List your company

Cardiac diagnosis on MIMIC-IV-Ext

54.9F1@3EvidenceRL with GRPO
32.95638.65344.3550.047Mar 20, 2026

Evaluation results

MethodLinks
EvidenceRL with GRPOBackbone=Gemma-3-27B2026.0354.946.248.444.4714.111.469
EvidenceRL with GRPOBackbone=Llama-3.2-3B2026.0354.57778.761.62.48.63.487.5
EvidenceRL with GRPOBackbone=Llama-3.1-8B2026.0353.958.961.155.34.314.27.276.1
Self-RAGBackbone=Gemma-3-27B2026.0353.325.827.434.59.517.517.754.3
fDPOBackbone=Gemma-3-27B2026.0353.162.564.648.54.411.6777.7
Reasoning OnlyBackbone=Gemma-3-27B2026.0352.237.13939.18.515.513.763.1
Self-RAGBackbone=Llama-3.3-70B2026.0351.639.942.141916.311.666.3
Reasoning OnlyBackbone=Llama-3.3-70B2026.0351.346.949.443.88.514.39.371.9
EvidenceRL with GRPOBackbone=Gemma-3-12B2026.0349.33840.53810.617.210.365.9
SFTBackbone=Llama-3.2-3B2026.0349.134.836.541.311.515.512.267
Self ConsistencyBackbone=Gemma-3-27B2026.0348.645.948.139.68.713.910.168.7
fDPOBackbone=Gemma-3-12B2026.0347.937.639.93710.216.51165.6
SFTBackbone=GPT-oss-20B2026.0347.731.833.13812.115.712.863.5
SFTBackbone=Llama-3.1-8B2026.0347.224.926.537.913.317.513.861.4
Self-RAGBackbone=Gemma-3-12B2026.034717.51929.514.419.316.252.9
Self ConsistencyBackbone=Llama-3.3-70B2026.0346.851.253.241.29.212.17.874.5
EvidenceRL with GRPOBackbone=Gemma-3-4B2026.0346.8545642.47.813.96.676.4
Reasoning OnlyBackbone=Gemma-3-12B2026.0346.521.523.53114.419.214.356.3
Self-RAGBackbone=Gemma-3-4B2026.0346.546.248.739.28.218.17.870.4
fDPOBackbone=Gemma-3-4B2026.0346.368.970.847.65.99.83.186.4
Reasoning OnlyBackbone=GPT-oss-20B2026.0346-2.2-0.924.915.617.72640.2
fDPOBackbone=GPT-oss-20B2026.034614.515.729.913.116.520.848.4
Reasoning OnlyBackbone=Gemma-3-4B2026.0344.234.336.335.113.419.28.966
Self ConsistencyBackbone=GPT-oss-20B2026.03440.41.626.916.515.724.643.5
Reasoning OnlyBackbone=GPT-oss-120B2026.0343.618.319.730.814.415.517.953.3
Self ConsistencyBackbone=Gemma-3-4B2026.0341.948.350.537.310.614.96.275.3
Self ConsistencyBackbone=GPT-oss-120B2026.0341.225.126.732.315.216.11359.2
Self ConsistencyBackbone=Gemma-3-12B2026.034124.42628.716.4171259.7
SFTBackbone=Gemma-3-4B2026.0339.92.11.828.418.218.621.147.3
fDPOBackbone=Llama-3.1-8B2026.0339.751.45339.5911.27.376.7
Self-RAGBackbone=Llama-3.1-8B2026.0339.43033.131.513.619.99.662.7
Self ConsistencyBackbone=Llama-3.2-3B2026.0339.348.250.333.610.715.65.971.8
fDPOBackbone=Llama-3.2-3B2026.0338.762.463.637.97.911.43.682.2
Reasoning OnlyBackbone=Llama-3.1-8B2026.0338.219.421.327.417.720.211.556.6
Self-RAGBackbone=GPT-oss-120B2026.0337.111.412.929.915.716.418.252.6
Reasoning OnlyBackbone=Llama-3.2-3B2026.033745.347.631.811.616.56.272.5
SFTBackbone=Gemma-3-27B2026.0336.61.21.227.81819.721.545.7
Self-RAGBackbone=Llama-3.2-3B2026.0336.524.425.62617.318.510.459.6
Self-RAGBackbone=GPT-oss-20B2026.0336.50.21.725.716.317.623.742.7
SFTBackbone=Llama-3.3-70B2026.0335.28.91032.514.318.119.552.1
Self ConsistencyBackbone=Llama-3.1-8B2026.0334.825.827.626.817.619.78.760.7
SFTBackbone=Gemma-3-12B2026.0333.80.30.628.516.918.823.445.3