def main():
conf = Configs()
experiment = Experiment(writers={'sqlite'})
experiment.calc_configs(conf, {'epochs': 'random'}, ['set_seed', 'loop'])
experiment.add_models(dict(model=conf.model))
experiment.start(run=-1)
conf.loop()
def main():
conf = Configs()
experiment = Experiment(writers={'sqlite'})
experiment.calc_configs(conf, {'optimizer': 'adam_optimizer'},
['set_seed', 'main'])
experiment.add_models(dict(model=conf.model))
experiment.start()
conf.main()
def main():
conf = Configs()
experiment = Experiment(writers={'sqlite', 'tensorboard'})
experiment.calc_configs(conf,
{'epochs': 'random'},
['set_seed', 'main'])
experiment.add_models(dict(model=conf.model))
experiment.start()
conf.main()
def main():
conf = Configs()
experiment = Experiment(writers={'sqlite'})
experiment.calc_configs(conf,
{},
run_order=['set_seed', 'main'])
experiment.start()
conf.main()
from typing import List, Optional, NamedTuple
import torch
import torch.nn
from parser.load import EncodedFile, split_train_valid, load_files
from lab.experiment.pytorch import Experiment
from parser import tokenizer
# Configure the experiment
from parser.batch_builder import BatchBuilder
from parser.merge_tokens import InputProcessor
EXPERIMENT = Experiment(name="id_embeddings",
python_file=__file__,
comment="With ID embeddings",
check_repo_dirty=False,
is_log_python_file=False)
logger = EXPERIMENT.logger
# device to train on
device = torch.device("cuda:1")
cpu = torch.device("cpu")
class ModelOutput(NamedTuple):
decoded_input_logits: torch.Tensor
decoded_predictions: Optional[torch.Tensor]
probabilities: torch.Tensor
logits: torch.Tensor
"""
import argparse
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torchvision import datasets, transforms
from lab.experiment.pytorch import Experiment
# Declare the experiment
EXPERIMENT = Experiment(name="mnist_pytorch",
python_file=__file__,
comment="Test",
check_repo_dirty=False
)
logger = EXPERIMENT.logger
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
self.conv1 = nn.Conv2d(1, 20, 5, 1)
self.conv2 = nn.Conv2d(20, 50, 5, 1)
self.fc1 = nn.Linear(4 * 4 * 50, 500)
self.fc2 = nn.Linear(500, 10)
def forward(self, x):
from typing import NamedTuple, List, Tuple
import torch
import torch.nn
import parser.load
import parser.tokenizer
from lab import colors
from lab.experiment.pytorch import Experiment
from simple_model import SimpleLstmModel
from parser import tokenizer
# Experiment configuration to load checkpoints
EXPERIMENT = Experiment(name="simple_lstm_1000",
python_file=__file__,
comment="Simple LSTM All Data",
check_repo_dirty=False,
is_log_python_file=False)
logger = EXPERIMENT.logger
# device to evaluate on
device = torch.device("cuda:1")
# Beam search
BEAM_SIZE = 8
class Suggestions(NamedTuple):
codes: List[List[int]]
matched: List[int]
def create_exp():
return Experiment(name="test", comment="Test")