Pith. sign in

REVIEW 11 cited by

Neural Text Generation with Unlikelihood Training

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 1908.04319 v2 pith:EAIFARRA submitted 2019-08-12 cs.LG cs.CLstat.ML

classification cs.LGcs.CLstat.ML
keywords trainingbeammodelstandardtextunlikelihooddecodingdull
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
abstract

Neural text generation is a key tool in natural language applications, but it is well known there are major problems at its core. In particular, standard likelihood training and decoding leads to dull and repetitive outputs. While some post-hoc fixes have been proposed, in particular top-$k$ and nucleus sampling, they do not address the fact that the token-level probabilities predicted by the model are poor. In this paper we show that the likelihood objective itself is at fault, resulting in a model that assigns too much probability to sequences containing repeats and frequent words, unlike those from the human training distribution. We propose a new objective, unlikelihood training, which forces unlikely generations to be assigned lower probability by the model. We show that both token and sequence level unlikelihood training give less repetitive, less dull text while maintaining perplexity, giving superior generations using standard greedy or beam search. According to human evaluations, our approach with standard beam search also outperforms the currently popular decoding methods of nucleus sampling or beam blocking, thus providing a strong alternative to existing techniques.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 11 Pith papers

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

  1. MentalThink: Shaping Thoughts in Mental SVG World

    cs.AI 2026-07 conditional novelty 7.0 of 10

    MLLMs that generate and render SVG sketches as multi-turn intermediate reasoning steps reach 55.1% on VSIBench and 76.0% on MindCube, far above the Qwen2.5-VL-7B backbone.

  2. On the Fitness Landscape in the $NK$ Model

    math.PR 2025-08 unverdicted novelty 7.0 of 10

    For the NK fitness landscape with K/N tending to alpha, exact limits for free energy and maximum fitness are identified, together with the geometry of near-fittest peaks.

  3. Token-Level Off-Policy Learning for Faithful Generation Under Distribution Shift

    cs.CL 2026-07 conditional novelty 6.0 of 10

    A token-level correctness classifier trained with LoRA and then merged into the model boosts out-of-distribution factuality in summarization and translation.

  4. Bayesian Repetition Penalty: A Principled Adjacent-Conditional Framework for Reversing Attention Collapse in Autoregressive Language Models

    cs.AI 2026-07 conditional novelty 6.0 of 10

    A closed-form ratio of binomial conditional probabilities, comparing a token's recent count with its corpus rate, defines a self-normalizing logit bias that suppresses repetitive loops in autoregressive language models.

  5. LP-SFT: Local-Preserving Supervised Fine-Tuning via Multimodal Entropy Structure

    cs.CL 2026-07 conditional novelty 6.0 of 10

    Preserving adaptive non-label local structure from the base model during SFT improves the pass@1 vs pass@k trade-off and reduces catastrophic forgetting versus vanilla cross-entropy and recent SFT variants.

  6. Between Suppression and Collapse: Evaluating Narrative Unlearning with LENS

    cs.CL 2026-06 conditional novelty 6.0 of 10

    LENS measures narrative unlearning at four resistance levels and shows existing objectives can suppress a target frame at selected middle checkpoints, with transfer to attributed/contrastive prompts and rare residual ...

  7. Soft Reasoning: Navigating Solution Spaces in Large Language Models through Controlled Embedding Exploration

    cs.CL 2025-05 conditional novelty 6.0 of 10

    Soft Reasoning improves LLM accuracy on reasoning benchmarks by perturbing the embedding of the first generated token and searching the perturbation space with Bayesian optimization guided by the model's own verifier.

  8. Avoidance Decoding for Diverse Multi-Branch Story Generation

    cs.CL 2025-09 conditional novelty 5.0 of 10

    Avoidance Decoding penalizes token choices that resemble previously generated story branches, using a hybrid concept-level and narrative-level similarity penalty, and reports large diversity gains across several LLMs.

  9. Breaking the Likelihood Trap: Consistent Generative Recommendation with Graph-structured Model

    cs.IR 2025-10 conditional novelty 4.0 of 10

    CONGRATS uses a DAG-structured positional decoder and evaluator-in-the-loop training to generate more diverse and accurate recommendation lists, showing offline and Kuaishou A/B gains.

  10. GOLFer: Smaller LM-Generated Documents Hallucination Filter & Combiner for Query Expansion in Information Retrieval

    cs.IR 2025-06 conditional novelty 4.0 of 10

    GOLFer filters hallucinated sentences from small-LM-generated hypothetical documents and reweights the rest into the query, improving retrieval at lower cost than large LLM expansion.

  11. Multi-Agent Language Models: Advancing Cooperation, Coordination, and Adaptation

    cs.CL 2025-06 conditional novelty 3.0 of 10

    A thesis proposal repurposing two prior papers on LM agents for text games, framed as a path to theory-of-mind AI, with no new theory-of-mind evidence.

Pith tools