List your company

Text-to-Video Generation on DL3DV-10K 1K (test)

17.69PSNREpipolar-DPO
14.767615.526316.28517.0437Jan 30, 2026

Evaluation results

MethodLinks
Epipolar-DPOBase Model=CogVideoX-5B2026.0117.690.6180.5070.9710.5280.5794553.674948.67
Baseline-T2VBase Model=CogVideoX-5B2026.0117.530.6140.5080.9670.5330.584————
VideoGPABase Model=CogVideoX-5B2026.0117.310.6210.4950.9740.5190.54862.676742.6760.33
SFTBase Model=CogVideoX-5B2026.0117.070.5730.5630.9680.5860.71914.6723.6739.3315.33
GeoVideoBase Model=CogVideoX1.5-5B2026.0115.90.510.6430.8520.6730.8417.3644.4430.5618.06
Baseline-T2V15Base Model=CogVideoX1.5-5B2026.0115.570.480.5550.9760.5880.685————
VideoGPABase Model=CogVideoX1.5-5B2026.0114.880.5030.520.9820.5560.56760.4254.1752.0857.64