LLaMA-2-7b-chat Editing failed

```bash
CUDA_VISIBLE_DEVICES=7 python -m fastedit.editor \
--data data/example.json \
--model /path/to/Llama-2-7b-chat-hf \
--config llama-7b \
--template default

RuntimeError: probability tensor contains either `inf`, `nan` or element < 0
```

It will be `inf` in generate processing. This method is only use for pretrained model like `Llama-2-7b-hf`?

Start by reproducing the command through the fastedit.editor entry point with data/example.json, the llama-7b config, and the default template. Compare the failure with the Llama-2-7b-chat-hf model against the pretrained model mentioned in the report and trace where generation produces invalid probabilities. Done means the supported-model behavior is fixed or the chat-model limitation is clearly documented.

Written by the indexing model from the issue text.