I have trained (and annealed) a model on my data, using close to default training configs.
Is there a way for me to convert this trained pytorch checkpoint into a huggingface model, and then upload it?
When I tried doing so, it seems like the keys of the pytorch vs huggingface models are very different.