Plans for 13b version?
#5
by
rombodawg
- opened
Do you guys plan on making a 32k context llama 2 13b? And if so any idea when it would come out?
Yep, stay tuned :) We will probably have a instruct/chat version for 7B first to figure out the full recipe, and start to do larger models
Hi! Thanks for this model. For some reason this is the best open source model featuring both long context + instruct functionality. I've only got poor results from the other alternatives.
I'm really, really looking forward to the 13B version! What's the ETA for it?
Someone needs to port flash attention to Apple Silicon GPUs. Another great model I can't run :(