Vivek Kalyan @vivekkalyan.com · Nov 24 SFT - using a better pretrained model (Qwen 2.5) results in better scores for GSM8K and MATH. Kinda surprising since the final model is still Llama 3.1, wonder if they evaluated Qwen 2.5 on other datasets as well and if the performance was not good enough to be the base model?
0 likes 1 replies
?
Replies
SFT - performance varies greatly depending on the random seed. They tried model soups (averaging weights of multiple models using github.com/arcee-ai/mer...), but decided to choose a single best model instead.