m00x

11 ms·

151.
by m00x 2y ago
Yeah it's SF/Seattle/Chicago $400k is base + bonus
152.
by m00x 2y ago
midsize fintech "startup" 1500 employees with ~$2B ARR.
153.
by m00x 2y ago
Google's main product is ads, not search. Search is also not tech, it's a directory of websites. Pure tech products is GCP or AWS. Instagram's business is photos, Facebook is connections. Figma is a design tool, Slack is a wa
154.
by m00x 2y ago
If the H1B is paid $400k, is it really depressing salaries? It's more about having employees that are 10-15% better than the competition if you can, which makes you much more likely to beat them.
155.
by m00x 2y ago
That's ridiculous. 20% is already enough from top of band.
156.
by m00x 2y ago
We pay $400-700k for L4-L6 and we have a hard time finding qualified candidates.
157.
by m00x 2y ago
I do coding interviews 2-3x/week for a private, but large company. We do not lack candidates, but we lack qualified candidates. Most people that I interview have no clue how software actually works. Most are leetcode monkeys or just re
158.
by m00x 2y ago
Happened to me in Canada too, minus the painkillers.
159.
by m00x 2y ago
Time is a circle. All I know is that I wouldn't use Klarna's products. I work in fintech and the number of mistakes I've seen since devs started relying on AI has accelerated. I use it too, but I'm not an idiot about it.
160.
by m00x 2y ago
For most industries the tech isn't the primary product. Google sells ads, not tech Uber sells rides, not tech. Amazon sells a marketplace, not tech. Tesla sells cars, not tech. Tech is just the tool.
161.
by m00x 2y ago
Ah the junior dev starry-eyed mindset. Doing web dev isn't the only thing in the world. AI will absolutely accelerate development, but we still need so much more modernization across all industries. We still have factories running on
162.
by m00x 2y ago
Boeing all over again
163.
by m00x 2y ago
Everything is a distribution. Most Canadians spend 6h+ waiting for a doctor in the ER, and 20% of Canadians don't have access to a family doctor. Your brother was very lucky, or this is fake.
164.
by m00x 2y ago
I feel like giving teachers and schools more freedoms to implement these things would be a great overall effect. We've seen a large reduction of what teachers are able to do in the last few decades because the school districts have con
165.
by m00x 2y ago
It would be almost impossible to enforce. School smartphone ban has an obvious way to be enforced since it gives teachers the ability to do what they've wanted to do all along.
166.
by m00x 2y ago
In Canada you just die before you get a family doctor :) Last time I went to the ER, the wait was 10h for kidney stones. Thankfully I passed out for 4h in the waiting room from the pain and the stones passed by the time I saw a doctor, so a
167.
by m00x 2y ago
Overspent on the house? Not uncommon in Canada tbh. The country is in a really bad state real-estate wise.
168.
by m00x 2y ago
Stop being dramatic. Most companies appreciate white hats. OP didn't divulge the exploit and is being responsible.
169.
by m00x 2y ago
What have you implemented in CUDA?
170.
by m00x 2y ago
I'm running codeseeker 13B model on my macbook with no perf issues and I get a response within a few seconds. Running a specialist model makes more sense on small devices.
171.
by m00x 2y ago
Do you have the code for that test?
172.
by m00x 2y ago
It would be a good idea to start with some basic understanding of GPU, and realizing why this can't easily be done.
173.
by m00x 2y ago
It won't be 5x slower, it would be 20-50x slower if you would implement it as you said. You can't just "add more ram" to GPUs and have them work the same way. Memory access is completely different than on CPUs.
174.
by m00x 2y ago
Mac minis are great for this. They're cheap-ish and they can run quite large models at a decent speed if you run it with an MLX backend.
175.
by m00x 2y ago
You can just use a CPU in that case, no? You can run most ML inference on vectorized operations on modern CPUs at a fraction of the price.
176.
by m00x 2y ago
It's also impossible and it would need to be a CPU. CPUs and GPUs access memory very differently.
177.
by m00x 2y ago
Inference is still a lot faster on CUDA than on CPU. It's fine if you run it at home or on your laptop for privacy, but if you're serving those models at any scale, you're going to be using GPUs with CUDA. Inference is also a
178.
by m00x 2y ago
Because the CPU has to load the model in parts for every cycle so you're spending a lot of time on IO and it offsets processing. You're talking about completely different things here. It's fine if you're doing a few requ
179.
by m00x 2y ago
FLUX.1 D generation is about a minute at 20 steps on a 4080, but takes 35 minutes on the CPU.
180.
by m00x 2y ago
This is the most script kiddy comment I've seen in a while. llama.cpp is just inference, not training, and the CUDA backend is still the fastest one by far. No one is even close to matching CUDA on either training or inference. The clo

More ›