Good at coding?

#1
by Banaxi-Tech - opened

So this 1.3M model seems to perform shockingly good at code, on my benchmark it got 977 on Code, and my 25M model got 871 so its extremely good!

Wait, are you joking? I find it hard to believe a 1.3M parameter model bested a 25 million parameter model on coding.

Wait, I think its because you trained it on AxiomicLabs/NPset-2-Python-Edu

Well, its still awesome. Er-Tiny beat a 25M model at coding, at least based on your benchmark.

Banaxi-Tech changed discussion status to closed

Sign up or log in to comment