{ "tok": "out/tokenizer", "val": "out/val.bin", "base_bin": "out/train.bin", "base_split": 1020000000, "glob": "out2/*.bin", "p_synth": -1, "save": "teacher100m", "hidden": 768, "layers": 14, "heads": 12, "kv": 4, "ffn": 2048, "seq": 512, "micro": 32, "accum": 16, "tokens": 20000000000.0, "minutes": 295.0, "lr_muon": 0.02, "lr_adam": 0.006, "wd": 0.0, "warmup": 0.02, "decay": 0.25, "eval_every": 500, "save_every_min": 30, "teacher": "", "kd_alpha": 0.2, "mem_frac": 0, "compile": 1, "seed": 1337, "params": 91250176, "final_val": 2.132023814320564, "steps": 9621, "tokens_seen": 2522087424, "seconds": 17544.111047267914, "synth_B": 5.306053065, "edu_B": 16.31685136 }