diff --git a/README.md b/README.md
index 1b77cbae..0684f50a 100644
--- a/README.md
+++ b/README.md
@@ -8,8 +8,6 @@
git clone git@github.com:aigc-apps/PAI-RAG.git
```
-注：如果需要调用Open AI，需要使用新加坡开发机器，不能连通弹内Gitlab环境，需要手动将代码上传到新加坡机器
-
### Step2: 配置环境
本项目使用poetry进行管理，建议在安装环境之前先创建一个空环境。为了确保环境一致性并避免因Python版本差异造成的问题，我们指定Python版本为3.10。
@@ -36,7 +34,7 @@ poetry install
pai_rag run [--host HOST] [--port PORT] [--config CONFIG_FILE]
```
-现在你可以使用命令行向服务侧发送API请求，或者直接打开http://HOST:PORT。
+现在你可以使用命令行向服务侧发送API请求，或者直接打开http://localhost:8000
1.
diff --git a/src/pai_rag/app/api/models.py b/src/pai_rag/app/api/models.py
index 4ad55816..af0d421d 100644
--- a/src/pai_rag/app/api/models.py
+++ b/src/pai_rag/app/api/models.py
@@ -35,9 +35,13 @@ class RagResponse(BaseModel):
class LlmResponse(BaseModel):
answer: str
+class ContextDoc(BaseModel):
+ text: str
+ score: float
+ metadata: Dict
class RetrievalResponse(BaseModel):
- answer: str
+ docs: List[ContextDoc]
class KnowledgeInput(BaseModel):
diff --git a/src/pai_rag/app/web/rag_client.py b/src/pai_rag/app/web/rag_client.py
index af4cc467..1be5b5d3 100644
--- a/src/pai_rag/app/web/rag_client.py
+++ b/src/pai_rag/app/web/rag_client.py
@@ -76,7 +76,8 @@ def query_vector(self, text: str):
session_id = r.headers["x-session-id"]
response = dotdict(json.loads(r.text))
response.session_id = session_id
-
+ formatted_text = "\n\n".join([f"""[Doc {i+1}] [score: {doc["score"]}]\n{doc["text"]}""" for i,doc in enumerate(response["docs"])])
+ response["answer"] = formatted_text
return response
def add_knowledge(self, file_dir: str, enable_qa_extraction: bool):
diff --git a/src/pai_rag/app/web/ui.py b/src/pai_rag/app/web/ui.py
index caf0b299..1cc97753 100644
--- a/src/pai_rag/app/web/ui.py
+++ b/src/pai_rag/app/web/ui.py
@@ -25,7 +25,7 @@
#### \N{fire} Platform: [PAI](https://help.aliyun.com/zh/pai) / [PAI-EAS](https://www.aliyun.com/product/bigdata/learn/eas) / [PAI-DSW](https://pai.console.aliyun.com/notebook) \N{rocket} Supported VectorStores: [Hologres](https://www.aliyun.com/product/bigdata/hologram) / [ElasticSearch](https://www.aliyun.com/product/bigdata/elasticsearch) / [AnalyticDB](https://www.aliyun.com/product/apsaradb/gpdb) / [FAISS](https://python.langchain.com/docs/integrations/vectorstores/faiss)
- #### \N{fire} API Docs \N{rocket} View Tracing \N{fire} 欢迎加入【PAI】RAG答疑群 27370042974
+ #### \N{fire} API Docs \N{rocket} \N{fire} 欢迎加入【PAI】RAG答疑群 27370042974
"""
css_style = """
@@ -145,11 +145,7 @@ def respond(input_elements: List[Any]):
def create_ui():
with gr.Blocks(css=css_style) as homepage:
- gr.Markdown(
- value=welcome_message_markdown.format(
- EAS_TRACE_ENDPOINT=environ.get("EAS_ARIZE_PHOENIX_URL", "")
- )
- )
+ gr.Markdown(value=welcome_message_markdown)
with gr.Tab("\N{rocket} Settings"):
with gr.Row():
diff --git a/src/pai_rag/core/rag_application.py b/src/pai_rag/core/rag_application.py
index e100d5ba..93456852 100644
--- a/src/pai_rag/core/rag_application.py
+++ b/src/pai_rag/core/rag_application.py
@@ -9,6 +9,7 @@
RetrievalQuery,
RagResponse,
LlmResponse,
+ ContextDoc,
RetrievalResponse,
)
@@ -64,15 +65,11 @@ async def aquery_vectordb(self, query: RetrievalQuery) -> RetrievalResponse:
session_id = correlation_id.get() or DEFAULT_SESSION_ID
self.logger.info(f"Get session ID: {session_id}.")
node_results = await self.retriever.aretrieve(query.question)
- text_results = [
- score_node.node.get_content().replace("\n", " ")
+
+ docs = [ContextDoc(text = score_node.node.get_content(), metadata=score_node.node.metadata, score=score_node.score)
for score_node in node_results
]
- formatted_results = [
- f"**Doc [{i}]**: {text} \n" for i, text in enumerate(text_results)
- ]
- result = "\n".join(formatted_results)
- return RetrievalResponse(answer=result)
+ return RetrievalResponse(docs=docs)
async def aquery(self, query: RagQuery) -> RagResponse:
"""Query answer from RAG App asynchronously.
