ttul

37 ms·

601.
by ttul 3y ago
It's also why you can see galaxies in the background!
602.
by ttul 3y ago
Whatever they're doing in the background to magically run these huge models, I appreciate the hidden complexity a great deal. The Playground is wonderfully responsive and minimalist if you want to kick the tires on a variety of high qu
603.
by ttul 3y ago
One way to think about high-dimensional spaces is that there is a huge amount of volume relative to the number of coordinates in the space. Yes, there are infinitely many coordinates; however, each coordinate has more dimensions to it relat
604.
by ttul 3y ago
I appreciate the need for some sovereign oversight of telecoms for security reasons, but I have never understood why Canada disallows foreign companies to operate a telecom or mobile operator entirely. If Verizon or AT&T started install
605.
by ttul 3y ago
Every Canadian federal government eventually bows down to the four big media/telecom companies that collectively control the internet and mobile communications of the entire population. It’s a disgusting situation, but I’m not sure wha
606.
by ttul 3y ago
Quantization means reducing the number of bits used to encode each floating point number constituting a parameter in the model So instead of having billions of possible values per weight, you might have just 255. The model has to have its w
607.
by ttul 3y ago
Well, since Together did hire the author of Mamba…
608.
by ttul 3y ago
See my poorly educated answer above. I don’t think that’s how MoE actually works. A new mixture of experts is chosen for every new context.
609.
by ttul 3y ago
This being said, presumably if you’re running a huge farm of GPUs, you could put each expert onto its own slice of GPUs and orchestrate data to flow between GPUs as needed. I have no idea how you’d do this…
610.
by ttul 3y ago
Someone smarter will probably correct me, but I don’t think that is how MoE works. With MoE, a feed-forward network assesses the tokens and selects the best two of eight experts to generate the next token. The choice of experts can change w
611.
by ttul 3y ago
Yes. Keep in mind that, aside from their former board, they probably employ a good PR firm.
612.
by ttul 3y ago
I get the distinct sense that Anthropic needs some better product managers and application engineers. You can destroy a lot of business value by making stupid, avoidable moves like this.
613.
by ttul 3y ago
I beg to differ. If I start with a summary that was produced by a model that has an excellent broad understanding of all prior work, it can help to fill in gaps so that when I do read the paper later, I parse it better. I have ADHD and a le
614.
by ttul 3y ago
If Apple does start enforcing signed attestations, they will say that it's to reduce abuse. I have no doubt (being in the anti-abuse world) that spammers and phishing gangs will immediately begin using Beeper to spam iMessage users bec
615.
by ttul 3y ago
Hailing Jeremy Howard and his team at Fast.ai to make it so!
616.
by ttul 3y ago
Yeah, as the other comment here indicates, I used Claude to generate an initial summary, which I verified and edited before posting. I am teaching myself deep learning and I use Claude to analyze papers before diving in. Many papers are har
617.
by ttul 3y ago
The paper claims that Mamba outperforms Transformers on modeling audio waveforms. Once a signal gets into a computer, it is technically always discrete, but I think what you’re referring to here is the ability to model continuous signals li
618.
by ttul 3y ago
I got Claude to write a draft and then I scanned the paper to make sure it didn’t make any egregious errors. I cleaned up its output, removing some bullets that I felt didn’t add anything the abstract had already covered well. I also prompt
619.
by ttul 3y ago
This paper introduces a new class of models called selective state space models (S6 or selective SSMs). The key ideas and results are: 1. SSMs are a type of recurrent model that can scale linearly with sequence length, making them more effi
620.
by ttul 3y ago
Rumour has it that the NPCs will engage in realistic dialogues with the player that are generated by an LLM. By the time this game releases, running a reasonable LLM locally would not be out of the question. That alone would greatly increas
621.
by ttul 3y ago
Try $1B in one day when it releases!
622.
by ttul 3y ago
That’s great. I didn’t realize the hiring was so frosty right now. Maybe I should dust off the startup pants.
623.
by ttul 3y ago
I started my first startup in September 2001. We got amazing people for nothing and free office space. Post-2008, my series-A failed to produce any sales because the market was simply gone. But at least I could find and retain good engineer
624.
by ttul 3y ago
A very fair comment. I use an M2 MBP with max specs. It’s very powerful, but the Nvidia card draws a whole lot more power…
625.
by ttul 3y ago
Sure, they’d be well placed to do this.
626.
by ttul 3y ago
If you are a programmer at heart, ComfyUI will feel very comfortable (pun intended). It's basically a visual programming environment optimized for the type of compositional programming that machine learning models desire. The next thin
627.
by ttul 3y ago
Good lord. I can get a 2048x2048 upscaled output from a very complex ComfyUI workflow on a 4090 in 15 seconds. This includes three IPAdapter nodes, a sampling stage, a three-stage iterative latent upscaler, and multiple ControlNets. Macs ar
628.
by ttul 3y ago
You can use MailChannels for free via Cloudflare and there are no volume limits.
629.
by ttul 3y ago
I recall having a conversation in about 2016 with a lawyer who was looking into crypto for potential opportunities. After a thorough analysis, he said, “In a few years, all these people will be in jail.” He was right.
630.
by ttul 3y ago
In a democracy, those limitless powers are limited by voters. The government can only pass laws and implement regulations that are acceptable to a democratic public. To the extent that voters want to move their money freely around the world

More ›