Spaces:
Sleeping
Sleeping
tryout no plot
Browse files
streamlit_simulation/app.py
CHANGED
|
@@ -142,29 +142,29 @@ def load_lightgbm_model():
|
|
| 142 |
def load_transformer_model_and_dataset():
|
| 143 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 144 |
|
| 145 |
-
|
| 146 |
-
|
| 147 |
-
|
| 148 |
-
|
| 149 |
-
|
| 150 |
-
|
| 151 |
-
|
| 152 |
-
|
| 153 |
-
|
| 154 |
-
model.load_state_dict(torch.load(checkpoint_path, map_location=device))
|
| 155 |
-
model.to(device)
|
| 156 |
-
model.eval()
|
| 157 |
|
| 158 |
-
|
| 159 |
-
|
| 160 |
-
|
| 161 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 162 |
|
| 163 |
-
return model, test_dataset, device
|
| 164 |
-
|
| 165 |
-
except Exception as e:
|
| 166 |
-
st.error(f"❌ Fehler beim Laden des Transformer-Modells: {e}")
|
| 167 |
-
raise e
|
| 168 |
|
| 169 |
@st.cache_data
|
| 170 |
def load_data():
|
|
|
|
| 142 |
def load_transformer_model_and_dataset():
|
| 143 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 144 |
|
| 145 |
+
model = load_moment_model()
|
| 146 |
+
checkpoint_path = hf_hub_download(
|
| 147 |
+
repo_id="dlaj/energy-forecasting-files",
|
| 148 |
+
filename="transformer_model/model_final.pth",
|
| 149 |
+
repo_type="dataset"
|
| 150 |
+
)
|
| 151 |
+
model.load_state_dict(torch.load(checkpoint_path, map_location=device))
|
| 152 |
+
model.to(device)
|
| 153 |
+
model.eval()
|
|
|
|
|
|
|
|
|
|
| 154 |
|
| 155 |
+
csv_path = hf_hub_download(
|
| 156 |
+
repo_id="dlaj/energy-forecasting-files",
|
| 157 |
+
filename="data/processed/energy_consumption_aggregated_cleaned.csv",
|
| 158 |
+
repo_type="dataset"
|
| 159 |
+
)
|
| 160 |
+
|
| 161 |
+
# Datasets
|
| 162 |
+
train_dataset = InformerDataset(data_split="train", forecast_horizon=FORECAST_HORIZON, random_seed=13, csv_path=csv_path)
|
| 163 |
+
test_dataset = InformerDataset(data_split="test", forecast_horizon=FORECAST_HORIZON, random_seed=13, csv_path=csv_path)
|
| 164 |
+
test_dataset.scaler = train_dataset.scaler
|
| 165 |
+
|
| 166 |
+
return model, test_dataset, device
|
| 167 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 168 |
|
| 169 |
@st.cache_data
|
| 170 |
def load_data():
|
transformer_model/scripts/utils/informer_dataset_class.py
CHANGED
|
@@ -18,6 +18,7 @@ class InformerDataset:
|
|
| 18 |
data_stride_len: int = 1,
|
| 19 |
task_name: str = "forecasting",
|
| 20 |
random_seed: int = 42,
|
|
|
|
| 21 |
):
|
| 22 |
"""
|
| 23 |
Parameters
|
|
@@ -36,7 +37,7 @@ class InformerDataset:
|
|
| 36 |
|
| 37 |
self.seq_len = SEQ_LEN
|
| 38 |
self.forecast_horizon = forecast_horizon
|
| 39 |
-
self.full_file_path_and_name = DATA_PATH
|
| 40 |
self.data_split = data_split
|
| 41 |
self.data_stride_len = data_stride_len
|
| 42 |
self.task_name = task_name
|
|
|
|
| 18 |
data_stride_len: int = 1,
|
| 19 |
task_name: str = "forecasting",
|
| 20 |
random_seed: int = 42,
|
| 21 |
+
csv_path=None
|
| 22 |
):
|
| 23 |
"""
|
| 24 |
Parameters
|
|
|
|
| 37 |
|
| 38 |
self.seq_len = SEQ_LEN
|
| 39 |
self.forecast_horizon = forecast_horizon
|
| 40 |
+
self.full_file_path_and_name = csv_path if csv_path is not None else DATA_PATH
|
| 41 |
self.data_split = data_split
|
| 42 |
self.data_stride_len = data_stride_len
|
| 43 |
self.task_name = task_name
|