
      "epoch:3391 from 100000\n",
      "loss: 0.002050899853742294\n",
      "epoch:3392 from 100000\n",
      "loss: 0.0029895405937452324\n",
      "epoch:3393 from 100000\n",
      "loss: 0.002322655835655496\n",
      "epoch:3394 from 100000\n",
      "loss: 0.002841450030331316\n",
      "epoch:3395 from 100000\n",
      "loss: 0.0030294342938077534\n",
      "epoch:3396 from 100000\n",
      "loss: 0.002439814660633033\n",
      "epoch:3397 from 100000\n",
      "loss: 0.0024359132752351798\n",
      "epoch:3398 from 100000\n",
      "loss: 0.002112481507765551\n",
      "epoch:3399 from 100000\n",
      "loss: 0.00239610085378672\n",
      "epoch:3400 from 100000\n",
      "loss: 0.0026268705211123233\n",
      "epoch:3401 from 100000\n",
      "loss: 0.0026010104531337674\n",
      "epoch:3402 from 100000\n",
      "loss: 0.002684625857455103\n",
      "epoch:3403 from 100000\n",
      "loss: 0.002221362386308101\n",
      "epoch:3404 from 100000\n",
      "loss: 0.0023813558540268787\n",
      "epoch:3405 from 100000\n",
      "loss: 0.002658173407041886\n",
      "epoch:3406 from 100000\n",
      "loss: 0.0028433906340126214\n",
      "epoch:3407 from 100000\n",
      "loss: 0.002197055856299812\n",
      "epoch:3408 from 100000\n",
      "loss: 0.0027789941504465787\n",
      "epoch:3409 from 100000\n",
      "loss: 0.002445822440961365\n",
      "epoch:3410 from 100000\n",
      "loss: 0.00232186888177921\n",
      "epoch:3411 from 100000\n",
      "loss: 0.002477959413795361\n",
      "epoch:3412 from 100000\n",
      "loss: 0.0018619257750458473\n",
      "epoch:3413 from 100000\n",
      "loss: 0.0024861269767484373\n",
      "epoch:3414 from 100000\n",
      "loss: 0.002767942527100331\n",
      "epoch:3415 from 100000\n",
      "loss: 0.002606515994315828\n",
      "epoch:3416 from 100000\n",
      "loss: 0.0026728750311804106\n",
      "epoch:3417 from 100000\n",
      "loss: 0.0029266393352145315\n",
      "epoch:3418 from 100000\n",
      "loss: 0.002583414016015695\n",
      "epoch:3419 from 100000\n",
      "loss: 0.0023049345352319506\n",
      "epoch:3420 from 100000\n",
      "loss: 0.002674976326346723\n",
      "epoch:3421 from 100000\n",
      "loss: 0.0024263329238181086\n",
      "epoch:3422 from 100000\n",
      "loss: 0.002497864183510501\n",
      "epoch:3423 from 100000\n",
      "loss: 0.003114033653957904\n",
      "epoch:3424 from 100000\n",
      "loss: 0.0024878298567614365\n",
      "epoch:3425 from 100000\n",
      "loss: 0.002485161202599784\n",
      "epoch:3426 from 100000\n",
      "loss: 0.002642623184499371\n",
      "epoch:3427 from 100000\n",
      "loss: 0.002760127899307463\n",
      "epoch:3428 from 100000\n",
      "loss: 0.002308006418019204\n",
      "epoch:3429 from 100000\n",
      "loss: 0.0025805448631217587\n",
      "epoch:3430 from 100000\n",
      "loss: 0.002433614945971385\n",
      "epoch:3431 from 100000\n",
      "loss: 0.0021517119418774655\n",
      "epoch:3432 from 100000\n",
      "loss: 0.00221153379311545\n",
      "epoch:3433 from 100000\n",
      "loss: 0.0025049487511523694\n",
      "epoch:3434 from 100000\n",
      "loss: 0.0032259906143963235\n",
      "epoch:3435 from 100000\n",
      "loss: 0.0023219273917902683\n",
      "epoch:3436 from 100000\n",
      "loss: 0.003085075751016575\n",
      "epoch:3437 from 100000\n",
      "loss: 0.002617467253543282\n",
      "epoch:3438 from 100000\n",
      "loss: 0.002196978759698709\n",
      "epoch:3439 from 100000\n",
      "loss: 0.001666967869795144\n",
      "epoch:3440 from 100000\n",
      "loss: 0.0023335848686372174\n",
      "epoch:3441 from 100000\n",
      "loss: 0.0026269211252554214\n",
      "epoch:3442 from 100000\n",
      "loss: 0.002361846936452139\n",
      "epoch:3443 from 100000\n",
      "loss: 0.0021636008114001015\n",
      "epoch:3444 from 100000\n",
      "loss: 0.002370757692801817\n",
      "epoch:3445 from 100000\n",
      "loss: 0.0023418322755617282\n",
      "epoch:3446 from 100000\n",
      "loss: 0.0025069326524089774\n",
      "epoch:3447 from 100000\n",
      "loss: 0.0025613418265493237\n",
      "retrying for data\n",
      "epoch:3448 from 100000\n",
      "loss: 0.002702948939486305\n",
      "epoch:3449 from 100000\n",
      "loss: 0.0029947209391707474\n",
      "epoch:3450 from 100000\n",
      "loss: 0.0028704674986670398\n",
      "epoch:3451 from 100000\n",
      "loss: 0.0019710651945388236\n",
      "epoch:3452 from 100000\n",
      "loss: 0.0026627286472944434\n",
      "epoch:3453 from 100000\n",
      "loss: 0.0027675747385274008\n",
      "epoch:3454 from 100000\n",
      "loss: 0.002710575737239651\n",
      "epoch:3455 from 100000\n",
      "loss: 0.0024947230907059748\n",
      "epoch:3456 from 100000\n",
      "loss: 0.002254489438598739\n",
      "epoch:3457 from 100000\n",
      "loss: 0.002466229732647207\n",
      "epoch:3458 from 100000\n",
      "loss: 0.002301884526076492\n",
      "epoch:3459 from 100000\n",
      "loss: 0.002718977219046792\n",
      "epoch:3460 from 100000\n",
      "loss: 0.0018954040428411645\n",
      "epoch:3461 from 100000\n",
      "loss: 0.0023890387046776596\n",
      "epoch:3462 from 100000\n",
      "loss: 0.0023225846621244273\n",
      "epoch:3463 from 100000\n",
      "loss: 0.0026653585028952875\n",
      "epoch:3464 from 100000\n",
      "loss: 0.0027088785563762485\n",
      "epoch:3465 from 100000\n",
      "loss: 0.0029959234372824994\n",
      "epoch:3466 from 100000\n",
      "loss: 0.0022363808414240793\n",
      "epoch:3467 from 100000\n",
      "loss: 0.0024302070231101425\n",
      "epoch:3468 from 100000\n",
      "loss: 0.002533240755354882\n",
      "epoch:3469 from 100000\n",
      "loss: 0.0026506187316357455\n",
      "retrying for data\n",
      "epoch:3470 from 100000\n",
      "loss: 0.002329626553404296\n",
      "epoch:3471 from 100000\n",
      "loss: 0.0023680811742110564\n",
      "epoch:3472 from 100000\n",
      "loss: 0.002022666808613849\n",
      "epoch:3473 from 100000\n",
      "loss: 0.0026486784400390206\n",
      "epoch:3474 from 100000\n",
      "loss: 0.0027921036409281896\n",
      "epoch:3475 from 100000\n",
      "loss: 0.0026393452521062954\n",
      "epoch:3476 from 100000\n",
      "loss: 0.002306053714761447\n",
      "epoch:3477 from 100000\n",
      "loss: 0.0027176658829120015\n",
      "epoch:3478 from 100000\n",
      "loss: 0.002345123024156237\n",
      "epoch:3479 from 100000\n",
      "loss: 0.002827550784189725\n",
      "epoch:3480 from 100000\n",
      "loss: 0.0024646633561928866\n",
      "epoch:3481 from 100000\n",
      "loss: 0.002187360303537389\n",
      "epoch:3482 from 100000\n",
      "loss: 0.002326840476285952\n",
      "epoch:3483 from 100000\n",
      "loss: 0.0022315152489937873\n",
      "epoch:3484 from 100000\n",
      "loss: 0.0024220093535376463\n",
      "epoch:3485 from 100000\n",
      "loss: 0.002095227538148947\n",
      "epoch:3486 from 100000\n",
      "loss: 0.002449587175647139\n",
      "epoch:3487 from 100000\n",
      "loss: 0.002551748751585592\n",
      "epoch:3488 from 100000\n",
      "loss: 0.002659182839266805\n",
      "epoch:3489 from 100000\n",
      "loss: 0.002178542008095974\n",
      "epoch:3490 from 100000\n",
      "loss: 0.0021403197544884733\n",
      "epoch:3491 from 100000\n",
      "loss: 0.002279046762140524\n",
      "retrying for data\n",
      "epoch:3492 from 100000\n",
      "loss: 0.002651808296548097\n",
      "epoch:3493 from 100000\n",
      "loss: 0.0025027701519348966\n",
      "epoch:3494 from 100000\n",
      "loss: 0.0021996990463719612\n",
      "epoch:3495 from 100000\n",
      "loss: 0.00260449785577066\n",
      "epoch:3496 from 100000\n",
      "loss: 0.002869530989357827\n",
      "epoch:3497 from 100000\n",
      "loss: 0.0025839996135346155\n",
      "epoch:3498 from 100000\n",
      "loss: 0.0025427005532208113\n",
      "epoch:3499 from 100000\n",
      "loss: 0.0019302482402565602\n",
      "epoch:3500 from 100000\n",
      "loss: 0.0029019361854949466\n",
      "epoch:3501 from 100000\n",
      "loss: 0.002221355723236229\n",
      "epoch:3502 from 100000\n",
      "loss: 0.0026533850998475828\n",
      "epoch:3503 from 100000\n",
