Kuinox

9 ms·

91.
by Kuinox 1y ago
The em dashes exists in ChatGPT output because existing human text contains it, like journal articles.
92.
by Kuinox 1y ago
Remember, they only measured that the less time you spend on a task, the less you remember it.
93.
by Kuinox 1y ago
Just search it: https://www.theenglishkitchen.co/2015/07/fried-potato-salad.... https://www.food.com/recipe/southern-fried-potato-salad-2134... And simply french fries salad: https:&#x2
94.
by Kuinox 1y ago
> For example, ou can think about how french fries taste but unless you can feel the taste you won't be repulsed if someone suggested making salad out of your left over french fries. Deep fried potato salad is a thing and humans did
95.
by Kuinox 1y ago
You changed the quote: His statement was about a specific technology, an AI that make 3D character move like zombies. The author is also changing the subject of the quote. He said it reminded him of a disabled friend that this technology wa
96.
by Kuinox 1y ago
The online version is way worse, it also have a router that could route it to a random model.
97.
by Kuinox 1y ago
Well, the GPT-5 context windows offer roughly a little more than a MB
98.
by Kuinox 1y ago
It's because math problems allow to easily check that the solution is correct, it allow to do a lot of 'search': https://yellow-apartment-148.notion.site/AI-Search-The-Bitte...
99.
by Kuinox 1y ago
Most of the failures for theses simple logic question come from the inability to simply copy data accuratly. Logic is too abstract to be measured, but this single bench show something getting in it's way. I got another bench that s
100.
by Kuinox 1y ago
https://i.imgur.com/l2elIAv.png Difficulty is the amount of digits, small models struggle with 10 digits numbers, gemini and gpt-5 are very good recent models, gemini start failing before 40 digits, GPT-5 (the one by api, t
101.
by Kuinox 1y ago
LLMs already get enough working memory, they do not fail because of lack of working space.
102.
by Kuinox 1y ago
I always use the lowest temperature that I can input. But GPT-5 doesn't support a temperature setting. You'll get something like: { "error": { "message": "Unsupported value: 'temperature'
103.
by Kuinox 1y ago
Simple maths problems are simple logic problem. Here it doesn't even have to come up with a reasoning, it probably already memorised how to solve sums. Yet it fails at that, it shows it cannot solve logic problems if there are too
104.
by Kuinox 1y ago
I do think that competent humans can solve any arbitrary sum of 2 whole number with a pen, paper and time. LLMs can't do that.
105.
by Kuinox 1y ago
That's was the initial thinking of anyone which I explained this, it was also my speculation, but when you look in it's reasoning where it do the mistake, it correctly extract the digits out of the input token. As I say in anoth
106.
by Kuinox 1y ago
This is a simple sum of 2 whole number, the number are simply big. Most of the time they make a correct summation table but fail to copy correctly the sum result into a final result. That is not a tokenisation problem (you can change the
107.
by Kuinox 1y ago
It's specific model that run for maths. GPT-5 and Gemini 2.5 still cannot compute an arbitrary length sum of whole number without a calculator. I have a proceduraly generated benchmark of basic operations, LLMs gets better at it wi
108.
by Kuinox 1y ago
Mozilla have been building local AI for years, and it have been great. We got tons of local AI services of acceptable quality thanks to them: Translation, text to speech, speech to text. Let them cook, I'm sure a few of theses featur
109.
by Kuinox 1y ago
You dont need to burn the drone but the fiber cable.
110.
by Kuinox 1y ago
Wouldn't a flame thrower be crazy effective against a drone ?
111.
by Kuinox 1y ago
Avalonia is cross plateform unlike WPF.
112.
by Kuinox 1y ago
Author used copilot in rider, tbh it's one of the worst rider. Which llm model was used ? VSCode&VS copilot allow you to select it. I suspect theses bench were run on the default model, ChatGPT 4o, which is now more than a year
113.
by Kuinox 1y ago
> we would not be mystified as to how the polynomial interpolates between the points. We would not be surprised - we wouldnt know how the model resolve the problem. We wouldn't know if it is approximating, calculating the correct va
114.
by Kuinox 1y ago
GitHub wasn't used there. It was on Codeberg.
115.
by Kuinox 1y ago
I tried to find a person I know and it found only unrelated persons.
116.
by Kuinox 1y ago
The problem with your answer is that you make affirmations using logical fallacies. We both don't know how LLMs, and brains works to produce output. Any affirmation toward that without proof is affirming things without any basis. For
117.
by Kuinox 1y ago
Yeah we don't, that's one of the point of my reply, we don't know how LLMs works either.
118.
by Kuinox 1y ago
That's obviously because the brain is not generally intelligent it's just retrieving concepts from a high-dimensional statistically fit function. The extra info injects noise into the calculation which confounds it.
119.
by Kuinox 1y ago
I just tried FaceCheck and it doesnt seem to work at all.
120.
by Kuinox 1y ago
There are two kind of person who ask for a tldr: Someone who doesn't understand or care about the depths of works and only want the information because of this. And people who appreciate it but lacks time. If I see the markers of

More ›