Search results

What is the best practice of unsharding the model?

Status: Open.

#3704 In meta-pytorch/torchrec;

· Morphlng opened on Jan 29, 2026

Incorrect embedding_lookup results (zeroed out) when embedding_dim > 1024, despite FBGEMM limit being 2048

Status: Open.

#3673 In meta-pytorch/torchrec;

· tiankongdeguiji opened on Jan 20, 2026

<code>_pin_and_move</code> Can Break Program Concurrency

Status: Open.

#3667 In meta-pytorch/torchrec;

· leimao opened on Jan 15, 2026

BUG: Unknown type name '__torch__.torch.classes.fbgemm.AtomicCounter'

Status: Open.

#3643 In meta-pytorch/torchrec;

· intfish123 opened on Dec 30, 2025

how to use torch.distributed.checkpoint to save and load state dict

Status: Open.

#3567 In meta-pytorch/torchrec;

· haolujun opened on Nov 20, 2025

How can I export a trained model to the Triton inference server?

Status: Open.

#3561 In meta-pytorch/torchrec;

· intfish123 opened on Nov 19, 2025

How to convert DistributedModelParallel to quantize_inference_model and use torch.jit.script to save?

Status: Open.

#3559 In meta-pytorch/torchrec;

· intfish123 opened on Nov 19, 2025

test_installation.py does not work

Status: Open.

#3539 In meta-pytorch/torchrec;

· hetian127 opened on Nov 12, 2025

Could Torchrec support PyTorch's PrivateUse1 Dispatch Key?

Status: Open.

#3525 In meta-pytorch/torchrec;

· kwgqjj opened on Nov 7, 2025

Clarify <code>WARNING:torchrec.distributed.utils:Sharding Type is data_parallel, caching params will be ignored</code>

Status: Open.

#3505 In meta-pytorch/torchrec;

· gyulaz-htec opened on Nov 3, 2025

System OOM Error in Distributed Checkpoint Writer When using fused_uvm_caching Kernel

Status: Open.

#3399 In meta-pytorch/torchrec;

· tiankongdeguiji opened on Sep 29, 2025

confused with torchrec/distributed/sharding/dynamic_sharding.py

Status: Open.

#3395 In meta-pytorch/torchrec;

· haolujun opened on Sep 26, 2025