huggingface / huggingface/diffusers
[SD3] Incorrect stochastic sampling implementation
- Lenguaje dominante
- Python
- Estrellas
- 34.5k
- Forks
- 7.3k
- Merge medio
- 3 d 3 h
- PR fusionados (30 d)
- 91
Descripción
### Describe the bug
Ref to Algorithm 2 of [EDM](https://arxiv.org/abs/2206.00364), for a given sample $x_t$, noise is introduced to it reaching a higher noise level $\hat{t}$, then we evaluate network with $\hat{x_t}$, $\hat{t}$ as input. However, the current implementation evaluates network with $x_t$, $t$ as input, which is inconsistent from definition.
Current implementation is more similar with Euler-Maruyama in spirit, "One can interpret Euler–Maruyama as first adding
noise and then performing an ODE step, not from the intermediate state after noise injection, but
assuming that $x$ and $\sigma$ remained at the initial state at the beginning of the iteration step." quote from [EDM](https://arxiv.org/abs/2206.00364)
### Reproduction
no
### Logs
_No response_
### System Info
no
### Who can help?
@yiyixuxu @sayakpaul
Guía de contribución
Línea de trabajo
No file, test, or entry point is named. Start by locating the stochastic sampling implementation and compare its network inputs with Algorithm 2 of the linked EDM paper, focusing on whether the noise-adjusted sample and noise level are evaluated. Done means the implementation follows that algorithm and has coverage demonstrating the corrected inputs.
Escrito por el modelo de indexación a partir del texto del issue.
Evaluación
- Stack tecnológico
- python, pytorch
- Área
- machine-learning
- Tipo de issue
- Error
- Dificultad
- 4/5
- Tiempo estimado
- 3-5 días
- Estado de actividad
- Estancado
- Claridad
- Necesita aclaración
- Aptitud para principiantes
- 30/100