The authors report that, from training over 400 language models, they find that compute-optimal training scales model size and the number of training tokens equally, doubling tokens with every doubling of model size.
One recorded claim. The claim panel gives its qualification, the source passages with their place in each source, and how it was checked.
What it connects
Training Compute-Optimal Large Language Models → states its contribution → The authors report that, from training over 400 language models, they find that compute-optimal training scales model size and the number of training tokens equally, doubling tokens with every doubling of model size.
Where it appears
No collection page holds this claim. It is listed with its records in the explorer.