EnvScaler-Qwen3-8B is a tool-enhanced language model based on Qwen3-8B (Thinking Mode), trained using the
EnvScaler framework for tool-interactive agent tasks. This model has been trained through
Supervised Fine-Tuning (SFT) followed by
Reinforcement Learning (RL).
The training process enables the model to learn from both demonstration trajectories (SFT) and reinforcement signals (RL), resulting in improved performance on complex tool-interactive tasks.
1from transformers import AutoModelForCausalLM, AutoTokenizer
2
3model_name = "XXHStudyHard/EnvScaler-Qwen3-8B"
4tokenizer = AutoTokenizer.from_pretrained(model_name)
5model = AutoModelForCausalLM.from_pretrained(model_name)
6
7# Use with function calling interface
8# See EnvScaler project for full interaction examples
For full integration with tool-interactive environments, please refer to the
EnvScaler project documentation.
1@article{song2026envscaler,
2 title={EnvScaler: Scaling Tool-Interactive Environments for LLM Agent via Programmatic Synthesis},
3 author={Song, Xiaoshuai and Chang, Haofei and Dong, Guanting and Zhu, Yutao and Dou, Zhicheng and Wen, Ji-Rong},
4 journal={arXiv preprint arXiv:2601.05808},
5 year={2026}
6}
This model is licensed under the Apache 2.0 License, following the base Qwen3 model license.
For any questions or feedback, please contact:
songxiaoshuai@ruc.edu.cn
