what's the origin train data
#5
by
baby1
- opened
could you share the original training dataset
From my understanding, the base model is llama 13B using vicuna's train.py to train share-gpt data ?
Refer to the base repo for information
https://huggingface.co/lmsys/vicuna-13b-delta-v0
it
Refer to the base repo for information
https://huggingface.co/lmsys/vicuna-13b-delta-v0
this is the delta weight
When we training , which dataset of sharegpt we applyed?