Functions — github.com/datawhalechina/all-in-rag

hub / github.com/datawhalechina/all-in-rag / functions

Functions495 in github.com/datawhalechina/all-in-rag

⨍ Functions 495◇ Types & classes 94

↓ 51 callersFunction_pcfg

(url='', hf_hub='', filename='', mean=None, std=None)

code/C3/visual_bge/visual_bge/eva_clip/pretrained.py:18

↓ 22 callersMethodencode

(self, image=None, text=None)

code/C3/visual_bge/visual_bge/modeling.py:114

↓ 20 callersMethod_safe_truncate

安全截取字符串，处理None值 Args: text: 输入文本 max_length: 最大长度 Returns: 截取后的字符串

code/C9/rag_modules/milvus_index_construction.py:48

↓ 14 callersMethodhas_collection

检查集合是否存在 Returns: 集合是否存在

code/C9/rag_modules/milvus_index_construction.py:461

↓ 12 callersMethodcreate_index

创建向量索引 Returns: 是否创建成功

code/C9/rag_modules/milvus_index_construction.py:162

↓ 10 callersMethod__init__

( self, dim, num_heads=8, qkv_bias=True, scaled_co

code/C3/visual_bge/visual_bge/eva_clip/transformer.py:151

↓ 10 callersMethodclose

关闭资源连接

code/C9/rag_modules/graph_rag_retrieval.py:697

↓ 7 callersMethod__init__

( self, in_features, hidden_features=None, out_features=None, act_

code/C3/visual_bge/visual_bge/eva_clip/eva_vit_model.py:48

↓ 7 callersFunction_env

(name: str, default: str | None = None)

Extra-chapter/PowerRAG-SDK-Text-QA/code/main.py:13

↓ 6 callersMethodhybrid_search

混合检索：三路召回（图键值双层 + 向量 + BM25）→ RRF 融合

code/C9/rag_modules/hybrid_retrieval.py:756

↓ 6 callersMethodload_collection

加载集合到内存 Returns: 是否加载成功

code/C9/rag_modules/milvus_index_construction.py:474

↓ 6 callersMethodquery

执行Text2SQL查询

code/C4/text2sql/text2sql_agent.py:38

↓ 6 callersMethodsearch

搜索相关的知识库信息

code/C4/03_text2sql_demo_v2.py:223

↓ 6 callersMethodsearch

执行搜索

code/C3/work_hybrid_multimodal_search.py:257

↓ 5 callersFunction_ntuple

(n)

code/C3/visual_bge/visual_bge/eva_clip/utils.py:277

↓ 5 callersMethodencode_text_hybrid

使用BGE-M3编码文本，返回稀疏和密集向量

code/C3/work_hybrid_multimodal_search.py:75

↓ 5 callersFunctionimage_transform

( image_size: int, is_train: bool, mean: Optional[Tuple[float, ...]] = None, s

code/C3/visual_bge/visual_bge/eva_clip/transform.py:60

↓ 5 callersFunctionload_state_dict

(checkpoint_path: str, map_location: str='cpu', model_key: str='model|module|state_dict', is_openai: bool=Fals

code/C3/visual_bge/visual_bge/eva_clip/factory.py:80

↓ 4 callersMethod_build_context

构建上下文字符串 Args: docs: 文档列表 max_length: 最大长度 Returns: 格式化的上下文字符串

code/C8/rag_modules/generation_integration.py:364

↓ 4 callersFunction_clean_tag

(tag: str)

code/C3/visual_bge/visual_bge/eva_clip/pretrained.py:191

↓ 4 callersMethod_make_layer

(self, planes, blocks, stride=1)

code/C3/visual_bge/visual_bge/eva_clip/modified_resnet.py:132

↓ 4 callersFunction_safe_get

(obj: Any, attr: str, default: Any = None)

