List your company

Selective Prediction on TriviaQA (test)

100Power (α=0.1)LEC
36.372852.891469.4185.9286Dec 1, 2025

Evaluation results

MethodLinks
LECLLM=Qwen2.5-14B2025.1210071.93100100100
UCB-CLPLLM=LLaMA-3.1-70B2025.1210099.35100100100
UCB-HFDLLM=LLaMA-3.1-70B2025.1210095.03100100100
LECLLM=LLaMA-3.1-70B2025.1210099.96100100100
UCB-CLPLLM=Qwen2.5-14B2025.1299.8762.4100100100
UCB-HFDLLM=Qwen2.5-14B2025.1297.18—100100100
LECLLM=LLaMA-3.1-8B2025.1296.1275.38100100100
LECLLM=OpenChat-3.52025.1295.2172.3100100100
UCB-CLPLLM=LLaMA-3.1-8B2025.1293.9671.43100100100
LECLLM=Vicuna-13B-V1.52025.1293.4265.45100100100
UCB-CLPLLM=OpenChat-3.52025.1292.9466.84100100100
UCB-HFDLLM=LLaMA-3.1-8B2025.1290.3953.39100100100
UCB-CLPLLM=Vicuna-13B-V1.52025.1289.4456.02100100100
UCB-HFDLLM=OpenChat-3.52025.1288.8460.91100100100
LECLLM=Qwen2.5-7B2025.1288.548.89100100100
UCB-HFDLLM=Vicuna-13B-V1.52025.1283.6452.41100100100
UCB-CLPLLM=Qwen2.5-7B2025.1283.3139.05100100100
UCB-HFDLLM=Qwen2.5-7B2025.1273.96—99.9100100
LECLLM=Vicuna-7B-V1.52025.1265.08—92.08100100
LECLLM=Qwen2.5-3B2025.1259.9827.0690.81100100
UCB-CLPLLM=Vicuna-7B-V1.52025.1256.86—86.3100100
UCB-CLPLLM=Qwen2.5-3B2025.1252.1923.7685.54100100
UCB-HFDLLM=Vicuna-7B-V1.52025.1242.28—80.6899.99100
UCB-HFDLLM=Qwen2.5-3B2025.1238.82—77.72100100