arXiv:2306.05426cs.LG
SequenceMatch: Imitation Learning for Autoregressive Sequence Modelling with Backtracking
Abstract
In many domains, autoregressive models can attain high likelihood on the task of predicting the next observation. However, this maximum-likelihood (MLE) objective does not necessarily match a downstream use-case of autoregressively generating high-quality sequences. The MLE objective weights sequences proportionally to their frequency under the data distribution, with no guidance for the model's behaviour out of distribution (OOD): leading to compounding error during autoregressive generation. In order to address this compounding error problem, we formulate sequence generation as an imitation learning (IL) problem. This allows us to minimize a variety of divergences between the distribution of sequences generated by an autoregressive model and sequences from a dataset, including divergences with weight on OOD generated sequences. The IL framework also allows us to incorporate backtracking by introducing a backspace action into the generation process. This further mitigates the compounding error problem by allowing the model to revert a sampled token if it takes the sequence OOD. Our resulting method, SequenceMatch, can be implemented without adversarial training or architectural changes. We identify the SequenceMatch-$χ^2$ divergence as a more suitable training objective for autoregressive models which are used for generation. We show that empirically, SequenceMatch training leads to improvements over MLE on text generation with language models and arithmetic.
There's been a few attempts at training a backspace token, though. e.g.: https://arxiv.org/abs/2502.04404 https://arxiv.org/abs/2306.05426
ijkView on HNParent context
There have been attempts to give LLMs backspace tokens. Since no frontier model uses it I can only guess it doesn't scale as well as just letting it correct itself in COT https://arxiv.org/abs/2306.05426
krackersView on HNParent context