ilaksh

19 ms·

361.
by ilaksh 11mo ago
This is ageism and it's rampant and very problematic. He is ageist against himself almost as much as society is. You should not need to do anything special to be a 40 or 50 year old (or whatever) and fit in with a group. This is one of
362.
by ilaksh 11mo ago
Yes, that is what they were suggesting in the interview, which I think is not quite accurate, so I replied with the comment above.
363.
by ilaksh 11mo ago
It means it is trained/tuned for function (tool) calling (e.g. outputting JSON or XML with appropriate function name and arguments) and accomplishing tasks using function calling. In this case it's also trained/tuned for comp
364.
by ilaksh 11mo ago
The question of how emotions function and how they might be related to value functions is absolutely central to that discussion and very relevant to his field. Doing fundamental AI research definitely involves adjacent fields like neurobiol
365.
by ilaksh 11mo ago
They are not optimized to waste tokens. That is absolutely ridiculous. All of the LLM providers have been struggling from day one to meet demand. They are not trying to provide outputs that create more demand. In fact, for example, Opus 4.5
366.
by ilaksh 11mo ago
I'm so easily influenced. I came very close to immediately ordering Mexican food on DoorDash.
367.
by ilaksh 11mo ago
Really interesting how many layers of competition were going on. Not only were different computer manufacturers competing with each other, but different generations of computers from that manufacturer were competing, like Atari ST versus At
368.
by ilaksh 11mo ago
Actually, I think this idea is a really practical and interesting way to apply some types of text-to-image and image-to-3d models to user generated games, as an alternative to the heavy frame-by-frame uber world model generation where every
369.
by ilaksh 11mo ago
I skimmed over this, but did I see a reference sandbox implementation? And then basically the chat UI interacts with that with postMessage (and receiving) and forwards tool calls to the MCP server. Does it also forward tool calls the MCP se
370.
by ilaksh 11mo ago
4-5 times faster with minimal change in quality seems like a clear upgrade in efficiency.
371.
by ilaksh 11mo ago
okay since Gemini 3 is AI mode now, I switched from the free perplexity back to google as being my search default.
372.
by ilaksh 11mo ago
The only one it doesn't win is SWE bench which it is significantly behind Claude Sonnet. You just can't take down Sonnet.
373.
by ilaksh 11mo ago
Telnyx seems to be down for me. Actually I lied, I think it is working. at least call connected.
374.
by ilaksh 11mo ago
When this kind of thing happens it makes me feel better about my own programming problems. I wonder if it has anything to do with the replicate.com purchase? Probably not.
375.
by ilaksh 11mo ago
Okay so if sending raw binary is a normal use case, does that mean HTTP/3 is a misnomer? Or are they confusing me with the way they present Web transport ad being equivalent to HTTP/3 and WebTransport just sending binary data? Als
376.
by ilaksh 11mo ago
Don't pretend that 98% of startups don't actually dream of an exit or payday.
377.
by ilaksh 11mo ago
From an AI integration perspective, I am hopeful that Cloudflare may be able to improve "performance on the cheap" for replicate's models a little bit. Replicate has had multiple ways to deploy for auto scaling and you can ju
378.
by ilaksh 11mo ago
I think the solution is public efficient protocols, p2p, content centric networking, etc. We have enough technology to be able too avoid having to run searches through centralized services. Maybe p2p or federalized AI agents, or just decent
379.
by ilaksh 1y ago
That's true but I also think despite being wrong about the capabilities of LLMs, LeCun has been right in that variations of LLMs are not an appropriate target for long term research that aims to significantly advance AI. Especially at
380.
by ilaksh 1y ago
The prompt just said to iterate until they were satisfied. Adding something like "don't be afraid to change your approach or make significant revisions" would probably give different results.
381.
by ilaksh 1y ago
ok.. when I tried to use pay-as-you-go it was unusable for me because there were a ton of 429s and 503s. one test it was just constant for a few seconds when I tried it, 429 or 503. I am using it for a voice application though so retrying c
382.
by ilaksh 1y ago
What kind of plan do you have with Cerebras? It seems like something like that would need one of the $1500/month plans at least if there were more than a handful of customers.
383.
by ilaksh 1y ago
I have a comprehensive agent system. Currently working on SIP calls with different models including realtime. https://github.com/runvnc/mindroot
384.
by ilaksh 1y ago
Is it Trek or Super Star Trek? https://github.com/philspil66/Super-Star-Trek
385.
by ilaksh 1y ago
Why isn't it more popular if it's so good? Because popularity and merit are not the same thing. People confuse them all the time though. Check out the most popular music today. Like the top ten songs currently. Do you think those
386.
by ilaksh 1y ago
You can build a first version of that in a week or less. I think it could be great. The challenge for me has always been getting other people to use things I've built. But why not try it?
387.
by ilaksh 1y ago
You're right to point out the speed, but it is just not true that GUIs can't be fast. You can have screens that load quickly and shortcut keys etc. You just have to make fast navigation and data entry a high priority. The assumpti
388.
by ilaksh 1y ago
Very interesting and ambitious project and nice design. I hope the author will be able to comment here. I'm interested to hear about the plans or capabilities in R' or Radiance for things like concurrent programming, asynchronous&
389.
by ilaksh 1y ago
Whoops didn't notice the bad autocorrect.
390.
by ilaksh 1y ago
How hard would it be to add creatine collaboration to Marimo?

More ›