Saved in:
Bibliographic Details
Main Authors: Zhang, Xiao, Li, Miao, Wu, Ji
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2406.01976
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909215961907200
author Zhang, Xiao
Li, Miao
Wu, Ji
author_facet Zhang, Xiao
Li, Miao
Wu, Ji
contents Language models can learn sophisticated language understanding skills from fitting raw text. They also unselectively learn useless corpus statistics and biases, especially during finetuning on domain-specific corpora. In this paper, we propose a simple modification to causal language modeling called conditional finetuning, which performs language modeling conditioned on a context. We show that a context can "explain away" certain corpus statistics and make the model avoid learning them. In this fashion, conditional finetuning achieves selective learning from a corpus, learning knowledge useful for downstream tasks while avoiding learning useless corpus statistics like topic biases. This selective learning effect leads to less forgetting and better stability-plasticity tradeoff in domain finetuning, potentially benefitting lifelong learning with language models.
format Preprint
id arxiv_https___arxiv_org_abs_2406_01976
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Conditional Language Learning with Context
Zhang, Xiao
Li, Miao
Wu, Ji
Computation and Language
Language models can learn sophisticated language understanding skills from fitting raw text. They also unselectively learn useless corpus statistics and biases, especially during finetuning on domain-specific corpora. In this paper, we propose a simple modification to causal language modeling called conditional finetuning, which performs language modeling conditioned on a context. We show that a context can "explain away" certain corpus statistics and make the model avoid learning them. In this fashion, conditional finetuning achieves selective learning from a corpus, learning knowledge useful for downstream tasks while avoiding learning useless corpus statistics like topic biases. This selective learning effect leads to less forgetting and better stability-plasticity tradeoff in domain finetuning, potentially benefitting lifelong learning with language models.
title Conditional Language Learning with Context
topic Computation and Language
url https://arxiv.org/abs/2406.01976