Datasets:

nuprl

/

MultiPL-E

like 45

Follow

Northeastern University Programming Research Lab 25

I would like to ask how the performance of your large model was achieved. how to generate a grade according your datasets?Was it done using a specific algorithm, and is there any related paper on this?

If I want to use your pipeline but modify some of the data in the dataset to prevent the model from training on this specific data, how should I proceed?

Thanks for your remarkable job. I love it!