align xpu's autocast behavior w/ cuda by using device agnostic torch APIs
@ArthurZucker, pls help review, thx very much.
cc @IlyasMoutawwakil
ci failure seems not brought by my PR
@ArthurZucker @IlyasMoutawwakil could you help review and comment? Thx very much
ci failure maybe because of the instable ci env
CI seems clear now! cc @IlyasMoutawwakil
@IlyasMoutawwakil , could you help review? Thx very much.
@Rocketknight1 , do you know who else need review this PR after Ilyas approved? Thx.
run-slow: qwen2_5_omni, gemma, phimoe, qwen2_moe, gpt2, distilbert
This comment contains run-slow, running the specified jobs:
models: ['models/distilbert', 'models/gemma', 'models/gpt2', 'models/phimoe', 'models/qwen2_5_omni', 'models/qwen2_moe'] quantizations: [] ...
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.