ismailmaj

3 ms·

1.
by ismailmaj 12d ago
5 for sure is unbearable, 4.8 is a reasonable sweet spot, 4.6 tends to just agree with whatever I say. But lately I haven't downgraded because 5 is so much better at tool use, so I just accept the cost of Fable for chatting and hope Op
2.
by ismailmaj 12d ago
FYI Mistral at launch just dropped the weights without any model architecture mentioned. Most of the OSS models follow the same architecture which is Llama +- a few things, so it wasn't too hard for people to make it work.
3.
by ismailmaj 17d ago
Assuming that current LLMs don't have a "self" whatever that means, what makes you think it won't emerge after enough intelligence? Also it's not even necessary, it's sufficient for it to be aligned with human
4.
by ismailmaj 18d ago
Pay isn't the main problem with Paris, it's quality of life. Housing near work is mostly old Haussmannian buildings that are rent capped and where demand far exceeds supply, so most of the stock is unmaintained. You either accept
5.
by ismailmaj 18d ago
All original scientist co-founders of H have left. From what I've seen in the past few months, Mistral is the unique European lab still trying to compete (I wouldn't count Poolside as EU).
6.
by ismailmaj 25d ago
Obviously they do not, the same way Amazon isn't treating their drivers poorly, it is just third party contractors "surprisingly" not doing their job well.
7.
by ismailmaj 1mo ago
It feels really weird to see this when I keep seeing the CEO and CTO of HuggingFace push for local inference to win.
8.
by ismailmaj 1mo ago
even for just in-memory quick data analysis?
9.
by ismailmaj 2mo ago
There is something very SIMD-coded in GPU programming which is coalesced stores/loads, if a warp (32 threads) handles contiguous memory, it will create ~4 transactions instead of 32.
10.
by ismailmaj 2mo ago
Given the trend of revisions, I wouldn't be surprised if the real loss is around 6 figures, May 2026 revision went from 172k to 63k.
11.
by ismailmaj 2mo ago
Unsure if it's LLMs that fail at tabular data or its just that tree boosting are spectacular at that task.
12.
by ismailmaj 2mo ago
For my projects, I can produce with LLMs a codebase that I'd be proud of, but it requires extensive use of the plan mode and nagging anytime the implementation looks much more complex than the feature at hand requires. Anytime I tried
13.
by ismailmaj 2mo ago
it's 80% less cost, not 80% in efficiency gains, could be that Luna was overpriced to begin with, we don't have much info on the models themselves. Assuming the efficiency gains are real, I feel like something has to give, maybe w
14.
by ismailmaj 2mo ago
While the model itself is great at hiding the answer, the AI summary keeps giving it away, it's so annoying and AFAIK it cannot be disabled.
15.
by ismailmaj 2mo ago
I'd pay good money to see OpenAI "oh fuck" war rooms.
16.
by ismailmaj 3mo ago
I think I would be horrified looking at your Claude API bill.
17.
by ismailmaj 3mo ago
Apparently plus users do not have access to Sol, so I'm really worried about the ugly Terra Pelican.
18.
by ismailmaj 3mo ago
I don't know what the day to day would be like but it was a ML research engineer role.
19.
by ismailmaj 3mo ago
I'm pretty confident I would've been able to come up to the solution in better circumstances, maybe even without hints. But it was clear in this case that the interviewer just took a question from the company's bank of questi
20.
by ismailmaj 3mo ago
I got that recently, I really didn't like that question. For the n-th percentile version, the obvious solution is sorting and it takes 10 seconds to get to that point, 5 minutes of implementation with tests. Good. It's all downhil
21.
by ismailmaj 3mo ago
High interest rates and instability makes it a bad environment to invest, it raises the bar on speculative/growth investment, and tech is mainly that. Mix that with heavy AI bills, there isn't a lot of budget left for hiring.
22.
by ismailmaj 3mo ago
From talking to someone that worked at YouTube for 15 years, they still had a lot of core Python code in 2016 that was legacy from the OG company/team and that code needed to be transitioned to follow the Google way of doing things in
23.
by ismailmaj 3mo ago
I'll pick small company, thank you.
24.
by ismailmaj 3mo ago
I'm not getting the Opus 4.8 switch for coding, supposedly given how fast I reached the usage limit, which is kind of nice.
25.
by ismailmaj 3mo ago
It's great for defining fancy floats used in machine learning e.g. https://github.com/zml/zml/blob/33ced8fa078b3c7c8c709bd526ae...
26.
by ismailmaj 3mo ago
they're still kings for training, though I've heard Anthropic is training now on JAX+TPU setup, so might not be a monopoly in that segment.
27.
by ismailmaj 3mo ago
I can talk about France specifically, the working age population feels squeezed currently because a big portion of their wages goes to pay pensions, let's be specific: for a 100k paid by your employer, 46k goes to you after all taxes,
28.
by ismailmaj 3mo ago
> The risk of being cut off from strategically important technologies, once minimal, is now very real. Good timing given the Fable 5 story.
29.
by ismailmaj 4mo ago
My main read on this is that European governments will more aggressively invest in regional AI labs. For Nvidia chips you could've deluded yourself that the US is just anti-china, that position is harder to argue for right now.
30.
by ismailmaj 4mo ago
In my experience a lot of companies try really hard to be data oriented and try to find objective metrics for impact, sometimes it’s good, often it’s bad. Like LOC count, PR count, time in meetings or time spent at the office. Enough of thi

More ›