---
title: Causal Diffusion Transformers for Generative Modeling
url: https://www.ml-quant.com/papers/arxiv/2412.12095/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-09-26
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: arXiv:2412.12095
source_url: https://arxiv.org/abs/2412.12095
featured: 2024-12-18
citations: 34
topic: Econometrics & Forecasting
---


# Causal Diffusion Transformers for Generative Modeling

The article discusses Causal Diffusion, a framework that enhances diffusion models' performance and allows a seamless shift between autoregressive and diffusion generation modes, achieving top results on the ImageNet generation benchmark.

- Source: https://arxiv.org/abs/2412.12095
- Identifier: arXiv:2412.12095
- Released: 2024-12-16
- First featured: Quant Letter No. 79 (2024-12-18): https://www.ml-quant.com/issues/2024-12-18/
- Citations (Semantic Scholar): 34
- Published in: not yet
- Topic: Econometrics & Forecasting

## Related

- [Systematic comparison of deep generative models applied to multivariate financial time series](https://www.ml-quant.com/papers/arxiv/2412.06417/): The study contrasts deep generative models (DGMs) and parametric models for creating financial time series, highlighting the advantages of DGMs in an implied volatility trading task.
- [Hidformer: Transformer-Style Neural Network in Stock Price Forecasting](https://www.ml-quant.com/papers/arxiv/2412.19932/): The research explores the application of the Hidformer model, a Transformer-based neural network, in stock price forecasting, combining technical analysis and advanced machine learning for improved accuracy.
- [Time-Causal VAE: Robust Financial Time Series Generator](https://www.ml-quant.com/papers/arxiv/2411.02947/): The article discusses a time-causal variational autoencoder (TC-VAE) that generates reliable financial time series data. This data closely mirrors real market distribution and is beneficial for financial optimization tasks.
- [Can GANs Learn the Stylized Facts of Financial Time Series?](https://www.ml-quant.com/papers/arxiv/2410.09850/): The study examines the capability of Generative Adversarial Networks in learning complex financial time series patterns, highlighting that their performance is greatly influenced by the generator architecture chosen.
- [CoFinDiff: Controllable Financial Diffusion Model for Time Series Generation](https://www.ml-quant.com/papers/arxiv/2503.04164/): CoFinDiff, a new synthetic financial data generation model, is designed to overcome the shortcomings of traditional statistical models, producing synthetic data that matches any conditions.
- [A Financial Time Series Denoiser Based on Diffusion Models](https://www.ml-quant.com/papers/arxiv/2409.02138/): Using the diffusion model as a denoiser for financial time series data improves data predictability and trading performance, leading to more profitable trades and enhanced trading efficiency.