Rate this Page
★ ★ ★ ★ ★
recipes/compiling_optimizer_lr_scheduler

Note

Go to the end to download the full example code.

(beta) Running the compiled optimizer with an LR Scheduler#

Created On: May 21, 2024 | Last Updated: May 21, 2024 | Last Verified: Nov 05, 2024

Author: Michael Lazos

The optimizer is a key algorithm for training any deep learning model. In this example, we will show how to pair the optimizer, which has been compiled using torch.compile, with the LR schedulers to accelerate training convergence.

Note

This tutorial requires PyTorch 2.3.0 or later.

Model Setup#

For this example, we’ll use a simple sequence of linear layers.

import torch

# Create simple model
model = torch.nn.Sequential(
    *[torch.nn.Linear(1024, 1024, False, device="cuda") for _ in range(10)]
)
input = torch.rand(1024, device="cuda")

# run forward pass
output = model(input)

# run backward to populate the grads for our optimizer below
output.sum().backward()

Setting up and running the compiled optimizer with LR Scheduler#

In this section, we’ll use the Adam optimizer with LinearLR Scheduler and create a helper function to wrap the step() call for each of them in torch.compile().

Note

torch.compile is only supported on CUDA devices that have a compute capability of 7.0 or higher.

# exit cleanly if we are on a device that doesn't support ``torch.compile``
if torch.cuda.get_device_capability() < (7, 0):
    print("Exiting because torch.compile is not supported on this device.")
    import sys
    sys.exit(0)

# !!! IMPORTANT !!! Wrap the lr in a Tensor if we are pairing the
# the optimizer with an LR Scheduler.
# Without this, torch.compile will recompile as the value of the LR
# changes.
opt = torch.optim.Adam(model.parameters(), lr=torch.tensor(0.01))
sched = torch.optim.lr_scheduler.LinearLR(opt, total_iters=5)

@torch.compile(fullgraph=False)
def fn():
    opt.step()
    sched.step()


# Warmup runs to compile the function
for _ in range(5):
    fn()
    print(opt.param_groups[0]["lr"])

tensor(0.0047)
tensor(0.0060)
tensor(0.0073)
tensor(0.0087)
tensor(0.0100)

Extension: What happens with a non-tensor LR?#

For the curious, we will show how to peek into what happens with torch.compile when we don’t wrap the LR in a tensor.

# No longer wrap the LR in a tensor here
opt = torch.optim.Adam(model.parameters(), lr=0.01)
sched = torch.optim.lr_scheduler.LinearLR(opt, total_iters=5)

@torch.compile(fullgraph=False)
def fn():
    opt.step()
    sched.step()

# Setup logging to view recompiles
torch._logging.set_logs(recompiles=True)

# Warmup runs to compile the function
# We will now recompile on each iteration
# as the value of the lr is mutated.
for _ in range(5):
    fn()

