Constant fold initializer for DQ node #23366
Draft
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
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
Comment on lines
+304
to
+306
| // Dequantize initializer using ORT ConstantFolding optimizer for dq node if initializer has specific(? TBD) data type. | ||
| // This feature is required by some NPU's. | ||
| // "0": disable. ORT doesn't constant fold the DQ node. [DEFAULT] |
There was a problem hiding this comment.
Suggested change
| // Dequantize initializer using ORT ConstantFolding optimizer for dq node if initializer has specific(? TBD) data type. | |
| // This feature is required by some NPU's. | |
| // "0": disable. ORT doesn't constant fold the DQ node. [DEFAULT] | |
| // Dequantize initializer using ORT ConstantFolding optimizer for dq node if initializer has specific(? TBD) data type. | |
| // This feature is required by some NPU's. | |
| // "0": disable. ORT doesn't constant fold the DQ node. [DEFAULT] |
Comment on lines
+736
to
+738
| false /*skip_dequantize_linear*/, | ||
| false /*dequantize_initializer_for_dequantize_linear*/, | ||
| empty_config_options), |
There was a problem hiding this comment.
Suggested change
| false /*skip_dequantize_linear*/, | |
| false /*dequantize_initializer_for_dequantize_linear*/, | |
| empty_config_options), | |
| false /*skip_dequantize_linear*/, | |
| false /*dequantize_initializer_for_dequantize_linear*/, | |
| empty_config_options), |
|
In the selection of DQ nodes to constant fold, how can user/EP specify which data type of initializers to be considered? |
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Outdated suggestions cannot be applied.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.
Some NPUs require weights/initializers to be in FP32, FP16, INT8, UINT8 and INT4 if consumed by Q/DQ nodes.
In other words, ORT needs to dequantize "specific data type" initializers to FP32 for them.
This PR leverages ORT ConstantFolding optimizer to dequantize initializer for DQ node if the initializer has a specific data type.