🚩 Report
#4
by
vanakema
- opened
This is incorrectly labeled as bf16, but it is an (I assume accidental) reupload of a 4-bit quant
@reach-vb
@pcuenq
@osanseviero
@awni
Hey admins, just wanted to raise this to y'all to see if maybe you can remove it. Once it is, I can rename the corrected model I uploaded here: mlx-community/Meta-Llama-3.1-70B-Instruct-bf16-CORRECTED
It was an accident.
Fixing it ;)
Thanks for flagging it!
Indeed it's actually a quantized model instead of half-precision
Uploading the fix right now.
Fixed ✅