BAAI发布AREX-2:基于Qwen3.8的27B长周期Agent模型
BAAI/AREX-2 - 27B - Agent model based on Qwen3.8 27B
BAAI发布的这款27B Agent模型展示了极具价值的测试时自改进机制,且实现了从代码到深度研究的跨域能力迁移,对探索低成本长链路推理有重要参考意义。
"AREX-2 is a 27B-parameter long-horizon agent model from the Beijing Academy of Artificial Intelligence (BAAI). It learns to improve a solution over multiple test-time rounds: propose, measure, reflect, and revise.
AREX-2 是北京智源人工智能研究院(BAAI)推出的拥有 270 亿参数的长周期智能体模型。它通过学习在多个测试轮次中改进解决方案来提升能力:提出、评估、反思和修订。
AREX-2 is trained on machine-learning and algorithmic-programming tasks with verifiable feedback, together with the existing AREX deep-research data. The learned self-improvement behavior transfers to deep research without adding new search trajectories.
AREX-2 在带有可验证反馈的机器学习和算法编程任务上进行训练,并结合现有的 AREX 深度研究数据。所习得的自我改进行为能够迁移至深度研究任务,而无需新增搜索轨迹。
- Architecture: Dense Qwen3.8-compatible multimodal model
- Parameters: 27B
- Context length: 262,144 tokens
- 架构:稠密型 Qwen3.8 兼容的多模态模型
- 参数量:270 亿
- 上下文长度:262,144 tokens
Key features
关键特性
- Long-horizon self-improvement: turns extra test-time rounds into useful solution refinement.
- Feedback-driven reflection: reads scores, logs, errors, and timings to decide what to change next.
- Cross-domain performance: training on coding and machine-learning tasks also improves the model's deep-research performance.
- Long-horizon reasoning: sustains productive iteration as the task budget grows."
- 长周期自我改进:将额外的测试轮次转化为有效的方案优化。
- 反馈驱动反思:读取分数、日志、错误和耗时信息,以决定下一步的修改方向。
- 跨领域性能:在编码和机器学习任务上的训练也能提升模型的深度研究表现。
- 长周期推理:随着任务预算的增加,能够维持高效的生产性迭代。"
Gguf's - https://huggingface.co/mradermacher/AREX-2-GGUF
Gguf 格式 - https://huggingface.co/mradermacher/AREX-2-GGUF
更进一步:量化金融体系
看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力