Script for using large language models to assign quality ratings to training examples in bulk.
Has two backends, vLLM and exllama. Outputs ratings to a shelve file.
Currently pretty specifically tailored to my workflow. Feel free to reach out to me with any questions, suggestions, or feedback.