Skip to content

Repository files navigation

Counting-Stars-Local

Fork from Counting-Stars.

Usage

  1. Prepare a long context like ./harry_potter_all.txt
  2. generate test data: python gen_test_data_en.py
    • You may need to change prompt to get it right
  3. run evaluation: python test -m <model_path>
  4. insert result label into json file like: "answer": "{\"little_penguin\": [1, 2, 3, 4]}",
    • three field was required in json file: answer, sky_size, reference_counting_results
  5. generate visualized graph

Tips

Small model may not return the right answer like: [1, 2, 3, 4]. The viz_multi_page.py script will extract every number in the "answer" filed and reader each graph a single page in the pdf file.

Terms

  • M: How many stars
  • N: How long the context should be
    • interval = MAX_LEN/N
    • sky_size = [1 x interval, 2 x interval, ..., N x interval]
      • each sky contain M stars

About

Counting-Stars scripts for evaluating local llm.

Resources

Stars

2 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages