Huggingface Transformer - GPT2 resume training from saved checkpoint

Question

Resuming the GPT2 finetuning, implemented from run_clm.py

Does GPT2 huggingface has a parameter to resume the training from the saved check