Convrot Models

📦 INT8 Quantized Model Collection (ConvRot) This repository contains a collection of popular generative models quantized to using the method for optimized inference with minimal quality loss. Almost all models were quantized from the original weights using the script with the attribute for calibration. --- ## 🛠️ Quantization Details - BF16 → INT8 ConvRot - ConvRot (Convolution Rotation) - for per-layer clipping optimization (couple has ) - (Comfy-Org) was first quantized in ComfyUI using , then convert to format via --- ## 📂 Included Models | Model Name | Original Source | Notes | |------------|----------------|-------| | | | LTX 2.3 22B | — | | | | | Flux-2 dev | via ComfyUI-INT8-Fast | | | Flux-2 dev | — | | | Flux-2 dev | — | | | Flux-2 Klein 4B | — | | | Flux-2 Klein 9B | — | | | Flux-2 Klein 9B KV | — | | | Qwen Image 2512 | — | | | Qwen Image Edit 2511 | — | | | Z-Image | — | | | Z-Image Turbo | — | | | Krea 2 Raw | — | | | Krea 2 Turbo | — | | | Ideogram 4 | via dequant FP8 | | | Ideogram 4 Uncond. | via dequant FP8 | | | | | Trellis 2 | — | | | Pixal 3D | — | | | | | Gemma 4 E4B IT | — | | | Gemma 2 2B IT | — | | | Qwen 3.5 4B | — | | | Qwen 3 VL 8B | — | | | Qwen 3 VL 4B | — | | | Qwen 3 8B | — | | | Qwen 3 4B | — | | | Qwen 2.5 VL 7B | — | | | Mistral 3 Small for Flux2 | — | > The Ideogram 4 models were sourced from and added to this repository for convenience purposes. --- ## 📜 Licenses Each model retains the from its respective base model. Please refer to the original model cards for licensing terms: - - - - - - - - - - - - - - - - - - - --- ## 🙏 Acknowledgments - for the quantization script and bf16 models - for and conversion tools - for nodes - All original model authors and teams