training dict |
|---|
{
"sequence_length": 1024,
"batch_size": 16,
"gradient_accumulation_steps": 1,
"learning_rate": 0.0003,
"min_lr_ratio": 0.1,
"weight_decay": 0.1,
"adam_beta1": 0.9,
"adam_beta2": 0.95,
"adam_epsilon": 1e-8,
"max_grad_norm": 1,
"seed": 1337,
"gradient_checkpointing": true,
"attention_backend": ... |
No dataset card yet
- Downloads last month
- 5,083