Back to search
Top Contributors
173
commits
138
commits
98
commits
63
commits
58
commits
51
commits
42
commits
40
commits
33
commits
28
commits
A PyTorch native platform for training generative AI models
Stars
5.2k
Forks
777
Watchers
5.2k
Open Issues
480
Overall repository health assessment
No package.json found
This might not be a Node.js project
173
commits
138
commits
98
commits
63
commits
58
commits
51
commits
42
commits
40
commits
33
commits
28
commits
[GraphTrainer][AutoDev] Enable cudagraph in aot_fx_trace mode (#2868)
d98fa54View on GitHub[compile] Share SimpleFSDP wrapper class across same-type module instances (#2754)
e43f252View on GitHub[graph_trainer] Add graph module dumping section to CLAUDE.md (#2902)
2c6bb1cView on GitHub[graph_trainer] Enable precompile with regional_inductor (#2883)
4a57715View on GitHub[FSDP2] lazy import FSDPMeshInfo/ShardPlacementResult for older PyTorch (#2888)
ec1b1c9View on GitHub[graph_trainer] Parameterize AutoDev board configuration (#2892)
70f005eView on GitHubSet cast_forward_inputs=False for LLM model's MixedPrecisionPolicy (#2881)
21c4913View on GitHub[Recreate of 2753] [graph_trainer] Replace trace_module/run_traced_module with minimal_fx_tracer API (#2882)
390da99View on GitHubFix UnicodeDecodeError in integration test runner (#2863)
06dfdafView on GitHubBump pypa/gh-action-pypi-publish from 1.13.0 to 1.14.0 in the github-actions group (#2880)
dbe5745View on GitHub[GraphTrainer][AutoDev] Accept float inputs in CUDAGraphWrapper (#2876)
762a24eView on GitHub