config.json

{
  "vocab_size": 32768,
  "max_seq_len": 4096,
  "n_layer": 44,
  "n_embd": 768,
  "n_head": 12,
  "n_kv_head": 4,
  "mlp_hidden": 2048,
  "rope_base": 10000.0,
  "rms_eps": 1e-06,
  "qk_norm": true,
  "z_loss": 0.0,

  "model_type": "gouda-cheddar",
  "architectures": [
    "ChedarForCausalLM"
  ],

  "auto_map": {
    "AutoConfig": "configuration_cheddar.CheddarConfig",
    "AutoModelForCausalLM": "modeling_cheddar.CheddarForCausalLM"
  },

  "name": "Cheddar-2.0",
  "torch_dtype": "bfloat16",
  "tie_word_embeddings": true
}