{ "architectures": [ "GridTSTForTimeSeriesPrediction" ], "attention_dropout": 0.1, "attention_strategy": "channel_first", "d_model": 128, "dropout": 0.2, "ffn_dim": 256, "head_dropout": 0.0, "init_std": 0.2, "label_len": 192, "model_type": "gridtst", "norm_type": "batchnorm", "num_channels": 21, "num_heads": 16, "num_layers": 4, "num_patches": 6, "patch_len": 64, "qkv_bias": true, "revin_affine": true, "seq_len": 336, "stride": 64, "torch_dtype": "float32", "transformers_version": "4.35.2" }