      "loss: 0.002204892625258026\n",
      "epoch:3504 from 100000\n",
      "loss: 0.0019705473734745738\n",
      "epoch:3505 from 100000\n",
      "loss: 0.0028667220900615575\n",
      "epoch:3506 from 100000\n",
      "loss: 0.0025128529955225076\n",
      "epoch:3507 from 100000\n",
      "loss: 0.0027367416952301937\n",
      "epoch:3508 from 100000\n",
      "loss: 0.0021199356671388614\n",
      "epoch:3509 from 100000\n",
      "loss: 0.002179063100527401\n",
      "epoch:3510 from 100000\n",
      "loss: 0.0026270634173215512\n",
      "epoch:3511 from 100000\n",
      "loss: 0.0027366648387783243\n",
      "epoch:3512 from 100000\n",
      "loss: 0.002720074472392753\n",
      "epoch:3513 from 100000\n",
      "loss: 0.002834971187269275\n",
      "epoch:3514 from 100000\n",
      "loss: 0.002318361034554073\n",
      "epoch:3515 from 100000\n",
      "loss: 0.0022716008393217635\n",
      "epoch:3516 from 100000\n",
      "loss: 0.0022777326150276167\n",
      "epoch:3517 from 100000\n",
      "loss: 0.0028368151812401265\n",
      "epoch:3518 from 100000\n",
      "loss: 0.0021153806491724936\n",
      "epoch:3519 from 100000\n",
      "loss: 0.0028053409087078762\n",
      "epoch:3520 from 100000\n",
      "loss: 0.002577882726882308\n",
      "epoch:3521 from 100000\n",
      "loss: 0.001831256776426926\n",
      "epoch:3522 from 100000\n",
      "loss: 0.002655555630440176\n",
      "epoch:3523 from 100000\n",
      "loss: 0.0026636047390411477\n",
      "epoch:3524 from 100000\n",
      "loss: 0.002829371037250894\n",
      "epoch:3525 from 100000\n",
      "loss: 0.0025980711762372266\n",
      "epoch:3526 from 100000\n",
      "loss: 0.003067102001296007\n",
      "epoch:3527 from 100000\n",
      "loss: 0.003018921679199593\n",
      "epoch:3528 from 100000\n",
      "loss: 0.0029762561517543418\n",
      "epoch:3529 from 100000\n",
      "loss: 0.002643632071965385\n",
      "epoch:3530 from 100000\n",
      "loss: 0.0020582813242242537\n",
      "epoch:3531 from 100000\n",
      "loss: 0.0024409575597701405\n",
      "epoch:3532 from 100000\n",
      "loss: 0.0030211222305069896\n",
      "epoch:3533 from 100000\n",
      "loss: 0.0026580536161304735\n",
      "epoch:3534 from 100000\n",
      "loss: 0.0021143256454649872\n",
      "epoch:3535 from 100000\n",
      "loss: 0.002437590146257662\n",
      "epoch:3536 from 100000\n",
      "loss: 0.0023923823717950654\n",
      "epoch:3537 from 100000\n",
      "loss: 0.002982522994017245\n",
      "epoch:3538 from 100000\n",
      "loss: 0.002611328688146841\n",
      "epoch:3539 from 100000\n",
      "loss: 0.002046147120125852\n",
      "epoch:3540 from 100000\n",
      "loss: 0.002946891991001621\n",
      "epoch:3541 from 100000\n",
      "loss: 0.0023114952666105637\n",
      "epoch:3542 from 100000\n",
      "loss: 0.002494083923920698\n",
      "epoch:3543 from 100000\n",
      "loss: 0.002695641044851982\n",
      "epoch:3544 from 100000\n",
      "loss: 0.0030311119254520236\n",
      "epoch:3545 from 100000\n",
      "loss: 0.0029482326047514107\n",
      "epoch:3546 from 100000\n",
      "loss: 0.002432359292839692\n",
      "epoch:3547 from 100000\n",
      "loss: 0.0020220660714755923\n",
      "epoch:3548 from 100000\n",
      "loss: 0.0028485930582462515\n",
      "epoch:3549 from 100000\n",
      "loss: 0.003014311112337964\n",
      "epoch:3550 from 100000\n",
      "loss: 0.0023015045183854\n",
      "epoch:3551 from 100000\n",
      "loss: 0.0022430137987612397\n",
      "epoch:3552 from 100000\n",
      "loss: 0.002358652046240195\n",
      "epoch:3553 from 100000\n",
      "loss: 0.0026711636329537214\n",
      "epoch:3554 from 100000\n",
      "loss: 0.002438104939216146\n",
      "epoch:3555 from 100000\n",
      "loss: 0.0025373896884843816\n",
      "epoch:3556 from 100000\n",
      "loss: 0.0025991536204656995\n",
      "epoch:3557 from 100000\n",
      "loss: 0.002632402125215094\n",
      "epoch:3558 from 100000\n",
      "loss: 0.002477496984369054\n",
      "epoch:3559 from 100000\n",
      "loss: 0.0028182778808059084\n",
      "epoch:3560 from 100000\n",
      "loss: 0.0027804945486593624\n",
      "epoch:3561 from 100000\n",
      "loss: 0.002928498569332305\n",
      "epoch:3562 from 100000\n",
      "loss: 0.002655760528909923\n",
      "epoch:3563 from 100000\n",
      "loss: 0.0027542937096427522\n",
      "epoch:3564 from 100000\n",
      "loss: 0.002582170620769375\n",
      "epoch:3565 from 100000\n",
      "loss: 0.0025389434707108194\n",
      "epoch:3566 from 100000\n",
      "loss: 0.0025918845652199707\n",
      "epoch:3567 from 100000\n",
      "loss: 0.0030417789872672074\n",
      "epoch:3568 from 100000\n",
      "loss: 0.0019851300916684522\n",
      "epoch:3569 from 100000\n",
      "loss: 0.0024218428355311516\n",
      "epoch:3570 from 100000\n",
      "loss: 0.003249398745651888\n",
      "epoch:3571 from 100000\n",
      "loss: 0.002366696045015715\n",
      "epoch:3572 from 100000\n",
      "loss: 0.0024975097721267048\n",
      "epoch:3573 from 100000\n",
      "loss: 0.002608435530760289\n",
      "epoch:3574 from 100000\n",
      "loss: 0.0023509523347229333\n",
      "epoch:3575 from 100000\n",
      "loss: 0.002919250100546833\n",
      "epoch:3576 from 100000\n",
      "loss: 0.002516432978637795\n",
      "epoch:3577 from 100000\n",
      "loss: 0.0027851796469917645\n",
      "epoch:3578 from 100000\n",
      "loss: 0.0019851040659375485\n",
      "epoch:3579 from 100000\n",
      "loss: 0.0024660825748412663\n",
      "epoch:3580 from 100000\n",
      "loss: 0.0028396036248068413\n",
      "epoch:3581 from 100000\n",
      "loss: 0.0019730374949062934\n",
      "epoch:3582 from 100000\n",
      "loss: 0.002210983183630333\n",
      "epoch:3583 from 100000\n",
      "loss: 0.0026400218491460237\n",
      "epoch:3584 from 100000\n",
      "loss: 0.0021446976621462\n",
      "epoch:3585 from 100000\n",
      "loss: 0.0024328990251092364\n",
      "epoch:3586 from 100000\n",
      "loss: 0.0030306115081089047\n",
      "epoch:3587 from 100000\n",
      "loss: 0.002623062134347265\n",
      "epoch:3588 from 100000\n",
      "loss: 0.0025242058133017053\n",
      "epoch:3589 from 100000\n",
      "loss: 0.002171566981957085\n",
      "epoch:3590 from 100000\n",
      "loss: 0.0029579259653473855\n",
      "epoch:3591 from 100000\n",
      "loss: 0.00238570865370491\n",
      "epoch:3592 from 100000\n",
      "loss: 0.002241797671331369\n",
      "epoch:3593 from 100000\n",
      "loss: 0.0026595060142540206\n",
      "epoch:3594 from 100000\n",
      "loss: 0.0023889387310873644\n",
      "epoch:3595 from 100000\n",
      "loss: 0.0022050670375506343\n",
      "epoch:3596 from 100000\n",
      "loss: 0.002361106018611281\n",
      "epoch:3597 from 100000\n",
      "loss: 0.0025605481220623005\n",
      "epoch:3598 from 100000\n",
      "loss: 0.0021254074547414348\n",
      "epoch:3599 from 100000\n",
      "loss: 0.0026049197958855075\n",
      "epoch:3600 from 100000\n",
      "loss: 0.0023803175086811734\n",
      "epoch:3601 from 100000\n",
      "loss: 0.002552851129166811\n",
      "epoch:3602 from 100000\n",
      "loss: 0.0025045111014208032\n",
      "epoch:3603 from 100000\n",
      "loss: 0.002755535947869703\n",
      "epoch:3604 from 100000\n",
      "loss: 0.0029397426723960507\n",
      "epoch:3605 from 100000\n",
      "loss: 0.00239717516831206\n",
      "epoch:3606 from 100000\n",
      "loss: 0.00268032762699022\n",
      "epoch:3607 from 100000\n",
      "loss: 0.002103202903072088\n",
      "epoch:3608 from 100000\n",
      "loss: 0.0027265022201277134\n",
      "epoch:3609 from 100000\n",
      "loss: 0.0023032407702601176\n",
      "epoch:3610 from 100000\n",
      "loss: 0.0026068427923178206\n",
      "epoch:3611 from 100000\n",
      "loss: 0.0020792805426594896\n",
      "epoch:3612 from 100000\n",
      "loss: 0.0024460465308209223\n",
      "epoch:3613 from 100000\n",
      "loss: 0.0028661113139207828\n",
      "epoch:3614 from 100000\n",
      "loss: 0.002481595841345552\n",
      "epoch:3615 from 100000\n",
      "loss: 0.002727724091869277\n",
      "epoch:3616 from 100000\n",
      "loss: 0.002720239084936793\n",
