File size: 1,654 Bytes
3a93d0e 6e9de70 8c54d99 3a93d0e | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 | NOTICE
======
Qwen is licensed under the Qwen RESEARCH LICENSE AGREEMENT, Copyright (c) 2026
Hangzhou Tongyi Laboratory Technology Co., Ltd. All Rights Reserved.
This repository is a derivative work of Qwen-Image-2.1. The full agreement is in
`LICENSE`, a copy of it is given to every recipient of these files.
Modified files, as required by section 2.b of the agreement:
transformer/config.json added the `text_fusion_config` key, `_class_name` changed to
QwenImage21FusionTransformer2DModel
transformer/*.safetensors re-saved in fp16, plus the 66 `text_fusion.*` tensors of the
adapter from the image21-08b-text-encoder-adapter project (adapter_v11)
vae/ copied unchanged (fp32)
scheduler/ modified: dynamic shifting off, plain static `shift` 5.0, `shift_terminal`
removed (was 0.02) — the sdxs-micro schedule
text_encoder/ different model: Qwen3.5-0.8B instead of Qwen3-VL-8B. Upstream files are
re-saved in fp16 (bf16 -> fp16, single `model.safetensors` instead of
the sharded bf16 file); `config.json` is unchanged. Modified under the
Apache License 2.0, section 4.b.
tokenizer/, processor/ the Qwen3.5-0.8B tokenizer and processor, byte-identical to upstream
Files added by this repository: pipeline.py, transformer.py, example.py, model_index.json,
media/, README.md.
Improved using Qwen.
The Qwen3.5-0.8B text encoder is redistributed under the Apache License 2.0, see
`LICENSE-Qwen3.5-0.8B`.
|