Skip to content

Pull requests: pytorch/FBGEMM

New pull request New
384 Open 2,963 Closed
384 Open 2,963 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Enable AMD BF16 Grouped Gemm cla signed fb-exported
#3526 opened Dec 22, 2024 by jwfromm Loading…
2
change pmt require grad to false when detached cla signed fb-exported
#3525 opened Dec 22, 2024 by duduyi2013 Loading…
2
remove output dtype restriction in ssd tbe cla signed fb-exported
#3524 opened Dec 22, 2024 by duduyi2013 Loading…
3
Check src & dst dtypes in allgather to prevent silent failures. cla signed fb-exported
#3523 opened Dec 20, 2024 by ChenheliHua Loading…
2
Improve performance of prefill mode FP8 Grouped Gemm cla signed fb-exported
#3522 opened Dec 20, 2024 by jwfromm Loading…
3
Add fused_moe kernel to ck_extension cla signed fb-exported
#3518 opened Dec 19, 2024 by sijiac Loading…
3
Cherry-pick CK PR #1636 for fp8 GEMM rowwise for 70B Prefill cla signed fb-exported
#3517 opened Dec 19, 2024 by zjing14 Loading…
5
env variable to select rounding mode cla signed fb-exported
#3515 opened Dec 19, 2024 by hhyuanf Loading…
6
Optimzed backward pass for ROCm devices (pt 2) ciflow/rocm cla signed fb-exported module: rocm
#3511 opened Dec 18, 2024 by q10 Loading…
6
Add NEON transpose kernel for Float cla signed
#3510 opened Dec 18, 2024 by annop-w Loading…
1
Back out "Manual loop unroll for rocm inference" ciflow/rocm cla signed fb-exported module: rocm
#3506 opened Dec 15, 2024 by brad-mengchi Loading…
4
[fbgemm_gpu] Add support for CUDA 12.6 builds in OSS cla signed
#3503 opened Dec 13, 2024 by q10 Loading…
2
FBGEMM kernel changes cla signed fb-exported
#3496 opened Dec 11, 2024 by csmiler Loading…
2
migrate "jagged_flash_attention" cla signed fb-exported
#3490 opened Dec 10, 2024 by brad-mengchi Loading…
5
sparse_ema_options (fbgemm) cla signed fb-exported
#3473 opened Dec 6, 2024 by spcyppt Loading…
6
Remove numpy requirement cla signed fb-exported
#3470 opened Dec 6, 2024 by spcyppt Loading…
2
Optimzed backward pass for ROCm devices (#3367) ciflow/rocm cla signed fb-exported module: rocm
#3468 opened Dec 6, 2024 by q10 Loading…
9
TEST Pin torch to 2.6.0.dev20241125 cla signed
#3452 opened Dec 4, 2024 by atalman Loading…
3
Use GEMM kernel for KleidiAI to accelerate FP16Benchmark cla signed
#3440 opened Dec 3, 2024 by milpuz01 Loading…
1
Make check_feature_gate_key PT2 compatible cla signed fb-exported
#3426 opened Nov 30, 2024 by sryap Loading…
2
Make check_feature_gate_key PT2 compatible cla signed fb-exported
#3425 opened Nov 30, 2024 by sryap Loading…
2
Add NEON and SVE implementations for Float16 conversions cla signed
#3424 opened Nov 28, 2024 by annop-w Loading…
3
Fix CUDA-11.4 build issue cla signed fb-exported
#3422 opened Nov 27, 2024 by malfet Loading…
2
Support sending using lengths to TBE instead of just offsets (#2557) cla signed fb-exported
#3420 opened Nov 26, 2024 by PaulZhang12 Loading…
2
[fbgemm_gpu] Build GenAI for ROCm in OSS [WIP] ciflow/rocm cla signed module: rocm
#3415 opened Nov 25, 2024 by q10 Loading…
3
ProTip! Adding no:label will show everything without a label.