BERTIN: Efficient Pre-Training of a Spanish Language Model using Perplexity Sampling
This work enables small teams to train language models more efficiently on a limited budget, addressing a resource bottleneck in NLP.
The authors tackled the problem of inefficient pre-training of large language models by introducing perplexity sampling, a data-centric technique that reduces pre-training steps by half and data usage by one-fifth while achieving comparable or better results on certain tasks.
The pre-training of large language models usually requires massive amounts of resources, both in terms of computation and data. Frequently used web sources such as Common Crawl might contain enough noise to make this pre-training sub-optimal. In this work, we experiment with different sampling methods from the Spanish version of mC4, and present a novel data-centric technique which we name $\textit{perplexity sampling}$ that enables the pre-training of language models in roughly half the amount of steps and using one fifth of the data. The resulting models are comparable to the current state-of-the-art, and even achieve better results for certain tasks. Our work is proof of the versatility of Transformers, and paves the way for small teams to train their models on a limited budget. Our models are available at this $\href{https://huggingface.co/bertin-project}{URL}$.