How was this trained ?
#1
by JohanJam - opened
Hello, how did you train this ? Which dataset ? Thanks.
Hi, @JohanJam . We trained this using the SFT library, employing the use of LoRA for the quantization of the necessary model weights.
The datasets varied across the various models we fine tuned, we could help you locate them, however, some are not publicly available.
Do let us know what you need more clarity on. We're in a bit of a fix since we're cooped up with lots of activities, but we'd try to give you the needed info.
Regards, Benedict.