llama3 instruct models have a list of eos_token_ids instead of a single token id. Before this change, this triggered an assertion in the optimum-neuron code during generation.
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.
What does this PR do?
llama3 instruct models have a list of eos_token_ids instead of a single token id. Before this change, this triggered an assertion in the optimum-neuron code during generation.