Skip to content
RED SAGSkills
hard20 pts · first answer countsgpu-computingperformancedistributed-training

Computing MFU for a training run

You train a 7B model at 2,500 tokens per second per A100 (312 TFLOPS BF16 peak). Using 6 × parameters FLOPs per token, what is the model FLOPs utilisation (MFU)?

Choose one answer

Discussion 0

Join free to post in the discussion.

Related questions

More GPUs and distributed training questionsMore Performance and debugging questionsMore Distributed Training questions