going back to cuda 12, removing hf mixins

ben_publication/BENv2ImageClassifier.py

@@ -7,18 +7,17 @@ from typing import List
 import lightning.pytorch as pl
 import torch
 import torch.nn.functional as F
-from configilm import ConfigILM
 from configilm.ConfigILM import ILMConfiguration
 from configilm.ConfigILM import ILMType
 from configilm.extra.BENv2_utils import NEW_LABELS
 from configilm.extra.CustomTorchClasses import LinearWarmupCosineAnnealingLR
 from configilm.metrics import get_classification_metric_collection
-from huggingface_hub import PyTorchModelHubMixin
 
 __author__ = "Leonard Hackel - BIFOLD/RSiM TU Berlin"
 
 
-class BENv2ImageEncoder(pl.LightningModule, PyTorchModelHubMixin):
+# class BENv2ImageEncoder(pl.LightningModule, PyTorchModelHubMixin):
+class BENv2ImageEncoder(pl.LightningModule):
     """
     Wrapper around a pytorch module, allowing this module to be used in automatic
     training with pytorch lightning.
@@ -39,7 +38,7 @@ class BENv2ImageEncoder(pl.LightningModule, PyTorchModelHubMixin):
         assert config.classes == 19
         self.mock_params = torch.nn.Linear(config.classes, config.classes)
         self.model = lambda x: self.mock_params(torch.rand((x.shape[0], config.classes)).to(x.device))
-        #self.model = ConfigILM.ConfigILM(config)
+        # self.model = ConfigILM.ConfigILM(config)
         self.val_output_list: List[dict] = []
         self.test_output_list: List[dict] = []
         self.loss = torch.nn.BCEWithLogitsLoss()
@@ -114,7 +113,6 @@ class BENv2ImageEncoder(pl.LightningModule, PyTorchModelHubMixin):
         super().on_validation_epoch_start()
         self.val_output_list = []
 
-
     def on_validation_epoch_end(self):
         avg_loss = torch.stack([x["loss"] for x in self.val_output_list]).mean()
         self.log("val/loss", avg_loss)

pyproject.toml

@@ -10,14 +10,6 @@ python = ">=3.10, <3.12"
 configilm = { extras = ["full"], version = "^0.6.3" }
 wandb = "^0.17.1"
 numpy = "^1.26.4"
-torch = {version = "^2.3.1+cu118", source = "pytorch-gpu-src"}
-torchvision = {version = "^0.18.1+cu118", source = "pytorch-gpu-src"}
-torchaudio = {version = "^2.3.1+cu118", source = "pytorch-gpu-src"}
-
-[[tool.poetry.source]]
-name = "pytorch-gpu-src"
-url = "https://download.pytorch.org/whl/cu118"
-priority = "explicit"
 
 [build-system]
 requires = ["poetry-core"]