{ "architectures": [ "transformerModel" ], "batch_size": 64, "dropout": 0.1, "ffn_num_hiddens": 64, "ffn_num_input": 32, "key_size": 32, "lr": 0.005, "model_type": "custom_transformer", "norm_shape": [ 32 ], "num_epochs": 200, "num_heads": 4, "num_hiddens": 32, "num_layers": 2, "num_steps": 10, "query_size": 32, "src_vocab_len": 184, "tgt_vocab": 201, "torch_dtype": "float32", "transformers_version": "4.45.2", "value_size": 32 }