Skip to content

Actions: deepspeedai/DeepSpeed

Actions

python

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
python.yml
2,345 workflow runs
2,345 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
Drop the documented grad_hooks ZeRO option, which does not exist python #16248: Pull request #8242 opened by vineethsaivs
Action required vineethsaivs:fix/zero-config-grad-hooks-doc
vineethsaivs:fix/zero-config-grad-hooks-doc
Action required
[AutoTP] Replace tp_shard process-wide globals with per-model AutoTPMeta python #16247: Pull request #8241 opened by delock
1m 43s gma/autotp-per-model-meta
gma/autotp-per-model-meta
1m 43s
python python #16246: Scheduled
1m 20s master
master
1m 20s
[DeepCompile] Fix staticmethod handling on Python 3.9 python #16245: Pull request #8240 opened by baremetaldevx86
Action required baremetaldevx86:fix/7433-staticmethod-descriptor
baremetaldevx86:fix/7433-staticmethod-descriptor
Action required
Add native (DeepNVMe) host-memory pinning backend for accelerators python #16244: Pull request #8211 synchronize by sfc-gh-truwase
1m 36s tjruwase/native-pin-memory
tjruwase/native-pin-memory
1m 36s
python python #16243: Merge group checks requested
1m 56s
1m 56s
Filter --include against the real slots, not against itself python #16242: Pull request #8239 opened by vineethsaivs
Action required vineethsaivs:fix/launcher-include-real-slots
vineethsaivs:fix/launcher-include-real-slots
Action required
python python #16241: Merge group checks requested
1m 25s
1m 25s
Split native host pinning into standalone pin_memory op python #16240: Pull request #8236 synchronize by sfc-gh-truwase
1m 28s tjruwase/pin-memory-op
tjruwase/pin-memory-op
1m 28s
python python #16239: Merge group checks requested
1m 45s
1m 45s
Unmanaged gradient accumulation: ZeRO offload support python #16238: Pull request #8225 synchronize by delock
1m 27s sfc-gh-truwase/gas_mgmt_offload
sfc-gh-truwase/gas_mgmt_offload
1m 27s
(1/2) Implementing Compiler Pass for AutoTP python #16237: Pull request #8204 synchronize by therealnaveenkamal
Action required therealnaveenkamal:feature/autotp
therealnaveenkamal:feature/autotp
Action required
Fix compilation error for torch 2.12+ python #16236: Pull request #8238 opened by hwchen2017
1m 29s hongwei/c++_version
hongwei/c++_version
1m 29s
Split native host pinning into standalone pin_memory op python #16235: Pull request #8236 synchronize by sfc-gh-truwase
1m 26s tjruwase/pin-memory-op
tjruwase/pin-memory-op
1m 26s
Split native host pinning into standalone pin_memory op python #16234: Pull request #8236 synchronize by sfc-gh-truwase
1m 30s tjruwase/pin-memory-op
tjruwase/pin-memory-op
1m 30s
Add configurable sum gradient reduction python #16233: Pull request #8232 synchronize by zupengwang
Action required zupengwang:feature/gradient-allreduce-sum
zupengwang:feature/gradient-allreduce-sum
Action required
Keep the elastic batch size within max_train_batch_size python #16232: Pull request #8237 opened by vineethsaivs
Action required vineethsaivs:fix/elasticity-batch-over-max
vineethsaivs:fix/elasticity-batch-over-max
Action required
python python #16231: Scheduled
1m 17s master
master
1m 17s
Split native host pinning into standalone pin_memory op python #16230: Pull request #8236 synchronize by sfc-gh-truwase
1m 26s tjruwase/pin-memory-op
tjruwase/pin-memory-op
1m 26s
Split native host pinning into standalone pin_memory op python #16229: Pull request #8236 synchronize by sfc-gh-truwase
1m 21s tjruwase/pin-memory-op
tjruwase/pin-memory-op
1m 21s
Split native host pinning into standalone pin_memory op python #16228: Pull request #8236 opened by sfc-gh-truwase
1m 34s tjruwase/pin-memory-op
tjruwase/pin-memory-op
1m 34s
Unmanaged gradient accumulation: ZeRO offload support python #16227: Pull request #8225 synchronize by sfc-gh-truwase
1m 28s sfc-gh-truwase/gas_mgmt_offload
sfc-gh-truwase/gas_mgmt_offload
1m 28s
[AutoTP] Complete uneven sharding and universal checkpoint support python #16226: Pull request #8185 synchronize by delock
6m 26s jinyouzhi:uneven
jinyouzhi:uneven
6m 26s
python python #16225: Scheduled
1m 45s master
master
1m 45s
Add an opt-in DeepEP transport for the AutoEP expert all-to-all python #16224: Pull request #8213 synchronize by yh0903
Action required yh0903:autoep-deepep-backend
yh0903:autoep-deepep-backend
Action required