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

Zero-shot Reasoning on Downstream Tasks (LMB, PIQA, HellaSwag, OPQA, ARC)

32.75LAMBADA (LMB) Accuracy

FLT

11.502817.018922.53528.0511May 11, 2026
Updated 14d ago

Evaluation Results

MethodLinks
2026.05
32.7565.0737.0533.253.8728.4141.72
2026.05
31.1564.8537.0335.252.8127.2441.38
2026.05
31.1264.6835.7933.256.0628.4141.54
2026.05
30.6463.8135.7231.851.0926.3639.9
2026.05
30.5664.8536.2432.853.7827.8141
2026.05
30.3562.8936.333452.8126.6240.05
2026.05
30.0764.2535.4334.651.6427.1340.52
2026.05
29.9264.6336.3331.452.6526.2740.2
2026.05
29.5362.9435.8132.451.9727.6440.04
2026.05
28.3964.4135.943250.9628.0739.96
2026.05
26.3761.3131.3632.843.2224.3136.56
2026.05
20.5858.6527.762838.8424.2333.01
2026.05
20.3959.2427.7129.3939.9423.4633.35
2026.05
20.0459.6327.827.239.5622.4432.77
2026.05
19.7558.8627.528.239.623.5432.9
2026.05
19.6559.8427.5628.239.6824.6533.26
2026.05
19.2558.9727.9127.837.4523.0332.4
2026.05
19.0359.1428.225.638.8823.8932.45
2026.05
18.759.4627.662738.9723.6332.57
2026.05
18.3959.1427.4825.839.5622.4432.13
2026.05
16.3357.5626.7928.236.6523.2931.47
2026.05
12.3256.92626.836.4822.8630.22