R6_k2048_hz_nat_s2/config.json

{
  "run_name": "R6_k2048_hz_nat_s2",
  "tokenizer": "vq",
  "patch": 32,
  "ctx_span": 4096,
  "fine_span": 1024,
  "enc_hidden": 64,
  "fsq_levels": [
    8,
    6,
    5,
    5,
    5
  ],
  "vq_codes": 2048,
  "vq_dim": 64,
  "vq_beta": 0.25,
  "vq_decay": 0.99,
  "vq_entropy_w": 0.0,
  "vq_warm_steps": 2000,
  "vq_warm_decay": 0.9,
  "aux_code_ce": 0.2,
  "aux_code_offsets": [
    1
  ],
  "aux_side_codes": false,
  "ce_smooth_eps": 0.0,
  "ce_smooth_k": 8,
  "aux_horizon_ce": 0.2,
  "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": [
    7500,
    15000,
    22500,
    30000
  ]
}