{
  "architectures": [
    "GPTS25ForCausalLM"
  ],
  "model_type": "gpt-s2.5",
  "vocab_size": 8192,
  "hidden_size": 256,
  "num_hidden_layers": 4,
  "num_attention_heads": 8,
  "num_key_value_heads": 2,
  "head_dim": 32,
  "intermediate_size": 768,
  "max_position_embeddings": 512,
  "rope_theta": 10000.0,
  "norm_type": "rmsnorm",
  "norm_eps": 1e-06,
  "mlp_activation": "swiglu",
  "tie_word_embeddings": true,
  "bias": false,
  "torch_dtype": "float32",
  "transformers_version": "4.50.0",
  "parameter_count": 5114112
}