chiisana

If memory serves, 175B parameters is for the GPT3 model, not even the 3.5 model that caught the world by surprise; and they have not disclosed parameter space for GPT4, 4o, and o1 yet. If memory also serves, 3 was primarily English, and had only a relatively small set of words (I think 50K or something to that effect) it was considering as next token candidates. Now that it is able to work in multiple languages and multi modal, the parameter space must be much much larger.

The amount of things it can do now is incredible, but our perceived incremental improvements on LLM will probably slow down (due to the pace fitting to the predicted lines in log space)… until the next big thing (neural nets > expert systems > deep learning > LLM > ???). Such an exciting time we’re in!

Edit: found it. Roughly 50K tokens for input output embedding, in GPT3. 3Blue1Brown has a really good explanation here for anyone interested: https://youtu.be/wjZofJX0v4M

The models are not wrong. The models are nothing but a statistical model that’s really good at predicting the next word that is likely to follow base on prior information given. It doesn’t have understanding of the context of the words, just that statistically they’re likely to follow. As such, all LLM outputs are correct to their design.

The users’ assumption/expectation of the output being factual is what is wrong. Hallucination is a fancy word in attempt make the users not feel as upset when the output passage doesn’t match their assumption/expectation.

Not a space I’m familiar with, but a friend of mine was all over Habitica and mentioned you could self host it. Is this something that might fit what you’re looking for?

Multiple compose file, each in their own directory for a stack of services. Running Lemmy? It goes to ~/compose_home/lemmy, with binds for image resized and database as folders inside that directory. Running website? It goes to ~/compose_home/example.com, with its static files, api, and database binds all as folders inside that. Etc etc. Use gateway reverse proxy (I prefer Traefik but each to their own) and have each stack join the network to expose only what you’d need.

Back up is easy, snapshot the volume bind (stop any service individually as needed); moving server for specific stack is easy, just move the directory over to a new system (update gateway info if required); upgrading is easy, just upgrade individual stack and off to the races.

Pulling all stacks into a single compose for the system as a whole is nuts. You lose all the flexibility and gain… nothing?