General discussion/quants.
Thank you very much for both this, and prime punch my good sir.
Good news after all. Hopefully it does well for my formatting needs too.
We'll see benchs from V4-32K vs V4.20-32K?
v4 baseline back, submitted v4 32k, and 4.20 32k.
Just curious to see the difference with Layla in the mix, V4 didn't feel like a regression at all for me. Leaderboard is kind of a meme too but it's good to check.
Oh honestly i had more trouble with formatting and it felt a tad dumber overall in regards to v4.
And i anticipate v4 32k/ 4.20 to score around 69, but either way if people like them and they perform well i dont care too much.
a tad dumber overall in regards to v4
Well considering how Layla does, it's expected, I think it's a bit of a trade depending on your needs, V4 and V4.20 can be used interchangeably.
Maybe I'll run into noticable issues later.
Never mind, she is dumber. Was doing some multi-character. Well then, it happens to the best of them.