ethanfel/Qwen3-VL-4B-Heretic-ComfyUI-ConvRot-INT8
This repository contains a true dynamic-ConvRot INT8 conversion of DreamFast/Qwen3-VL-4b-Heretic, packaged as a single-file ComfyUI text encoder for Krea 2.
Fonte do modelo
- Tipo
- Other
- Tamanho do arquivo
- 4,50 GB
- Modelo base
- Qwen3
- Versão
- vRevision c4ac8d8
- Atualizado
- 28 de set. de 2026
- Downloads
- 0
- Fonte
- Hugging Face
Fontes
1 fonte
Fontes
1 fonteVerificado 28 de set.
Artefatos de modelo
1 artefato
Artefatos de modelo
1 artefatoqwen3-vl-4b-heretic_int8_dynamic_convrot.safetensors
safetensors · 4,50 GB · SHA-256 c766f6f078bb…4778 · Hugging Face
BaixarTrechos de fonte
2 trechos
Trechos de fonte
2 trechos--- license: apache-2.0 base_model: - DreamFast/Qwen3-VL-4b-Heretic tags: - qwen3-vl - qwen3 - comfyui - krea - krea2 - text-encoder - int8 - convrot - abliteration - uncensored - quantized --- # Qwen3-VL-4B Heretic — Dynamic ConvRot INT8 for ComfyUI Krea 2 This repository contains a true dynamic-ConvRot INT8 conversion of [DreamFast/Qwen3-VL-4b-Heretic](https://huggingface.co/DreamFast/Qwen3-VL-4b-Heretic), packaged as a single-file ComfyUI text encoder for Krea 2. Krea 2 uses Qwen3-VL-4B-Instruct conditioning, not the text-only Qwen3-4B architecture. The Heretic source is an abliterated derivative of [Qwen/Qwen3-VL-4B-Instruct](https://huggingface.co/Qwen/Qwen3-VL-4B-Instruct). ## Downloads | Precision | File | Size | Notes | | --- | --- | ---: | --- | | INT8 | `qwen3-vl-4b-heretic_int8_dynamic_convrot.safetensors` | 4.84 GB | Hosted in this repository; recommended | | BF16 | [Upstream BF16 safetensors](https://huggingface.co/DreamFast/Qwen3-VL-4b-Heretic-ComfyUI/blob/main/qwen3-vl-4b-heretic.safetensors) | 8.88 GB | Publicly hosted by DreamFast; not duplicated here | INT8 SHA-256: ```text c766f6f078bbfe7474699faaceb9484251b0c7c8dd4bf2d5730b320feca54778 ``` ## Quantization Built with [silveroxides/convert_to_quant](https://github.com/silveroxides/convert_to_quant) 1.3.1 at commit `ee83a4d90d18f52824c3b6b23e45bc7514fb027e`. - 356 INT8 weights with per-row FP32 scales - All 356 quantized layers contain `convrot: true` ComfyUI metadata - Dynamic ConvRot group sizes: - 216 layers at 256 - 72 layers at 1,024 - 68 layers at 4,096 - 357 embeddings, norms, biases, and other sensitive tensors remain BF16 - Deterministic seed 42 - Prodigy learned rounding Conversion command: ```bash ctq \ -i qwen3-vl-4b-heretic.safetensors \ -o qwen3-vl-4b-heretic_int8_dynamic_convrot.safetensors \ --int8 \ --scaling_mode row \ --dynamic-convrot \ --comfy_quant \ --save-quant-metadata \ --low-memory \ --device cuda \ --heur \ --exclude-layers \ '(embed|norm|bias|lm_head|spiece_model|multi_modal_projector|patch_embed|patch_conv)' \ --verbose NORMAL \ --manual-seed 42 ``` ## Fidelity versus BF16 Four varied prompts were encoded through ComfyUI's Krea 2 loader with BF16 as the reference. | Metric | INT8 dynamic ConvRot | | --- | ---: | | Mean cosine similarity | 0.999853 | | Mean relative L2 error | 2.13% | | Mean absolute error | 0.03439...
Source context: 0 downloads · 17 likes · Repo ethanfel/Qwen3-VL-4B-Heretic-ComfyUI-ConvRot-INT8
