mlx-community
/

Qwen-Image-Flash-8bit

Model card Files Files and versions

Qwen-Image-Flash-8bit

29.5 GB

Ctrl+K

Ctrl+K

1 contributor

History: 10 commits

xocialize's picture

transformer: int8 DiT (attn+FFN+modulation, group 64)

677924a verified about 10 hours ago

scheduler
scheduler: static shift-3 FlowMatchEuler about 10 hours ago
text_encoder
text_encoder: int8 Qwen2.5-VL-7B language model (no vision tower) about 10 hours ago
tokenizer
tokenizer (incl. converted tokenizer.json) about 10 hours ago
transformer
transformer: int8 DiT (attn+FFN+modulation, group 64) about 10 hours ago
vae
vae: unquantized AutoencoderKLQwenImage about 10 hours ago
.gitattributes

1.58 kB
tokenizer (incl. converted tokenizer.json) about 10 hours ago
LICENSE

11.3 kB
LICENSE (Apache-2.0, additional information per upstream) about 10 hours ago
NOTICE

1.25 kB
NOTICE: NVIDIA OML attribution + statement of quantization changes about 10 hours ago
README.md

4.42 kB
README: int8 tier — measured quality, footprints, and why there is no 4-bit tier about 10 hours ago
model_index.json

443 Bytes
model_index.json about 10 hours ago