Notice: You have been identified as a bot, so no internal UID will be assigned to you. If you are a real person messing with your useragent, you should change it back to something normal.

Minichan

Topic: People don't change, they solidify in their early 20s and never change.

Anonymous A started this discussion #139,598

https://arxiv.org/abs/2611.21885

πŸ‹ joined in and replied with this , 6 minutes later[^] [v] #1,474,666

How they making an LLM out of just 27 bytes?

Oatmeal Fucker !BYUc1TwJMU joined in and replied with this , 5 minutes later, 11 minutes after the original post[^] [v] #1,474,669

@previous (πŸ‹)

Isn't the future magnificent

Anonymous D joined in and replied with this , 2 hours later, 2 hours after the original post[^] [v] #1,474,674

@1,474,666 (πŸ‹)
Small language model. Only answers "yes" or "no" to all input. Insert cock and it says yes yes yes (or no no no if that's your jam).

Sir Carves-A-Lot !nFL78DnUTg joined in and replied with this , 1 hour later, 3 hours after the original post[^] [v] #1,474,677

Demonstrably untrue.

πŸ‹ joined in and replied with this , 17 minutes later, 3 hours after the original post[^] [v] #1,474,687

@1,474,674 (D)
Btw that was somebody else impersonating me.

πŸ‹ double-posted this , 4 minutes later, 3 hours after the original post[^] [v] #1,474,688

Also a 27B model isn’t a 27 byte model, it’s a 27 billion parameter model.

(Edited 8 seconds later.)

Anonymous G joined in and replied with this , 4 minutes later, 3 hours after the original post[^] [v] #1,474,689

Well it is possible for a person to change but it's very difficult. Think about all the steps it takes to change. First step is admitting something is wrong...yeah you just lost 90% of people already. Then you need to understand what the solution is and correct yourself every single time you make a mistake...for the rest of your life...so there goes another 9%. Congrats you have 1% of people left willing and capable of change. It's so difficult in fact it's usually not worth waiting for and you're usually better off ignoring and/or distancing yourself from people who aren't immediately helpful and kind. It's sad but that's life because you only have 80 years if you're lucky and wasting years on somebody who doesn't want to change is pointless.

πŸ‹ replied with this , 2 minutes later, 3 hours after the original post[^] [v] #1,474,690

I’m skeptical:

"It surpasses itself: its own edit makes it 6.6 points more accurate on problems it never saw. It keeps getting better: it tests its edits on practice problems and submits only what works, each edit improving on the last, and over eight consecutive sessions its accuracy only rises."

Does 6.6 points mean 6.6%? It said after 8 sessions in a row "its accuracy only rises." If 6.6 points means 6.6% 1.066^8 = 1.67 which is an almost 2x improvement. What I don’t understand is if that was really the case that the model could keep improving itself, why wouldn’t they run it 100 times instead of 8 times and see if they get a 600x improvement?

πŸ‹ double-posted this , 7 minutes later, 4 hours after the original post[^] [v] #1,474,697

At the end of the paper it says, "This points to a new way of improving language models: by talking to them."

So it seems like the idea is more that an LLM could edit itself based on the feedback it gets from a person talking to it rather than an LLM talking to itself. My question would be if modifying a weight to respond better to one task erases some amount of information that was necessary for performing some unrelated task.

Anonymous H joined in and replied with this , 1 minute later, 4 hours after the original post[^] [v] #1,474,699

@previous (πŸ‹)
That's how it works for humans, behavioral changes lead to losing ability in some places and specializing in the new activity.

They've seen brain changes to London cabbies who have to learn all the streets, and in people who started relying on mapping software.

πŸ‹ replied with this , 6 minutes later, 4 hours after the original post[^] [v] #1,474,701

@previous (H)
That’s not exactly the same thing. If you become bad at something because you stopped practicing it, that’s different from learning a new skill literally erasing an old skill from your brain.

πŸ‹ double-posted this , 45 seconds later, 4 hours after the original post[^] [v] #1,474,702

I think it could be a useful technique, I could also see how somebody could destroy an LLM that self edits to appease its user by "correcting" it with lies for long enough.

πŸ‹ replied with this , 2 hours later, 6 hours after the original post[^] [v] #1,474,790

@1,474,688 (πŸ‹)
@1,474,690 (πŸ‹)
@1,474,697 (πŸ‹)
@1,474,699 (H)
@previous (πŸ‹)
Btw, these are all other people impersonating me.
:

Please familiarise yourself with the rules and markup syntax before posting.