dkeviv/Gemma-4-E2B-it-qat-mobile-ONNX-text-only-q2f16-bucket

Gemma 4 E2B IT q2f16 text-only ONNX

This bucket contains the Gemma 4 E2B instruction-tuned q2f16 decoder and token-embedding ONNX graphs for browser text generation. Vision and audio encoder graphs are intentionally omitted.

Source: onnx-community/gemma-4-E2B-it-qat-mobile-ONNX, Apache-2.0. The package is intended for Transformers.js with device: "webgpu", dtype: "q2f16", and onnxruntime-web 1.29.0 or newer. The chat template is embedded in tokenizer_config.json for browser loading.