Pith. sign in

REVIEW 6 cited by

Value Augmented Sampling for Language Model Alignment and Personalization

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 2405.06639 v1 pith:NSNEQT2L submitted 2024-05-10 cs.LG cs.AIcs.CL

classification cs.LGcs.AIcs.CL
keywords valuellmsoptimizationpolicyadaptationaugmentedco-trainingcost
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Aligning Large Language Models (LLMs) to cater to different human preferences, learning new skills, and unlearning harmful behavior is an important problem. Search-based methods, such as Best-of-N or Monte-Carlo Tree Search, are performant, but impractical for LLM adaptation due to their high inference cost. On the other hand, using Reinforcement Learning (RL) for adaptation is computationally efficient, but performs worse due to the optimization challenges in co-training the value function and the policy. We present a new framework for reward optimization, Value Augmented Sampling (VAS), that can maximize different reward functions using data sampled from only the initial, frozen LLM. VAS solves for the optimal reward-maximizing policy without co-training the policy and the value function, making the optimization stable, outperforming established baselines, such as PPO and DPO, on standard benchmarks, and achieving comparable results to Best-of-128 with lower inference cost. Unlike existing RL methods that require changing the weights of the LLM, VAS does not require access to the weights of the pre-trained LLM. Thus, it can even adapt LLMs (e.g., ChatGPT), which are available only as APIs. In addition, our algorithm unlocks the new capability of composing several rewards and controlling the extent of each one during deployment time, paving the road ahead for the future of aligned, personalized LLMs.

Discussion (0). Sign in to comment.

Forward citations

Cited by 6 Pith papers

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

  1. Selective Safety Steering via Value-Filtered Decoding

    cs.LG 2026-05 unverdicted novelty 7.0 of 10

    Value-filtered decoding steers LLM outputs for safety at decoding time using a value criterion with an explicit bound on false interventions controlled by one threshold hyperparameter.

  2. Controlling Multimodal LLMs via Reward-guided Decoding

    cs.CV 2025-08 conditional novelty 6.0 of 10

    MRGD guides MLLM decoding with a learned hallucination reward and a detector-based recall reward, allowing users to trade off object precision, recall, and test-time compute while reducing object hallucinations on CHA...

  3. PICACO: Pluralistic In-Context Value Alignment of LLMs via Total Correlation Optimization

    cs.CL 2025-07 conditional novelty 6.0 of 10

    PICACO optimizes a meta-instruction by maximizing total correlation between intended values and LLM responses, and reports consistent though modest gains across five value sets and three target models.

  4. From Outcomes to Processes: Guiding PRM Learning from ORM for Inference-Time Alignment

    cs.CL 2025-06 conditional novelty 6.0 of 10

    SP-PRM trains process reward models from outcome reward models via score and preference consistency, improving reward-guided search on dialogue, summarization, and reasoning tasks.

  5. Safe Inference-Time Alignment via Lagrangian Reward Augmentation

    cs.LG 2026-07 conditional novelty 5.5 of 10

    Dualizing Safe RLHF yields a one-dimensional convex calibration of λ that defines a drop-in safety-aware reward for Best-of-N and token-level inference-time decoders.

  6. Test-time reward-guided alignment of language models by importance sampling on pre-logit space

    cs.LG 2025-10 conditional novelty 5.0 of 10

    AISP is a training-free decode-time alignment method: Gaussian-perturb LLM pre-logits, score sampled responses with a reward model, and iteratively shift the perturbation mean by reward-weighted importance sampling.

Pith tools