      "epoch:3617 from 100000\n",
      "loss: 0.0021889917782971224\n",
      "epoch:3618 from 100000\n",
      "loss: 0.0027051630424068662\n",
      "epoch:3619 from 100000\n",
      "loss: 0.0024296604120479515\n",
      "epoch:3620 from 100000\n",
      "loss: 0.0023180338631334507\n",
      "epoch:3621 from 100000\n",
      "loss: 0.0028262997775527765\n",
      "epoch:3622 from 100000\n",
      "loss: 0.0020977690206507305\n",
      "epoch:3623 from 100000\n",
      "loss: 0.0029663060682025844\n",
      "epoch:3624 from 100000\n",
      "loss: 0.0034366797957545714\n",
      "epoch:3625 from 100000\n",
      "loss: 0.0024399184189860534\n",
      "epoch:3626 from 100000\n",
      "loss: 0.003221390775564714\n",
      "epoch:3627 from 100000\n",
      "loss: 0.0027221210077641445\n",
      "epoch:3628 from 100000\n",
      "loss: 0.00242219175191849\n",
      "epoch:3629 from 100000\n",
      "loss: 0.0019173773272029848\n",
      "epoch:3630 from 100000\n",
      "loss: 0.002548050882808184\n",
      "epoch:3631 from 100000\n",
      "loss: 0.002354846023678192\n",
      "epoch:3632 from 100000\n",
      "loss: 0.0024022187485002178\n",
      "epoch:3633 from 100000\n",
      "loss: 0.0028123143766265457\n",
      "epoch:3634 from 100000\n",
      "loss: 0.002532784591494419\n",
      "epoch:3635 from 100000\n",
      "loss: 0.002329561023543647\n",
      "epoch:3636 from 100000\n",
      "loss: 0.0029998052227711014\n",
      "epoch:3637 from 100000\n",
      "loss: 0.002408097524181585\n",
      "epoch:3638 from 100000\n",
      "loss: 0.0021795859605049372\n",
      "epoch:3639 from 100000\n",
      "loss: 0.00245738738755108\n",
      "epoch:3640 from 100000\n",
      "loss: 0.002842233362798652\n",
      "epoch:3641 from 100000\n",
      "loss: 0.0027591449407395885\n",
      "epoch:3642 from 100000\n",
      "loss: 0.002486191883861011\n",
      "epoch:3643 from 100000\n",
      "loss: 0.002605147891699744\n",
      "epoch:3644 from 100000\n",
      "loss: 0.0025675791876835774\n",
      "epoch:3645 from 100000\n",
      "loss: 0.0021892464836810177\n",
      "epoch:3646 from 100000\n",
      "loss: 0.0023248751436142356\n",
      "epoch:3647 from 100000\n",
      "loss: 0.002446293419353651\n",
      "epoch:3648 from 100000\n",
      "loss: 0.0026110486488448714\n",
      "epoch:3649 from 100000\n",
      "loss: 0.002734756711845421\n",
      "epoch:3650 from 100000\n",
      "loss: 0.0027472731543696227\n",
      "epoch:3651 from 100000\n",
      "loss: 0.0027766228614609645\n",
      "epoch:3652 from 100000\n",
      "loss: 0.002327564504298607\n",
      "epoch:3653 from 100000\n",
      "loss: 0.002750652619340599\n",
      "epoch:3654 from 100000\n",
      "loss: 0.0020737270535278185\n",
      "epoch:3655 from 100000\n",
      "loss: 0.0029016861190029886\n",
      "epoch:3656 from 100000\n",
      "loss: 0.002651740179629769\n",
      "epoch:3657 from 100000\n",
      "loss: 0.0025710675963637186\n",
      "epoch:3658 from 100000\n",
      "loss: 0.0022436670116263713\n",
      "epoch:3659 from 100000\n",
      "loss: 0.002823471466143701\n",
      "epoch:3660 from 100000\n",
      "loss: 0.003171651641508788\n",
      "epoch:3661 from 100000\n",
      "loss: 0.0029005238323520643\n",
      "epoch:3662 from 100000\n",
      "loss: 0.002670650414472675\n",
      "epoch:3663 from 100000\n",
      "loss: 0.0022045141556361614\n",
      "epoch:3664 from 100000\n",
      "loss: 0.003170001761375829\n",
      "epoch:3665 from 100000\n",
      "loss: 0.0027044116071976987\n",
      "epoch:3666 from 100000\n",
      "loss: 0.002888331203564576\n",
      "epoch:3667 from 100000\n",
      "loss: 0.002998000190373773\n",
      "epoch:3668 from 100000\n",
      "loss: 0.002330031384644826\n",
      "epoch:3669 from 100000\n",
      "loss: 0.002746276755715371\n",
      "epoch:3670 from 100000\n",
      "loss: 0.0021829254027920797\n",
      "epoch:3671 from 100000\n",
      "loss: 0.0024723545952554105\n",
      "epoch:3672 from 100000\n",
      "loss: 0.00199221622949608\n",
      "epoch:3673 from 100000\n",
      "loss: 0.002696159203537718\n",
      "epoch:3674 from 100000\n",
      "loss: 0.0026928306704405713\n",
      "epoch:3675 from 100000\n",
      "loss: 0.0022528354521114125\n",
      "epoch:3676 from 100000\n",
      "loss: 0.002147917606379224\n",
      "epoch:3677 from 100000\n",
      "loss: 0.0023758292298339256\n",
      "epoch:3678 from 100000\n",
      "loss: 0.0025698138537393334\n",
      "epoch:3679 from 100000\n",
      "loss: 0.0022729985080900406\n",
      "epoch:3680 from 100000\n",
      "loss: 0.0030947107973133825\n",
      "epoch:3681 from 100000\n",
      "loss: 0.002879117334202874\n",
      "epoch:3682 from 100000\n",
      "loss: 0.0024511779742510953\n",
      "epoch:3683 from 100000\n",
      "loss: 0.002694253738354746\n",
      "epoch:3684 from 100000\n",
      "loss: 0.0026510197876916664\n",
      "epoch:3685 from 100000\n",
      "loss: 0.002681340698330814\n",
      "epoch:3686 from 100000\n",
      "loss: 0.0023453599580420814\n",
      "epoch:3687 from 100000\n",
      "loss: 0.00253620252908604\n",
      "epoch:3688 from 100000\n",
      "loss: 0.0027428569315191\n",
      "epoch:3689 from 100000\n",
      "loss: 0.0030389786841880095\n",
      "epoch:3690 from 100000\n",
      "loss: 0.0024679907573670334\n",
      "epoch:3691 from 100000\n",
      "loss: 0.002658890651716206\n",
      "epoch:3692 from 100000\n",
      "loss: 0.002283869142416961\n",
      "epoch:3693 from 100000\n",
      "loss: 0.002560806274956917\n",
      "epoch:3694 from 100000\n",
      "loss: 0.003069554062477664\n",
      "epoch:3695 from 100000\n",
      "loss: 0.002863633061608084\n",
      "epoch:3696 from 100000\n",
      "loss: 0.0020762856428007126\n",
      "epoch:3697 from 100000\n",
      "loss: 0.0022118304962077673\n",
      "epoch:3698 from 100000\n",
      "loss: 0.0019821790798459915\n",
      "epoch:3699 from 100000\n",
      "loss: 0.002249123715927226\n",
      "epoch:3700 from 100000\n",
      "loss: 0.00241354049040865\n",
      "epoch:3701 from 100000\n",
      "loss: 0.00255355790051226\n",
      "epoch:3702 from 100000\n",
      "loss: 0.002823086116450213\n",
      "epoch:3703 from 100000\n",
      "loss: 0.001914861569275672\n",
      "epoch:3704 from 100000\n",
      "loss: 0.0024884937166405052\n",
      "epoch:3705 from 100000\n",
      "loss: 0.002090927799201836\n",
      "epoch:3706 from 100000\n",
      "loss: 0.002497240779121057\n",
      "epoch:3707 from 100000\n",
      "loss: 0.0024574527829486215\n",
      "epoch:3708 from 100000\n",
      "loss: 0.0023408290805235765\n",
      "epoch:3709 from 100000\n",
      "loss: 0.002685225656506418\n",
      "epoch:3710 from 100000\n",
      "loss: 0.003187805091528162\n",
      "epoch:3711 from 100000\n",
      "loss: 0.002547286726155562\n",
      "epoch:3712 from 100000\n",
      "loss: 0.0025069436654518995\n",
      "epoch:3713 from 100000\n",
      "loss: 0.002025092732765188\n",
      "epoch:3714 from 100000\n",
      "loss: 0.002861233187331891\n",
      "epoch:3715 from 100000\n",
      "loss: 0.002618923501813697\n",
      "epoch:3716 from 100000\n",
      "loss: 0.002864010413361484\n",
      "epoch:3717 from 100000\n",
      "loss: 0.002147313757916436\n",
      "epoch:3718 from 100000\n",
      "loss: 0.0022445645913933276\n",
      "epoch:3719 from 100000\n",
      "loss: 0.0026686141329150814\n",
      "epoch:3720 from 100000\n",
      "loss: 0.0027626237513800334\n",
      "epoch:3721 from 100000\n",
      "loss: 0.0024392380610720465\n",
      "epoch:3722 from 100000\n",
      "loss: 0.001931616294655214\n",
      "epoch:3723 from 100000\n",
      "loss: 0.0024913197937905807\n",
      "epoch:3724 from 100000\n",
      "loss: 0.0026766964491287126\n",
      "epoch:3725 from 100000\n",
      "loss: 0.002862904112760134\n",
      "epoch:3726 from 100000\n",
      "loss: 0.002596312159369063\n",
      "epoch:3727 from 100000\n",
      "loss: 0.0024943962363010996\n",
      "epoch:3728 from 100000\n",
      "loss: 0.0022790338121438936\n",
      "epoch:3729 from 100000\n",
      "loss: 0.002631709762937362\n",
      "epoch:3730 from 100000\n",
