kaiokendev
/

superhot-13b-16k-no-rlhf-test

Model card Files Files and versions Community

Resources

View closed (0)

Possibility that Claude/ChatGPT uses similar techniques on adjusting RoPE sampling rate?

#4 opened almost 2 years ago by

PPL chart for 16k models?

#3 opened almost 2 years ago by

7B, 33B and 65B versions?

#2 opened almost 2 years ago by

Difference between this and 8k version?

#1 opened almost 2 years ago by