Summary of Changes
- Switch to multistage dockerfile which greatly reduced the size of the image
- Refactor image scripts to remove `launch_training` and call `sft_trainer` directly.
- Note that this affects the error codes returned from `sft_trainer` to user error code _1_ and internal error code _203_.
- In addition, this affects the logging as parameter parsing logging is moved into `sft_trainer` which is harder to view.
What's Changed
* Switch to multistage dockerfile by tharapalanivel in https://github.com/foundation-model-stack/fms-hf-tuning/pull/154
* refactor: remove launch_training and call sft_trainer directly by anhuong in https://github.com/foundation-model-stack/fms-hf-tuning/pull/164
* docs: consolidate configs, add kfto config by anhuong in https://github.com/foundation-model-stack/fms-hf-tuning/pull/170
* fix: bloom model can't run with flash-attn by anhuong in https://github.com/foundation-model-stack/fms-hf-tuning/pull/173
* Update README.md for Lora modules by Ssukriti in https://github.com/foundation-model-stack/fms-hf-tuning/pull/174
**Full Changelog**: https://github.com/foundation-model-stack/fms-hf-tuning/compare/v0.2.0...v0.3.0