Release v0.1.5: Final release

Please follow LLaMA-Efficient-Tuning for further updates.

Release v0.1.4

• Advanced configurations in Web UI
• Fix device map #318
• Fix API (downgrade pydantic<2.0.0)
• Code simplification #321
• Add datasets (ShareGPT, RefGPT [1])

[1] https://github.com/sufengniu/RefGPT

Release v0.1.3

• Web UI: source_prefix, max_length, dev set
• Bug fix: replace the Literal type with Enum for pydantic [1] #294
• Bug fix: definition of the chat function in stream_chat.py #317
• Add Web demo #309

[1] pydantic/pydantic#5821, tiangolo/sqlmodel#67

Release v0.1.2

• Fix gradient accumulation in PPO Trainer #299
• Better Web UI
  • checkpoint manager
  • add locales
  • save predictions #306

Release v0.1.1

• Thank @codemayq for building a better Web UI.

Release v0.1.0

update readme