iicd/xautodl
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Prototype MAML

Browse Source
This commit is contained in:
D-X-Y 2021-05-10 09:42:42 +08:00
parent 5457dcf042
commit 755c7c90cf
1 changed files with 67 additions and 75 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

142
exps/LFNA/basic-maml.py
View File

@ -1,8 +1,8 @@
##################################################### #####################################################
# Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.04 # # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.04 #
##################################################### #####################################################
# python exps/LFNA/basic-maml.py --env_version v1 # # python exps/LFNA/basic-maml.py --env_version v1 --hidden_dim 16 --inner_step 5
# python exps/LFNA/basic-maml.py --env_version v2 # # python exps/LFNA/basic-maml.py --env_version v2
##################################################### #####################################################
import sys, time, copy, torch, random, argparse import sys, time, copy, torch, random, argparse
from tqdm import tqdm from tqdm import tqdm
@ -30,73 +30,71 @@ from lfna_utils import lfna_setup, TimeData
class MAML: class MAML:
"""A LFNA meta-model that uses the MLP as delta-net.""" """A LFNA meta-model that uses the MLP as delta-net."""
def __init__(self, container, criterion, meta_lr, inner_lr=0.01, inner_step=1): def __init__(
self, network, criterion, epochs, meta_lr, inner_lr=0.01, inner_step=1
):
self.criterion = criterion self.criterion = criterion
self.container = container # self.container = container
self.network = network
self.meta_optimizer = torch.optim.Adam( self.meta_optimizer = torch.optim.Adam(
self.container.parameters(), lr=meta_lr, amsgrad=True self.network.parameters(), lr=meta_lr, amsgrad=True
)
self.meta_lr_scheduler = torch.optim.lr_scheduler.MultiStepLR(
optimizer,
milestones=[
int(epochs * 0.25),
int(epochs * 0.5),
int(epochs * 0.75),
],
gamma=0.3,
) )
self.inner_lr = inner_lr self.inner_lr = inner_lr
self.inner_step = inner_step self.inner_step = inner_step
self._best_info = dict(state_dict=None, score=None)
print("There are {:} weights.".format(w_container.numel()))
def adapt(self, model, dataset): def adapt(self, dataset):
# create a container for the future timestamp # create a container for the future timestamp
y_hat = model.forward_with_container(dataset.x, self.container) container = self.network.get_w_container()
loss = self.criterion(y_hat, dataset.y)
grads = torch.autograd.grad(loss, self.container.parameters())
fast_container = self.container.additive( for k in range(0, self.inner_step):
[-self.inner_lr * grad for grad in grads] y_hat = self.network.forward_with_container(dataset.x, container)
) loss = self.criterion(y_hat, dataset.y)
import pdb grads = torch.autograd.grad(loss, container.parameters())
pdb.set_trace() container = container.additive([-self.inner_lr * grad for grad in grads])
w_container.requires_grad_(True) return container
containers = [w_container]
for idx, dataset in enumerate(seq_datasets): def predict(self, x, container=None):
x, y = dataset.x, dataset.y if container is not None:
y_hat = model.forward_with_container(x, containers[-1]) y_hat = self.network.forward_with_container(x, container)
loss = criterion(y_hat, y) else:
gradients = torch.autograd.grad(loss, containers[-1].tensors) y_hat = self.network(x)
with torch.no_grad(): return y_hat
flatten_w = containers[-1].flatten().view(-1, 1)
flatten_g = containers[-1].flatten(gradients).view(-1, 1)
input_statistics = torch.tensor([x.mean(), x.std()]).view(1, 2)
input_statistics = input_statistics.expand(flatten_w.numel(), -1)
delta_inputs = torch.cat((flatten_w, flatten_g, input_statistics), dim=-1)
delta = self.delta_net(delta_inputs).view(-1)
delta = torch.clamp(delta, -0.5, 0.5)
unflatten_delta = containers[-1].unflatten(delta)
future_container = containers[-1].no_grad_clone().additive(unflatten_delta)
# future_container = containers[-1].additive(unflatten_delta)
containers.append(future_container)
# containers = containers[1:]
meta_loss = []
temp_containers = []
for idx, dataset in enumerate(seq_datasets):
if idx == 0:
continue
current_container = containers[idx]
y_hat = model.forward_with_container(dataset.x, current_container)
loss = criterion(y_hat, dataset.y)
meta_loss.append(loss)
temp_containers.append((dataset.timestamp, current_container, -loss.item()))
meta_loss = sum(meta_loss)
w_container.requires_grad_(False)
# meta_loss.backward()
# self.meta_optimizer.step()
return meta_loss, temp_containers