      "loss: 0.002362835919313966\n",
      "epoch:3731 from 100000\n",
      "loss: 0.0027196660279287244\n",
      "epoch:3732 from 100000\n",
      "loss: 0.0027256174263072808\n",
      "epoch:3733 from 100000\n",
      "loss: 0.0025321798460282707\n",
      "epoch:3734 from 100000\n",
      "loss: 0.002741463873292105\n",
      "epoch:3735 from 100000\n",
      "loss: 0.0024150532934470448\n",
      "epoch:3736 from 100000\n",
      "loss: 0.002361063156897103\n",
      "epoch:3737 from 100000\n",
      "loss: 0.00201078856781578\n",
      "epoch:3738 from 100000\n",
      "loss: 0.0026946195959567376\n",
      "epoch:3739 from 100000\n",
      "loss: 0.00212963756513318\n",
      "epoch:3740 from 100000\n",
      "loss: 0.0023249356687244926\n",
      "epoch:3741 from 100000\n",
      "loss: 0.0023027001279842807\n",
      "epoch:3742 from 100000\n",
      "loss: 0.0023481124489705962\n",
      "epoch:3743 from 100000\n",
      "loss: 0.002482290038045676\n",
      "epoch:3744 from 100000\n",
      "loss: 0.0029171442265294445\n",
      "epoch:3745 from 100000\n",
      "loss: 0.0024440165742021236\n",
      "epoch:3746 from 100000\n",
      "loss: 0.0020300172841416497\n",
      "epoch:3747 from 100000\n",
      "loss: 0.002366876439737098\n",
      "epoch:3748 from 100000\n",
      "loss: 0.0022719345726898155\n",
      "epoch:3749 from 100000\n",
      "loss: 0.0021171509991972925\n",
      "epoch:3750 from 100000\n",
      "loss: 0.0023821171943154695\n",
      "epoch:3751 from 100000\n",
      "loss: 0.0029655303843298952\n",
      "epoch:3752 from 100000\n",
      "loss: 0.0025131288338116065\n",
      "epoch:3753 from 100000\n",
      "loss: 0.002738445534220091\n",
      "epoch:3754 from 100000\n",
      "loss: 0.002222914475368043\n",
      "epoch:3755 from 100000\n",
      "loss: 0.0026717123223534145\n",
      "epoch:3756 from 100000\n",
      "loss: 0.001984161710765875\n",
      "epoch:3757 from 100000\n",
      "loss: 0.002228847536457579\n",
      "epoch:3758 from 100000\n",
      "loss: 0.0022376473996246204\n",
      "epoch:3759 from 100000\n",
      "loss: 0.0028352470074253233\n",
      "epoch:3760 from 100000\n",
      "loss: 0.00241468063791217\n",
      "epoch:3761 from 100000\n",
      "loss: 0.0022907850307092303\n",
      "epoch:3762 from 100000\n",
      "loss: 0.002268650514892556\n",
      "epoch:3763 from 100000\n",
      "loss: 0.002041396212973723\n",
      "epoch:3764 from 100000\n",
      "loss: 0.0029953770968376148\n",
      "epoch:3765 from 100000\n",
      "loss: 0.0019216018060319584\n",
      "epoch:3766 from 100000\n",
      "loss: 0.002537476840132058\n",
      "epoch:3767 from 100000\n",
      "loss: 0.002766842459493546\n",
      "epoch:3768 from 100000\n",
      "loss: 0.0021866126902239102\n",
      "epoch:3769 from 100000\n",
      "loss: 0.0025479054389165867\n",
      "epoch:3770 from 100000\n",
      "loss: 0.0026054199027498726\n",
      "epoch:3771 from 100000\n",
      "loss: 0.002583455262509915\n",
      "epoch:3772 from 100000\n",
      "loss: 0.0025252460802107635\n",
      "epoch:3773 from 100000\n",
      "loss: 0.0029409605495800406\n",
      "epoch:3774 from 100000\n",
      "loss: 0.0024572623781864422\n",
      "epoch:3775 from 100000\n",
      "loss: 0.0028932909061154533\n",
      "epoch:3776 from 100000\n",
      "loss: 0.002439311947313172\n",
      "epoch:3777 from 100000\n",
      "loss: 0.0029834111968511934\n",
      "epoch:3778 from 100000\n",
      "loss: 0.0021839174106901282\n",
      "epoch:3779 from 100000\n",
      "loss: 0.0024897901991352\n",
      "epoch:3780 from 100000\n",
      "loss: 0.0026415247051829738\n",
      "epoch:3781 from 100000\n",
      "loss: 0.002585483488388718\n",
      "epoch:3782 from 100000\n",
      "loss: 0.002766496985515232\n",
      "epoch:3783 from 100000\n",
      "loss: 0.002302351937032654\n",
      "epoch:3784 from 100000\n",
      "loss: 0.00245076077004569\n",
      "epoch:3785 from 100000\n",
      "loss: 0.0025308105841475026\n",
      "epoch:3786 from 100000\n",
      "loss: 0.002763791795103998\n",
      "epoch:3787 from 100000\n",
      "loss: 0.0022946271350861025\n",
      "epoch:3788 from 100000\n",
      "loss: 0.0022184096113733176\n",
      "epoch:3789 from 100000\n",
      "loss: 0.0022881518959820824\n",
      "epoch:3790 from 100000\n",
      "loss: 0.0029216595788454924\n",
      "epoch:3791 from 100000\n",
      "loss: 0.003157443576725427\n",
      "epoch:3792 from 100000\n",
      "loss: 0.002040849914806131\n",
      "epoch:3793 from 100000\n",
      "loss: 0.002929265716431928\n",
      "epoch:3794 from 100000\n",
      "loss: 0.0027577109855769777\n",
      "epoch:3795 from 100000\n",
      "loss: 0.0024340738771115866\n",
      "epoch:3796 from 100000\n",
      "loss: 0.0023569015325790588\n",
      "epoch:3797 from 100000\n",
      "loss: 0.002170647635352907\n",
      "epoch:3798 from 100000\n",
      "loss: 0.0025269948252173435\n",
      "epoch:3799 from 100000\n",
      "loss: 0.002431837028410655\n",
      "epoch:3800 from 100000\n",
      "loss: 0.002445878393473322\n",
      "epoch:3801 from 100000\n",
      "loss: 0.002099528808301443\n",
      "epoch:3802 from 100000\n",
      "loss: 0.0024974786429083906\n",
      "epoch:3803 from 100000\n",
      "loss: 0.002514556000733137\n",
      "epoch:3804 from 100000\n",
      "loss: 0.0031868753838466546\n",
      "epoch:3805 from 100000\n",
      "loss: 0.002241032266226739\n",
      "epoch:3806 from 100000\n",
      "loss: 0.002840662490825707\n",
      "epoch:3807 from 100000\n",
      "loss: 0.002404192196877375\n",
      "epoch:3808 from 100000\n",
      "loss: 0.0023913765274983234\n",
      "epoch:3809 from 100000\n",
      "loss: 0.0027584163896108294\n",
      "epoch:3810 from 100000\n",
      "loss: 0.0026835787592744964\n",
      "epoch:3811 from 100000\n",
      "loss: 0.0022288430054686614\n",
      "epoch:3812 from 100000\n",
      "loss: 0.0024116177665973737\n",
      "epoch:3813 from 100000\n",
      "loss: 0.0028164625184530223\n",
      "epoch:3814 from 100000\n",
      "loss: 0.0023873377190852807\n",
      "epoch:3815 from 100000\n",
      "loss: 0.0022305474526547187\n",
      "epoch:3816 from 100000\n",
      "loss: 0.0024587334960983753\n",
      "epoch:3817 from 100000\n",
      "loss: 0.00292495683666516\n",
      "epoch:3818 from 100000\n",
      "loss: 0.00239716329355133\n",
      "epoch:3819 from 100000\n",
      "loss: 0.0025554841526371774\n",
      "epoch:3820 from 100000\n",
      "loss: 0.002363986498266968\n",
      "epoch:3821 from 100000\n",
      "loss: 0.0017374064130422084\n",
      "epoch:3822 from 100000\n",
      "loss: 0.002453050428869119\n",
      "epoch:3823 from 100000\n",
      "loss: 0.0025874990394925135\n",
      "epoch:3824 from 100000\n",
      "loss: 0.0028479264413476812\n",
      "epoch:3825 from 100000\n",
      "loss: 0.0020386107651120255\n",
      "epoch:3826 from 100000\n",
      "loss: 0.002376823376238235\n",
      "epoch:3827 from 100000\n",
      "loss: 0.002330534233834669\n",
      "epoch:3828 from 100000\n",
      "loss: 0.0021215904666149754\n",
      "epoch:3829 from 100000\n",
      "loss: 0.0022112209125282334\n",
      "epoch:3830 from 100000\n",
      "loss: 0.002021113113727324\n",
      "epoch:3831 from 100000\n",
      "loss: 0.0026385049760904167\n",
      "epoch:3832 from 100000\n",
      "loss: 0.0021526299015590666\n",
      "epoch:3833 from 100000\n",
      "loss: 0.002446452295458812\n",
      "epoch:3834 from 100000\n",
      "loss: 0.0025718605465669953\n",
      "epoch:3835 from 100000\n",
      "loss: 0.00226797293061054\n",
      "epoch:3836 from 100000\n",
      "loss: 0.0024435115973915345\n",
      "epoch:3837 from 100000\n",
      "loss: 0.0021694760206401043\n",
      "epoch:3838 from 100000\n",
      "loss: 0.002330860338304319\n",
      "epoch:3839 from 100000\n",
      "loss: 0.002531258897022326\n",
      "epoch:3840 from 100000\n",
      "loss: 0.002220915674712387\n",
      "epoch:3841 from 100000\n",
      "loss: 0.002917103838484536\n",
      "epoch:3842 from 100000\n",
      "loss: 0.0019363698317675926\n",
      "epoch:3843 from 100000\n",
      "loss: 0.002316859940805216\n",
