WARP: On the Benefits of Weight Averaged Rewarded Policies

• URL: http://arxiv.org/abs/2406.16768v1
• Date: Mon, 24 Jun 2024 16:24:34 GMT
• Title: WARP: On the Benefits of Weight Averaged Rewarded Policies
• Authors: Alexandre Ramé, Johan Ferret, Nino Vieillard, Robert Dadashi, Léonard Hussenot, Pierre-Louis Cedoz, Pier Giuseppe Sessa, Sertan Girgin, Arthur Douillard, Olivier Bachem,
• Abstract summary: We introduce a novel alignment strategy named Weight Averaged Rewarded Policies (WARP) WARP merges policies in the weight space at three distinct stages. Experiments with GEMMA policies validate that WARP improves their quality and alignment, outperforming other open-source LLMs.
• Score: 66.95013068137115
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Reinforcement learning from human feedback (RLHF) aligns large language models (LLMs) by encouraging their generations to have high rewards, using a reward model trained on human preferences. To prevent the forgetting of pre-trained knowledge, RLHF usually incorporates a KL regularization; this forces the policy to remain close to its supervised fine-tuned initialization, though it hinders the reward optimization. To tackle the trade-off between KL and reward, in this paper we introduce a novel alignment strategy named Weight Averaged Rewarded Policies (WARP). WARP merges policies in the weight space at three distinct stages. First, it uses the exponential moving average of the policy as a dynamic anchor in the KL regularization. Second, it applies spherical interpolation to merge independently fine-tuned policies into a new enhanced one. Third, it linearly interpolates between this merged model and the initialization, to recover features from pre-training. This procedure is then applied iteratively, with each iteration's final model used as an advanced initialization for the next, progressively refining the KL-reward Pareto front, achieving superior rewards at fixed KL. Experiments with GEMMA policies validate that WARP improves their quality and alignment, outperforming other open-source LLMs.

Related papers

• Contrastive Policy Gradient: Aligning LLMs on sequence-level scores in a supervised-friendly fashion [44.95386817008473]
  We introduce Contrastive Policy Gradient, or CoPG, a simple and mathematically principled new RL algorithm that can estimate the optimal policy even from off-policy data. We show this approach to generalize the direct alignment method IPO (identity preference optimization) and classic policy gradient. We experiment with the proposed CoPG on a toy bandit problem to illustrate its properties, as well as for finetuning LLMs on a summarization task.
  arXiv  Detail & Related papers  (2024-06-27T14:03:49Z)
• Improving Reward-Conditioned Policies for Multi-Armed Bandits using Normalized Weight Functions [8.90692770076582]
  Recently proposed reward-conditioned policies (RCPs) offer an appealing alternative in reinforcement learning. We show that RCPs are slower to converge and have inferior expected rewards at convergence, compared with classic methods. We refer to this technique as generalized marginalization, whose advantage is that negative weights for policies conditioned on low rewards can make the resulting policies more distinct from them.
  arXiv  Detail & Related papers  (2024-06-16T03:43:55Z)
• REBEL: Reinforcement Learning via Regressing Relative Rewards [59.68420022466047]
  We propose REBEL, a minimalist RL algorithm for the era of generative models. In theory, we prove that fundamental RL algorithms like Natural Policy Gradient can be seen as variants of REBEL. We find that REBEL provides a unified approach to language modeling and image generation with stronger or similar performance as PPO and DPO.
  arXiv  Detail & Related papers  (2024-04-25T17:20:45Z)
• Fine-Tuning Language Models with Reward Learning on Policy [68.70065254564642]
  Reinforcement learning from human feedback (RLHF) has emerged as an effective approach to aligning large language models (LLMs) to human preferences. Despite its popularity, (fixed) reward models may suffer from inaccurate off-distribution. We propose reward learning on policy (RLP), an unsupervised framework that refines a reward model using policy samples to keep it on-distribution.
  arXiv  Detail & Related papers  (2024-03-28T10:02:10Z)
• COPR: Continual Human Preference Learning via Optimal Policy Regularization [56.1193256819677]
  Reinforcement Learning from Human Feedback (RLHF) is commonly utilized to improve the alignment of Large Language Models (LLMs) with human preferences. We propose the Continual Optimal Policy Regularization (COPR) method, which draws inspiration from the optimal policy theory.
  arXiv  Detail & Related papers  (2024-02-22T02:20:08Z)
• WARM: On the Benefits of Weight Averaged Reward Models [63.08179139233774]
  We propose Weight Averaged Reward Models (WARM) to mitigate reward hacking. Experiments on summarization tasks, using best-of-N and RL methods, shows that WARM improves the overall quality and alignment of LLM predictions.
  arXiv  Detail & Related papers  (2024-01-22T18:27:08Z)
• Nash Learning from Human Feedback [86.09617990412941]
  We introduce an alternative pipeline for the fine-tuning of large language models using pairwise human feedback. We term this approach Nash learning from human feedback (NLHF) We present a novel algorithmic solution, Nash-MD, founded on the principles of mirror descent.
  arXiv  Detail & Related papers  (2023-12-01T19:26:23Z)

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.