import numpy as np import torch import torch.optim as optim from tensorboard_logger import configure, log_value from torch.autograd.variable import Variable from src.Utils import read_config from src.Generator.generator import Generator from src.Models.loss import losses_joint from src.Models.models import CsgNet, ParseModelOutput from src.Utils.learn_utils import LearningRate from src.Utils.train_utils import prepare_input_op, Callbacks import time if len(sys.argv) > 1: config = read_config.Config(sys.argv[1]) else: config = read_config.Config("config.yml") model_name = config.model_path.format(config.proportion, config.top_k, config.hidden_size, config.batch_size, config.optim, config.lr, config.weight_decay, config.dropout, "mix", config.mode) print(config.config) config.write_config("log/configs/{}_config.json".format(model_name)) configure("log/tensorboard/{}".format(model_name), flush_secs=5) callback = Callbacks(config.batch_size, "log/db/{}".format(model_name)) callback.add_element(["train_loss", "test_loss", "train_mse", "test_mse"])
import torch import torch.optim as optim from torch.autograd.variable import Variable from src.Utils import read_config from src.Generator.shapenet_generator import get_shapenet_data from src.Models.loss import losses_joint from src.Models.models import CsgNet, ParseModelOutput from src.Utils.learn_utils import LearningRate from src.Utils.train_utils import prepare_input_op, Callbacks import deepdish as dd from torch.utils.data import DataLoader from vis_voxels import vis_voxels device = torch.device("cuda") config = read_config.Config("config.yml") with open("draws.txt", "r") as file: unique_draws = file.readlines() unique_draws = [x.strip() for x in unique_draws] primitives = dd.io.load('data/primitives.h5') max_len = 7 def _col(samples): return samples def get_csgnet(): csgnet = CsgNet(grid_shape=[64, 64, 64], dropout=config.dropout, mode=config.mode,