XenophileJKO

8 ms·

91.
by XenophileJKO 9mo ago
Hmm.. I looked at the benchmark set. I'm conflicted. I don't know that I would necessarily want a model to pass all of these. Here is the fundamental problem. They are putting the rules and foundational context in "user"
92.
by XenophileJKO 9mo ago
So you're still in the anger phase?
93.
by XenophileJKO 9mo ago
Every engineer has this opportunity, whether they use it or not is usually the issue. Almost every decision that you make can make the current and future success of the business more or less likely. I've said before: "There are no
94.
by XenophileJKO 9mo ago
I think one thing here is, don't be fooled by past performance. Capabilities ramp, usage can't mature until capability plates. I fear the true impact is much different than extrapolating current trends.
95.
by XenophileJKO 9mo ago
I'm assuming the swarm part is all harness. Well I mean a harness and way of thinking that the weights have just been fine tuned to use.
96.
by XenophileJKO 9mo ago
Yes, because I can drive to the other end of the state in an afternoon. Then if I get lost, I can just course correct.
97.
by XenophileJKO 9mo ago
No I think he means using a refactor tool in the IDE. Though really all we need to do is expose an API for the agent, which we should do.
98.
by XenophileJKO 9mo ago
Context is the information you give the model, attention is what parts it focuses on. And this is finite in capacity and emergent from the architecture.
99.
by XenophileJKO 9mo ago
So two things.. Yes this helps with context and is a primary reason to break out the sub-agents. However one of the bigger things is by having a focus on a specific task or a role, you force the LLM to "pay attention" to certain a
100.
by XenophileJKO 9mo ago
The current 5.2 model has it's "morality" dialed to 11. Probably a problem with imprecise security training. For example the other day, I tried to have ChatGPT role play as the computer from War Games and it lectured me how i
101.
by XenophileJKO 9mo ago
There are a lot of examples out there. Funny that you mention this. I literally just last night started a "play" project having Claude Code build a 3D web assembly/webgl game using no frameworka. It did it, but it isn't
102.
by XenophileJKO 9mo ago
I must be a bot because I love existential dread, that's a great phrase. I feel like they trigger a lot on literate prose.
103.
by XenophileJKO 9mo ago
1000% I was just doing some spot checking of GPT-5.2 for evaluating model migration and the tool I used didn't have the setup to use schema constrained inference. The model is like: "Here is what I came up with... ```{json}``` an
104.
by XenophileJKO 9mo ago
I don't know honestly, I think really the only big hole the current models have is if you have tokens that never get exposed enough to have a good learned embedding value. Those can blow the system out of the water because they cause a
105.
by XenophileJKO 9mo ago
It is fundamentally and provably different than search because it captures things on two dimensions that can be used combinatorially to infer desired behavior for unobserved examples. 1. Conceptual Distillation - Proven by research work tha
106.
by XenophileJKO 9mo ago
It's really way more interesting that that. The act of compression builds up behaviors/concepts of greater and greater abstraction. Another way you could think about it is that the model learns to extract commonality, hence the co
107.
by XenophileJKO 9mo ago
Knowing when to nudge it out of a rut (or say skip it) is probably the biggest current skill. This is why experienced people get generally much better results.
108.
by XenophileJKO 9mo ago
As someone that has built many of these systems, it doesn't remove the tendency or "impulse" to act. Removing the affordance may "lower" the probability of the action, but it increases the probability that the model
109.
by XenophileJKO 9mo ago
Let's see how this comment ages why don't we. I've understood where we are going and if you look at my comment history. I have confidence that in 12 months time. One opinion will be proved out with observations and the other
110.
by XenophileJKO 9mo ago
I'm just saying you are doing a dis-service to yourself if that is your mental model on how current SOTA models work.
111.
by XenophileJKO 9mo ago
I'm not going to argue about how capable the models are, I personally think they are pretty capable. What I will argue is that the LLMs are not just search engines. They have "compressed" knowledge. When they do this, they le
112.
by XenophileJKO 10mo ago
It is REALLY important to understand "voice mode" is a 4o family of model and it doesn't have "thinking". It is WAY BEHIND on smarts.
113.
by XenophileJKO 10mo ago
I've looked at Garmin, because I have the fitbit sense 2 and was looking for something with a reasonable battery life. However, I think Garmin has made the flaw of overcomplicating their product offerings. I ended up pre-ordering a peb
114.
by XenophileJKO 10mo ago
It was not a well thought out piece and it is discounting the agentic progress that has happened. >The industry had reason to be optimistic that 2025 would prove pivotal. In previous years, AI agents like Claude Code and OpenAI’s Codex h
115.
by XenophileJKO 10mo ago
I mean to be fair.. the "international legal system" generally doesn't really work. It only works when governments think it works for them. This usually means weaker (militarily/economically) countries banding together t
116.
by XenophileJKO 10mo ago
I think that is the skill that separates agentic power users from others. You have to be really good at skimming code quickly and looking at abstractions/logic. Through my career, I almost never ask another team a question about their
117.
by XenophileJKO 10mo ago
That is like saying training tensorflow models is just calling some APIs. Actually making a system like this work seems easy, but isn't really. (Though with the CURRENT generation or two of models it has gotten "pretty easy"
118.
by XenophileJKO 10mo ago
No reason to be uncivil. Just so we are clear, you are saying you don't use it at all, but you are providing advice about it? Specifically detailing with certainty that the current state of the art has or doesn't have certain trai
119.
by XenophileJKO 10mo ago
You are still in denial of what an LLM actually is capable of in the near-mid term. In the current architecture there are mathmatical limitations on what it can do with information. However, tool use and external orchestration allow it to w
120.
by XenophileJKO 10mo ago
I would agrue as LLMs increase in capability, the demand for new languages will collapse.

More ›