Great model, BUT it loops!
#13
by ZeroWw - opened
This is a great and fast model, but when presented with a logic problem it LOOPS indefinitely both with reasoning on or off.
Try to increase temperature to 1, and/or use a less quantized version. Yes it loops a lot, but it stops repeating itself, if you increase temperature enough.
With llama.cpp, llama-cli I found not specifying top_k or top_p or temp, gave no 'loops', but using suggested top_p, top_k params gave repetition.