Troubleshooting Unsupported Models#

Sometimes a model listed on the Hugging Face Hub may not work with NeMo AutoModel. If you encounter any such model, please open a GitHub issue with the model ID and any stack trace you see.

Common Issues#

Issue

Example Error

Solution

Model has explicitly disabled training in its definition code

—

Request support via a GitHub issue. We can add the model through our custom registry.

Model requires a newer transformers version

The checkpoint you are trying to load has model type deepseek_v32 but Transformers does not recognize this architecture.

Upgrade transformers (and NeMo AutoModel if needed), or open a GitHub issue.

Model upper-bounds transformers, requiring an older version

—

Open a GitHub issue.

Unsupported checkpoint format

OSError: meta-llama/Llama-2-70b does not appear to have a file named pytorch_model.bin, model.safetensors, ...

Open a GitHub issue or request the model publisher to share a SafeTensors checkpoint.

These cases typically stem from upstream packaging or dependency constraints. You would encounter the same issues when using transformers directly, as AutoModel mirrors the familiar load and fine-tune semantics.

Steps to Try#

  1. Upgrade NeMo AutoModel to a release that supports the required transformers version. See Installation.

  2. Enable remote code — if the model uses custom code, set trust_remote_code: true in your model: config. See Hugging Face API Compatibility.

  3. Open a GitHub issue with the model ID and error so we can prioritize support or add a registry-backed implementation.