Bahushruth

Hey I tried using this class on a toy model + data and I get this error `RuntimeError: Trying to backward through the graph a second time, but the saved...

This works thank you! One more thing that I noticed was that sometimes loss is nan `Epoch 046: | Loss: 0.02856 | Acc: 65.750 Epoch 047: | Loss: nan |...

I will now implement this in my competition notebook and let you know if we make progress.

I will wait for your implementation of the class no worries. Currently, I had 0.2 to 0.3 improvement in AUC and significant improvement in ACC. And I followed a bit...

Hi, Really cool that you guys found the repo interesting. Initially it was just a quick thing my friend and I hacked together. Based on the feedback from folks we...

Has anyone found a way to capture the contents of the screen and not just the desktop background?

> I also found these bindings https://github.com/PotatoSpudowski/fastLLaMa > > Some feature suggestions, mostly about low level capabilities: > > * Accessing the output classifier activations from python, enabling sampling and...

@amitsingh19975 implemented a [hack](https://github.com/PotatoSpudowski/fastLLaMa/compare/main...feature/adapter_quant). Is this correct? If yes we can maybe raise a PR for this library We are a little sceptical about int4 tho

Cosine similarity is a score ranging from -1 to 0 values closer to -1 have a greater cosine similarity. May I ask what your end score was and how well...

Interesting, I will think about ways to do this meanwhile for now as you mentioned the only proper way is to remove it from site packages directly. Let's keep this...