Skip to content

[QEff. Finetuning] Adding TP+DDP support in HF Trainer stack#907

Draft
smedhe wants to merge 3 commits intoquic:ft_experimental_v1from
smedhe:hf_trainer_tp_ddp
Draft

[QEff. Finetuning] Adding TP+DDP support in HF Trainer stack#907
smedhe wants to merge 3 commits intoquic:ft_experimental_v1from
smedhe:hf_trainer_tp_ddp

Conversation

@smedhe
Copy link
Copy Markdown
Contributor

@smedhe smedhe commented Apr 6, 2026

Adding TP+DDP support in the new hf trainer stack. It is still an experimental feature due to transformers and accelerate versions mismatch.
Further TO-DO:
load_in_4_bits flag is not supported in v5.1.0, commented out since it is not used
need to add steps for running tp+ddp in the config.md

smedhe added 2 commits April 6, 2026 09:14
Signed-off-by: Sharvari Medhe <smedhe@qti.qualcomm.com>
Signed-off-by: Sharvari Medhe <smedhe@qti.qualcomm.com>
@smedhe smedhe marked this pull request as draft April 6, 2026 09:20
Signed-off-by: Sharvari Medhe <smedhe@qti.qualcomm.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant