ComfyUI Node: 🍳 Kitchen NVFP4 Converter

🍳 Kitchen NVFP4 Converter:

The ConvertToNVFP4 node is designed to facilitate the conversion of model weights into the NVFP4 format, a specialized data representation that optimizes the performance of AI models, particularly in environments utilizing NVIDIA hardware. This node is part of the ComfyUI Kitchen suite, which aims to enhance the efficiency and speed of model processing by leveraging advanced quantization techniques. By converting model weights to NVFP4, you can achieve significant reductions in memory usage and computational overhead, making it ideal for deploying large-scale models on resource-constrained devices. The node intelligently handles different model types and adapts its conversion strategy accordingly, ensuring compatibility and optimal performance across various AI applications.

🍳 Kitchen NVFP4 Converter Input Parameters:

model_name

The model_name parameter specifies the name of the diffusion model you wish to convert. It is crucial as it determines the source model whose weights will be transformed into the NVFP4 format. This parameter accepts a list of available model filenames, ensuring you select the correct model for conversion.

output_filename

The output_filename parameter defines the name of the output file where the converted model will be saved. This is a string parameter with a default value of "model-nvfp4". It allows you to specify a custom name for the converted model file, making it easier to organize and identify different versions of your models.

model_type

The model_type parameter indicates the type of model being converted. It offers several options, such as "Z-Image-Turbo", "Flux.1-dev", and "LTX-2-19b-dev-or-distilled", among others. This parameter is essential as it influences the conversion process by determining the specific quantization and optimization strategies applied to the model weights. The default value is "Z-Image-Turbo".

device

The device parameter specifies the hardware device on which the conversion process will be executed. It can be set to either "cuda" or "cpu", with "cuda" being the default option. This parameter is important because it affects the speed and efficiency of the conversion process, with CUDA-enabled devices typically offering faster performance due to GPU acceleration.

🍳 Kitchen NVFP4 Converter Output Parameters:

status

The status output parameter provides a string message indicating the success or failure of the conversion process. This output is crucial for verifying that the model weights have been successfully converted to the NVFP4 format and saved to the specified output file. It helps you quickly identify any issues that may have occurred during the conversion.

🍳 Kitchen NVFP4 Converter Usage Tips:

• Ensure that the model_name parameter is correctly set to the desired diffusion model to avoid conversion errors.
• Use the device parameter to leverage GPU acceleration by selecting "cuda" for faster conversion times, especially when working with large models.
• Customize the output_filename to keep track of different model versions and conversions, which can be helpful for model management and deployment.

🍳 Kitchen NVFP4 Converter Common Errors and Solutions:

Model file not found

• Explanation: This error occurs when the specified model_name does not match any available model files.
• Solution: Verify that the model_name is correctly specified and corresponds to an existing model file in the designated directory.

Unsupported model type

• Explanation: This error arises when the model_type parameter is set to a value not supported by the node.
• Solution: Ensure that the model_type is one of the predefined options, such as "Z-Image-Turbo" or "LTX-2-19b-dev-or-distilled".

Device not available

• Explanation: This error happens when the selected device is not available or improperly configured.
• Solution: Check that the specified device is correctly set to either "cuda" or "cpu" and that the necessary hardware and drivers are properly installed and configured.