config.json

{
  "hidden_size": 768,
  "num_layers": 12,
  "intermediate_size": 3072,
  "num_attention_heads": 12,
  "num_key_value_heads": 4,
  "max_position_embeddings": 1024,
  "vocab_size": 50257,
  "rope_theta": 10000.0,
  "rms_norm_eps": 1e-06,
  "hidden_act": "gelu_pytorch_tanh",
  "attention_dropout": 0.0,
  "hidden_dropout": 0.0,
  "tie_word_embeddings": false,
  "initializer_range": 0.02,
  "architectures": [
    "GemmaForCausalLM"
  ],
  "auto_map": {
    "AutoConfig": "model.GemmaConfig",
    "AutoModelForCausalLM": "model.GemmaForCausalLM"
  },
  "model_type": "gemma_custom",
  "torch_dtype": "float32"
}