PR #7975 fix gemma4 num attention head bugs

fix gemma4 num attention head bugs

ming.lee committed 56 days ago

revise logic of fallback for text config

ming.lee committed 55 days ago

Merge branch 'master' into master

mingxiang1006 committed 50 days ago

fix gemma4 num attention head bugs

ming.lee committed 50 days ago

revise logic of fallback for text config

ming.lee committed 50 days ago

fix(fp_quantizer): fix UB and negative shift warnings in fp_quantize_impl.cu (#7973)

Cursx committed 50 days ago

fix(op_builder): avoid duplicate/wrong -gencode flags (#7974)

Cursx committed 49 days ago

Rename dequantization template parameters (#7976)

Flamefire committed 49 days ago

Avoid CUDA reinit error in CI tests (#7977)

tohtana committed 49 days ago

Merge branch 'master' of https://github.com/mingxiang1006/DeepSpeed

ming.lee committed 49 days ago

remove fix here comment

ming.lee committed 48 days ago

update text fallback

ming.lee committed 42 days ago

Merge branch 'master' into master

delock committed 42 days ago

DCO remediation sign-off

ming.lee committed 42 days ago

DeepSpeed fix gemma4 num attention head bugs #7975 Closed