jatins

4 ms·

31.
by jatins 2mo ago
As I said: If you couldn't be bothered to write it, don't expect people to read it. I find it disrespectful to your audience That said my own views on what is/isnt slop is evolving i understand your pov
32.
by jatins 2mo ago
slop README. If you couldn't be bothered to write it, don't expect people to read it.
33.
by jatins 2mo ago
it’s probably just some informal Slack messages between colleagues that is being described as “reports from users”. There is no Report here
34.
by jatins 2mo ago
I found there Deep Research quite good in terms of browsing the web, better than Gemini's. Though it blew through ~40$ in one task so it was very expensive, probably running at insane margins.
35.
by jatins 2mo ago
Core of my comment is 1/ being wrong on timing is being wrong 2/ There is no clear definition of bubble burst. Public markets were down 10% in July and many stocks like SNDK fell 50% -- is this a bubble burst?
36.
by jatins 2mo ago
Ed wrote a post back in 2024 where he claimed OpenAI would fail in 2 years if they didn’t do the, seemingly impossible at that time, things like raising more money than any company before etc. And OpenAI did all of that and is still alive t
37.
by jatins 2mo ago
Is there anyone at X (including xai cofounders) who has been there more than 2 years?
38.
by jatins 2mo ago
> If anthropic wins, Google copies their approach, buys anthropic for cheap or both Google buys Anthropic for cheap? How?
39.
by jatins 2mo ago
I don’t think so. If Kimi and Deepseek can launch models better than Gemini with much much lesser resources then it is increasingly looking like an organization issue at Google
40.
by jatins 2mo ago
I don’t disagree with this if you code for a hobby. Buy if you code for a job, good luck justifying this to management. “Yeah Claude already gave me the solution, I’ll take the rest of the week to type it out”
41.
by jatins 2mo ago
> Last thought is that the domain of worlds/games exposes a weakness in LLMs: they can't easily audit their work because they aren't able to efficiently and natively perceive videos or play games within them. Again comes b
42.
by jatins 3mo ago
It is. I just wish people would throw a plain text page with a paragraph instead of this because you bet I am just scrolling through it without reading as soon as Claudism appears
43.
by jatins 3mo ago
> On the other hand, What is the other hand here? They could do slow/costly illegal think #1 instead of the fast/cheaper illegal thing #2 that they currently do?
44.
by jatins 3mo ago
> they've just been able to use the synthetic data off of them to get the leap in quality from Opus. > not a distilled version of Mythos or Fable isnt distilled == trained on synthetic data and reasoning traces?
45.
by jatins 3mo ago
> Honestly no idea what I’d do if I were in government. It's not ordinary citizens' job to figure out solutions to these problems. Government has teams of intelligence agencies to answer these questions. Encrypted chat is funda
46.
by jatins 3mo ago
Indian government also blocked Telegram a while back to prevent exam leaks and stopped Whatsapp's rollout of usernames (which has some legitimate concerns) but more or less it wants complete China-like control of it's citizens&#x2
47.
by jatins 3mo ago
Better than Fable 5 on all but 3 evals. Has Anthropic ever mentioned how do Opus and Fable differ? It used to be Haiku < Sonnet < Opus in terms of params. Where does Fable fit in this?
48.
by jatins 3mo ago
What an unnecessarily negative and cynical reaction to what is neutral at worst and positive at best
49.
by jatins 3mo ago
> Progress takes time and there can be mistakes is basically the story of our civilization tho Yeah, no one is doubting it. People are just asking to not be lied to about how much progress has been made
50.
by jatins 3mo ago
As open source as their timeline algorithm?
51.
by jatins 3mo ago
I do wish I had more friends
52.
by jatins 3mo ago
All the signs of slop, short one line paragraphs. Tested a passage on Pangram and 100% AI generated btw https://www.pangram.com/history/f6b20d8e-4eb4-4bc2-b325-ad39...
53.
by jatins 3mo ago
Why are you telling it that? No one is arguing LLMs are perfect but at this point dismissing them as stochastic parrots seems a bit silly
54.
by jatins 3mo ago
> insane, unproven, unpredictable autocomplete math equation It won IMO, solved Erdos problems. At what point will you stop saying that?
55.
by jatins 3mo ago
rewrites feel like an area where LLMs are better suited than humans imo It’s mostly grunt work and LLMs are well suited for translation tasks (iirc transformers arch was originally invented for translation)
56.
by jatins 3mo ago
Extremely thorough and well written. I was hoping it’d end in a “so how much did this cost?” so that others team looking at similar migrations have an estimate on what they can expect
57.
by jatins 3mo ago
oh wow, a blast from the past. I remember a site called staticdrop or statichost like a decade ago, before vercel and everything that did the same thing Good to see great ideas making a comeback
58.
by jatins 3mo ago
> friendly employees Where are the employees? There are so few employees (other than cashier) on the ground in a football field sized warehouse. Good luck finding someone if you have a question
59.
by jatins 3mo ago
If I see this code in a new company I just joined I am quitting on day 1
60.
by jatins 4mo ago
> Webernetes is intended to be used to make interactive Kubernetes content;

More ›