open_lm icon indicating copy to clipboard operation
open_lm copied to clipboard

Parameter input rotary-freq

Open jmercat opened this issue 9 months ago • 1 comments

This allows to change the rotary positional embedding frequency parameter. This is useful given the more recent approaches: LLaMA 1&2 used 10000 which is the default value here. LLaMA 3 uses 500000, Mistral uses 1000000 LLaMA long context extension works usually increase from 10000 to 100000.

jmercat avatar Apr 30 '24 02:04 jmercat