Upload Part3.md

Part3.md

@@ -0,0 +1,12 @@
+### Fine Tuning
+
+Fine tuning is when using a pretrained model, train it on a dataset specific to the task. The steps for fine tuning a pre_trained model can be given as follows:
+
+1. Tokenize each sentence using RoBERTa 
+2. Set training parameters
+3. Train the model
+4. Assess: Using ROC-AUC for this problem
+
+### Challenge
+
+The biggest challenge in this task is to get the correct label for each piece of news. As we can't verify news ourselves, I managed to choose news that is objective and has been confirmed or falsified.