{
  "architectures": [
    "Gemma3ForCausalLM"
  ],
  "model_type": "gemma3",
  "torch_dtype": "bfloat16",
  "vocab_size": 262208,
  "hidden_size": 2560,
  "intermediate_size": 10240,
  "num_hidden_layers": 34,
  "num_attention_heads": 8,
  "num_key_value_heads": 4,
  "head_dim": 256,
  "hidden_activation": "gelu_pytorch_tanh",
  "max_position_embeddings": 8192,
  "rms_norm_eps": 1e-06,
  "tie_word_embeddings": true,
  "transformers_version": "4.49.0"
}