Why does a double BOS token get added to my InferenceResource.chat_completion calls and cause and check_double_bos_eos error?

[thaddavis]

When I make this call I get an error?

client.inference.chat_completion(
        messages=[
            UserMessage(
                content="hello world, write me a 3 word poem about the moon",
                role="user",
            ),
        ],
        model_id="meta-llama/Llama-3.2-1B-Instruct",
        stream=False,
    )

[yanxi0830]

Wondering what is the inference provider you are using? It looks like ollama from the logs, if using ollama, which model are you using from ollama run?