Open-Sora
Open-Sora copied to clipboard
hpcaitech
Open-Sora: Democratizing Efficient Video Production for All
Please consider enabling the Use of pytorch's scaled_dot_product_attention as an alternative for those with older GPU's See this example for another product. https://github.com/HiDream-ai/HiDream-I1/pull/27
@zhengzangw @SimonWXW Hello, In the technical report, it is mentioned that the video DC-AE model is further tuned using DINOv2 distillation loss after the initial training. However, I noticed that...
This PR fixes condition config mutation in `prepare_visual_condition` during training. Specifically, because dictionaries are passed by reference, even a single short video would be enough to completely remove `v2v_head`, `v2v_tail`,...
I run multi gpu inference on v1.2 and set nproc_per_node=6. Could you please clarify Why this happened? How to fix it? Thank you ! [rank0]: Traceback (most recent call last):...
executed cmd such as below and got exception: torchrun --nproc_per_node 1 --standalone scripts/diffusion/inference.py configs/diffusion/inference/t2i2v_256px.py --save-dir samples --prompt "raining, sea" [11/11/25 02:34:23] INFO colossalai - colossalai - INFO: /root/miniconda3/lib/python3.10/site-packages/colossalai/initialize.py:75 launch INFO...
Hello Open-Sora team, I would like to request access to the pretrained weights for Open-Sora v1.3, especially for image-to-video generation (I2V). My Hugging Face username is nufisiaz Thank you!
Aspect Ratio: 16:9 horizontal Style: Photorealistic, raw found-footage aesthetic Camera: First-person perspective (FPV), selfie-style, low angle looking up at the character Device Simulation: iPhone handheld, slight shake for realism Scene...
When using multiple GPUs in combination with `torchrun --nproc_per_node 1 ...` inference saves the decoder output only on main process ignoring the output coming from the rest of the GPUs....
