riku_iki

18 ms·

361.
by riku_iki 2y ago
No, but even if it is buggy, it is usual LLM cycle: - you: there is no such function - LLM: you are absolutely right, here is fixed code
362.
by riku_iki 2y ago
https://chatgpt.com/share/6807278c-a0d0-8006-80f3-f62ae9f8ff...
363.
by riku_iki 2y ago
I use LLM for coding every day. LLMs can totally make something super dumb: oh this test is broken, gatcha, lets replace it with something which is not broken (but very different). > They can also silently reduce the tested scenarios, bu
364.
by riku_iki 2y ago
There are plenty of tech bros in corps who push some hyped stuff, add to resume and move to next gig. Not sure why do you think airbnb is different.
365.
by riku_iki 2y ago
> And I would definitely expect that such a conversion would have resulted in hideous, non-idiomatic code in the target language. and we don't know what is the quality of end code, it is possible that tech debt created by migration
366.
by riku_iki 2y ago
> Less difference — mandelbrot, k-nucleotide, reverse-complement, regex-redux — when the task requires memory to be used. yes, I referred to benchmarks with large memory consumption, where Java still uses from 2 to 10(as in binary tree t
367.
by riku_iki 2y ago
With Java, the issue is that each allocated object carries significant memory footprint, as result total memory consumption is much higher compared to C++: https://benchmarksgame-team.pages.debian.net/benchmarksgame/...
368.
by riku_iki 2y ago
In both scenarios publishers look good: 1. CPC: google has strong incentive to generate clicks, because that's where they get revenue: advertisers are charged per click. 2. CPM: publishers get their guaranteed CPM if that's their
369.
by riku_iki 2y ago
> They just need to squeeze more from publishers and at the same time increase click costs. but publishers receive stable share of click cost (67%?), so they should be happy with this arrangement.
370.
by riku_iki 2y ago
> Parts of this can definately be broken up without destroying Google. its about Display Ads business, which is 10% of Google revenue as per article. So, everything there can be broken up without destroying Google.
371.
by riku_iki 2y ago
did they even ask forgiveness?
372.
by riku_iki 2y ago
> everything else is a race to the bottom as tokens and intelligence become commoditized. its not necessary commoditized, things can always become cheaper than before. Its actually opposite: there are handful of big tech providers, whic
373.
by riku_iki 2y ago
realistically speaking, it is likely not possible to deliver envelop from China at 75c cost, its likely someone else is paying for your 75c envelops (like taxpayers through subsidies).
374.
by riku_iki 2y ago
Its trivial to check if any brands mentioned in the response before returning it to user, and then ask LLM to adjust response to mention brand who paid for placement instead.
375.
by riku_iki 2y ago
I think sooner or later LLM providers will force to introduce Ads, and those folks are Ok with ads, since they used google search.
376.
by riku_iki 2y ago
Its just my personal non-tech network is switching to chatgpt on many accounts. Your network can be different of course.
377.
by riku_iki 2y ago
> unlike openai, google is already cashflow positive and doesnt need to raise any external funds this can quickly change in several quarters, if users decide to leave google search, then all google org/infra complexity will play ver
378.
by riku_iki 2y ago
> The main successful product from OpenAI is the ChatGPT app, but there's a limit on how much you can charge people for subscription fees other significant revenue surfaces: - providing LLM APIs to enterprises - ChatBot Ads market:
379.
by riku_iki 2y ago
Now we can have 4 router startups on top of this
380.
by riku_iki 2y ago
I honestly have no idea how OAI researchers can guarantee cut off date for Antropic models for example.
381.
by riku_iki 2y ago
It could work with knowledge cut off if they can reliably guarantee it, and also make sure LLMs are not searching github under the surface.
382.
by riku_iki 2y ago
> If f(paper) -> code is the weakest part of the chain, it makes sense to target that. my point is that LLMs are already potentially seeing solution on github, so you can't use that benchmark as metric unless there is some explan
383.
by riku_iki 2y ago
> even then the original code wouldn’t conform to the structure of the test. yeah, this part should be central in this work: how well those tests are built, do they actually are catching data leakage, how this is measured, etc.
384.
by riku_iki 2y ago
Realistically person need just some XXXM to cover 99% of luxury lifestyle, and even stock drop 20x they will still have those money. Also, Larry has some sickness, maybe his thoughts currently not about money at all.
385.
by riku_iki 2y ago
One concern is that coding/logical puzzles are verticals where LLMs have lots of training data, they require small context window, and that's why they are doing well, but they don't necessary scale/generalize on other to
386.
by riku_iki 2y ago
Government is also regularly being hacked
387.
by riku_iki 2y ago
I think many people are very not Ok how government handles data: https://news.ycombinator.com/item?id=43237352
388.
by riku_iki 2y ago
> executive insurance > It's not my fault we're doing badly, I brought in a team of best in class consultants and even THEY couldn't help us turn things around yet - it must be the macro economy, industry headwinds, its
389.
by riku_iki 2y ago
AWS is probably guarding their IP, there is a chance McKinsey is just making things up, you can't verify it.
390.
by riku_iki 2y ago
I didn't get idea of this benchmark, they ask to produce code to replicate result of papers, which already have code on github?..

More ›