Skip to content

Actions: liuliu/ccv

analyze

Actions

Loading...
Loading

Create status badge

Loading
100 workflow runs
100 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

Support activation NHWC while W in NCHW for conv_transpose in CUDNN.
analyze #100: Commit b2bcf7f pushed by liuliu
February 25, 2025 01:17 8m 50s unstable
February 25, 2025 01:17 8m 50s
Add conv3d support on MPS backend.
analyze #99: Commit 19f4960 pushed by liuliu
January 24, 2025 06:28 1h 35m 4s unstable
January 24, 2025 06:28 1h 35m 4s
Fix an issue with cudnn conv3d.
analyze #98: Commit 819aa2a pushed by liuliu
January 24, 2025 05:18 16m 54s unstable
January 24, 2025 05:18 16m 54s
Fix more minor issues with convolution construction and group norm.
analyze #97: Commit c0bd3a5 pushed by liuliu
January 22, 2025 02:52 13m 8s unstable
January 22, 2025 02:52 13m 8s
Add support for conv3d on call to cudnn side.
analyze #96: Commit d157996 pushed by liuliu
January 21, 2025 17:51 1h 15m 42s unstable
January 21, 2025 17:51 1h 15m 42s
Add some code to expand convolution to 3d (still work in progress).
analyze #95: Commit e3ccc41 pushed by liuliu
January 20, 2025 00:56 8m 54s unstable
January 20, 2025 00:56 8m 54s
Fix bug where grid.z is larger than 65535, there is a launch failure.
analyze #94: Commit f3a29ea pushed by liuliu
January 18, 2025 01:29 8m 57s unstable
January 18, 2025 01:29 8m 57s
Disable splitkv for most cases (only decoding) due to excessive RAM u…
analyze #93: Commit 3fec0ed pushed by liuliu
January 16, 2025 19:44 1h 8m 31s unstable
January 16, 2025 19:44 1h 8m 31s
Add a basic bench for grad.
analyze #92: Commit 645e6f3 pushed by liuliu
January 1, 2025 17:26 29m 35s unstable
January 1, 2025 17:26 29m 35s
Add a test case for larger than 2^16 grid dimension.
analyze #91: Commit ca844cf pushed by liuliu
December 22, 2024 16:55 1h 46m 19s unstable
December 22, 2024 16:55 1h 46m 19s
Add a final fix for MFAv2 where we compute the gid ourselves.
analyze #90: Commit 85efcf8 pushed by liuliu
December 20, 2024 23:15 1h 40m 59s unstable
December 20, 2024 23:15 1h 40m 59s
Increase testing length.
analyze #89: Commit 05c73e6 pushed by liuliu
December 20, 2024 21:30 1h 19m 9s unstable
December 20, 2024 21:30 1h 19m 9s
Revert the parameter selection since it is adversarial on M4.
analyze #88: Commit a16537f pushed by liuliu
December 17, 2024 00:33 39m 2s unstable
December 17, 2024 00:33 39m 2s
Need better way to organize the estimations.
analyze #87: Commit 3bff50d pushed by liuliu
December 11, 2024 05:50 2h 43m 1s unstable
December 11, 2024 05:50 2h 43m 1s
Fix build break.
analyze #86: Commit e40d046 pushed by liuliu
December 11, 2024 05:49 53m 17s unstable
December 11, 2024 05:49 53m 17s
Do fp16 for sdpa_bench.
analyze #85: Commit 0adf85a pushed by liuliu
December 11, 2024 03:42 31m 46s unstable
December 11, 2024 03:42 31m 46s
Add sdpa bench.
analyze #84: Commit aafc394 pushed by liuliu
December 11, 2024 00:37 1h 9m 38s unstable
December 11, 2024 00:37 1h 9m 38s
Fix a crash on upsample if rheight / rwidth < 1 (not good result, but…
analyze #83: Commit 07512af pushed by liuliu
December 10, 2024 18:01 25m 41s unstable
December 10, 2024 18:01 25m 41s
Implement the logic to handle failures with cufile.
analyze #82: Commit 2a59318 pushed by liuliu
December 4, 2024 23:31 9m 3s unstable
December 4, 2024 23:31 9m 3s
Fix a bug when reinit tensor, we didn't do so with stride.
analyze #81: Commit 98eb262 pushed by liuliu
December 3, 2024 21:38 13m 16s unstable
December 3, 2024 21:38 13m 16s
Use new exec_dep logic for gradient checkpointing.
analyze #80: Commit ce93a48 pushed by liuliu
December 3, 2024 19:55 20m 12s unstable
December 3, 2024 19:55 20m 12s
Update from the failed branch where we try to replace core exec_dep t…
analyze #79: Commit 787623a pushed by liuliu
December 3, 2024 18:14 20m 23s unstable
December 3, 2024 18:14 20m 23s
Missing the file.
analyze #78: Commit 77b1346 pushed by liuliu
December 3, 2024 01:21 1h 15m 32s unstable
December 3, 2024 01:21 1h 15m 32s
Reorganize code a little bit in preparation for using chain decomposi…
analyze #77: Commit d60bb2a pushed by liuliu
December 3, 2024 01:15 32m 49s unstable
December 3, 2024 01:15 32m 49s
Use chain decomposition for exec_dep computation in memory_reduction.
analyze #76: Commit c174631 pushed by liuliu
December 3, 2024 00:06 46m 18s unstable
December 3, 2024 00:06 46m 18s