Use default device obtained from architecture #165
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
PicoCentauri
changed the title
PicoCentauri
force-pushed
the
default-device
branch
3 times, most recently
from
bd30dd9
to
9d23279
Compare
Comment on lines
69
to
74
| if desired_device == "multi-cuda" and torch.cuda.device_count() < 1: | ||
| raise ValueError( | ||
| "Requested device 'multi-gpu' or 'multi-cuda', but found only one CUDA " | ||
| "device. If you want to run on a single GPU, please use 'gpu' or " | ||
| "'cuda' instead." | ||
| ) |
There was a problem hiding this comment.
< 2. Good catch.
|
|
||
|
|
||
| def default_device(_root_: BaseContainer) -> str: | ||
| """Custom OmegaConf resolver to find the default precision of an architecture. |
There was a problem hiding this comment.
Default device (not dtype), I also think this is not doing what the docstring says. This is picking the device for training, not finding the default device of the architecture
There was a problem hiding this comment.
Sorry, I just copied the docstring from the other function and forgot to change. Should be changed in the latest version.
|
Otherwise, I've tested the code a bit and it works |
PicoCentauri
force-pushed
the
default-device
branch
from
9d23279
to
28121aa
Compare
frostedoyster
approved these changes
Apr 11, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Similar to #155 within this PR the device is dynamically selected based on the architecture and the available devices on the system. This only applied if the user did not specify a device manually in the
options.yaml.To achieve this, I slightly changed the API of
pick_devices. Therequested_deviceparameter is now optional and renamed todesired_device. If this parameter is not given the device is solely picked based on the list ofarchitecture_devicesand devices available on the current system. I also made theget_available_devicesfunction private because it is only of usage within thepick_devicesfunction.📚 Documentation preview 📚: https://metatensor-models--165.org.readthedocs.build/en/165/