debug.py

from dataclasses import dataclass, field
import matplotlib.pyplot as plt
import io
import csv
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib
import importlib
import os
import functools
import itertools
import torch

import torch.nn as nn
import numpy as np
import tensorflow as tf
import tensorflow_datasets as tfds
import tensorflow_gan as tfgan
import tqdm
import io
import inspect
sns.set(font_scale=2)
sns.set(style="whitegrid")

import models
from models import utils as mutils
from models import ncsnv2
from models import ncsnpp
from models import ddpm as ddpm_model
from models import layerspp
from models import layers
from models import normalization

#from configs.ncsnpp import cifar10_continuous_ve as configs
from configs.ddpm import cifar10_continuous_vp as configs
config = configs.get_config()

checkpoint = torch.load('exp/ddpm_continuous_vp.pth')

#score_model = ncsnpp.NCSNpp(config)
score_model = ddpm_model.DDPM(config)
score_model.load_state_dict(checkpoint)
score_model = score_model.eval()
x = torch.ones(8, 3, 32, 32)
y = torch.tensor([1] * 8)
breakpoint()
with torch.no_grad():
  score = score_model(x, y)