PVNet 5.3.0__py3-none-any.whl → 5.3.5__py3-none-any.whl

pvnet/models/late_fusion/late_fusion.py

@@ -46,6 +46,7 @@ class LateFusionModel(BaseModel):
         include_generation_history: bool = False,
         include_sun: bool = True,
         include_time: bool = False,
+        t0_embedding_dim: int = 0,
         location_id_mapping: dict[Any, int] | None = None,
         embedding_dim: int = 16,
         forecast_minutes: int = 30,
@@ -85,6 +86,8 @@ class LateFusionModel(BaseModel):
             include_generation_history: Include generation yield data.
             include_sun: Include sun azimuth and altitude data.
             include_time: Include sine and cosine of dates and times.
+            t0_embedding_dim: Shape of the embedding of the init-time (t0) of the forecast. Not used
+                if set to 0.
             location_id_mapping: A dictionary mapping the location ID to an integer. ID embedding is
                 not used if this is not provided.
             embedding_dim: Number of embedding dimensions to use for location ID.
@@ -119,6 +122,7 @@ class LateFusionModel(BaseModel):
         self.include_pv = pv_encoder is not None
         self.include_sun = include_sun
         self.include_time = include_time
+        self.t0_embedding_dim = t0_embedding_dim
         self.location_id_mapping = location_id_mapping
         self.embedding_dim = embedding_dim
         self.add_image_embedding_channel = add_image_embedding_channel
@@ -246,6 +250,8 @@ class LateFusionModel(BaseModel):
             # Update num features
             fusion_input_features += 32
 
+        fusion_input_features += self.t0_embedding_dim
+
         if include_generation_history:
             # Update num features
             fusion_input_features += self.history_len + 1
@@ -321,6 +327,9 @@ class LateFusionModel(BaseModel):
             time = self.time_fc1(time)
             modes["time"] = time
 
+        if self.t0_embedding_dim>0:
+            modes["t0_embed"] = x["t0_embedding"]
+
         out = self.output_network(modes)
 
         if self.use_quantile_regression:

pvnet/models/late_fusion/site_encoders/encoders.py

@@ -158,7 +158,7 @@ class SingleAttentionNetwork(AbstractSitesEncoder):
         super().__init__(sequence_length, num_sites, out_features)
         self.sequence_length = sequence_length
         self.target_id_embedding = nn.Embedding(target_id_dim, out_features)
-        self.id_embedding = nn.Embedding(num_sites, id_embed_dim)
+        self.site_id_embedding = nn.Embedding(num_sites, id_embed_dim)
         self._ids = nn.parameter.Parameter(torch.arange(num_sites), requires_grad=False)
         self.use_id_in_value = use_id_in_value
         self.key_to_use = key_to_use
@@ -224,7 +224,7 @@ class SingleAttentionNetwork(AbstractSitesEncoder):
         site_seqs, batch_size = self._encode_inputs(x)
 
         # site ID embeddings are the same for each sample
-        id_embed = torch.tile(self.id_embedding(self._ids), (batch_size, 1, 1))
+        id_embed = torch.tile(self.site_id_embedding(self._ids), (batch_size, 1, 1))
         # Each concated (site sequence, site ID embedding) is processed with encoder
         x_seq_in = torch.cat((site_seqs, id_embed), dim=2).flatten(0, 1)
         key = self._key_encoder(x_seq_in)

pvnet/optimizers.py

@@ -65,7 +65,7 @@ class AbstractOptimizer(ABC):
     """
 
     @abstractmethod
-    def __call__(self):
+    def __call__(self, model: Module):
         """Abstract call"""
         pass
 
