| { | |
| "baseline": { | |
| "learning_rate": 0.0003, | |
| "batch_size": 32 | |
| }, | |
| "large-lr": { | |
| "learning_rate": 0.001, | |
| "batch_size": 32 | |
| }, | |
| "cosine-sched": { | |
| "learning_rate": 0.0005, | |
| "batch_size": 64 | |
| }, | |
| "warmup-long": { | |
| "learning_rate": 0.0005, | |
| "batch_size": 128 | |
| }, | |
| "dropout-0.2": { | |
| "learning_rate": 0.0003, | |
| "batch_size": 64 | |
| } | |
| } |