typo in new doc?
In the beginning:
"DeepSeek-R1-0528 is DeepSeek's new update to their R1 reasoning model. The 671B parameter model's unquantized version requires 700GB of disk space whilst the quantized dynamic 1.78-bit version uses 33.8GB (-75% reduction in size)."
Shouldn't it be "around 151GB" from next section?
In the beginning:
"DeepSeek-R1-0528 is DeepSeek's new update to their R1 reasoning model. The 671B parameter model's unquantized version requires 700GB of disk space whilst the quantized dynamic 1.78-bit version uses 33.8GB (-75% reduction in size)."
Shouldn't it be "around 151GB" from next section?
Hi thanks for pointing it out! Docs are still very much a work in progress! If there's any other errors please let us know :)
Quants first, docs later works for me :)
@CHNtentes @gghfez 8bit and 16bit are up
4bit, 3bit etc should be up within an hour
Dynamic imatrix coming within the next 6 hrs
@CHNtentes @gghfez 8bit and 16bit are up
4bit, 3bit etc should be up within an hour
Dynamic imatrix coming within the next 6 hrs
Thanks! Honestly I'm a bit worried about you guys. Are you super human who don't need rest? Please do not wear yourself out...