-
|
After last update is no more possible training on <8gb? Before update all working and consuming around 6gb, now always is cuda out of memory, in the same settings/ |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 2 replies
-
|
yes even lora doens't work anymore :( before it used less than 5 GB. I have 8 GB vram. It tries to allocate the same vram wether i check lora or not. Both times: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 8.00 GiB total capacity; 7.14 GiB already allocated; 0 bytes free; 7.28 GiB reserved in total by PyTorch) |
Beta Was this translation helpful? Give feedback.
-
|
can confirm this, could use a batch of 4 with DB and 12 with lora, now can't do more than 4 with lora and 1 with regular DB. hopefully @d8ahazard will confirm what's going on and fix |
Beta Was this translation helpful? Give feedback.
-
|
yes, in order to start learning again on 8gb, you need to install a separate web-gui for Lora cloneofsimo/lora#102 |
Beta Was this translation helpful? Give feedback.
yes even lora doens't work anymore :( before it used less than 5 GB. I have 8 GB vram. It tries to allocate the same vram wether i check lora or not. Both times: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 8.00 GiB total capacity; 7.14 GiB already allocated; 0 bytes free; 7.28 GiB reserved in total by PyTorch)