      "epoch:3844 from 100000\n",
      "loss: 0.00239561897205931\n",
      "epoch:3845 from 100000\n",
      "loss: 0.0025704117049372144\n",
      "epoch:3846 from 100000\n",
      "loss: 0.0027140459314125565\n",
      "epoch:3847 from 100000\n",
      "loss: 0.0017409262545982074\n",
      "epoch:3848 from 100000\n",
      "loss: 0.002310230680990344\n",
      "epoch:3849 from 100000\n",
      "loss: 0.002926119157791618\n",
      "epoch:3850 from 100000\n",
      "loss: 0.0027718886719299007\n",
      "epoch:3851 from 100000\n",
      "loss: 0.0023597957030006\n",
      "retrying for data\n",
      "epoch:3852 from 100000\n",
      "loss: 0.0025055886847837883\n",
      "epoch:3853 from 100000\n",
      "loss: 0.0024041266403855843\n",
      "epoch:3854 from 100000\n",
      "loss: 0.0025249652572227888\n",
      "epoch:3855 from 100000\n",
      "loss: 0.0028338930958966557\n",
      "epoch:3856 from 100000\n",
      "loss: 0.0027520290220479637\n",
      "epoch:3857 from 100000\n",
      "loss: 0.002410376497607558\n",
      "epoch:3858 from 100000\n",
      "loss: 0.002257739777462575\n",
      "epoch:3859 from 100000\n",
      "loss: 0.0022818098866537184\n",
      "epoch:3860 from 100000\n",
      "loss: 0.002534011268494396\n",
      "epoch:3861 from 100000\n",
      "loss: 0.0023537555771895313\n",
      "epoch:3862 from 100000\n",
      "loss: 0.0028509350970011837\n",
      "epoch:3863 from 100000\n",
      "loss: 0.0029755225785095263\n",
      "epoch:3864 from 100000\n",
      "loss: 0.002636703586517797\n",
      "epoch:3865 from 100000\n",
      "loss: 0.0024216046711558192\n",
      "epoch:3866 from 100000\n",
      "loss: 0.002119973780622786\n",
      "epoch:3867 from 100000\n",
      "loss: 0.0026808361120345126\n",
      "epoch:3868 from 100000\n",
      "loss: 0.0024878647648449714\n",
      "epoch:3869 from 100000\n",
      "loss: 0.0026031284199348192\n",
      "epoch:3870 from 100000\n",
      "loss: 0.001723533213095152\n",
      "epoch:3871 from 100000\n",
      "loss: 0.0029551075936069537\n",
      "epoch:3872 from 100000\n",
      "loss: 0.0021485536095724456\n",
      "epoch:3873 from 100000\n",
      "loss: 0.002211350231846154\n",
      "epoch:3874 from 100000\n",
      "loss: 0.0028379624214665\n",
      "epoch:3875 from 100000\n",
      "loss: 0.0020790635960281634\n",
      "epoch:3876 from 100000\n",
      "loss: 0.0027182471810647257\n",
      "epoch:3877 from 100000\n",
      "loss: 0.0025870769760842904\n",
      "epoch:3878 from 100000\n",
      "loss: 0.0027354572984563674\n",
      "epoch:3879 from 100000\n",
      "loss: 0.002519993917900365\n",
      "epoch:3880 from 100000\n",
      "loss: 0.002545873224363504\n",
      "epoch:3881 from 100000\n",
      "loss: 0.0025566886384353893\n",
      "epoch:3882 from 100000\n",
      "loss: 0.0024309081377964503\n",
      "epoch:3883 from 100000\n",
      "loss: 0.00216652852341781\n",
      "epoch:3884 from 100000\n",
      "loss: 0.0027467350903975785\n",
      "epoch:3885 from 100000\n",
      "loss: 0.0025425456072980523\n",
      "epoch:3886 from 100000\n",
      "loss: 0.0022054036966778767\n",
      "epoch:3887 from 100000\n",
      "loss: 0.002446314909064995\n",
      "epoch:3888 from 100000\n",
      "loss: 0.0023100995270510793\n",
      "epoch:3889 from 100000\n",
      "loss: 0.0023881400677225884\n",
      "epoch:3890 from 100000\n",
      "loss: 0.0023129304414908347\n",
      "epoch:3891 from 100000\n",
      "loss: 0.002679216803443296\n",
      "epoch:3892 from 100000\n",
      "loss: 0.002620541947635502\n",
      "epoch:3893 from 100000\n",
      "loss: 0.0019098310001766095\n",
      "epoch:3894 from 100000\n",
      "loss: 0.0029322514338900874\n",
      "epoch:3895 from 100000\n",
      "loss: 0.002879371008106091\n",
      "epoch:3896 from 100000\n",
      "loss: 0.003035931511902845\n",
      "epoch:3897 from 100000\n",
      "loss: 0.0028271410485842807\n",
      "epoch:3898 from 100000\n",
      "loss: 0.002671993069441214\n",
      "epoch:3899 from 100000\n",
      "loss: 0.0026545092400311887\n",
      "epoch:3900 from 100000\n",
      "loss: 0.00288024609386639\n",
      "epoch:3901 from 100000\n",
      "loss: 0.0026395153987550657\n",
      "epoch:3902 from 100000\n",
      "loss: 0.0025724891206664324\n",
      "epoch:3903 from 100000\n",
      "loss: 0.002106486417801534\n",
      "epoch:3904 from 100000\n",
      "loss: 0.003071483079366999\n",
      "epoch:3905 from 100000\n",
      "loss: 0.002593009550594161\n",
      "epoch:3906 from 100000\n",
      "loss: 0.0027607933918147864\n",
      "epoch:3907 from 100000\n",
      "loss: 0.002620199442304738\n",
      "epoch:3908 from 100000\n",
      "loss: 0.002556558937968134\n",
      "epoch:3909 from 100000\n",
      "loss: 0.0025146303389647073\n",
      "epoch:3910 from 100000\n",
      "loss: 0.0024593211167314166\n",
      "epoch:3911 from 100000\n",
      "loss: 0.0022367057876522267\n",
      "epoch:3912 from 100000\n",
      "loss: 0.002027415777376973\n",
      "epoch:3913 from 100000\n",
      "loss: 0.002306089170360792\n",
      "epoch:3914 from 100000\n",
      "loss: 0.00216429984413935\n",
      "epoch:3915 from 100000\n",
      "loss: 0.0026128924153567823\n",
      "retrying for data\n",
      "epoch:3916 from 100000\n",
      "loss: 0.0028102785722410317\n",
      "epoch:3917 from 100000\n",
      "loss: 0.002689729266876384\n",
      "epoch:3918 from 100000\n",
      "loss: 0.0033965109103917257\n",
      "epoch:3919 from 100000\n",
      "loss: 0.0029591370719685983\n",
      "epoch:3920 from 100000\n",
      "loss: 0.0030030056996679377\n",
      "epoch:3921 from 100000\n",
      "loss: 0.0021476030773044386\n",
      "epoch:3922 from 100000\n",
      "loss: 0.002453205739357145\n",
      "epoch:3923 from 100000\n",
      "loss: 0.002449875761300291\n",
      "epoch:3924 from 100000\n",
      "loss: 0.002212583075532848\n",
      "epoch:3925 from 100000\n",
      "loss: 0.0024001686138319656\n",
      "epoch:3926 from 100000\n",
      "loss: 0.0026930402843561296\n",
      "epoch:3927 from 100000\n",
      "loss: 0.002360738289070241\n",
      "epoch:3928 from 100000\n",
      "loss: 0.0032847880153497044\n",
      "epoch:3929 from 100000\n",
      "loss: 0.0022048057363122098\n",
      "epoch:3930 from 100000\n",
      "loss: 0.002370426626043809\n",
      "epoch:3931 from 100000\n",
      "loss: 0.00298643225937667\n",
      "epoch:3932 from 100000\n",
      "loss: 0.002848656649575787\n",
      "epoch:3933 from 100000\n",
      "loss: 0.0025225587708064268\n",
      "epoch:3934 from 100000\n",
      "loss: 0.0022447258819369154\n",
      "epoch:3935 from 100000\n",
      "loss: 0.0025326083955974354\n",
      "epoch:3936 from 100000\n",
      "loss: 0.0015933554279286\n",
      "epoch:3937 from 100000\n",
      "loss: 0.0030447238667079546\n",
      "epoch:3938 from 100000\n",
      "loss: 0.0025166719164815277\n",
      "epoch:3939 from 100000\n",
      "loss: 0.0025831099634672228\n",
      "epoch:3940 from 100000\n",
      "loss: 0.002419942562369215\n",
      "epoch:3941 from 100000\n",
      "loss: 0.0021696739725030056\n",
      "epoch:3942 from 100000\n",
      "loss: 0.0025317923197292203\n",
      "epoch:3943 from 100000\n",
      "loss: 0.002417497551618908\n",
      "epoch:3944 from 100000\n",
      "loss: 0.0028821934493237222\n",
      "epoch:3945 from 100000\n",
      "loss: 0.0022455403275216668\n",
      "epoch:3946 from 100000\n",
      "loss: 0.002852116313164288\n",
      "epoch:3947 from 100000\n",
      "loss: 0.0024804942986520473\n",
      "epoch:3948 from 100000\n",
      "loss: 0.002341281649577809\n",
      "epoch:3949 from 100000\n",
      "loss: 0.002520977036326144\n",
      "epoch:3950 from 100000\n",
      "loss: 0.002254343450289298\n",
      "epoch:3951 from 100000\n",
      "loss: 0.002386481522989925\n",
      "epoch:3952 from 100000\n",
      "loss: 0.0023942382676693796\n",
      "epoch:3953 from 100000\n",
      "loss: 0.0026620083248189985\n",
      "epoch:3954 from 100000\n",
      "loss: 0.0021834446679633857\n",
      "epoch:3955 from 100000\n",
      "loss: 0.00276000104216223\n",
      "epoch:3956 from 100000\n",
      "loss: 0.00280545241702157\n",
