LLM核心论文23篇Sentiment Neuron: Learning to Generate Reviews and Discovering SentimentGPT-1: Improving Language Understanding by Generative Pre-TrainingScaling Law: Scaling Laws for Neural Language ModelsGPT-3: Language Models are Few-Shot Learners价值对齐InstructGPT: Training language models to follow instructions with human feedbackConstitutional AI: Harmlessness from AI Feedback架构Transformer: Attention is All You NeedT5: Exploring the Limits of Transfer Learning with a Unified Text-to-Text TransformerBERT: Pre-training of Deep Bidirectional Transformers for Language Understanding轻量微调Rethinking Efficient Tuning Methods from a Unified Perspective推理时算法Chain-of-Thought: Chain-of-Thought Prompting Elicits Reasoning in Large Language Models榜单MMLU: Measuring Massive Multitask Language UnderstandingMATH: Measuring Mathematical Problem Solving With the MATH Dataset多模态Multimodal Few-Shot Learning with Frozen Language ModelsCLIP: Learning Transferable Visual Models From Natural Language SupervisionFlamingo: a Visual Language Model for Few-Shot Learning高观点Pretrained Transformers as Universal Computation EnginesLarge Language Models as General Pattern MachinesAn Observation on Generalization讲座Ilyas talk at GTC2023Alec Radford on LM (Youtube)过去与未来Learning Meaning in Natural Language Processing — The Semantics Mega-ThreadWhat will GPT-2030 look like或许有25年的25篇 但是我还没整理 待更新