百度发布 ERNIE Evolve 自动化研究系统，可自主改进训练与推理
百度搞了个自动做研究的系统 ERNIE Evolve，给定目标就能自己改代码跑实验，在 NanoGPT Speedrun 上拿下 SOTA，还把 ERNIE 5.5 Flash 的 GPU 推理吞吐拉高 33.7%。做工程优化的开发者可以看看它怎么把实验闭环自动化。
百度推出 ERNIE Evolve，一个面向 AI 训练与推理优化的自动化研究系统。它接收明确的研究目标后，自主选择研究方向、修改代码并运行实验，只有通过独立评估与审查的改进才会保留，且每轮发现会指导下一轮迭代。
在公开基准上，该系统在 NanoChat AutoResearch 和 NanoGPT Speedrun 上达到 SOTA；在 ERNIE 5.5 Flash 推理测试中，确定性 GPU 推理吞吐提升 33.7%，单请求解码速度提升 14.9%。百度将此视为迈向递归自我改进（RSI）的实用一步，下一步计划把该方法扩展到训练系统、模型能力与评估环节。
这类由系统自主完成研究闭环的做法此前多见于学术界的 AlphaEvolve 等项目，百度将其落到自家模型推理优化上，意味着自动化研究开始进入生产环境验证阶段。
Introducing ERNIE Evolve, our automated research system for improving AI training and inference. ⚙️
Given a clear research goal, it selects research directions, modifies code and runs experiments. Only improvements that pass independent evaluation and review are kept, and each round’s findings guide the next.
From public benchmarks to ERNIE 5.5 Flash inference tests, here’s what it achieved: - SOTA on NanoChat AutoResearch and NanoGPT Speedrun - 33.7% higher deterministic GPU inference throughput - 14.9% faster single-request decoding on GPUs
This is a practical step toward recursive self-improvement (RSI). Next, we plan to extend this approach to training systems, model capabilities and evaluation.
This report is categorized under AITOP's "Models" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include ERNIE Evolve, Baidu, 递归自我改进, 推理优化.
Source attribution: curated from 百度 AI Baidu (@Baidu_Inc).
