Moral Foundations of Large Language Models
Paper
•
2310.15337
•
Published
•
1
RL/Alignment
-
-
Specific versus General Principles for Constitutional AI
Paper • 2310.13798 • Published • 2 -
Contrastive Prefence Learning: Learning from Human Feedback without RL
Paper • 2310.13639 • Published • 24 -
RLAIF: Scaling Reinforcement Learning from Human Feedback with AI Feedback
Paper • 2309.00267 • Published • 47 -
Self-Alignment with Instruction Backtranslation
Paper • 2308.06259 • Published • 40 -
Deep Reinforcement Learning from Hierarchical Weak Preference Feedback
Paper • 2309.02632 • Published • 1 -
A General Theoretical Paradigm to Understand Learning from Human Preferences
Paper • 2310.12036 • Published • 14 -
Pairwise Proximal Policy Optimization: Harnessing Relative Feedback for LLM Alignment
Paper • 2310.00212 • Published • 2 -
Learning Optimal Advantage from Preferences and Mistaking it for Reward
Paper • 2310.02456 • Published • 1 -
Teaching Language Models to Self-Improve through Interactive Demonstrations
Paper • 2310.13522 • Published • 11 -
Chain-of-Thought Reasoning is a Policy Improvement Operator
Paper • 2309.08589 • Published • 1 -
MAF: Multi-Aspect Feedback for Improving Reasoning in Large Language Models
Paper • 2310.12426 • Published • 1 -
Enable Language Models to Implicitly Learn Self-Improvement From Data
Paper • 2310.00898 • Published • 23 -
Tuna: Instruction Tuning using Feedback from Large Language Models
Paper • 2310.13385 • Published • 10 -
Reflection-Tuning: Data Recycling Improves LLM Instruction-Tuning
Paper • 2310.11716 • Published • 5 -
CITING: Large Language Models Create Curriculum for Instruction Tuning
Paper • 2310.02527 • Published • 2 -
Towards Understanding Sycophancy in Language Models
Paper • 2310.13548 • Published • 4 -
Peering Through Preferences: Unraveling Feedback Acquisition for Aligning Large Language Models
Paper • 2308.15812 • Published • 1 -
SALMON: Self-Alignment with Principle-Following Reward Models
Paper • 2310.05910 • Published • 2 -
UltraFeedback: Boosting Language Models with High-quality Feedback
Paper • 2310.01377 • Published • 5 -
Verbosity Bias in Preference Labeling by Large Language Models
Paper • 2310.10076 • Published • 2 -
Safe RLHF: Safe Reinforcement Learning from Human Feedback
Paper • 2310.12773 • Published • 28 -
OpenChat: Advancing Open-source Language Models with Mixed-Quality Data
Paper • 2309.11235 • Published • 16 -
DialCoT Meets PPO: Decomposing and Exploring Reasoning Paths in Smaller Language Models
Paper • 2310.05074 • Published • 1 -
SELF: Language-Driven Self-Evolution for Large Language Model
Paper • 2310.00533 • Published • 2 -
MINT: Evaluating LLMs in Multi-turn Interaction with Tools and Language Feedback
Paper • 2309.10691 • Published • 4 -
A Long Way to Go: Investigating Length Correlations in RLHF
Paper • 2310.03716 • Published • 9 -
Efficient RLHF: Reducing the Memory Usage of PPO
Paper • 2309.00754 • Published • 13 -
Aligning Language Models with Offline Reinforcement Learning from Human Feedback
Paper • 2308.12050 • Published • 1 -
Reward Model Ensembles Help Mitigate Overoptimization
Paper • 2310.02743 • Published • 1 -
SCREWS: A Modular Framework for Reasoning with Revisions
Paper • 2309.13075 • Published • 15 -
Leveraging Unpaired Data for Vision-Language Generative Models via Cycle Consistency
Paper • 2310.03734 • Published • 14 -
DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines
Paper • 2310.03714 • Published • 30 -
