Language Model Evaluation Beyond Perplexity

• URL: http://arxiv.org/abs/2106.00085v2
• Date: Wed, 2 Jun 2021 03:41:11 GMT
• Title: Language Model Evaluation Beyond Perplexity
• Authors: Clara Meister, Ryan Cotterell
• Abstract summary: We analyze whether text generated from language models exhibits the statistical tendencies present in the human-generated text on which they were trained. We find that neural language models appear to learn only a subset of the tendencies considered, but align much more closely with empirical trends than proposed theoretical distributions.
• Score: 47.268323020210175
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We propose an alternate approach to quantifying how well language models learn natural language: we ask how well they match the statistical tendencies of natural language. To answer this question, we analyze whether text generated from language models exhibits the statistical tendencies present in the human-generated text on which they were trained. We provide a framework--paired with significance tests--for evaluating the fit of language models to these trends. We find that neural language models appear to learn only a subset of the tendencies considered, but align much more closely with empirical trends than proposed theoretical distributions (when present). Further, the fit to different distributions is highly-dependent on both model architecture and generation strategy. As concrete examples, text generated under the nucleus sampling scheme adheres more closely to the type--token relationship of natural language than text produced using standard ancestral sampling; text from LSTMs reflects the natural language distributions over length, stopwords, and symbols surprisingly well.

Related papers

• Linguistically Grounded Analysis of Language Models using Shapley Head Values [2.914115079173979]
  We investigate the processing of morphosyntactic phenomena by leveraging a recently proposed method for probing language models via Shapley Head Values (SHVs) Using the English language BLiMP dataset, we test our approach on two widely used models, BERT and RoBERTa, and compare how linguistic constructions are handled. Our results show that SHV-based attributions reveal distinct patterns across both models, providing insights into how language models organize and process linguistic information.
  arXiv  Detail & Related papers  (2024-10-17T09:48:08Z)
• On the Proper Treatment of Tokenization in Psycholinguistics [53.960910019072436]
  The paper argues that token-level language models should be marginalized into character-level language models before they are used in psycholinguistic studies. We find various focal areas whose surprisal is a better psychometric predictor than the surprisal of the region of interest itself.
  arXiv  Detail & Related papers  (2024-10-03T17:18:03Z)
• What do Large Language Models Learn beyond Language? [10.9650651784511]
  We find that pretrained models significantly outperform comparable non-pretrained neural models. Experiments surprisingly reveal that the positive effects of pre-training persist even when pretraining on multi-lingual text or computer code. Our findings suggest a hitherto unexplored deep connection between pre-training and inductive learning abilities of language models.
  arXiv  Detail & Related papers  (2022-10-21T23:43:13Z)
• Model Criticism for Long-Form Text Generation [113.13900836015122]
  We apply a statistical tool, model criticism in latent space, to evaluate the high-level structure of generated text. We perform experiments on three representative aspects of high-level discourse -- coherence, coreference, and topicality. We find that transformer-based language models are able to capture topical structures but have a harder time maintaining structural coherence or modeling coreference.
  arXiv  Detail & Related papers  (2022-10-16T04:35:58Z)
• Transparency Helps Reveal When Language Models Learn Meaning [71.96920839263457]
  Our systematic experiments with synthetic data reveal that, with languages where all expressions have context-independent denotations, both autoregressive and masked language models learn to emulate semantic relations between expressions. Turning to natural language, our experiments with a specific phenomenon -- referential opacity -- add to the growing body of evidence that current language models do not well-represent natural language semantics.
  arXiv  Detail & Related papers  (2022-10-14T02:35:19Z)
• Integrating Linguistic Theory and Neural Language Models [2.870517198186329]
  I present several case studies to illustrate how theoretical linguistics and neural language models are still relevant to each other. This thesis contributes three studies that explore different aspects of the syntax-semantics interface in language models.
  arXiv  Detail & Related papers  (2022-07-20T04:20:46Z)
• Dependency-based Mixture Language Models [53.152011258252315]
  We introduce the Dependency-based Mixture Language Models. In detail, we first train neural language models with a novel dependency modeling objective. We then formulate the next-token probability by mixing the previous dependency modeling probability distributions with self-attention.
  arXiv  Detail & Related papers  (2022-03-19T06:28:30Z)
• Uncovering More Shallow Heuristics: Probing the Natural Language Inference Capacities of Transformer-Based Pre-Trained Language Models Using Syllogistic Patterns [9.031827448667086]
  We explore the shallows used by transformer-based pre-trained language models (PLMs) that are fine-tuned for natural language inference (NLI) We find evidence that the models rely heavily on certain shallows, picking up on symmetries and asymmetries between premise and hypothesis.
  arXiv  Detail & Related papers  (2022-01-19T14:15:41Z)
• Language Models are Few-shot Multilingual Learners [66.11011385895195]
  We evaluate the multilingual skills of the GPT and T5 models in conducting multi-class classification on non-English languages. We show that, given a few English examples as context, pre-trained language models can predict not only English test samples but also non-English ones.
  arXiv  Detail & Related papers  (2021-09-16T03:08:22Z)
• Stochastic Natural Language Generation Using Dependency Information [0.7995360025953929]
  This article presents a corpus-based model for generating natural language text. Our model encodes dependency relations from training data through a feature set, then produces a new dependency tree for a given meaning representation. We show that our model produces high-quality utterances in aspects of informativeness and naturalness as well as quality.
  arXiv  Detail & Related papers  (2020-01-12T09:40:11Z)

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.