{ "architectures": [ "LlamaForCausalLM" ], "model_type": "llama", "hidden_size": 8, "intermediate_size": 16, "num_attention_heads": 2, "num_key_value_heads": 2, "num_hidden_layers": 1, "vocab_size": 32, "max_position_embeddings": 64, "rms_norm_eps": 1e-06, "bos_token_id": 1, "eos_token_id": 2, "tie_word_embeddings": false, "torch_dtype": "float32" }