List your company
ResearchFormerly WizWand

Medical Question Answering on PubMedQA

86Pass@1MedXIAOHE
19.75236.95154.1571.349Sep 28, 2025Nov 7, 2025Dec 17, 2025Jan 26, 2026Mar 7, 2026Apr 16, 2026May 27, 2026
Updated 5mo ago

Evaluation results

MethodLinks
MedXIAOHEMode=Thinking mode, Decoding=Greedy2026.0286
Gemini 3.0 ProDecoding=Greedy2026.0280.8
GPT-5.2 ThinkingMode=Thinking mode, Decoding=Greedy2026.0279.8
Gemini 2.5 ProDecoding=Greedy2026.0275.6
SHIFTTrain Size=0.1%2026.0574.2
SC-EntropyTrain Size=0.2%2026.0572.4
SC-EntropyTrain Size=0.1%2026.0571.6
CoT SimilarityTrain Size=0.1%2026.0571.4
ClusterTrain Size=0.1%2026.0571
CoT SimilarityTrain Size=0.2%2026.0571
SHIFTTrain Size=0.2%2026.0570.8
Random (avg)Train Size=0.2%2026.0570.72
Random (avg)Train Size=0.1%2026.0570.56
CoreSetTrain Size=0.2%2026.0570.4
Q-PPLTrain Size=0.2%2026.0570
Qwen3-1.7BTrain Size=100%2026.0569.6
Q-PPLTrain Size=0.1%2026.0569.6
A-PPLTrain Size=0.2%2026.0569.2
A-PPLTrain Size=0.1%2026.0569
ClusterTrain Size=0.2%2026.0569
CoreSetTrain Size=0.1%2026.0568.6
Qwen3-1.7BTrain Size=N/A2026.0566.2
Qwen2.5-7B + GRPO w/ VERL.Base Model=Qwen2.5-7B, Training=GRPO w/ VERL.2025.0947.8
Qwen2.5-7B + GRPOBase Model=Qwen2.5-7B, Training=GRPO2025.0946.6
Qwen2.5-7B + PPOBase Model=Qwen2.5-7B, Training=PPO2025.0945.8
Qwen2.5-7B + PPO w/ VERL.Base Model=Qwen2.5-7B, Training=PPO w/ VERL.2025.0945.4
Qwen2.5-7BBase Model=Qwen2.5-7B2025.0943.6
Mathstral-7B-v0.1 + GRPO w/ VERL.Base Model=Mathstral-7B-v0.1, Training=GRPO w/ VERL.2025.0924.4
Mathstral-7B-v0.1 + PPOBase Model=Mathstral-7B-v0.1, Training=PPO2025.0923.1
Mathstral-7B-v0.1 + GRPOBase Model=Mathstral-7B-v0.1, Training=GRPO2025.0922.6
Mathstral-7B-v0.1Base Model=Mathstral-7B-v0.12025.0922.4
Mathstral-7B-v0.1 + PPO w/ VERL.Base Model=Mathstral-7B-v0.1, Training=PPO w/ VERL.2025.0922.3