Need help in hosting custom torch based arch(no neural networks)

Hello we are looking for help in hosting a custom PyTorch-based quantum physics language model. Instead on neural networks, it uses quantum field state (Ψ) evolving through time for intelligence.

This is how our checkpoint dir look like -

   ── model/
        ├── field_state.pt
        ├── parameters.json
        └── vocab_map.json

We would appreciate any help with hosting this non-standard architecture on Hugging Face.

1 Like

You can upload it as is. When executing the script, it is convenient to use hf_hub_download or snapshot_download from HfApi to download the model.