Gotta remember they were trained off of the internet. Which is to say the largest body of people loadly professing the opinions are fact and refusing to say otherwise.
This. It’s deceptive af. My old job we had our holiday party at a KBBQ spot and all 12 of us went through like 30 bottles. I remember looking up and thinking, wtf where did all these bottles come from?!.. oh wait I’m blasted. God I miss that little company, such a fun job.
Look brother I don’t agree with your lifestyle but I will defend to the death your right to rawdog seventeen other anonymous partners in the local motel 6.
This is flavored soju, which is usually around 10-12% abv and is sweetened. Very drinkable. Unflavored soju is a little less friendly if you don’t like tasting alcohol.
I’ve never had the flavored ones, but not really. It’s actually a very smooth drink. It’s a neutral spirit made from sweet potato so it tastes kind of like vodka, but without the bite because it’s half the alcohol content.
This might be happening because of the ‘elegant’ (incredibly hacky) way openai encodes multiple languages into their models. Instead of using all character sets, they use a modulo operator on each character, to make all Unicode characters represented by a small range of values. On the back end, it somehow detects which language is being spoken, and uses that character set for the response. Seeing as the last line seems to be the same mathematical expression as what you asked, my guess is that your equation just happened to perfectly match some sentence that would make sense in the weird language.
Can’t find the exact source–I’m on mobile right now–but the code for the gpt-2 encoder uses a utf-8 to unicode look up table to shrink the vocab size. github.com/openai/gpt-2/blob/master/…/encoder.py
There are bindings in java and c++, but python is the industry standard for AI. The libraries for machine learning are actually written in c++, but use python language bindings. Python doesn’t tend to slow things down since machine learning is gpu-bound anyway. There are also library specific programming languages which urges the user to make pythonic code that can be compiled into c++.
I suppose it’s conceivable that there’s a bug in converting between different representations of Unicode, but I’m not buying and of this “detected which language is being spoken” nonsense or the use of character sets. It would just use Unicode.
The modulo idea makes absolutely no sense, as LLMs use tokens, not characters, and there’s soooooo many tokens. It would make no sense to make those tokens ambiguous.
I completely agree that it’s a stupid way of doing things, but it is how openai reduced the vocab size of gpt-2 & gpt-3. As far as I know–I have only read the comments in the source code– the conversion is done as a preprocessing step. Here’s the code to gpt-2: github.com/openai/gpt-2/blob/master/…/encoder.py I did apparently make a mistake, as the vocab reduction is done through a lut instead of a simple mod.
Yeah and of the entire subgroup they pick furry talk to be drunk? Near every one of those cat fuckers are alcoholics. At minimum triple this. To really get that meow meow talk you’d need to convince one of those teetotaler tiny girls to drink but you know she won’t. Instead you’re trying to get an alcoholic 190lb man named Blaze drunk enough to start that kitty babble? Try six lol.
I had a good friend in college who had never touched alcohol before, finally convinced him to come out for a beer and he threw up wasted all over the table after 1 pint of beer
ok so I don’t really drink alcohol, but some online friends got me to try soju. Grabbed some from the store and… I don’t get it. very much an alcohol taste to it. I can taste the plum in it, but it’s just not that great imo.
As an alcoholic, alcohol literally tastes bad and your brain only starts liking it once it recognizes the high. I mean it’s literally ethanol, a poison
I’m not an alcoholic, but used to drink wodka regularly when going out like decades ago. Even today when I smell non scented hand sanitizer, I can feel a little buzz in my brain.
sopuli.xyz
Oldest