{ "architectures": [ "MistralForCausalLM" ], "attention_dropout": 0.0, "bos_token_id": null, "dtype": "float32", "eos_token_id": 14001, "head_dim": null, "hidden_act": "silu", "hidden_size": 384, "initializer_range": 0.02, "intermediate_size": 1536, "max_position_embeddings": 640, "model_type": "mistral", "num_attention_heads": 6, "num_hidden_layers": 8, "num_key_value_heads": 2, "pad_token_id": 14000, "rms_norm_eps": 1e-06, "rope_theta": 10000.0, "sliding_window": 640, "tie_embeddings": true, "tie_word_embeddings": false, "transformers_version": "4.57.4", "use_cache": false, "vocab_size": 14002 }