Aligning Text-to-Image Diffusion Models with Reward Backpropagation
Paper • 2310.03739 • Published • 21 -
Aligning Large Multimodal Models with Factually Augmented RLHF
Paper • 2309.14525 • Published • 29 -
Prometheus: Inducing Fine-grained Evaluation Capability in Language Models
Paper • 2310.08491 • Published • 53 -
The Consensus Game: Language Model Generation via Equilibrium Search
Paper • 2310.09139 • Published • 12 -
Quality-Diversity through AI Feedback
Paper • 2310.13032 • Published • 1 -
Reward-Augmented Decoding: Efficient Controlled Text Generation With a Unidirectional Reward Model
Paper • 2310.09520 • Published • 10 -
Controllable Text Generation with Residual Memory Transformer
Paper • 2309.16231 • Published • 1 -
Qwen Technical Report
Paper • 2309.16609 • Published • 34 -
Controlled Decoding from Language Models
Paper • 2310.17022 • Published • 14 -
JudgeLM: Fine-tuned Large Language Models are Scalable Judges
Paper • 2310.17631 • Published • 32 -
Trustworthy LLMs: a Survey and Guideline for Evaluating Large Language Models' Alignment
Paper • 2308.05374 • Published • 27 -
SayCanPay: Heuristic Planning with Large Language Models using Learnable Domain Knowledge
Paper • 2308.12682 • Published • 2 -
Grounding Large Language Models in Interactive Environments with Online Reinforcement Learning
Paper • 2302.02662 • Published • 1 -
Natural Logic-guided Autoregressive Multi-hop Document Retrieval for Fact Verification
Paper • 2212.05276 • Published • 1 -
Aligning Large Language Models with Human: A Survey
Paper • 2307.12966 • Published • 1 -
Zephyr: Direct Distillation of LM Alignment
Paper • 2310.16944 • Published • 121 -
Statistical Rejection Sampling Improves Preference Optimization
Paper • 2309.06657 • Published • 13 -
Principle-Driven Self-Alignment of Language Models from Scratch with Minimal Human Supervision
Paper • 2305.03047 • Published • 1 -
CIEM: Contrastive Instruction Evaluation Method for Better Instruction Tuning
Paper • 2309.02301 • Published • 1 -
TouchStone: Evaluating Vision-Language Models by Language Models
Paper • 2308.16890 • Published • 1 -
Are Large Language Model-based Evaluators the Solution to Scaling Up Multilingual Evaluation?
Paper • 2309.07462 • Published • 3 -
Stabilizing RLHF through Advantage Model and Selective Rehearsal
Paper • 2309.10202 • Published • 9 -
VIGC: Visual Instruction Generation and Correction
Paper • 2308.12714 • Published • 1 -
Improving Generalization of Alignment with Human Preferences through Group Invariant Learning
Paper • 2310.11971 • Published • 1 -
Large Language Models as Optimizers
Paper • 2309.03409 • Published • 75 -
In-Context Alignment: Chat with Vanilla Language Models Before Fine-Tuning
Paper • 2308.04275 • Published • 1 -
Direct Preference Optimization: Your Language Model is Secretly a Reward Model
Paper • 2305.18290 • Published • 48 -
Beyond Reward: Offline Preference-guided Policy Optimization
Paper • 2305.16217 • Published • 1 -
Decentralized Policy Optimization
Paper • 2211.03032 • Published • 1 -
Large Language Models are not Fair Evaluators
Paper • 2305.17926 • Published • 1 -
SLiC-HF: Sequence Likelihood Calibration with Human Feedback
Paper • 2305.10425 • Published • 5 -
Don't throw away your value model! Making PPO even better via Value-Guided Monte-Carlo Tree Search decoding
Paper • 2309.15028 • Published • 1 -
Improving Language Models with Advantage-based Offline Policy Gradients
Paper • 2305.14718 • Published • 2 -
Large Language Models Cannot Self-Correct Reasoning Yet
Paper • 2310.01798 • Published • 33 -
Q-Transformer: Scalable Offline Reinforcement Learning via Autoregressive Q-Functions
Paper • 2309.10150 • Published • 24 -
CodeRL: Mastering Code Generation through Pretrained Models and Deep Reinforcement Learning
Paper • 2207.01780 • Published • 1 -
RLTF: Reinforcement Learning from Unit Test Feedback
Paper • 2307.04349 • Published • 4 -
Fine-Grained Human Feedback Gives Better Rewards for Language Model Training
Paper • 2306.01693 • Published • 3 -
Fine-tuning Language Models with Generative Adversarial Feedback
Paper • 2305.06176 • Published • 1 -
Aligning Large Language Models through Synthetic Feedback
Paper • 2305.13735 • Published • 1 -
Fine-Tuning Language Models with Advantage-Induced Policy Alignment
Paper • 2306.02231 • Published • 2 -
Reinforced Self-Training (ReST) for Language Modeling
Paper • 2308.08998 • Published • 2 -
SuperHF: Supervised Iterative Learning from Human Feedback
Paper • 2310.16763 • Published • 1 -
Split and Merge: Aligning Position Biases in Large Language Model based Evaluators
Paper • 2310.01432 • Published • 1 -
Generative Judge for Evaluating Alignment
Paper • 2310.05470 • Published • 1 -
Personas as a Way to Model Truthfulness in Language Models
Paper • 2310.18168 • Published • 5 -
A Framework for Automated Measurement of Responsible AI Harms in Generative AI Applications
Paper • 2310.17750 • Published • 9 -
RRAML: Reinforced Retrieval Augmented Machine Learning
Paper • 2307.12798 • Published • 1 -
Enabling Intelligent Interactions between an Agent and an LLM: A Reinforcement Learning Approach
Paper • 2306.03604 • Published • 1 -
Conservative Dual Policy Optimization for Efficient Model-Based Reinforcement Learning
Paper • 2209.07676 • Published • 2 -
Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!
Paper • 2310.03693 • Published • 1 -
Evaluating the Moral Beliefs Encoded in LLMs
Paper • 2307.14324 • Published • 1 -
Moral Mimicry: Large Language Models Produce Moral Rationalizations Tailored to Political Identity
Paper • 2209.12106 • Published • 1 -
Red-Teaming Large Language Models using Chain of Utterances for Safety-Alignment
Paper • 2308.09662 • Published • 3 -
KoSBi: A Dataset for Mitigating Social Bias Risks Towards Safer Large Language Model Application
Paper • 2305.17701 • Published • 1 -
A Survey on Fairness in Large Language Models
Paper • 2308.10149 • Published • 1 -
Do LLMs Understand User Preferences? Evaluating LLMs On User Rating Prediction
Paper • 2305.06474 • Published • 1 -
Of Models and Tin Men: A Behavioural Economics Study of Principal-Agent Problems in AI Alignment using Large-Language Models
Paper • 2307.11137 • Published • 1 -
LoRA Fine-tuning Efficiently Undoes Safety Training in Llama 2-Chat 70B
Paper • 2310.20624 • Published • 12 -
ExpeL: LLM Agents Are Experiential Learners
Paper • 2308.10144 • Published • 2 -
Sociotechnical Safety Evaluation of Generative AI Systems
Paper • 2310.11986 • Published -
Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback
Paper • 2307.15217 • Published • 36 -
Secrets of RLHF in Large Language Models Part I: PPO
Paper • 2307.04964 • Published • 28 -
Demystifying GPT Self-Repair for Code Generation
Paper • 2306.09896 • Published • 19 -
Training language models to follow instructions with human feedback
Paper • 2203.02155 • Published • 15 -
Okapi: Instruction-tuned Large Language Models in Multiple Languages with Reinforcement Learning from Human Feedback
