OpenAssistant

/

codellama-13b-oasst-sft-v10

like 65

Hello Andreas,

Sorry if I'm being ignorant and asking wrong questions.

Is it possible to merge this model with Open-Orca/OpenOrcaxOpenChat-Preview2-13B just like how they merged OpenOrca and Platypus to create Open-Orca/OpenOrca-Platypus2-13B.

The reason I'm asking this question is because OpenAssistant/codellama-13b-oasst-sft-v10 scored very good in HumanEval+ benchmarks

while Open-Orca/OpenOrcaxOpenChat-Preview2-13B has very good scores in lm-evaluation-harness

So I'm just wondering if we can merge both to create a model that performs well in all tasks and can become the ultimate goto model in 13B params range.

Thanks.