-
Running the app takes some time to load the model into memory , and since we're using quantized version, llm.to('cuda') is not made use of.
The answers from the RAG are pretty decent given that t…
-
Hi :
I am wondering if I want to train baseline++ with few shot training method, I just change the dataloader class ,right? Because now in train.py, the baseline and baseline++ are not trained …
-
Hello, I hope this message finds you well.
I noticed a function called Weighted_GAP in both PFENet.py and PFENet2Plus.py. I'm curious about the specific purpose of this function and how the paramet…
-
Hi, thangs for your work. I have a 16gb gpu and in order to run the training process, i change the support_shot from 30 to 20 and the batch size from 8 to 2.But i don't know whether it is better to re…
-
Good day
I'm trying to train the MRCNN on my Anaconda environment. I'm using Python 3.6 and on a Windows 8.1. I believe I've managed to install TF (`version 1.15.0`) and KERAS (`version 2.2.4-tf`) …
-
We train the network to make it classify a sample with unknown class, however, the network need label itself as part of input, In real applications, only those samples with unknown labels need to be c…
-
I have a little question about the testing.
In the testing part, the query image always comes from the 5 support classes. But, in practice, there is a totally unknown query coming, how we choose …
-
As the title suggests, we need to find out if "Learning Plans" link from user profile can be switched off OR not to keep User Profile area bare minimum from Participant point of view.
cc: @Pro-Paul…
-
, in which they continue "hybrid …