ResearchBenchmarksAlternative Annotator Test on BillsFollow85Doc-Level Fit ProbPROXANN27.842.6557.572.35Jul 1, 2025Evaluation ResultsMethodMethodLinksDoc-Level Fit ProbDoc-Level Rank ProbTopic-Level Fit ProbTopic-Level Rank ProbPROXANNLLM Backbone=GPT-4oLLM Backbone=GPT-4o2025.0785717775PROXANNLLM Backbone=Qwen-3 32BLLM Backbone=Qwen-3 32B2025.0767687470PROXANNLLM Backbone=Llama-3.3...LLM Backbone=Llama-3.3-70B2025.0766677060PROXANNLLM Backbone=Qwen-3 8BLLM Backbone=Qwen-3 8B2025.0766578043PROXANNLLM Backbone=Qwen-2.5-72BLLM Backbone=Qwen-2.5-72B2025.0761717865PROXANNLLM Backbone=Llama-3.1-8BLLM Backbone=Llama-3.1-8B2025.0730531444