File size: 8,488 Bytes
9826ba0 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 |
---
license: mit
pipeline_tag: text-generation
library_name: transformers
---
# Stable-DiffCoder-8B-Base
<div align="left" style="line-height: 1;">
<a href="https://bytedance-stable-diffcoder.github.io/" target="_blank" style="margin: 2px;">
<img alt="Homepage" src="https://img.shields.io/badge/Seed--Coder-Homepage-a468fe?color=a468fe&logoColor=white" style="display: inline-block; vertical-align: middle;"/>
</a>
<a href="https://arxiv.org/abs/xxx" target="_blank" style="margin: 2px;">
<img alt="Technical Report" src="https://img.shields.io/badge/arXiv-Technical%20Report-brightgreen?logo=arxiv&logoColor=white" style="display: inline-block; vertical-align: middle;"/>
</a>
<a href="https://huggingface.co/ByteDance-Seed" target="_blank" style="margin: 2px;">
<img alt="Hugging Face" src="https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-ByteDance%20Seed-536af5?color=536af5&logoColor=white" style="display: inline-block; vertical-align: middle;"/>
</a>
<a href="https://github.com/ByteDance-Seed/Stable-DiffCoder/blob/master/LICENSE" style="margin: 2px;">
<img alt="License" src="https://img.shields.io/badge/License-MIT-f5de53?color=f5de53&logoColor=white" style="display: inline-block; vertical-align: middle;"/>
</a>
</div>
## Introduction
We are thrilled to introduce Stable-DiffCoder, which is a strong code diffusion large language model. Built directly on the Seed-Coder architecture, data, and training pipeline, it introduces a block diffusion continual pretraining (CPT) stage with a tailored warmup and block-wise clipped noise schedule.
Under identical architecture and data settings, we systematically analyze and design an efficient diffusion training pipeline that is not only stable but also potentially lifts the model’s performance ceiling. With this recipe, Stable-DiffCoder demonstrates overall performance improvements compared to its autoregressive (AR) counterpart across a broad set of code benchmarks, while any-order modeling improves structured code handling for editing and reasoning, and diffusion-based corruption aids learning for low-resource programming languages.
Notably, with only CPT followed by supervised fine-tuning, Stable-DiffCoder further surpasses many strong ∼8B AR and diffusion-based code models. These results demonstrate that diffusion-based training can improve code modeling quality beyond what AR training alone can achieve, even under tightly controlled data and architecture constraints.
<p align="center">
<img width="100%" src="imgs/intro_performance.png">
</p>
This repo contains the **Stable-DiffCoder-8B-Base** model, which has the following features:
- Type: Mask Diffusion Language Models
- Training Stage: Pretraining
- Data Source: GitHub data, code-related web data
- Training Tokens: 1.3 trillion
- Supports: Code completion, code infilling (Fill-in-the-Middle)
- Context Length: 8192
## Model Downloads
| Model Name | Length | Download | Notes |
|---------------------------------------------------------|--------|------------------------------------|-----------------------|
| 👉 **Stable-DiffCoder-8B-Base** | 8K | 🤗 [Model](https://huggingface.co/ByteDance-Seed/Stable-DiffCoder-8B-Base) | Pretrained on our model-centric code data. |
| Stable-DiffCoder-8B-Instruct | 8K | 🤗 [Model](https://huggingface.co/ByteDance-Seed/Stable-DiffCoder-8B-Instruct) | Instruction-tuned for alignment with user intent. |
## Requirements
You need to install a specific version of `transformers` to adapt to the inference process:
```bash
pip install transformers==4.46.2
```
## Explanation of Inference Parameters
- `steps`: Number of steps for diffusion generation
- `gen_length`: Maximum length of the generated output
- `block_length`: Length of the diffusion block, with a default value of 4
- `temperature`: Temperature for generation, with a default value of 0.0
- `remasking`: Remasking strategy, optional values are 'low_confidence' or 'random', default value is 'low_confidence' (for principle, refer to [LLADA](https://github.com/ML-GSAI/LLaDA))
- `tokenizer`: Tokenizer used for text encoding and decoding
- `shift`: Whether to shift the output to the right by one position (similar to AutoRegressive/AR), default value is False
- `threshold`: Threshold for decoding (range: 0-1.0), default value is None; a smaller value results in faster decoding speed (for principle, refer to [Fast-DLLM](https://github.com/NVlabs/Fast-dLLM))
- `eos_id`: ID of the end-of-sequence token, default value is `tokenizer.eos_token_id`
## Quickstart
Here is a simple example demonstrating how to load the model and generate code.
