{ "_name_or_path": "/workspace/pretraining/local-models/training/2024-02-05_16-59-29", "architectures": [ "MistralForCausalLM" ], "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 864, "initializer_range": 0.02, "intermediate_size": 2382, "max_position_embeddings": 1024, "model_type": "mistral", "num_attention_heads": 12, "num_hidden_layers": 18, "num_key_value_heads": 2, "rms_norm_eps": 1e-06, "rope_theta": 10000.0, "sliding_window": 4096, "tie_word_embeddings": true, "torch_dtype": "float32", "transformers_version": "4.34.1", "use_cache": true, "vocab_size": 50257 }