---
pipeline_tag: text-generation
license: apache-2.0
base_model: Qwen/Qwen3-8B
library_name: zeromodels
language:
- en
tags:
- keras
- zeromodels
- qwen3
- qwen3-dense
- text-generation
- pytorch
- jax
- tf
---

# Run Qwen3 with Keras 3: JAX, PyTorch, or TensorFlow

[![GitHub](https://img.shields.io/badge/GitHub-ZeroModels-181717?logo=github)](https://github.com/IMvision12/ZeroModels) [![Docs](https://img.shields.io/badge/Docs-Qwen3-1f6feb)](https://imvision12.github.io/ZeroModels/qwen3/) [![HuggingFace](https://img.shields.io/badge/HuggingFace-Qwen3-ffd21e?logo=huggingface&logoColor=black)](https://huggingface.co/collections/zeromodels/qwen3-6a8eae26d6249f084e68731c)

# zeromodels/qwen3-8b

Pure-**Keras 3** conversion of [`Qwen/Qwen3-8B`](https://huggingface.co/Qwen/Qwen3-8B) for [zeromodels](https://github.com/IMvision12/ZeroModels). One implementation runs unmodified on **TensorFlow / Torch / JAX**. This is a **dense** Qwen3; weights are stored in **bfloat16**.

For model details, license, and usage terms, see the upstream [model card](https://huggingface.co/Qwen/Qwen3-8B).

Paper: [Qwen3 Technical Report (arXiv:2505.09388)](https://arxiv.org/abs/2505.09388) · [HF Papers](https://huggingface.co/papers/2505.09388)

Paper: [YaRN: Efficient Context Window Extension of Large Language Models (arXiv:2309.00071)](https://arxiv.org/abs/2309.00071) · [HF Papers](https://huggingface.co/papers/2309.00071)

## ✨ Quick start

```python
import os
os.environ["KERAS_BACKEND"] = "torch"  # or "jax" / "tensorflow"

from zeromodels.models.qwen3 import Qwen3TextGenerate, Qwen3Tokenizer

model = Qwen3TextGenerate.from_weights("zeromodels/qwen3-8b")
tokenizer = Qwen3Tokenizer.from_weights("zeromodels/qwen3-8b")

inputs = tokenizer("Give me a short introduction to large language models.")
outputs = model.generate(**inputs, max_new_tokens=128)
print(tokenizer.decode(outputs[0]))
```

Load any Qwen3 variant the same way with `from_weights("zeromodels/<variant>")`:

| Variant | Hub |
| --- | --- |
| `qwen3-0.6b` | [zeromodels/qwen3-0.6b](https://huggingface.co/zeromodels/qwen3-0.6b) |
| `qwen3-0.6b-base` | [zeromodels/qwen3-0.6b-base](https://huggingface.co/zeromodels/qwen3-0.6b-base) |
| `qwen3-1.7b` | [zeromodels/qwen3-1.7b](https://huggingface.co/zeromodels/qwen3-1.7b) |
| `qwen3-1.7b-base` | [zeromodels/qwen3-1.7b-base](https://huggingface.co/zeromodels/qwen3-1.7b-base) |
| `qwen3-4b` | [zeromodels/qwen3-4b](https://huggingface.co/zeromodels/qwen3-4b) |
| `qwen3-4b-base` | [zeromodels/qwen3-4b-base](https://huggingface.co/zeromodels/qwen3-4b-base) |
| `qwen3-8b` | [zeromodels/qwen3-8b](https://huggingface.co/zeromodels/qwen3-8b) |
| `qwen3-8b-base` | [zeromodels/qwen3-8b-base](https://huggingface.co/zeromodels/qwen3-8b-base) |
| `qwen3-14b` | [zeromodels/qwen3-14b](https://huggingface.co/zeromodels/qwen3-14b) |
| `qwen3-14b-base` | [zeromodels/qwen3-14b-base](https://huggingface.co/zeromodels/qwen3-14b-base) |
| `qwen3-32b` | [zeromodels/qwen3-32b](https://huggingface.co/zeromodels/qwen3-32b) |
| `qwen3-30b-a3b` | [zeromodels/qwen3-30b-a3b](https://huggingface.co/zeromodels/qwen3-30b-a3b) |
| `qwen3-30b-a3b-instruct-2507` | [zeromodels/qwen3-30b-a3b-instruct-2507](https://huggingface.co/zeromodels/qwen3-30b-a3b-instruct-2507) |

## Special Thanks

A huge thank you to the Qwen team at Alibaba for creating and releasing these models.

License: Apache 2.0.