Training Compute-Optimal Large Language Models

by   Jordan Hoffmann, et al.

We investigate the optimal model size and number of tokens for training a transformer language model under a given compute budget. We find that current large language models are significantly undertrained, a consequence of the recent focus on scaling language models whilst keeping the amount of training data constant. By training over language models ranging from 70 million to over 16 billion parameters on 5 to 500 billion tokens, we find that for compute-optimal training, the model size and the number of training tokens should be scaled equally: for every doubling of model size the number of training tokens should also be doubled. We test this hypothesis by training a predicted compute-optimal model, , that uses the same compute budget as but with 70B parameters and 4× more more data. uniformly and significantly outperforms (280B), GPT-3 (175B), Jurassic-1 (178B), and Megatron-Turing NLG (530B) on a large range of downstream evaluation tasks. This also means that uses substantially less compute for fine-tuning and inference, greatly facilitating downstream usage. As a highlight, reaches a state-of-the-art average accuracy of 67.5% on the MMLU benchmark, greater than a 7% improvement over .


page 1

page 2

page 3

page 4


Scaling Data-Constrained Language Models

The current trend of scaling language models involves increasing both pa...

An Information-Theoretic Analysis of Compute-Optimal Neural Scaling Laws

We study the compute-optimal trade-off between model and training data s...

Scaling Laws for Generative Mixed-Modal Language Models

Generative language models define distributions over sequences of tokens...

Reconsidering the Past: Optimizing Hidden States in Language Models

We present Hidden-State Optimization (HSO), a gradient-based method for ...

Meta-Learning Fast Weight Language Models

Dynamic evaluation of language models (LMs) adapts model parameters at t...

Staged Training for Transformer Language Models

The current standard approach to scaling transformer language models tra...

How BPE Affects Memorization in Transformers

Training data memorization in NLP can both be beneficial (e.g., closed-b...

Please sign up or login with your details

Forgot password? Click here to reset