svachalek

8 ms·

181.
by svachalek 6mo ago
True. The prompt reads: Run the following Python: ```
182.
by svachalek 6mo ago
I'm sure this makes the billionaire class happy but there are some legit economics involved. We all want frontier class models running on our home PC but that takes 100x or 1000x the computer we've been running. The market isn&#x2
183.
by svachalek 6mo ago
Is it really privacy washing? This author assumed people would think AI was local but I don't know anyone like that. Everyone I know who's not deep in tech only uses big hosted models and isn't aware that anything else exists
184.
by svachalek 6mo ago
The latest small models are now reliable enough at simple tools like web search I think. It's just afaik none of the user friendly harnesses like ollama or LMStudio have a real one-click setup flow for this. You'll need to downloa
185.
by svachalek 6mo ago
I've gotten it to work with other models. They've got to be perfectly aligned usually, in terms of provider, quantization etc. Might be a bit before you can get a matched set.
186.
by svachalek 6mo ago
By the laws of physics, it's pretty clear we don't. The same chemical and electromagnetic interactions that drive everything around us are active in our brains, causing us to do things and feel things. We feel like we're in c
187.
by svachalek 6mo ago
This is for Claude Code. It's not hard to install/run Claude Code but it might be a little intimidating if you're not used to command lines. I think in the regular Claude app, you can set up something like this as a "pro
188.
by svachalek 6mo ago
Depends what you call easy but LMStudio is a drag and drop installation and can run thousands of different models.
189.
by svachalek 6mo ago
Ha. It's a thought, at least. I've been programming for about 45 years and have seen the industry go through idioms like fashion. I remember the source to one popular video game, I think it was Quake, was organized into 26 files.
190.
by svachalek 6mo ago
It's statistics based on human coding patterns and human thought patterns (encoded in natural language) though. It's not I AM MODEL 527R1, I AM TOTALLY LOGICAL AND DO NOT UNDERSTAND YOUR HUMAN EMOTIONAL STATES.
191.
by svachalek 6mo ago
Yeah it seems based on 2023 research which is ancient, back when we didn't have coding agents at all, and on some 1980s sci fi concepts of "how machines think" (beedeeboop) rather than the all too human coding agents we have.
192.
by svachalek 6mo ago
I think they're conflating/confusing a bunch of different things here. E-ink tags let stores run sales more often, offer "happy hour" time of day discounts, etc. It's not so much individualized (other than probably
193.
by svachalek 6mo ago
This is assuming children should be on social media at all, which I for one would debate.
194.
by svachalek 6mo ago
There's a reason, you just won't like it.
195.
by svachalek 6mo ago
It's still pretty good. TADS is a more modern alternative, and the one I would go with, but basically these haven't been commercially viable products since the 80s so there's not a lot going into it. Arguably, the right answe
196.
by svachalek 6mo ago
Worth building with VC capital maybe. A small team putting together an app that pulled in $20M per year should be pretty pleased with that.
197.
by svachalek 6mo ago
Or that the government offered Meta $50 for a list of agitators and they said why not. Given Meta's track record it's totally on brand.
198.
by svachalek 6mo ago
Yeah recently it had an issue getting OIDC working and decided to implement its own, throwing in a few thousand extra lines. I'm sure there were no security holes created in there at all. /s
199.
by svachalek 6mo ago
Weirdly enough I feel like this is how the Vision Pro Youtube app works. I'm not sure why.
200.
by svachalek 6mo ago
My last ER trip ended up over $1,000,000. Fortunately after out-of-pocket maximum it was all covered by insurance but this kind of debt would be life ending to the uninsured.
201.
by svachalek 6mo ago
My school district (Poway, one of the better ones in California) is using AI extensively to grade schoolwork.
202.
by svachalek 7mo ago
This is a fair point. When people talk about LLM writing they're always picking on its visible tics and clear flaws. It's a lot more uncomfortable to talk about the things it does better than most of us. There is a lot of precisio
203.
by svachalek 7mo ago
I think this is on point, I've really started to think about LLMs in terms of attention budget more than tokens. There's only so many things they can do at once, which ones are most important to you?
204.
by svachalek 7mo ago
Tribal chiefs are not authoritarians? Because basically every Stone Age village has one.
205.
by svachalek 7mo ago
The Qwen Plus models should be compared to Gemini, not Gemma.
206.
by svachalek 7mo ago
They write code differently but that doesn't mean that's the kind of code they prefer to read. Don't ascribe too much intention to a stochastic process. Their coding style is above all else a symptom of their very limited con
207.
by svachalek 7mo ago
1. There are small local models that have the capabilities of frontier models a year ago 2. They aren't harvesting your data for government files or training purposes 3. They won't be altered overnight to push advertising or a pol
208.
by svachalek 7mo ago
Every contract I have to agree to these days has a "valid until unilaterally invalidated" clause. It feels like we're all just going through the motions.
209.
by svachalek 7mo ago
Exactly. There's a clear alternative in my mind, one I'm sure is objectionable in its own way but I think is the least evil of the three: require providers to label their content and make them liable for it. This allows parents to
210.
by svachalek 7mo ago
I really feel like we have not encountered the same stupid people. Most stupid people I know respond to every question with some form of will-not-attempt. What's 74 times 2? Use a calculator! Should I drive or walk to the car wash? Not

More ›