Sana
Sana copied to clipboard
NVlabs
SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer
when will controlnet training script be released?
Our SANA-Sprint models focus on timestep distillation, achieving high-quality generation with 1-4 inference steps. Refer to [SANA-Sprint Manual](https://github.com/NVlabs/Sana/blob/main/asset/docs/sana_sprint.md) for more details. After the PR merged: https://github.com/huggingface/diffusers/pull/11074 pip install git+https://github.com/huggingface/diffusers ```python...
I trained a Efficient-Large-Model/Sana_1600M_1024px/checkpoints/Sana_1600M_1024px.pth to generate image in a certain style(camera angle and composition), the checkpoints are in .pth I want to use this script below to dreambooth LoRA finetune...
Hi, I'm trying to reproduce the Sana-Sprint experiment and have two questions regarding the configuration and dataset used for training the student model: 1. **Teacher Checkpoint for Student Initialization:** As...
(Sana) (base) lzha0538@gpu1-3w-1:~/2Dediting/Sana$ /home/lzha0538/miniconda3/envs/Sana/bin/python /home/lzha0538/2Dediting/Sana/SanaControlnet.py 2025-04-01 11:48:34 - [Sana] - INFO - Sampler flow_dpm-solver, flow_shift: 4.0 2025-04-01 11:48:34 - [Sana] - INFO - Inference with torch.float16, PAG guidance layer: [8]...
I tried [Inference with 8bit](https://github.com/NVlabs/Sana/blob/main/asset/docs/8bit_sana.md) with SANA1.5 1.6B and the image quality is worse than the demo. Then I tried device_map, the inference on 8g VRAM is about 65% slower...
In the paper, it is discussed that SANA-Sprint disable `1000t` in training, which leads to better training stability. However, in the code, SANA-Sprint still seems to keep the `1000t`? I...
