Chinese contentNace.AI 开源 Drex 1.5：9B 决策模型，输出选项概率而非文本
Nace.AI 开源了个 9B 的决策模型 Drex 1.5，不生成文本，一次前向传播直接给每个选项打概率，还支持 128K 上下文，适合做选项评分场景。
#128K上下文
3 stories · 0 in 7 days · 0 in 30 days
Stories, products and papers tagged "128K上下文", newest first.
Oct 11
Oct 2
NVIDIA Long-Transduction Agents Study
This research reveals critical limitations in current AI agent capabilities that developers must account for when building long-running applications. The significant accuracy drops at longer context lengths suggest practical constraints for enterprise implementations requiring sustained task performance. Developers should consider these findings when designing agent workflows that maintain state across thousands of operations, particularly for financial or data-intensive applications.
Jul 28
Models14:32
Chinese content全球首个Agentic扩散模型：边行动边纠错，128K上下文追平自回归扩散模型也能跑Agent任务了，128K上下文还能边做边改，效果和自回归差不多，值得看看。