{ | |
"model_type": "CustomModel", | |
"architecture": "CustomModel", | |
"input_size": 512, | |
"hidden_size": 128, | |
"output_size": 768, | |
"vocab_size": 30522, // Example vocab size, adjust according to your tokenizer's vocabulary | |
"layer_norm_epsilon": 1e-12, | |
"hidden_dropout_prob": 0.1, | |
"num_attention_heads": 12, // Adjust if your model uses attention mechanisms | |
"attention_probs_dropout_prob": 0.1, | |
"intermediate_size": 3072, // Example size, adjust based on your model's architecture | |
"num_hidden_layers": 12, // Adjust based on your model's depth | |
"initializer_range": 0.02 | |
} | |