@@ -129,19 +129,18 @@ class EmbAdamWReduceLROnPlateau(AbstractOptimizer):
             {"params": decay, "weight_decay": self.weight_decay},
             {"params": no_decay, "weight_decay": 0.0},
         ]
+        monitor = "quantile_loss/val" if model.use_quantile_regression else "MAE/val"
         opt = torch.optim.AdamW(optim_groups, lr=self.lr, **self.opt_kwargs)
-
         sch = torch.optim.lr_scheduler.ReduceLROnPlateau(
             opt,
             factor=self.factor,
             patience=self.patience,
             threshold=self.threshold,
         )
-        sch = {
-            "scheduler": sch,
-            "monitor": "quantile_loss/val" if model.use_quantile_regression else "MAE/val",
+        return {
+            "optimizer": opt,
+            "lr_scheduler": {"scheduler": sch, "monitor": monitor},
         }
-        return [opt], [sch]
 
 
 class AdamWReduceLROnPlateau(AbstractOptimizer):
@@ -153,15 +152,13 @@ class AdamWReduceLROnPlateau(AbstractOptimizer):
         patience: int = 3, 
         factor: float = 0.5, 
         threshold: float = 2e-4, 
-        step_freq=None, 
         **opt_kwargs,
     ):
         """AdamW optimizer and reduce on plateau scheduler"""
-        self._lr = lr
+        self.lr = lr
         self.patience = patience
         self.factor = factor
         self.threshold = threshold
-        self.step_freq = step_freq
         self.opt_kwargs = opt_kwargs
 
     def _call_multi(self, model):
@@ -169,7 +166,7 @@ class AdamWReduceLROnPlateau(AbstractOptimizer):
 
         group_args = []
 
-        for key in self._lr.keys():
+        for key in self.lr.keys():
             if key == "default":
                 continue
 
@@ -178,43 +175,38 @@ class AdamWReduceLROnPlateau(AbstractOptimizer):
                 if param_name.startswith(key):
                     submodule_params += [remaining_params.pop(param_name)]
 
-            group_args += [{"params": submodule_params, "lr": self._lr[key]}]
+            group_args += [{"params": submodule_params, "lr": self.lr[key]}]
 
         remaining_params = [p for k, p in remaining_params.items()]
         group_args += [{"params": remaining_params}]
-
-        opt = torch.optim.AdamW(
-            group_args, 
-            lr=self._lr["default"] if model.lr is None else model.lr, 
-            **self.opt_kwargs,
+        monitor = "quantile_loss/val" if model.use_quantile_regression else "MAE/val"
+        opt = torch.optim.AdamW(group_args, lr=self.lr["default"], **self.opt_kwargs)
+        sch = torch.optim.lr_scheduler.ReduceLROnPlateau(
+            opt,
+            factor=self.factor,
+            patience=self.patience,
+            threshold=self.threshold,
         )
-        sch = {
-            "scheduler": torch.optim.lr_scheduler.ReduceLROnPlateau(
-                opt,
-                factor=self.factor,
-                patience=self.patience,
-                threshold=self.threshold,
-            ),
-            "monitor": "quantile_loss/val" if model.use_quantile_regression else "MAE/val",
+        return {
+            "optimizer": opt,
+            "lr_scheduler": {"scheduler": sch, "monitor": monitor},
         }
 
-        return [opt], [sch]
 
     def __call__(self, model):
         """Return optimizer"""
-        if not isinstance(self._lr, float):
+        if isinstance(self.lr, dict):
             return self._call_multi(model)
         else:
-            default_lr = self._lr if model.lr is None else model.lr
-            opt = torch.optim.AdamW(model.parameters(), lr=default_lr, **self.opt_kwargs)
+            monitor = "quantile_loss/val" if model.use_quantile_regression else "MAE/val"
+            opt = torch.optim.AdamW(model.parameters(), lr=self.lr, **self.opt_kwargs)
             sch = torch.optim.lr_scheduler.ReduceLROnPlateau(
                 opt,
                 factor=self.factor,
                 patience=self.patience,
                 threshold=self.threshold,
             )
-            sch = {
-                "scheduler": sch,
-                "monitor": "quantile_loss/val" if model.use_quantile_regression else "MAE/val",
+            return {
+                "optimizer": opt,
+                "lr_scheduler": {"scheduler": sch, "monitor": monitor},
             }
-            return [opt], [sch]

pvnet/training/lightning_module.py

@@ -109,7 +109,7 @@ class PVNetLightningModule(pl.LightningModule):
         losses = self._calculate_common_losses(y, y_hat)
         losses = {f"{k}/train": v for k, v in losses.items()}
 
-        self.log_dict(losses, on_step=True, on_epoch=True)
+        self.log_dict(losses, on_step=True, on_epoch=True, batch_size=y.size(0))
 
         if self.model.use_quantile_regression:
             opt_target = losses["quantile_loss/train"]
@@ -256,11 +256,28 @@ class PVNetLightningModule(pl.LightningModule):
 
