Refactor ascend kernels#17

Closed

yao-fengchen wants to merge 12 commits intorefactor-ascendfrom

refactor-ascend-kernels

Commits on Aug 22, 2024

support ascend using infer_ext

CyCle1024
authored and
yao-fengchen
committed
fix(ascend): make infer_ext using TND format q,k,v in paged_token_attention

CyCle1024
authored and
yao-fengchen
committed
support ascend using infer_ext
yao-fengchen
committed
feat: support ascend moe_gating_topk_softmax
yao-fengchen
committed
feat: change infer_ext ops function param order (#2)

CyCle1024
authored and
yao-fengchen
committed
ascend: align attention mask to 32bytes (#7)

JackWeiw
authored and
yao-fengchen
committed
fix attn args (#9)

jinminxi104
authored and
yao-fengchen
committed
fix: expand shape of attn_mask (#10)

JackWeiw
authored and
yao-fengchen
committed
feat: udpate infer_ext ops interface (#13)

CyCle1024
authored and
yao-fengchen
committed
rename infer_ext to dlinfer

CyCle1024
authored and
yao-fengchen
committed
format code
yao-fengchen
committed
Support internlm 2.5 (#14)

pdx1989
authored and
yao-fengchen
committed