feat: Add native Gemma3 support for ONNX export #92
+17
−0
Conversation
This file contains hidden or 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
- Add Gemma3OnnxConfig class with proper configuration - Register gemma3 model type for text generation and classification tasks - Add Gemma3 to supported architectures and test mappings - Set minimum transformers version requirement to 4.50.0 - Follow same pattern as existing Gemma/Gemma2 implementations Fixes: ValueError when exporting Gemma3 models to ONNX format Resolves: 'gemma3 model, that is a custom or unsupported architecture' error
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.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.
Add Gemma3 VLM Support for ONNX Export
🎯 Overview
This PR adds comprehensive support for Gemma3 Vision Language Model (VLM) variants in the ONNX export functionality, enabling users to export both Gemma2 base models and PaliGemma VLM models to ONNX format.
1. Gemma2 Base Model Support
gemma2models.gemma2.Gemma2OnnxConfigdefault- Basic inferencedefault-with-past- Optimized inference with KV cachecausal-lm- Causal language modelingcausal-lm-with-past- Optimized causal LM with KV cachesequence-classification- Text classificationtoken-classification- Token-level classification2. PaliGemma VLM Support
paligemmamodels.paligemma.PaliGemmaOnnxConfigdefault- Basic inferencevision2seq-lm- Vision-to-sequence language modeling