         # Calculate the persistance losses - we only need to do this once per training run
         # not every epoch
-        if self.current_epoch == 0:
+        if self.current_epoch==0:
+            # Need to find last valid value before forecast
+            target_data = batch["generation"]
+            history_data = target_data[:, :-(self.model.forecast_len)]
+            
+            # Find where values aren't dropped
+            valid_mask = history_data >= 0
+            
+            # Last valid value index for each sample
+            flipped_mask = valid_mask.float().flip(dims=[1])
+            last_valid_indices_flipped = torch.argmax(flipped_mask, dim=1)
+            last_valid_indices = history_data.shape[1] - 1 - last_valid_indices_flipped
+            
+            # Grab those last valid values
+            batch_indices = torch.arange(
+                history_data.shape[0], 
+                device=history_data.device
+            )
+            last_valid_values = history_data[batch_indices, last_valid_indices]
+
             y_persist = (
-                batch["generation"][:, -(self.model.forecast_len + 1)]
-                .unsqueeze(1)
-                .expand(-1, self.model.forecast_len)
+                last_valid_values.unsqueeze(1).expand(-1, self.model.forecast_len)
             )
             mae_step_persist, mse_step_persist = self._calculate_step_metrics(y, y_persist)
             self._val_persistence_horizon_maes.append(mae_step_persist)
@@ -272,7 +289,7 @@ class PVNetLightningModule(pl.LightningModule):
             )
 
         # Log the metrics
-        self.log_dict(losses, on_step=False, on_epoch=True)
+        self.log_dict(losses, on_step=False, on_epoch=True, batch_size=y.size(0))
 
     def on_validation_epoch_end(self) -> None:
         """Run on epoch end"""

{pvnet-5.3.0.dist-info → pvnet-5.3.5.dist-info}/METADATA

@@ -1,12 +1,12 @@
 Metadata-Version: 2.4
 Name: PVNet
-Version: 5.3.0
+Version: 5.3.5
 Summary: PVNet
 Author-email: Peter Dudfield <info@openclimatefix.org>
 Requires-Python: <3.14,>=3.11
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: ocf-data-sampler>=0.6.0
+Requires-Dist: ocf-data-sampler>=1.0.9
 Requires-Dist: numpy
 Requires-Dist: pandas
 Requires-Dist: matplotlib

