Show HN: LoRA can make distributed fine-tuning far faster

Posted by ericyu3 on Thursday August 31, 2023 @10:43PM
from the show-hn dept.

( 2 comments | 5 points | on HN )

Re:Show HN: LoRA can make distributed fine-tuning far faster by jbentleylong on Thursday August 31, 2023 @11:07PM (#37346333)

LoRA AWS swarms are 60x 1 GPU throughput for fine-tuning?? Kind of a big find

[ Parent | on HN ]

[dead]

Re:Show HN: LoRA can make distributed fine-tuning far faster by Tostino on Friday September 01, 2023 @02:05PM (#37354260)

Generally good as long as you target all layers.

[ Parent | on HN ]