v0.8.0

What's new in 0.8.0 (2024-01-11)

These are the changes in inference v0.8.0.

New features

• FEAT: qwen 1.8b gptq by @codingl2k1 in #869
• FEAT: docker compose support by @Minamiyama in #868
• FEAT: Simple OAuth2 system by @ChengjieLi28 in #793
• FEAT: Chat vl web UI by @codingl2k1 in #882
• FEAT: Yi chat gptq by @codingl2k1 in #876

Enhancements

• ENH: Stream use xoscar generator by @codingl2k1 in #859
• ENH: UI supports registering custom gptq models by @ChengjieLi28 in #875
• ENH: make the size param of *_to_image more compatible by @liunux4odoo in #881
• BLD: Update package-lock.json by @aresnow1 in #886
• REF: Add model_hub property in EmbeddingModelSpec by @aresnow1 in #877

Bug fixes

• BUG: Fix image model b64_json output by @codingl2k1 in #874
• BUG: fix libcuda.so.1: cannot open shared object file by @superhuahua in #883
• BUG: Fix auto recover kwargs by @codingl2k1 in #885

Documentation

• DOC: docker image translation by @aresnow1 in #865
• DOC: register model with model_family by @ChengjieLi28 in #863
• DOC: Add OpenAI Client API doc by @codingl2k1 in #864
• DOC: add docker instructions by @onesuper in #878

New Contributors

• @superhuahua made their first contribution in #883

Full Changelog: v0.7.5...v0.8.0