Alignment
26 milestones sit on this thread. For 21 of them it is the main thread. The others also belong to another thread, which is named under them.
2017
- RLHF originDeep Reinforcement Learning from Human PreferencesTraining method
2019
- RLHF for LMsFine-Tuning Language Models from Human PreferencesTraining method
2020
- RLHF for summarisationLearning to Summarize from Human FeedbackTraining method
2021
- FLANFinetuned Language Models Are Zero-Shot LearnersTraining method
- T0Multitask Prompted Training Enables Zero-Shot Task GeneralizationModel
- HHH assistantA General Language Assistant as a Laboratory for AlignmentResearch paper
- WebGPTWebGPT: Browser-assisted question-answering with human feedbackMain thread: Representation and retrievalResearch paper
2022
- LaMDALaMDA: Language Models for Dialog ApplicationsModel
- InstructGPTTraining language models to follow instructions with human feedbackTraining method
- STaRSTaR: Bootstrapping Reasoning With ReasoningTraining method
- HH-RLHFTraining a Helpful and Harmless Assistant with RLHFTraining method
- Flan-T5 / Flan-PaLMScaling Instruction-Finetuned Language ModelsMain thread: Scale and dataTraining method
- ChatGPTChatGPTProduct
- GPT-3.5GPT-3.5 seriesModel
- Constitutional AIConstitutional AI: Harmlessness from AI FeedbackTraining method
- Self-InstructSelf-Instruct: Aligning Language Models with Self-Generated InstructionsMain thread: Scale and dataTraining method
2023
2024
- GRPODeepSeekMath: Pushing the Limits of Mathematical ReasoningTraining method
- Claude 3The Claude 3 model familyModel
- Chatbot ArenaChatbot Arena: An Open Platform for Evaluating LLMs by Human PreferenceMain thread: Scale and dataBenchmark
- o1OpenAI o1Model
2025
- Claude 3.7Claude 3.7 SonnetModel