| { | |
| "name": "smoke", | |
| "seed": 0, | |
| "d_model": 1024, | |
| "depth": 32, | |
| "char_window": 256, | |
| "attn": "flex", | |
| "compile": true, | |
| "qk_norm": true, | |
| "seq_len": 8192, | |
| "micro_batch": 4, | |
| "grad_accum": 2, | |
| "lr": 0.0003, | |
| "wd": 0.1, | |
| "clip": 1.0, | |
| "warmup_frac": 0.04, | |
| "decay_frac": 0.2, | |
| "tier_weights": [ | |
| 1.0, | |
| 1.0, | |
| 0.3 | |
| ], | |
| "lam": 0.1, | |
| "total_steps": 150, | |
| "num_workers": 16, | |
| "log_every": 5, | |
| "ckpt_every": 100, | |
| "eval_every_ckpt": true, | |
| "eval_n": 64, | |
| "out_dir": "$STOICHEIA_DATA/runs/pretrain_smoke" | |
| } |