Activity

1. RobinQu

   added this to First GA Release of instinct.cppon Apr 3, 2024
2. RobinQu

   converted this from a draft issue on Apr 3, 2024
3. RobinQu

   added this to the v0.1.2 milestone on Apr 3, 2024
4. RobinQu commented on Apr 5, 2024

   

   RobinQu

   on Apr 5, 2024

   OwnerAuthor

   More actions

   more complex rag pipeline may invole agent frameworks #18

   Adaptive RAG

   

   https://github.com/langchain-ai/langgraph/blob/70c1c996a4c9fe8df518bcd849b3c6453dd0d58b/examples/rag/langgraph_adaptive_rag.ipynb
5. RobinQu commented on May 20, 2024

   

   RobinQu

   on May 20, 2024

   Last edited by RobinQu

   OwnerAuthor

   More actions

   Rerank, Colbert, ...

   Related work

   Methods

   • Keyword based - BM25
   • Cross-encoders BCE reranker
   • Late-interaction - ColBERT, BGE-M3
     • 2004.12832v2.pdf
     • https://zhuanlan.zhihu.com/p/683483778

   Opensourced Projects

   • Official COLBERT implementation: https://github.com/bclavie/RAGatouille
   • BCE family: https://github.com/netease-youdao/BCEmbedding
   • BAAI BGE family: https://github.com/FlagOpen/FlagEmbedding
   • BCE & BGE in C++: https://github.com/li-plus/chatglm.cpp

   

   

   

   Conlcusion

   • Both BCE and BGE family can be regarded as SOTA.
   • For multilanguage use case, valillan RAGatouille lags behind.
   • As Late-interaction models are faster in inference. bge-m3 is prefered as first ranking methods in RAG pipeline.
6. RobinQu

   modified the milestones: v0.1.2, v0.1.3 on May 21, 2024
7. RobinQu

   moved this to In Progress in First GA Release of instinct.cppon May 21, 2024
8. RobinQu

   modified the milestones: v0.1.3, v0.1.4 on May 21, 2024
9. RobinQu commented on May 21, 2024

   

   RobinQu

   on May 21, 2024

   Last edited by RobinQu

   OwnerAuthor

   More actions

   Timeline

   • BGE M3 in GGUF format #21
     • experiemntes with BERT + BGE M3 model for better understanding of forward pass, 3 days , 1 day.
     • bge-m3 inference code. A standalone repo may be needed. 7 days, 3 days.
   • First version of file-search tool for assistant-api #20
     • file and filebatch controllers, services, mappers. 3 days.
     • duckdb vectordb operator, retriever factory, 2 days
     • DuckDB: search with filter, delete with filter, 1 day
     • summarization pipeline for files, 2 days
     • file object handler, 2 days
     • file batch background jobs to update its progress, 1 day
     • search tool 3 days.
     • (optional) parallel react agent executor, 2 days
     • integration test 3 days.
     • eval test HF QA datast. 3 days.
10. RobinQu commented on May 27, 2024

    

    RobinQu

    on May 27, 2024

    OwnerAuthor

    More actions

    OpenAI officials parameter for RAG: https://platform.openai.com/docs/assistants/tools/file-search/how-it-works

    By default, the file_search tool uses the following settings:
    Chunk size: 800 tokens
    Chunk overlap: 400 tokens
    Embedding model: text-embedding-3-large at 256 dimensions
    Maximum number of chunks added to context: 20 (could be fewer)

    Supported file formats: https://platform.openai.com/docs/assistants/tools/file-search/supported-files

    
11. RobinQu

    moved this from In Progress to Done in First GA Release of instinct.cppon Jun 14, 2024
12. RobinQu

    changed the title [-]Adanvanced RAG implementations[/-] [+]Modular RAG implementations: reranker, multi-retrievers[/+] on Jun 14, 2024
13. RobinQu

    closed this as completedon Jun 14, 2024