what is the difference between oasst-sft-7-llama-30b-xor oasst-sft-6-llama-30b-xor?
#1
by
Surface
- opened
I just want to know the answer.
I believe the 7 means 7 epochs (vs 6 epochs) of training for the open assistant dataset
I believe the 7 means 7 epochs (vs 6 epochs) of training for the open assistant dataset
SFT-7 means the 7th training run. SFT 7 has slightly updated data/training parameters compared to SFT 6.
OllieStanley
changed discussion status to
closed