{ "architectures": ["MistralForCausalLM"], "model_type": "mistral", "hidden_size": 4096, "intermediate_size": 14336, "num_attention_heads": 32, "num_hidden_layers": 8, "tie_word_embeddings": false, "vocab_size": 32000, "max_position_embeddings": 4096, "attention_dropout": 0.1, "hidden_dropout": 0.1, "initializer_range": 0.02, "layer_norm_eps": 1e-5, "use_cache": true, "pad_token_id": 0, "bos_token_id": 1, "eos_token_id": 2, "torch_dtype": "bfloat16", "transformers_version": "4.38.1", "parameters": 673800000000000 }