File size: 3,462 Bytes
080da24 1bf1a13 9d59c93 080da24 9d59c93 c0cfb4c 9d59c93 c0cfb4c 9d59c93 c0cfb4c 08950dc c0cfb4c 08950dc 9d59c93 08950dc 9d59c93 08950dc 9d59c93 08950dc 9d59c93 08950dc 9d59c93 08950dc | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 | ---
license: mit
library_name: transformers
pipeline_tag: text-generation
tags:
- long-context
- reinforcement-learning
- rlvr
- grpo
- multitask
---
# ✨ GoLongRL-4B
We present **GoLongRL**, a fully open-source, capability-oriented post-training recipe for long-context reinforcement learning with verifiable rewards (RLVR).
| Resource | Link |
|---|---|
| 📝 Paper | [ArXiv (2605.19577)](https://arxiv.org/abs/2605.19577) |
| 🤗 HF Paper | [GoLongRL](https://huggingface.co/papers/2605.19577) |
| 💻 Code | [GitHub Repository](https://github.com/xiaoxuanNLP/GoLongRL) |
| 📂 Collection | [GoLongRL Collection](https://huggingface.co/collections/Kwai-Klear/golongrl) |
| 📊 Dataset | [GoLongRL Dataset](https://huggingface.co/datasets/Kwai-Klear/GoLongRL) |
| 📧 Contact | xiao_xuan_zi_666@163.com & suzhenpeng13@163.com |
## 📌 Overview
<div align="center">
<img src="main_result.png" width="100%"/>
<sub>Overall performance comparison on long-context benchmarks (DocMath, LongBench-V2, Frames, MRCR, CorpusQA, LBV1-QA).</sub>
</div>
**GoLongRL-4B** achieves strong long-context performance at the 4B scale. Our framework combines the following:
1. **Capability-Oriented Dataset (23K samples, 9 task types).** Guided by a taxonomy of long-context capabilities, the dataset covers precise retrieval, comprehension, exhaustive retrieval, numerical reasoning, structured extraction, structured matching, graded ranking, sequence ordering, and summarization. Each task is paired with its natural evaluation metric (EM, Accuracy, F1, math_verify, IoU, SubEM, NDCG, Pairwise, ROUGE-L) as the reward function.
2. **TMN-Reweight.** To address optimization challenges from heterogeneous rewards, we propose TMN-Reweight, which combines task-level mean normalization for cross-task reward scale alignment with difficulty-adaptive weighting for more reliable advantage estimation. It provides a modest but consistent improvement over vanilla GRPO.
3. **Full Open Release.** We publicly release the complete dataset, the four-phase construction pipeline, and all training code.
### Key Results
- Under the same vanilla GRPO setup, our dataset alone outperforms the closed-source QwenLong-L1.5 dataset.
- TMN-Reweight further improves average performance to 63.0 at 4B scale, surpassing QwenLong-L1.5 with its specialized AEPO algorithm.
- General capabilities (MMLU-Pro, AIME24/25, GPQA) are preserved or improved, with substantial gains on dialogue memory (LongMemEval +13.6) and agentic memory benchmarks.
## 🔍 Evaluation
Evaluation uses **QwenLong-Benchmarks**, covering three capability dimensions:
| Dimension | Benchmarks |
|-----------|-----------|
| **Long-Context** | LongBench-V2, MRCR (≤128K / 128K–512K / 512K–1M), Frames, LongBench QA, DocMath, CorpusQA (≤128K / ≤1M) |
| **General** | MMLU-Pro, AIME 2024/2025, GPQA-Diamond |
| **Memory** | BFCL-V4 (memory subset), LongMemEval |
## 🤝 Citation
```bibtex
@misc{lv2026golongrlcapabilityorientedlongcontext,
title={GoLongRL: Capability-Oriented Long Context Reinforcement Learning with Multitask Alignment},
author={Minxuan Lv and Tiehua Mei and Tanlong Du and Junmin Chen and Zhenpeng Su and Ziyang Chen and Ziqi Wang and Zhennan Wu and Ruotong Pan and jian Liang and Ruiming Tang and Han Li},
year={2026},
eprint={2605.19577},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2605.19577},
}
``` |