lossolo

12 ms·

181.
by lossolo 11mo ago
> US's goal there wasn't to win anything since there's nothing of value there War is only a tool, dominating a country or region militarily is not the same as winning a war if you have not achieved its political goals. In
182.
by lossolo 11mo ago
> Everyone I know who did try testosterone or steroids thought they were doing it the safe way (HCG, PCT, limited cycle length) but became unable to produce enough endogenous testosterone by their 40s even with SERMs. Then your sample is
183.
by lossolo 11mo ago
Well, you can always do the same thing that an LLM would: open SEO spam ranking sites "best humidifiers 2025", filled with referral links to Amazon or other sellers, which basically copy product descriptions and assign rankings th
184.
by lossolo 11mo ago
It's hyper individualism that is fueling this type of capitalism, the notion that you owe nothing to anyone and other people in society are not your concern, society itself is not your concern, it’s only about what you want etc. It&#x
185.
by lossolo 11mo ago
I've had problems solved incorrectly and edge cases missed by Sonnet and by other LLMs (ChatGPT, Gemini) and the other way around too. Once they saw the other model's answer, they admitted their "critical mistake". It&#
186.
by lossolo 11mo ago
Apple hardware is great, but the software has become complete garbage. On my Apple Watch I have regularly occurring hiccups where the whole UI freezes, especially when I go into training mode to pick an exercise. On my iPhone, after the liq
187.
by lossolo 11mo ago
> after being one of the biggest contributors to the problem. How many "Made in China" products do you have at home right now? Who is contributing to the problem?
188.
by lossolo 11mo ago
LLMs can "edit" code, but as you say, they do it differently from diffusion models. They operate directly on long text sequences and use much more context, which is one reason they currently work better for coding. Diffusion model
189.
by lossolo 11mo ago
They use bidirectional attention between modalities, not within the same modality. This doesn’t change much in the context you're referring to (coding). How do you think "thinking" works in current SOTA models like GPT-5-Thin
190.
by lossolo 11mo ago
We have already found limitations of the current LLM paradigm, even if we don't have a theorem saying transformers can never be AGI. Scaling laws show that performance keeps improving with more params, data + compute but only following
191.
by lossolo 11mo ago
We're sorry, but this video is not available. :(
192.
by lossolo 11mo ago
It went from 1k -> 2k -> 4k to 8k, so I guess the next will be 16k and then 32k..
193.
by lossolo 11mo ago
So what was wrong with his isolated benchmark code that he shared here?
194.
by lossolo 11mo ago
It's getting out of control. Just yesterday, I was turning on remote desktop on Windows 11. It showed as enabled, with a label indicating that it was listening on port 3389, but it wasn't accessible from a remote host. I checked o
195.
by lossolo 11mo ago
This shouldn't be downvoted because it's stating facts. RCEs for critical infrastructure/OSes are very rare, they don't just grow on trees. I agree that OP exaggerated by saying that any government can buy whatever RCE t
196.
by lossolo 11mo ago
You mean like a demo for simple stuff? Something like hello world type tasks? The small models you mentioned earlier are incapable of doing anything genuinely useful for daily use. The few tasks they can handle are easier and faster to just
197.
by lossolo 11mo ago
Small models handle simple, low context tasks most of the time correctly. But for more complex tasks, they fail due to insufficient training capacity and too few parameters to integrate the necessary relationships.
198.
by lossolo 11mo ago
No. They will go out of context and collapse into loops.
199.
by lossolo 11mo ago
Nothing will change. They will go out of context and collapse into loops.
200.
by lossolo 11mo ago
This is false. You can buy whole H100 clusters in China and Alibaba, Bytedance, Tencent etc have enough cards for training and inference. Shenzhen 2025 https://imgur.com/a/r6tBkN3
201.
by lossolo 11mo ago
This doesn't work like that. An analogy would be giving a 5 year old a task that requires the understanding of the world of an 18 year old. It doesn't matter whether you give that child 5 minutes or 10 hours, they won't be ca
202.
by lossolo 11mo ago
Same, I also use links from HN threads and I donated to archive.is in the past. I don't want them gone.
203.
by lossolo 11mo ago
Minority Report when?
204.
by lossolo 11mo ago
I noticed that too, it's so annoying.
205.
by lossolo 1y ago
> Infamously, the M1 Ultra GPU at 120W equals the performance of the RTX 3090 at 320W That's not true.
206.
Nvidia becomes first company to reach $5T valuation, fueled by AI boom(cnbc.com)
13 points by lossolo 1y ago|1 comments
207.
by lossolo 1y ago
These concerns are exaggerated. I've been running on Hetzner, OVH and friends for 20 years. During that time I've had only two issues, one about 15 years ago when a PSU failed on one of the servers, and another a few years ago whe
208.
by lossolo 1y ago
You can easily fit below 10 billion for the whole datacenter, then you only pay for electricity + maintenance + staff. 100k GPUs cost a few billion USD, that's more than enough to train frontier models, run experiments, and serve model
209.
by lossolo 1y ago
Last time, I gave claude an invoice and asked it to change one item on it, it did so nicely and gave me the new invoice. Good thing I noticed it had also changed the bank account number.. The more complicated the spreadsheet and the more de
210.
by lossolo 1y ago
"Sam Altman says ChatGPT will soon sext with verified adults" https://news.ycombinator.com/item?id=45582874

More ›