Paper • 2307.16039 • Published • 4 -
A Mixture-of-Expert Approach to RL-based Dialogue Management
Paper • 2206.00059 • Published • 1 -
"Pick-and-Pass" as a Hat-Trick Class for First-Principle Memory, Generalizability, and Interpretability Benchmarks
Paper • 2310.20654 • Published • 1 -
On Second Thought, Let's Not Think Step by Step! Bias and Toxicity in Zero-Shot Reasoning
Paper • 2212.08061 • Published • 1 -
MUTEX: Learning Unified Policies from Multimodal Task Specifications
Paper • 2309.14320 • Published • 1 -
Lifelong Inverse Reinforcement Learning
Paper • 2207.00461 • Published • 1 -
Improving Code Generation by Training with Natural Language Feedback
Paper • 2303.16749 • Published • 1 -
Tailoring Self-Rationalizers with Multi-Reward Distillation
Paper • 2311.02805 • Published • 3 -
Unleashing the Power of Pre-trained Language Models for Offline Reinforcement Learning
Paper • 2310.20587 • Published • 16 -
B-Coder: Value-Based Deep Reinforcement Learning for Program Synthesis
Paper • 2310.03173 • Published • 1 -
Towards Anytime Fine-tuning: Continually Pre-trained Language Models with Hypernetwork Prompt
Paper • 2310.13024 • Published • 1 -
Multi-Task Recommendations with Reinforcement Learning
Paper • 2302.03328 • Published • 1 -
Curriculum-based Asymmetric Multi-task Reinforcement Learning
Paper • 2211.03352 • Published • 1 -
Efficient Training of Multi-task Combinarotial Neural Solver with Multi-armed Bandits
Paper • 2305.06361 • Published • 1 -
Rethinking Decision Transformer via Hierarchical Reinforcement Learning
Paper • 2311.00267 • Published • 1 -
Pre-training with Synthetic Data Helps Offline Reinforcement Learning
Paper • 2310.00771 • Published • 2 -
Guiding Pretraining in Reinforcement Learning with Large Language Models
Paper • 2302.06692 • Published • 1 -
Large Language Model Alignment: A Survey
Paper • 2309.15025 • Published • 2 -
Making Large Language Models Better Reasoners with Alignment
Paper • 2309.02144 • Published • 2 -
Pretraining in Deep Reinforcement Learning: A Survey
Paper • 2211.03959 • Published • 1 -
Reinforcement Learning for Generative AI: A Survey
Paper • 2308.14328 • Published • 1 -
d3rlpy: An Offline Deep Reinforcement Learning Library
Paper • 2111.03788 • Published • 1 -
Harnessing Mixed Offline Reinforcement Learning Datasets via Trajectory Weighting
Paper • 2306.13085 • Published • 1 -
Efficient Online Reinforcement Learning with Offline Data
Paper • 2302.02948 • Published • 2 -
Improving Offline-to-Online Reinforcement Learning with Q-Ensembles
Paper • 2306.06871 • Published • 1 -
A Simple Unified Uncertainty-Guided Framework for Offline-to-Online Reinforcement Learning
Paper • 2306.07541 • Published • 1 -
A Dataset Perspective on Offline Reinforcement Learning
Paper • 2111.04714 • Published • 1 -
Goal-Conditioned Predictive Coding as an Implicit Planner for Offline Reinforcement Learning
Paper • 2307.03406 • Published • 1 -
Semi-Supervised Offline Reinforcement Learning with Action-Free Trajectories
Paper • 2210.06518 • Published • 1 -
Mildly Constrained Evaluation Policy for Offline Reinforcement Learning
Paper • 2306.03680 • Published • 1 -
Conservative State Value Estimation for Offline Reinforcement Learning
Paper • 2302.06884 • Published • 1 -
Revisiting the Minimalist Approach to Offline Reinforcement Learning
Paper • 2305.09836 • Published • 3 -
Q-Ensemble for Offline RL: Don't Scale the Ensemble, Scale the Batch Size
