GraphWiz/LLaMA2-7B-RFT

1{2  "JK": "last",3  "_name_or_path": "/hpc2hdd/home/yli258/olddata/yuhanli/LLaMA_v2_ckpts/hf/Llama-2-7b-hf",4  "architectures": [5    "LlamaForCausalLM"6  ],7  "attention_bias": false,8  "attention_dropout": 0.0,9  "bos_token_id": 1,10  "dropout_ratio": 0,11  "end_token_id": 2,12  "eos_token_id": 2,13  "gnn_emb_dim": 300,14  "gnn_type": "gin",15  "graph_pooling": "mean",16  "hidden_act": "silu",17  "hidden_size": 4096,18  "initializer_range": 0.02,19  "intermediate_size": 11008,20  "max_position_embeddings": 4096,21  "model_type": "llama",22  "num_attention_heads": 32,23  "num_hidden_layers": 32,24  "num_key_value_heads": 32,25  "num_layer": 5,26  "num_tasks": 1,27  "pad_token_id": 2,28  "pretraining_tp": 1,29  "rms_norm_eps": 1e-05,30  "rope_scaling": null,31  "rope_theta": 10000.0,32  "tie_word_embeddings": false,33  "torch_dtype": "float16",34  "transformers_version": "4.36.2",35  "use_cache": true,36  "vocab_size": 3200037}38