CogVideoX is an open-source video generation model similar to QingYing. The table below displays the list of video generation models we currently offer, along with their foundational information.

Model Name	CogVideoX1.5-5B (Latest)	CogVideoX1.5-5B-I2V (Latest)	CogVideoX-2B	CogVideoX-5B	CogVideoX-5B-I2V
Release Date	November 8, 2024	November 8, 2024	August 6, 2024	August 27, 2024	September 19, 2024
Video Resolution	1360 * 768	Min(W, H) = 768 768 ≤ Max(W, H) ≤ 1360 Max(W, H) % 16 = 0	720 * 480
Number of Frames	Should be 16N + 1 where N <= 10 (default 81)	Should be 8N + 1 where N <= 6 (default 49)
Inference Precision	BF16 (Recommended), FP16, FP32, FP8*, INT8, Not supported: INT4	*FP16(Recommended)*, BF16, FP32, FP8, INT8, Not supported: INT4	BF16 (Recommended), FP16, FP32, FP8*, INT8, Not supported: INT4
Single GPU Memory Usage	SAT BF16: 76GB diffusers BF16: from 10GB* diffusers INT8(torchao): from 7GB*	SAT FP16: 18GB diffusers FP16: 4GB minimum* diffusers INT8 (torchao): 3.6GB minimum*	SAT BF16: 26GB diffusers BF16 : 5GB minimum* diffusers INT8 (torchao): 4.4GB minimum*
Multi-GPU Memory Usage	*BF16: 24GB using diffusers**	*FP16: 10GB using diffusers**	*BF16: 15GB using diffusers**
Inference Speed (Step = 50, FP/BF16)	Single A100: ~1000 seconds (5-second video) Single H100: ~550 seconds (5-second video)	Single A100: ~90 seconds Single H100: ~45 seconds	Single A100: ~180 seconds Single H100: ~90 seconds
Prompt Language	English*
Prompt Token Limit	224 Tokens	226 Tokens
Video Length	5 seconds or 10 seconds	6 seconds
Frame Rate	16 frames / second	8 frames / second
Position Encoding	3d_rope_pos_embed	3d_sincos_pos_embed	3d_rope_pos_embed	3d_rope_pos_embed + learnable_pos_embed
Download Link (Diffusers)	🤗 HuggingFace 🤖 ModelScope 🟣 WiseModel	🤗 HuggingFace 🤖 ModelScope 🟣 WiseModel	🤗 HuggingFace 🤖 ModelScope 🟣 WiseModel	🤗 HuggingFace 🤖 ModelScope 🟣 WiseModel	🤗 HuggingFace 🤖 ModelScope 🟣 WiseModel
Download Link (SAT)	🤗 HuggingFace 🤖 ModelScope 🟣 WiseModel	SAT

Data Explanation

Testing with the diffusers library enabled all optimizations included in the library. This scheme has not been tested on non-NVIDIA A100/H100 devices. It should generally work with all NVIDIA Ampere architecture or higher devices. Disabling optimizations can triple VRAM usage but increase speed by 3-4 times. You can selectively disable certain optimizations, including:

pipe.enable_sequential_cpu_offload()
pipe.vae.enable_slicing()
pipe.vae.enable_tiling()

In multi-GPU inference, enable_sequential_cpu_offload() optimization needs to be disabled.
Using an INT8 model reduces inference speed, meeting the requirements of lower VRAM GPUs while retaining minimal video quality degradation, at the cost of significant speed reduction.
PytorchAO and Optimum-quanto can be used to quantize the text encoder, Transformer, and VAE modules, reducing CogVideoX’s memory requirements, making it feasible to run the model on smaller VRAM GPUs. TorchAO quantization is fully compatible with torch.compile, significantly improving inference speed. FP8 precision is required for NVIDIA H100 and above, which requires source installation of torch, torchao, diffusers, and accelerate. Using CUDA 12.4 is recommended.
Inference speed testing also used the above VRAM optimizations, and without optimizations, speed increases by about 10%. Only diffusers versions of models support quantization.
Models support English input only; other languages should be translated into English during prompt crafting with a larger model.

Note

Use SAT for inference and fine-tuning SAT version models. Check our GitHub for more details.

Getting Started Quickly 🤗

This model supports deployment using the Hugging Face diffusers library. You can follow the steps below to get started.

We recommend that you visit our GitHub to check out prompt optimization and conversion to get a better experience.

