pytorch
/
opacus
/
15000828402
80%
|
Ran
|
Jobs
3
|
Files
123
|
Run time
1min
|
Badge
Embed ▾
README BADGES
|
coverage: 85.925% (+0.1%) from 85.811%
push
github
facebook-github-botAdd norm_grad_sample for RMSNorm (#755) Summary: Pull Request resolved: https://github.com/pytorch/opacus/pull/755 Llama3 model has RMSNorm and currently, we use functorch to support FGC for RMSNorm. This causes FSDP to rely on the root node for all_gather call of the RMSNorm layers. Adding norm_grad_sample method for RMSNorm to support layer-wise FSDP for this layer. Reviewed By: HuanyuZhang Differential Revision: D74334633 fbshipit-source-id: 399e98422
50 of 50 new or added lines in 3 files covered. (100.0%)
5360 of 6238 relevant lines covered (85.92%)
1.91 hits per line
Jobs
| ID | Job ID | Ran | Files | Coverage | |
|---|---|---|---|---|---|
| 1 | run-3 - 15000828402.1 | 67 |
48.53 |
GitHub Action Run | |
| 2 | run-1 - 15000828402.2 | 122 |
85.77 |
GitHub Action Run | |
| 3 | run-2 - 15000828402.3 | 122 |
85.76 |
GitHub Action Run |
