Add Whisper for the task "automatic-speech-recognition" w/o. KV cache

[JingyaHuang]

What does this PR do?

Export

• With CLI

optimum-cli export neuron --model openai/whisper-tiny --task automatic-speech-recognition --batch_size 1 --audio_sequence_length 100 --sequence_length 128 --auto_cast none  whisper_tiny_neuronx/

• With API

from optimum.neuron import NeuronWhisperForConditionalGeneration
save_directory = "whisper_tiny_neuronx/"

# 1. Export
compiler_args = {"auto_cast": "matmul", "auto_cast_type": "bf16"}
input_shapes = {"batch_size": 1, "audio_sequence_length": 100, "sequence_length": 128}
neuron_model = NeuronWhisperForConditionalGeneration.from_pretrained(
    model_id,
    export=True,
    **compiler_args,
    **input_shapes,
)
# Save locally or upload to the HuggingFace Hub
neuron_model.save_pretrained(save_directory)

Inference

neuron_model = NeuronWhisperForConditionalGeneration.from_pretrained(save_directory)
predicted_ids = neuron_model.generate(input_features)
transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
print(transcription[0])

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests?

[HuggingFaceDocBuilderDev]

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.