You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have pretrain a 13B model with "pos-emb": "rotary" and "seq-length": 2048. I want to fine tune the model with larger seq-length, such as 4096. Can this fine tune improve the model performance or affect the position embedding?
The text was updated successfully, but these errors were encountered:
I have pretrain a 13B model with
"pos-emb": "rotary"
and"seq-length": 2048
. I want to fine tune the model with largerseq-length
, such as4096
. Can this fine tune improve the model performance or affect the position embedding?The text was updated successfully, but these errors were encountered: