Q-SFT: Q-Learning for Language Models via Supervised Fine-Tuning

• URL: http://arxiv.org/abs/2411.05193v1
• Date: Thu, 07 Nov 2024 21:36:52 GMT
• Title: Q-SFT: Q-Learning for Language Models via Supervised Fine-Tuning
• Authors: Joey Hong, Anca Dragan, Sergey Levine,
• Abstract summary: Value-based reinforcement learning can learn effective policies for a wide range of multi-turn problems. Current value-based RL methods have proven particularly challenging to scale to the setting of large language models. We propose a novel offline RL algorithm that addresses these drawbacks, casting Q-learning as a modified supervised fine-tuning problem.
• Score: 62.984693936073974
• License:
• Abstract: Value-based reinforcement learning (RL) can in principle learn effective policies for a wide range of multi-turn problems, from games to dialogue to robotic control, including via offline RL from static previously collected datasets. However, despite the widespread use of policy gradient methods to train large language models for single turn tasks (e.g., question answering), value-based methods for multi-turn RL in an off-policy or offline setting have proven particularly challenging to scale to the setting of large language models. This setting requires effectively leveraging pretraining, scaling to large architectures with billions of parameters, and training on large datasets, all of which represent major challenges for current value-based RL methods. In this work, we propose a novel offline RL algorithm that addresses these drawbacks, casting Q-learning as a modified supervised fine-tuning (SFT) problem where the probabilities of tokens directly translate to Q-values. In this way we obtain an algorithm that smoothly transitions from maximizing the likelihood of the data during pretraining to learning a near-optimal Q-function during finetuning. Our algorithm has strong theoretical foundations, enjoying performance bounds similar to state-of-the-art Q-learning methods, while in practice utilizing an objective that closely resembles SFT. Because of this, our approach can enjoy the full benefits of the pretraining of language models, without the need to reinitialize any weights before RL finetuning, and without the need to initialize new heads for predicting values or advantages. Empirically, we evaluate our method on both pretrained LLMs and VLMs, on a variety of tasks including both natural language dialogue and robotic manipulation and navigation from images.

Related papers

• Multi-turn Reinforcement Learning from Preference Human Feedback [41.327438095745315]
  Reinforcement Learning from Human Feedback (RLHF) has become the standard approach for aligning Large Language Models with human preferences. Existing methods work by emulating the preferences at the single decision (turn) level. We develop novel methods for Reinforcement Learning from preference feedback between two full multi-turn conversations.
  arXiv  Detail & Related papers  (2024-05-23T14:53:54Z)
• How Can LLM Guide RL? A Value-Based Approach [68.55316627400683]
  Reinforcement learning (RL) has become the de facto standard practice for sequential decision-making problems by improving future acting policies with feedback. Recent developments in large language models (LLMs) have showcased impressive capabilities in language understanding and generation, yet they fall short in exploration and self-improvement capabilities. We develop an algorithm named LINVIT that incorporates LLM guidance as a regularization factor in value-based RL, leading to significant reductions in the amount of data needed for learning.
  arXiv  Detail & Related papers  (2024-02-25T20:07:13Z)
• Offline Q-Learning on Diverse Multi-Task Data Both Scales And Generalizes [100.69714600180895]
  offline Q-learning algorithms exhibit strong performance that scales with model capacity. We train a single policy on 40 games with near-human performance using up-to 80 million parameter networks. Compared to return-conditioned supervised approaches, offline Q-learning scales similarly with model capacity and has better performance, especially when the dataset is suboptimal.
  arXiv  Detail & Related papers  (2022-11-28T08:56:42Z)
• Offline RL for Natural Language Generation with Implicit Language Q Learning [87.76695816348027]
  Large language models can be inconsistent when it comes to completing user specified tasks. We propose a novel RL method, that combines both the flexible utility framework of RL with the ability of supervised learning. In addition to empirically validating ILQL, we present a detailed empirical analysis situations where offline RL can be useful in natural language generation settings.
  arXiv  Detail & Related papers  (2022-06-05T18:38:42Z)
• Text Generation with Efficient (Soft) Q-Learning [91.47743595382758]
  Reinforcement learning (RL) offers a more flexible solution by allowing users to plug in arbitrary task metrics as reward. We introduce a new RL formulation for text generation from the soft Q-learning perspective. We apply the approach to a wide range of tasks, including learning from noisy/negative examples, adversarial attacks, and prompt generation.
  arXiv  Detail & Related papers  (2021-06-14T18:48:40Z)
• AWAC: Accelerating Online Reinforcement Learning with Offline Datasets [84.94748183816547]
  We show that our method, advantage weighted actor critic (AWAC), enables rapid learning of skills with a combination of prior demonstration data and online experience. Our results show that incorporating prior data can reduce the time required to learn a range of robotic skills to practical time-scales.
  arXiv  Detail & Related papers  (2020-06-16T17:54:41Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.