tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Decouple generate_during_eval from wandb to support other visualizers (#2849) [skip ci]

Browse Source 
* Add generate_during_eval for mlflow for dpo

* Decouple generate_during_eval from wandb
This commit is contained in:
Dhruv Mullick
2025-07-02 06:04:40 -06:00
committed by GitHub
parent 327b4e48e9
commit d1224db8f4
2 changed files with 2 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/axolotl/core/trainers/dpo/__init__.py
View File

@@ -28,7 +28,7 @@ class DPOStrategy:
training_args_kwargs["max_completion_length"] = None training_args_kwargs["max_completion_length"] = None
training_args_kwargs["max_length"] = cfg.sequence_len training_args_kwargs["max_length"] = cfg.sequence_len
training_args_kwargs["max_prompt_length"] = cfg.sequence_len training_args_kwargs["max_prompt_length"] = cfg.sequence_len
training_args_kwargs["generate_during_eval"] = cfg.use_wandb training_args_kwargs["generate_during_eval"] = cfg.dpo_generate_during_eval
if cfg.dpo_use_weighting is not None: if cfg.dpo_use_weighting is not None:
training_args_kwargs["use_weighting"] = cfg.dpo_use_weighting training_args_kwargs["use_weighting"] = cfg.dpo_use_weighting
if cfg.dpo_padding_free is not None: if cfg.dpo_padding_free is not None:

1
src/axolotl/utils/schemas/config.py
View File

@@ -146,6 +146,7 @@ class AxolotlInputConfig(
dpo_label_smoothing: float | None = None dpo_label_smoothing: float | None = None
dpo_norm_loss: bool | None = None dpo_norm_loss: bool | None = None
dpo_padding_free: bool | None = None dpo_padding_free: bool | None = None
dpo_generate_during_eval: bool | None = None
datasets: ( datasets: (
Annotated[ Annotated[