Skip to content

Actions: NVIDIA/TransformerEngine

Build

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
3,090 workflow run results
3,090 workflow run results

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

TP communication overlap: enable the overlap between GEMM chunk at Ho…
Build #6111: Pull request #1311 opened by erhoo82
November 4, 2024 17:28 1h 13m 5s erhoo82:tp_rs_bf16
November 4, 2024 17:28 1h 13m 5s
[TE/JAX] XLA FFI calls for three cast transpose functions
Build #6110: Pull request #1310 synchronize by pre-commit-ci bot
November 4, 2024 17:06 1h 12m 15s huanghua1994:xla-ffi-act-trans
November 4, 2024 17:06 1h 12m 15s
[TE/JAX] XLA FFI calls for layer norm and RMS norm
Build #6108: Pull request #1290 synchronize by phu0ngng
November 4, 2024 15:59 1h 12m 14s huanghua1994:xla-custom-call-ffi
November 4, 2024 15:59 1h 12m 14s
[PyTorch] Fix autocast deprecation warnings
Build #6107: Pull request #1277 synchronize by yaox12
November 4, 2024 09:48 1h 10m 16s yaox12:xiny/fix_autocast_warning
November 4, 2024 09:48 1h 10m 16s
[PyTorch] Userbuffers support in operation-based API
Build #6098: Pull request #1142 synchronize by timmoon10
November 1, 2024 21:17 1h 13m 44s timmoon10:ub-ops
November 1, 2024 21:17 1h 13m 44s
[JAX] Expose cp params to jax DPA api
Build #6096: Pull request #1292 synchronize by mgoldfarb-nvidia
November 1, 2024 20:40 1h 15m 44s kocchop:faysal/expose-cp-to-jax-dpa
November 1, 2024 20:40 1h 15m 44s
[JAX] Fix for Disable FusedAttn with FFI by default
Build #6095: Pull request #1304 synchronize by phu0ngng
November 1, 2024 19:43 1h 10m 51s phu0ngng:fused_attn_ffi
November 1, 2024 19:43 1h 10m 51s
[JAX] Fix for Disable FusedAttn with FFI by default
Build #6094: Pull request #1304 opened by phu0ngng
November 1, 2024 15:49 1h 13m 42s phu0ngng:fused_attn_ffi
November 1, 2024 15:49 1h 13m 42s
[PyTorch] Make FP8 MHA work with RoPE when CP is on
Build #6093: Pull request #1297 synchronize by yaox12
November 1, 2024 04:32 1h 9m 28s yaox12:xiny/fp8_mha_with_rope_cp
November 1, 2024 04:32 1h 9m 28s
[PyTorch] Make FP8 MHA work with RoPE when CP is on
Build #6092: Pull request #1297 synchronize by yaox12
November 1, 2024 04:30 1h 6m 42s yaox12:xiny/fp8_mha_with_rope_cp
November 1, 2024 04:30 1h 6m 42s
[PyTorch] Userbuffers support in operation-based API
Build #6090: Pull request #1142 synchronize by pre-commit-ci bot
October 31, 2024 23:05 1h 10m 16s timmoon10:ub-ops
October 31, 2024 23:05 1h 10m 16s
[PyTorch] Userbuffers support in operation-based API
Build #6089: Pull request #1142 synchronize by timmoon10
October 31, 2024 23:04 1h 15m 1s timmoon10:ub-ops
October 31, 2024 23:04 1h 15m 1s
[JAX] Expose cp params to jax DPA api
Build #6088: Pull request #1292 synchronize by mgoldfarb-nvidia
October 31, 2024 22:21 1h 14m 14s kocchop:faysal/expose-cp-to-jax-dpa
October 31, 2024 22:21 1h 14m 14s
[PyTorch] Add heuristics for intializing FP8 params
Build #6087: Pull request #1300 synchronize by timmoon10
October 31, 2024 21:54 1h 9m 33s timmoon10:fp8-heuristic
October 31, 2024 21:54 1h 9m 33s
Support using fp16 master weights and fp16/fp8 optimizer states in FusedAdam
Build #6086: Pull request #1078 synchronize by timmoon10
October 31, 2024 20:46 1h 13m 14s kunlunl:mx_fp16
October 31, 2024 20:46 1h 13m 14s