Tongda Xu (@xutongda) profile picture
Joined November 2022
Photo by Tongda Xu (@xutongda)Photo by Tongda Xu (@xutongda)Photo by Tongda Xu (@xutongda)Photo by Tongda Xu (@xutongda)
🚀 Can we build a VQ-VAE without training one? Our ICML 2026 paper introduces Gaussian Quant (GQ): ✅ Train only a Gaussian VAE ✅ Convert it into a VQ-VAE without any additional training ✅ Random Gaussian noise codebook Paper: arxiv.org/abs/2512.06609
2
23
164
10,918
Tongda Xu retweeted
Does GenAI create new knowledge? arxiv.org/abs/2602.06021 gives * 1st explicit characterization of diffusion model's generalization * more precise than offered by classical stat. learning theory * systematic integration of various inductive biases (training+architecture+inference)
4
28
1
179
19,107
We provide an explaination for why reconstruction correlates negatively to gFID, by showing that the latent space optimized for reconstruction prefers separable modes, which leads to halluciation.
53
We provide an explanation for why iFID correlates well with gFID, and why reconstruction metrics are negatively correlated with gFID, by showing that latent space with low iFID avoids hallucation brought by mode interpolation.
53
For each element in the dataset, we retrieve its nearest neighbor (NN) in the latent space and interpolate their latent representations. We then decode the interpolated latent and compute the FID between the decoded samples and the original dataset.
44
A year later we finally scale this idea up to 256^2 images. Though the technical approach is kind of different from DiffC. We use inversion to replace correlation communication to make a real codec. Preprint: arxiv.org/abs/2401.08920 Code: github.com/tongdaxu/Idempote…
Replying to @lucastheis
We only require a single unconditionally trained diffusion model (used across bit-rates) where previous approaches use encoder and decoder transforms, entropy models, VGG, and adversarial models to achieve realism.
1
364