Comment by winddude

I got worse than 1 token/sec, and yes, wasn't impressed with bloom results, but I believe it's also very foreign language heavy. I haven't tried it yet but I believe flexGen benchmarked faster as well.