Share your thoughts, 1 month free Claude Pro on usSee more

Retrieval-Augmented Question Answering on RAGQA Leaderboard (test)

85AVG Score

SFT+RL(Rhybrid)

Updated 1mo ago

Evaluation Results

Method	Links
SFT+RL(Rhybrid) 2026.02		85	86.2	82.4	71.5	92.8	93.2	89.4	79.8
SFT+RL(RCLR) 2026.02		85	86.9	82.5	72.7	93.8	94.7	89.5	74.6
SFT+RL(Rhybrid) 2026.02		84.9	87.1	83.2	73.9	94	94.9	88.9	72.6
SFT+RL(Rtotal) 2026.02		84.2	86.5	82.5	73.2	93.7	95.4	86.4	71.8
SFT+RL(Rtotal) 2026.02		83.9	85	81.3	70.7	92.6	91.4	86.6	80
SFT+RL(RCLR) 2026.02		83.8	84.4	81.6	69.3	92.3	92.5	88.1	78.6
SFT+RL(Racc) 2026.02		83.6	84.9	81.5	70.5	92.6	91.1	86.7	78
SFT+RL(Racc) 2026.02		83.1	85.4	82.3	70.6	93.2	93.8	85.9	70.4
SFT+RL(Rhybrid) 2026.02		83.1	83	81.9	70.5	94.9	94.4	79.3	78
SFT+RL(Rcite) 2026.02		82.4	82.6	79.7	68.1	91.8	91.6	85.8	77.2
SFT+RL(Rcite) 2026.02		82.4	83.4	80.4	69.5	93.3	94	85.2	70.8
SFT+RL(RCLR) 2026.02		82.3	82.7	81.5	69	93.7	93.5	78.7	76.8
SFT 2026.02		81.6	82	79.6	66.8	91.4	89.6	84.7	77.2
SFT+RL(Rtotal) 2026.02		81.2	82.2	81.1	69.1	92.5	93.4	78.3	71.6
SFT+RL(Rhybrid) 2026.02		80.4	82.1	80.7	65.2	92.1	90.7	77.6	74.8
SFT 2026.02		80.3	81.1	79.3	64.9	92	92.2	84.5	68.2
SFT+RL(Racc) 2026.02		80	80.8	80.5	66.2	93.2	92.5	76.7	70.4
SFT+RL(RCLR) 2026.02		79.7	80.4	79.9	64.4	91.6	89.6	76.6	75.6
SFT+RL(Rtotal) 2026.02		79	78.9	78.6	64.8	91.6	90.4	76.4	74.8
SFT+RL(Rcite) 2026.02		78.9	79.2	79	65.3	91.8	93	75.5	68.4
Qwen3-235B-A22B-Instruct-2507 2026.02		78	84.9	82.8	65.3	93.8	87.6	78.2	53.2
SFT 2026.02		77.6	78.3	78.7	63.4	92.3	92.2	73.8	64.2
SFT+RL(Rcite) 2026.02		77.2	78.3	78.8	62.9	91.8	90.4	74.9	63.2
SFT+RL(Racc) 2026.02		77	79.8	79.6	64.3	91.1	89.7	76.6	58.2
Qwen3-30B-A3B-Instruct-2507 2026.02		75.5	81.4	81.9	55.8	94.2	86.7	75.9	52.4
SFT 2026.02		72	74.6	76.1	60.2	87.7	88.6	74.1	43
Qwen3-32B 2026.02		69.3	77.9	73.2	48.6	89.4	76.4	68.3	51.6
Qwen3-14B 2026.02		63.2	68.8	69.4	37.6	86.8	72.7	65.1	42
Qwen3-8B 2026.02		61.9	71.4	71	44.5	90.2	74	64.7	17.6