-
Notifications
You must be signed in to change notification settings - Fork 478
Insights: pytorch/FBGEMM
September 14, 2024 – September 21, 2024
Overview
-
0 Active issues
-
- 0 Merged pull requests
- 5 Open pull requests
- 0 Closed issues
- 0 New issues
Loading
Could not load contribution data
Please try again later
Loading
5 Pull requests opened by 5 people
-
Tuned fp8 gemm for LDM cases
#3142 opened
Sep 17, 2024 -
mark autograd function traceable
#3143 opened
Sep 17, 2024 -
Bfloat16 support for fbgemm::pack_segments_forward/backward_cpu
#3154 opened
Sep 19, 2024 -
use float instead of bool for should_ema and should_swap
#3158 opened
Sep 20, 2024 -
[fbgemm_gpu] Add workflow for running only on non-PyTorch infrastructure
#3159 opened
Sep 21, 2024
3 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.
-
Fp8 hipstream fix
#3127 commented on
Sep 20, 2024 • 4 new comments -
remove NOT rocm check from gen_ai compile
#3126 commented on
Sep 21, 2024 • 1 new comment -
AssertionError: Per channel weight observer is not supported yet for ConvTranspose{n}d.
#1576 commented on
Sep 20, 2024 • 0 new comments