【PPMat Dataset No.2】Jarvis-DFT3D数据集适配

ppmat/datasets/jarvis_dataset.py

@@ -897,9 +897,15 @@ def filter_unvalid_by_property(self):
             data = self.property_data[property_name]
             reserve_idx = []
             for i, data_item in enumerate(data):
-                if isinstance(data_item, str) or (
-                    data_item is not None and not math.isnan(data_item)
-                ):
+                # Convert 'na' strings to NaN for proper filtering
+                if isinstance(data_item, str):
+                    if data_item.lower() in ['na', 'nan', 'none', '']:
+                        data_item = np.nan
+                    else:
+                        # Skip non-numeric strings (they're invalid for numeric properties)
+                        continue
+                # Keep only valid numeric values (not None, not NaN)
+                if data_item is not None and not math.isnan(data_item):
                     reserve_idx.append(i)
             for key in self.property_data.keys():
                 self.property_data[key] = [
@@ -979,9 +985,22 @@ def __getitem__(self, idx: int):
             data["structure_array"] = self.get_structure_array(structure)
         for property_name in self.property_names:
             if property_name in self.property_data:
-                data[property_name] = np.array(
-                    [self.property_data[property_name][idx]]
-                ).astype("float32")
+                value = self.property_data[property_name][idx]
+                # Check for 'na' strings - these should have been filtered out during initialization
+                if isinstance(value, str) and value.lower() in ['na', 'nan', 'none', '']:
+                    raise ValueError(
+                        f"Found invalid property value '{value}' at index {idx} for property "
+                        f"'{property_name}'. This should have been filtered out during dataset "
+                        f"initialization. Please ensure 'filter_unvalid=True' is set and "
+                        f"consider clearing the cache to regenerate filtered data."
+                    )
+                # Check for NaN values - these should also have been filtered out
+                if value is not None and (isinstance(value, float) and math.isnan(value)):
+                    raise ValueError(
+                        f"Found NaN value at index {idx} for property '{property_name}'. "
+                        f"This should have been filtered out during dataset initialization."
+                    )
+                data[property_name] = np.array([value]).astype("float32")
             else:
                 raise KeyError(f"Property {property_name} not found.")