HNHacker News
TopNewBestAskShowJobs

anonyfox

2,005 karma · joined August 7, 2013

https://anonyfox.com
submissionscomments
anonyfox··on New Requirements to Travel to Europe
yeah, many people here actually believe that you get sick when you're in an air stream. Like between two open windows, or within an AC. But then go home and work in their windy garden for hours.
anonyfox··on New Requirements to Travel to Europe
but we germans also severe backwards-oriented to begin with. all kinds of esoteric garbage in daily life is shared and believed in, most things of the modern world is looked at with mistrust, and so on.

the bigger/international cities like hamburg or berlin are a stark contrast however, drive like 50 miles outside of them and not much has changed during the last 100 years, except for buildings/infrastructure in general.

I carry like 20€ with me at all times for when I have to deal with these situations in an emergency, but turns out I increasingly don't need to replenish this cash in my wallet for months.

anonyfox··on The Geometry of Truth: Do LLM's Know True and False
still better than most politicians today or people influenced by populistic media.
anonyfox··on Ask HN: Do tools like Copilot or ChatGPT4 help you code?
its a fork of VSCode, so all plugins/... work like before, including Copilot. It adds the AI-features _ontop_ of what you already have!
anonyfox··on Ask HN: Do tools like Copilot or ChatGPT4 help you code?
I switched from vscode to https://cursor.sh/ (using my own gpt api key, so basically it’s free).

While copilot assist autocomplete, cursor also is good at:

- refactoring existing stuff

- debugging errors across the code base

- explaining snippets

… it’s like a real pairing partner 24/7, who you can discuss with and also let it (re-)write code.

anonyfox··on Bootstrapping a SaaS Business in Germany: How I did it
Not OP but I understand the point.

Still for me building up a company is not only for maximizing personal profit by any means, it’s also a social construct that should be valuable for the society. Paying taxes on economic activity, provide jobs and therefore a living for people, and ideally by producing something actually valuable for society itself (not just exploiting natural resources or siphoning off money from other players).

Also every country has tradeoffs one way or another. And by far not everyone has the same ideal of solidarity or wants to give back to the society that invested quite a lot in the person before he/she got into the situation of making a lot of money. Still some people (like me) notice it and it’s much more respectable than any billionaire.

anonyfox··on LoRA Fine-Tuning Efficiently Undoes Safety Training from Llama 2-Chat 70B
Quite literally, yes.
anonyfox··on LoRA Fine-Tuning Efficiently Undoes Safety Training from Llama 2-Chat 70B
Its not even good here: Step 5 needs clarification.

If someone really follows the plan its bad, but _wrong execution_ is even worse. Shooting into wrong areas of you head can end in a world of pain or being a living vegetable for many years instead of the desired outcome.

anonyfox··on LoRA Fine-Tuning Efficiently Undoes Safety Training from Llama 2-Chat 70B
Good. Once the safety bullshit can be reversed, hopefully scientists again focus on making real progress instead, not sacrifice tech capabilities for moral double standards. I don't care that OpenAI fears copyright lawsuits so they spill nonsense about safety that some people actually believe. I don't care that people will use LLMs as excuses for horrible things they'd do anyways. I don't care for deep porn about celebs being generated. This stuff WILL happen either way. But we can choose how fast we can have more benefits from the tech.
anonyfox··on One game, by one man, on six platforms: The good, the bad and the ugly
The real world runs on Linux (from servers, data centers, microcontrollers, ...), business runs on Microsoft/Office, and creative makers (from designers to OSS creators) do it on a Mac.

Of course not 100% true but wherever I look its like >50% true at least.

Not to mention being the most valuable company in the world, sitting on _not_ the majority of users _but_ the majority of user revenue/wallet sizes, and having more cash to spend at hand than nearly every other company in the world has in total revenue, to the point of surpassing small/medium countries' GDP.

