pottokao/Qwen-Image-2.1-Text-Encoder-Heretic-GGUF

Showcase: descriptive style names (remove photographer names, incl. CJK); fix asset URLs

Document Q6_K and Q8_0 quantizations

Add qwen3vl_8b_heretic-Q8_0.gguf (verified heretic source; vision intact; Q6_K embd protected at Q8_0)

Add qwen3vl_8b_heretic-Q6_K.gguf (verified heretic source; vision intact; Q6_K embd protected at Q8_0)

Attribute showcase images to the ComfyUI-QwenImage-PhotoStyles node

README: showcase links open the image viewer instead of downloading

README: showcase images link to full size

README: remove showcase image 03

README: Q4_K_M file table under the showcase

README: showcase note as small text under the images

README: showcase made with Q4_K_M throughout

README: AI-generated / style homage disclaimer on showcase

README: remove showcase image 17

README: showcase captions (style)

README: all-GGUF showcase

README: all-GGUF table linking the DiT GGUF repo

README: move patch explanation to the bottom, cover both TE and DiT fixes

README: two-node manual install (git or ZIP), guide link, DiT fix note

README: recommended sampler settings (25 steps, cfg 1 -> 3 schedule for text)

README: bf16 single file now in this repo; sizes as shown on HF

Add bf16 single-file ComfyUI text encoder (any device)

README: add PE-T2I / PE-I2I rewriter links, real rewrite example, NVFP4 showcase

README: don't-rename note, editing verified

README: GGUF text encoder now works via ComfyUI-GGUF-Qwen3VL-TE add-on; reorganize

Add which-file-to-download table (FP8/CUDA, bf16/non-CUDA, GGUF/llama.cpp) at top

Add evidence links (ComfyUI issue #9868, official docs) for the GGUF-TE platform gap

Add FP8 text encoder (verified generates images in official Qwen-Image-2.1 workflow)

ComfyUI has no GGUF path for the QI2.1 text encoder yet; add working FP8 safetensors as the ComfyUI option

Add prominent warning: GGUF not usable as QI2.1 text encoder in ComfyUI; point to W4A8

Add NOTICE (Apache-2.0 attribution + modifications)

Replace LICENSE with Apache-2.0

Relicense to Apache-2.0 (weights are Qwen3-VL-8B-Instruct, Apache-2.0)

Add comfyui tag (usable via the ComfyUI-GGUF custom node)

Upload README.md with huggingface_hub

Upload LICENSE with huggingface_hub

Upload folder using huggingface_hub

initial commit