{ "architectures": [ "MistralForCausalLM" ], "attention_dropout": 0.0, "bos_token_id": null, "dtype": "float32", "eos_token_id": 1, "head_dim": null, "hidden_act": "silu", "hidden_size": 32, "initializer_range": 0.02, "intermediate_size": 256, "max_position_embeddings": 78, "model_type": "mistral", "num_attention_heads": 1, "num_hidden_layers": 2, "num_key_value_heads": 1, "pad_token_id": 2, "rms_norm_eps": 1e-06, "rope_theta": 10000.0, "sliding_window": 64, "tie_word_embeddings": false, "transformers_version": "4.57.3", "use_cache": false, "vocab_size": 1200 }