beta6 w4a8 please?
do you plan to release beta6 w4a8?
Yes, TenStrip. Please upload w4a8 version if you can!
You can get some w4 and w6 hybrid versions of beta6 on my huggingface repo here: https://huggingface.co/LokkenJP/TenEros_Max_optimized_learned
Thanks a lot! i tried it but the video looks choppy, it says it is 24 frames but looks like 16 fps. could it be my workflow or could it be due the compression? i was using beta5 w4a8 before and it still works the best for me!
That happens sometimes depending on prompting I believe, and much more if you use loras or exotic nodes like H3 cache or Sparse attn. I guess it might be the base TenStrip used for its merge. If you want to be sure, try to test the HQ w6 18GB quantization I did of Beta6 with a barebones workflow. This one is as good as it gets under 20GB and I've been using it without issues for several days now.
ok, i'll try it. thanks!