```python
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
device = 'cuda'
model = AutoModelForCausalLM.from_pretrained('Stable_DiffCoder-8B-Base', trust_remote_code=True, torch_dtype=torch.bfloat16).to(device).eval()
tokenizer = AutoTokenizer.from_pretrained('Stable_DiffCoder-8B-Base', trust_remote_code=True)
prompt = 'Write a quick sort algorithm.'
input_ids = tokenizer(prompt)['input_ids']
input_ids = torch.tensor(input_ids).to(device).unsqueeze(0)
out = model.generate(input_ids, steps=128, gen_length=128, block_length=4, temperature=0., remasking='low_confidence', tokenizer=tokenizer, shift=False, threshold=None, eos_id=tokenizer.eos_token_id)
print(tokenizer.decode(out[0][input_ids.shape[1]:], skip_special_tokens=True))
```
## Fill-in-the-Middle (FIM) Example
Stable-DiffCoder-8B-Base natively supports Fill-in-the-Middle (FIM) tasks, where the model is given a prefix and a suffix and asked to predict the missing middle content. This allows for code infilling scenarios such as completing a function body or inserting missing logic between two pieces of code.
```python
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
device = 'cuda'
model = AutoModelForCausalLM.from_pretrained('ByteDance-Seed/Stable-DiffCoder-8B-Base', trust_remote_code=True, torch_dtype=torch.bfloat16).to(device).eval()
tokenizer = AutoTokenizer.from_pretrained('ByteDance-Seed/Stable-DiffCoder-8B-Base', trust_remote_code=True)
prefix = "def add_numbers(a, b):\n "
suffix = "\n return result"
# Combine prefix and suffix following the FIM format
prompt = '<[fim-suffix]>' + suffix + '<[fim-prefix]>' + prefix + '<[fim-middle]>'
input_ids = tokenizer(prompt)['input_ids']
input_ids = torch.tensor(input_ids).to(device).unsqueeze(0)
out = model.generate(input_ids, steps=64, gen_length=64, block_length=4, temperature=0., remasking='low_confidence', tokenizer=tokenizer, shift=False, threshold=None, eos_id=tokenizer.eos_token_id)
print(tokenizer.decode(out[0][input_ids.shape[1]:], skip_special_tokens=True))
```
## Evaluation
Stable-DiffCoder-8B-Base has been evaluated on code generation, code completion, and code reasoning benchmarks, achieving state-of-the-art performance among ~8B open-source models.
| | DeepSeek-Coder-6.7B-Base | OpenCoder-8B-Base | Qwen2.5-Coder-7B | Seed-Coder-8B-Base | Stable-DiffCoder-8B-Base |
|------------|:------------------------:|:-----------------:|:----------------:|:------------------:|:------------------------:|
| HumanEval | 47.6 | 66.5 | 72.0 | 77.4 | **79.3** |
| MBPP | 70.2 | 79.9 | 79.4 | 82.0 | **83.6** |
| MultiPL-E | 44.7 | 61.0 | 58.8 | 67.6 | **71.2** |
| CRUXEval-O | 41.0 | 43.9 | 56.0 | 54.8 | **60.0** |
For detailed benchmark performance, please refer to our [📑 Technical Report](https://github.com/ByteDance-Seed/Stable-DiffCoder/blob/master/Stable-DiffCoder.pdf).
## License
This project is licensed under the MIT License. See the [LICENSE file](https://github.com/ByteDance-Seed/Stable-DiffCoder/blob/master/LICENSE) for details.
## Citation
If you find our work helpful, feel free to give us a cite.
```
@misc{seed2026stablediffcoder,
title={{Stable-DiffCoder}: Pushing the Frontier of Code Diffusion Large Language Model},
author={{ByteDance Seed} and Chenghao Fan and Wen Heng and Bo Li and Sichen Liu and Yuxuan Song and Jing Su and Xiaoye Qu and Kai Shen and Wei Wei},
year={2026}
}
``` |