文章预览
LG - 机器学习 CV - 计算机视觉 CL - 计算与语言 AS - 音频与语音 1、[LG] Adaptive Conformal Inference by Betting 2、[AS] ETTA：Elucidating the Design Space of Text-to-Audio Models 3、[LG] Aviary：training language agents on challenging scientific tasks 4、[CL] Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs 5、[AS] TangoFlux：Super Fast and Faithful Text to Audio Generation with Flow Matching and Clap-Ranked Preference Optimization 摘要：通过投注进行自适应共形推理、文本到音频模型设计空间探索、用挑战性科学任务对语言Agent进行训练、o1类LLM的过度思考、通过流匹配和Clap排名偏好优化实现超快且忠实的文本到音频生成 1、[LG] Adaptive Conformal Inference by Betting A Podkopaev, D Xu, K Lee [Walmart Global Tech] 通过投注进行自适应共形推理 要点: 问题：传统的共形预测依赖于数据可交换性的假设，但在真实场景（如时间序列或具有分
………………………………
