List your company

Sentence-level length-controlled question answering on ELI5 (test)

0.08MAEICL+Feedback
-0.07880.99312.0653.1369Jan 5, 2026

Evaluation results

MethodLinks
ICL+FeedbackBackbone=Qwen3-8B, Strategy=ICL+Feedback2026.010.0892.87.69
FeedbackBackbone=LLaMA-3.1-8B-Instruct, Strategy=Feedback2026.010.1195.757.05
FeedbackBackbone=Qwen3-8B, Strategy=Feedback2026.010.1291.157.51
ICL+FeedbackBackbone=LLaMA-3.1-8B-Instruct, Strategy=ICL+Feedback2026.010.387.137.18
SFT+FeedbackBackbone=Qwen3-8B, Strategy=SFT+Feedback2026.010.478.68.11
SFT+FeedbackBackbone=LLaMA-3.1-8B-Instruct, Strategy=SFT+Feedback2026.011.5444.47.45
Qwen3-8BBackbone=Qwen3-8B, Strategy=Baseline2026.011.8248.677.57
SFTBackbone=Qwen3-8B, Strategy=SFT2026.012.2119.188.18
ICLBackbone=Qwen3-8B, Strategy=ICL2026.012.7434.577.71
SFTBackbone=LLaMA-3.1-8B-Instruct, Strategy=SFT2026.013.4311.127.5
LLaMA-3.1-8B-InstructBackbone=LLaMA-3.1-8B-Instruct, Strategy=Baseline2026.013.5233.457.26
ICLBackbone=LLaMA-3.1-8B-Instruct, Strategy=ICL2026.014.0527.847.31