Closed mohamed-abdelaziz721 closed 1 year ago
Hi there, the codet5-base checkpoint is the one trained with the proposed identifier-aware denoising pretraining objective, while "task control codes" are only used for the multi-task finetuning. So you need to finetune CodeT5-base checkpoint on downstream tasks or directly use our released checkpoints.
I was using hugging faces codet5-base to try code generation and understanding tasks.
I cannot find documentation anywhere that indicates how to use "task control codes" for different input types.
I am trying to do something like Figure 1 in the original paper