import os from typing import Dict, List, Any from PIL import Image import jax from transformers import ViTFeatureExtractor, AutoTokenizer, FlaxVisionEncoderDecoderModel class PreTrainedPipeline(): def __init__(self, path=""): model_dir = os.path.join(path, "ckpt_epoch_3_step_6900") self.model = FlaxVisionEncoderDecoderModel.from_pretrained(model_dir) self.feature_extractor = ViTFeatureExtractor.from_pretrained(model_dir) self.tokenizer = AutoTokenizer.from_pretrained(model_dir) max_length = 16 num_beams = 4 self.gen_kwargs = {"max_length": max_length, "num_beams": num_beams} @jax.jit def _generate(pixel_values): output_ids = self.model.generate(pixel_values, **self.gen_kwargs).sequences return output_ids self.generate = _generate # compile the model image_path = os.path.join(path, 'val_000000039769.jpg') image = Image.open(image_path) self(image) image.close() def __call__(self, inputs: "Image.Image") -> List[str]: """ Args: Return: """ pixel_values = self.feature_extractor(images=inputs, return_tensors="np").pixel_values output_ids = self.generate(pixel_values) preds = self.tokenizer.batch_decode(output_ids, skip_special_tokens=True) preds = [pred.strip() for pred in preds] return preds