| { | |
| "channels": 16, | |
| "d_ff": 3072, | |
| "d_kv": 64, | |
| "d_model": 768, | |
| "dense_act_fn": "relu", | |
| "dropout_rate": 0.1, | |
| "feed_forward_proj": "relu", | |
| "in_channels": 3, | |
| "initializer_factor": 1.0, | |
| "is_gated_act": false, | |
| "layer_norm_epsilon": 1e-06, | |
| "load_weights": false, | |
| "model_max_length": 512, | |
| "model_name": "t5-base", | |
| "model_type": "tilt_base_configuration", | |
| "n_positions": 512, | |
| "num_classes": 61, | |
| "num_decoder_layers": 12, | |
| "num_heads": 12, | |
| "num_layers": 12, | |
| "num_pool_layers": 3, | |
| "output_past": true, | |
| "output_size": [ | |
| 3, | |
| 3 | |
| ], | |
| "relative_attention_max_distance": 128, | |
| "relative_attention_num_buckets": 32, | |
| "sampling_ratio": 2, | |
| "spatial_scale": 0.125, | |
| "transformers_version": "4.27.4", | |
| "use_cache": true, | |
| "vocab_size": 32128 | |
| } | |