These are good examples of why a non-smartphone is valuable and a smartphone is not necessary. Also the linked article states that they will be allowed to use their phone when given circumstances require, which I think covers the cases you outlined
> Also the linked article states that they will be allowed to use their phone when given circumstances require, which I think covers the cases you outlined
Yes. I was discussing things from the thread's title and the arguments in the thread that phones aren't useful at school. The actual action that Finland is taking (Children aren't allowed to use cellphones at school without permission, effectively) is reasonable. But it's also not what the title of this thread says.
While you haven't used linear algebra, someone working in MLE would probably find it useful to understand. Now I'm not saying you need to take a linear algebra class to understand matrix/vector operations, but it would be useful as it comes up.
At my work, we self-host some models and have found that for anything remotely similar to RAG or use cases that are very specific, the quantized models have proven to be more than sufficient. This helps us keep them running on smaller infra and generally lower costs
Personally I've noticed major changes in performance between different quantisations of the same model.
Mistral's large 123B model works well (but slowly) at 4-bit quantisation, but if I knock it down to 2.5-bit quantisation for speed, performance drops to the point where I'm better off with a 70B 4-bit model.
This makes me reluctant to evaluate new models in heavily quantised forms, as you're measuring the quantisation more than the actual model.
That's a fair point - the trick with dynamic quants is we selectively choose not to quantize many components - ie attention is left at 4 or 6bit, just the MoE parts are 1.5bit (-1, 0, 1)
There are distilled versions like Qwen 1.5, 3, 14, 32, Llama 8, 70, but those are distilled - if you want to run the original R1, then the quants are currently the only way.
But I agree quants do affect perf - hence the trick for MoEs is to not quantize specific areas!
I have had a really good experience using Apple Maps for public transit. Earlier this year I went to NYC for the first time as an adult and it was super easy to use for finding which train to get on. Had a similar experience in Europe this fall as well.
This is awesome and I am happy to read that she was able to remember the device and asks if things have been added to it. My parents have just retired and I wonder if something like this would be advantageous to introduce prior to signs of memory loss. My grandmother had Alzheimers and while it is different than the amnesia that OP references, her memories were lost in reverse chronological order (can't remember where her keys are, but could remember where her last job was, later could not remember that last job, but could remember her first job, etc). So introducing this prior to those recent memory lapses could help solidify that device in my parents head so that they could benefit from it even if they do start to exhibit that behavior.
