Lora & push merged model to HF

[noelo]

Hi,
I'm just getting starting with axolotl and fine tuning models. When using lora/qlora I noticed that only the adapter weights are pushed to my huggingface repo. Is is possible to have these weights automatically merged with the original model and then be pushed. The axolotl.cli.merge_lora cli option also doesn't push the merged models. Am I missing something obvious here?

[noelo]

You can always use the huggingface-cli package to upload the model if the code doesn't do it automatically