Add dora_ft example xpu support #2700
Conversation
Signed-off-by: Liu, Kaixuan <[email protected]>
Signed-off-by: Liu, Kaixuan <[email protected]>
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
|
|
||
| # Setup device | ||
| device = torch.device(device) | ||
| device = torch.accelerator.current_accelerator().type if hasattr(torch, "accelerator") else "cuda" |
There was a problem hiding this comment.
I think it would be better to allow the user to pass the device explicitly, i.e. revert this change. But the default could be "auto" , in which case the device is inferred like in this line. WDYT?
There was a problem hiding this comment.
Yes, I agree. Have updated the code.
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
|
@kaixuanliu Is there anything left to do? The PR is still marked as "draft". |
|
@BenjaminBossan , there is a bnb dependency here: https://github.com/huggingface/peft/pull/2700/files#diff-743cf3ad364387a1e75c4e38d779be602a91d82622846cb4ce0afdd3ed0eac3cR1013, I am not sure if it is OK to merge this PR first. |
Signed-off-by: Liu, Kaixuan <[email protected]>
Signed-off-by: Liu, Kaixuan <[email protected]>
Signed-off-by: Liu, Kaixuan <[email protected]>
| "top_p = 0.9\n", | ||
| "temperature = 0.7\n", | ||
| "user_question = \"What is the purpose of quantization in LLMs?\"\n", | ||
| "#TODO: XPU support depneds on this PR in bitsandbytes:https://github.com/bitsandbytes-foundation/bitsandbytes/pull/1692\n", |
There was a problem hiding this comment.
Let's wait for that PR to be merged first, then we can remove the comment and merge this PR.
|
This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread. |
|
not stale, still waiting for the bnb PR |
* support for synchronization ref-model added * support for synchronization ref-model added * tests for sync_ref_model added * Update tests/test_grpo_trainer.py Co-authored-by: Quentin Gallouédec <[email protected]> * split and fix test * style * doc * move after init to ensure accelerator exists * Update tests/test_grpo_trainer.py * style --------- Co-authored-by: Quentin Gallouédec <[email protected]> Co-authored-by: Quentin Gallouédec <[email protected]>
|
@BenjaminBossan The dependency PR is merged. Pls help review. Thx~ |
Signed-off-by: Liu, Kaixuan <[email protected]>
|
The Hub seems to be quite unreliable right now, I'll re-trigger CI another time. |
No description provided.