Vchitect / Latte

Latte: Latent Diffusion Transformer for Video Generation.

Vchitect/Latte Issues

trained and sample result very strange （我自己训练复现的效果很奇怪）
Closed 2 months ago20
how to place and preprocess these datasets
Updated 7 days ago1
Question: model code and design choices
Updated 15 days ago1
视频帧率
Updated 15 days ago1
question on t2v model training
Updated 16 days ago1
有关时长的问题。
Updated 18 days ago3
Preprocess of UCF101
Updated 21 days ago14
About video VAE
Updated 21 days ago1
Questions about the *0.18215 and /0.18215 operation
Closed 23 days ago2
Is autoregression possible?
Updated a month ago3
No positional embeddings in LatteT2V?
Closed a month ago
Evaluate the FVD？
Closed a month ago5
image_size = [256,512]
Updated a month ago4
FaceForensics数据集
Updated a month ago1
Some weights of AutoencoderKL were not initialized from the model checkpoint at /path/to/Latte/t2v_required_models/ and are newly initialized because the shapes did not match:
Updated 2 months ago2
Train code of t2v？
Updated 2 months ago2
CUDA out of memory
Updated 2 months ago4
anyone meet zero grad?零梯度？
Closed 2 months ago1
What is the difference between Latte and ViViT?
Updated 2 months ago2
RuntimeError: "slow_conv2d_cpu" not implemented for 'Half'
Closed 2 months ago2
Issue about "LayerNormKernelImpl" not implemented for 'Half'
Closed 2 months ago5
RuntimeError: "LayerNormKernelImpl" not implemented for 'Half'
Updated 2 months ago1
Extra key in ucf101.pt
Updated 2 months ago7
Why choose these datasets and why not compare with pika, SVD or Gen2?
Updated 2 months ago1
how to get 2048 videos for computing FVD?
Updated 2 months ago3
How can I turn on the autoregressive mode to generate >16 frame videos?
Updated 2 months ago1
Code Reuse
Updated 2 months ago5
Latte的实时微信讨论组
Updated 2 months ago6
About Training Speed
Updated 2 months ago3
t2i代码疑问
Updated 2 months ago1
About resume checkpoint
Updated 2 months ago2
Implementation of compression frame patch embedding (Fig. 3b)
Closed 2 months ago2
diffusion noise modify
Updated 2 months ago1
About evaluation
Updated 2 months ago3
如何复现主页展示的t2v效果？
Updated 2 months ago3
t2v只支持16帧吗？我改成更多比如32帧就啥都看不到了
Updated 2 months ago1
Can you provide the code for DDIM sampler
Updated 2 months ago1
为什么会这样报错呢在运行sample.py模块的时候Traceback (most recent call last): File "C:\Users\Dell\Desktop\Project\Latte-main\sample\sample.py", line 29, in <module> from models import get_models File "C:\Users\Dell\Desktop\Project\Latte-main\models\__init__.py", line 7, in <module> from .latte_t2v import LatteT2V File "C:\Users\Dell\Desktop\Project\Latte-main\models\latte_t2v.py", line 11, in <module> from diffusers.models.embeddings import get_1d_sincos_pos_embed_from_grid, ImagePositionalEmbeddings, CaptionProjection, PatchEmbed, CombinedTimestepSizeEmbeddings ImportError: cannot import name 'CaptionProjection' from 'diffusers.models.embeddings'
Updated 2 months ago1
T2V with >16 vedio_length output random noises
Closed 3 months ago3
Training BatchSize
Updated 2 months ago5
torchrun --nnodes=1 --nproc_per_node=2 train_with_img.py --config ./configs/sky/sky_img_train.yaml error
Updated 3 months ago2
Re-implementation err on ffs experiment
Updated 3 months ago1
Non-consecutive added token '<extra_id_99>' found.
Closed 3 months ago2
Discriminative tasks
Updated 3 months ago1
Excellent work, will there be an official support of images to vedio (like sora) ?
Updated 3 months ago4
What is
Updated 3 months ago2
Cannot find model：LatteT2V.from_pretrained_2d
Closed 3 months ago1
Does Latte support multiple GPUs
Closed 3 months ago4
run bash sample/t2v.sh error
Closed 3 months ago4
run bash sample/t2v.sh,but why?
Updated 3 months ago1