onnxruntime
[JS/WebGU] Support fp16 in Attention by performing the computation in fp32.
#20486

Merged

[JS/WebGU] Support fp16 in Attention by performing the computation in fp32. #20486

satyajandhyala merged 2 commits into main from sajandhy/webgpu_attention_fp16_support_fit

Take case fp16 support

4191156a

satyajandhyala changed the title ~~[JS/WebGU] Take case fp16 support~~ [JS/WebGU] Take case fp16 support in Attention by performing the computation in fp32. 1 year ago

Clean-up the code related to RelativePositionBias.

7c61867c

guschmue added ep:WebGPU

satyajandhyala changed the title ~~[JS/WebGU] Take case fp16 support in Attention by performing the computation in fp32.~~ [JS/WebGU] Support fp16 in Attention by performing the computation in fp32. 1 year ago

satyajandhyala requested a review from

guschmue 1 year ago

satyajandhyala requested a review from

fs-eire 1 year ago

satyajandhyala added release:1.18.0

satyajandhyala marked this pull request as ready for review 1 year ago

guschmue approved these changes on 2024-04-27

satyajandhyala merged 736cbb39 into main 1 year ago

satyajandhyala deleted the sajandhy/webgpu_attention_fp16_support_fit branch 1 year ago

sophies927 added triage:approved

yihonglyu added cherry-picked

yihonglyu added rel-merged

Reviewers

guschmue

fs-eire

Assignees

No one assigned

Labels

triage:approved ep:WebGPU release:1.18.0 cherry-picked rel-merged

Milestone

No milestone

onnxruntime [JS/WebGU] Support fp16 in Attention by performing the computation in fp32. #20486 Merged

[JS/WebGU] Support fp16 in Attention by performing the computation in fp32. #20486

onnxruntime
[JS/WebGU] Support fp16 in Attention by performing the computation in fp32.
#20486

Merged