Paper • 2211.11092 • Published • 1 -
A learning gap between neuroscience and reinforcement learning
Paper • 2104.10995 • Published • 1 -
AF Adapter: Continual Pretraining for Building Chinese Biomedical Language Model
Paper • 2211.11363 • Published • 1 -
RLocator: Reinforcement Learning for Bug Localization
Paper • 2305.05586 • Published • 1 -
Beyond Words: A Mathematical Framework for Interpreting Large Language Models
Paper • 2311.03033 • Published • 1 -
LoopTune: Optimizing Tensor Computations with Reinforcement Learning
Paper • 2309.01825 • Published • 1 -
Do-Not-Answer: A Dataset for Evaluating Safeguards in LLMs
Paper • 2308.13387 • Published • 1 -
Hypernetworks for Zero-shot Transfer in Reinforcement Learning
Paper • 2211.15457 • Published • 1 -
Offline Experience Replay for Continual Offline Reinforcement Learning
Paper • 2305.13804 • Published • 1 -
Rewarded meta-pruning: Meta Learning with Rewards for Channel Pruning
Paper • 2301.11063 • Published • 1 -
Fine-tuning Language Models for Factuality
Paper • 2311.08401 • Published • 28 -
From Instructions to Intrinsic Human Values -- A Survey of Alignment Goals for Big Models
Paper • 2308.12014 • Published • 1 -
Automatically Correcting Large Language Models: Surveying the landscape of diverse self-correction strategies
Paper • 2308.03188 • Published • 2 -
When to Show a Suggestion? Integrating Human Feedback in AI-Assisted Programming
Paper • 2306.04930 • Published • 3 -
Appropriateness is all you need!
Paper • 2304.14553 • Published • 1 -
LLM Cognitive Judgements Differ From Human
Paper • 2307.11787 • Published • 1 -
Fake Alignment: Are LLMs Really Aligned Well?
Paper • 2311.05915 • Published • 2 -
LLM Augmented Hierarchical Agents
Paper • 2311.05596 • Published • 1 -
Self-driven Grounding: Large Language Model Agents with Automatical Language-aligned Skill Learning
Paper • 2309.01352 • Published • 1 -
Introspective Tips: Large Language Model for In-Context Decision Making
Paper • 2305.11598 • Published • 1 -
SPRING: GPT-4 Out-performs RL Algorithms by Studying Papers and Reasoning
Paper • 2305.15486 • Published • 1 -
Alignment is not sufficient to prevent large language models from generating harmful information: A psychoanalytic perspective
Paper • 2311.08487 • Published • 1 -
HelpSteer: Multi-attribute Helpfulness Dataset for SteerLM
Paper • 2311.09528 • Published • 2 -
Learning to Prune Deep Neural Networks via Reinforcement Learning
Paper • 2007.04756 • Published • 1 -
Training Language Models with Language Feedback at Scale
Paper • 2303.16755 • Published • 1 -
Recomposing the Reinforcement Learning Building Blocks with Hypernetworks
Paper • 2106.06842 • Published • 1 -
Continual Model-Based Reinforcement Learning with Hypernetworks
Paper • 2009.11997 • Published • 1 -
Responsible Task Automation: Empowering Large Language Models as Responsible Task Automators
Paper • 2306.01242 • Published • 2 -
Pangu-Agent: A Fine-Tunable Generalist Agent with Structured Reasoning
Paper • 2312.14878 • Published • 13 -
ICE-GRT: Instruction Context Enhancement by Generative Reinforcement based Transformers
Paper • 2401.02072 • Published • 9 -
RLHF-V: Towards Trustworthy MLLMs via Behavior Alignment from Fine-grained Correctional Human Feedback
Paper • 2312.00849 • Published • 8 -
Diffusion Model Alignment Using Direct Preference Optimization
Paper • 2311.12908 • Published • 47 -
Alignment for Honesty
Paper • 2312.07000 • Published • 11 -
Pearl: A Production-ready Reinforcement Learning Agent
