4 Releases published by 1 person

• v1.2.0-rc1

  published Jun 6, 2025

• v1.2.0-rc2

  published Jun 6, 2025

• v1.2.0-rc3

  published Jun 6, 2025

• v1.2.0

  published Jun 6, 2025

25 Pull requests opened by 16 people

• Support RecMetrics new options in Pyper config

  #2987 opened May 21, 2025

• Fix undeclared type in colab for sharding examples

  #2989 opened May 21, 2025

• Enable fx tracing for maybe_td_to_kjt

  #2991 opened May 21, 2025

• Migrate to correct logger interface

  #2996 opened May 23, 2025

• Add missing Pyre mode headers]

  #3044 opened Jun 5, 2025

• Create an example for MPZCH

  #3063 opened Jun 9, 2025

• Revert D76061842

  #3066 opened Jun 9, 2025

• Benchamrk framework for torchrec

  #3072 opened Jun 9, 2025

• Simplify fqns_to_feature_names in Delta Tracker

  #3081 opened Jun 11, 2025

• Test commit

  #3084 opened Jun 11, 2025

• add support for List[torch.Tensor]

  #3088 opened Jun 12, 2025

• OSS TorchRec Internal MPZCH modules (#3017)

  #3089 opened Jun 12, 2025

• Change job name while uploading sharding plan to Manifold

  #3092 opened Jun 13, 2025

• Added Optimizer configuration that supports optimizer type, learning rate, momentum, and weight decay configurations.

  #3094 opened Jun 13, 2025

• Use Index Select instead of permute operation reduction in remote ro compute time

  #3097 opened Jun 13, 2025

• Added model config for supporting SparseNN, TowerSparseNN, TowerCollectionSparseNN models.

  #3104 opened Jun 17, 2025

• Support ebc for kv zch

  #3106 opened Jun 17, 2025

• Added Optimizer configuration that supports optimizer type, learning rate, momentum, and weight decay configurations.

  #3107 opened Jun 17, 2025

• kvzch use new operator in model publish

  #3108 opened Jun 17, 2025

• Added model wrappers for DeepFM and DLRM

  #3115 opened Jun 18, 2025

• fix validate nightly binaries

  #3117 opened Jun 20, 2025

• skip python 3.13t for pip install fbgemm-gpu

  #3118 opened Jun 20, 2025

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

  #3119 opened Jun 20, 2025

• Update docstrings for stride_per_key_per_rank

  #3120 opened Jun 20, 2025

• Add usage example to KeyedJaggedTensor docstring

  #3121 opened Jun 20, 2025

2 Issues closed by 2 people

• Compatibility with PyTorch 2.7.0 and FBGEMM_GPU 1.2.0?

  #2971 closed Jun 6, 2025

• [Question] Is there a way to run torchrec on sm_100?

  #2854 closed May 27, 2025

10 Issues opened by 6 people

• Which lightning strategy to use with torchrec optimizers?

  #3114 opened Jun 18, 2025

• Torchrec for older pytorch and fbgemm

  #3113 opened Jun 18, 2025

• 'SplitTableBatchedEmbeddingBagsCodegen' object has no attribute 'get_learning_rate' when upgrading to 1.2.0

  #3099 opened Jun 16, 2025

• I want to use CPU-based distributed approach to train a small recommendation model. Is there a demo available for me to refer to?

  #3033 opened Jun 3, 2025

• Does TorchReD support distributed training of large-scale recommender systems on CPU?

  #3031 opened Jun 3, 2025

• [Bug]: ManagedCollisionEmbeddingCollection returns all-zero embeddings after applying apply_optimizer_in_backward with RowWiseAdagrad

  #3030 opened Jun 3, 2025

• If I train with CPU and the embedding layer takes up too much space, how can I solve this problem?

  #3029 opened Jun 3, 2025

• For training recommender systems using CPU, are there any optimization capabilities for embeddings?

  #3028 opened Jun 3, 2025

• [Documentation] more accurate discription of behavior of KeyedJaggedTensor

  #3013 opened May 29, 2025

• [BUG]: Cannot copy out of meta tensor; no data!

  #3006 opened May 28, 2025

5 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.

• roadmap 2025 H1 H2

  #2825 commented on May 27, 2025 • 0 new comments

• OOM when I create a 196GB embedding.

  #2858 commented on May 27, 2025 • 0 new comments

• Does torchrec support implicit feedback datasets?

  #2873 commented on May 27, 2025 • 0 new comments

• ManagedCollisionEmbeddingBagCollectionSharder cannot support CPU

  #2958 commented on Jun 6, 2025 • 0 new comments

• mark metricmodule methods as experimental

  #2983 commented on May 28, 2025 • 0 new comments