transformers
[`distributed`] decoupled tp_plan and ep_plan
#48859
Merged

Commits
  • Add dense and expert device mesh views with a MeshManager
    3outeille committed 29 days ago
  • cleaning
    3outeille committed 29 days ago
  • clean
    3outeille committed 29 days ago
  • update
    3outeille committed 29 days ago
  • clean doc
    3outeille committed 29 days ago
  • Use ep_size in the Kimi K2.5 expert parallel example
    3outeille committed 29 days ago
  • Enable expert parallelism in the Mega MoE example
    3outeille committed 29 days ago
  • warn once
    3outeille committed 29 days ago
  • Add dense and expert device mesh views with a MeshManager
    3outeille committed 29 days ago
  • cleaning
    3outeille committed 29 days ago
  • update
    3outeille committed 29 days ago
  • clean doc
    3outeille committed 29 days ago
  • Update src/transformers/distributed/mixin.py
    3outeille committed 29 days ago
  • style: ruff import block formatting in distributed config
    3outeille committed 29 days ago
  • add more comments
    3outeille committed 29 days ago
  • Decouple tp_plan and ep_plan for expert parallelism.
    3outeille committed 29 days ago
  • Match expert paths with a regex and keep only expert rules under token dispatch
    3outeille committed 29 days ago
  • revert change on warning
    3outeille committed 29 days ago
  • Apply suggestion from @ArthurZucker
    3outeille committed 29 days ago
  • cleaning
    3outeille committed 29 days ago
  • Add dense and expert device mesh views with a MeshManager
    3outeille committed 26 days ago
  • cleaning
    3outeille committed 26 days ago
  • clean
    3outeille committed 26 days ago
  • update
    3outeille committed 26 days ago
  • clean doc
    3outeille committed 26 days ago
  • Use ep_size in the Kimi K2.5 expert parallel example
    3outeille committed 26 days ago
  • Enable expert parallelism in the Mega MoE example
    3outeille committed 26 days ago
  • warn once
    3outeille committed 26 days ago
  • Add dense and expert device mesh views with a MeshManager
    3outeille committed 26 days ago
  • cleaning
    3outeille committed 26 days ago
  • update
    3outeille committed 26 days ago
  • clean doc
    3outeille committed 26 days ago
  • Update src/transformers/distributed/mixin.py
    3outeille committed 26 days ago
  • style: ruff import block formatting in distributed config
    3outeille committed 26 days ago
  • add more comments
    3outeille committed 26 days ago
  • Decouple tp_plan and ep_plan for expert parallelism.
    3outeille committed 26 days ago
  • Match expert paths with a regex and keep only expert rules under token dispatch
    3outeille committed 26 days ago
  • revert change on warning
    3outeille committed 26 days ago
  • Apply suggestion from @ArthurZucker
    3outeille committed 26 days ago
  • cleaning
    3outeille committed 26 days ago
  • fix
    3outeille committed 26 days ago
  • Merge branch 'ep-mesh-infrastructure' of github.com:huggingface/transformers into ep-mesh-infrastructure
    3outeille committed 26 days ago
  • Merge branch 'ep-plans' of github.com:huggingface/transformers into ep-plans
    3outeille committed 26 days ago
  • revert
    3outeille committed 26 days ago
  • Merge branch 'main' into ep-mesh-infrastructure
    3outeille committed 26 days ago
  • Merge branch 'ep-mesh-infrastructure' into ep-plans
    3outeille committed 26 days ago
  • typo
    3outeille committed 26 days ago
  • Merge branch 'main' into ep-mesh-infrastructure
    3outeille committed 26 days ago
  • clearer
    3outeille committed 26 days ago
  • Merge branch 'ep-mesh-infrastructure' of github.com:huggingface/transformers into ep-mesh-infrastructure
    3outeille committed 26 days ago
  • Merge branch 'main' into ep-mesh-infrastructure
    3outeille committed 25 days ago
  • clean
    3outeille committed 25 days ago
  • Merge branch 'ep-mesh-infrastructure' of github.com:huggingface/transformers into ep-mesh-infrastructure
    3outeille committed 25 days ago
  • Merge branch 'ep-mesh-infrastructure' into ep-plans
    3outeille committed 25 days ago
  • cleaning
    3outeille committed 25 days ago
  • Merge branch 'ep-plans' of github.com:huggingface/transformers into ep-plans
    3outeille committed 25 days ago
  • Merge branch 'main' into ep-mesh-infrastructure
    3outeille committed 25 days ago
  • Merge branch 'ep-mesh-infrastructure' into ep-plans
    3outeille committed 25 days ago
  • Update src/transformers/distributed/utils.py
    3outeille committed 25 days ago
  • Merge branch 'main' into ep-mesh-infrastructure
    3outeille committed 25 days ago
  • Update src/transformers/distributed/configuration_utils.py
    3outeille committed 25 days ago
  • linting
    3outeille committed 25 days ago
  • Merge branch 'ep-mesh-infrastructure' of github.com:huggingface/transformers into ep-mesh-infrastructure
    3outeille committed 25 days ago
  • fix comment
    3outeille committed 25 days ago
  • Merge branch 'ep-mesh-infrastructure' into ep-plans
    3outeille committed 25 days ago
  • Merge branch 'main' into ep-mesh-infrastructure
    3outeille committed 25 days ago
  • Merge branch 'ep-mesh-infrastructure' into ep-plans
    3outeille committed 25 days ago
  • Merge branch 'main' into ep-plans
    3outeille committed 25 days ago
Loading