Actions: vllm-project/vllm-ascend
Actions
9,223 workflow run results
9,223 workflow run results
FULL_DECODE_ONLY
mode for GQA/MHA models
ascend test
#9315:
Pull request #2128
synchronize
by
yiz-liu
FULL_DECODE_ONLY
mode for GQA/MHA models
ascend test
#9310:
Pull request #2128
synchronize
by
yiz-liu
AscendAttentionMetadataBuilder
for better extensibility and make the builder class of torchair extend from it
ascend test
#9309:
Pull request #2375
synchronize
by
shen-shanshan
torchair_attention
to torchair
dir
ascend test
#9305:
Pull request #2017
synchronize
by
shen-shanshan