Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction · HackerPapers discussions

Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction — mentioned by peter_peter

Mon, 04 Aug 2025 18:58:24 GMT

Location: Southern California Remote: Yes Willing to relocate: Yes Technologies: C++, Python, Hugging Face Resume: Can share on request Email: peterpeternews@gmail.com I recently graduated a PhD in theoretical computer science. I have a strong mathematical background and software engineering experience. Recently I have been interested in machine learning. Right now I am working on a transformer based autoregressive image generation model (see for example https://arxiv.org/abs/2404.02905 ). I am looking for any professional experience in the machine learning domain. I am willing to work part…

Paper and discussion history. A mention is not an endorsement.

Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction — mentioned by cubefox

Mon, 19 May 2025 16:34:19 GMT

It's nice that this contains a comparison between diffusion models that are used for image models, and the autoregressive models that are used for LLMs. But recently (2024 NeuIPS paper of the year) there was a new paper on autoregressive image modelling that apparently outperforms diffusion models: https://arxiv.org/abs/2404.02905 The innovation is that it doesn't predict image patches (like older autoregressive image models) but somehow does some sort of "next scale" or "next resolution" prediction. In the past, autoregressive image models did not perform as well as diffusion models, which…

Paper and discussion history. A mention is not an endorsement.

Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction — mentioned by famouswaffles

Wed, 09 Apr 2025 14:24:54 GMT

This video is very good. https://youtu.be/EzDsrEvdgNQ?si=EWp3U1GMkwg1bMQQ One thing i'd add is that generating the tokens at the target resolution from the start is no longer the only approach to autoregressive image generation. Rather than predicting each patch at the target resolution right away, it starts with the image (as patches) at a very small resolution and increasingly scales up. Paper here - https://arxiv.org/abs/2404.02905

Paper and discussion history. A mention is not an endorsement.

Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction — mentioned by andy12_

Wed, 09 Apr 2025 08:12:32 GMT

I mean, it could very well be that it generates image patches autoregressively, but in a pyramidal way (first a very low resolution version, the "canvas", and then each individual patch). This is very similar to VAR [1] We can't really be sure until OpenAI tells us. [1] https://arxiv.org/pdf/2404.02905

Paper and discussion history. A mention is not an endorsement.

Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction — mentioned by famouswaffles

Tue, 08 Apr 2025 13:30:56 GMT

It's kind of clear that for every request, it generates a new image entirely. Some people are speculating a diffusion decoder but i think it's more likely an implementation of VAR - https://arxiv.org/abs/2404.02905. So rather than predicting each patch at the target resolution right away, it starts with the image (as patches) at a very small resolution and increasingly scales up. I guess that could make it hard for the model to learn to just copy and paste image tokens for editing like it might for text.

Paper and discussion history. A mention is not an endorsement.

Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction — mentioned by famouswaffles

Tue, 08 Apr 2025 12:09:50 GMT

Open AI have both said it's native image generation and autoregressive. It has the signs of it too. It's probably an implementation of VAR (https://arxiv.org/abs/2404.02905) - autoregressive image generation with a small twist. Rather than predict every token at the target resolution directly, start with predicting it at a small resolution, cranking it higher and higher until the desired resolution.

Paper and discussion history. A mention is not an endorsement.

Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction — mentioned by famouswaffles

Wed, 26 Mar 2025 01:35:40 GMT

There is a more recent approach to auto-regressive image generation. Rather than predicting the next patch at the target resolution one by one, it predicts the next resolution. That is, the image at a small resolution followed by the image at a higher resolution and so on. https://arxiv.org/abs/2404.02905

Paper and discussion history. A mention is not an endorsement.

Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction — mentioned by famouswaffles

Mon, 14 Oct 2024 20:23:50 GMT

The most popular method using autoregression in image generation space is to predict image patches/tokens and not pixels, though that still scales worse than diffusion. A fairly new but promising approach for autoregression that seems to scale as well as diffusion is predicting the next image scale/resolution rather than the next image patch. https://arxiv.org/abs/2404.02905

Paper and discussion history. A mention is not an endorsement.

Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction — HN submission

Sat, 05 Oct 2024 14:15:51 GMT

Visual Autoregressive Modeling: Image Generation via Next-Resolution Prediction

1 point; 1 discussion reply on HN. Paper and source history. A submission is not an endorsement.

Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction — HN submission

Fri, 05 Apr 2024 12:41:02 GMT

Visual Autoregressive Modeling: Scalable Image Generation W NextScale Prediction

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.