spirosdim

import torch
from detectron2.solver.build import get_default_optimizer_params
from detectron2.solver.build import maybe_add_gradient_clipping

class MyTrainer(DefaultTrainer):
    @classmethod
    def build_optimizer(cls, cfg, model):
        """
        Build an optimizer from config.
        """

spirosdim

import PIL, os
from tqdm import tqdm
import numpy as np
import torch
from torch.utils.data import Dataset

class ImgDataset(Dataset):
    def __init__(self, data_dir, extension='.jpg', transform=None):
        self.data_dir = data_dir
        self.image_files  = [f for f in os.listdir(data_dir) if f.endswith(extension)]

spirosdim

import torch
from detectron2.modeling import build_model
from detectron2.checkpoint import DetectionCheckpointer
import detectron2.data.transforms as T
from detectron2.data.detection_utils import read_image


class MyPredictor:
    """
    source: https://github.com/facebookresearch/detectron2/blob/08f617cc2c9276a48e7e7dc96ae946a5df23af3f/detectron2/engine/defaults.py#L252

spirosdim

class MyTrainer(DefaultTrainer): 
  
    @classmethod
    def build_train_loader(cls, cfg):
        mapper_train = DatasetMapper(cfg, 
                       is_train=True, 
                       augmentations=[ #Apply a sequence of augmentations.
                                      T.RandomFlip(prob=0.5, horizontal=False, vertical=True),
                                      T.RandomFlip(prob=0.5, horizontal=True, vertical=False),
                                      T.RandomRotation(angle=[0, 90, 180, 270], sample_style="choice"),

spirosdim

import torch
import torch.nn as nn
import json
from pathlib import Path
from transformers import AutoConfig, AutoTokenizer, AutoModel

tokenizer = AutoTokenizer.from_pretrained('distilbert-base-uncased')

class Tagger(nn.Module):
    """

spirosdim

from pathlib import Path
import torch
import torch.nn as nn
from transformers import AutoConfig, AutoTokenizer, AutoModel

class Tagger(nn.Module):
    """
    Minimal model class just for inference
    """
    def __init__(self):