One needs to run api.py as well as post_api.py , in order to get inference , argparse can be used to pass arguments for inference , can also change / modify the code as well for input , model screenshots of run can be found in gdrive: metrics : you can check with N-grams whether its using the words / ngrams from the train data or not , to better justify the generation .

docker command to build image from Dockerfile : docker built -t lyric_gen . command to run a docker : docker run -d -p 8000:8000 lyric_gen

Pretrained Text generation model can be finetuned with gpt-2-simple also can be done with huggingface .

model checkpoint :