Transformer Training Compute

Calculate transformer training compute from multiple technical inputs.

How this calculator works

Enter model parameters (b), training tokens (t), hardware efficiency (%). Select Calculate to apply the displayed formula and review the labeled results.

Formula / method

Training FLOPs ≈ 6 × parameters × training tokens

Worked example

Use the documented default inputs to model transformer training compute.

Assumptions and limitations

Inputs represent stable average operating conditions and the displayed equation applies.

This is a planning model; real systems and measurements can require domain-specific corrections.