Viewing profile — himata4113

From my experience their capabilities are extremely narrow and generally perform terribly when faced with issues outside comparatively narrow training data.

It's not that their strategy is to train smaller models, it's the only choice they have. Training SOTA takes anywhere from 1.5b to 150b. We don't know the real cost of training for…

I doubt it, a lot of what I use llms for is on the very edge of what these models are trained for.

feels like a harness issue, bug fixes work out fine with omp.sh as long as it can be tested properly. Untestable things tend to lead to hallucinated results though.

Compare the performance of a 980 and a 5050 and I am sure that will answer your question. Also models baked into the silicon are able to achieve efficiency that is simply impossibl…

ehh, they're pretty good at automated performance research and bug fixes, especially when spanned across hundreds of them.

Neither of these points feel true anymore. Models are very much predictable these days (except anthropic models). The real issue stems from letting them work on their own for far t…

I've seen someone rewrite a wireguard proxy in rust for performance without realizing that it's just a daemon that interacts with the kernel to setup routing... But hey, at least t…

It's not only cheap, it's actually free since if you are not using your 95th percentile you are losing money.

The hard part of enrichment is getting to around 20%, after that it actually becomes easier. They're slow-decay isotopes yes, but I am fairly sure you can build a multi-stage devic…

No, it's straight up gambling mechanics. Open any trading card game and tell me that it is not filled with mechanics that effectively resemble gambling.

I don't think 10x density is possible without getting into nuclear, at least not with the chemistries we know of today, everything at a certain point becomes an explosive. Speaking…

Roblox should be outright banned in the EU or at least the top 100 games all qualify for what would be considered lootboxes/gambling and generally predatory behavior. Basically eve…

There's a pretty large paper on the topic: https://arxiv.org/pdf/2606.15497 , but results are somewhat mixed.

Everyone including myself has attempted this relentlessly, it just doesn't work out beyond some arbitrary improvement in specific tested categories instead of broader capability in…

it's actually the end of the beginning, there's walls that self-improving models hit that are never overcome even when given vast amounts of time.

We know that at the end of the day AI intelligence is closely tied to invested compute. The fact that automated AI research might spark some kind of exponencial increase in capabil…

AI is nowhere near the intelligence of a very educated person that has innate talent for problem solving. It does solve the problem of applying human intelligence on problems that …

I would love having the capability of just writing json-esque SQL that has the most basic type checks, alas we live in purgatory with weird build systems to validate queries.

"No implementation detail, algorithm, or configuration is contained in this document by design." + odd page cuts, it's as-if no human has ever looked at this before uploading it.

You can have sol write code and terra will find a plethora of bugs and rip the code apart. In reality this is just the nature of advisory prompting and why advisor from omp.sh is s…

So it's okay for organized crime to do it, but not normal plebs? The former is obviously a lot more harmful and in reality this is just a wakeup call to get their shit together. Pe…

What I really started to notice is that SOTA models are really good at putting themselves out of the job. We can see this already with GPT how luna can do 90% of what sol is used f…

Demand #1 weakens america and everyone around them Demand #2 Why does this matter? The answer was that it does not. ( https://news.ycombinator.com/item?id=49007610 ) Demand #3 This…

it IS cheap (once the weights are released) and it will only become CHEAPER. For around $3700 a month (via loan purchased hardware + energy cost) you can run around 32 concurrent i…