I think the real story is a guy who replaced all his customer service with chatbots is now pushing code at the Department of Veterans Affairs, a place with actual life or death stakes if a customer can’t get the answers they need because they’re stuck in chatbot hell.
Our third grader just started viola this school year and I was amazed at how cheap the rental was! I was expecting to get raked over the coals monthly, but it was just a little over $100 for the entire year. The chin rest was one thing we didn’t have to pay for, the schools have the younger kids use a big foam sponge attached with a rubber band.
That absolutely works for me! I play multiple instruments and have found that the early morning is the best time mentally for me to devote uninterrupted time to practice and playing. I’m also fortunate enough to have a basement with another floor between my cacophony and my sleeping family
I think (from personal experience) talking with a good mental health professional would really help with your current state of mind and the pressure you’re feeling.
The model displayed in the drop-down when you access the web library is the default that will be pulled. Compare the size and hash to the more detailed model listing below it and you will see what quantization you have.
Example: the default model weights for Llama 3.3 70b, after hitting the “view all” have this hash and size listed next to it - a6eb4748fd29 • 43GB
Now scroll down through the list and you will find the one that matches that hash and size is “70b-instruct-q4_K_M”. That tells you that the default weights for Llama 3.3 70B from Ollama are 4-bit quantized (q4) while the “K_M” tells you a bit about what techniques were used during quantization to balance size and performance.
If you hit the drop-down menu for the size of the model, then tap “view all”, you will see the size and hash of the model you have selected and can compare it to the full list below it that has the quantization specs in the name.
Wow, I completely forgot about Dale Brown! That was a very specific time in my life when I was into Tom Clancy due to my dad having all his books on a shelf and I had run out of Clancy novels to read.
I love that he actually posts % complete widgets at the top of his website showing how far along he is in the multiple books he authors at once. Closest thing to a Grafana dashboard for authors we’re going to get!
One of my most treasured possessions is a cloth Velcro patch display panel holding all my patches from various military deployments. The official unit patches are good and all, but nothing holds a candle to the creativity of military personnel with access to a patch making shop while deployed.
Yeah, I’m specifically responding to the parent’s comment about the 48GB tier. When you’re looking in that range, it’s usually because you want to pack in as much vram as possible into your rack space, so consumer level cards are off the table. I definitely agree multiple 3090 is the way to go if you aren’t trying to host models for smaller scale enterprise use, which is where 48GB cards shine.
At the 48GB level, L40S are great cards and very cost effective. If you aren’t aiming for constant uptime on several >70B models at once, they’re for sure the way to go!
My kids are just getting into Minecraft on the switch and I am a bit confused by the options for hosting a server. I’d love to hop on and play with them on a self-hosted server, what’s the current best way to do that or is that level of cross-play between PC and Switch on a self-hosted solution even possible?
Edit: I don’t currently play, which is why I’m not familiar with what’s out there
I would highly recommend Pianote if you’re looking for an online lessons resource. They have tons of YouTube videos you can check out to see if their style works for you, then you can subscribe to the actual site when you’re ready for a more structured learning path. It did wonders for me getting back into piano after a few decades off!
Yes, Langchain has tooling specifically for connecting to Ollama that can be chained with other tooling in their library to pull in your documents, chunk them, and store them for RAG. See here for a good example notebook:
No, we don’t get funding through grants. Our work is either direct sponsored work or internal research funded through overhead. We aren’t a federally funded R&D center like Laurence Livermore or MIT Lincoln Labs, we are a university affiliated research center.
We exist in this nice middle ground between academia and pure profit companies where we do a bunch of research and exploratory work without the pressures of trying to get tenure, while also getting to push real products out to our sponsors but without ever worrying about getting laid off because we didn’t meet some growth number for the quarter.
That sounds exactly like my employer, the JHU Applied Physics Lab. We’re a university affiliated research center, so we aren’t tied to the government pay scale and can provide a better career advancement outlook.