dspy ollama+llama2:13b redundant output

ollama+llama2:13b redundant output

Open Ren-Xin opened this issue 11 months ago • 3 comments

Mar 01 '24 12:03 Ren-Xin

Hello @Ren-Xin I think the output generation is the model property.

Mar 02 '24 03:03 insop

Thanks， @insop So, what should I do to reproduce the results of llama2:13b in the paper? Now his results are very poor.

Mar 02 '24 06:03 Ren-Xin

I think if you'd use few-shot learning examples, say through BootstrapFewShot, and the model will perform better.

Mar 03 '24 22:03 insop