Extra-chapter/PowerRAG-SDK-Text-QA/code/main.py:33

↓ 4 callersFunctioncalc_response_score

(results, metric)

code/C6/01_llamaindex_evaluation_example.py:90

↓ 4 callersMethodencode_mm

(self, images:torch.Tensor, texts)

code/C3/visual_bge/visual_bge/modeling.py:224

↓ 4 callersMethodencode_multimodal

编码多模态内容（图像+文本）

code/C3/work_hybrid_multimodal_search.py:69

↓ 4 callersMethodencode_text

编码文本为密集向量

code/C4/03_text2sql_demo_v2.py:21

↓ 4 callersFunctionget_pretrained_cfg

(model: str, tag: str)

code/C3/visual_bge/visual_bge/eva_clip/pretrained.py:227

↓ 4 callersMethodload_data

加载知识库数据

code/C4/03_text2sql_demo_v2.py:91

↓ 3 callersMethod_insert_data

插入数据

code/C4/text2sql/knowledge_base.py:131

↓ 3 callersMethodbatch_process_recipes

批量处理菜谱目录 - 支持断点续传和分批保存

code/C9/agent(代码系ai生成)/recipe_ai_agent.py:909

↓ 3 callersMethodcreate_collection

创建Collection

code/C3/work_hybrid_multimodal_search.py:159

↓ 3 callersFunctioncreate_model

( model_name: str, pretrained: Optional[str] = None, precision: str = 'fp32',

code/C3/visual_bge/visual_bge/eva_clip/factory.py:211

↓ 3 callersFunctiondownload_pretrained

( cfg: Dict, force_hf_hub: bool = False, cache_dir: Union[str, None] = None, )

code/C3/visual_bge/visual_bge/eva_clip/pretrained.py:304

↓ 3 callersMethodencode_image

(self, images)

code/C3/visual_bge/visual_bge/modeling.py:308

↓ 3 callersMethodencode_query

编码查询（支持图像+文本或仅文本）

code/C3/work_multimodal_dragon_search.py:75

↓ 3 callersMethodexport_to_neo4j_csv

导出为Neo4j导入格式的CSV - 支持合并批次数据

code/C9/agent(代码系ai生成)/recipe_ai_agent.py:1133

↓ 3 callersFunctionfreeze_batch_norm_2d

Converts all `BatchNorm2d` and `SyncBatchNorm` layers of provided module into `FrozenBatchNorm2d`. If `module` is itself an instance of eithe

code/C3/visual_bge/visual_bge/eva_clip/utils.py:237

↓ 3 callersMethodgenerate_adaptive_answer

智能统一答案生成自动适应不同类型的查询，无需预先分类

code/C9/rag_modules/generation_integration.py:38

↓ 3 callersMethodgenerate_concept_id

生成新的概念ID

code/C9/agent(代码系ai生成)/recipe_ai_agent.py:591

↓ 3 callersMethodhybrid_search

混合检索 - 结合向量检索和BM25检索，使用RRF重排 Args: query: 查询文本 top_k: 返回结果数量 Returns: 检索到的文档列表

code/C8/rag_modules/retrieval_optimization.py:50

↓ 3 callersMethodlock

lock modules Args: unlocked_groups (int): leave last n layer groups unlocked (default: 0)

code/C3/visual_bge/visual_bge/eva_clip/timm_model.py:80

↓ 3 callersMethodnormalize_amount

标准化用量表达 Args: amount: 原始用量 unit: 单位 Returns: Tuple[标准化后的用量, 估算数值]

code/C9/agent(代码系ai生成)/amount_normalizer.py:90

↓ 3 callersFunctionrotate_half

(x)

code/C3/visual_bge/visual_bge/eva_clip/rope.py:23

↓ 3 callersMethodsave_batch_data

保存当前批次数据

code/C9/agent(代码系ai生成)/recipe_ai_agent.py:525

↓ 3 callersMethodsave_progress

保存处理进度

code/C9/agent(代码系ai生成)/recipe_ai_agent.py:491

↓ 3 callersMethodset_grad_checkpointing

(self, enable=True)

code/C3/visual_bge/visual_bge/eva_clip/model.py:238

↓ 3 callersMethodsimilarity_search

相似度搜索 Args: query: 查询文本 k: 返回结果数量 Returns: 相似文档列表

code/C8/rag_modules/index_construction.py:121

↓ 2 callersMethod__init__

(self, inplanes, planes, stride=1)

