Skip to content

Actions: NVIDIA/TransformerEngine

Actions

Blossom-CI

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
blossom-ci.yml
2,500+ workflow runs
2,500+ workflow runs

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
Adaptive Compression Blossom-CI #11067: Issue comment #2807 (comment) created by ShawRong
1s
1s
Why NVFP4 Inference (50 PFLOPS) Outperforms Training (35 PFLOPS) on Rubin GPU？ Blossom-CI #11066: Issue comment #2565 (comment) created by FTi130
1s
1s
Adaptive Compression Blossom-CI #11065: Issue comment #2807 (comment) created by FTi130
1s
1s
Plumb FP8+THD Blossom-CI #11064: Issue comment #2994 (comment) created by bbuschkaemper
2s
2s
[Common] MXFP8: Add swizzled-scale fast path for cast-only quantization Blossom-CI #11063: Issue comment #3338 (comment) created by Oleg-Goncharov
1s
1s
[PyTorch] Enable NVFP4 row-scaled (per-token) backward for GroupedLinear Blossom-CI #11062: Issue comment #3324 (comment) created by timmoon10
2s
2s
Gate FA4 and stabilize attention test imports Blossom-CI #11061: Issue comment #3341 (comment) created by greptile-apps Bot
2s
2s
Add 2d quant for mxfp8 Blossom-CI #11060: Issue comment #2634 (comment) created by ptrendx
9s
9s
[Common][PyTorch] EP dispatch with unfused MXFP8 quantization Blossom-CI #11059: Issue comment #3270 (comment) created by phu0ngng
2s
2s
Plumb FP8+THD Blossom-CI #11058: Issue comment #2994 (comment) created by sudhakarsingh27
1s
1s
Plumb FP8+THD Blossom-CI #11057: Issue comment #2994 (comment) created by sudhakarsingh27
1s
1s
Optionally pad QK/V for better kernel selection Blossom-CI #11056: Issue comment #3339 (comment) created by greptile-apps Bot
1s
1s
Add per-sequence causal policy to packed THD attention Blossom-CI #11055: Issue comment #3274 (comment) created by desh2608
12s
12s
[Common] MXFP8: Add swizzled-scale fast path for cast-only quantization Blossom-CI #11054: Issue comment #3338 (comment) created by Oleg-Goncharov
2s
2s
NVRTC NVFP4 quantization kernels Blossom-CI #11053: Issue comment #3301 (comment) created by ptrendx
1s
1s
[Common] MXFP8: Add swizzled-scale fast path for cast-only quantization Blossom-CI #11052: Issue comment #3338 (comment) created by greptile-apps Bot
1s
1s
[Bug] use_single_group_runtime_offsets arg is not supported for ScaledSrelu Blossom-CI #11051: Issue comment #3328 (comment) created by ksivaman
1s
1s
[Common] Fix pointer arithmatic to generate correct LDS/STS instructions Blossom-CI #11050: Issue comment #3291 (comment) created by ptrendx
1s
1s
[Common] Fix pointer arithmatic to generate correct LDS/STS instructions Blossom-CI #11049: Issue comment #3291 (comment) created by kainzhong
10s
10s
[Bug] FusedAttention THD + learnable softmax backward IMA (cuDNN err 700) on Hopper Blossom-CI #11048: Issue comment #3249 (comment) created by KshitijLakhani
1s
1s
[common] Improved performance of Group MXFP8 kernels Blossom-CI #11047: Issue comment #3337 (comment) created by greptile-apps Bot
1s
1s
Refactor GroupedLinear quantization dispatch Blossom-CI #11046: Issue comment #3326 (comment) created by negvet
1s
1s
[PyTorch] Fine-grained recipe docs Blossom-CI #11045: Issue comment #3336 (comment) created by greptile-apps Bot
1s
1s
[Bug] FusedAttention THD + learnable softmax backward IMA (cuDNN err 700) on Hopper Blossom-CI #11044: Issue comment #3249 (comment) created by fy1214
1s
1s
Blossom-CI Blossom-CI #11043: created by fy1214
8s
8s