{ "architectures": [ "PianoT5Gemma" ], "attention_dropout": 0.0, "bos_token_id": 2, "classifier_dropout_rate": 0.0, "decoder": { "attention_bias": false, "attention_dropout": 0.0, "attn_logit_softcapping": 50.0, "cross_attention_hidden_size": 768, "dropout_rate": 0.0, "eos_token_id": 3, "final_logit_softcapping": 30.0, "head_dim": 128, "hidden_activation": "gelu_pytorch_tanh", "hidden_size": 768, "initializer_range": 0.02, "intermediate_size": 3072, "is_decoder": true, "layer_types": [ "sliding_attention", "full_attention" ], "max_position_embeddings": 8192, "model_type": "t5_gemma_module", "num_attention_heads": 8, "num_hidden_layers": 2, "num_key_value_heads": 4, "query_pre_attn_scalar": 256, "rms_norm_eps": 1e-06, "rope_theta": 10000.0, "sliding_window": 4096, "use_cache": true, "vocab_size": 5389 }, "dropout_rate": 0.0, "encoder": { "attention_bias": false, "attention_dropout": 0.0, "attn_logit_softcapping": 50.0, "dropout_rate": 0.0, "eos_token_id": 3, "final_logit_softcapping": 30.0, "head_dim": 128, "hidden_activation": "gelu_pytorch_tanh", "hidden_size": 768, "initializer_range": 0.02, "intermediate_size": 3072, "layer_types": [ "sliding_attention", "full_attention", "sliding_attention", "full_attention", "sliding_attention", "full_attention", "sliding_attention", "full_attention", "sliding_attention", "full_attention" ], "max_position_embeddings": 8192, "model_type": "t5_gemma_module", "num_attention_heads": 8, "num_hidden_layers": 10, "num_key_value_heads": 4, "query_pre_attn_scalar": 256, "rms_norm_eps": 1e-06, "rope_theta": 10000.0, "sliding_window": 4096, "use_cache": true, "vocab_size": 5389 }, "eos_token_id": 3, "hidden_size": 768, "initializer_range": 0.02, "is_encoder_decoder": true, "mask_token_id": 1, "model_type": "t5gemma", "pad_token_id": 0, "pedal_start": 5261, "pitch_start": 5, "play_token_id": 4, "timing_start": 261, "torch_dtype": "bfloat16", "transformers_version": "4.54.0", "use_cache": true, "valid_id_range": [ [ 5, 133 ], [ 261, 5252 ], [ 133, 261 ], [ 261, 5261 ], [ 5261, 5389 ], [ 5261, 5389 ], [ 5261, 5389 ], [ 5261, 5389 ] ], "velocity_start": 133, "vocab_size": 5389 }