Modelpublic

brewbadgertim/Huihui-Qwen3-VL-4B-Instruct-abliterated-Quants

sourceHugging Faceapache-2.0updated 3mo agoView on Hugging Face

0likes

Model Card

Huihui-Qwen3-VL-4B-Instruct-abliterated — ComfyUI Edition

This repo packages the huihui-ai/Huihui-Qwen3-VL-4B-Instruct-abliterated Vision-Language Model in two ready-to-use formats for ComfyUI:

| File | Size | Format

| Huihui-Qwen3-VL-4B-Instruct-abliterated.safetensors | 8.88 GiB | bf16

| Huihui-Qwen3-VL-4B-Instruct-abliterated-int8.safetensors | 4.13 GiB | int8

—Base model: `huihui-ai/Huihui-Qwen3-VL-4B-Instruct-abliterated` (apache-2.0)
—Origin model: `Qwen/Qwen3-VL-4B-Instruct`
—Abliteration: only the text part was processed (not the vision part), so the visual encoder behaves identically to the upstream model.

The INT8 conversion was done on GPU (NVIDIA RTX 5080) in ~6.5 seconds using Star Nodes' 'Star Ultimate Model Converter' inside ComfyUI desktop.

—License: apache-2.0 (inherited from upstream Qwen/Qwen3-VL-4B-Instruct).
—Abliteration notice: This is an uncensored variant. The safety filtering has been significantly reduced, potentially generating sensitive, controversial, or inappropriate content. Use with caution. See the upstream model card for the full disclaimer.
—No warranty. Users are solely responsible for any consequences arising from use of this model.