Gating Schnell on HF broke a lot of open source code

[josephrocca]

@timudk Just in case y'all didn't realize, gating the Schnell model on huggingface has broken a lot of code on Github. Quantization libs like deepcompressor, lora trainers, paper replication code, some comfyui stuff, etc. I don't understand the reason for gating an Apache 2.0 model - requiring people sign up to HF, and give you their name and email. Seems like a dick move, but I figure you may have good reasons (e.g. legal stuff or whatever). In any case, just letting you know in case it was an oversight. Not a big deal in the grand scheme of things, of course.