11 Pull requests opened by 6 people

• Revert D77709565

  #3173 opened Jul 9, 2025

• YAML config support for pipeline benchmarking

  #3180 opened Jul 10, 2025

• Fix compatibility issue with EmbeddingBag for IEN Publish

  #3181 opened Jul 10, 2025

• Add GPU RE for MPZCH module tests

  #3185 opened Jul 11, 2025

• Enable Changing the # of shards for CW resharding

  #3188 opened Jul 14, 2025

• Temporary Commit at 7/14/2025, 3:15:40 PM

  #3189 opened Jul 14, 2025

• Refactor benchmark utilities: centralize core logic into `_run_benchmark_core`

  #3191 opened Jul 14, 2025

• Modify sharding_single_rank_test_single_process method to test E2E deterministic results when random_seed is set

  #3193 opened Jul 15, 2025

• Fix test failure: torchrec/distributed/tests/test_quant_sequence_model_parallel.py::QuantSequenceModelParallelTest::test_sharded_quant_kv_zch

  #3194 opened Jul 15, 2025

• Fix failing CI failures on Github workflow

  #3195 opened Jul 15, 2025

• Fix skipped tests for test_model_parallel_gloo

  #3196 opened Jul 15, 2025

1 Issue closed by 1 person

• OOM when I create a 196GB embedding.

  #2858 closed Jul 11, 2025

1 Issue opened by 1 person

• [Bug] `FP16` returns `FP32` and `BF16` fails during sharding

  #3187 opened Jul 14, 2025

7 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.

• message: 'AttributeError: ''OptimizerArgs'' object has no attribute ''learning_rate''' when running train_dlrm.py

  #3170 commented on Jul 14, 2025 • 0 new comments

• Can NVIDIA-Merlin/HierarchicalKV integrated to torchrec as EmbeddingBag Extension?

  #2923 commented on Jul 15, 2025 • 0 new comments

• Create an example for MPZCH

  #3063 commented on Jul 15, 2025 • 0 new comments

• Zero Collision Hash Benchmark Framework

  #3127 commented on Jul 14, 2025 • 0 new comments

• Operator level microbenchmarking

  #3154 commented on Jul 9, 2025 • 0 new comments

• Add tensorboard to display training and evaluation metrics

  #3163 commented on Jul 14, 2025 • 0 new comments

• batched_embedding_kernel int32 support behind jk

  #3164 commented on Jul 10, 2025 • 0 new comments