40亿参数模型在消费级硬件上实现高效精准问答。
Jan-nano Technical Report
- 用多阶段强化学习训练，不用传统文本预测。
- 简单问答准确率达83.2%，支持128K上下文长度。
- 适合追求低资源高效率的智能应用开发。
大多数语言模型面临强大能力与高算力需求之间的根本矛盾。我们通过Jan-nano——一个40亿参数的语言模型，打破这一限制：它不追求全知全能，而是专注于快速定位信息。该模型基于Qwen3-4B，采用新型多阶段可验证奖励强化学习（RLVR）系统进行微调，完全摒弃了依赖下一个词预测的监督微调（SFT）方式。在集成MCP的前提下，其在SimpleQA基准测试中达到83.2%的准确率，且可在消费级硬件上运行。128K的上下文长度表明，智能的本质不在于规模，而在于策略。
原文摘要 · Abstract (English)
Most language models face a fundamental tradeoff where powerful capabilities require substantial computational resources. We shatter this constraint with Jan-nano, a 4B parameter language model that redefines efficiency through radical specialization: instead of trying to know everything, it masters the art of finding anything instantly. Fine-tuned from Qwen3-4B using our novel multi-stage Reinforcement Learning with Verifiable Rewards (RLVR) system that completely eliminates reliance on next token prediction training (SFT), Jan-nano achieves 83.2% on SimpleQA benchmark with MCP integration while running on consumer hardware. With 128K context length, Jan-nano proves that intelligence isn't about scale, it's about strategy.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。