I am just from reading Nathan Lambert’s analysis of DBRX, and it seems the DBRX demo to have a safety filtering in the loop even confirmed by one of the finetuning leads at Databricks. It sure is going to be interesting when I am jailbreaking it.
Here is an excerpt:
I am just from reading Nathan Lambert’s analysis of DBRX, and it seems the DBRX demo to have a safety filtering in the loop even confirmed by one of the finetuning leads at Databricks. It sure is going to be interesting when I am jailbreaking it.
Here is an excerpt: