clbrmbr

13 ms·

211.
by clbrmbr 1y ago
I have actually had some really great flow evenings lately, the likes of which I have not enjoyed in many years, precisely because of AI-assisted coding. The trick is to break the task down in to components that are of moderate complexity s
212.
by clbrmbr 1y ago
Wow epic job on the presentation. Love the interactive content and streaming. Presumably you generated a special API key and put a limit on the spend haha.
213.
by clbrmbr 1y ago
I received a METS edition of the Prose Merlin as a kid. Lots of fun reading that book. Indeed the arbitrary and self-inconsistent spelling was initially a turn off, but now I see it as more of a fuzzy orthography. I like alternating between
214.
by clbrmbr 1y ago
You should actually try it. The really big models (4 and 4.5, sadly not 4o) have truly breathtaking ability to dig up hidden gems that have a really low profile on the internet. The recommendations also seem to cut through all the SEO and r
215.
by clbrmbr 1y ago
But it’s so much weaker than 4.5 in broader tasks… maybe more optimized against benchmarks but it’s just no replacement for a huge model.
216.
by clbrmbr 1y ago
Maybe a reference to the OpenAI models loving to output em-dashes?
217.
by clbrmbr 1y ago
If so, the loss of fidelity versus 4.5 is really noticeable and a loss for numerous applications. (Finding a vegan restaurant in a random city neighborhood, for example.)
218.
by clbrmbr 1y ago
The deprecation of GPT-4.5 makes me sad. It's an amazing model with great world-knowledge and subtly. It KNOWS THINGS that, on a quick experiment, 4.1 just does not. 4.5 could tell me what I would see from a random street corner in New
219.
by clbrmbr 1y ago
I had a lot of fun building rockets in a similar way as a kid. We also tried PVC but discovered that it fails in a dangerous way, after which we used only copper. I had fun programming a PIC16 in assembler to read from the accelerometer and
220.
by clbrmbr 2y ago
I got one German “w” when using the following prompt, but most of the “w” were still pronounced as liquids rather than labial fricatives. > Speak with an exaggerated German accent, pronouncing all “w” as “v”
221.
by clbrmbr 2y ago
Got this issue in my ‘24 Camry with the volume. If I had the music cranked up previously and then I turned on the car there’s a lag before you can turn off or turn down the volume. I love the CarPlay feature, but I also miss good old 90s er
222.
by clbrmbr 2y ago
This. There’s obvious differences in salvinorin and psilocybin effects and both are well represented in the data. But keep in mind the current paper only looks at existing categories of visual effect. The second study might include a finer
223.
by clbrmbr 2y ago
Volodymyr, can you comment on your organization's approach to safety? It would appear that your model will be a step-change in the capabilities/FLOP, which could change the scale of applications. Also being a novel architecture, h
224.
by clbrmbr 2y ago
That sounds like it would be far less computationally efficient. How can there be an efficiency gain over autoregression?
225.
by clbrmbr 2y ago
Wait, do we know there to be PMBs passing through us like neutrinos? Is this a new kind of cosmic ray?
226.
by clbrmbr 2y ago
I just today reviewed a PR with a default insecure option. But here we’re working on local networks where there’s no way to get a certificate because there’s not a domain name that points to the local IP address. At least with HTTPS over th
227.
by clbrmbr 2y ago
> the real winners in all this: the programmers who saw the chaos coming and refused to play along. The ones who […] went deep into systems programming, AI interpretability, or high-performance computing. These are the people who actuall
228.
by clbrmbr 2y ago
I have found it very difficult to use aider as it makes many changes I don’t always agree with. Personally, I either use Cursor to make edits, or ask oh one to write a whole script or single-file module.
229.
by clbrmbr 2y ago
Wireshark is to tcpdump as stratoshark is to strace. Did I get the analogy right?
230.
by clbrmbr 2y ago
As a user of PyPI, what’s a best practice to protect against compromised libraries? I fear that freezing the version number is inadequate because attackers (who don’t forget, control the dependency) could change the git tag and redeploy a c
231.
by clbrmbr 2y ago
It could be interesting to find the shortest solution. But the site is not set up for verifying solutions in various languages.
232.
by clbrmbr 2y ago
I once designed and configured a satellite-microwave hybrid network for a large US customer with field offices around Borneo. I’ll never forget making the leased line handoffs back in Jakarta. I had exactly zero experience doing this so I g
233.
by clbrmbr 2y ago
I see “just” as an invitation to win the argument by agreeing with the person. “Ooo yeah good idea, I think we could. Can you help me think through this?” — And then they ideally proceed to come around to essentially what you wanted to do a
234.
by clbrmbr 2y ago
How they tested the cable during laying: > At least once in every five minutes the cable was used as a condenser. A charge from a battery of about one hundred cells was given and in turn discharged through a properly shunted Sullivan gal
235.
by clbrmbr 2y ago
“Failed to generate result”
236.
by clbrmbr 2y ago
That’s a problem for Europe but not so much the USA. We should not have a Chinese company controlling our information infrastructure to this degree. It’s divestiture or ban, not just a ban.
237.
by clbrmbr 2y ago
I used to be super productive at the raw keyboard. Then RSI got to me. But with CoPilot, I’m back to my normal productivity. For me it’s a life-saver as it allows fast typing with minimal hand strain.
238.
by clbrmbr 2y ago
Definitely something investors should be aware of so that they can further optimize their decision making. However, scores of physical attractiveness are going to be strongly influenced by the character & charisma of the speaker. And I
239.
by clbrmbr 2y ago
Don’t key prefixes take care of the need for an arbitrary number of tables?
240.
by clbrmbr 2y ago
IIUC, a Gemini-based system could translate the natural language questions into Lean, but in the blog post they don’t really commit to whether this was done just to generate training data or was used in the competition.

More ›