pith. sign in

arxiv: 2009.14109 · v2 · pith:LPOQOCQMnew · submitted 2020-09-29 · 💻 cs.CL

Improving Low Compute Language Modeling with In-Domain Embedding Initialisation

classification 💻 cs.CL
keywords datain-domainlanguageembeddingsimproveinputmillionmodel
0
0 comments X
read the original abstract

Many NLP applications, such as biomedical data and technical support, have 10-100 million tokens of in-domain data and limited computational resources for learning from it. How should we train a language model in this scenario? Most language modeling research considers either a small dataset with a closed vocabulary (like the standard 1 million token Penn Treebank), or the whole web with byte-pair encoding. We show that for our target setting in English, initialising and freezing input embeddings using in-domain data can improve language model performance by providing a useful representation of rare words, and this pattern holds across several different domains. In the process, we show that the standard convention of tying input and output embeddings does not improve perplexity when initializing with embeddings trained on in-domain data.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

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

  1. Smarter, Better, Faster, Longer: A Modern Bidirectional Encoder for Fast, Memory Efficient, and Long Context Finetuning and Inference

    cs.CL 2024-12 unverdicted novelty 5.0

    ModernBERT is a new bidirectional encoder model achieving SOTA performance on diverse classification and retrieval benchmarks while offering superior speed and memory efficiency for long-context inference.