HNHacker News
TopNewBestAskShowJobs

villish

101 karma · joined August 3, 2024

submissionscomments
villish··on Jellyfin 12.0
I like them both, but I can never give up plexamp. I’ve tried navidrome and various jellyfin clients but I keep going back.
villish··on GPT-6 Astra in code review: Gains, privacy, and cost
(58% on OpenRouter)

I believe both things can be true. OAI/Anthropic selling more tokens than ever, and downloadable weight models increasing in market share.

I doubt OpenAI would copy Fables pricing if it wasn’t financially beneficial. They can only burn through VC funding for so long.

villish··on Artificial Analysis Intelligence Index v4.2
> different benchmarks for different domains.

Labs know the only thing the public even discusses on model releases are benchmarks, so they devote a majority of training on just benchmaxxing. It’s marketing.

Muse Spark looks great in benchmarks. Everyone I know who has tried it (Rust & C++ projects) has determined it’s a resounding “meh”. That doesn’t mean it’s not a great tool for frontend devs, I wouldn’t know.

villish··on GPT-6 Astra in code review: Gains, privacy, and cost
That likely won’t change if other competitors don’t take the lead at some point. If companies are willing to pay top dollar for the best models AND they get to extract as much money from Chinese labs distilling Astra/Fable it makes no sense to lower prices. Obviously not great for everyday users who don’t have unlimited money.
villish··on Artificial Analysis Intelligence Index v4.2
Hopefully this wakes people up from this addiction to benchmarks when discussing various AI models. Different model families have strengths and weaknesses in various domains, but those are never discussed.
villish··on IBM Bob
“Bobcoins are the consumption-based billing metric used for Bob subscriptions. They provide a transparent, predictable way to measure and bill for your Bob usage.”

I’m getting sick of these mobile game currency style credits/coins that obfuscate costs or usage. It’s honestly pretty gross.

Just telling us exactly what we get for our money.

villish··on GPT-6 Astra makes major gains in the Artificial Analysis Coding Agent Index
I have no faith in these benchmarks. Muse 1.3 shouldn’t even be in the same conversation yet it scores above GPT 5.6 & 6.0
villish··on K2 Horizon: A connected fleet of six open models
Frontier. Everything is frontier. K2 not to be confused with the other K2, or K3 that is also frontier.
villish··on Google avoids a breakup of its ad tech business
Yeah and how was it done? Why did all (60?) of the trump campaigns lawsuits get thrown out due to lack of evidence?

Why did the democrats just simply forget to do it again?

How many hoops will you have to jump through to answer these questions?

villish··on Can I opt out of my input or output data being used for training?
Pretty much. Europe is basically not at all involved in the AI race and is reliant on US and China.
villish··on Path to Astra: critical capabilities and frontier safeguards
Explosive/Chemical based is my line of thinking. A Fable level model with zero safe guards planning and coaching someone is within the realm of possibility now.

That is much more dangerous than a man with a gun.

villish··on Can I opt out of my input or output data being used for training?
To have at least a choice of using a European trained model. Downloadable weights is not open source. Mistral is able to respond to any regulatory queries about training data and concerns.
villish··on Sweden pauses Denmark power cable over EU grid rules
> Or is this just a dog whistle for the HN crowd who like to complain about anything related to EU regulations?

Only topped by the number of people who love to cheer for US vs China while living in neither.

villish··on Path to Astra: critical capabilities and frontier safeguards
> These are not really analogous: direct lethality of LLMs is quite low.

I imagine an LLM with no safeguards and a psychopathic mind would be considerably more dangerous than a gun. I don’t mean only for hacking. Though everyone in the world suddenly having a pocket expert hacker should be taken seriously.

I find it completely insane when people bash the labs for even considering safety. The entitlement is off the charts.

villish··on Path to Astra: critical capabilities and frontier safeguards
Should everyone have access to guns too? I ask because there seems to be a disconnect where a lot of people who live in countries with gun control don’t want AI offensive capabilities to be regulated.

I presume your country doesn’t have AI sovereignty so no matter what you won’t have access to models aligned with your beliefs.

villish··on My local model setup on an M4 Pro Mac Mini
Keep in mind these downloadable models use 3-10x the amount of tokens as well. You really can’t beat a couple $20 subscriptions.

https://quesma.com/benchmarks/babaisbench/

villish··on Apple caught off guard by AI demand for Mac Mini and Mac Studio
If you have been using cloud hosted models, you will be severely disappointed with what you’d be able to run on 16GB VRAM. You will spend most of your time fighting with the model to fix its mistakes.
villish··on The EU has begun enforcing the AI Act: first RFIs to model providers
Why would they not apply to all models? The models with downloadable weights aren’t “open”. Meaning no one can actually give a summary about their training content and security. Which is what they’re asking US labs.
villish··on The EU has begun enforcing the AI Act: first RFIs to model providers
The article only lists US labs, I wonder if that was an oversight. Chinese models are also available to Europeans through API providers. Surely they risk fines too.. right?
villish··on Hy4 preview
Countries that aren’t competitive need access to training datasets so that they may train their own similarly capable models and be sure of the inputs. Governments cannot blindly trust open weight models from China and the US.
villish··on GLM-5.3 is now open-weight
What quant are you running and tps?
villish··on Trade (and Tariffs)
> but agree on not "bullying" allies for no reason and little profit.

This is what confuses me most about Trumps tactics. These are longtime friends and allies who would’ve gladly worked together to try to improve any perceived imbalances.

I cannot fathom treating a friend this way. It makes no sense to me and is destructive.

villish··on Qwen3.8-Flash-Next
Thats only important if running it locally is critical for privacy reasons or just as a hobby.

Time has a cost in business. If a model needs 30 million tokens to achieve a similar result as another that can do it in 10 million, that 60 tokens per second will take a long time.

villish··on Disrupting a new covert influence campaign from Russia
Evangelical Christians are the target audience. They are strong supporters of Israel because of their interpretation of the bible.
villish··on Why some US restaurants are banning tips
> Part of European dining is the social aspect

So like every culture in the world then?

villish··on Your Open Source Model Could Have a Hidden Time-Release Backdoor
Both propriety and open weight models are closed source. Open source would include training data so anyone could train a similarly capable model.
villish··on There's no reason for software to be slow anymore
Such as? There are alternatives to almost anything you can think of already. All popular sites and services have clones. What more do you want?

The issue I think you’ll run into is that they’re lacking users and kind of just worse versions. Usually because people who say things like this, don’t actually want to be personally inconvenienced to switch.

villish··on Japan tried to build an operating system for the world, the US intervened
Did you forget about what happened with Hong Kong?
villish··on U.S. Department of Energy Launches the Genesis Open Models Initiative
I have no doubt companies like Microsoft, Amazon, and Google will rush to give them all the data they want in order to keep those government contracts flowing.
villish··on Muse Code and Muse Spark 1.2
Muse 1.1 performed relatively well according to benchmarks, putting it within spitting distance of the premier models. However, based on the results I got from it and the review videos I watched, it wasn’t even close.

Opus 5 is incredible at making games. Almost like a generation better than other models from my experience. You won't see that if you just look at the popular benchmarks..

You have to test each model on your actual use case to see how well it really performs.

← PreviousPage 2 of 5Next →