Gemma 27B Is actually quite good, but "narrow." Its super low context and seems to be hyper optimized for short chatbot-arena style questions.

This is the stuff I love to know so thanks for sharing. I will be pulling Command R tomorrow.

I manually specify what models to pull. I’m not running anything too crazy. My largest model is gemma27B. But I’ve worked with dolphin-mistral which was fun.

I read localllama through redlib but I don’t contribute. I am not technical enough to contribute and I don’t understand the math.

I have been looking at YouTube for some videos to try to explain it, but I haven’t found anything that is in the sweet spot between “video for non-technical people” and “video for people with PhD and quantum physics”

I’m running Nvidia on Ubuntu. I’ll give exllama a shot.

I think it’s amazing. I’m running Ollama with a bunch of open-source llms. You’re right. It’s so good. The problem is keeping up to date on what the newest development is.

The pace of progress is so fast and it’s really difficult to know what the cool kids are experimenting with this moment.

Is this because FIDO2 is flawed, the yubikey hardware design is flawed or both?

I don't have an invite to CGpeers. If I did, I would offer it to you.

If you're familiar with private trackers, I have an extra invitation to one, if you want it. You can PM me.