Pith. sign in

REVIEW 8 cited by

WARP: On the Benefits of Weight Averaged Rewarded Policies

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2406.16768 v1 pith:TJ5GE6UH submitted 2024-06-24 cs.LG cs.AI

classification cs.LGcs.AI
keywords policieswarpinitializationmodelrewardweightalignmentaveraged
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original 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.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 8 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. SMOPD: Multi-Reward Reinforcement Learning via Specialize-and-Merge Online Policy Distillation

    cs.LG 2026-08 conditional novelty 6.0 of 10

    SMOPD trains one reward-specialized teacher per reward and merges them via token-level online distillation into a balanced student, reporting gains over GDPO on tool-calling format compliance and safety alignment.

  2. REVA-PO: Stabilizing Reinforcement Learning for Chest X-ray Report Generation

    cs.CV 2026-07 conditional novelty 6.0 of 10

    REVA-PO stabilizes GRPO-style RL for CXR report generation via response-level adaptive KL weights and validation-anchored policy resets, reporting new SOTA BLEU and clinical F1 scores.

  3. Multi-Turn On-Policy Distillation with Prefix Replay

    cs.LG 2026-07 conditional novelty 6.0 of 10

    ReOPD offline-distills multi-turn agentic LLMs via teacher-prefix replay plus step-decay sampling, matching online OPD accuracy at ≥4× speed with zero tool calls.

  4. Parameter-Efficient Interventions for Enhanced Model Merging

    cs.CV 2024-12 conditional novelty 6.0 of 10

    Applying ReFT-style interventions at every transformer block of a merged model improves multi-task accuracy beyond post-hoc single-layer repair, and slicing the representation keeps the parameter cost low.

  5. How to Merge Your Multimodal Models Over Time?

    cs.LG 2024-12 conditional novelty 6.0 of 10

    A systematic study of temporal model merging shows that initialization and deployment choices matter far more than the merging technique, with EMA-style weight interpolation as the best practice.

  6. WSM: Decay-Free Learning Rate Schedule via Checkpoint Merging for LLM Pre-training

    cs.CL 2025-07 conditional novelty 5.0 of 10

    Checkpoint merging during constant-LR training can replace LR decay and yields improved LLM benchmark scores over Warmup-Stable-Decay.

  7. Rethinking Mixture-of-Agents: Is Mixing Different Large Language Models Beneficial?

    cs.CL 2025-02 conditional novelty 5.0 of 10

    An ensemble built from repeated samples of a single strong LLM outperforms the standard multi-model Mixture-of-Agents on several benchmarks.

  8. If You Can't Use Them, Recycle Them: Optimizing Merging at Scale Mitigates Performance Tradeoffs

    cs.CL 2024-12 conditional novelty 5.0 of 10

    Optimizing linear merge weights on 16 104B generalist checkpoints via CMA-ES reduces task tradeoffs and beats individual models and simple merge baselines.

Pith tools