70% Size, 100% Accuracy: Lossless LLM Compression for Efficient GPU Inference via Dynamic-Length Float
Paper
•
2504.11651
•
Published
•
28
Tianyi Zhang
LeanQuant
AI & ML interests
None yet
Recent Activity
new activity
about 5 hours ago
DFloat11/Wan2.2-T2V-A14B-2-DF11:This is interesting
posted
an
update
1 day ago
🚀 Full-Quality Wan2.2 Video Generation on a single 24GB GPU — Powered by DFloat11
We just released the DFloat11 compressed Wan2.2 models. Now you can run full-quality Wan2.2 video generation on a single 24GB GPU, thanks to DFloat11 compression and CPU offloading.
🔗 Image-to-Video: https://huggingface.co/DFloat11/Wan2.2-I2V-A14B-DF11
🔗 Text-to-Video: https://huggingface.co/DFloat11/Wan2.2-T2V-A14B-DF11
updated
a model
1 day ago
DFloat11/Wan2.2-I2V-A14B-2-DF11
Organizations
authored
a
paper
4 months ago