{
  "vocab_size": 8192,
  "d_model": 512,
  "n_layers": 8,
  "head_dim": 64,
  "n_experts": 64,
  "routed_experts": 4,
  "context": 1024,
  "rope_base": 1024.0,
  "qb_bias_space": "logit",
  "attention_backend": "torch",
  "expert_backend": "torch",
  "norm_lm": 5.0143386949929845,
  "norm_max": 101.90990763664634,
  "lr_lm_head": 0.003915577057540501,
  "lr_matrix": 0.023986659477334044,
  "lr_embed": 0.5154617146617039,
  "lr_norm": 0.643620249153108,
  "adam_beta1": 0.9591653429217373,
  "adam_beta2": 0.9570474948230828,
  "muon_beta": 0.9155117103190143,
  "matrix_rms": 0.7216605510948313,
  "embed_init_std": 0.17257427654581495,
  "embed_norm": 5.981044067725889,
  "residual_scale": 2.4049379753472047,
  "norm_init": 1.06850357265372,
  "router_temperature": 1.0961720778130266,
  "warmup_steps": 106,
  "micro_batch_size": 128,
  "global_batch_tokens": 524288,
  "total_tokens": 1500000000,
  "compile": true,
  "data_dir": "data/fineweb10B_sp8192",
  "eval_every": 0,
  "save_every": 0,
  "qb_diagnostics": false,
  "qb_logit_histogram": "compact",
  "qb_load_profile": "uniform",
  "qb_lower_slack": 0.0,
  "qb_upper_slack": 0.0,
  "qb_inverse_rank_offset": null
}