All Workflows build_and_test.yml documentation.yml nightly.yml release.yml

Actor

All actors zhangyiss

Status

All status Success Failure Waiting Running

Add a 2-pass col reduce for CUDA (#2863)

#59:Scheduled

main

2025-12-05 14:33:11 +08:00

0s

Update gumbel function signature parameters (#2868)

#57:Scheduled

main

2025-12-05 14:33:09 +08:00

0s

Fix export scatters (#2852)

#53:Scheduled

main

2025-12-04 14:33:10 +08:00

0s

[CUDA] Migrate conv code to new cuDNN APIs (#2847)

#50:Scheduled

main

2025-12-03 14:33:08 +08:00

0s

[CUDA] Use cuDNN attention when T_q != T_kv (#2843)

#48:Scheduled

main

2025-12-02 14:33:08 +08:00

0s

[CUDA] Use cuDNN attention when T_q != T_kv (#2843)

#47:Scheduled

main

2025-11-29 16:38:06 +08:00

0s

[CUDA] Use cuDNN attention when T_q != T_kv (#2843)

#46:Scheduled

main

2025-11-28 14:33:08 +08:00

0s

[CUDA] Support array mask in SDPA (#2822)

#44:Scheduled

main

2025-11-27 14:33:10 +08:00

0s

Fix `mx.core.linspace` type annotation (#2820)

#39:Scheduled

main

2025-11-26 14:33:09 +08:00

0s

Force cudaGraphExec reinstantiation when clusters are used (#2813)

#36:Scheduled

main

2025-11-25 14:33:10 +08:00

0s

patch bump for future version (#2804)

#30:Scheduled

main

2025-11-22 16:38:10 +08:00

0s

Fix macos release target and linux arm release (#2802)

#28:Scheduled

main

2025-11-21 14:33:10 +08:00

0s

[CUDA] cuDNN backward attention (#2762)

#24:Scheduled

main

2025-11-20 14:33:09 +08:00

0s

Fix building with CUDA < 12.8 (#2782)

#18:Scheduled

main

2025-11-19 14:33:10 +08:00

0s

Use std::optional for mask_arr arg (#2763)

#14:Scheduled

main

2025-11-18 14:33:12 +08:00

0s

[CUDA] cuDNN forward attention (#2743)

#12:Scheduled

main

2025-11-15 16:38:12 +08:00

0s

minor debugging for publishing (#2739)

#11:Scheduled

main

2025-11-14 14:33:10 +08:00

0s

Linux on arm (#2751)

#10:Scheduled

main

2025-11-13 14:33:11 +08:00

0s

Compatibility with pip-installed openmpi (#2741)

#8:Scheduled

main

2025-11-12 14:33:09 +08:00

0s

Compatibility with pip-installed openmpi (#2741)

#7:Scheduled

main

2025-11-11 14:33:09 +08:00

0s

Shapeless support for `zeros/ones_like` (#2726)

#6:Scheduled

main

2025-11-08 16:38:10 +08:00

0s

[CUDA] Reduce use of managed memory (#2725)

#5:Scheduled

main

2025-11-07 14:33:15 +08:00

0s

don't test when doing release (#2734)

#4:Scheduled

main

2025-11-06 14:33:09 +08:00

0s

skip self-hosted runners on forks (#2730)

#3:Scheduled

main

2025-11-05 14:33:10 +08:00

0s

use faster dequant for fp4 qmv (#2720)

#2:Scheduled

main

2025-11-04 14:33:08 +08:00

0s

Fix the order of hosts in the ring (#2718)

#1:Scheduled

main

2025-11-01 16:38:09 +08:00

0s