README.md

---
license: apache-2.0
---

# Seed-Coder-8B-Base

## Introduction
**Seed-Coder-8B-Base** is an 8-billion-parameter foundation model tailored for code understanding and generation. It is designed to provide developers with a powerful, general-purpose code model capable of handling a wide range of coding tasks.  
It features:
- Pre-trained on a **massively curated corpus**, filtered using **LLM-based techniques** to ensure **high-quality real-world code**, **text-code alignment data**, and **synthetic datasets**, resulting in cleaner and more effective learning signals.
- Excels at **code completion** and supports **Fill-in-the-Middle (FIM)** tasks, enabling it to predict missing code spans given partial contexts.
- Robust performance across **various programming languages** and **code reasoning scenarios**, making it ideal for downstream finetuning or direct use in code generation systems.
- **Long-context support** up to 32K tokens, enabling it to handle large codebases, multi-file projects, and extended editing tasks.

Seed-Coder-8B-Base serves as the foundation for Seed-Coder-8B-Instruct and Seed-Coder-8B-reasoning.

## Requirements
You will need to install the latest versions of `transformers` and `accelerate`:

```bash
pip install -U transformers accelerate
```

## Quickstart

Here is a simple example demonstrating how to load the model and perform code generation using the Hugging Face `pipeline` API:

```python
import transformers
import torch

model_id = "ByteDance-Seed/Seed-Coder-8B-Base"

pipeline = transformers.pipeline(
    "text-generation",
    model=model_id,
    model_kwargs={"torch_dtype": torch.bfloat16},
    device_map="auto",
)

output = pipeline("def say_hello_world():", max_new_tokens=100)
print(output[0]["generated_text"])
```

### Fill-in-the-Middle (FIM) Example

Seed-Coder-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.

A typical usage flow:

```python
import transformers
import torch

model_id = "/mnt/bn/daoguang/ckpts/Bytedance/Doubao-Coder-base/P6Dense"

pipeline = transformers.pipeline(
    "text-generation",
    model=model_id,
    model_kwargs={"torch_dtype": torch.bfloat16},
    device_map="auto",
)

# You can concatenate a prefix, a special FIM separator token, and a suffix
prefix = "def add_numbers(a, b):\n    "
suffix = "\n    return result"

# Combine prefix and suffix following the FIM format
fim_input = '<[fim-suffix]>' + suffix + '<[fim-prefix]>' + prefix + '<[fim-middle]>'

output = pipeline(fim_input, max_new_tokens=512)
print(output[0]["generated_text"])
```

## Evaluation

Seed-Coder-8B-Base has been internally evaluated across a variety of code understanding and generation benchmarks.  
It demonstrates strong capabilities in:
- Fluent and contextually appropriate code completion.
- Reasoning about code structure and inferring missing logic.
- Generalizing across different programming languages, coding styles, and codebases.

For detailed benchmark results, please refer to our [📑 paper](https://arxiv.org/pdf/xxx.xxxxx).

## Citation

If you find Seed-Coder helpful, please consider citing our work:

```
@article{zhang2025seedcoder,
    title={Seed-Coder: Let the Code Model Curate Data for Itself},
    author={Xxx},
    year={2025},
    eprint={2504.xxxxx},
    archivePrefix={arXiv},
    primaryClass={cs.CL},
    url={https://arxiv.org/abs/xxxx.xxxxx}, 
}
```