IC-601Lightweight LLMs exhibit high judgment uncertainty (disagreement ratio exceeding 50% for Qwen2-1.5B) when making repeated binary checklist evaluations, with uncertainty increasing as model size decreases

Tianjun Wei, Wei Wen, Ruizhi Qiao, Xing Sun, Jianghong Ma

SourceRocketEval: Efficient automated LLM evaluation via grading checklist

The paper measures the ratio of disagreement across repeated samplings of binary yes/no judgments on WildBench checklist items for five released LLMs. When asked to judge whether a response satisfies a specific criterion, the models produce inconsistent answers across samplings. Qwen2-1.5B shows a disagreement ratio exceeding 50% across 3 sampling results, while larger models like Llama-3-70B show much lower disagreement. The paper identifies a clear correlation between model size and consistency, with smaller models tending to produce higher inconsistency. This high uncertainty is identified as a primary reason why lightweight LLMs fail as reliable judges in chain-of-thought evaluation settings.

Evidence
correlational
Key metric
Qwen2-1.5B disagreement ratio exceeding 50% across 3 sampling results; Llama-3-70B shows lowest disagreement across 7 samplings
Caveat
Measured only on WildBench checklist items (binary yes/no questions); the paper does not quantify the exact disagreement ratio for each model in the text, only describing the trend in Figure 4.
Model
Qwen 2 Qwen2-1.5B, 7B, 72B, Llama 3 8B, 70B
Concepts
Failure mode, Scale-dependent behaviour
Datasets
WildBench [eval]
Related work
MT-Bench [context], WildBench [context]
Related findings
IC-602
Extraction
automatic-extraction