Updated Pt Train Fine Tune Classification Transformer Model node and added a workflow and documentation.

docs/fine_tuning_huggning_face_models_for_classification.md

@@ -54,4 +54,18 @@ Specify **Pt Train Transformer Classification Model** and configure training par
 
 ## Expected Accuracy
 
-Using this workflow, you should achieve around 92-93% validation accuracy on the Hugging Face version of the IMDB dataset with the pretrained DistilBERT model.
+Using this workflow, you should achieve around 92-93% validation accuracy on the Hugging Face version of the IMDB dataset with the pretrained DistilBERT model.
+
+# Fine-tuning by freezing the pretrained module layers
+
+ComfyUI-Pt-Wrapper also offers a node for fine-tuning by freezing the pretrained module layers.  
+See `examples/workflows/distilbert_classification_2.json`.
+
+![Workflow](images/distilbert_classification_2.png)
+
+Add **Pt Train Fine Tune Classification Transformer Model** node and specify the number of epochs to freeze in `freeze pretrained module epochs`.  
+Also specify the total number of epochs in `epochs`.
+
+For example, if you set `freeze_pretrained_module_epochs` to 1 and `epochs` to 4, then the weights for the pretrained module are frozen during the first epoch, and full fine-tuning takes place from epochs 2 to 4.
+
+Using this example workflow, you can expect a validation accuracy of 92% or 93%.

modules/pytorch_wrapper/pt_train_fine_tune_classification_transformer_model.py

@@ -51,7 +51,6 @@ class PtTrainFineTuneClassificationTransformerModel:
                 "classification_metrics": ("BOOLEAN", {"default":True})
             },
             "optional": {
-                "scheduler": ("PTLRSCHEDULER", {}),
                 "val_loader": ("PTDATALOADER", {})
             }
         }

pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "pt-wrapper"
 description = "PyTorch extension for ComfyUI featuring extensive PyTorch wrapper nodes for seamless tensor operations and PyTorch model training."
-version = "3.5.0"
+version = "3.5.1"
 license = {file = "LICENSE"}
 
 [project.urls]