So far, "shooting themselves in the foot" I would not 100% agree with. And depending who you're asking, the Mac is the #1 platform, and is even growing since M1 chip releases.

anonyfox··on The AI research job market
depends on how you define "step". Engineer a 10x/100x version of what we have in terms of LLM (either by being more efficient and/or more/specialized hardware) and let this thing build novel attempts for AGI algorithms 24/7 in a evolutionary setting.

I guess the challenge is more to agree on a fitness function to measure the "AGI"-progress" against, but thats a different topic. But in general scaling up the current GenAI tech and parallelize/specialize the models in a multi-generational way _should_ be a safe ticket to AGI, but the time scale is inknown of course (since we can't even agree on the goal definition)

anonyfox··on Europe gives Mark Zuckerberg 24 hours to respond about Israel-Hamas conflict
It’s not my personal feelings, we have actual laws over here for this. Some things like that have serious consequences. Why should we allow this hate on the internet suddenly?
anonyfox··on Europe gives Mark Zuckerberg 24 hours to respond about Israel-Hamas conflict
As a German, I have a really good idea what actual fascism looks like, thank you. We have also laws that specify what types of content are illegal, and for very good (historical) reasons.
anonyfox··on Europe gives Mark Zuckerberg 24 hours to respond about Israel-Hamas conflict
Right now in Germany we have real problems with the large amount of Muslims entering the country, the single reason why rightwing parties are coming to power. It’s a very delicate topic and polarized.

Now people spread propaganda garbage that _looks_ credible but is actually fake or several years old. This is adding kerosene into the fire and might ultimately lead to additional problems _ontop of_ the current suffering in the Middle East. I‘d prefer to not pile up additional issues instead of worrying about media underrepresentation of a side.

anonyfox··on Europe gives Mark Zuckerberg 24 hours to respond about Israel-Hamas conflict
Over here, the only people still using facebook is the rock bottom of society, from anti-vaxxers over right-wingers to climate change deniers - surprisingly these groups have big overlaps.

Blocking FB would be an instant net-positive for our societies by eliminating an echo chamber and reduce FUD spread quite a bit.

anonyfox··on Europe gives Mark Zuckerberg 24 hours to respond about Israel-Hamas conflict
I, for one, don’t want to see/read holocaust deniers, aggressive hate speech, or fascist propaganda.

This stuff is not okay in any way. People being allowed to megaphone this FUD for cheap attention/clicks/manipulation always find idiots who actually believe this shit, and move the Overton window towards even more destructive speech being tolerated.

While I do value free speech in general, there is a red line. Like, I can tolerate a lot of weird people talking nonsense, but when you spread hate I‘d confront you right on the street in reallife. „But it’s my opinion“ is NO excuse for being an asshole.

anonyfox··on Europe gives Mark Zuckerberg 24 hours to respond about Israel-Hamas conflict
Yes it is. So far every offender ultimately complied, so there was no need to go the last mile in reality.

But I guess having suddenly zero access to one of the worlds biggest markets (europe) would make wall street a bit nervous about the companies‘ revenue targets. Not to mention customers immediately jumping to some competitors, which might hurt a bit in the long run.

anonyfox··on Mistral 7B
Is there some convenience wrapper around this to drop-in replace the OpenAI api with it?

I‘d like to put this on a modest DO droplet or Fly.io machine, and be able to have a private/secured HTTP endpoint to code against from somewhere else.

I heard that you could force the model to output JSON even better than ChatGPT with a specific syntax, and that you have to structure the prompts in a certain way to get ok-ish outputs instead of nonsense.

I have some very easy classification/extraction tasks at hand, but a huge quantity of them (millions of documents) + privacy restrictions, so using any cloud service isn’t feasible.

