|
|
|
|
|
|
by jimmySixDOF
631 days ago
|
|
|
tl;dr:
> We are able to reproduce the model benchmark scores initially claimed and are sharing the eval code. That is a big deal considering all the accusations flying around at the time so I hope this forensic update checks out and everyone who jumped to conclusions a couple of weeks ago takes a step back to reflect. |
|
|