Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Natural Language Understanding and Reasoning on Standard Downstream Benchmarks Two-Shot (val)

56.86ARC-E Accuracy (Normalized)

AdaGC

45.607248.528651.4554.3714Feb 16, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.02
56.8629.6159.3657.8933.673.9957.6226.4685.953.47
2025.02
55.8128.5860.756.543373.7256.7525.5183.252.64
2025.02
53.8328.4258.6955.6633.873.0754.1425.1281.851.61
2025.02
52.9528.6756.1555.6935.473.0756.4326.8882.852
2025.02
52.8629.156.4853.7631.873.0755.7226.0382.651.27
2025.02
51.9426.8857.5553.3632.472.3154.1426.6381.650.75
2025.02
50.3427.3958.8152.9634.271.1654.0625.3779.950.47
2025.02
47.2625.650.3146.4432.269.6452.3325.0777.847.41
2025.02
47.125.7756.5443.973068.8852.9626.0977.247.61
2025.02
46.5525.8549.7645.713070.0253.225.6977.747.16
2025.02
46.0426.1949.7247.513169.754.3824.9878.547.56