long-t5-tglobal-xl-qmsum-wip / eval_results.json
Peter Szemraj
add post-training checkpoint
a0560e1
raw
history blame
325 Bytes
{
"epoch": 3.0,
"eval_gen_len": 80.8,
"eval_loss": 2.050546884536743,
"eval_rouge1": 35.3881,
"eval_rouge2": 11.509,
"eval_rougeL": 23.1543,
"eval_rougeLsum": 31.3295,
"eval_runtime": 365.9424,
"eval_samples": 100,
"eval_samples_per_second": 0.273,
"eval_steps_per_second": 0.273
}