---
license: apache-2.0
library_name: transformers
pipeline_tag: text-generation
base_model:
- Qwen/Qwen2.5-Coder-7B-Instruct
datasets:
- TIGER-Lab/FIM-Midtraining-400K
tags:
- code
- software-engineering
- fim
---

# FIM-Mid-7B

[📄 Paper](https://arxiv.org/abs/2607.12463) · [💻 GitHub](https://github.com/TIGER-AI-Lab/FIM-Midtraining) · [🤗 Dataset](https://huggingface.co/datasets/TIGER-Lab/FIM-Midtraining-400K) · [🤗 Collection](https://huggingface.co/collections/TIGER-Lab/fim-midtraining)

**FIM-Mid-7B** is the mid-trained checkpoint of the FIM 7B pipeline: `Qwen2.5-Coder-7B-Instruct` after function-aware FIM mid-training, **before** agent post-training. Post-training this checkpoint on R2E-Gym agent trajectories produces [TIGER-Lab/FIM-7B](https://huggingface.co/TIGER-Lab/FIM-7B).

It is released for reproducibility and further post-training. The paper deliberately never scores mid-training-only checkpoints — a FIM-only model has degraded instruction-following and cannot be compared fairly against instruction-tuned baselines; every reported gain is one that *survives* post-training.

## Training

- **Base model**: [`Qwen/Qwen2.5-Coder-7B-Instruct`](https://huggingface.co/Qwen/Qwen2.5-Coder-7B-Instruct)
- **FIM mid-training**: [`midtraining/configs/fim_midtrain.yaml`](https://github.com/TIGER-AI-Lab/FIM-Midtraining/blob/main/midtraining/configs/fim_midtrain.yaml) on [TIGER-Lab/FIM-Midtraining-400K](https://huggingface.co/datasets/TIGER-Lab/FIM-Midtraining-400K) — AdamW, lr `1.0e-5`, cosine schedule, warmup ratio `0.1`, weight decay `0.05`, one epoch, sequence length `32768`, bf16 (as-run copy: [`FIM_Midtrain_7B.yaml`](https://github.com/TIGER-AI-Lab/FIM-Midtraining/blob/main/midtraining/configs/FIM_Midtrain_7B.yaml))
- **Post-training**: none — see [TIGER-Lab/FIM-7B](https://huggingface.co/TIGER-Lab/FIM-7B) for the post-trained agent model

## Serve with vLLM

A standard Qwen2.5 checkpoint; no overrides needed at its native 32768 context:

```bash
CUDA_VISIBLE_DEVICES=0 \
python -m vllm.entrypoints.openai.api_server \
  --model TIGER-Lab/FIM-Mid-7B \
  --served-model-name FIM-Mid-7B \
  --host 127.0.0.1 \
  --port 8400 \
  --tensor-parallel-size 1 \
  --max-model-len 32768 \
  --gpu-memory-utilization 0.9 \
  > vllm_fim_mid7b.log 2>&1 &
```

## Post-training

To reproduce FIM-7B, run R2E-Gym trajectory SFT from this checkpoint — the exact config is [`posttraining/r2egym/FIM_Posttrain_7B.yaml`](https://github.com/TIGER-AI-Lab/FIM-Midtraining/blob/main/posttraining/r2egym/FIM_Posttrain_7B.yaml) (LLaMA-Factory, full fine-tuning, lr `1.0e-5`, 2 epochs, cutoff 32768), which already points at this repo id. See [`posttraining/r2egym/`](https://github.com/TIGER-AI-Lab/FIM-Midtraining/tree/main/posttraining/r2egym) for the walkthrough.

## Citation

```bibtex
@article{wang2026fim,
  title={Function-Aware Fill-in-the-Middle as Mid-Training for Coding Agent Foundation Models},
  author={Wang, Yubo and Liang, Jiarong and Zhang, Yuxuan and Liu, Xuye and Wei, Cong and Zhang, Yuyu and Nie, Ping and Chen, Wenhu},
  journal={arXiv preprint arXiv:2607.12463},
  year={2026}
}
```