MrezaPRZ/sql-encoder

1{2  "_name_or_path": "deepseek-ai/deepseek-coder-1.3b-instruct",3  "architectures": [4    "LlamaForSequenceClassification"5  ],6  "attention_bias": false,7  "attention_dropout": 0.0,8  "bos_token_id": 32013,9  "eos_token_id": 32021,10  "hidden_act": "silu",11  "hidden_size": 2048,12  "id2label": {13    "0": "LABEL_0"14  },15  "initializer_range": 0.02,16  "intermediate_size": 5504,17  "label2id": {18    "LABEL_0": 019  },20  "max_position_embeddings": 16384,21  "model_type": "llama",22  "num_attention_heads": 16,23  "num_hidden_layers": 24,24  "num_key_value_heads": 16,25  "pad_token_id": 32014,26  "pretraining_tp": 1,27  "rms_norm_eps": 1e-06,28  "rope_scaling": {29    "factor": 4.0,30    "type": "linear"31  },32  "rope_theta": 100000,33  "tie_word_embeddings": false,34  "torch_dtype": "bfloat16",35  "transformers_version": "4.37.2",36  "use_cache": false,37  "vocab_size": 3225638}39