Skip to content

Actions: NVIDIA/TransformerEngine

Actions

Build

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
build.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
[JAX] Limit max logit attr to forward Build #20440: Pull request #3428 synchronize by jberchtold-nvidia
6h 0m 8s KshitijLakhani:klakhani/fix/jax-max-logit-ffi-attr
KshitijLakhani:klakhani/fix/jax-max-logit-ffi-attr
6h 0m 8s
Thread tanh logit softcapping through FlashAttention (FA2, opt-in FA3) Build #20439: Pull request #3391 synchronize by nvegesna-netizen
6h 5m 10s nvegesna-netizen:nvegesna/gemma2-softcap-core
nvegesna-netizen:nvegesna/gemma2-softcap-core
6h 5m 10s
Thread tanh logit softcapping through FlashAttention (FA2, opt-in FA3) Build #20438: Pull request #3391 synchronize by nvegesna-netizen
22m 27s nvegesna-netizen:nvegesna/gemma2-softcap-core
nvegesna-netizen:nvegesna/gemma2-softcap-core
22m 27s
Thread tanh logit softcapping through FlashAttention (FA2, opt-in FA3) Build #20437: Pull request #3391 synchronize by nvegesna-netizen
2m 59s nvegesna-netizen:nvegesna/gemma2-softcap-core
nvegesna-netizen:nvegesna/gemma2-softcap-core
2m 59s
Thread tanh logit softcapping through FlashAttention (FA2, opt-in FA3) Build #20436: Pull request #3391 synchronize by nvegesna-netizen
12m 53s nvegesna-netizen:nvegesna/gemma2-softcap-core
nvegesna-netizen:nvegesna/gemma2-softcap-core
12m 53s
Fix TE EP MOE tests aborting insted of skipping when < 4 GPUs are available Build #20435: Pull request #3431 opened by fheinecke
6h 0m 7s fheinecke:fred/fix-ep-moe-test-1
fheinecke:fred/fix-ep-moe-test-1
6h 0m 7s
Thread tanh logit softcapping through FlashAttention (FA2, opt-in FA3) Build #20434: Pull request #3391 synchronize by nvegesna-netizen
22m 5s nvegesna-netizen:nvegesna/gemma2-softcap-core
nvegesna-netizen:nvegesna/gemma2-softcap-core
22m 5s
Thread tanh logit softcapping through FlashAttention (FA2, opt-in FA3) Build #20433: Pull request #3391 synchronize by nvegesna-netizen
7m 1s nvegesna-netizen:nvegesna/gemma2-softcap-core
nvegesna-netizen:nvegesna/gemma2-softcap-core
7m 1s
[PyTorch] Share CUDA graph memory across dynamic CP variants Build #20432: Pull request #3353 synchronize by xiaoyao0115
1h 55m 47s xiaoyao0115:test/dcp-cg-gap-fix
xiaoyao0115:test/dcp-cg-gap-fix
1h 55m 47s
[Common] Improved performance of Group MXFP8 kernels Build #20431: Pull request #3337 synchronize by Oleg-Goncharov
6h 0m 10s Oleg-Goncharov:pr_mxfp8_group_kernels_optimization
Oleg-Goncharov:pr_mxfp8_group_kernels_optimization
6h 0m 10s
[Common/PyTorch] Grouped weighted-SwiGLU MXFP8 kernel Build #20430: Pull request #3315 synchronize by Oleg-Goncharov
6h 0m 9s cael-ling:feature/mxfp8-group-swiglu-recompute
cael-ling:feature/mxfp8-group-swiglu-recompute
6h 0m 9s
[PyTorch] Allow CP P2P transport group overrides Build #20429: Pull request #3420 synchronize by xiaoyao0115
2h 2m 39s xiaoyao0115:cp-p2p-transport
xiaoyao0115:cp-p2p-transport
2h 2m 39s
[Common] Split grouped activation build Build #20428: Pull request #3430 opened by harryzhou2000
6h 0m 19s harryzhou2000:hhanyu/grouped-activation-tu-split
harryzhou2000:hhanyu/grouped-activation-tu-split
6h 0m 19s
[PyTorch] Add no-load-balance THD all-gather CP Build #20427: Pull request #3221 synchronize by sudhakarsingh27
6h 0m 7s sudhakarsingh27:sudhakars/relaxed-load-balancing-all-gather
sudhakarsingh27:sudhakars/relaxed-load-balancing-all-gather
6h 0m 7s
[PyTorch] Reduce CUDA graph memory retention Build #20426: Pull request #3427 synchronize by buptzyb
6h 0m 8s buptzyb:codex/te-warmup-output-lifetime
buptzyb:codex/te-warmup-output-lifetime
6h 0m 8s
[PyTorch] Fused GDN attention Build #20425: Pull request #3351 synchronize by ksivaman
6h 0m 8s ksivaman:gdn_attention
ksivaman:gdn_attention
6h 0m 8s
Add CUDA wheels as build-time requirements for build isolation Build #20424: Pull request #3253 synchronize by ksivaman
10m 26s fheinecke:fred/add-build-time-requirements-1
fheinecke:fred/add-build-time-requirements-1
10m 26s
fix: guard backward_dw() bias-grad backfill against a legitimately-empty bgrad Build #20423: Pull request #3400 synchronize by ksivaman
6h 0m 6s nvegesna-netizen:nvegesna/fix-backward-dw-bgrad-none
nvegesna-netizen:nvegesna/fix-backward-dw-bgrad-none
6h 0m 6s
[JAX] Optimize MoE block Build #20422: Pull request #3354 synchronize by tdophung
6h 0m 14s jberchtold-nvidia:jberchtold/moeblock-debug
jberchtold-nvidia:jberchtold/moeblock-debug
6h 0m 14s
[Common] Experimental CuTeDSL MXFP8 backends in C++ via TVM-FFI Build #20421: Pull request #3137 synchronize by pre-commit-ci Bot
9m 32s kainzhong:cutedsl_mxfp8_common
kainzhong:cutedsl_mxfp8_common
9m 32s
[Common] Experimental CuTeDSL MXFP8 backends in C++ via TVM-FFI Build #20420: Pull request #3137 synchronize by kainzhong
2m 18s kainzhong:cutedsl_mxfp8_common
kainzhong:cutedsl_mxfp8_common
2m 18s
[Common] Default EP comm kernels to 32 SMs; keep shuffle/preprocess on all SMs Build #20419: Pull request #3422 synchronize by phu0ngng
6h 0m 9s phu0ngng:ep-sm-defaults
phu0ngng:ep-sm-defaults
6h 0m 9s
[JAX] Fix GEMM partitioning to reduce over nested contracting mesh axes Build #20418: Pull request #3429 opened by phu0ngng
6h 0m 9s phu0ngng:jax/fix_gemm_partitioning
phu0ngng:jax/fix_gemm_partitioning
6h 0m 9s
Enable FA4 for context-parallel attention Build #20417: Pull request #3149 synchronize by sudhakarsingh27
6h 0m 7s sudhakarsingh27:fa4-cp-exploration
sudhakarsingh27:fa4-cp-exploration
6h 0m 7s
Add a backwards linear function to be used with the fused mla q up-proj Build #20416: Pull request #3330 synchronize by pre-commit-ci Bot
6h 0m 10s chaseblock:lin_bwd
chaseblock:lin_bwd
6h 0m 10s