Convrot Models
This repository contains a collection of popular generative models quantized to INT8 using the ConvRot method for optimized inference with minimal quality loss.
Source du modèle
- Type
- Other
- Taille du fichier
- 2,99–30,8 GB
- Modèle de base
- Z-Image Turbo
- Version
- vRevision eb66826
- Mis à jour
- 1 oct. 2026
- Téléchargements
- 0
- Tâches
- Édition d'image
- Source
- Hugging Face
Description de la source
📦 INT8 Quantized Model Collection (ConvRot)
This repository contains a collection of popular generative models quantized to INT8 using the ConvRot method for optimized inference with minimal quality loss.
Almost all models were quantized from the original BF16 weights using the quant_int8_convrot.py script with the --mseclip attribute for calibration.
🛠️ Quantization Details
- Base Precision: BF16 → INT8 ConvRot
- Method: ConvRot (Convolution Rotation)
- Calibration:
--mseclipfor per-layer clipping optimization (couple has--absmax) - Tool:
quant_int8_convrot.py(Comfy-Org)
flux-2-dev_int8_convrot_quantops.safetensors was first quantized in ComfyUI using ComfyUI-QuantOps, then convert to int8_convrot format via convert_to_comfy.py
📂 Included Models
| Model Name | Original Source | Notes |
|---|---|---|
| Checkpoints | ||
ltx-2.3-22b-dev_int8_convrot.safetensors | LTX 2.3 22B | — |
| Diffusion models | ||
flux-2-dev_int8_convrot_quantops.safetensors | Flux-2 dev | via ComfyUI-INT8-Fast |
flux-2-dev_int8_convrot_mseclip.safetensors | Flux-2 dev | — |
flux-2-dev_int8_convrot_absmax.safetensors | Flux-2 dev | — |
flux-2-klein-4b_int8_convrot.safetensors | Flux-2 Klein 4B | — |
flux-2-klein-9b_int8_convrot.safetensors | Flux-2 Klein 9B | — |
flux-2-klein-9b-kv_int8_convrot.safetensors | Flux-2 Klein 9B KV | — |
qwen_image_2512_int8_convrot.safetensors | Qwen Image 2512 | — |
qwen_image_edit_2511_int8_convrot.safetensors | Qwen Image Edit 2511 | — |
z_image_int8_convrot.safetensors | Z-Image | — |
z_image_turbo_int8_convrot.safetensors | Z-Image Turbo | — |
krea2_raw_int8_convrot.safetensors | Krea 2 Raw | — |
krea2_turbo_int8_convrot.safetensors | Krea 2 Turbo | — |
ideogram4_int8_convrot.safetensors | Ideogram 4 | via dequant FP8 |
ideogram4_unconditional_int8_convrot.safetensors | Ideogram 4 Uncond. | via dequant FP8 |
| 3D models | ||
trellis_2_int8_convrot.safetensors | Trellis 2 | — |
pixal3d_int8_convrot.safetensors | Pixal 3D | — |
| Text encoders | ||
gemma4_e4b_it_int8_convrot.safetensors | Gemma 4 E4B IT | — |
gemma_2_2b_it_elm_int8_convrot.safetensors | Gemma 2 2B IT | — |
qwen3.5_4b_int8_convrot.safetensors | Qwen 3.5 4B | — |
qwen3vl_8b_int8_convrot.safetensors | Qwen 3 VL 8B | — |
qwen3vl_4b_int8_convrot.safetensors | Qwen 3 VL 4B | — |
qwen_3_8b_int8_convrot.safetensors | Qwen 3 8B | — |
qwen_3_4b_int8_convrot.safetensors | Qwen 3 4B | — |
qwen_2.5_vl_7b_int8_convrot.safetensors | Qwen 2.5 VL 7B | — |
mistral_3_small_flux2_int8_convrot.safetensors | Mistral 3 Small for Flux2 | — |
Note: The Ideogram 4 models were sourced from Comfy-Org/Ideogram-4 and added to this repository for convenience purposes.
📜 Licenses
Each model retains the original license from its respective base model. Please refer to the original model cards for licensing terms:
- LTX 2.3
- Flux.2 dev
- Flux.2 klein 9b
- Flux.2 klein 9b-kv
- Flux.2 klein 4b
- Qwen-Image-2512
- Qwen-Image-Edit-2511
- Z-Image
- Z-Image Turbo
- Krea 2
- Ideogram 4
- Trellis.2
- Pixal3D
- Gemma
- Qwen
- Mistral
🙏 Acknowledgments
- Comfy-Org for the quantization script and bf16 models
- BobJohnson24 for
ComfyUI-INT8-Fastand conversion tools - silveroxides for
ComfyUI-QuantOpsnodes - All original model authors and teams
Sources
1 source
Sources
1 sourceVérifié 1 oct.
Artefacts du modèle
28 artefacts
Artefacts du modèle
28 artefactscheckpoints/ltx-2.3-22b-dev_int8_convrot.safetensors
safetensors · 23,9 GB · SHA-256 1853dbf29f63…47ba · Hugging Face
Téléchargerdiffusion_models/3d/pixal3d_int8_convrot.safetensors
safetensors · 5,20 GB · SHA-256 3003b0f95d6d…141f · Hugging Face
Téléchargerdiffusion_models/3d/trellis_2_int8_convrot.safetensors
safetensors · 4,89 GB · SHA-256 2c3efb637d48…20d7 · Hugging Face
Téléchargerdiffusion_models/flux/flux-2-dev_int8_convrot_quantops.safetensors
safetensors · 30,7 GB · SHA-256 e5febc3033a4…487c · Hugging Face
Téléchargerdiffusion_models/flux/flux-2-klein-4b_int8_convrot.safetensors
safetensors · 3,79 GB · SHA-256 f2ba417fcd5b…bb07 · Hugging Face
Téléchargerdiffusion_models/flux/flux-2-klein-9b_int8_convrot.safetensors
safetensors · 8,79 GB · SHA-256 031aa8bd2ebe…ffca · Hugging Face
Téléchargerdiffusion_models/flux/flux-2-klein-9b-kv_int8_convrot.safetensors
safetensors · 8,79 GB · SHA-256 c77a5030a03c…b672 · Hugging Face
Téléchargerdiffusion_models/flux/flux2-dev_int8_convrot_absmax.safetensors
safetensors · 30,8 GB · SHA-256 c4b7c7ff8aa4…b564 · Hugging Face
Téléchargerdiffusion_models/flux/flux2-dev_int8_convrot_mseclip.safetensors
safetensors · 30,8 GB · SHA-256 b4f9d748a38b…2096 · Hugging Face
Téléchargerdiffusion_models/ideogram/ideogram4_int8_convrot.safetensors
safetensors · 8,93 GB · SHA-256 a91640029434…163d · Hugging Face
Téléchargerdiffusion_models/ideogram/ideogram4_unconditional_int8_convrot.safetensors
safetensors · 8,93 GB · SHA-256 cd03ed94f244…c23d · Hugging Face
Téléchargerdiffusion_models/krea/krea2_raw_int8_convrot.safetensors
safetensors · 13,0 GB · SHA-256 d15476640b6e…a787 · Hugging Face
Téléchargerdiffusion_models/krea/krea2_turbo_int8_convrot.safetensors
safetensors · 13,0 GB · SHA-256 9377e6b8fea4…1777 · Hugging Face
Téléchargerdiffusion_models/qwen/qwen_image_2512_int8_convrot.safetensors
safetensors · 19,1 GB · SHA-256 cd6f8c0388cd…9ee2 · Hugging Face
Téléchargerdiffusion_models/qwen/qwen_image_edit_2511_int8_convrot.safetensors
safetensors · 19,1 GB · SHA-256 5fe66af54bdb…8eab · Hugging Face
Téléchargerdiffusion_models/z_image/z_image_int8_convrot.safetensors
safetensors · 5,75 GB · SHA-256 022661fa7ae8…64ff · Hugging Face
Téléchargerdiffusion_models/z_image/z_image_turbo_int8_convrot.safetensors
safetensors · 5,75 GB · SHA-256 d2f0574bb0a5…31c6 · Hugging Face
Téléchargertext_encoders/gemma_2_2b_it_elm_int8_convrot.safetensors
safetensors · 2,99 GB · SHA-256 8f53c4e831c0…6a05 · Hugging Face
Téléchargertext_encoders/gemma4_12b_int8_convrot.safetensors
safetensors · 12,2 GB · SHA-256 40c40e27e7e4…80dc · Hugging Face
Téléchargertext_encoders/gemma4_e4b_it_int8_convrot.safetensors
safetensors · 10,8 GB · SHA-256 c6bfb2a0534a…7930 · Hugging Face
Téléchargertext_encoders/mistral_3_small_flux2_int8_convrot.safetensors
safetensors · 17,2 GB · SHA-256 d058dded377b…2c4b · Hugging Face
Téléchargertext_encoders/qwen_2.5_vl_7b_int8_convrot.safetensors
safetensors · 8,92 GB · SHA-256 5ae7646f584c…99c4 · Hugging Face
Téléchargertext_encoders/qwen_3_4b_int8_convrot.safetensors
safetensors · 4,11 GB · SHA-256 60bd28906e4e…1c56 · Hugging Face
Téléchargertext_encoders/qwen_3_8b_int8_convrot.safetensors
safetensors · 8,79 GB · SHA-256 d7c84f7e1b45…b4cc · Hugging Face
Téléchargertext_encoders/qwen3.5_4b_int8_convrot.safetensors
safetensors · 4,98 GB · SHA-256 c09c01e44741…dcc0 · Hugging Face
Téléchargertext_encoders/qwen3.5_9b_int8_convrot.safetensors
safetensors · 11,0 GB · SHA-256 ee6a6df6a6cf…9ca6 · Hugging Face
Téléchargertext_encoders/qwen3vl_4b_int8_convrot.safetensors
safetensors · 4,53 GB · SHA-256 975604665916…e41a · Hugging Face
Téléchargertext_encoders/qwen3vl_8b_int8_convrot.safetensors
safetensors · 9,37 GB · SHA-256 ae3439a4b4a0…f7df · Hugging Face
TéléchargerExtraits de sources
3 extraits
Extraits de sources
3 extraits--- license: other tags: - quantized - int8 - comfyui - diffusion_models - convrot - text_encoders --- # 📦 INT8 Quantized Model Collection (ConvRot) This repository contains a collection of popular generative models quantized to **INT8** using the **ConvRot** method for optimized inference with minimal quality loss. Almost all models were quantized from the original **BF16** weights using the `quant_int8_convrot.py` script with the `--mseclip` attribute for calibration. --- ## 🛠️ Quantization Details - **Base Precision:** BF16 → INT8 ConvRot - **Method:** ConvRot (Convolution Rotation) - **Calibration:** `--mseclip` for per-layer clipping optimization (couple has `--absmax`) - **Tool:** [`quant_int8_convrot.py`](https://github.com/Comfy-Org/comfy-model-tools/blob/main/quant_int8_convrot.py) (Comfy-Org) `flux-2-dev_int8_convrot_quantops.safetensors` was first quantized in ComfyUI using [ComfyUI-QuantOps](https://github.com/silveroxides/ComfyUI-QuantOps), then convert to `int8_convrot` format via [`convert_to_comfy.py`](https://github.com/BobJohnson24/ComfyUI-INT8-Fast/blob/main/convert_to_comfy.py) --- ## 📂 Included Models | Model Name | Original Source | Notes | |------------|----------------|-------| |**Checkpoints** |`ltx-2.3-22b-dev_int8_convrot.safetensors` | LTX 2.3 22B | — | | |**Diffusion models** | `flux-2-dev_int8_convrot_quantops.safetensors` | Flux-2 dev | via ComfyUI-INT8-Fast | | `flux-2-dev_int8_convrot_mseclip.safetensors` | Flux-2 dev | — | | `flux-2-dev_int8_convrot_absmax.safetensors` | Flux-2 dev | — | | `flux-2-klein-4b_int8_convrot.safetensors` | Flux-2 Klein 4B | — | | `flux-2-klein-9b_int8_convrot.safetensors` | Flux-2 Klein 9B | — | | `flux-2-klein-9b-kv_int8_convrot.safetensors` | Flux-2 Klein 9B KV | — | | `qwen_image_2512_int8_convrot.safetensors` | Qwen Image 2512 | — | | `qwen_image_edit_2511_int8_convrot.safetensors` | Qwen Image Edit 2511 | — | | `z_image_int8_convrot.safetensors` | Z-Image | — | | `z_image_turbo_int8_convrot.safetensors` | Z-Image Turbo | — | | `krea2_raw_int8_convrot.safetensors` | Krea 2 Raw | — | | `krea2_turbo_int8_convrot.safetensors` | Krea 2 Turbo | — | | `ideogram4_int8_convrot.safetensors` | Ideogram 4 | via dequant FP8 | | `ideogram4_unconditional_int8_convrot.safetensors` | Ideogram 4 Uncond. | via dequant FP8 | | |**3D models** | `trellis_2_int8_convrot.safetensors` | Trellis 2 | —...
Source context: 0 downloads · 23 likes · Repo supermind/int8_convrot_models
Source context: 0 downloads · 6 likes · Repo supermind/int8_convrot_models
