Issues

• [MSVC] Different linkage

  pytorch/FBGEMM#6030 · Filter by author SunBlack opened on Jul 20, 2026

• _fbgemm_torch_compat_table is outdated and emits false compatibility warnings

  pytorch/FBGEMM#5953 · Filter by author bwnotfound opened on Jun 25, 2026

• AttributeError: '_OpNamespace' 'fbgemm' object has no attribute 'get_bucket_sorted_indices_and_bucket_tensor'

  pytorch/FBGEMM#5666 · Filter by author YINUO1221 opened on Apr 21, 2026
  2

• FBGEMM-GPU v1.5.0 Cannot Run on SM70 and SM75 Architectures

  pytorch/FBGEMM#5481 · Filter by author tiankongdeguiji opened on Mar 16, 2026
  6

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

  pytorch/FBGEMM#5326 · Filter by author tiankongdeguiji opened on Jan 20, 2026
  9

• CUDA 13 builds not pushed to pypi index

  pytorch/FBGEMM#5321 · Filter by author winglian opened on Jan 16, 2026
  1

• vllm serve pytorch/gemma-3-12b-it-FP8 failed with fbgemm-gpu-genai 1.4.2 on B200

  pytorch/FBGEMM#5221 · Filter by author huydhn opened on Dec 13, 2025
  1

• How to apply gradient clip in fused optimizer?

  pytorch/FBGEMM#5176 · Filter by author acmilannesta opened on Nov 28, 2025

• include <torch/csrc/jit/runtime/static/ops.h> No such file or directory

  pytorch/FBGEMM#5175 · Filter by author fmo-mt opened on Nov 27, 2025
  3

• Release Schedule for FBGEMM_GPU 1.4.0 with PyTorch 2.9.0 Compatibility

  pytorch/FBGEMM#5104 · Filter by author tiankongdeguiji opened on Nov 10, 2025
  8

• TypeError when using string as callable in Triton kernel with tl.constexpr

  pytorch/FBGEMM#5070 · Filter by author CaesarKingW opened on Oct 30, 2025

• Release timeline for 1.4 / 1.4.1 wheels (tags exist but no PyPI artifacts for cuda 13.0)

  pytorch/FBGEMM#5067 · Filter by author foreverYoungGitHub opened on Oct 28, 2025
  3

• split_embeddings_cache/lru_cache_populate.cu compiled error

  pytorch/FBGEMM#5036 · Filter by author gzj-ai opened on Oct 21, 2025

• Building fbgemm_genai from source getting Killed

  pytorch/FBGEMM#5005 · Filter by author mavibirdesmi opened on Oct 15, 2025
  1

• RuntimeError: Tried to instantiate class 'fbgemm.DramKVEmbeddingCacheWrapper', but it does not exist! Ensure that it is registered via torch::class_

  pytorch/FBGEMM#4861 · Filter by author tiankongdeguiji opened on Sep 11, 2025
  3

• unable to import fbgemm_gpu

  pytorch/FBGEMM#4754 · Filter by author vkuzo opened on Aug 21, 2025
  12

• Various problems on arm64

  pytorch/FBGEMM#4679 · Filter by author omegacoleman opened on Aug 12, 2025

• dense_embedding_codegen_lookup_function CPU is uncorrect with GPU

  pytorch/FBGEMM#4586 · Filter by author JoheyHan opened on Jul 28, 2025
  2

• Does INT8 Quantized Convolution Still Involve Floating-Point Operations？

  pytorch/FBGEMM#4474 · Filter by author zengwz opened on Jul 11, 2025

• How can I turn on the Warp specialization in the grouped_gemm op?

  pytorch/FBGEMM#4321 · Filter by author linlang1837 opened on Jun 11, 2025
  4

• version compatiblity with import lib

  pytorch/FBGEMM#4198 · Filter by author jerryzh168 opened on May 27, 2025
  2

• MoE layers test triton compilation error

  pytorch/FBGEMM#4192 · Filter by author moderato opened on May 27, 2025
  3

• Do you plan to support heterogeneous schedules for FusedEmbeddingBag operations?

  pytorch/FBGEMM#4152 · Filter by author PanZaifeng opened on May 19, 2025

• error happened when torch export onnx

  pytorch/FBGEMM#4148 · Filter by author cwy0110 opened on May 19, 2025

• jagged_to_padded_dense: an autograd kernel was not registered to the Autograd key(s) but we are trying to backprop through it

  pytorch/FBGEMM#4115 · Filter by author RyanZotti opened on May 13, 2025
  5