{ | |
"vocab_size": 32000, | |
"hidden_size": 512, | |
"num_heads": 8, | |
"num_encoder_layers": 6, | |
"num_decoder_layers": 6, | |
"dropout_rate": 0.1, | |
"max_position_embeddings": 512, | |
"activation_function": "gelu", | |
"initializer_range": 0.02 | |
} | |
{ | |
"vocab_size": 32000, | |
"hidden_size": 512, | |
"num_heads": 8, | |
"num_encoder_layers": 6, | |
"num_decoder_layers": 6, | |
"dropout_rate": 0.1, | |
"max_position_embeddings": 512, | |
"activation_function": "gelu", | |
"initializer_range": 0.02 | |
} | |