OpenWebText
PulseAugur coverage of OpenWebText — every cluster mentioning OpenWebText across labs, papers, and developer communities, ranked by signal.
3 天有情绪数据
-
新的DSL框架增强了非自回归生成模型
研究人员引入了离散随机定位（DSL），一种新的连续状态非自回归生成框架。该方法旨在通过提供一种对信噪比不变的更灵活的表示来改进现有的离散扩散模型。在OpenWebText数据集上，使用DSL对预训练模型进行微调已显示出分布忠实度的显著提高，甚至支持以更少的步数进行更快采样。
-
New research enhances diffusion language model efficiency and scalability
Researchers are exploring new methods to improve the efficiency and scalability of diffusion language models (DLMs) for generating long sequences of text. One approach, Block Approximate Sparse Attention (BA-Att), accel…
-
New LLM training methods boost efficiency and error recovery
Researchers have developed new techniques for improving the efficiency of training large language models (LLMs). One method, Step Rejection Fine-Tuning (SRFT), leverages unsuccessful training trajectories by assessing t…
-
OpenAI launches GPT-5.5 Instant, while NRGPT explores energy-based GPT alternatives
OpenAI has updated ChatGPT with GPT-5.5 Instant, enhancing its default model for more accurate responses and better personalization. This upgrade aims to reduce hallucinations and provide clearer, more tailored interact…