Ok good, to know, I hope that information can be accessed without an account, I deleted my Twitter account long before Musk took over and it became X and I don't plan to create an X account now.
I’m certain it is a Llama model. I got it head-to-head with Llama 3.1 70B in a creative writing task, and the generated poems started with the same few lines and stayed extremely close in content even after they started to diverge. I’d guess it is a Llama 3.5 type release.
shroddy@reddit
Do they reveal their mystery models somewhere after some time? I had a few different ones recently and want to know what they were.
DuckyBlender@reddit
Yes, usually when the model officially releases they reveal what model it was
shroddy@reddit
Do you know where that is? I was looking on their blog and the leaderboard.
Physical_Manu@reddit
Usually X.
shroddy@reddit
Ok good, to know, I hope that information can be accessed without an account, I deleted my Twitter account long before Musk took over and it became X and I don't plan to create an X account now.
Physical_Manu@reddit
X without an account has only been getting worse and worse. For some reason it shows you posts out of order and limits how much you can see.
shroddy@reddit
Yeah, it is really bad, I hope lmsys will start to post it elsewhere too.
auradragon1@reddit
Is it good?
Salty-Garage7777@reddit
Came up a couple of times side by side with the best and, IMO, it's way worse.
soup9999999999999999@reddit
Idk I choose it over claude-3-5-sonnet more than once.
Salty-Garage7777@reddit
Interesting! What do you find it better than Sonnet at? What use cases made you choose it?
umarmnaq@reddit (OP)
It's roughly on the same level as GPT-4 (no o), and LLaMA 3 70b
phhusson@reddit
gosh wtf are you sending sms on boot for. please don't be writing a malware.
And if you're doing IoT stuff I'm curious what (I do Android OS development for work and opensource projects)
CheatCodesOfLife@reddit
From the comments about restricted timeframe, Probably enforcing screen time for his kids or something
a_slay_nub@reddit
Restricting your kid's SMS messages during specific timeframes seems like a recipe for disaster. What if it's 1am and they get hurt?
mrjackspade@reddit
The code isn't blocking SMS messaging, it's sending message if the device is on.
There's nothing stopping anyone from turning it on and sending a message if they get hurt.
OrangeESP32x99@reddit
Well, they should probably be in bed at 1am lol
MMAgeezer@reddit
Of course - but they won't always be.
We have to hope for the best and plan for the worst.
Imjustmisunderstood@reddit
OP’s way is the highway
umarmnaq@reddit (OP)
It was just a testing app which checks if the device is turned on during a specified timeframe, and sends an SMS to a number.
R4_Unit@reddit
I’m certain it is a Llama model. I got it head-to-head with Llama 3.1 70B in a creative writing task, and the generated poems started with the same few lines and stayed extremely close in content even after they started to diverge. I’d guess it is a Llama 3.5 type release.
Terminator857@reddit
gemini-test has done well for me. I suspect it is gemini 2.0. I was not inspired by pumpkin_pie results.
guyinalabcoat@reddit
yep, gemini-test has been killing it. I end up nearly always picking it, even over Sonnet 3.5 and 4o.
RuslanAR@reddit
Meta model?
RuthlessCriticismAll@reddit
hallucination
ambient_temp_xeno@reddit
Nemotron is LLaMA 3.1 based, though. So maybe it's 405b Nemotron?
Journeyj012@reddit
Apparently it's way worse, so I'm assuming some small model
fractalcrust@reddit
i'd throw in a fake origin in the system prompt tbh
MoffKalast@reddit
Must've been trained on some synthetic Meta data at least.
Famous-Associate-436@reddit
Maybe its the grok-beta model xAI is beta-testing?
Additional_Ad_7718@reddit
Thanks giving surprise