My experience with them is that the are taught to cover as much syntax and libraries as possible, without spending time learning how solve problems and develop their own algorithms. They (in general) expect to follow predefined recipes.
You might need to play around with the default settings. One of the first models I tried running on my Mac was really slow.. Turns out it was preallocating a long context window that wouldn't fit in the GPU memory, so it ran on the CPU.
We had this at Clemson University about 10 years ago. I think the whole design came out of an MIT lab though. There's video of it still up here: https://www.clemsongis.org/clemson-sandbox
We do use WebLLM and a hosted Weaviate database, but there are complaints about speed (both retrieval and time to first token as the context will get big). The Gemma 3n "nesting doll" approach sounds like it could be useful .. but haven't found anyone specifically doing it to add domain specific knowledge.
Which small model is good for fine tuning to various enterprise data sets? Our business units are wanting to run small models in browser and on mobile devices, without dealing with RAG and cloud resources.
If you need to support several languages, you're going to have to have a zoo of models. Small ones just can't handle that many; and they especially aren't good enough for distribution, we only use them for understanding.
Kinda of strange to pick an example that is just wrong. It's supposed to be written from 1840 and says Paris is the seat of Napoleon almost 20 years after he died.