Skip to content

Actions: microsoft/DeepSpeed

nv-accelerate-v100

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
4,906 workflow runs
4,906 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

Stage3: Use new torch grad accumulation hooks API
nv-accelerate-v100 #12572: Pull request #6773 synchronize by loadams
December 19, 2024 20:32 4m 19s deepcharm:stage3-use-new-grad-acc-api
December 19, 2024 20:32 4m 19s
Fix: forbid repeated deepspeed.initialize on training objects
nv-accelerate-v100 #12571: Pull request #6874 synchronize by traincheck-team
December 19, 2024 18:37 Action required traincheck-team:fix-6848-forbid-repeated-init
December 19, 2024 18:37 Action required
Add the missing view operations from sequence parallel(async).
nv-accelerate-v100 #12570: Pull request #6750 synchronize by loadams
December 19, 2024 17:39 Action required inkcherry:ds_overlap_fix
December 19, 2024 17:39 Action required
Zero2: avoid graph breaks in torch.compile by using param_idx
nv-accelerate-v100 #12569: Pull request #6803 synchronize by loadams
December 19, 2024 17:36 19m 18s nelyahu:zero2_param_idx
December 19, 2024 17:36 19m 18s
Cleanup ops/transformer/inference tests
nv-accelerate-v100 #12567: Pull request #6830 synchronize by loadams
December 19, 2024 17:32 12m 40s loadams/transformers-inference
December 19, 2024 17:32 12m 40s
Cleanup ops/transformer/inference tests
nv-accelerate-v100 #12566: Pull request #6830 synchronize by loadams
December 19, 2024 17:27 5m 5s loadams/transformers-inference
December 19, 2024 17:27 5m 5s
Cleanup ops/transformer/inference tests
nv-accelerate-v100 #12565: Pull request #6830 synchronize by loadams
December 19, 2024 17:25 2m 41s loadams/transformers-inference
December 19, 2024 17:25 2m 41s
hpu_accelerator: use torch.use_deterministic_algorithms
nv-accelerate-v100 #12562: Pull request #6897 opened by nelyahu
December 19, 2024 07:23 12m 18s nelyahu:patch-2
December 19, 2024 07:23 12m 18s
nv-accelerate-v100
nv-accelerate-v100 #12561: Scheduled
December 19, 2024 00:07 56m 45s master
December 19, 2024 00:07 56m 45s
Allow to compile collective for PT > 2.3
nv-accelerate-v100 #12560: Pull request #6674 reopened by loadams
December 18, 2024 21:53 2h 44m 47s nelyahu:compile_collectives
December 18, 2024 21:53 2h 44m 47s
Allow to compile collective for PT > 2.3
nv-accelerate-v100 #12559: Pull request #6674 synchronize by loadams
December 18, 2024 21:07 39m 26s nelyahu:compile_collectives
December 18, 2024 21:07 39m 26s
Copy #6674: Allow to compile collective for PT > 2.3
nv-accelerate-v100 #12558: Pull request #6894 opened by loadams
December 18, 2024 21:01 53m 58s loadams/test-compile-collectives
December 18, 2024 21:01 53m 58s
Fix checkpointable_layers Logic
nv-accelerate-v100 #12557: Pull request #6881 synchronize by Quentin-Anthony
December 18, 2024 20:25 2h 14m 25s Quentin-Anthony:qanthony/fix-act-recomp
December 18, 2024 20:25 2h 14m 25s
Support latest transformers with DSChat
nv-accelerate-v100 #12555: Pull request #6711 synchronize by loadams
December 18, 2024 20:24 1h 52m 38s loadams/fix-ds-chat-transformers
December 18, 2024 20:24 1h 52m 38s
Training ops kernels: Speeding up the Llama-based MoE architectures
nv-accelerate-v100 #12554: Pull request #6734 synchronize by loadams
December 18, 2024 19:27 Action required RezaYazdaniAminabadi:tops-kernels
December 18, 2024 19:27 Action required
Add the missing view operations from sequence parallel(async).
nv-accelerate-v100 #12553: Pull request #6750 synchronize by loadams
December 18, 2024 18:59 Action required inkcherry:ds_overlap_fix
December 18, 2024 18:59 Action required
Fix error caused by all_reduce call in domino
nv-accelerate-v100 #12552: Pull request #6880 synchronize by hwchen2017
December 18, 2024 18:02 1h 32m 20s hongwei/fix_domino_allreduce
December 18, 2024 18:02 1h 32m 20s
Stage3: Use new torch grad accumulation hooks API
nv-accelerate-v100 #12551: Pull request #6773 synchronize by loadams
December 18, 2024 17:55 17m 44s deepcharm:stage3-use-new-grad-acc-api
December 18, 2024 17:55 17m 44s
Zero2: avoid graph breaks in torch.compile by using param_idx
nv-accelerate-v100 #12550: Pull request #6803 synchronize by loadams
December 18, 2024 17:55 25m 53s nelyahu:zero2_param_idx
December 18, 2024 17:55 25m 53s
Update version.txt after 0.16.2 release
nv-accelerate-v100 #12549: Pull request #6893 opened by loadams
December 18, 2024 17:52 16m 35s AutoPR/0.16.2
December 18, 2024 17:52 16m 35s
Inference ops unit test failures/fixes
nv-accelerate-v100 #12546: Pull request #6879 synchronize by loadams
December 18, 2024 16:53 17m 55s loadams/inference-ops-test-repro
December 18, 2024 16:53 17m 55s
Stage3: Use new torch grad accumulation hooks API
nv-accelerate-v100 #12545: Pull request #6773 synchronize by loadams
December 18, 2024 16:51 15m 9s deepcharm:stage3-use-new-grad-acc-api
December 18, 2024 16:51 15m 9s