🏴‍☠️ No more private imports.

It should still pass github actions.

perbert/data/__init__.py

@@ -0,0 +1,2 @@
+from .collators import Collator, DecayCollator, HuggingfaceCollator
+from .datamodules import TextDataModule

perbert/data/collators.py

@@ -1,4 +1,3 @@
-# pyright: reportPrivateImportUsage=false
 from __future__ import annotations
 
 from abc import abstractmethod

perbert/data/datamodules.py

@@ -1,4 +1,3 @@
-# pyright: reportPrivateImportUsage=false
 from __future__ import annotations
 
 from typing import Any
@@ -65,7 +64,6 @@ def setup(self, stage: str | LightningStage) -> None:
             assert Splits.TEST in self.datasets.keys(), self.datasets
 
     def _init_collator(self) -> Collator:
-
         tokenizer = AutoTokenizer.from_pretrained(self.cfg["data"]["tokenizer"])
         mask_prob = self.cfg["model"]["lm"]["mask_prob"]
 

perbert/data/datasets/mappers.py

@@ -1,4 +1,3 @@
-# pyright: reportPrivateImportUsage=false
 from __future__ import annotations
 
 from abc import abstractmethod

perbert/data/datasets/prepare.py

@@ -1,4 +1,3 @@
-# pyright: reportPrivateImportUsage=false
 from __future__ import annotations
 
 import typing

perbert/data/datasets/wrappers.py

@@ -1,4 +1,3 @@
-# pyright: reportPrivateImportUsage=false
 from __future__ import annotations
 
 from abc import abstractmethod

perbert/models/__init__.py

@@ -0,0 +1,3 @@
+from .init import bert_init, emb_init, layernorm_init, linear_init
+from .length_schedulers import LengthScheduler
+from .models import Model, OptimizerType

perbert/models/init.py

@@ -1,4 +1,3 @@
-# pyright: reportPrivateImportUsage=false
 from typing import Callable
 
 import loguru

perbert/models/models.py

@@ -1,4 +1,3 @@
-# pyright: reportPrivateImportUsage=false
 from __future__ import annotations
 
 import typing

perbert/trainer.py

@@ -1,4 +1,3 @@
-# pyright: reportPrivateImportUsage=false
 from __future__ import annotations
 
 import typing
@@ -15,7 +14,8 @@
     RichModelSummary,
     RichProgressBar,
 )
-from pytorch_lightning.loggers import Logger, TensorBoardLogger, WandbLogger
+from pytorch_lightning.loggers import Logger, TensorBoardLogger
+from pytorch_lightning.loggers.wandb import WandbLogger
 
 
 class Trainer(PLTrainer):