When can we anticipate the release of the DPO version?
#3
by HR1777 - opened
Could you please provide us with information regarding the estimated release date of the DPO version of bagel-34b-v0.4? Additionally, could you offer some insight into the improvements made in bagel-34b-v0.4 compared to bagel-34b-v0.2?"
I may actually re-train the base model because this one has some issues with random tokens being generated with sglang/vllm, likely due to chatml tokens. Will provide updates.
That's a good news. We are waiting for it and the new DPO version.
HR1777 changed discussion status to closed