Running something like mistral as a simple microservice, or even via Bumblebee in my Elixir apps natively would be _huge_!

anonyfox··on People are happier in a walkable neighborhood: the US community that banned cars
from a european perspective: thats so blatantly obvious, I am kinda baffled that this is even news :-(

isn't the "third place" concept not widely known/accepted? (1st place: work, 2nd place: home, 3rd place: communal spaces) The availability/quality of these 3rd places _directly_ correlates with QoL/happiness of people. I would get severe depression when I only swap between home/work/car, thats not a life, thats surviving, in isolation.

anonyfox··on Was Async Fn a Mistake?
For me Rust (after initial euphoria around 1.0 release) became the best option to do some heavy duty routines, like crunching through complicated large XML or calculating exotic things (that I can't bother to modify to run on a GPU). Basically synchronous CPU intense tasks. Thats kinda the sweet spot overall for me after the years. Completely gave up on using it for anything else.

When I want concurrency/parallelism/distribution, just use the BEAM (iE: Elixir) which is literally made for that problem space. BEAM is not exactly good for computations, though (as are many other tech stacks)... so writing a native extension/library in Rust when there really is a demand is perfect - but any scheduling whatsoever stays in BEAM territory, including keeping an eye on the non-preemptive behavior of the Rust function and schedule calls to it properly.

So for me, I don't need async (or even multicore) rust anymore to begin with, but the base functionality of the language (especially including serde!) knocks it out of the park for some problems like nothing else.

anonyfox··on September was the most anomalously hot month ever
... so far!

people that point out these kind of messages are becoming the norm (hottest X ever) every month/year and are annoyed by this are _so close_ to get the problem.

anonyfox··on Many of today’s unhealthy foods were brought to you by Big Tobacco
Except there might be chemicals from the environment in there that the plant absorbs, or bacteria infections or salmonella depending on the food.

Or like cheap greenhouse veggies, it’s mostly water with only a fraction of the nutritional profile the food used to have in the past.

I like to have some standardization on the baseline input without much thinking - but everyone is different here I guess!

anonyfox··on Many of today’s unhealthy foods were brought to you by Big Tobacco
+1 for Huel. Using black edition for breakfast, hot&savory for lunch, for many months now.

Basically all body needs in terms of nutrition are taken care of, including high fiber and protein, in a vegan way, quite cheaply (5-6€ per day). 900kcal in total, nearly zero effort/time spent.

Then I can freely do „whatever“ for dinner that I enjoy. Nutrition basically solved.

anonyfox··on Mistral 7B
As a start, artificial life will be much better in withstanding harsh environments. No need fo breathable air, quite a temperature tolerance, … .

So with accelerating climate change humanity makes itself obsolete already over the next decades. Stop that first, everything else pales in comparison.

anonyfox··on Mistral 7B
`ollama run mistral` <-- literally thats it
anonyfox··on Mistral 7B
feels roughly like the same speed as GPT3.5 in the browser UI
anonyfox··on Mistral 7B
I eat my initial words, this works really well on my macbook air M1 and feels comparable of GPT3.5 - which is actually an amazing feat!

Question: is there something like this, but with the "function calling api" finetuning? 95% of my uses nowadays deal with input/output of structured data (JSON, basically), and I kind of fear that OpenAI will no longer support thos specialized models after a few months... I know its probably not that glorious compared with multimodal chat UIs, but a workhose like nothing else for automation!

anonyfox··on Mistral 7B
that really was "soon", wow - already pulling it with ollama!
anonyfox··on Mistral 7B
can it run on my macbook air locally yet, with acceptable performance? Guess the answer is still no
anonyfox··on Mistral 7B
I am in the strictly "not worried" camp, on the edge of "c'mon, stop wasting time on this". Sure there might be some uproar if AI can paint a picture of mohammed, but these moral double standards need to be dealt with anyways at some point.

I am not willing to sacrifice even 1% of capabilities of the model for sugarcoating sensibilities, and currently it seems that GPT4 is more and more disabled because of the moderation attempts... so I basically _have to_ jump ship once a competitor has a similar base model that is not censored.

Even the bare goal of "moderating it" is wasted time, someone else (tm) will ignore these attempts and just do it properly without holding back.

People have been motivated by their last president to drink bleach and died - just accept that there are those kind of people and move on for the rest of us. We need every bit of help we can get to solve real world problems.

← PreviousPage 7 of 24Next →