Release v0.2.14

• LangChain integration with OpenAI-compatible API server (doc)
• Enhanced OpenAI-compatible API server (doc)
• More model support (https://chat.lmsys.org/?arena)
  • Guanaco, WizardLM, GPT4ALL, and more.
• Bug fixes

Release v0.2.9

• Added OpenAI-compatible API server (https://github.com/lm-sys/FastChat/blob/main/docs/openai_api.md). Now you can use FastChat as a local drop-in replacement for OpenAI APIs.
• Supported Bard and Claude. Now Google's Bard and Anthropic's Claude are available at arena.lmsys.org
• Supported various open-source models (e.g., MPT-Chat-7B, Phoenix-inst-chat-7b)
• Other bug fixes

Release v0.2.5

• Released FastChat-T5 compatible with commercial usage. Checkout weights.
• Announced ChatBot Arena for comparing open LLMs. Checkout demo.
• Increased the model support coverage. The following models are tested with FastChat CLI and GUI.
  • Vicuna, Alpaca, LLaMA, Koala
  • lmsys/fastchat-t5-3b-v1.0
  • BlinkDL/RWKV-4-Raven
  • databricks/dolly-v2-12b
  • OpenAssistant/oasst-sft-1-pythia-12b
  • project-baize/baize-lora-7B
  • StabilityAI/stablelm-tuned-alpha-7b
  • THUDM/chatglm-6b

Release v0.2.3

• Added a Chatbot Arena where you can compare open large models side-by-side (https://arena.lmsys.org/)
  • Now we support Vicuna, Koala, OpenAssistant, Dolly, ChatGLM, StableLM, Alpaca, LLaMA
• Fix bugs