List your company

Open-ended Question Answering on NarrativeQA (Acc, BLEU, F1)

73.77Accuracyw/t BoT
56.724461.149765.57570.0003May 20, 2026

Evaluation results

MethodLinks
w/t BoTModel=Qwen2.5-1.5b-instruct, Geometric Component=Barycentric Transport2026.0573.7754.8573.82
w/t CDModel=Qwen2.5-1.5b-instruct, Geometric Component=Cosine Dispersion2026.0572.5953.7273.25
w/t KLEModel=Qwen2.5-1.5b-instruct2026.0572.0352.1172.39
w/t SEModel=Qwen2.5-1.5b-instruct2026.0571.7652.272.24
CAREModel=Qwen2.5-1.5b-instruct2026.0571.451.8471.65
GRPOModel=Qwen2.5-1.5b-instruct2026.0566.2348.7167.02
EMPOModel=Qwen2.5-1.5b-instruct2026.0557.3841.3260.96