1 paper
Ruifeng Yuan, Yizhi Li, Yaxin Du +9
Existing auto-research benchmarks often entangle multiple sources of improvement, including training frameworks, hyperparameters, compute budgets, and data, making it difficult to…