跳到主内容
@wquguru
精选70Chubby♨️产品发布/更新

DeepSeek V4 Pro 搭配J-Space框架,多项Agent基准提升

Someone has combined DeepSeek V4 Pro 0813 with a harness called J-Space, which a…

原文
发到 X

Someone has combined DeepSeek V4 Pro 0813 with a harness called J-Space, which apparently reduces errors in the extended reasoning/tool-use process.

有人将DeepSeek V4 Pro 0813与一个名为J-Space的框架结合,这似乎减少了扩展推理/工具使用过程中的错误。

In the benchmarks shown, DeepSeek's performance improves significantly in some cases: Terminal Bench 87.9 → 90.1, NL2Repo 61.5 → 73.4, CyberGym 83.3 → 86.8, DeepSWE 62.7 → 72.0, and Toolathlon 74.1 → 79.5.

在展示的基准测试中,DeepSeek的性能在某些情况下显著提升: Terminal Bench 87.9 → 90.1, NL2Repo 61.5 → 73.4, CyberGym 83.3 → 86.8, DeepSWE 62.7 → 72.0,以及 Toolathlon 74.1 → 79.5。

This combination also outperforms Fable 5 and Opus 4.8 in several agentic/coding benchmarks.

这种组合在多个代理/编码基准测试中也优于Fable 5和Opus 4.8。

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

进入量化体系 →

关联讨论

同一事件的更多信源
评论称某工具并非J-Space实现,纯框架侧改进
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)原文

相似阅读

另一事件,读法相近