Skip to content

Commit

Permalink
Feat: Support internvl2 and internvl stream (#2079)
Browse files Browse the repository at this point in the history
  • Loading branch information
@amumu96
amumu96 authored Aug 16, 2024
1 parent f5229a2 commit e4d2257
Show file tree
Hide file tree
Showing 10 changed files with 1,278 additions and 271 deletions.
4 changes: 2 additions & 2 deletions xinference/model/llm/__init__.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -127,7 +127,7 @@ def _install():
from .transformers.minicpmv26 import MiniCPMV26Model
from .transformers.qwen_vl import QwenVLChatModel
from .transformers.yi_vl import YiVLChatModel
from .vllm.core import VLLMChatModel, VLLMModel
from .vllm.core import VLLMChatModel, VLLMModel, VLLMVisionModel

try:
from .transformers.omnilmm import OmniLMMModel
Expand All @@ -145,7 +145,7 @@ def _install():
]
)
SGLANG_CLASSES.extend([SGLANGModel, SGLANGChatModel])
VLLM_CLASSES.extend([VLLMModel, VLLMChatModel])
VLLM_CLASSES.extend([VLLMModel, VLLMChatModel, VLLMVisionModel])
MLX_CLASSES.extend([MLXModel, MLXChatModel])
TRANSFORMERS_CLASSES.extend(
[
Expand Down
185 changes: 176 additions & 9 deletions xinference/model/llm/llm_family.json
View file
Original file line number Diff line number Diff line change
Expand Up @@ -7083,43 +7083,210 @@
"model_format": "pytorch",
"model_size_in_billions": 2,
"quantizations": [
"none"
"4-bit",
"8-bit",
"none"
],
"model_id": "OpenGVLab/Mini-InternVL-Chat-2B-V1-5",
"model_revision": "ce3f67acff17281bacbf4b156f402a0580fb9605"
"model_revision": "ecbbd21dcf38caa74d925967b997167b0c7b3f47"
},
{
"model_format": "pytorch",
"model_size_in_billions": 4,
"quantizations": [
"4-bit",
"8-bit",
"none"
],
"model_id": "OpenGVLab/Mini-InternVL-Chat-4B-V1-5",
"model_revision": "ce1559ddf9d87f5130aa5233b0e93b95e4e4161a"
},
{
"model_format": "pytorch",
"model_size_in_billions": 26,
"quantizations": [
"none"
"4-bit",
"8-bit",
"none"
],
"model_id": "OpenGVLab/InternVL-Chat-V1-5",
"model_revision": "e822119e5806946ce128043023a73d715ecabf8d"
"model_revision": "9db32d9127cac0c85961e169d75da57a18a847b1"
}
],
"prompt_style": {
"style_name": "INTERNVL",
"system_prompt": "You are InternLM (书生·浦语), a helpful, honest, and harmless AI assistant developed by Shanghai AI Laboratory (上海人工智能实验室).",
"roles": [
"<|im_start|>user",
"<|im_start|>assistant"
],
"intra_message_sep": "<|im_end|>",
"stop_token_ids": [
2,
92543,
92542
],
"stop": [
"</s>",
"<|im_end|>",
"<|im_start|>"
]
}
},
{
"version": 1,
"context_length": 32768,
"model_name": "internvl2",
"model_lang": [
"en",
"zh"
],
"model_ability": [
"chat",
"vision"
],
"model_description": "InternVL 2 is an open-source multimodal large language model (MLLM) to bridge the capability gap between open-source and proprietary commercial models in multimodal understanding. ",
"model_specs": [
{
"model_format": "pytorch",
"model_size_in_billions": 1,
"quantizations": [
"4-bit",
"8-bit",
"none"
],
