5 Pull requests opened by 5 people

• convert stride_per_key_per_rank to tensor inside KJT

  #2959 opened May 8, 2025

• Support R-squared

  #2964 opened May 12, 2025

• Swap CPU TBE lookup kernel with SSD TBE lookup kernel for tables being offloaded to SSD

  #2965 opened May 13, 2025

• Remove some directories

  #2967 opened May 13, 2025

• Support DI sharding for FPE_EBC

  #2968 opened May 13, 2025

2 Issues opened by 2 people

• ManagedCollisionEmbeddingBagCollectionSharder cannot support CPU

  #2958 opened May 8, 2025

• CUDA error: an Illegal memory access was encountered

  #2957 opened May 7, 2025

6 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Allow a context manager to be called around apply_jit

  #2927 commented on May 8, 2025 • 0 new comments

• {Tests-WIP}[Torchrec] Add context manager to use next batch context for postprocs

  #2939 commented on May 8, 2025 • 0 new comments

• Add support for pipelined postproc

  #2940 commented on May 8, 2025 • 0 new comments

• Add padding in dynamic sharding for tensors before all2all

  #2944 commented on May 12, 2025 • 0 new comments

• DistributedModelParallel resharding Interface

  #2945 commented on May 14, 2025 • 0 new comments

• Update KJT stride calculation logic to be based off of inverse_indices for VBE KJTs.

  #2949 commented on May 13, 2025 • 0 new comments