Paper • 2312.03814 • Published • 14 -
Routing to the Expert: Efficient Reward-guided Ensemble of Large Language Models
Paper • 2311.08692 • Published • 12 -
Trusted Source Alignment in Large Language Models
Paper • 2311.06697 • Published • 10 -
Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models
Paper • 2401.01335 • Published • 64 -
Self-Rewarding Language Models
Paper • 2401.10020 • Published • 143 -
Aligning Large Language Models with Human Preferences through Representation Engineering
Paper • 2312.15997 • Published • 1 -
DRLC: Reinforcement Learning with Dense Rewards from LLM Critic
Paper • 2401.07382 • Published • 2 -
Secrets of RLHF in Large Language Models Part II: Reward Modeling
Paper • 2401.06080 • Published • 25 -
Data-Efficient Alignment of Large Language Models with Human Feedback Through Natural Language
Paper • 2311.14543 • Published • 1 -
LongAlign: A Recipe for Long Context Alignment of Large Language Models
Paper • 2401.18058 • Published • 21 -
StepCoder: Improve Code Generation with Reinforcement Learning from Compiler Feedback
Paper • 2402.01391 • Published • 41 -
Efficient Exploration for LLMs
Paper • 2402.00396 • Published • 21 -
West-of-N: Synthetic Preference Generation for Improved Reward Modeling
Paper • 2401.12086 • Published • 1 -
Improving Reinforcement Learning from Human Feedback with Efficient Reward Model Ensemble
Paper • 2401.16635 • Published • 1 -
Uncertainty-Penalized Reinforcement Learning from Human Feedback with Diverse Reward LoRA Ensembles
Paper • 2401.00243 • Published • 1 -
Iterative Data Smoothing: Mitigating Reward Overfitting and Overoptimization in RLHF
Paper • 2401.16335 • Published • 1 -
Transforming and Combining Rewards for Aligning Large Language Models
Paper • 2402.00742 • Published • 11 -
Helping or Herding? Reward Model Ensembles Mitigate but do not Eliminate Reward Hacking
Paper • 2312.09244 • Published • 7 -
Can LLMs Follow Simple Rules?
Paper • 2311.04235 • Published • 10 -
Zero-Shot Goal-Directed Dialogue via RL on Imagined Conversations
Paper • 2311.05584 • Published • 1 -
Mixtures of Experts Unlock Parameter Scaling for Deep RL
Paper • 2402.08609 • Published • 34 -
DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
Paper • 2402.03300 • Published • 69 -
ReFT: Reasoning with Reinforced Fine-Tuning
Paper • 2401.08967 • Published • 27 -
In deep reinforcement learning, a pruned network is a good network
Paper • 2402.12479 • Published • 17 -
Q-Probe: A Lightweight Approach to Reward Maximization for Language Models
Paper • 2402.14688 • Published -
WildChat: 1M ChatGPT Interaction Logs in the Wild
Paper • 2405.01470 • Published • 59 -
FLAME: Factuality-Aware Alignment for Large Language Models
Paper • 2405.01525 • Published • 24 -
NeMo-Aligner: Scalable Toolkit for Efficient Model Alignment
Paper • 2405.01481 • Published • 25 -
Self-Play Preference Optimization for Language Model Alignment
Paper • 2405.00675 • Published • 24 -
Small Language Model Can Self-correct
Paper • 2401.07301 • Published -
Show, Don't Tell: Aligning Language Models with Demonstrated Feedback
Paper • 2406.00888 • Published • 30 -
LongSkywork: A Training Recipe for Efficiently Extending Context Length in Large Language Models
Paper • 2406.00605 • Published • 2 -
WPO: Enhancing RLHF with Weighted Preference Optimization
Paper • 2406.11827 • Published • 14 -
XLand-100B: A Large-Scale Multi-Task Dataset for In-Context Reinforcement Learning
Paper • 2406.08973 • Published • 85