---
title: LLaDA-8B-Base
canonical_url: "https://www.modelscope.cn/models/GSAI-ML/LLaDA-8B-Base"
md_url: "https://www.modelscope.cn/models/GSAI-ML/LLaDA-8B-Base.md"
repository: GSAI-ML/LLaDA-8B-Base
last_updated: 2025-10-22
license: mit
pipeline_tag: text-generation
tasks:
  - text-generation
model_type:
  - llada
architectures:
  - LLaDAModelLM
parameters: 8.0B
tensor_type:
  - BF16
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 1333
stars: 0
---

# LLaDA-8B-Base

> LLaDA-8B-Base - GSAI-ML 在 ModelScope 开源的模型。We introduce LLaDA, a diffusion model with an unprecedented 8B scale, trained entirely from scratch, rivaling LLaMA3 8B in performance.

GSAI-ML/LLaDA-8B-Base 是 ModelScope 魔搭社区上的 8.0B 参数text-generation模型，采用 mit 许可。

- **Repository**: GSAI-ML/LLaDA-8B-Base
- **License**: mit
- **Tasks**: text-generation
- **Parameters**: 8.0B
- **Downloads**: 1333
- **Stars**: 0
- **Last updated**: 2025-10-22

Source: https://www.modelscope.cn/models/GSAI-ML/LLaDA-8B-Base

---

# LLaDA-8B-Base

We introduce LLaDA, a diffusion model with an unprecedented 8B scale, trained entirely from scratch, rivaling LLaMA3 8B in performance.

[Project Page](https://ml-gsai.github.io/LLaDA-demo/)

[Code](https://github.com/ML-GSAI/LLaDA)

## Updates
[2025-10-21] We have modified modeling_llada.py to support the input of attention_mask.