Skip to content
PR Checks

[CUDA] Support FP8 (E4M3) KV Cache for Group Query Attention #12841

Sign in to view logs

[CUDA] Support FP8 (E4M3) KV Cache for Group Query Attention

[CUDA] Support FP8 (E4M3) KV Cache for Group Query Attention #12841

Triggered via pull request February 14, 2026 04:12
@tianleiwutianleiwu
synchronize #27321
tlwu/20260211/gqa_fp8_kv_cache
Status Success
Total duration 2m 23s
Artifacts –

pr_checks.yml

on: pull_request
Suggest fixes
2m 8s
Suggest fixes
Fit to window
Zoom out
Zoom in

Annotations

4 warnings
Suggest fixes
The `set-output` command is deprecated and will be disabled soon. Please upgrade to using Environment Files. For more information see: https://github.blog/changelog/2022-10-11-github-actions-deprecating-save-state-and-set-output-commands/
Suggest fixes
The `set-output` command is deprecated and will be disabled soon. Please upgrade to using Environment Files. For more information see: https://github.blog/changelog/2022-10-11-github-actions-deprecating-save-state-and-set-output-commands/
Suggest fixes
The `set-output` command is deprecated and will be disabled soon. Please upgrade to using Environment Files. For more information see: https://github.blog/changelog/2022-10-11-github-actions-deprecating-save-state-and-set-output-commands/
Suggest fixes
The `set-output` command is deprecated and will be disabled soon. Please upgrade to using Environment Files. For more information see: https://github.blog/changelog/2022-10-11-github-actions-deprecating-save-state-and-set-output-commands/