drob518

9 ms·

181.
by drob518 3mo ago
Interesting, thanks. Looking at the model card on Huggingface, it’s combining the Qwythos and Qwable fine tunes from Empero.
182.
by drob518 3mo ago
Evidence?
183.
by drob518 3mo ago
I tried it, too, and it got stuck in some loops where it couldn’t recover. Shame, it was promising for the same reason as Bonsai’s models.
184.
by drob518 3mo ago
Yep, that’s the question. I asked just that when Bonsai’s first models got released. Super interesting if we can push the parameter count over 100B with 1.125 bit quantization and still keep pretty good performance versus 16-bit 100B models
185.
by drob518 3mo ago
Same here. I’m excited to have a model that might be usable on a 16 GB laptop.
186.
by drob518 3mo ago
This is going in a good direction.
187.
by drob518 3mo ago
Got it. Yea, if those are your priorities then it certainly makes sense to write your own.
188.
by drob518 3mo ago
Right. And this is safe now since Java Time classes are immutable. That’s key. Before, you had to worry about someone modifying the object.
189.
by drob518 3mo ago
I do something similar for Java Time local dates. Financial data in particular has lots of redundant date info and benefits from being memoized. Converting to epoch millis also works.
190.
by drob518 3mo ago
I prefer to see C++ as a failed experiment that just keeps going and going rather than garbage. The software industry learned a lot from it, both good and bad. But yea, I haven’t programmed in it since the late 1990s.
191.
by drob518 3mo ago
Yep, and sometimes just a small code change can flip on boxing.
192.
by drob518 3mo ago
Yes, and verbosity (thinking) is a huge factor.
193.
by drob518 3mo ago
Sure, if your fundamental issue is “bloat,” you can always write a less general-purpose system that is smaller. No doubt about that.
194.
by drob518 3mo ago
On the issue of the bubble, I’m right there with you, 100%. I’m not sure that “subsidized” is the right word for Anthropic’s or OpenAI’s pricing, though. I’d say it’s forward-priced. Supposedly, they have claimed that inference by itself is
195.
by drob518 3mo ago
Agreed that the node cesspool is a risk. That said, what batteries does Pi have that you don’t want?
196.
by drob518 3mo ago
Sure, fair enough. Clearly, if they increase costs by too much, people will go to their competitors, but those competitors also make money selling tokens, so the whole industry is incentivized to inflate token consumption up to the point of
197.
by drob518 3mo ago
That’s certainly doable, but then you need to create all the add-ons you would have added to Pi. IMO, Pi stands in that sweet spot between being very minimal while still offering a catalog of pluggable functionality that you can add to it.
198.
by drob518 3mo ago
I use Pi and love it, but the base install is extremely minimal (a handful of tools, no subagents, etc.). That’s on purpose. Sure, you can add more with packages, but that’s not the base install.
199.
by drob518 3mo ago
Exactly. It’s minimal to start but the sky is the limit at the other end. I love the plugability of Pi.
200.
by drob518 3mo ago
Of course it is. How could it be anything different? Clearly, that’s how these companies make money.
201.
by drob518 3mo ago
Well, Anthropic and OpenAI make money selling tokens, so…
202.
by drob518 3mo ago
Right, so it’s a trade off between contexts. There are two reasons to use subagents, parallelism and tailoring of context. For the second, there is the “personality” of the subagents as well as how much context is injected from the main age
203.
by drob518 3mo ago
I think the commenter (who is not me) is saying to use Fable as the main agent but then use lesser models for your subagents so that you get the advantages of Fable to plan but then the subagents don’t cost as much, and may be more focused
204.
by drob518 3mo ago
Cool, you do you.
205.
by drob518 3mo ago
Clearly, you’re not going to do it for just Fibonacci. And yes, you can always rewrite the agent code, quit, and restart to modify the agent. This is just like using an LLM to modify Pi and then restarting, which is done today. But being ab
206.
by drob518 3mo ago
So, writing an agent in Lisp is interesting but not particularly novel. If there’s a big idea here it’s giving Lisp’s eval function to the AI as a tool. Yes, AIs write Python and Bash all day long already, but those scripts are run out of p
207.
by drob518 3mo ago
Agreed. You can tell when someone “gets it” or not by whether they see sexprs as a problem to be overcome or a benefit to be embraced. Count me firmly in the benefit camp.
208.
by drob518 3mo ago
Seems to me that OpenAI has a culture of questionable ethics that includes this incident but goes way beyond it. This seems very “on brand” for them.
209.
by drob518 3mo ago
It does. In Prolog, data is code is data as well.
210.
by drob518 3mo ago
Clojure programmer here. Refactoring is the only time I miss static typing.

More ›