{pvnet-5.3.0.dist-info → pvnet-5.3.5.dist-info}/RECORD

@@ -1,14 +1,14 @@
 pvnet/__init__.py,sha256=TAZm88TJ5ieL1XjEyRg1LciIGuSScEucdAruQLfM92I,25
 pvnet/datamodule.py,sha256=wc1RQfFhgW9Hxyw7vrpFERhOd2FmjDsO1x49J2erOYk,5750
 pvnet/load_model.py,sha256=P1QODX_mJRnKZ_kIll9BlOjK_A1W4YM3QG-mZd-2Mcc,3852
-pvnet/optimizers.py,sha256=1N4b-Xd6QiIrcUU8cbU326bbFC0BvMNIV8VYWtGILJc,6548
+pvnet/optimizers.py,sha256=DZ74KcFQV226zwu7-qAzofTMTYeIyScox4Kqbq30WWY,6440
 pvnet/utils.py,sha256=L3MDF5m1Ez_btAZZ8t-T5wXLzFmyj7UZtorA91DEpFw,6003
 pvnet/models/__init__.py,sha256=owzZ9xkD0DRTT51mT2Dx_p96oJjwDz57xo_MaMIEosk,145
 pvnet/models/base_model.py,sha256=V-vBqtzZc_c8Ho5hVo_ikq2wzZ7hsAIM7I4vhzGDfNc,16051
 pvnet/models/ensemble.py,sha256=USpNQ0O5eiffapLPE9T6gR-uK9f_3E4pX3DK7Lmkn2U,2228
 pvnet/models/late_fusion/__init__.py,sha256=Jf0B-E0_5IvSBFoj1wvnPtwYDxs4pRIFm5qHv--Bbps,26
 pvnet/models/late_fusion/basic_blocks.py,sha256=_cYGVyAIyEJS4wd-DEAXQXu0br66guZJn3ugoebWqZ0,1479
-pvnet/models/late_fusion/late_fusion.py,sha256=kQUnyqMykmwc0GdoFhNXYStJPrjr3hFSvUNe8FumVx4,15260
+pvnet/models/late_fusion/late_fusion.py,sha256=r05RJvw2-ZQgWJobOGq1g4rlMJQjGM0UzG3syA4T0qo,15617
 pvnet/models/late_fusion/encoders/__init__.py,sha256=bLBQdnCeLYhwISW0t88ZZBz-ebS94m7ZwBcsofWMHR4,51
 pvnet/models/late_fusion/encoders/basic_blocks.py,sha256=DGkFFIZv4S4FLTaAIOrAngAFBpgZQHfkGM4dzezZLk4,3044
 pvnet/models/late_fusion/encoders/encoders3d.py,sha256=9fmqVHO73F-jN62w065cgEQI_icNFC2nQH6ZEGvTHxU,7116
@@ -17,13 +17,13 @@ pvnet/models/late_fusion/linear_networks/basic_blocks.py,sha256=RnwdeuX_-itY4ncM
 pvnet/models/late_fusion/linear_networks/networks.py,sha256=exEIz_Z85f8nSwcvp4wqiiLECEAg9YbkKhSZJvFy75M,2231
 pvnet/models/late_fusion/site_encoders/__init__.py,sha256=QoUiiWWFf12vEpdkw0gO4TWpOEoI_tgAyUFCWFFpYAk,45
 pvnet/models/late_fusion/site_encoders/basic_blocks.py,sha256=iEB_N7ZL5HMQ1hZM6H32A71GCwP7YbErUx0oQF21PQM,1042
-pvnet/models/late_fusion/site_encoders/encoders.py,sha256=DcTV2LeZ0pSZpGFmsmPEqYIhmPQeYCNi3UM406zHm14,11310
+pvnet/models/late_fusion/site_encoders/encoders.py,sha256=PemEUa_Wv5pFWw3usPKEtXcvs_MX2LSrO6nhldO_QVk,11320
 pvnet/training/__init__.py,sha256=FKxmPZ59Vuj5_mXomN4saJ3En5M-aDMxSs6OttTQOcg,49
-pvnet/training/lightning_module.py,sha256=57sT7bPCU7mJw4EskzOE-JJ9JhWIuAbs40_x5RoBbA8,12705
+pvnet/training/lightning_module.py,sha256=hmvne9DQauWpG61sRK-t8MTZRVwdywaEFCs0VFVRuMs,13522
 pvnet/training/plots.py,sha256=7JtjA9zIotuoKZ2l0fbS-FZDB48TcIk_-XLA2EWVMv4,2448
 pvnet/training/train.py,sha256=Sry2wYgggUmtIB-k_umFts7xMr2roEL76NCu9ySbLUY,4107
-pvnet-5.3.0.dist-info/licenses/LICENSE,sha256=tKUnlSmcLBWMJWkHx3UjZGdrjs9LidGwLo0jsBUBAwU,1077
-pvnet-5.3.0.dist-info/METADATA,sha256=b4Ki0jGoNNEd1VopMvR5p-iasCi0ZVtGwA-RfoHRCWw,16479
-pvnet-5.3.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-pvnet-5.3.0.dist-info/top_level.txt,sha256=4mg6WjeW05SR7pg3-Q4JRE2yAoutHYpspOsiUzYVNv0,6
-pvnet-5.3.0.dist-info/RECORD,,
+pvnet-5.3.5.dist-info/licenses/LICENSE,sha256=tKUnlSmcLBWMJWkHx3UjZGdrjs9LidGwLo0jsBUBAwU,1077
+pvnet-5.3.5.dist-info/METADATA,sha256=rIlZGmFiIzkMpG_5U-6SrsdDW6fIke667JAG79g3KN4,16479
+pvnet-5.3.5.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+pvnet-5.3.5.dist-info/top_level.txt,sha256=4mg6WjeW05SR7pg3-Q4JRE2yAoutHYpspOsiUzYVNv0,6
+pvnet-5.3.5.dist-info/RECORD,,