facebook-llama/custom_code

1{2  "_name_or_path": "/Users/guynachshon/Documents/code/research/temp/glm_large_chinese",3  "architectures": [4    "GLMModel"5  ],6  "attention_dropout_prob": 0.1,7  "attention_scale": 1.0,8  "block_position_encoding": true,9  "checkpoint_activations": false,10  "checkpoint_num_layers": 1,11  "embedding_dropout_prob": 0.1,12  "hidden_size": 1024,13  "initializer_range": 0.02,14  "max_sequence_length": 1024,15  "model_type": "glm",16  "num_attention_heads": 16,17  "num_layers": 24,18  "output_dropout_prob": 0.1,19  "output_predict": true,20  "parallel_output": true,21  "pool_token": "cls",22  "relative_encoding": false,23  "spell_func": "lstm",24  "spell_length": null,25  "torch_dtype": "float32",26  "transformers_version": "4.35.2",27  "vocab_size": 5004828}29