-
-
Notifications
You must be signed in to change notification settings - Fork 15k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
python3Packages.{torch,torch-bin}: 2.0.1 -> 2.1.1 #259068
Conversation
Binaries are missing for |
Thanks for putting this together so quickly @GaetanLepage ! |
ddb3fcb
to
0e89147
Compare
Good timing, I'm almost done testing #258328. |
On macos 10.15 this appears to break |
Almost forgot about this. |
Awesome! |
a44cf7a
to
9a6cb67
Compare
@GaetanLepage I'm still unable to build PyTorch or the test I use to verify it (https://github.com/connorbaker/nix-cuda-test) nix build github:ConnorBaker/nix-cuda-test#nix-cuda-test --override-input nixpkgs "github:GaetanLepage/nixpkgs/8672a90ccb487e1a6ca4e07314610e9aa8f22fca" fails with (full log: https://gist.github.com/ConnorBaker/b5ee6bbcc2c078fd13ba285e13c59f6b) CMake Error at /nix/store/vnhl4zdy7igx9gd3q1d548vwzz15a9ma-cmake-3.27.7/share/cmake-3.27/Modules/FindPackageHandleStandardArgs.cmake:230 (message):
Could NOT find CUDAToolkit (missing: CUDAToolkit_INCLUDE_DIR) (found
version "11.8.89")
Call Stack (most recent call first):
/nix/store/vnhl4zdy7igx9gd3q1d548vwzz15a9ma-cmake-3.27.7/share/cmake-3.27/Modules/FindPackageHandleStandardArgs.cmake:600 (_FPHSA_FAILURE_MESSAGE)
/nix/store/vnhl4zdy7igx9gd3q1d548vwzz15a9ma-cmake-3.27.7/share/cmake-3.27/Modules/FindCUDAToolkit.cmake:938 (find_package_handle_standard_args)
cmake/public/cuda.cmake:59 (find_package)
cmake/Dependencies.cmake:44 (include)
CMakeLists.txt:722 (include) I believe you need to add these changes to your patch: https://github.com/ConnorBaker/nixpkgs/blob/6d8c6ac313ec6327a0477238703a0f160fcec445/pkgs/development/python-modules/torch/0001-cmake-Use-cmake-provided-FindCUDAToolkit.cmake.patch#L58-L59. |
8672a90
to
edbb0e4
Compare
|
edbb0e4
to
63c6160
Compare
I'm able to build and run PyTorch to train a small model: $ nom build github:ConnorBaker/nix-cuda-test#nix-cuda-test --override-input nixpkgs "github:GaetanLepage/nixpkgs/63c61609df86dbb0dc581f80afe80f870e62d3f6" --builders ''
$ ./result/bin/nix-cuda-test
Seed set to 42
Using bfloat16 Automatic Mixed Precision (AMP)
GPU available: True (cuda), used: True
TPU available: False, using: 0 TPU cores
IPU available: False, using: 0 IPUs
HPU available: False, using: 0 HPUs
Files already downloaded and verified
Files already downloaded and verified
LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0]
| Name | Type | Params
-----------------------------------------------
0 | criterion | CrossEntropyLoss | 0
1 | model | ViT | 86.3 M
-----------------------------------------------
86.3 M Trainable params
0 Non-trainable params
86.3 M Total params
345.317 Total estimated model params size (MB)
Epoch 9: 100%|████████████████████████████████████████████████████████████████████████████████| 781/781 [01:28<00:00, 8.83it/s, v_num=29, train_loss=2.350, val_loss=2.330]`Trainer.fit` stopped: `max_epochs=10` reached.
Epoch 9: 100%|████████████████████████████████████████████████████████████████████████████████| 781/781 [01:30<00:00, 8.65it/s, v_num=29, train_loss=2.350, val_loss=2.330] Notably, @SomeoneSerge @samuela @Madouura barring any surprise developments I plan to merge this tomorrow. Is that acceptable? |
Result of 26 packages marked as broken and skipped:
30 packages failed to build:
514 packages built:
|
Fairseq is overdue for a new release, the maintainers have been contacted, you can ignore the failure. |
The PR's contents looks sane, but why is |
63c6160
to
acbfdb5
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Built python311Packages.torchWithRocm
locally.
No idea why the nixpkgs-review
build failed, but works on my machine.
LGTM.
Merging -- thank you @GaetanLepage! |
Description of changes
Update
torch
andtorch-bin
.Changelog: https://github.com/pytorch/pytorch/releases/tag/v2.1.1
cc @teh @thoughtpolice @tscholak
Things done
sandbox = true
set innix.conf
? (See Nix manual)nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD"
. Note: all changes have to be committed, also see nixpkgs-review usage./result/bin/
)