FSDP1 -> FSDP2 (#2760)

* FSDP2 args migration implementation

This commit implements the migration to FSDP2 arguments including:
- FSDP2 support with LoRA training
- DPO integration with FSDP2
- Model loading fixes and refactoring
- CPU offloading and PEFT handling
- Test updates and CI improvements
- Bug fixes for dtype errors and various edge cases
This commit is contained in:
salman
2025-07-12 15:18:01 +01:00
committed by GitHub
parent eb662557a7
commit d6e4a611e5
27 changed files with 1357 additions and 436 deletions

View File

@@ -40,13 +40,13 @@ use_cpu: false
Configure your model to use FSDP in the Axolotl yaml. For example:
```yaml
fsdp:
- full_shard
- auto_wrap
fsdp_version: 2
fsdp_config:
fsdp_offload_params: true
fsdp_state_dict_type: FULL_STATE_DICT
fsdp_transformer_layer_cls_to_wrap: LlamaDecoderLayer
offload_params: true
state_dict_type: FULL_STATE_DICT
auto_wrap_policy: TRANSFORMER_BASED_WRAP
transformer_layer_cls_to_wrap: LlamaDecoderLayer
reshard_after_forward: true
```
All you have to do now is launch using accelerate as you would usually do on each machine and voila, the processes will start once you have launched accelerate on every machine.