      "epoch:3957 from 100000\n",
      "loss: 0.002666503031349521\n",
      "epoch:3958 from 100000\n",
      "loss: 0.0026993690004246673\n",
      "epoch:3959 from 100000\n",
      "loss: 0.0025823565441669416\n",
      "epoch:3960 from 100000\n",
      "loss: 0.0028376153428553152\n",
      "epoch:3961 from 100000\n",
      "loss: 0.0022821872951084288\n",
      "epoch:3962 from 100000\n",
      "loss: 0.0025766600563770226\n",
      "epoch:3963 from 100000\n",
      "loss: 0.0021623997544679696\n",
      "epoch:3964 from 100000\n",
      "loss: 0.0027097322007989533\n",
      "epoch:3965 from 100000\n",
      "loss: 0.002249208322979257\n",
      "epoch:3966 from 100000\n",
      "loss: 0.002723065708210015\n",
      "epoch:3967 from 100000\n",
      "loss: 0.003387025798247123\n",
      "epoch:3968 from 100000\n",
      "loss: 0.002024579374079849\n",
      "epoch:3969 from 100000\n",
      "loss: 0.0023812658403130627\n",
      "epoch:3970 from 100000\n",
      "loss: 0.0030277522511141797\n",
      "epoch:3971 from 100000\n",
      "loss: 0.002608729187500103\n",
      "epoch:3972 from 100000\n",
      "loss: 0.0024692267108150645\n",
      "epoch:3973 from 100000\n",
      "loss: 0.0023758900088921564\n",
      "epoch:3974 from 100000\n",
      "loss: 0.002643430764379673\n",
      "epoch:3975 from 100000\n",
      "loss: 0.002046792958367405\n",
      "epoch:3976 from 100000\n",
      "loss: 0.0030708459092068097\n",
      "epoch:3977 from 100000\n",
      "loss: 0.002365063771065934\n",
      "epoch:3978 from 100000\n",
      "loss: 0.0025035834058826367\n",
      "epoch:3979 from 100000\n",
      "loss: 0.002881978837990573\n",
      "epoch:3980 from 100000\n",
      "loss: 0.0023961756662060907\n",
      "epoch:3981 from 100000\n",
      "loss: 0.002738644529884482\n",
      "epoch:3982 from 100000\n",
      "loss: 0.002417202768356219\n",
      "epoch:3983 from 100000\n",
      "loss: 0.0019142640109635067\n",
      "epoch:3984 from 100000\n",
      "loss: 0.002553814676232946\n",
      "epoch:3985 from 100000\n",
      "loss: 0.002403422279996903\n",
      "epoch:3986 from 100000\n",
      "loss: 0.0021029309249627204\n",
      "epoch:3987 from 100000\n",
      "loss: 0.003006951417205528\n",
      "epoch:3988 from 100000\n",
      "loss: 0.002767332240750875\n",
      "epoch:3989 from 100000\n",
      "loss: 0.0024137731623454783\n",
      "epoch:3990 from 100000\n",
      "loss: 0.002203153672866165\n",
      "epoch:3991 from 100000\n",
      "loss: 0.002688870655674691\n",
      "epoch:3992 from 100000\n",
      "loss: 0.002601621681407096\n",
      "epoch:3993 from 100000\n",
      "loss: 0.0024603334025812273\n",
      "epoch:3994 from 100000\n",
      "loss: 0.002543995135908972\n",
      "epoch:3995 from 100000\n",
      "loss: 0.0027525599089983643\n",
      "epoch:3996 from 100000\n",
      "loss: 0.0025034136271244734\n",
      "epoch:3997 from 100000\n",
      "loss: 0.0025573756054058094\n",
      "epoch:3998 from 100000\n",
      "loss: 0.002881856794772375\n",
      "epoch:3999 from 100000\n",
      "loss: 0.003193423056586653\n",
      "epoch:4000 from 100000\n",
      "loss: 0.002461534500127982\n",
      "epoch:4001 from 100000\n",
      "loss: 0.002588279233890489\n",
      "epoch:4002 from 100000\n",
      "loss: 0.0022326062111233114\n",
      "epoch:4003 from 100000\n",
      "loss: 0.0023056696448122693\n",
      "epoch:4004 from 100000\n",
      "loss: 0.0025143416647637196\n",
      "epoch:4005 from 100000\n",
      "loss: 0.0024986247235290193\n",
      "epoch:4006 from 100000\n",
      "loss: 0.002826656600774413\n",
      "epoch:4007 from 100000\n",
      "loss: 0.0019955362465395865\n",
      "epoch:4008 from 100000\n",
      "loss: 0.002395128681229153\n",
      "epoch:4009 from 100000\n",
      "loss: 0.0029036107034698944\n",
      "epoch:4010 from 100000\n",
      "loss: 0.002571326829638565\n",
      "epoch:4011 from 100000\n",
      "loss: 0.002838019128233782\n",
      "epoch:4012 from 100000\n",
      "loss: 0.002225517618541062\n",
      "epoch:4013 from 100000\n",
      "loss: 0.0029988694593896525\n",
      "epoch:4014 from 100000\n",
      "loss: 0.002547510763633909\n",
      "epoch:4015 from 100000\n",
      "loss: 0.0027218357124638715\n",
      "epoch:4016 from 100000\n",
      "loss: 0.002620279515127777\n",
      "epoch:4017 from 100000\n",
      "loss: 0.002566726638633554\n",
      "epoch:4018 from 100000\n",
      "loss: 0.00281132967592157\n",
      "epoch:4019 from 100000\n",
      "loss: 0.002188412116794325\n",
      "epoch:4020 from 100000\n",
      "loss: 0.0025111583717603025\n",
      "epoch:4021 from 100000\n",
      "loss: 0.002859603220031204\n",
      "epoch:4022 from 100000\n",
      "loss: 0.002517627117555321\n",
      "epoch:4023 from 100000\n",
      "loss: 0.0021783803665584855\n",
      "epoch:4024 from 100000\n",
      "loss: 0.0022460292415189542\n",
      "epoch:4025 from 100000\n",
      "loss: 0.0022995989619687407\n",
      "epoch:4026 from 100000\n",
      "loss: 0.00207053820594183\n",
      "epoch:4027 from 100000\n",
      "loss: 0.0023725009115054263\n",
      "epoch:4028 from 100000\n",
      "loss: 0.002536112525177714\n",
      "epoch:4029 from 100000\n",
      "loss: 0.0022177315382236884\n",
      "epoch:4030 from 100000\n",
      "loss: 0.002191799328741695\n",
      "epoch:4031 from 100000\n",
      "loss: 0.0023977423568908307\n",
      "epoch:4032 from 100000\n",
      "loss: 0.002017852971178513\n",
      "epoch:4033 from 100000\n",
      "loss: 0.0023892532166769342\n",
      "epoch:4034 from 100000\n",
      "loss: 0.002161627158841384\n",
      "epoch:4035 from 100000\n",
      "loss: 0.0018441293733388875\n",
      "epoch:4036 from 100000\n",
      "loss: 0.002389506472724179\n",
      "epoch:4037 from 100000\n",
      "loss: 0.002527964335257593\n",
      "epoch:4038 from 100000\n",
      "loss: 0.0023585567563770837\n",
      "epoch:4039 from 100000\n",
      "loss: 0.002576367901042431\n",
      "epoch:4040 from 100000\n",
      "loss: 0.0023520026688856888\n",
      "epoch:4041 from 100000\n",
      "loss: 0.002853081215164366\n",
      "epoch:4042 from 100000\n",
      "loss: 0.002494725910565876\n",
      "epoch:4043 from 100000\n",
      "loss: 0.002528020918759921\n",
      "epoch:4044 from 100000\n",
      "loss: 0.003110497132496448\n",
      "epoch:4045 from 100000\n",
      "loss: 0.002095620306263868\n",
      "epoch:4046 from 100000\n",
      "loss: 0.0028338211875507113\n",
      "epoch:4047 from 100000\n",
      "loss: 0.0026257299051906102\n",
      "epoch:4048 from 100000\n",
      "loss: 0.0024480033998770523\n",
      "epoch:4049 from 100000\n",
      "loss: 0.002223191552843673\n",
      "epoch:4050 from 100000\n",
      "loss: 0.002266186756173738\n",
      "epoch:4051 from 100000\n",
      "loss: 0.0026381745982462235\n",
      "epoch:4052 from 100000\n",
      "loss: 0.0026582933389249774\n",
      "epoch:4053 from 100000\n",
      "loss: 0.0021694236473308592\n",
      "epoch:4054 from 100000\n",
      "loss: 0.0022681586631563277\n",
      "epoch:4055 from 100000\n",
      "loss: 0.002684154476810363\n",
      "epoch:4056 from 100000\n",
      "loss: 0.002908053235216812\n",
      "epoch:4057 from 100000\n",
      "loss: 0.0029314179771517956\n",
      "epoch:4058 from 100000\n",
      "loss: 0.0024476569802232007\n",
      "epoch:4059 from 100000\n",
      "loss: 0.0021643102545283455\n",
      "epoch:4060 from 100000\n",
      "loss: 0.0030419150666602945\n",
      "epoch:4061 from 100000\n",
      "loss: 0.0022314830049481316\n",
      "epoch:4062 from 100000\n",
      "loss: 0.002322548120588408\n",
      "epoch:4063 from 100000\n",
      "loss: 0.0032635609584303893\n",
      "epoch:4064 from 100000\n",
      "loss: 0.0024275107013522756\n",
      "epoch:4065 from 100000\n",
      "loss: 0.002071937262670076\n",
      "epoch:4066 from 100000\n",
      "loss: 0.002789422390122809\n",
      "retrying for data\n",
      "epoch:4067 from 100000\n",
      "loss: 0.00220149656612989\n",
      "epoch:4068 from 100000\n",
      "loss: 0.0022344273178021012\n",
      "epoch:4069 from 100000\n",
      "loss: 0.0022371629485462563\n",
