Skip to content

Actions: deepspeedai/DeepSpeed

Actions

python

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
python.yml
2,244 workflow runs
2,244 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
Unmanaged gradient accumulation: ZeRO stage 2 support python #16085: Pull request #8203 synchronize by sfc-gh-truwase
2m 8s sfc-gh-truwase/gas_mgmt_zero2
sfc-gh-truwase/gas_mgmt_zero2
2m 8s
Wiring up offload_opt_states python #16084: Pull request #8186 synchronize by pengdurice
5m 56s pengdurice:peng-better-deepcompile-v2
pengdurice:peng-better-deepcompile-v2
5m 56s
fix(zero3): async grad offload + pinned offload buffers by default python #16083: Pull request #8207 opened by delock
1m 38s delock:gma/async_zero_offload
delock:gma/async_zero_offload
1m 38s
[WIP][AutoTP] Complete uneven sharding and universal checkpoint support python #16082: Pull request #8185 synchronize by jinyouzhi
-1s jinyouzhi:uneven
jinyouzhi:uneven
-1s
python python #16081: Merge group checks requested
1m 32s
1m 32s
[WIP][AutoTP] Complete uneven sharding and universal checkpoint support python #16080: Pull request #8185 synchronize by jinyouzhi
5m 25s jinyouzhi:uneven
jinyouzhi:uneven
5m 25s
python python #16079: Merge group checks requested
1m 29s
1m 29s
Add Triton grouped-GEMM for MoE experts on Ampere/Ada python #16078: Pull request #8180 synchronize by hwchen2017
1m 30s hongwei/group_gemm
hongwei/group_gemm
1m 30s
Consolidate LR scheduler tensor-LR tests into a contract suite python #16077: Pull request #8206 synchronize by delock
2m 57s delock:gma/lr-scheduler-lr-tensor-test
delock:gma/lr-scheduler-lr-tensor-test
2m 57s
[WIP][AutoTP] Complete uneven sharding and universal checkpoint support python #16076: Pull request #8185 synchronize by jinyouzhi
Action required jinyouzhi:uneven
jinyouzhi:uneven
Action required
Consolidate LR scheduler tensor-LR tests into a contract suite python #16075: Pull request #8206 opened by delock
1m 40s delock:gma/lr-scheduler-lr-tensor-test
delock:gma/lr-scheduler-lr-tensor-test
1m 40s
python python #16074: Merge group checks requested
1m 30s
1m 30s
python python #16073: Scheduled
1m 23s master
master
1m 23s
python python #16072: Merge group checks requested
1m 27s
1m 27s
python python #16071: Merge group checks requested
1m 25s
1m 25s
Fix AutoEP ZeRO-1/2 universal conversion python #16070: Pull request #8198 synchronize by tohtana
1m 30s tohtana:tohtana/fix-autoep-zero12-universal-checkpoint
tohtana:tohtana/fix-autoep-zero12-universal-checkpoint
1m 30s
python python #16069: Merge group checks requested
1m 28s
1m 28s
Fix DeepCompile ZeRO-3 gathered parameter ownership python #16068: Pull request #8157 synchronize by tohtana
1m 34s tohtana:tohtana/deepcompile-z3-gather-ownership
tohtana:tohtana/deepcompile-z3-gather-ownership
1m 34s
Stabilize DeepCompile ZeRO-3 memory scheduling python #16067: Pull request #8169 synchronize by tohtana
1m 29s tohtana:tohtana/deepcompile-zero3-memory-stability-scheduler-standalone
tohtana:tohtana/deepcompile-zero3-memory-stability-scheduler-standalone
1m 29s
python python #16066: Merge group checks requested
1m 21s
1m 21s
python python #16065: Merge group checks requested
1m 27s
1m 27s
python python #16064: Merge group checks requested
1m 34s
1m 34s
Preserve tensor learning rates in OneCycle python #16063: Pull request #8205 opened by tohtana
1m 30s tohtana:tohtana/fix-onecycle-tensor-lr-init
tohtana:tohtana/fix-onecycle-tensor-lr-init
1m 30s
python python #16062: Merge group checks requested
1m 30s
1m 30s
MoE dispatch: fix silent Tutel + TP corruption, speed up native and Tutel paths python #16061: Pull request #8195 synchronize by jinyouzhi
1m 31s jinyouzhi:moe
jinyouzhi:moe
1m 31s