List your company

Reading Comprehension on SQuAD (Accuracy)

84.36AccuracyMeZO
30.727244.651158.57572.4989Feb 21, 2024Jul 1, 2024Nov 10, 2024Mar 21, 2025Jul 31, 2025Dec 9, 2025Apr 20, 2026

Evaluation results

MethodLinks
MeZOBackbone=OPT-13B, Evaluation Protocol=ZO-tuning2026.0484.36
MixtralParameters=8x22B2024.0784.1
AdaLeZOBackbone=OPT-13B, Evaluation Protocol=ZO-tuning2026.0483.52
GemmaParameters=7B2024.0781.8
Llama 3Parameters=70B2024.0781.8
Llama 3Parameters=405B2024.0781.8
MeZOBackbone=OPT-30B, Evaluation Protocol=ZO-tuning2026.0480.31
AdaLeZOBackbone=OPT-30B, Evaluation Protocol=ZO-tuning2026.0480.19
Llama 3Parameters=8B2024.0777
MistralParameters=7B2024.0773.2
Zero-shotBackbone=OPT-30B, Evaluation Protocol=Zero-shot2026.0446.5
Zero-shotBackbone=OPT-13B, Evaluation Protocol=Zero-shot2026.0446.2
INTRADOCChunk Length (L)=8K2024.0241.32
INTRADOCChunk Length (L)=2K2024.0239.91
BM25ChunkChunk Length (L)=8K2024.0239.45
BM25ChunkChunk Length (L)=2K2024.0237.91
UNIChunkChunk Length (L)=2K2024.0237.33
MIXChunkChunk Length (L)=2K2024.0236.7
UNIChunkChunk Length (L)=8K2024.0234.94
MIXChunkChunk Length (L)=8K2024.0232.79