bug: mmlu evaluation scores are constant

[jalling97]

Steps to reproduce

1. Deploy LFAI
2. Run evaluations against deployed instance
3. View MMLU results

Expected result

• Scores to very across each topic category in MMLU

Actual Result

• All scores are the same for each category (approx 0.69, which is relatively high)

Visual Proof (screenshots, videos, text, etc)

INFO:root:MMLU task scores:
                            Task    Score
0  high_school_european_history  0.69697
1               business_ethics  0.69697

Additional Context

May need to investigate the underlying implementation of mmlu in deepeval