API Sequence Generation Training
Dataset
Our dataset is available at https://zenodo.org/record/7549828#.Y8j164ZBy3I
annotation
CUDA_VISIBLE_DEVICES=<gpu_id> python single_run.py --max_length 64 --norm True --batch_size 32 --output_dir <output_directory> --epoch 30 --training_file <training_data_path> --valid_file <validation_data_path>annotation + SO Title
CUDA_VISIBLE_DEVICES=<gpu_id> python dual_run.py --max_length 64 --norm True --batch_size 32 --output_dir <output_directory> --epoch 30 --training_file <training_data_path> --valid_file <validation_data_path>annotation + SO Title + SO API
CUDA_VISIBLE_DEVICES=<gpu_id> python dual_api_run.py --max_length 64 --norm True --batch_size 32 --output_dir <output_directory> --epoch 30 --training_file <training_data_path> --valid_file <validation_data_path>Evaluation
single-annotation
CUDA_VISIBLE_DEVICES=<gpu_id> python evaluation.py --model_type 0 --max_length 64 --load_model_path <model_path>/pytorch_model.bin --test_filename <test_file_path> --output_dir <output_path>annotation + SO Title
CUDA_VISIBLE_DEVICES=<gpu_id> python evaluation.py --model_type 1 --max_length 64 --load_model_path <model_path>/pytorch_model.bin --test_filename <test_file_path> --output_dir <output_path>annotation + SO Title + SO API
CUDA_VISIBLE_DEVICES=<gpu_id> python evaluate.py --model_type 2 --load_model_path 13-Oct-dual/checkpoint-best-bleu/pytorch_model.bin --test_filename <test_file_path> --output_dir <output_path>Calculate the results
single-annotation
python calculate_bleu_score.py --reference <evaluation_output_path>/test_ref.csv --candidate <evaluation_output_path>/test_hyp.csv