ladberg

11 ms·

121.
by ladberg 3y ago
Also see: https://langcc.io/
122.
by ladberg 3y ago
The short film is probably a better version of the story: https://www.youtube.com/watch?v=9mbp0DugfCA
123.
by ladberg 3y ago
I don't think there's anything revolutionary, just a lot of parts working very well in tandem: - Multiple cameras per eye, and at a very short distance from your eye - The screen is fixed relative to the cameras for all devices, t
124.
by ladberg 3y ago
I used to work on it and have spent tons of time in the headset. The eye tracking is next-level and it's really the only platform that exists with eye tracking as a primary input method. I'm pretty confident it will greatly improv
125.
by ladberg 3y ago
Jekyll is used by default for GitHub Pages sites, so I'm sure companies like GitHub would appreciate the savings from speedups like this.
126.
by ladberg 3y ago
I think their "spaces" (i.e. running tons of models for free with a nice UI) are widely used.
127.
by ladberg 3y ago
The point of static sites is to avoid the need for any rendering/compute server-side, which this accomplishes! E.g. you could use this in your GitHub Pages site, which is just served from a CDN without any real backend.
128.
by ladberg 3y ago
Eh, I don't disagree that unsafe languages will continue to be used, but I disagree with ease of compiler design as the reason. You are comparing one of the easier languages to write a compiler for (C) with one of the hardest (Rust), a
129.
by ladberg 3y ago
I think "unique among big tech companies" and "used to be the norm" can both be true! It definitely has pros and cons though: I've worked on a handful of teams at Apple and was very lucky to have been able to work o
130.
by ladberg 3y ago
You're thinking about the wrong bandwidth here. The article is talking about going from the GPU's RAM <-> GPU cores (i.e. through load/store instructions in a cuda kernel), not from CPU's RAM <-> GPU's R
131.
by ladberg 3y ago
I kinda feel like this is DOA until modules are more widespread. Any new cppfront code in an existing C++ codebase can't include any existing header files which automatically excludes it from any C++ codebase I've worked on (or an
132.
by ladberg 3y ago
I think they were hoping that the off-the-shelf model the author used was a novel technique trained on unlabeled data, not a dataset labeled by humans.
133.
by ladberg 3y ago
Define city speeds? Because this is pretty objectively false at the speeds cars usually drive in all the cities I've lived in. Even the pdf you linked has an EV being recorded louder than an ICE at only a single speed out of all th
134.
by ladberg 3y ago
Would recommend Build Bench to help gain a mental model for how different styles of code impact build times: https://build-bench.com/
135.
by ladberg 3y ago
Eh, CO2 is a pretty low energy molecule so doing anything with it usually takes energy input. I'd believe in a ton of other wacky sci-fi ideas turning into reality before that one. That said, I'm hopeful we could use cheap energ
136.
by ladberg 3y ago
Yes, but using .reality files is better for stuff like games in RealityKit. For example it would contain a GPU-ready texture instead of a jpeg (and a lot of other differences), so it's more in-line with how assets are typically loaded
137.
by ladberg 3y ago
Idk, to me this reads pretty clearly remote (for this model in particularly): > My LLM tool provides command-line access to a wide variety of language models, both via web APIs and self-hosted on your own machine ... The brand new llm-re
138.
Paul Debevec talks early CGI breakthroughs with VFX artists [video](youtube.com)
2 points by ladberg 3y ago|0 comments
139.
by ladberg 3y ago
C doesn't even originally have true/false, I think you may be conflating the two concepts that "any nonzero int is truthy" and "boolean expressions evaluate to ints". The standard mandates that boolean expressi
140.
by ladberg 3y ago
I feel like person you're replying to specifically implied that OpenAI does it differently in that aspect, and they were also high up at Netflix so is likely very familiar :)
141.
by ladberg 3y ago
> Eyes are not required for proprioception. No, but they will mess with and override proprioception. I've used the Vision Pro and the first time you try to take a drink with it you'll probably miss your mouth. I think the main
142.
by ladberg 3y ago
The 68k is per millisecond , which lines up with 68M per second.
143.
by ladberg 3y ago
Does anyone actually care about their looks on a plane? I can't imagine prioritizing what other plane passengers think of me over my personal comfort and ability to get through sitting in a cramped seat all day...
144.
by ladberg 3y ago
Bleecker and 7th Ave For comparison: I'm trying to beat the numerous dollar pizza and food carts nearby, not normal "fast food" like the more expensive Five Guys on that intersection.
145.
by ladberg 3y ago
Not sure if you're being facetious or just don't understand the reality of living in NYC... I can probably walk to a dozen different big grocery stores in 15m and they're ALL more expensive than the cheap fast food in the sam
146.
by ladberg 3y ago
I don't mean the cost of my time that work pays me, just how much money I'd personally pay to avoid doing something something I don't like (schlepping grocery bags on the subway, doing dishes). In general cooking is the fun p
147.
by ladberg 3y ago
> There are even people so confused about food prices that they make claims that fast food is cheaper than groceries. I live in an expensive part of NYC and have to go decently far out (by subway, I don't have a car) to find groceri
148.
by ladberg 3y ago
Yep, I pretty much exclusively use Apple Maps in the US and it's great but when I'm on vacation it's often unusable. Most recently, I was in Georgia (country) and navigation of any kind is nonexistent. Like often times I woul
149.
by ladberg 3y ago
> Not sure if all this is true, but I did not feel like finding out. It is most definitely not true
150.
by ladberg 3y ago
In my opinion it's significantly more meaningful than blockchain hype. Blockchain doesn't add any value whatsoever for 99% of the use-cases that companies were advertising it for, but AI can immediately and cheaply create output t

More ›