File size: 1,391 Bytes
63fba53 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 | import torch
from abc import ABC, abstractmethod
class AbstractProcessor(ABC):
"""
Abstract class for processors to show what methods they need to implement.
Processors handle text encoding and image preprocessing.
"""
@abstractmethod
def encode_text(self, prompt):
pass
@abstractmethod
def preprocess_images(self, images: list):
pass
class FlamingoProcessor(AbstractProcessor):
"""
Processor class for Flamingo.
"""
def __init__(self, tokenizer, vision_processor):
"""
OF does not use same vision processor, image_processor only transforms single image
"""
self.tokenizer = tokenizer
self.vision_processor = vision_processor
def encode_text(self, prompt):
self.tokenizer.padding_side = "left"
# For generation padding tokens should be on the left
return self.tokenizer([prompt],
return_tensors="pt",
)
def preprocess_images(self, images: list):
vision_x = [self.vision_processor(im).unsqueeze(0) for im in images]
vision_x = torch.cat(vision_x, dim=0)
return vision_x
def clean_generation(response):
"""
for some reason, the open-flamingo based model slightly changes the input prompt (e.g. prepends <unk>, an adds some spaces)
"""
return response.replace('<unk> ', '').strip() |