      "epoch:4070 from 100000\n",
      "loss: 0.002432956239573514\n",
      "epoch:4071 from 100000\n",
      "loss: 0.0027360605964190654\n",
      "epoch:4072 from 100000\n",
      "loss: 0.002288572629652208\n",
      "epoch:4073 from 100000\n",
      "loss: 0.002782592145081253\n",
      "epoch:4074 from 100000\n",
      "loss: 0.0030356702372955624\n",
      "epoch:4075 from 100000\n",
      "loss: 0.0023416001695863997\n",
      "epoch:4076 from 100000\n",
      "loss: 0.0020503803611120475\n",
      "epoch:4077 from 100000\n",
      "loss: 0.002723611539039439\n",
      "epoch:4078 from 100000\n",
      "loss: 0.00263802260792545\n",
      "epoch:4079 from 100000\n",
      "loss: 0.0025181341679854086\n",
      "epoch:4080 from 100000\n",
      "loss: 0.0022074290852742706\n",
      "epoch:4081 from 100000\n",
      "loss: 0.00260627764463095\n",
      "epoch:4082 from 100000\n",
      "loss: 0.0021888499662026106\n",
      "epoch:4083 from 100000\n",
      "loss: 0.002513781807436999\n",
      "epoch:4084 from 100000\n",
      "loss: 0.002610641781132017\n",
      "epoch:4085 from 100000\n",
      "loss: 0.002197278999574337\n",
      "epoch:4086 from 100000\n",
      "loss: 0.002828722335124212\n",
      "epoch:4087 from 100000\n",
      "loss: 0.002580472749968976\n",
      "epoch:4088 from 100000\n",
      "loss: 0.0022829691048116274\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "Cell \u001b[0;32mIn[14], line 73\u001b[0m\n\u001b[1;32m     70\u001b[0m loss_total \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n\u001b[1;32m     71\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m i \u001b[38;5;129;01min\u001b[39;00m \u001b[38;5;28mrange\u001b[39m(epochs):\n\u001b[0;32m---> 73\u001b[0m     loss \u001b[38;5;241m=\u001b[39m \u001b[43mtrainer\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mtrain_step\u001b[49m\u001b[43m(\u001b[49m\u001b[43munet_number\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m \u001b[49m\u001b[43munet_to_train\u001b[49m\u001b[43m,\u001b[49m\u001b[43mmax_batch_size\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m \u001b[49m\u001b[43mbatch_size\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m     74\u001b[0m     \u001b[38;5;28mprint\u001b[39m(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mepoch:\u001b[39m\u001b[38;5;124m'\u001b[39m \u001b[38;5;241m+\u001b[39m \u001b[38;5;28mstr\u001b[39m(i\u001b[38;5;241m+\u001b[39m\u001b[38;5;241m1\u001b[39m) \u001b[38;5;241m+\u001b[39m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m from \u001b[39m\u001b[38;5;124m'\u001b[39m \u001b[38;5;241m+\u001b[39m \u001b[38;5;28mstr\u001b[39m(epochs))\n\u001b[1;32m     76\u001b[0m     \u001b[38;5;66;03m#print(\"At sample seen \" + str(my_dataset.no_of_samples_seen()))\u001b[39;00m\n",
      "File \u001b[0;32m~/anaconda3/envs/ds2f_m_i/lib/python3.8/site-packages/imagen_pytorch/trainer.py:622\u001b[0m, in \u001b[0;36mImagenTrainer.train_step\u001b[0;34m(self, unet_number, **kwargs)\u001b[0m\n\u001b[1;32m    619\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mcreate_train_iter()\n\u001b[1;32m    621\u001b[0m kwargs \u001b[38;5;241m=\u001b[39m {\u001b[38;5;124m'\u001b[39m\u001b[38;5;124munet_number\u001b[39m\u001b[38;5;124m'\u001b[39m: unet_number, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs}\n\u001b[0;32m--> 622\u001b[0m loss \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mstep_with_dl_iter\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mtrain_dl_iter\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    623\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mupdate(unet_number \u001b[38;5;241m=\u001b[39m unet_number)\n\u001b[1;32m    624\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m loss\n",
      "File \u001b[0;32m~/anaconda3/envs/ds2f_m_i/lib/python3.8/site-packages/imagen_pytorch/trainer.py:640\u001b[0m, in \u001b[0;36mImagenTrainer.step_with_dl_iter\u001b[0;34m(self, dl_iter, **kwargs)\u001b[0m\n\u001b[1;32m    638\u001b[0m dl_tuple_output \u001b[38;5;241m=\u001b[39m cast_tuple(\u001b[38;5;28mnext\u001b[39m(dl_iter))\n\u001b[1;32m    639\u001b[0m model_input \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mdict\u001b[39m(\u001b[38;5;28mlist\u001b[39m(\u001b[38;5;28mzip\u001b[39m(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mdl_tuple_output_keywords_names, dl_tuple_output)))\n\u001b[0;32m--> 640\u001b[0m loss \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mforward\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43m{\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mmodel_input\u001b[49m\u001b[43m}\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    641\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m loss\n",
      "File \u001b[0;32m~/anaconda3/envs/ds2f_m_i/lib/python3.8/site-packages/imagen_pytorch/trainer.py:136\u001b[0m, in \u001b[0;36mcast_torch_tensor.<locals>.inner\u001b[0;34m(model, *args, **kwargs)\u001b[0m\n\u001b[1;32m    133\u001b[0m args, kwargs_values \u001b[38;5;241m=\u001b[39m all_args[:split_kwargs_index], all_args[split_kwargs_index:]\n\u001b[1;32m    134\u001b[0m kwargs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mdict\u001b[39m(\u001b[38;5;28mtuple\u001b[39m(\u001b[38;5;28mzip\u001b[39m(kwargs_keys, kwargs_values)))\n\u001b[0;32m--> 136\u001b[0m out \u001b[38;5;241m=\u001b[39m \u001b[43mfn\u001b[49m\u001b[43m(\u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    137\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m out\n",
      "File \u001b[0;32m~/anaconda3/envs/ds2f_m_i/lib/python3.8/site-packages/imagen_pytorch/trainer.py:999\u001b[0m, in \u001b[0;36mImagenTrainer.forward\u001b[0;34m(self, unet_number, max_batch_size, *args, **kwargs)\u001b[0m\n\u001b[1;32m    996\u001b[0m     total_loss \u001b[38;5;241m+\u001b[39m\u001b[38;5;241m=\u001b[39m loss\u001b[38;5;241m.\u001b[39mitem()\n\u001b[1;32m    998\u001b[0m     \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mtraining:\n\u001b[0;32m--> 999\u001b[0m         \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43maccelerator\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbackward\u001b[49m\u001b[43m(\u001b[49m\u001b[43mloss\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1001\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m total_loss\n",
      "File \u001b[0;32m~/anaconda3/envs/ds2f_m_i/lib/python3.8/site-packages/accelerate/accelerator.py:1987\u001b[0m, in \u001b[0;36mAccelerator.backward\u001b[0;34m(self, loss, **kwargs)\u001b[0m\n\u001b[1;32m   1985\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m\n\u001b[1;32m   1986\u001b[0m \u001b[38;5;28;01melif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mscaler \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n\u001b[0;32m-> 1987\u001b[0m     \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mscaler\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mscale\u001b[49m\u001b[43m(\u001b[49m\u001b[43mloss\u001b[49m\u001b[43m)\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbackward\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1988\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m   1989\u001b[0m     loss\u001b[38;5;241m.\u001b[39mbackward(\u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n",
