Pith. sign in

REVIEW 9 cited by

Sequential Monte Carlo Steering of Large Language Models using Probabilistic Programs

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 2306.03081 v2 pith:SRRECEPP submitted 2023-06-05 cs.AI cs.CLcs.PLstat.CO

classification cs.AIcs.CLcs.PLstat.CO
keywords languageprobabilisticsteeringcarlogenerationllmsmodelsmonte
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Even after fine-tuning and reinforcement learning, large language models (LLMs) can be difficult, if not impossible, to control reliably with prompts alone. We propose a new inference-time approach to enforcing syntactic and semantic constraints on the outputs of LLMs, called sequential Monte Carlo (SMC) steering. The key idea is to specify language generation tasks as posterior inference problems in a class of discrete probabilistic sequence models, and replace standard decoding with sequential Monte Carlo inference. For a computational cost similar to that of beam search, SMC can steer LLMs to solve diverse tasks, including infilling, generation under syntactic constraints, and prompt intersection. To facilitate experimentation with SMC steering, we present a probabilistic programming library, LLaMPPL (https://github.com/probcomp/hfppl), for concisely specifying new generation tasks as language model probabilistic programs, and automating steering of LLaMA-family Transformers.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 9 Pith papers

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

  1. Imprompt: A Language Framework for Prompt Programming

    cs.AI 2026-07 conditional novelty 7.0 of 10

    Prompt programs can be typed like programs and compiled in different ways; Imprompt formalizes two such compilers and tests them on tax-law reasoning and data-labeling tasks.

  2. Depth-Entropy Guided Sampling for Training-Free LLM Reasoning

    cs.LG 2026-06 conditional novelty 7.0 of 10

    Layer-wise entropy collapse depth is a weak per-candidate signal that, compounded inside MCMC power sampling, yields state-of-the-art training-free LLM reasoning accuracy.

  3. Reinforced sequential Monte Carlo for amortised sampling

    cs.LG 2025-10 conditional novelty 6.0 of 10

    A method that trains neural samplers using SMC-collected off-policy samples and an importance-weighted replay buffer improves mode coverage on multi-modal targets.

  4. Composable Effect Handling for Programming LLM-integrated Scripts

    cs.PL 2025-07 conditional novelty 6.0 of 10

    Effect handlers can separate LLM script workflow from I/O and concurrency, enabling a reported 10.88x average speedup in a Tree-of-Thoughts case study.

  5. Syntax Repair as Language Intersection

    cs.FL 2025-07 conditional novelty 6.0 of 10

    Bounded syntax repair is formalized as the intersection of a context-free grammar with an acyclic Levenshtein automaton, and the paper shows this intersection can be built, enumerated, and reranked efficiently.

  6. Integrating Neural and Symbolic Components in a Model of Pragmatic Question-Answering

    cs.CL 2025-06 conditional novelty 6.0 of 10

    A neuro-symbolic Rational Speech Act model with LLM proposers and evaluators predicts human question-answer patterns about as well as the fully hand-specified probabilistic model.

  7. Uncertainty-Aware Simulation-Based Inference for Operations Research with Large Language Models

    cs.LG 2026-07 conditional novelty 5.0 of 10

    Lookahead resampling with entropy- and power-based rewards steers LLM decoding toward OR formulations whose short simulated continuations are most concentrated, giving reported pass@1 gains not yet separated from adde...

  8. Rare Event Analysis of Large Language Models

    cs.LG 2026-02 conditional novelty 5.0 of 10

    Using annealed transition path sampling plus MBAR reweighting, the authors estimate TinyStories-8M completion probabilities for extreme ARI and log-probability values that are unobservable by direct sampling.

  9. Syntactic Control of Language Models by Posterior Inference

    cs.CL 2025-06 conditional novelty 4.0 of 10

    SMC sampling with a tetratagger shaping function and a POS-bigram proposal lifts constituency-tree match from 12-35% to about 93% F1 on GPT2-large and Llama3-8B.

Pith tools