Enable constant values as inputs to linear layers in PyTorch #1076
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
elad-c
requested changes
May 19, 2024
| @@ -42,7 +42,8 @@ def __init__(self, | |||
| reuse_group: str = None, | |||
| quantization_attr: Dict[str, Any] = None, | |||
| has_activation: bool = True, | |||
| is_custom: bool = False | |||
| is_custom: bool = False, | |||
| has_positional_weights: bool = False | |||
There was a problem hiding this comment.
This is not needed. See below
| @@ -96,6 +99,15 @@ def get_has_activation(self): | |||
| """ | |||
| return self.has_activation | |||
|
|
|||
| def get_has_positional_weights(self): | |||
There was a problem hiding this comment.
change to has_positional_weights which is a @property, that checks whether there are positional weights (i.e. check for integers in the keys of the weights dictionary).
| graph.get_in_stats_collector(n), | ||
| fw_impl=fw_impl) | ||
| if n.has_positional_weights: | ||
| for candidate_qc in n.candidates_quantization_cfg: |
| @@ -67,7 +67,8 @@ def substitute(self, | |||
| return graph | |||
|
|
|||
| # Check if convolution and residual satisfy the collapsing conditions, otherwise skip substitution | |||
| if len(graph.get_next_nodes(first_node)) > 1 or len(graph.get_prev_nodes(second_node)) != 2: | |||
| if (len(graph.get_next_nodes(first_node)) > 1 or len(graph.get_prev_nodes(first_node)) < 1 or | |||
There was a problem hiding this comment.
isn't it clearer to write (not blabla==1)?
| @@ -33,7 +33,7 @@ def node_builder(n: BaseNode) -> Module: | |||
|
|
|||
| framework_attr = copy.copy(n.framework_attr) | |||
| node_instance = n.type(**framework_attr) | |||
| node_instance.load_state_dict({k: torch.tensor(v) for k, v in n.weights.items()}, strict=False) | |||
| node_instance.load_state_dict({k: torch.tensor(v) for k, v in n.weights.items() if isinstance(k, str)}, strict=False) | |||
There was a problem hiding this comment.
Add comment to explain why only str keys are used
elad-c
approved these changes
May 19, 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.
Enable constant values as inputs to linear layers in PyTorch
Pull Request Description:
Checklist before requesting a review: