lm_eval/tasks/tinyBenchmarks/tinyBenchmarks.yaml (16 lines of code) (raw):
group: tinyBenchmarks
task:
- task: tinyArc
num_fewshot: 25
- task: tinyGSM8k
num_fewshot: 5
- task: tinyMMLU
num_fewshot: 0
- task: tinyWinogrande
num_fewshot: 5
- task: tinyHellaswag
num_fewshot: 10
- task: tinyTruthfulQA
num_fewshot: 0
metadata:
version: 0.0