--- title: Video-GenCompression canonical_url: "https://www.modelscope.cn/models/kocham/Video-GenCompression" md_url: "https://www.modelscope.cn/models/kocham/Video-GenCompression.md" repository: kocham/Video-GenCompression chinese_name: "视频生成式压缩" last_updated: 2025-05-11 license: other parameters: 7.7B tensor_type: - BF16 - F32 library_name: - safetensors - tensorflow - pytorch frameworks: - Pytorch downloads: 25 stars: 2 tags: - nvidia - cosmos --- # Video-GenCompression > Video-GenCompression - kocham 在 ModelScope 开源的模型。Cosmos-Tokenize1: A suite of image and video tokenizers kocham/Video-GenCompression 是 ModelScope 魔搭社区上的 7.7B 参数机器学习模型，采用 other 许可。 - **Repository**: kocham/Video-GenCompression - **License**: other - **Parameters**: 7.7B - **Tags**: nvidia, cosmos - **Downloads**: 25 - **Stars**: 2 - **Last updated**: 2025-05-11 Source: https://www.modelscope.cn/models/kocham/Video-GenCompression --- # **Cosmos-Tokenize1**: A suite of image and video tokenizers [**Cosmos**](https://huggingface.co/collections/nvidia/cosmos-tokenize1-67d354326c497d65d5cf7ee9) | [**Code**](https://github.com/nvidia-cosmos/cosmos-predict1) | [**Paper**](https://arxiv.org/abs/2501.03575) | [**Paper Website**](https://research.nvidia.com/labs/dir/cosmos-predict1) # Model Overview ## Description: **Cosmos-Tokenize1** is a suite of visual tokenizers for images and videos that delivers various compression rates while maintaining high reconstruction quality. Cosmos Tokenizer can serve as an effective and efficient building block in both diffusion-based and autoregressive models for image and video generation. This model is ready for commercial use. Our tokenizers come in two types: **Continuous** (C) and **Discrete** (D), each with **Image** (I) and **Video** (V) variants: * Continuous tokenizers encode visual data into continuous latent embeddings, as shown in latent diffusion models like [Stable Diffusion](https://github.com/CompVis/stable-diffusion). These embeddings are suitable for models that generate data by sampling from continuous distributions. * Discrete tokenizers encode visual data into discrete latent codes, mapping them into quantized indices, as seen in autoregressive transformers such as [VideoPoet](https://sites.research.google/videopoet/). This discretization is required for models that generate data by optimizing the cross-entropy loss, such as the GPT models. | | Continuous ( C ) | Discrete ( D ) | | ------------------|---------------------|---------------------| | **Images ( I )** | Cosmos-Tokenizer-CI | Cosmos-Tokenizer-DI | | **Videos ( V )** | Cosmos-Tokenizer-CV | Cosmos-Tokenizer-DV | **Model**: [Cosmos-Tokenize1-CV8x8x8-360p](https://huggingface.co/nvidia/Cosmos-Tokenize1-CV8x8x8-720p), a continuous video tokenizer with 8x temporal compression, and 8x8 spatial compression rate. **Model Developer**: NVIDIA ## Model Versions The Cosmos-Tokenize1 includes the following tokenizers: * **Continuous Tokenizers** * [Cosmos-Tokenize1-CI8x8-360p](https://huggingface.co/nvidia/Cosmos-Tokenize1-CI8x8-360p) (8x8 spatial compression, 360p and above) * [Cosmos-Tokenize1-CI16x16-360p](https://huggingface.co/nvidia/Cosmos-Tokenize1-CI16x16-360p) (16x16 spatial compression, 360p and above) * [Cosmos-Tokenize1-CV4x8x8-360p](https://huggingface.co/nvidia/Cosmos-Tokenize1-CV4x8x8-360p) (4x temporal compression, 8x8 spatial compression, 360p and above, 49 frames context) * [Cosmos-Tokenize1-CV8x8x8-720p](https://huggingface.co/nvidia/Cosmos-Tokenize1-CV8x8x8-720p) (8x temporal compression, 8x8 spatial compression, 720p and above, 121 frames context) * **Discrete Tokenizers** * [Cosmos-Tokenize1-DI8x8-360p](https://huggingface.co/nvidia/Cosmos-Tokenize1-DI8x8-360p) (8x8 spatial compression, 360p and above) * [Cosmos-Tokenize1-DI16x16-360p](https://huggingface.co/nvidia/Cosmos-Tokenize1-DI16x16-360p) (16x16 spatial compression, 360p and above) * [Cosmos-Tokenize1-DV4x8x8-360p](https://huggingface.co/nvidia/Cosmos-Tokenize1-DV4x8x8-360p) (4x temporal compression, 8x8 spatial compression, 360p and above, 49 frames context) * [Cosmos-Tokenize1-DV8x16x16-720p](https://huggingface.co/nvidia/Cosmos-Tokenize1-DV8x16x16-720p) (8x temporal compression, 16x16 spatial compression, 720p and above, 49 frames context) ### License: This model is released under the [NVIDIA Open Model License](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license). For a custom license, please contact [cosmos-license@nvidia.com](mailto:cosmos-license@nvidia.com). Under the NVIDIA Open Model License, NVIDIA confirms: * Models are commercially usable. * You are free to create and distribute Derivative Models. * NVIDIA does not claim ownership to any outputs generated using the Models or Derivative Models. **Important Note**: If you bypass, disable, reduce the efficacy of, or circumvent any technical limitation, safety guardrail or associated safety guardrail hyperparameter, encryption, security, digital rights management, or authentication mechanism contained in the Model, your rights under [NVIDIA Open Model License Agreement](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license) will automatically terminate. ## Model Architecture: Cosmos-Tokenize1-CV8x8x8-720p is a lightweight and computationally efficient architecture, featuring a temporally causal design. Specifically, we employ causal temporal convolution and causal temporal attention layers to preserve the natural temporal order of video frames, ensuring seamless tokenization of images and videos using a single unified network architecture. The encoder and decoder form a symmetrical pair, which are mirrors of each other. The encoder starts with a 2-level [Haar wavelet](https://link.springer.com/book/10.1007/978-3-319-04295-4) transform layer, which down-samples inputs by a factor of 4 in both spatial and temporal dimensions. Likewise, the decoder ends with an inverse wavelet transform. We employ the vanilla autoencoder (AE) formulation to model the latent space for continuous tokenizers. For discrete tokenizers, we adopt the [Finite-Scalar-Quantization](https://openreview.net/forum?id=8ishA3LxN8) (FSQ) as the latent space quantizer. ![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/638fb8cf2380ffd99caf8c2a/gQH5n9iCEtqZc7uutUwdL.jpeg) ## Input/Output Specifications ### Encoder * **Input** * **Type:** Images or Videos * **Format:** RGB (Red, Green, Blue) * **Properties:** * **Resolution:** Minimum: 256px (shorter side). Maximum: Up to 4K * **Video Length:** Up to 8 seconds for 1080p videos (bounded by A100 80G GPU memory; higher resolutions will have shorter supported durations) * **Output** * **Type:** Tokens * **Properties:** * Continuous-valued feature vectors with a dimensionality of 16 ### Decoder * **Input** * **Type:** Tokens * **Properties:** * Continuous-valued feature vectors with a dimensionality of 16 * **Output** * **Type** Images or Videos (matching input type) * **Format:** RGB (Red, Green, Blue) * **Properties:** * **Resolution:** Same as input resolution * **Video Length:** Same as input video length ## Software Integration (Required For NVIDIA Models Only): **Runtime Engine(s):** * [Cosmos-Predict1](https://github.com/nvidia-cosmos/cosmos-predict1) **Supported Hardware Microarchitecture Compatibility:** * NVIDIA Ampere (e.g., A100) * NVIDIA Hopper (e.g., H100) Note: We have only tested Cosmos-Tokenize1 with BF16 precision on Ampere and Hopper GPUs. If you are using older versions of NVIDIA GPUs (e.g., NVIDIA Volta GPUs), you may need to switch to FP32 precision. **Operating System(s):** * Linux (We have not tested on other operating systems.) # Usage Inference Engines: * See [Cosmos-Predict1](https://github.com/nvidia-cosmos/cosmos-predict1) (PyTorch) for details. # Evaluation ## Tokenization Comparison We have evaluated the additional **Cosmos Tokenizer** models on [DAVIS](https://davischallenge.org/) video benchmark dataset. | Tokenizer | Compression Ratio | Height | Num. of Frames | Quantization | PSNR (DAVIS) | SSIM (DAVIS) | rFVD (DAVIS) | |-----------|------------------|--------------|--------------|--------------|--------------|--------------|--------------| | CogVideoX | 4×4×4 | - | - | VAE | 31.74 | 0.860 | 19.58 | | OmniTokenizer | 4×8×8 | - | - | VAE | 29.04 | 0.710 | 117.66 | | Cosmos-Tokenizer-CV | 4×8×8 | 720 | 49 | AE | **35.28** | **0.890** | **15.93** | | Cosmos-Tokenizer-CV | 8×8×8 | 720 | 49 | AE | 34.10 | 0.850 | 30.16 | | Cosmos-Tokenizer-CV | 8×8×8 | 720 | 121 | AE | 34.32 | 0.867 | 23.49 | | Cosmos-Tokenizer-CV | 8×16×16 | 720 | 49 | AE | 32.55 | 0.770 | 93.82 | * We compare with the state-of-the-art discrete video tokenizer, [OmniTokenizer](https://github.com/FoundationVision/OmniTokenizer). * Evaluation metrics: * Peak Signal-to-Noise Ratio (PSNR) * Structural Similarity (SSIM) * Reconstruction Fréchet Video Distance (rFVD) ## Runtime Comparison The following table shows the number of parameters and the averaged encoding and decoding times per image or video frame, measured on a single A100 80GB GPU. For comparison, we also list the parameters and average speeds of prior state-of-the-art tokenizer(s) with the same compression ratio. | Tokenizer | Resolution | Compression Ratio | Parameters | Time (ms) | |----------------|------------|-------------------|------------|-----------| | CogVideoX | 720x1280 | 4×8×8 | 216M | 414 | | OmniTokenizer | 720x1280 | 4×8×8 | 54M | 82.9 | | Cosmos-Tokenizer-CV | 720x1280 | 4×8×8 | 105M | **34.8** | Note: We benchmarked the runtime for images under the 8x8 compression and videos under the 4×8×8 compression. Tokenizers with different compression ratios are not included in this comparison. ## Ethical Considerations NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their internal model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse. ### Plus Plus (++) Promise We value you, the datasets, the diversity they represent, and what we have been entrusted with. This model and its associated data have been: * Verified to comply with current applicable disclosure laws, regulations, and industry standards. * Verified to comply with applicable privacy labeling requirements. * Annotated to describe the collector/source (NVIDIA or a third-party). * Characterized for technical limitations. * Reviewed to ensure proper disclosure is accessible to, maintained for, and in compliance with NVIDIA data subjects and their requests. * Reviewed before release. * Tagged for known restrictions and potential safety implications. For more detailed information on ethical considerations for this model, please see the subcards of Explainability, Bias, Safety & Security, and Privacy below. Please report security vulnerabilities or NVIDIA AI Concerns [here](https://www.nvidia.com/en-us/support/submit-security-vulnerability/). ### Bias Field | Response :---------------------------------------------------------------------------------------------------|:--------------- Participation considerations from adversely impacted groups [protected classes](https://www.senate.ca.gov/content/protected-classes) in model design and testing: | None Measures taken to mitigate against unwanted bias: | None ### Explainability Field | Response :------------------------------------------------------------------------------------------------------|:--------------------------------------------------------------------------------- Intended Application & Domain: | Tokenization of images and videos Model Type: | Auto-Encoder Intended Users: | Generative AI developers for image and video generation models Output: | Images/Videos and Latent Tokens Describe how the model works: | Compresses and decompresses visual input (image/video). Technical Limitations: | Due to tokenizer compression limitations, some visual information (such as small text and other structured fine details) may not be reconstructed accurately. The tokenizers may not produce as high of a reconstruction results for videos with low resolution, e.g. less than 320p. Verified to have met prescribed NVIDIA quality standards: | Yes Performance Metrics: | Peak Signal-to-Noise Ratio (PSNR), Structural Similarity (SSIM), Reconstruction Fréchet Video Distance (rFVD), Reconstruction Fréchet Inception Distance (rFID), Latency Potential Known Risks: | Tokenizer's output can parse all forms of input, including what may be considered toxic, offensive, or indecent. Licensing: | [NVIDIA Open Model License](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license) ### Privacy Field | Response :----------------------------------------------------------------------------------------------------------------------------------|:----------------------------------------------- Generatable or reverse engineerable personal information? | No Protected class data used to create this model? | None Known Was consent obtained for any personal data used? | None Known How often is dataset reviewed? | Before Release Is there provenance for all datasets used in training? | Yes Does data labeling (annotation, metadata) comply with privacy laws? | Yes ### Safety Field | Response :---------------------------------------------------|:---------------------------------- Model Application(s): | Tokenization of images and videos Describe the life critical impact (if present). | None Known Use Case Restrictions: | [NVIDIA Open Model License](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license) Model and dataset restrictions: | The Principle of least privilege (PoLP) is applied limiting access for dataset generation and model development. Restrictions enforce dataset access during training, and dataset license constraints adhered to. Model checkpoints are made available on Hugging Face, and may become available on cloud providers' model catalog.