| { | |
| "data": "data/pile_10B_seed1234.bin", | |
| "out_dir": "checkpoints/long_warmup", | |
| "model_size": "31M", | |
| "readout_lr_mult": 1.0, | |
| "peak_lr": 0.001, | |
| "weight_decay": 0.1, | |
| "wu_weight_decay": -1.0, | |
| "warmup_steps": 2860, | |
| "global_batch": 1024, | |
| "micro_batch": 8, | |
| "max_tokens": 10000000000.0, | |
| "max_steps": 0, | |
| "max_hours": 11.0, | |
| "val_blocks": 256, | |
| "seed": 0, | |
| "data_seed": 1234, | |
| "precision": "fp16", | |
| "log_every": 20, | |
| "latest_every": 100, | |
| "n_params": 30494720, | |
| "tokens_per_step": 2097152, | |
| "wu_weight_decay_effective": 0.1 | |
| } |