0.1.0 (2023-11-30)
Features
- Improved documentation using Fern
- Fastest ingestion through different ingestions modes ([#1309] (#1309))
- Add sources to completions APIs and UI
- Add simple Basic auth
- Add basic CORS
- Add "search in docs" to UI
- LLM and Embeddings model separate configuration
- Allow using a system prompt in the API to modify the LLM behaviour
- Expose configuration of the model execution such as max_new_tokens
- Multiple prompt styles support for different models
- Update to Gradio 4
- Document deletion API
- Sagemaker support
- Disable Gradio Analytics (#1165) (6583dc8)
- Drop loguru and use builtin
logging(#1133) (64c5ae2) - enable resume download for hf_hub_download (#1249) (4197ada)
- move torch and transformers to local group (#1172) (0d677e1)
- Qdrant support (#1228) (03d1ae6)
- Added wipe command to easy up vector database reset
Bug Fixes
- Docker and sagemaker setup (#1118) (895588b)
- fix pytorch version to avoid wheel bug (#1123) (24cfddd)
- Remove global state (#1216) (022bd71)
- sagemaker config and chat methods (#1142) (a517a58)
- typo in README.md (#1091) (ba23443)
- Windows 11 failing to auto-delete tmp file (#1260) (0d52002)
- Windows permission error on ingest service tmp files (#1280) (f1cbff0)