You can deploy this model and start earning money today!

Fate-Zero/
Archer-Code-1.5B

reinforcement-learning

llm

reasoning

code

Views

No views yet

✨ Archer

🏹️ Reinforcement Learning for Enhanced Reasoning in LLMs 🎯

Overview

The Archer series focuses on research into RL algorithms and training for medium and small-scale models, aiming to deepen the community's understanding of the fundamental principles of reinforcement learning (RL) on large language models (LLMs). All released content will be comprehensively open-sourced to advance community research development.

_{Archer significantly improves the reasoning performance upon DAPO and outperforms previous 1.5B-level SOTA reasoning models.}

Archer is an open-source initiative enhancing reasoning in large language models through scalable, rule-governed reinforcement learning. We provide full-stack reproducibility including:

Training code and pipelines
Curated datasets
Trained models
Complete training logs

Current Models:

Archer-Code-1.5B - SOTA among similarly-sized models.

Evaluation

We conduct evaluation on both mathematical and coding benchmarks. Due to the high variance of the outputs from reasoning models, we report avg@K (pass@1 performance averaged over K outputs) and pass@K for each benchmark. The detailed results are shown in the table below.

Method	AIME24	AIME25	AMC23	MATH-500	Minerva	Olympiad	Avg.
avg@64	pass@64	avg@64	pass@64	avg@64	pass@64	avg@4	pass@4	avg@8	pass@8	avg@4	pass@4
DeepSeek-R1-1.5B	30.6	80.0	23.5	63.3	70.7	100.0	83.6	92.4	27.6	48.2	44.6	59.4	46.8
DAPO	42.1	80.0	28.6	56.7	80.3	97.5	87.6	94.6	29.2	46.3	53.2	65.8	53.5
DeepScaleR-1.5B	42.0	83.3	29.0	63.3	81.3	100.0	87.7	93.6	30.3	51.1	50.7	61.0	53.5
FastCuRL-1.5B-V3	48.1	80.0	32.7	60.0	86.4	95.0	89.8	94.0	33.6	50.0	55.3	64.3	57.7
Nemotron-1.5B	48.0	76.7	33.1	60.0	86.1	97.5	90.6	93.6	35.3	47.8	59.2	66.8	58.7
Archer-Math-1.5B	48.7	83.3	33.8	70.0	86.0	97.5	90.8	94.4	35.7	51.1	59.3	67.1	59.1

Method	LCB v5 (2024.08.01–2025.02.01)	LCB v6 (2025.02.01–2025.05.01)	Avg.
avg@8	pass@8	avg@16	pass@16
DeepSeek-R1-1.5B	16.7	29.0	17.2	34.4	17.0
DAPO	26.0	40.5	27.6	43.5	26.8
DeepCoder-1.5B	23.3	39.1	22.6	42.0	23.0
Nemotron-1.5B	26.1	35.5	29.5	42.8	27.8
Archer-Code-1.5B	29.4	43.7	30.2	45.8	29.8

Technical Report

Stabilizing Knowledge, Promoting Reasoning: Dual-Token Constraints for RLVR

Acknowledgements

We build our model upon DeepSeek-R1-Distill-Qwen-1.5B.
Training was carried out with a modified version of verl.

Citation

Please cite the following:

bibtex

@article{wang2025stabilizing,
  title={Stabilizing Knowledge, Promoting Reasoning: Dual-Token Constraints for RLVR},
  author={Wang, Jiakang and Liu, Runze and Zhang, Fuzheng and Li, Xiu and Zhou, Guorui},
  journal={arXiv preprint arXiv:2507.15778},
  year={2025}
}

Fate-Zero/Archer-Code-1.5B