Unable to evaluate few-shot results.

[whwang299]

Hello!
Thank you so much for sharing your work.
I found some (really) minor issues in the script as described below.

1. Unable to evaluate few-shot results.
   To fix the issue,

This line

zeroshot_lexglue/evaluate_perfomance.py

Line 70 in d23e81e

folder_name = f'_few_shot-predictions' if args.few_shot_k else 'zero-shot-predictions'

should be changed to

folder_name = f'few-shot-predictions' if args.few_shot_k else 'zero-shot-predictions'

2. Unable to import data in [task-name]_instruction.py.
To fix the issue,
the following line may be added before import data.

import sys
sys.path.append(os.path.dirname(os.path.dirname(os.path.realpath(__file__))))

(for instance the 3rd line of unfairtos_instruction.py

zeroshot_lexglue/build_instructions/unfairtos_instructions.py

Line 3 in d23e81e

from data import DATA_DIR

)

3. This is more like a question.
   I could not find the examples used for few-shot evaluation of unfair-tos, ledgar, and eurlex.

Thanks!

Wonseok