Merge pull request #13 from openclimatefix/issue/rm-nowcasting-utils

Issue/rm nowcasting utils

pvnet_summation/models/base_model.py

@@ -6,16 +6,12 @@
 import torch
 import torch.nn.functional as F
 import wandb
-from nowcasting_utils.models.loss import WeightedLosses
-from nowcasting_utils.models.metrics import (
-    mae_each_forecast_horizon,
-    mse_each_forecast_horizon,
-)
 from pvnet.models.base_model import BaseModel as PVNetBaseModel
 from pvnet.models.base_model import PVNetModelHubMixin
 from pvnet.models.utils import (
     MetricAccumulator,
     PredAccumulator,
+    WeightedLosses,
 )
 from pvnet.optimizers import AbstractOptimizer
 
@@ -187,8 +183,9 @@ def validation_step(self, batch: dict, batch_idx):
             "MSE/val": F.mse_loss(y_sum, y),
             "MAE/val": F.l1_loss(y_sum, y),
         }
-        mse_each_step = mse_each_forecast_horizon(output=y_sum, target=y)
-        mae_each_step = mae_each_forecast_horizon(output=y_sum, target=y)
+
+        mse_each_step = torch.mean((y_sum - y) ** 2, dim=0)
+        mae_each_step = torch.mean(torch.abs(y_sum - y), dim=0)
         gsp_sum_losses.update({f"MSE_horizon/step_{i:02}": m for i, m in enumerate(mse_each_step)})
         gsp_sum_losses.update({f"MAE_horizon/step_{i:02}": m for i, m in enumerate(mae_each_step)})
 

requirements.txt

@@ -1,7 +1,6 @@
-nowcasting_utils
-ocf_datapipes>=3.0.0
-pvnet>=2.6.2
-ocf_ml_metrics
+ocf_datapipes==3.0.*
+pvnet==2.6.*
+ocf_ml_metrics==0.0.*
 numpy
 pandas
 matplotlib