truncation_side=left as parameter in GPT2LMHeadModel.from_pretrained() #1
Description
In the few-shot-classification task during the evaluation when we run the evaluation using the shell command, when we use gpt2 as the task_lm we run into a TypeError. I believe this is due to the mistake of adding truncation_side=left as a parameter to the function call GPT2LMHeadModel.from_pretrained() in the fsc_evaluator.py script which applies mostly to tokenizers rather than models.
