HF transformers integration (checkpoint transfer)

Hi @jenjsun, I see that you were the last project member to commit here. In case you have access to the HF hub repos (https://huggingface.co/google/videoprism-base-f16r288, etc.), I'd like to seek your cooperation to transfer the checkpoints of the HF transformers implementation. Once the checkpoints are transferred, the transformers PR will be merged with main and these models can be used natively. In case it is preferred to keep the google/videoprism repos for the flax checkpoints only, new repos will be created by HF staffs and most likely the models will be named videoprism-hf.

You may find the converted checkpoints and config files here:
https://huggingface.co/MHRDYN7/videoprism-base-f16r288
https://huggingface.co/MHRDYN7/videoprism-large-f8r288
https://huggingface.co/MHRDYN7/videoprism-lvt-base-f16r288
https://huggingface.co/MHRDYN7/videoprism-lvt-large-f8r288

I'll be creating PRs for all these in the relevant google repos.

Thank You

Review the Hugging Face Transformers PR and the four converted checkpoint and config repositories linked in the issue. Confirm which google repositories or new videoprism-hf repositories should receive the checkpoints and what access is required. Done means the checkpoints are transferred and the Transformers integration can be merged.