Install the required dependencies

shell

# diffusers (from source)
# transformers>=4.46.2
# accelerate>=1.1.1
# imageio-ffmpeg>=0.5.1
pip install git+https://github.com/huggingface/diffusers
pip install --upgrade transformers accelerate diffusers imageio-ffmpeg

Run the code

python

import torch
from diffusers import CogVideoXPipeline
from diffusers.utils import export_to_video

prompt = "A panda, dressed in a small, red jacket and a tiny hat, sits on a wooden stool in a serene bamboo forest. The panda's fluffy paws strum a miniature acoustic guitar, producing soft, melodic tunes. Nearby, a few other pandas gather, watching curiously and some clapping in rhythm. Sunlight filters through the tall bamboo, casting a gentle glow on the scene. The panda's face is expressive, showing concentration and joy as it plays. The background includes a small, flowing stream and vibrant green foliage, enhancing the peaceful and magical atmosphere of this unique musical performance."

pipe = CogVideoXPipeline.from_pretrained(
    "THUDM/CogVideoX1.5-5B",
    torch_dtype=torch.bfloat16
)

pipe.enable_sequential_cpu_offload()
pipe.vae.enable_tiling()
pipe.vae.enable_slicing()

video = pipe(
    prompt=prompt,
    num_videos_per_prompt=1,
    num_inference_steps=50,
    num_frames=81,
    guidance_scale=6,
    generator=torch.Generator(device="cuda").manual_seed(42),
).frames[0]

export_to_video(video, "output.mp4", fps=8)

Quantized Inference

PytorchAO and Optimum-quanto can be used to quantize the text encoder, transformer, and VAE modules to reduce CogVideoX's memory requirements. This allows the model to run on free T4 Colab or GPUs with lower VRAM! Also, note that TorchAO quantization is fully compatible with torch.compile, which can significantly accelerate inference.

python

# To get started, PytorchAO needs to be installed from the GitHub source and PyTorch Nightly.
# Source and nightly installation is only required until the next release.

import torch
from diffusers import AutoencoderKLCogVideoX, CogVideoXTransformer3DModel, CogVideoXImageToVideoPipeline
from diffusers.utils import export_to_video
from transformers import T5EncoderModel
from torchao.quantization import quantize_, int8_weight_only

quantization = int8_weight_only

text_encoder = T5EncoderModel.from_pretrained("THUDM/CogVideoX1.5-5B", subfolder="text_encoder",
                                              torch_dtype=torch.bfloat16)
quantize_(text_encoder, quantization())

transformer = CogVideoXTransformer3DModel.from_pretrained("THUDM/CogVideoX1.5-5B", subfolder="transformer",
                                                          torch_dtype=torch.bfloat16)
quantize_(transformer, quantization())

vae = AutoencoderKLCogVideoX.from_pretrained("THUDM/CogVideoX1.5-5B", subfolder="vae", torch_dtype=torch.bfloat16)
quantize_(vae, quantization())

# Create pipeline and run inference
pipe = CogVideoXImageToVideoPipeline.from_pretrained(
    "THUDM/CogVideoX1.5-5B",
    text_encoder=text_encoder,
    transformer=transformer,
    vae=vae,
    torch_dtype=torch.bfloat16,
)

pipe.enable_model_cpu_offload()
pipe.vae.enable_tiling()
pipe.vae.enable_slicing()

prompt = "A little girl is riding a bicycle at high speed. Focused, detailed, realistic."
video = pipe(
    prompt=prompt,
    num_videos_per_prompt=1,
    num_inference_steps=50,
    num_frames=81,
    guidance_scale=6,
    generator=torch.Generator(device="cuda").manual_seed(42),
).frames[0]

export_to_video(video, "output.mp4", fps=8)

Additionally, these models can be serialized and stored using PytorchAO in quantized data types to save disk space. You can find examples and benchmarks at the following links:

Further Exploration

Feel free to enter our GitHub, where you'll find:

More detailed technical explanations and code.
Optimized prompt examples and conversions.
Detailed code for model inference and fine-tuning.
Project update logs and more interactive opportunities.
CogVideoX toolchain to help you better use the model.
INT8 model inference code.

Model License

This model is released under the CogVideoX LICENSE.

Citation

@article{yang2024cogvideox,
  title={CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer},
  author={Yang, Zhuoyi and Teng, Jiayan and Zheng, Wendi and Ding, Ming and Huang, Shiyu and Xu, Jiazheng and Yang, Yuanming and Hong, Wenyi and Zhang, Xiaohan and Feng, Guanyu and others},
  journal={arXiv preprint arXiv:2408.06072},
  year={2024}
}