clinton254/patch_tst

1{2  "activation_function": "gelu",3  "architectures": [4    "PatchTSTForClassification"5  ],6  "attention_dropout": 0.0,7  "bias": true,8  "channel_attention": false,9  "channel_consistent_masking": false,10  "context_length": 512,11  "d_model": 128,12  "distribution_output": "student_t",13  "do_mask_input": null,14  "dropout": 0.0,15  "ff_dropout": 0.0,16  "ffn_dim": 512,17  "head_dropout": 0.0,18  "init_std": 0.02,19  "loss": "mse",20  "mask_type": "random",21  "mask_value": 0,22  "model_type": "patchtst",23  "norm_eps": 1e-05,24  "norm_type": "batchnorm",25  "num_attention_heads": 4,26  "num_forecast_mask_patches": [27    228  ],29  "num_hidden_layers": 3,30  "num_input_channels": 3,31  "num_parallel_samples": 100,32  "num_targets": 4,33  "output_range": null,34  "patch_length": 16,35  "patch_stride": 16,36  "path_dropout": 0.0,37  "pooling_type": "mean",38  "positional_dropout": 0.0,39  "positional_encoding_type": "sincos",40  "pre_norm": true,41  "prediction_length": 24,42  "random_mask_ratio": 0.5,43  "scaling": "std",44  "share_embedding": true,45  "share_projection": true,46  "torch_dtype": "float32",47  "transformers_version": "4.37.2",48  "unmasked_channel_indices": null,49  "use_cls_token": false50}51