optimum/bert-base-cased-swag-neuronx

INFERENTIA2 ONLY

from transformers import AutoTokenizer
from optimum.neuron import NeuronBertForMultipleChoice

input_shapes = {"batch_size": 1, "sequence_length": 128, "num_choices": 4}
compiler_args = {"auto_cast": "matmul", "auto_cast_type": "bf16"}
neuron_model = NeuronBertForMultipleChoice.from_pretrained(
    "ZzRanger/bert-base-uncased-finetuned-swag",
    export=True,
    **input_shapes,
    **compiler_args,
)
# Save locally
neuron_model.save_pretrained("bert_base_cased_swag_neuronx")
neuron_model.push_to_hub(
    "bert_base_cased_swag_neuronx",
    repository_id="optimum/bert-base-cased-swag-neuronx",  # Replace with your HF Hub repo id
)