You must log in or # to comment.
Fully open is extremely good, it uses entirely open datasets and has all training and inference code. We need more things like that.
Just checked this out on HF and this shit isn’t even on the top 100 in any leaderboard. What gives?
