summaryrefslogtreecommitdiffstats
path: root/src/rag
AgeCommit message (Collapse)AuthorLines
2024-10-22refactor: several improvements (#943)sigoden-2/+2
- rename agents_config_dir to agents_data_dir - do not save empty role name while saving the session - rename delete keys (roles/sessions/rags/agents => role/session/rag/agent-data) - improve some printing messages
2024-10-20refactor: RAG bm25 (#939)sigoden-2/+2
2024-10-17refactor: improve RAG search (#931)sigoden-253/+74
2024-09-24feat: add AICHAT_EMBEDDINGS_RETRY_LIMIT (#882)sigoden-26/+9
2024-09-24feat: add retry logic to embedding/rerank api calls (#879)sigoden-6/+41
2024-09-24refactor: embeddings/rerank fn accept ref data (#878)sigoden-2/+2
2024-09-22feat: add batch_size to RAG yaml (#876)sigoden-4/+16
2024-09-14feat: add `.source rag` repl command (#871)sigoden-5/+31
2024-09-09refactor: abandon config `rag_min_score_rerank` (#852)sigoden-17/+23
2024-09-08feat: save rag in YAML instead of bin (#848)sigoden-10/+82
2024-09-08feat: support RAG-scoped rag_top_k and rag_reranker_model (#847)sigoden-17/+69
2024-09-08feat: tolerate failure to load some rag files (#846)sigoden-33/+45
2024-09-03feat: use dynamic batch size for embedding (#826)sigoden-3/+13
2024-08-12feat: support builtin website crawling (recursive_url) (#786)sigoden-12/+15
2024-07-29fix: unable to rebuild agent rag (#763)sigoden-1/+1
2024-07-28feat: ask for confirmation when some rag documents fail to load (#760)sigoden-19/+46
2024-07-11refactor: rag add documents (#700)sigoden-2/+12
2024-07-10refactor: fetch image as data url (#694)sigoden-2/+2
2024-07-10feat: `--file/.file` can load dirs (#693)sigoden-119/+4
2024-07-03refactor: update config.example.yaml and readme, fix typos (#682)sigoden-4/+4
2024-07-02refactor: improve RAG (#676)sigoden-2/+8
2024-07-02refactor: improve RAG (#675)sigoden-44/+16
2024-07-01feat: support `.rebuild rag` repl command(#672)sigoden-256/+249
2024-06-28feat: `.file`/`--file` support URLs (#665)sigoden-115/+11
2024-06-27refactor: smart document splitter (#662)sigoden-36/+56
2024-06-27feat: implement native rag url loader (#660)sigoden-9/+57
2024-06-27refactor: agent rag use `documents` field other than `embeddings` dir (#658)sigoden-4/+8
2024-06-26feat: rag load websites (#655)sigoden-110/+285
2024-06-26feat: custom rag document loaders (#650)sigoden-45/+45
2024-06-25refactor: rag chunk contains path metadata (#647)sigoden-17/+9
2024-06-25feat: support rag-dedicated clients (jina and voyageai) (#645)sigoden-0/+1
2024-06-23refactor: embedding model add price and dimension (#636)sigoden-3/+6
2024-06-21refactor: rename model.max_concurrent_chunks to model.max_batch_size (#626)sigoden-5/+5
2024-06-21refactor: improve rag (#622)sigoden-40/+66
2024-06-21feat: support rerank (#620)sigoden-35/+83
2024-06-20feat: rag hybrid search (#618)sigoden-13/+260
2024-06-19refactor: rag add rag_minimum_score config (#617)sigoden-15/+12
2024-06-14refactor: add/modify rag-related config (#599)sigoden-28/+82
2024-06-13refactor: change spinner text for rag searching (#596)sigoden-1/+1
2024-06-13feat: add rag and bot related cli options (#595)sigoden-0/+3
2024-06-12feat: rag splitter supports languages (#593)sigoden-80/+287
2024-06-12refactor: load pdf/docx files (#592)sigoden-33/+40
2024-06-12refactor: rag default_chunk_size (#588)sigoden-12/+25
2024-06-11feat: support bot (#579)sigoden-8/+11
* feat: support bots * refactor with RoleLike * improve exiting session * make bot works with rag * refactor repl assert state * add bot banner * repl complete bots according bots.txt * fix on windows * remove threadpool executing function callings * adjust repl left_prompt * move bot config to global config.yaml * `.bot` throw err if funciton callings is not configured
2024-06-06refactor: config refinement (#570)sigoden-3/+2
2024-06-05feat: support RAG (#560)sigoden-0/+1135
* feat: support RAG * support more embeddings models and implement concurrent embedding api * show the progress of addings paths * ignore embedding context when saving message * embedding model max_chunk_size => default_chunk_size * support pdf and pandoc formats (docx, epub, ipynb)