gpt-3.5-turbo-0125: solved 14/144 = 9.7%, total cost = 0.895176
gpt-4o-2024-05-13: solved 80/144 = 55.56%, total cost = 12.0396
There are 3 tasks that cause OpenDevin to hung. Need to investigate. For now, treat them as failure.
Create visualization for MINT benchmark & upload results (#2)
- add some outputs (da7aaba4e08844c20e458ec7d554243dfc5349bf)
- add viz tab for mint (38a40d14b423e340d494308aa2871d4736c78053)
- add complete math output for gpt-4o (7d377c3c477760c429dfd7f5583f998a0edf0968)
- add complete mmlu output for gpt-4o (0948b4d436b173e9b5b165a8de4ee2099cd30f33)
- add complete humaneval output for gpt-4o (45710d94e2a58ead357ebdc378981aac09e042e4)
- add complete theoremqa output for gpt-4o (841a9486d33eece9d9cdb03bb15b02093656a1ad)
Co-authored-by: Ryan Tran <ryanhoangt@users.noreply.huggingface.co>
