Transformers
PyTorch
Graphcore
bert
Generated from Trainer
Inference Endpoints
bert-large-uncased / all_results.json
sergiopperez's picture
Update BERT large uncased checkpoint after running phase 1 (SL 128) and phase 2 (SL 512)
387585b
raw
history blame
203 Bytes
{
"epoch": 2.04,
"train_loss": 0.02294661615032911,
"train_runtime": 3034.1773,
"train_samples": 16407928,
"train_samples_per_second": 11004.826,
"train_steps_per_second": 0.672
}