SEBIS/code_trans_t5_base_code_comment_generation_java_multitask

The CodeTrans model is based on the t5-base model architecture . It has its own SentencePiece vocabulary model . It can be used on unparsed and untokenized java code . The model could be used to generate the description for the java function or be fine-tuned on other java code tasks . It was trained on a single TPU Pod V3-8 for 460,000 steps in total, using sequence length 512 (batch size 4096) It has a total of approximately 220M parameters and was trained using the encoder-decoder architecture . The optimizer used is AdaFactor with inverse square root learning rate schedule for pre-training. The model was,