LZ_Khan

7 ms·

61.
by LZ_Khan 1y ago
Learning is much easier now. For example, on college level math texts sometimes you get stuck on a problem and aren't sure what you're missing. You could hire a tutor, go to office hours if you're still a uni student, or fumb
62.
by LZ_Khan 1y ago
I don't see the problem as it's not like you're forced to use their image generation model.
63.
by LZ_Khan 1y ago
Counterexample- Facebook made Threads which has similar # users as Twitter now.
64.
by LZ_Khan 1y ago
I feel like being at the bottom of the value chain is a mis-categorization. If you consider base LLM model as their sole offering I agree with you, but these companies have shown an eagerness to eat their way up the value chain. Agent mode,
65.
by LZ_Khan 1y ago
This is just the exact same culture as Deepmind minus the "everything on Slack" bulletpoint.
66.
by LZ_Khan 1y ago
I just shorted it as well.. 3 days before you. And I got roasted. Invest with caution.
67.
by LZ_Khan 1y ago
I do feel like large scale LLM vulnerabilities will be the real Y2K
68.
by LZ_Khan 1y ago
Is this the new Y2k?
69.
by LZ_Khan 1y ago
readability would probably be the sticking point
70.
by LZ_Khan 1y ago
Honestly I think this is a problem of over-engineering and simply allowing the user to press a button when he wants to start talking and press it when he's done is good enough. Or even a codeword for start and finish. We don't nee
71.
by LZ_Khan 1y ago
First of all, congratulations. That's a huge feat and you seem to have overcome a lot of hurdles to make it. That being said, I find it a bit discouraging that small-team passion projects with even the best product-market fit and minim
72.
by LZ_Khan 1y ago
Well I assume that's what Anthropic and Google are doing with Claude Code and Firebase Studio. The main thing is marketing and userbase, which shouldn't be underestimated.
73.
by LZ_Khan 1y ago
Um.. wasn't this what was mentioned was going to happen in AI 2027? "In a few rigged demos, it even lies in more serious ways, like hiding evidence that it failed on a task, in order to get better ratings."
74.
by LZ_Khan 2y ago
Annnnd this is why Google bought Wiz huh.
75.
by LZ_Khan 2y ago
I feel like covid started the remote work revolution and AI cheating is about to end it. Or at least bring back on-site interviews.
76.
by LZ_Khan 2y ago
How does a company like that go bankrupt? They sell kits for freaking $200. How is that unprofitable?
77.
by LZ_Khan 2y ago
I work in the field. The amount of jargon is indeed large but it's not out of the ordinary. It's simply how things are referred to. If the author explained what everything is the content would span a textbook. That being said I do
78.
by LZ_Khan 2y ago
Camera feed processing latency would be my guess. The system needs to make sense of a continuous video feed so moving slower reduces how much happens in between frames.
79.
by LZ_Khan 2y ago
I think we all share that feeling, even as a software engineer and I see AI writing 90% of my code for me.
80.
by LZ_Khan 2y ago
I just find it insane that we're bootstrapping reinforcement learning and world planning on top of basic next token prediction. I'm amazed that it works, but also amazed that this is the approach being prioritized.
81.
by LZ_Khan 2y ago
Also ai generated
82.
by LZ_Khan 2y ago
Shouldn't there be a comparison in performance on common benchmarks to other models? Like a 7B toponet model vs a 7B Llama model? As a layperson I don't understand why topology is a thing to optimize for.
83.
by LZ_Khan 2y ago
I actually think what DeepSeek did will slow down AI progress. What's the incentive to spend billions developing frontier models if once it's released some shady orgs in unregulated countries can just scrape your model outputs,
84.
Ask HN: Did DeepSeek train on ChatGPT outputs?
4 points by LZ_Khan 2y ago|1 comments
85.
by LZ_Khan 2y ago
You can't be a killer app if your competitor is just as good and free.
86.
by LZ_Khan 2y ago
Hype. We were already plateauing with next token prediction and letting the models think out loud has simply pushed the frontier a little bit in purely test-taking domains.
87.
by LZ_Khan 2y ago
Note: This is not an actual model but rather an announcement of an effort to reproduce the R1 model.
88.
by LZ_Khan 2y ago
I'm not saying the extension is unlawful. The person I replied to seemed to imply an addendum to a law should be less enforceable than the primary focus of the law. I was simply addressing that statement.
89.
by LZ_Khan 2y ago
The winner of the election is often the party that spent more money on political advertising, so I'm sure this is a well known phenomenon.
90.
by LZ_Khan 2y ago
And yet the law is the law. There's no premise that says the manner in which a law is passed determines its enforceability.

More ›