"model_id": "OpenGVLab/InternVL2-1B",
"model_revision": "a9fc14aea824b6ea1d44f8778cad6b35512c4ce1"
},
{
"model_format": "pytorch",
"model_size_in_billions": 2,
"quantizations": [
"4-bit",
"8-bit",
"none"
],
"model_id": "OpenGVLab/InternVL2-2B",
"model_revision": "422ad7c6335917bfb514958233955512338485a6"
},
{
"model_format": "awq",
"model_size_in_billions": 2,
"quantizations": [
"Int4"
],
"model_id": "OpenGVLab/InternVL2-2B-AWQ",
"model_revision": "701bc3fc098a8a3b686b3b4135cfb77202be89e0"
},
{
"model_format": "pytorch",
"model_size_in_billions": 4,
"quantizations": [
"4-bit",
"8-bit",
"none"
],
"model_id": "OpenGVLab/InternVL2-4B",
"model_revision": "b50544dafada6c41e80bfde2f57cc9b0140fc21c"
},
{
"model_format": "awq",
"model_size_in_billions": 4,
"quantizations": [
"Int4"
],
"model_id": "OpenGVLab/InternVL2-8B-AWQ",
"model_revision": "9f1a4756b7ae18eb26d8a22b618dfc283e8193b3"
},
{
"model_format": "pytorch",
"model_size_in_billions": 8,
"quantizations": [
"4-bit",
"8-bit",
"none"
],
"model_id": "OpenGVLab/InternVL2-8B",
"model_revision": "3bfd3664dea4f3da628785f5125d30f889701253"
},
{
"model_format": "pytorch",
"model_size_in_billions": 26,
"quantizations": [
"4-bit",
"8-bit",
"none"
],
"model_id": "OpenGVLab/InternVL2-26B",
"model_revision": "b9f3c7e6d575b0115e076a3ffc46fd20b7586899"
},
{
"model_format": "awq",
"model_size_in_billions": 26,
"quantizations": [
"Int8"
"Int4"
],
"model_id": "OpenGVLab/InternVL2-26B-AWQ",
"model_revision": "469e0019ffd251e22ff6501a5c2321964e86ef0d"
},
{
"model_format": "pytorch",
"model_size_in_billions": 40,
"quantizations": [
"4-bit",
"8-bit",
"none"
],
"model_id": "OpenGVLab/InternVL2-40B",
"model_revision": "725a12063bb855c966e30a0617d0ccd9e870d772"
},
{
"model_format": "awq",
"model_size_in_billions": 40,
"quantizations": [
"Int4"
],
"model_id": "OpenGVLab/InternVL2-40B-AWQ",
"model_revision": "d92e140f6dfe8ea9679924c6a31898f42c4e1846"
},
{
"model_format": "pytorch",
"model_size_in_billions": 76,
"quantizations": [
"4-bit",
"8-bit",
"none"
],
"model_id": "OpenGVLab/InternVL2-Llama3-76B",
"model_revision": "cf7914905f78e9e3560ddbd6f5dfc39becac494f"
},
{
"model_format": "awq",
"model_size_in_billions": 76,
"quantizations": [
"Int4"
],
"model_id": "OpenGVLab/InternVL-Chat-V1-5-{quantization}",
"model_revision": "acaaed06937c603ab04f084216ecb0268160f538"
"model_id": "OpenGVLab/InternVL2-Llama3-76B-AWQ",
"model_revision": "1bc796bf80f2ebc7d6a14c15f55217a4600d50a4"
}
],
"prompt_style": {
"style_name": "INTERNLM2",
"style_name": "INTERNVL",
"system_prompt": "You are InternLM (书生·浦语), a helpful, honest, and harmless AI assistant developed by Shanghai AI Laboratory (上海人工智能实验室).",
"roles": [
"<|im_start|>user",
"<|im_start|>assistant"
],
"intra_message_sep": "<|im_end|>",
"stop_token_ids": [
2,
92543,
92542
],
"stop": [
"<|im_end|>"
"</s>",
"<|im_end|>",
"<|im_start|>"
]
}
},
Expand Down
Loading

0 comments on commit e4d2257

Please sign in to comment.