config.json

{
  "card": 2048,
  "n_q": 32,
  "dep_q": 0,
  "delays": [
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0
  ],
  "dim": 2048,
  "text_card": 8000,
  "existing_text_padding_id": 3,
  "num_heads": 16,
  "num_layers": 16,
  "hidden_scale": 4.125,
  "causal": true,
  "layer_scale": null,
  "context": 750,
  "max_period": 100000.0,
  "gating": "silu",
  "norm": "rms_norm_f32",
  "positional_embedding": "rope",
  "depformer_dim": 1024,
  "depformer_num_heads": 16,
  "depformer_num_layers": 6,
  "depformer_dim_feedforward": null,
  "depformer_multi_linear": true,
  "depformer_pos_emb": "none",
  "depformer_weights_per_step": true,
  "conditioners": {},
  "cross_attention": false,
  "model_id": {
    "sig": "70f8f0ea",
    "epoch": 500
  },
  "lm_gen_config": {
    "temp": 0.0,
    "temp_text": 0.0,
    "top_k": 250,
    "top_k_text": 50
  },
  "stt_config": {
    "audio_delay_seconds": 0.5,
    "audio_silence_prefix_seconds": 0.0
  },
  "model_type": "stt",
  "mimi_name": "mimi-pytorch-e351c8d8@125.safetensors",
  "tokenizer_name": "tokenizer_en_fr_audio_8000.model"
}