List your company

Zero-shot Evaluation on 8 Downstream Tasks (ARC-C, BoolQ, HellaSwag, MMLU, OBQA, PIQA, RTE, WinoGrande)

45ARC-C AccuracyUNPRUNED
30.75234.45138.1541.849Nov 26, 2024

Evaluation results

MethodLinks
UNPRUNEDACT.=2.7B, MEM.=100%, Speedup=1.0x, Evaluation protocol=zero-shot2024.114579.577.361.243.680.367.969.365.5
CD-MoE-S#L=6, ACT.=2.5B, MEM.=78.3%, Speedup=1.22x, SFT status=W/O SFT, Evaluation protocol=zero-shot2024.1141.175.872.9564178.967.268.162.6
MC-SMoE#L=6, ACT.=2.7B, MEM.=73.0%, Speedup=1.0x, SFT status=W/O SFT, Evaluation protocol=zero-shot2024.1140.976.171.553.239.777.570.564.761.8
CD-MoE-SR#L=6, ACT.=2.7B, MEM.=79.7%, Speedup=1.19x, SFT status=W/O SFT, Evaluation protocol=zero-shot2024.1140.277.272.154.939.47871.166.462.4
LayerDrop#L=6, ACT.=2.3B, MEM.=78.3%, Speedup=1.23x, SFT status=W/O SFT, Evaluation protocol=zero-shot2024.1134.462.663.145.432.873.654.562.653.6
BlockDrop#L=6, ACT.=2.2B, MEM.=76.1%, Speedup=1.36x, SFT status=W/O SFT, Evaluation protocol=zero-shot2024.1131.362.257.438.43270.752.758.250.4