MGVQ (16x16x8)

Reported on 4 benchmarks across 1 task · 1 paper · 2 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Medical4 results

Image ReconstructiononImageNet
FID· 2025-07-10
0.49
SOTA
MGVQ: Could VQ-VAE Beat VAE? A Generalizable Tokenizer with Multi-group Quantization arXiv:2507.07997
Image ReconstructiononImageNet
SSIM· 2025-07-10
0.787
SOTA
MGVQ: Could VQ-VAE Beat VAE? A Generalizable Tokenizer with Multi-group Quantization arXiv:2507.07997
Image ReconstructiononImageNet
LPIPS· 2025-07-10
0.086
best: 0.066 (OptVQ (16x16x8))
MGVQ: Could VQ-VAE Beat VAE? A Generalizable Tokenizer with Multi-group Quantization arXiv:2507.07997
Image ReconstructiononImageNet
PSNR· 2025-07-10
24.7
best: 27.57 (OptVQ (16x16x8))
MGVQ: Could VQ-VAE Beat VAE? A Generalizable Tokenizer with Multi-group Quantization arXiv:2507.07997