      "File \u001b[0;32m~/anaconda3/envs/ds2f_m_i/lib/python3.8/site-packages/torch/_tensor.py:492\u001b[0m, in \u001b[0;36mTensor.backward\u001b[0;34m(self, gradient, retain_graph, create_graph, inputs)\u001b[0m\n\u001b[1;32m    482\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m has_torch_function_unary(\u001b[38;5;28mself\u001b[39m):\n\u001b[1;32m    483\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m handle_torch_function(\n\u001b[1;32m    484\u001b[0m         Tensor\u001b[38;5;241m.\u001b[39mbackward,\n\u001b[1;32m    485\u001b[0m         (\u001b[38;5;28mself\u001b[39m,),\n\u001b[0;32m   (...)\u001b[0m\n\u001b[1;32m    490\u001b[0m         inputs\u001b[38;5;241m=\u001b[39minputs,\n\u001b[1;32m    491\u001b[0m     )\n\u001b[0;32m--> 492\u001b[0m \u001b[43mtorch\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mautograd\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbackward\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m    493\u001b[0m \u001b[43m    \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mgradient\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mretain_graph\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mcreate_graph\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43minputs\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43minputs\u001b[49m\n\u001b[1;32m    494\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[0;32m~/anaconda3/envs/ds2f_m_i/lib/python3.8/site-packages/torch/autograd/__init__.py:251\u001b[0m, in \u001b[0;36mbackward\u001b[0;34m(tensors, grad_tensors, retain_graph, create_graph, grad_variables, inputs)\u001b[0m\n\u001b[1;32m    246\u001b[0m     retain_graph \u001b[38;5;241m=\u001b[39m create_graph\n\u001b[1;32m    248\u001b[0m \u001b[38;5;66;03m# The reason we repeat the same comment below is that\u001b[39;00m\n\u001b[1;32m    249\u001b[0m \u001b[38;5;66;03m# some Python versions print out the first line of a multi-line function\u001b[39;00m\n\u001b[1;32m    250\u001b[0m \u001b[38;5;66;03m# calls in the traceback and some print out the last line\u001b[39;00m\n\u001b[0;32m--> 251\u001b[0m \u001b[43mVariable\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_execution_engine\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mrun_backward\u001b[49m\u001b[43m(\u001b[49m\u001b[43m  \u001b[49m\u001b[38;5;66;43;03m# Calls into the C++ engine to run the backward pass\u001b[39;49;00m\n\u001b[1;32m    252\u001b[0m \u001b[43m    \u001b[49m\u001b[43mtensors\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    253\u001b[0m \u001b[43m    \u001b[49m\u001b[43mgrad_tensors_\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    254\u001b[0m \u001b[43m    \u001b[49m\u001b[43mretain_graph\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    255\u001b[0m \u001b[43m    \u001b[49m\u001b[43mcreate_graph\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    256\u001b[0m \u001b[43m    \u001b[49m\u001b[43minputs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    257\u001b[0m \u001b[43m    \u001b[49m\u001b[43mallow_unreachable\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m    258\u001b[0m \u001b[43m    \u001b[49m\u001b[43maccumulate_grad\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m    259\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "from imagen_pytorch import Unet, Imagen, ImagenTrainer,NullUnet\n",
    "from imagen_pytorch.data import Dataset\n",
    "\n",
    "unet0 = NullUnet()  # add a placeholder \"null\" unet for the base unet\n",
    "\n",
    "unet1 = Unet(\n",
    "    dim = unet1_dim,\n",
    "    cond_dim = 512,\n",
    "    dim_mults = (1, 2, 4, 8),\n",
    "    num_resnet_blocks = 3,\n",
    "    layer_attns = (False, True, True, True),\n",
    "    layer_cross_attns = (False, True, True, True)\n",
    ")\n",
    "\n",
    "unet2 = Unet(\n",
    "    dim = unet2_dim,\n",
    "    cond_dim = 512,\n",
    "    dim_mults = (1, 2, 4, 8),\n",
    "    num_resnet_blocks = (2, 4, 8, 8),\n",
    "    layer_attns = (False, False, False, True),\n",
    "    layer_cross_attns = (False, False, False, True)\n",
    ")\n",
    "\n",
    "if(ignore_image_guide):\n",
    "    unets = (unet1, unet2)\n",
    "    image_sizes= (unet1_image_size, image_size)\n",
    "    unet_to_train = UNET #1\n",
    "    #start_image_or_video = None\n",
    "else:\n",
    "    unets = (unet0,unet1, unet2)\n",
    "    image_sizes= (begin_with_image_size,unet1_image_size, image_size)\n",
    "    unet_to_train = UNET + 1 #2\n",
    "    #start_image_or_video = input2[:1,:]\n",
    "\n",
    "#print(input2[:1,:])\n",
    "#print(input2)\n",
    "\n",
    "imagen = Imagen(\n",
    "    unets = unets,\n",
    "    image_sizes = image_sizes,\n",
    "    timesteps = timesteps,\n",
    "    cond_drop_prob = 0.1\n",
    ").cuda()\n",
    "\n",
    "\n",
    "\n",
    "trainer = ImagenTrainer(\n",
    "    imagen = imagen,\n",
    "    split_valid_from_train = False, # whether to split the validation dataset from the training\n",
    "    checkpoint_every = save_model_every,\n",
    "    checkpoint_path = model_filename,\n",
    "    max_checkpoints_keep = max_checkpoints_keep,\n",
    "    only_train_unet_number = unet_to_train\n",
    ").cuda()\n",
    "\n",
    "\n",
    "trainer.add_train_dataloader(my_dataloader)\n",
    "\n",
    "#if not os.path.exists(imagen_samples_folder):\n",
    "#    os.makedirs(imagen_samples_folder)\n",
    "\n",
    "#my_file = Path(model_filename)\n",
    "#if my_file.is_file():\n",
    "#    print('Using model file ' + model_filename)\n",
    "#    trainer.load(model_filename)\n",
    "\n",
    "#trainer.load_from_checkpoint_folder()\n",
    "#print(epochs)\n",
    "pp = 0\n",
    "loss_total = None\n",
    "for i in range(epochs):\n",
    "    \n",
    "    loss = trainer.train_step(unet_number = unet_to_train,max_batch_size = batch_size)\n",
    "    print('epoch:' + str(i+1) + ' from ' + str(epochs))\n",
    "    \n",
    "    #print(\"At sample seen \" + str(my_dataset.no_of_samples_seen()))\n",
    "\n",
    "    print(f'loss: {loss}')\n",
    "    loss_total = plot_loss(my_dataset.no_of_samples_seen(),loss,not pp%(16 * 250),loss_total)\n",
    "    #print(loss_list)\n",
    "    #trainer.save_to_checkpoint_folder()\n",
    "    \n",
    "    last_sample_no = my_dataset.no_of_samples_seen()\n",
    "\n",
    "    if not pp%(16 * 100):\n",
    "\n",
    "        if(UNET == 1):\n",
    "            with open(model_filename + '/last_sample_no1.picke', 'wb') as handle:\n",
    "                pickle.dump(last_sample_no, handle)\n",
    "        else:\n",
    "            with open(model_filename + '/last_sample_no2.picke', 'wb') as handle:\n",
    "                pickle.dump(last_sample_no, handle)\n",
    "\n",
    "    pp = pp + 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "checkpoint saved to /data_store/checkpoint/checkpoint.215654.pt\n"
     ]
    }
   ],
   "source": [
    "trainer.save_to_checkpoint_folder()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "if(UNET == 1):\n",
    "    with open(model_filename + '/last_sample_no1.picke', 'wb') as handle:\n",
    "        pickle.dump(last_sample_no, handle)\n",
    "else:\n",
    "    with open(model_filename + '/last_sample_no2.picke', 'wb') as handle:\n",
    "        pickle.dump(last_sample_no, handle)\n",
    "loss_total = plot_loss(my_dataset.no_of_samples_seen(),loss,True,loss_total)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "my_dataset.destroy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from torchvision.utils import save_image\n",
    "from torchvision.transforms import Resize\n",
    "\n",
    "now =time.time()\n",
    "seconds = now\n",
    "output, input0 = my_dataset.__getitem__(305000)\n",
    "save_image(output,'ground_truth.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "resize_img_tensor = Resize((begin_with_image_size, begin_with_image_size))\n",
    "\n",
    "if(ignore_image_guide):\n",
    "    start_image_or_video = None\n",
    "else:\n",
    "    start_image_or_video = resize_img_tensor(output)\n",
    "\n",
    "#start_image_or_video = torch.unsqueeze(start_image_or_video,0)\n",
    "input0 = torch.unsqueeze(input0,0)\n",
    "output = torch.unsqueeze(output,0)\n",
    "start_image_or_video = torch.unsqueeze(start_image_or_video,0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "unet_to_stop = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "cond_scale = 2\n",
    "\n",
    "\n",
    "print(start_image_or_video.shape)\n",
    "print(input0.shape)\n",
    "print(output.shape)\n",
    "\n",
    "images = trainer.sample(text_embeds=input0,start_image_or_video = start_image_or_video,start_at_unet_number = 2  \n",
    "            ,stop_at_unet_number=unet_to_stop,batch_size = 1, return_pil_images = True,cond_scale=cond_scale) # returns List[Image]\n",
    "images[0].save('g'  + str(cond_scale) + '_generated.png')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "ds2f_m_i",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.16"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "78dccc615e96ab04385280185a87a524fe0822daf5dbad2f4bf2e7d7b28366a7"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}