Latest

Training on internalized skill workflows lets a 35B open-weights model beat far larger proprietary models on agentic benchmarks, shifting the edge from raw model scale to curated environment data. Builders running Claude Code or Codex harnesses can reuse the released environments and trajectories for their own SFT or RL pipelines. The results give Qwen-based deployments a concrete benchmark point against Claude Sonnet 4.6 and DeepSeek V4 Pro.

#SkillGym #Qwen3.5 #agents #reinforcement learning

Event

arXiv: DeepSeek@Zhilong Ge et al. (9)6 sourcesSource

Sep 24

Papers BuildersMulti-sourceSelected09:35

SkillGym Turns Human Agent Skills into Training Environments

The key finding is that skills can be internalized as model weights rather than injected at inference time, since the fine-tuned agent beats skill-assisted base models even without the skills provided. Agent engineers and fine-tuning practitioners can reuse the released environments and trajectories for SFT or RL with outcome-based rewards. The result also shows a 35B open Qwen-based model outscoring much larger closed models on SkillsBench, a useful cost comparison for teams building coding and tool-use agents.

#SkillGym #Qwen3.5 #agents #fine-tuning

Event

arXiv: DeepSeek@Zhilong Ge et al. (9)6 sourcesSource

Papers US vs China09:35

Chinese contentMWE-ECL 诊断基准：可检索的远距离上下文未必能改变模型的局部词 prior

这篇论文造了个诊断集，专门测长上下文模型“找到了但没用上”的问题，DeepSeek 两个模型都被测出检索满分但解读掉分，做长文本评测的可以看看。

#MWE-ECL #DeepSeek #长上下文 #评测基准

arXiv: DeepSeek@Wei He et al. (4)Source

Sep 17

AITOP16:09

Anthropic拟10月在新加坡设亚太第五个办事处开...

Anthropic拟10月在新加坡设亚太第五个办事处开放本地招聘数据显示，按人口比例计算，新加坡的Claude.ai使用水平在121个国家中排名第二。我们观察到，人工智能公司Anthropic计划于10月在新加坡设立其亚太地区的第五…

AITOPRead

Sep 14

Papers BuildersSelected12:57

RoofLang: Enabling AI-Driven Architecting of LLM Inference Systems

This matters because it introduces a new tool, RoofLang, that could significantly improve the performance of LLMs like DeepSeek V4-series by optimizing their inference architecture, potentially leading to faster processing and better user experience.

#RoofLang #LLM inference #DeepSeek V4 #AI optimization

Event

arXiv: DeepSeek@Ziyue Yang et al. (4)2 sourcesSource

Sep 3

PapersSelected73°11:50

Trace as State for Long-Context Transformers

The method significantly boosts performance on long-context reasoning tasks without changing transformer architecture, benefiting developers working with complex document analysis.

#Trace as State #DeepSeek V4 Pro Preview #GLM-5.2 #long-context reasoning

arXiv: DeepSeek@Xu Zou, Jie TangSource

Aug 21

Papers Policy10:35

Chinese contentReguSim：评估金融合规中的LLM智能体规则归因

这篇论文提出了一个评估LLM智能体在金融合规中规则归因的新方法，值得一读。它通过实验展示了规则指导如何减少违规行为，并强调了监控的重要性。

#ReguSim #金融合规 #LLM智能体 #规则归因

arXiv: DeepSeek@Yiyang Luo et al. (7)Source

Aug 14

PapersMulti-source74°12:27

Chinese contentAgent驱动设计硬件预取器MoP，性能超人类方案

这篇论文让AI Agent自己找预取器失败原因并造新方案，结果比人类设计的Alecto和Berti还强，值得看看。