{ "dropout": 0.04184815819561255, "feedforward_multiplier": 3.0, "hidden_size": 256, "max_sequence_length": 256, "num_attention_heads": 4, "num_classes": 4, "num_layers": 2, "pooling": "masked_mean", "rope_base": 10000.0, "vocab_size": 16000 }