Updated with the fixed text encoder.
Shrinking the LTX 2.5 text encoder a bit using W4A8. So far, it seems to hold up well, but I’d still consider it experimental for now.
Description
FAQ
Comments (2)
censure?
its the original only quantized so yes
Updated with the fixed text encoder.
Shrinking the LTX 2.5 text encoder a bit using W4A8. So far, it seems to hold up well, but I’d still consider it experimental for now.
censure?
its the original only quantized so yes