eliza555beth2002/DeepSeek-R1-Distill-Text2SQL-Merged

1{2    "architectures": [3        "LlamaForCausalLM"4    ],5    "attention_bias": false,6    "attention_dropout": 0.0,7    "bos_token_id": 128000,8    "torch_dtype": "bfloat16",9    "eos_token_id": 128001,10    "head_dim": 128,11    "hidden_act": "silu",12    "hidden_size": 4096,13    "initializer_range": 0.02,14    "intermediate_size": 14336,15    "max_position_embeddings": 131072,16    "mlp_bias": false,17    "model_type": "llama",18    "num_attention_heads": 32,19    "num_hidden_layers": 32,20    "num_key_value_heads": 8,21    "pad_token_id": 128004,22    "pretraining_tp": 1,23    "rms_norm_eps": 1e-05,24    "rope_scaling": {25        "factor": 8.0,26        "high_freq_factor": 4.0,27        "low_freq_factor": 1.0,28        "original_max_position_embeddings": 8192,29        "rope_type": "llama3"30    },31    "rope_theta": 500000.0,32    "tie_word_embeddings": false,33    "transformers_version": "4.57.0",34    "unsloth_fixed": true,35    "unsloth_version": "2025.10.1",36    "use_cache": true,37    "vocab_size": 12825638}