-
Notifications
You must be signed in to change notification settings - Fork 48
Enhance ViT policy with configurable trunk and update TF32 precision settings #4148
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
Conversation
|
Warning This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
This stack of pull requests is managed by Graphite. Learn more about stacking. |
This comment has been minimized.
This comment has been minimized.
54308be to
e295b66
Compare
6da3f04 to
be8dee0
Compare
| num_layers=trunk_num_resnet_layers, | ||
| pattern="A", # Axon blocks provide residual-like connections | ||
| post_norm=trunk_use_layer_norm, |
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.
NameError: Undefined variable reference
The code references trunk_num_resnet_layers and trunk_use_layer_norm in the components list default value, but these variables are not in scope at class definition time. Class attributes defined above (lines 165, 167) cannot be referenced by name in other class attribute default values.
This will cause an immediate NameError when the class is defined:
NameError: name 'trunk_num_resnet_layers' is not definedFix: Either use hardcoded values in the default or use _latent_dim pattern with underscore prefix, or move this logic to __post_init__ or the make_policy method where instance attributes are accessible.
| num_layers=trunk_num_resnet_layers, | |
| pattern="A", # Axon blocks provide residual-like connections | |
| post_norm=trunk_use_layer_norm, | |
| num_layers=4, # Default value | |
| pattern="A", # Axon blocks provide residual-like connections | |
| post_norm=True, # Default value |
Spotted by Graphite Agent
Is this helpful? React 👍 or 👎 to let us know.
|
Looks like there are a few issues preventing this PR from being merged!
If you'd like me to help, just leave a comment, like Feel free to include any additional details that might help me get this PR into a better state. You can manage your notification settings |

No description provided.