Perplexity 发布关于搜索增强模型后训练的研究，通过 SFT 和 RL 提升了 Qwen 模型的准确性与效率

We've published new research on how we post-train models for accurate search-augmented answers. Our SFT + RL pipeline improves search, citation quality, instruction following, and efficiency. With Qwen models, we match or beat GPT models on factuality at a lower cost.