def step(self): def step(self):
torch.nn.utils.clip_grad_norm_(self.delta_net.parameters(), 1.0) torch.nn.utils.clip_grad_norm_(self.container.parameters(), 1.0)
self.meta_optimizer.step() self.meta_optimizer.step()
self.meta_lr_scheduler.step()
def zero_grad(self): def zero_grad(self):
self.meta_optimizer.zero_grad() self.meta_optimizer.zero_grad()
def save_best(self, network, score):
if self._best_info["score"] is None or self._best_info["score"] < score:
state_dict = dict(
criterion=criterion,
network=network.state_dict(),
meta_optimizer=self.meta_optimizer.state_dict(),
meta_lr_scheduler=self.meta_lr_scheduler.state_dict(),
)
self._best_info["state_dict"] = state_dict
self._best_info["score"] = score
def main(args): def main(args):
logger, env_info = lfna_setup(args) logger, env_info, model_kwargs = lfna_setup(args)
model = get_model(dict(model_type="simple_mlp"), **model_kwargs)
total_time = env_info["total"] total_time = env_info["total"]
for i in range(total_time): for i in range(total_time):
@ -104,19 +102,12 @@ def main(args):
nkey = "{:}-{:}".format(i, xkey) nkey = "{:}-{:}".format(i, xkey)
assert nkey in env_info, "{:} no in {:}".format(nkey, list(env_info.keys())) assert nkey in env_info, "{:} no in {:}".format(nkey, list(env_info.keys()))
train_time_bar = total_time // 2 train_time_bar = total_time // 2
base_model = get_model(
dict(model_type="simple_mlp"),
act_cls="leaky_relu",
norm_cls="identity",
input_dim=1,
output_dim=1,
)
w_container = base_model.get_w_container()
criterion = torch.nn.MSELoss() criterion = torch.nn.MSELoss()
print("There are {:} weights.".format(w_container.numel()))
maml = MAML(w_container, criterion, args.meta_lr, args.inner_lr, args.inner_step) maml = MAML(
model, criterion, args.epochs, args.meta_lr, args.inner_lr, args.inner_step
)
# meta-training # meta-training
per_epoch_time, start_time = AverageMeter(), time.time() per_epoch_time, start_time = AverageMeter(), time.time()
@ -131,8 +122,7 @@ def main(args):
) )
maml.zero_grad() maml.zero_grad()
meta_losses = []
all_meta_losses = []
for ibatch in range(args.meta_batch): for ibatch in range(args.meta_batch):
sampled_timestamp = random.randint(0, train_time_bar) sampled_timestamp = random.randint(0, train_time_bar)
past_dataset = TimeData( past_dataset = TimeData(
@ -145,21 +135,23 @@ def main(args):
env_info["{:}-x".format(sampled_timestamp + 1)], env_info["{:}-x".format(sampled_timestamp + 1)],
env_info["{:}-y".format(sampled_timestamp + 1)], env_info["{:}-y".format(sampled_timestamp + 1)],
) )
maml.adapt(base_model, past_dataset) future_container = maml.adapt(model, past_dataset)
import pdb future_y_hat = maml.predict(future_dataset.x, future_container)
future_loss = maml.criterion(future_y_hat, future_dataset.y)
pdb.set_trace() meta_losses.append(future_loss)
meta_loss = torch.stack(meta_losses).mean()
meta_loss = torch.stack(all_meta_losses).mean()
meta_loss.backward() meta_loss.backward()
adaptor.step() maml.step()
debug_str = pool.debug_info(debug_timestamp)
logger.log("meta-loss: {:.4f}".format(meta_loss.item())) logger.log("meta-loss: {:.4f}".format(meta_loss.item()))
per_epoch_time.update(time.time() - start_time) per_epoch_time.update(time.time() - start_time)
start_time = time.time() start_time = time.time()
import pdb
pdb.set_trace()
logger.log("-" * 200 + "\n") logger.log("-" * 200 + "\n")
logger.close() logger.close()
@ -187,7 +179,7 @@ if __name__ == "__main__":
parser.add_argument( parser.add_argument(
"--meta_lr", "--meta_lr",
type=float, type=float,
default=0.01, default=0.1,
help="The learning rate for the MAML optimizer (default is Adam)", help="The learning rate for the MAML optimizer (default is Adam)",
) )
parser.add_argument( parser.add_argument(
@ -202,7 +194,7 @@ if __name__ == "__main__":
parser.add_argument( parser.add_argument(
"--meta_batch", "--meta_batch",
type=int, type=int,
default=5, default=10,
help="The batch size for the meta-model", help="The batch size for the meta-model",
) )
parser.add_argument( parser.add_argument(
@ -223,7 +215,7 @@ if __name__ == "__main__":
if args.rand_seed is None or args.rand_seed < 0: if args.rand_seed is None or args.rand_seed < 0:
args.rand_seed = random.randint(1, 100000) args.rand_seed = random.randint(1, 100000)
assert args.save_dir is not None, "The save dir argument can not be None" assert args.save_dir is not None, "The save dir argument can not be None"
args.save_dir = "{:}-{:}-d{:}".format( args.save_dir = "{:}-s{:}-{:}-d{:}".format(
args.save_dir, args.env_version, args.hidden_dim args.save_dir, args.inner_step, args.env_version, args.hidden_dim
) )
main(args) main(args)