Tensor parallelism for CUDA (#598)
* Add llama tensor split mode setting * Represent tensor as GPU split strategy * Put custom GPU split strategy last
N
Neil Mehta committed
5b959d6ed7854fc204dd9ec336aa9e164e7d7f61
Parent: 682b27f
Committed by GitHub <noreply@github.com>
on 5/28/2026, 6:27:56 PM