Mitigating Frequency Bias and Anisotropy in Language Model Pre-Training with Syntactic Smoothing

• URL: http://arxiv.org/abs/2410.11462v1
• Date: Tue, 15 Oct 2024 10:09:57 GMT
• Title: Mitigating Frequency Bias and Anisotropy in Language Model Pre-Training with Syntactic Smoothing
• Authors: Richard Diehl Martinez, Zebulon Goriely, Andrew Caines, Paula Buttery, Lisa Beinborn,
• Abstract summary: We introduce a method for quantifying the frequency bias of a language model. We then present a method for reducing the frequency bias of a language model by inducing a syntactic prior over token representations during pre-training. This approach results in better performance on infrequent English tokens and a decrease in anisotropy.
• Score: 6.726629754291751
• License:
• Abstract: Language models strongly rely on frequency information because they maximize the likelihood of tokens during pre-training. As a consequence, language models tend to not generalize well to tokens that are seldom seen during training. Moreover, maximum likelihood training has been discovered to give rise to anisotropy: representations of tokens in a model tend to cluster tightly in a high-dimensional cone, rather than spreading out over their representational capacity. Our work introduces a method for quantifying the frequency bias of a language model by assessing sentence-level perplexity with respect to token-level frequency. We then present a method for reducing the frequency bias of a language model by inducing a syntactic prior over token representations during pre-training. Our Syntactic Smoothing method adjusts the maximum likelihood objective function to distribute the learning signal to syntactically similar tokens. This approach results in better performance on infrequent English tokens and a decrease in anisotropy. We empirically show that the degree of anisotropy in a model correlates with its frequency bias.

Related papers

• The Fair Language Model Paradox [19.439996884827448]
  Large Language Models (LLMs) are widely deployed in real-world applications, yet little is known about their training dynamics at the token level. We show that as weight decay increases, low-frequency tokens are disproportionately depreciated. This is particularly concerning, as these neglected low-frequency tokens represent the vast majority of the token distribution in most languages.
  arXiv  Detail & Related papers  (2024-10-15T18:47:12Z)
• 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)
• Autoregressive Speech Synthesis without Vector Quantization [135.4776759536272]
  We present MELLE, a novel continuous-valued tokens based language modeling approach for text to speech synthesis (TTS) MELLE autoregressively generates continuous mel-spectrogram frames directly from text condition.
  arXiv  Detail & Related papers  (2024-07-11T14:36:53Z)
• Understanding and Mitigating Tokenization Bias in Language Models [6.418593476658017]
  State-of-the-art language models are autoregressive and operate on subword units known as tokens. We show that popular encoding schemes induce a sampling bias that cannot be mitigated with more training or data. We propose a novel algorithm to obtain unbiased estimates from any language model trained on tokenized data.
  arXiv  Detail & Related papers  (2024-06-24T17:38:02Z)
• SpeechAlign: Aligning Speech Generation to Human Preferences [51.684183257809075]
  We introduce SpeechAlign, an iterative self-improvement strategy that aligns speech language models to human preferences. We show that SpeechAlign can bridge the distribution gap and facilitate continuous self-improvement of the speech language model.
  arXiv  Detail & Related papers  (2024-04-08T15:21:17Z)
• Collaborative decoding of critical tokens for boosting factuality of large language models [57.504894664689]
  Finetuned and aligned models show improved abilities of instruction following and safe generation. The common practice of using sampling during generation also increases chances of hallucination. We introduce a collaborative decoding framework to harness the high factuality within pretrained models through the concept of critical tokens.
  arXiv  Detail & Related papers  (2024-02-28T01:53:37Z)
• MiLe Loss: a New Loss for Mitigating the Bias of Learning Difficulties in Generative Language Models [40.992566245706996]
  We propose a MiLe Loss function for mitigating the bias of learning difficulties with tokens. We train generative language models at different scales of 468M, 1.2B, and 6.7B parameters. Experiments reveal that models incorporating the proposed MiLe Loss can gain consistent performance improvement on downstream benchmarks.
  arXiv  Detail & Related papers  (2023-10-30T13:33:21Z)
• A Natural Bias for Language Generation Models [31.44752136404971]
  We show that we can endow standard neural language generation models with a separate module that reflects unigram frequency statistics as prior knowledge. We use neural machine translation as a test bed for this simple technique and observe that it: (i) improves learning efficiency; (ii) achieves better overall performance; and perhaps most importantly: appears to disentangle strong frequency effects.
  arXiv  Detail & Related papers  (2022-12-19T18:14:36Z)
• Quark: Controllable Text Generation with Reinforced Unlearning [68.07749519374089]
  Large-scale language models often learn behaviors that are misaligned with user expectations. We introduce Quantized Reward Konditioning (Quark), an algorithm for optimizing a reward function that quantifies an (un)wanted property. For unlearning toxicity, negative sentiment, and repetition, our experiments show that Quark outperforms both strong baselines and state-of-the-art reinforcement learning methods.
  arXiv  Detail & Related papers  (2022-05-26T21:11:51Z)
• Multi-timescale Representation Learning in LSTM Language Models [69.98840820213937]
  Language models must capture statistical dependencies between words at timescales ranging from very short to very long. We derived a theory for how the memory gating mechanism in long short-term memory language models can capture power law decay. Experiments showed that LSTM language models trained on natural English text learn to approximate this theoretical distribution.
  arXiv  Detail & Related papers  (2020-09-27T02:13:38Z)

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.