kaibioinfo

Hi, I have the same problem. I cannot get LORA run on a 24GB GPU. The problem is not the inference, it happens when running "model_pred = unet(noisy_latents, timesteps, encoder_hidden_states).sample"...

I'm currently playing around a bit to narrow down the error. What I can say is: - I can train complete finetuning (dreambooth) on my 24GB gpu even with using...

I think I solved the problem. The script is not using the xformer variant of the Lora cross attention processor. I will try to make a fix.

Here is the pull request: https://github.com/huggingface/diffusers/pull/2648

Put them into the same file and give them the same FEATURE_ID. They should be merged then. MZMine does also have a "export SIRIUS mgf" tool that should format the...

Woah, this is a mean one: the subtree of [C42H41O18 + 1] looks perfect: several hexose moieties, water losses - sure it gets high score and is picked, although the...

Hi, I think compounds above 1000 Da (and with so many high mass fragments) are too much for the ILP. I would suggest that we compute high mass compounds with...

Hi, the github repository is just a mirror of our lab-internal repository. With each release we merge all changes from our internal repository to github. That's why it seems like...

Hi, internally, [M+H]+ and [M]+ are treated equally (i.e., SIRIUS treats all [M]+ ions internally as [M+H]+ ions). The reason is that from the perspective of the mass spectrometry instrument,...

Is it so obvious? Peaks with 1 Da difference appear quite often and might just belong to a different formula. If an isotope peak is labeled as "H"-difference, it is...