| { | |
| "iteration": 50000, | |
| "best_val_loss": 5.49131965637207, | |
| "early_stopping_counter": 10, | |
| "learning_rate": 6e-06, | |
| "batch_size": 12, | |
| "block_size": 1024, | |
| "model_args": { | |
| "n_layer": 24, | |
| "n_head": 16, | |
| "n_embd": 1024, | |
| "block_size": 1024, | |
| "bias": false, | |
| "vocab_size": 4096, | |
| "dropout": 0.0 | |
| } | |
| } |