[float8nocompile] Add alternate Triton kernels for FP8 conversion which use atomic_max-based algo instead of reduction-based algo #3353
Triggered via pull request
December 20, 2024 21:52
Status
Cancelled
Total duration
2m 5s
Artifacts
–
float8_test.yml
on: pull_request
Matrix: test
Waiting for pending jobs
Annotations
1 error
Run Float8 Tests
Canceling since a higher priority waiting request for 'float8_test-Run Float8 Tests-refs/pull/1455/merge' exists
|