code/C3/visual_bge/visual_bge/eva_clip/modified_resnet.py:13

↓ 2 callersMethod_build_context

构建上下文信息

code/C4/text2sql/sql_generator.py:83

↓ 2 callersMethod_build_parent_doc_map

{str(node_id): 整篇父菜谱 Document}，由分块前的 data_module.documents 懒建一次。

code/C9/rag_modules/hybrid_retrieval.py:716

↓ 2 callersFunction_build_text_tower

( embed_dim: int, text_cfg: CLIPTextCfg, quick_gelu: bool = False, cast_dtype:

code/C3/visual_bge/visual_bge/eva_clip/model.py:173

↓ 2 callersFunction_build_vision_tower

( embed_dim: int, vision_cfg: CLIPVisionCfg, quick_gelu: bool = False, cast_dt

code/C3/visual_bge/visual_bge/eva_clip/model.py:92

↓ 2 callersMethod_categorize_synonyms_by_language

按语言分类同义词

code/C9/agent(代码系ai生成)/recipe_ai_agent.py:798

↓ 2 callersMethod_dist_gather_tensor

(self, t: Optional[torch.Tensor])

code/C3/visual_bge/visual_bge/modeling.py:364

↓ 2 callersMethod_fallback_parse

备用解析方法（基于规则）

code/C9/agent(代码系ai生成)/recipe_ai_agent.py:277

↓ 2 callersMethod_fallback_subgraph_extraction

降级子图提取

code/C9/rag_modules/graph_rag_retrieval.py:687

↓ 2 callersMethod_format_synonyms_for_neo4j

格式化同义词用于Neo4j导出

code/C9/agent(代码系ai生成)/recipe_ai_agent.py:840

↓ 2 callersMethod_get_node_neighbors

获取节点的邻居信息

code/C9/rag_modules/hybrid_retrieval.py:567

↓ 2 callersMethod_initialize_retrievers

初始化检索器

code/C9/main.py:185

↓ 2 callersMethod_paths_to_documents

将图路径转换为Document对象

code/C9/rag_modules/graph_rag_retrieval.py:586

↓ 2 callersFunction_require

(value: str | None, hint: str)

Extra-chapter/PowerRAG-SDK-Text-QA/code/main.py:20

↓ 2 callersFunction_rescan_model_configs

()

code/C3/visual_bge/visual_bge/eva_clip/factory.py:29

↓ 2 callersMethod_show_knowledge_base_stats

显示知识库统计信息

code/C9/main.py:202

↓ 2 callersMethod_tokenize_chinese

jieba 精确分词 + 停用词 / 空白 / 单字符过滤

code/C9/rag_modules/hybrid_retrieval.py:97

↓ 2 callersMethodanalyze_query

深度分析查询特征，决定最佳检索策略

code/C9/rag_modules/intelligent_query_router.py:65

↓ 2 callersMethodask_question

回答用户问题 Args: question: 用户问题 stream: 是否使用流式输出 Returns: 生成的回答或生成器

code/C8/main.py:129

↓ 2 callersFunctionbasic_clean

(text)

code/C3/visual_bge/visual_bge/eva_clip/tokenizer.py:60

↓ 2 callersFunctionbroadcat

(tensors, dim = -1)

code/C3/visual_bge/visual_bge/eva_clip/rope.py:7

↓ 2 callersMethodbuild_knowledge_base

构建知识库（如果需要）

code/C9/main.py:124

↓ 2 callersFunctionbuild_model_from_openai_state_dict

( state_dict: dict, quick_gelu=True, cast_dtype=torch.float16, )

code/C3/visual_bge/visual_bge/eva_clip/model.py:379

↓ 2 callersMethodbuild_recipe_documents

构建菜谱文档，集成相关的食材和步骤信息 Returns: 结构化的菜谱文档列表

code/C9/rag_modules/graph_data_preparation.py:179

↓ 2 callersFunctionbytes_to_unicode

Returns list of utf-8 byte and a corresponding list of unicode strings. The reversible bpe codes work on unicode strings. This means you

code/C3/visual_bge/visual_bge/eva_clip/tokenizer.py:26

↓ 2 callersMethodchunk_documents

Markdown结构感知分块 Returns: 分块后的文档列表

code/C8/rag_modules/data_preparation.py:139

↓ 2 callersMethodchunk_documents

对文档进行分块处理 Args: chunk_size: 分块大小 chunk_overlap: 重叠大小 Returns: 分块后的文

code/C9/rag_modules/graph_data_preparation.py:314

↓ 2 callersMethodcompute_similarity

(self, q_reps, p_reps)

code/C3/visual_bge/visual_bge/modeling.py:296

↓ 2 callersMethodcreate_collection

创建Milvus集合

code/C4/03_text2sql_demo_v2.py:65

↓ 2 callersMethoddim

返回维度信息，兼容原BGE-M3接口

code/C4/01_hybrid_search_v2.py:52

↓ 2 callersFunctiondownload_pretrained_from_hf

( model_id: str, filename: str = 'open_clip_pytorch_model.bin', revision=None,

code/C3/visual_bge/visual_bge/eva_clip/pretrained.py:293

↓ 2 callersFunctiondownload_pretrained_from_url

( url: str, cache_dir: Union[str, None] = None, )

code/C3/visual_bge/visual_bge/eva_clip/pretrained.py:239

↓ 2 callersMethodencode_image

(self, image_path: str)

code/C3/04_multi_milvus.py:30

↓ 2 callersMethodencode_multimodal

编码多模态内容（图像+文本）

code/C3/work_multimodal_dragon_search.py:88

↓ 2 callersMethodencode_text

encode text only

code/C3/visual_bge/visual_bge/modeling.py:182

↓ 2 callersMethodexport_to_csv

导出为CSV格式

code/C9/agent(代码系ai生成)/recipe_ai_agent.py:1043

↓ 2 callersMethodextract_recipe_info

使用AI提取菜谱信息

code/C9/agent(代码系ai生成)/recipe_ai_agent.py:134

↓ 2 callersMethodforward_features

(self, x, return_all_features=False)

code/C3/visual_bge/visual_bge/eva_clip/eva_vit_model.py:489

↓ 2 callersMethodgenerate_basic_answer

生成基础回答 Args: query: 用户查询 context_docs: 上下文文档列表 Returns: 生成的回答

code/C8/rag_modules/generation_integration.py:52

↓ 2 callersFunctionget_cast_dtype

(precision: str)

code/C3/visual_bge/visual_bge/eva_clip/model.py:83

↓ 2 callersMethodget_cast_dtype

(self)

code/C3/visual_bge/visual_bge/eva_clip/transformer.py:508

↓ 2 callersMethodget_entities_by_key

根据索引键获取实体

code/C9/rag_modules/graph_indexing.py:363

↓ 2 callersMethodget_extended_attention_mask

Makes broadcastable attention and causal masks so that future and masked tokens are ignored. Arguments: attention_mask (

code/C3/visual_bge/visual_bge/modeling.py:132

↓ 2 callersFunctionget_model_config

(model_name)

code/C3/visual_bge/visual_bge/eva_clip/factory.py:66

↓ 2 callersFunctionget_pairs

Return set of symbol pairs in a word. Word is represented as tuple of symbols (symbols being variable-length strings).

code/C3/visual_bge/visual_bge/eva_clip/tokenizer.py:48

↓ 2 callersMethodget_parent_documents

根据子块获取对应的父文档（智能去重） Args: child_chunks: 检索到的子块列表 Returns: 对应的父文档列表（去重，按相关性排序）

code/C8/rag_modules/data_preparation.py:314

↓ 2 callersFunctionget_pretrained_tag

(pretrained_model)

code/C3/visual_bge/visual_bge/eva_clip/factory.py:152

↓ 2 callersFunctionget_pretrained_url

(model: str, tag: str)

code/C3/visual_bge/visual_bge/eva_clip/pretrained.py:234

↓ 2 callersMethodget_route_statistics

获取路由统计信息

code/C9/rag_modules/intelligent_query_router.py:271

↓ 2 callersMethodget_statistics

获取键值对存储统计信息

code/C9/rag_modules/graph_indexing.py:375

↓ 2 callersMethodget_text_content

获取图像的文本描述内容

code/C3/work_multimodal_dragon_search.py:56

↓ 2 callersMethodget_text_content

获取图像的文本描述内容

code/C3/work_hybrid_multimodal_search.py:45

↓ 2 callersMethodgraph_rag_search

图RAG主搜索接口：整合所有图RAG能力

code/C9/rag_modules/graph_rag_retrieval.py:482

next →1–100 of 495, ranked by callers