NuminaMath-7B-TIR / eval_results.json
lewtun's picture
lewtun HF staff
Add AI-MO/deepseek-math-7b-sft-aimo_v51.2 checkpoint
167ee6b verified
raw
history blame
222 Bytes
{
"epoch": 4.0,
"eval_loss": 0.5301879048347473,
"eval_runtime": 0.5773,
"eval_samples": 99,
"eval_samples_per_second": 133.388,
"eval_steps_per_second": 3.465,
"perplexity": 1.699251576208271
}