Skip to content

Actions: meta-pytorch/autoparallel

Actions

Test CUDA

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
test_cuda.yml
1,711 workflow runs
1,711 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
Add doc_packed_attn: AutoParallel-shardable document-packed varlen attention Test CUDA #1715: Pull request #508 opened by fmassa
28m 57s fmassa/packed_doc_attention
fmassa/packed_doc_attention
28m 57s
Emit autoparallel solution as a TORCH_TRACE artifact (#507) Test CUDA #1714: Commit 14682dc pushed by fmassa
24m 40s main
main
24m 40s
Emit autoparallel solution as a TORCH_TRACE artifact Test CUDA #1713: Pull request #507 opened by fmassa
27m 6s fmassa/torch_trace_solution
fmassa/torch_trace_solution
27m 6s
Topo-span-bounded FSDP bucketing for AutoParallel (#481) Test CUDA #1712: Commit 7ad1142 pushed by fmassa
27m 44s main
main
27m 44s
Replay first-partitioner save decisions in torch.compile via _ReplayS… Test CUDA #1711: Commit c21df74 pushed by fmassa
27m 22s main
main
27m 22s
Replay first-partitioner save decisions in torch.compile via _ReplaySavesPartitioner Test CUDA #1710: Pull request #485 synchronize by fmassa
28m 9s fmassa/new_partitioner
fmassa/new_partitioner
28m 9s
Update to use CUDA 13.0 (#506) Test CUDA #1709: Commit 39421a2 pushed by fmassa
27m 51s main
main
27m 51s
Replay first-partitioner save decisions in torch.compile via _ReplaySavesPartitioner Test CUDA #1708: Pull request #485 synchronize by fmassa
35m 27s fmassa/new_partitioner
fmassa/new_partitioner
35m 27s
Replay first-partitioner save decisions in torch.compile via _ReplaySavesPartitioner Test CUDA #1707: Pull request #485 synchronize by fmassa
8m 7s fmassa/new_partitioner
fmassa/new_partitioner
8m 7s
Update to use CUDA 13.0 Test CUDA #1706: Pull request #506 opened by fmassa
24m 26s fmassa/update_ci
fmassa/update_ci
24m 26s
Replay first-partitioner save decisions in torch.compile via _ReplaySavesPartitioner Test CUDA #1705: Pull request #485 synchronize by fmassa
28m 11s fmassa/new_partitioner
fmassa/new_partitioner
28m 11s
Topo-span-bounded FSDP bucketing for AutoParallel Test CUDA #1704: Pull request #481 synchronize by fmassa
26m 7s fmassa/better_bucketing
fmassa/better_bucketing
26m 7s
Use upstream DTensor sharding strategies (#505) Test CUDA #1703: Commit 9fa2781 pushed by aditvenk
10h 2m 41s main
main
10h 2m 41s
Use upstream DTensor sharding strategies Test CUDA #1702: Pull request #505 synchronize by aditvenk
25m 59s aditvenk/stack/17
aditvenk/stack/17
25m 59s
Make AC recompute test topology-robust (#490) Test CUDA #1701: Commit 1aff15c pushed by aditvenk
21m 36s main
main
21m 36s
Use upstream DTensor sharding strategies Test CUDA #1700: Pull request #505 synchronize by aditvenk
25m 28s aditvenk/stack/17
aditvenk/stack/17
25m 28s
Use upstream DTensor sharding strategies Test CUDA #1699: Pull request #505 opened by aditvenk
10s aditvenk/stack/17
aditvenk/stack/17
10s
Use upstream DTensor reshape strategy Test CUDA #1698: Pull request #503 synchronize by aditvenk
29m 46s aditvenk/stack/15
aditvenk/stack/15
29m 46s
Use upstream DTensor scatter strategy Test CUDA #1697: Pull request #502 synchronize by aditvenk
29m 18s aditvenk/stack/14
aditvenk/stack/14
29m 18s
Use upstream DTensor index strategy Test CUDA #1696: Pull request #493 synchronize by aditvenk
30m 56s aditvenk/stack/5
aditvenk/stack/5
30m 56s
Use upstream DTensor convolution backward strategy Test CUDA #1695: Pull request #501 synchronize by aditvenk
29m 11s aditvenk/stack/13
aditvenk/stack/13
29m 11s
Use upstream DTensor select scatter strategy Test CUDA #1694: Pull request #497 synchronize by aditvenk
28m 39s aditvenk/stack/9
aditvenk/stack/9
28m 39s
Use upstream DTensor convolution strategy Test CUDA #1693: Pull request #500 synchronize by aditvenk
28m 5s aditvenk/stack/12
aditvenk/stack/12
28m 5s
Use upstream DTensor diagonal scatter strategy Test CUDA #1692: Pull request #496 synchronize by aditvenk
27m 55s aditvenk/stack/8
aditvenk/stack/8
27m 55s
Use upstream DTensor layer norm backward strategy Test CUDA #1691: Pull request #499 synchronize by aditvenk
28m 14s aditvenk/stack/11
aditvenk/stack/11
28m 14s