{ "run_name": "T0_fixed_s2", "tokenizer": "fixed", "patch": 32, "ctx_span": 4096, "pyramid_levels": [], "adaptive_hist_tokens": 96, "fine_span": 1024, "d_model": 256, "n_layers": 4, "n_heads": 4, "train_kmin": 4, "train_kmax": 32, "cpm_cmax": 16, "cpm_pmax": 0.4, "cpm_span_max": 4, "steps": 30000, "batch": 64, "n_variates": 8, "lr": 0.001, "normuon_lr": 0.0008, "weight_decay": 0.0001, "grad_clip": 1.0, "warmup_frac": 0.03, "decay_frac": 0.1, "seed": 2, "data_seed": 2, "pool": "short", "eval_every": 500, "ckpt_every": 2500, "gift_ckpts": [ 15000, 30000 ] }