merge main

examples/TCL_rag/config.yaml

@@ -1,33 +1,33 @@
 llm:
   name: openai
-  base_url: "https://api.gptsapi.net/v1"
-  api_key: "sk-2T06b7c7f9c3870049fbf8fada596b0f8ef908d1e233KLY2"
+  base_url: "xxx"
+  api_key: "xxx"
   model: "gpt-4.1-mini"
 
 embedding:
   name: huggingface
-  model_name: "/finance_ML/dataarc_syn_database/model/Qwen/qwen_embedding_0.6B"
+  model_name: "xxx"
   model_kwargs:
     device: "cuda:0"
 
 
 
 store:
   name: faiss
-  folder_path: /data/FinAi_Mapping_Knowledge/chenmingzhen/test_faiss_store
+  folder_path: xxx
 
 
 bm25:
   name: bm25
   k: 10
-  data_path: /data/FinAi_Mapping_Knowledge/chenmingzhen/tog3_backend/TCL/syn_table_data/data_all_clearn_short_chunk_with_caption_desc.json
+  data_path: xxx
 
 retriever:
   name: vectorstore
 
 reranker:
   name: qwen3
-  model_name_or_path: "/finance_ML/dataarc_syn_database/model/Qwen/qwen_reranker_0.6B"
+  model_name_or_path: "xxx"
   device_id: "cuda:0"
 
 dataset:

examples/TCL_rag/test.py

@@ -24,9 +24,8 @@
                 vector_store_config=vector_store_config,
                 bm25_retriever_config=bm25_retriever_config)
 
-    result = rag.invoke("毛细管设计规范按照什么标准",k=20)
+    result = rag.invoke("模块机传感器端子不防呆的改善方案是什么？由哪个部门负责？",k=20)
 
-    answer = rag.answer("毛细管设计规范按照什么标准",result)
-
-
-    print(answer)
+    for i in result:
+        print(i)
+        print("-"*100)

rag_factory/Retrieval/Retriever/Retriever_BM25.py

@@ -5,7 +5,7 @@
 from concurrent.futures import ThreadPoolExecutor
 from typing import Any, Callable, Dict, Iterable, List, Optional, Sequence
 from dataclasses import dataclass, field
-
+import uuid
 from pydantic import ConfigDict, Field, model_validator
 
 logger = logging.getLogger(__name__)
@@ -207,7 +207,7 @@ def from_texts(
                     f"与 texts 长度 ({len(texts_list)}) 不匹配"
                 )
         else:
-            ids_list = [None for _ in texts_list]
+            ids_list = [str(uuid.uuid4()) for _ in texts_list]
 
         # 预处理文本
         logger.info(f"正在预处理 {len(texts_list)} 个文本...")

rag_factory/Retrieval/Retriever/Retriever_MultiPath.py

@@ -50,8 +50,8 @@ def _get_relevant_documents(self, query: str, **kwargs: Any) -> List[Document]:
 
         Note:
             - 每个检索器的结果会被转换为RetrievalResult格式
-            - 支持多种输入格式：Document对象、字典格式、字符串等
-            - 融合后的结果会将score和rank信息保存在Document的metadata中
+            - 输入只会是Document对象
+            - 融合后的结果只返回排序好的Document对象
         """
         top_k = kwargs.get('top_k', 10)
 
@@ -65,43 +65,12 @@ def _get_relevant_documents(self, query: str, **kwargs: Any) -> List[Document]:
                 # 转换为RetrievalResult格式
                 formatted_results = []
                 for i, doc in enumerate(documents):
-                    if isinstance(doc, Document):
-                        # 如果是Document对象
-                        retrieval_result = RetrievalResult(
-                            document=doc,
-                            score=getattr(doc, 'score', 1.0),
-                            rank=i + 1
-                        )
-                    elif isinstance(doc, dict):
-                        # 如果返回的是字典格式，需要转换为Document对象
-                        content = doc.get('content', '')
-                        metadata = doc.get('metadata', {})
-                        doc_id = doc.get('id')
-
-                        document = Document(
-                            content=content,
-                            metadata=metadata,
-                            id=doc_id
-                        )
-
-                        retrieval_result = RetrievalResult(
-                            document=document,
-                            score=doc.get('score', 1.0),
-                            rank=i + 1
-                        )
-                    else:
-                        # 如果是字符串或其他格式，转换为Document对象
-                        document = Document(
-                            content=str(doc),
-                            metadata={},
-                            id=None
-                        )
-
-                        retrieval_result = RetrievalResult(
-                            document=document,
-                            score=1.0,
-                            rank=i + 1
-                        )
+                    # 输入只会是Document对象
+                    retrieval_result = RetrievalResult(
+                        document=doc,
+                        score=getattr(doc, 'score', 1.0),
+                        rank=i + 1
+                    )
                     formatted_results.append(retrieval_result)
 
                 all_results.append(formatted_results)
@@ -116,16 +85,10 @@ def _get_relevant_documents(self, query: str, **kwargs: Any) -> List[Document]:
 
         fused_results = self.fusion_method.fuse(all_results, top_k)
 
-        # 转换回Document格式
+        # 转换回Document格式，只返回排序好的Document对象
         documents = []
         for result in fused_results:
-            doc = result.document
-            # 将score和rank添加到metadata中以便保留
-            if doc.metadata is None:
-                doc.metadata = {}
-            doc.metadata['score'] = result.score
-            doc.metadata['rank'] = result.rank
-            documents.append(doc)
+            documents.append(result.document)
 
         return documents
 

requirements.txt

@@ -14,15 +14,13 @@ llama-index
 llama-index-core
 peewee
 
-mineru[core]
+
 rank_bm25
 faiss_gpu
 
 
-
-# streamlit 
+# for ocr
 PyMuPDF
-openai
 qwen_vl_utils
 transformers==4.51.3
 huggingface_hub
@@ -31,3 +29,6 @@ flash-attn==2.8.0.post2
 # for GLIBC 2.31, please use flash-attn==2.7.4.post1 instead of flash-attn==2.8.0.post2
 accelerate
 dashscope
+torch==2.7.0 torchvision==0.22.0 torchaudio==2.7.0 --index-url https://download.pytorch.org/whl/cu128
+
+mineru[core]