{
  "vocab_size": 50304,
  "max_seq_len": 1024,
  "d_model": 1024,
  "n_layers": 24,
  "n_heads": 16,
  "n_kv_heads": 4,
  "ffn_hidden_dim": 2730,
  "dropout": 0.0,
  "norm_eps": 1e-06,
  "rope_theta": 10000.0,
  "bias": false,
  "tie_weights": true,
  "use_flash": true,
  "use_qk_norm": true,
  "use_diff_attn": true,
  "use_mhc": false,
  "n_streams": 2,
  "mhc_every_n_layers": 1,
  "use_xsa": false,
  "architecture": "JarvisLM-GPT",
  "weights": "ema",
  "dtype": "bfloat16",
  "parameters": 315758848,
  "training_step": 10500,
  "source_checkpoint": "last.pt",
  "run": "jarvislm-v2-modern-315m-h200-5.5b"
}