Allow FP16 math in flash attention #24953
Return back to fp16 fa
7e1d816b
Make the min_value precision dependent
63f422b2
Update onnxruntime/contrib_ops/webgpu/bert/flash_attention.cc
18cd745a
qjia7
approved these changes
on 2025-06-05
fs-eire
approved these changes
on 2025-06-05
guschmue
approved these changes
on 2025-06-05
sushraja-msft
deleted the user/sushraja/fp16_fa branch 363 days ago
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub