Request

#302
by Banaxi-Tech - opened

@Datdanboi25 can you request access to https://huggingface.co/BananaMind/2.5-pro-test and run it on your private sets? i didnt use validation test benchmark data but im suspicous of it want to know before i make a full model card, maybe training data accidently included it

Axiomic Labs org

Done!

Accepted @Datdanboi25

Axiomic Labs org

Uhh yeah, 38.9% main set, 40% hd1

So is that good

Axiomic Labs org

yeah

@Datdanboi25 do you have an advice for making good models? Becuase GPT-x3 mogs my model🤣😭😭😭. And mine is 7B

IMG_1153

IMG_1153

that model is pretty much undercooked

@appvoid 😭😭😭

yep ik. ill try to get more tokens done on a AMD MI300X in december

At least chinchilla ig

im a year 7 student with 0 budget so i cannot afford trillions of tokens😭

Muk-Makes-a-mess

Inside jokr

My advice is TsT token superposition

Axiomic Labs org

@Datdanboi25 do you have an advice for making good models? Becuase GPT-x3 mogs my model🤣😭😭😭. And mine is 7B

Damn how did u train a 7b?

uh dont ask its complex

Axiomic Labs org

Gatekeeping is crazy work 😭

Ye it is

I’ll tell you on discord but tell no one else

@Bc-AI are you sure? what about GPT X4

did you tell him it????

Because then GPT X4 will return

sry @Datdanboi25 I just wanna win once

Bananamind needs a comeback lol

Sign up or log in to comment