You sure this is 248B params?
#3
by
unaidedelf87777
- opened
It seems you made a typo in your readme. It says this is 248b params, but the pytorch_model.bin is only 189 bytes, which means its almost empty. I dont know if this means that it failed to upload the model or what, but just letting ya know.
@unaidedelf87777 I think it may be quantized via a new advanced method
@pszemraj , there is no way to quantize a 240b param model down to 189bytes. From 2-300gb to 189bytes? Not gonna happen. That’s physically impossible.
no its an unfinished project