diff --git a/src/pai_rag/integrations/llms/paieas/README.md b/src/pai_rag/integrations/llms/paieas/README.md
deleted file mode 100644
index 798a44d4..00000000
--- a/src/pai_rag/integrations/llms/paieas/README.md
+++ /dev/null
@@ -1 +0,0 @@
-### PAI EAS llm integration
diff --git a/src/pai_rag/integrations/llms/paieas/__init__.py b/src/pai_rag/integrations/llms/paieas/__init__.py
deleted file mode 100644
index e4b3dd89..00000000
--- a/src/pai_rag/integrations/llms/paieas/__init__.py
+++ /dev/null
@@ -1,3 +0,0 @@
-from llama_index.llms.paieas.base import PaiEAS
-
-__all__ = ["PaiEAS"]
diff --git a/src/pai_rag/integrations/llms/paieas/llama_index/llms/paieas/base.py b/src/pai_rag/integrations/llms/paieas/base.py
similarity index 99%
rename from src/pai_rag/integrations/llms/paieas/llama_index/llms/paieas/base.py
rename to src/pai_rag/integrations/llms/paieas/base.py
index eeb79310..40b7edf5 100644
--- a/src/pai_rag/integrations/llms/paieas/llama_index/llms/paieas/base.py
+++ b/src/pai_rag/integrations/llms/paieas/base.py
@@ -79,7 +79,7 @@ def __init__(
@classmethod
def class_name(cls) -> str:
- return "PaiEasLLM"
+ return "PaiEAS"
@property
def metadata(self) -> LLMMetadata:
diff --git a/src/pai_rag/integrations/llms/paieas/poetry.lock b/src/pai_rag/integrations/llms/paieas/poetry.lock
deleted file mode 100644
index 4e5b56dc..00000000
--- a/src/pai_rag/integrations/llms/paieas/poetry.lock
+++ /dev/null
@@ -1,7 +0,0 @@
-# This file is automatically @generated by Poetry 1.8.2 and should not be changed by hand.
-package = []
-
-[metadata]
-lock-version = "2.0"
-python-versions = "^3.10"
-content-hash = "53f2eabc9c26446fbcc00d348c47878e118afc2054778c3c803a0a8028af27d9"
diff --git a/src/pai_rag/integrations/llms/paieas/pyproject.toml b/src/pai_rag/integrations/llms/paieas/pyproject.toml
deleted file mode 100644
index afa20346..00000000
--- a/src/pai_rag/integrations/llms/paieas/pyproject.toml
+++ /dev/null
@@ -1,21 +0,0 @@
-[build-system]
-requires = ["poetry-core"]
-build-backend = "poetry.core.masonry.api"
-
-[tool.llamahub]
-contains_example = false
-import_path = "llama_index.llms.paieas"
-
-[tool.llamahub.class_authors]
-PaiEas = "llama-index"
-
-[tool.poetry]
-name = "llama-index-llms-paieas"
-version = "0.1.0"
-description = ""
-authors = ["陆逊 "]
-readme = "README.md"
-packages = [{include = "llama_index/"}]
-
-[tool.poetry.dependencies]
-python = "^3.10"
diff --git a/src/pai_rag/integrations/llms/paieas/tests/test_pai_eas_llm.py b/src/pai_rag/integrations/llms/paieas/tests/test_pai_eas_llm.py
deleted file mode 100644
index ef72ee26..00000000
--- a/src/pai_rag/integrations/llms/paieas/tests/test_pai_eas_llm.py
+++ /dev/null
@@ -1,55 +0,0 @@
-from llama_index.llms.paieas.base import PaiEAS
-from llama_index.core.base.llms.types import ChatMessage
-import os
-
-
-def _get_eas_llm() -> PaiEAS:
- eas_url = os.environ.get("TEST_EAS_URL", None)
- eas_token = os.environ.get("TEST_EAS_TOKEN", None)
-
- if not eas_url or not eas_token:
- return None
-
- eas_llm = PaiEAS(endpoint=eas_url, token=eas_token, model_name="EasCustomModel")
- return eas_llm
-
-
-def test_pai_eas_llm_complete():
- eas_llm = _get_eas_llm()
- if eas_llm:
- response = eas_llm.complete("你好呀，最近怎么样？")
- assert len(response.text) > 0
-
-
-def test_pai_eas_llm_stream_complete():
- eas_llm = _get_eas_llm()
-
- if eas_llm:
- response = eas_llm.stream_complete("你好呀，最近怎么样？")
- text = None
- for r in response:
- text = r.text
-
- assert len(text) > 0
-
-
-def test_pai_eas_llm_chat():
- eas_llm = _get_eas_llm()
- if eas_llm:
- chat_messages = [ChatMessage(role="user", content="你好呀，最近怎么样？")]
- response = eas_llm.chat(chat_messages)
- print(response.message.content)
- assert len(response.message.content) > 0
-
-
-def test_pai_eas_llm_stream_chat():
- eas_llm = _get_eas_llm()
- if eas_llm:
- chat_messages = [ChatMessage(role="user", content="你好呀，最近怎么样？")]
- response = eas_llm.stream_chat(chat_messages)
-
- text = None
- for r in response:
- text = r.message.content
- print(text)
- assert len(text) > 0
diff --git a/src/pai_rag/modules/llm/llm_module.py b/src/pai_rag/modules/llm/llm_module.py
index 3da49a8e..d5417622 100644
--- a/src/pai_rag/modules/llm/llm_module.py
+++ b/src/pai_rag/modules/llm/llm_module.py
@@ -4,7 +4,7 @@
from llama_index.llms.openai import OpenAI
from llama_index.llms.azure_openai import AzureOpenAI
from llama_index.llms.dashscope import DashScope
-from pai_rag.integrations.llms.paieas import PaiEAS
+from pai_rag.integrations.llms.paieas.base import PaiEAS
from pai_rag.modules.base.configurable_module import ConfigurableModule
from pai_rag.modules.base.module_constants import MODULE_PARAM_CONFIG