创建

登录 ReadmeX

或

了解我们如何处理你的信息: 隐私政策

ReadmeX

陪你读懂 AI 动态

今天,想先了解什么?

选一份动态总结,或直接聊聊你的问题。

社区和关注动态总结只属于你自己。

资讯

RadixArk发布DaisyRose-27B智能体模型

AI 总结

RadixArk发布了首个模型DaisyRose-27B,该模型基于Qwen3.8-27B构建,并使用Miles进行强化学习后训练。RadixArk称,模型使用1200个精选的长程智能体任务,在144块H200 GPU上训练18天,最长会话达到100万token。厂商报告称,Terminal-Bench 4.1 pass@1从5.6%提升至12.0%,并在其他推理、指令遵循和编程基准上取得提升;Apache 2.0许可的BF16、FP8和NVFP4检查点已发布。

深度见解

此次发布提供开放权重检查点,并详细介绍长程智能体后训练方法,同时开源了可用于进一步实验的Miles强化学习框架。

Image 1: DaisyRose-27B: a daisy and a rose carrying the model name.

Today we are releasing DaisyRose-27B, RadixArk’s first model, built on Qwen3.8-27B and trained with reinforcement learning usingMiles. We trained it on long-horizon agent tasks with sessions up to 1M tokens, with just 1,200 curated tasks in 18 days on 144 H200 GPUs. DaisyRose-27B improves Terminal-Bench 4.1 pass@1 from 5.6% to 12.0%. Beyond Terminal-Bench 4.1, it also delivers gains on a variety of other reasoning, instruction following, and coding benchmarks.

We are releasingBF16,FP8, andNVFP4checkpoints for everyone to try and build on. In this post, we explain how carefully designed tasks and efficient infrastructure made these gains possible on a limited compute budget.

Image 2: DaisyRose-27B compared with eight models across seven benchmarks. DaisyRose scores 12.0 on Terminal Bench 4.1, 75.5 on Terminal Bench 2.1, 46.2 on DeepSWE v1.1, 94.7 on SWE Bench Pro v2, 34.3 on HLE (text only), 82.4 on IFBench, and 90.7 on GPQA Diamond.

#RadixArk#DaisyRose-27B#Qwen3.8-27B#Miles#Eigent AI#AgentENV#SGLang

原文约 5 分钟读完阅读原文
群聊

还没有评论,来说说你的看法。