Global ETD Search

Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.

Results

Showing 1 to 8 of 8 for “"Rlhf"”.

  1. Aligning Language Models Using Multi-Objective Deep Reinforcement Learning

    … is reinforcement learning from human feedback (RLHF). RLHF aims to optimize one objective based on human preferences. However, the cost of high-quality human feedback is enormous. Having all human annotators consistent in their opinions on desirable behaviors is also challenging. LLM alignment …

    brock Repository record for Aligning Language Models Using Multi-Objective Deep Reinforcement Learning (opens in a new tab)

  2. Inverse Constitutional AI

    … Empirical evaluation is conducted on the hh-rlhf dataset for training helpful and harmless AI assistants, as well as a synthetic dataset constructed by relabeling hh-rlhf samples with predefined principles. Results demonstrate promising capabilities in clustering semantically coherent topics …

    mit Repository record for Inverse Constitutional AI (opens in a new tab)

  3. Offline Reward Learning from Human Demonstrations and Feedback: A Linear Programming Approach

    … and reinforcement learning from human feedback (RLHF). Despite the successful application of these reward learning techniques across a wide range of tasks, a significant gap between theory and practice persists. This work aims to bridge this gap by introducing a novel linear programming (LP) …

    mit Repository record for Offline Reward Learning from Human Demonstrations and Feedback: A Linear Programming Approach (opens in a new tab)

  4. Adversarial Prompt Transformation for Systematic Jailbreaks of LLMs

    … Reinforcement Learning from Human Feedback (RLHF) to transform unsuccessful adversarial prompts into a successful jailbreak. Thus it learns a policy based on relation to existing jailbreak prompts that informs the generator LLM of what makes an adversarial prompt successful. This was …

    mit Repository record for Adversarial Prompt Transformation for Systematic Jailbreaks of LLMs (opens in a new tab)

  5. Steerable Alignment with Conditional Multiobjective Preference Optimization

    … as Reinforcement Learning from Human Feedback (RLHF) have provided useful paradigms for finetuning LLMs to produce outputs that are more consistent with human preferences. These approaches, however, assume that preferences are formed by a single, underlying reward model, which is likely …

    mit Repository record for Steerable Alignment with Conditional Multiobjective Preference Optimization (opens in a new tab)

  6. Prompt Injection Generation Using Small Language Models with Reinforcement Learning with Artificial Intelligence Feedback

    … like Reinforcement Learning with Human Feedback (RLHF) and keyword filtering have contributed to improving the robustness of these models, but these approaches are very resource-intensive and the models can still be vulnerable to malicious attacks like prompt injections and jailbreaking. One …

    mit Repository record for Prompt Injection Generation Using Small Language Models with Reinforcement Learning with Artificial Intelligence Feedback (opens in a new tab)

  7. Goal Inference from Open-Ended Dialog

    … diverse user goals. While offline methods like RLHF can represent various goals but require large datasets, our approach achieves similar flexibility with online efficiency. We extract natural language goal representations from conversations with Large Language Models (LLMs). We prompt an LLM to …

    mit Repository record for Goal Inference from Open-Ended Dialog (opens in a new tab)

  8. Training a massively multimodal transformer on YouTube data: pre-training and parameter efficient fine-tuning on HPC infrastructure

    Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2023-09-01 without embargo terms

    uiuc Repository record for Training a massively multimodal transformer on YouTube data: pre-training and parameter efficient fine-tuning on HPC infrastructure (opens in a new tab)