Linear Probing #38
Open
Linear Probing #38
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
fcogidi
requested changes
Jan 14, 2025
| @@ -2,6 +2,7 @@ | |||
|
|
|||
| from mmlearn.tasks.contrastive_pretraining import ContrastivePretraining | |||
| from mmlearn.tasks.ijepa import IJEPA | |||
| from mmlearn.tasks.linear_probing import LinearClassifierModule | |||
There was a problem hiding this comment.
Suggested change
| from mmlearn.tasks.linear_probing import LinearClassifierModule | |
| from mmlearn.tasks.linear_probing import LinearClassifier |
| @@ -11,4 +12,5 @@ | |||
| "IJEPA", | |||
| "ZeroShotCrossModalRetrieval", | |||
| "ZeroShotClassification", | |||
| "LinearClassifierModule", | |||
There was a problem hiding this comment.
Suggested change
| "LinearClassifierModule", | |
| "LinearClassifier", |
|
|
||
|
|
||
| @store(group="task", provider="mmlearn") | ||
| class LinearClassifierModule(L.LightningModule): |
There was a problem hiding this comment.
Suggested change
| class LinearClassifierModule(L.LightningModule): | |
| class LinearClassifier(L.LightningModule): |
|
|
||
| def __init__( | ||
| self, | ||
| # encoder: torch.nn.Module, |
| self, | ||
| # encoder: torch.nn.Module, | ||
| encoder: nn.Module, | ||
| model_checkpoint_path: Optional[str], # change name |
There was a problem hiding this comment.
Suggested change
| model_checkpoint_path: Optional[str], # change name | |
| model_checkpoint_path: Optional[str], |
| print("Encoder state dict loaded successfully") | ||
| except Exception as e: | ||
| print(f"Error loading state dict: {e}") | ||
| return model["rgb"] |
There was a problem hiding this comment.
Please generalize this as well. That is, allow the user to specify which key, if any, to extract.
| # Ignore specific keys | ||
| if keys_to_ignore: | ||
| state_dict = { | ||
| k: v for k, v in state_dict.items() if k not in keys_to_ignore |
There was a problem hiding this comment.
This is removing the keys from the state_dict
There was a problem hiding this comment.
Is the intention here to train a single linear classifier on all the datasets at once or a single linear classifier for each dataset?
| task: multiclass | ||
| num_classes: 4 | ||
| num_output_features: 512 | ||
| hidden_dims: [256] |
There was a problem hiding this comment.
Why add hidden layers? Are you following a certain recipe?
| num_output_features: 512 | ||
| hidden_dims: [256] | ||
| modality: rgb | ||
| encoder_checkpoint_path: /path/to/checkpoint |
There was a problem hiding this comment.
You can make this a required parameter like so:
Suggested change
| encoder_checkpoint_path: /path/to/checkpoint | |
| encoder_checkpoint_path: ??? |
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.
PR Type
[Feature]
Short Description
Linear probing task