LLMs Can Learn Complex Math from Just One Example: Researchers from University of Washington, Microsoft, and USC Unlock the Power of 1-Shot Reinforcement Learning with Verifiable Reward
"数据效率" 相关文章
OpenAI自曝GPT-4.5训练内幕：数据效率是关键，预训练仍然有用
OpenAI自曝GPT-4.5训练内幕：数据效率是关键，预训练仍然有用
OpenAI揭秘GPT-4.5训练：10万块GPU几乎全员上阵 出现“灾难性问题”
OpenAI 揭秘 GPT-4.5 训练：10 万块 GPU，几乎全员上阵，出现“灾难性问题”
OpenAI：现在只需 5~10 人即可从头重建 GPT-4，发展瓶颈已经从“算力”转变为“数据效率”
机器人泛化能力大幅提升：HAMSTER层次化方法和VLA尺度轨迹预测，显著提升开放世界任务成功率
This AI Paper from UC Berkeley Introduces a Data-Efficient Approach to Long Chain-of-Thought Reasoning for Large Language Models
少用33％数据，模型性能不变，陈丹琦团队用元数据来做降本增效
少用33％数据，模型性能不变，陈丹琦团队用元数据来做降本增效
Researchers from Princeton University Introduce Metadata Conditioning then Cooldown (MeCo) to Simplify and Optimize Language Model Pre-training
Google DeepMind Researchers Introduce Diffusion Augmented Agents: A Machine Learning Framework for Efficient Exploration and Transfer Learning
Unifying Vision and Language Models with Mohit Bansal - #636
