metadata
license: mit
base_model: Qwen/Qwen2.5-Coder-7B-Instruct
library_name: peft
datasets:
- TIGER-Lab/StopLoss-Training-Data
language:
- en
tags:
- lora
- agents
- coding-agents
- early-stopping
StopLoss-Qwen2.5-Coder-7B-Instruct
A LoRA adapter for Qwen/Qwen2.5-Coder-7B-Instruct that predicts whether a coding agent's run will fail. It is the Qwen2.5-Coder-7B-Instruct StopLoss monitor from StopLoss: Teaching Agents to Predict Failure and Stop Early.
🌐 Homepage | 📄 Paper | 💻 Code | 🤗 Training data
What it does
At tool steps 5, 8, 11, ..., the monitor reads the step count and the three most recent action–observation pairs of
a run. Its failure score is the probability of ABORT among the labels CONTINUE, DELIVER and ABORT, read from
the next-token logits in one forward pass. A run is stopped once the score reaches a threshold calibrated on
successful runs.
Usage
git clone https://github.com/TIGER-AI-Lab/StopLoss.git && cd StopLoss
pip install -e ".[model]"
python scripts/build_windows.py --runs my_runs.jsonl --out windows.jsonl --ks checkpoints
python scripts/score_windows.py --base Qwen/Qwen2.5-Coder-7B-Instruct \
--adapter TIGER-Lab/StopLoss-Qwen2.5-Coder-7B-Instruct \
--windows windows.jsonl --out scores.jsonl
This adapter is seed 0 of its backbone; the results in the paper use an ensemble of four seeds.