amd/Gemma-3-4b-it-mm-onnx-ryzenai-npu

By amd · ↓ 0 · ♥ 2 · Updated 2026-01-21

What's in this repo

This page lists the files in amd/Gemma-3-4b-it-mm-onnx-ryzenai-npu, including gemma-3-embedding.onnx, gemma-3-vision-npu.onnx, model.onnx.

3 ONNX models (.onnx files)
15 files, including 5 .json, 3 .onnx, 2 .data, 2 .bin, 1 .md, 1 .jinja
Last updated 2026-01-21

ONNX files

File	Size	Download
gemma-3-embedding.onnx	—	Download
gemma-3-vision-npu.onnx	—	Download
model.onnx	—	Download

How to use

These ONNX files can be loaded with ONNX Runtime. To inspect a model graph, upload the file to the Netron viewer. Download links point to the original files on Hugging Face.

README

amd/gemma-3-4b

Introduction

This model was created using Quark Quantization for the Decoder, followed by OGA Model Builder, and finalized with post-processing for NPU deployment.
Quantization Strategy
- AWQ / Group 128 / Asymmetric / BFP16 activations / UINT4 weights
Base model info:
Please refer to Gemma-3-4b-it for base model info.

Evaluation scores

The MMMU scores are, Music: 33.33, Marketing: 40, Math: 30, Clinical Medicine: 20, History: 40, Electronics: 13.3.
The perplexity measurement is run on the wikitext-2-raw-v1 (raw data) dataset provided by Hugging Face. Perplexity score measured for prompt length 2k is 16.825.

License

Model info

onnx zh en base_model:google/gemma-3-4b-it base_model:quantized:google/gemma-3-4b-it license:gemma region:us

↓ 0 ♥ 2

Created: 2025-12-12

Updated: 2026-01-21

View on Hugging Face

Files (15)

.gitattributes

README.md

chat_template.jinja

gemma-3-embedding.onnx ONNX

gemma-3-embedding.onnx.data

gemma-3-vision-npu.onnx ONNX

gemma-3-vision-npu.pb.bin

genai_config.json

model.onnx ONNX

model.onnx.data

model.pb.bin

processor_config.json

special_tokens_map.json

tokenizer.json

tokenizer_config.json

Related models

More models

No task type declared.