config.json

{
  "activation_dropout": 0.1,
  "architectures": [
    "SpeechLlamaForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.1,
  "bos_token_id": 0,
  "diffloss_d": 6,
  "diffloss_w": 1024,
  "dropout": 0.1,
  "eos_token_id": 2,
  "head_dim": 64,
  "hidden_act": "silu",
  "hidden_size": 1024,
  "initializer_range": 0.02,
  "intermediate_size": 2752,
  "max_position_embeddings": 2048,
  "mlp_bias": false,
  "model_type": "speech_llama",
  "noise_channels": 128,
  "num_attention_heads": 16,
  "num_hidden_layers": 12,
  "num_key_value_heads": 16,
  "pretraining_tp": 1,
  "rms_norm_eps": 1e-06,
  "rope_scaling": null,
  "rope_theta": 10000.0,
  "tie_word_embeddings": false,
  "torch_dtype": "float32",
  "training_cfg": 0.1,
  "transformers_version": "4.46.0.dev0",
  "use_cache": true,
  "vae_embed_dim": 128,
  "vocab_size": 16384
}