V0820 23:52:59.793000 36401 torch/_dynamo/guards.py:3508] [1/1] [__recompiles] Recompiling function wrapper in /usr/local/lib/python3.10/dist-packages/torch/optim/optimizer.py:496
V0820 23:52:59.793000 36401 torch/_dynamo/guards.py:3508] [1/1] [__recompiles]     triggered by the following guard failure(s):
V0820 23:52:59.793000 36401 torch/_dynamo/guards.py:3508] [1/1] [__recompiles]     - 1/0: Cache line invalidated because L['args'][0] got deallocated
V0820 23:52:59.830000 36401 torch/_dynamo/guards.py:3508] [2/1] [__recompiles] Recompiling function step in /usr/local/lib/python3.10/dist-packages/torch/optim/adam.py:213
V0820 23:52:59.830000 36401 torch/_dynamo/guards.py:3508] [2/1] [__recompiles]     triggered by the following guard failure(s):
V0820 23:52:59.830000 36401 torch/_dynamo/guards.py:3508] [2/1] [__recompiles]     - 2/0: Cache line invalidated because L['self'] got deallocated
V0820 23:53:02.624000 36401 torch/_dynamo/guards.py:3508] [2/2] [__recompiles] Recompiling function step in /usr/local/lib/python3.10/dist-packages/torch/optim/adam.py:213
V0820 23:53:02.624000 36401 torch/_dynamo/guards.py:3508] [2/2] [__recompiles]     triggered by the following guard failure(s):
V0820 23:53:02.624000 36401 torch/_dynamo/guards.py:3508] [2/2] [__recompiles]     - 2/1: ___as_tensor(self.param_groups[0]['lr']).item() == 0.003333333333333333  # (unknown source ___as_tensor(self.param_groups[0]['lr']).item(), please file a bug)
V0820 23:53:02.624000 36401 torch/_dynamo/guards.py:3508] [2/2] [__recompiles]     - 2/0: Cache line invalidated because L['self'] got deallocated
V0820 23:53:04.870000 36401 torch/_dynamo/guards.py:3508] [2/3] [__recompiles] Recompiling function step in /usr/local/lib/python3.10/dist-packages/torch/optim/adam.py:213
V0820 23:53:04.870000 36401 torch/_dynamo/guards.py:3508] [2/3] [__recompiles]     triggered by the following guard failure(s):
V0820 23:53:04.870000 36401 torch/_dynamo/guards.py:3508] [2/3] [__recompiles]     - 2/2: ___as_tensor(self.param_groups[0]['lr']).item() == 0.004666666666666667  # (unknown source ___as_tensor(self.param_groups[0]['lr']).item(), please file a bug)
V0820 23:53:04.870000 36401 torch/_dynamo/guards.py:3508] [2/3] [__recompiles]     - 2/1: ___as_tensor(self.param_groups[0]['lr']).item() == 0.003333333333333333  # (unknown source ___as_tensor(self.param_groups[0]['lr']).item(), please file a bug)
V0820 23:53:04.870000 36401 torch/_dynamo/guards.py:3508] [2/3] [__recompiles]     - 2/0: Cache line invalidated because L['self'] got deallocated
V0820 23:53:07.357000 36401 torch/_dynamo/guards.py:3508] [2/4] [__recompiles] Recompiling function step in /usr/local/lib/python3.10/dist-packages/torch/optim/adam.py:213
V0820 23:53:07.357000 36401 torch/_dynamo/guards.py:3508] [2/4] [__recompiles]     triggered by the following guard failure(s):
V0820 23:53:07.357000 36401 torch/_dynamo/guards.py:3508] [2/4] [__recompiles]     - 2/3: ___as_tensor(self.param_groups[0]['lr']).item() == 0.006000000000000001  # (unknown source ___as_tensor(self.param_groups[0]['lr']).item(), please file a bug)
V0820 23:53:07.357000 36401 torch/_dynamo/guards.py:3508] [2/4] [__recompiles]     - 2/2: ___as_tensor(self.param_groups[0]['lr']).item() == 0.004666666666666667  # (unknown source ___as_tensor(self.param_groups[0]['lr']).item(), please file a bug)
V0820 23:53:07.357000 36401 torch/_dynamo/guards.py:3508] [2/4] [__recompiles]     - 2/1: ___as_tensor(self.param_groups[0]['lr']).item() == 0.003333333333333333  # (unknown source ___as_tensor(self.param_groups[0]['lr']).item(), please file a bug)
V0820 23:53:07.357000 36401 torch/_dynamo/guards.py:3508] [2/4] [__recompiles]     - 2/0: Cache line invalidated because L['self'] got deallocated
V0820 23:53:09.609000 36401 torch/_dynamo/guards.py:3508] [2/5] [__recompiles] Recompiling function step in /usr/local/lib/python3.10/dist-packages/torch/optim/adam.py:213
V0820 23:53:09.609000 36401 torch/_dynamo/guards.py:3508] [2/5] [__recompiles]     triggered by the following guard failure(s):
V0820 23:53:09.609000 36401 torch/_dynamo/guards.py:3508] [2/5] [__recompiles]     - 2/4: ___as_tensor(self.param_groups[0]['lr']).item() == 0.007333333333333335  # (unknown source ___as_tensor(self.param_groups[0]['lr']).item(), please file a bug)
V0820 23:53:09.609000 36401 torch/_dynamo/guards.py:3508] [2/5] [__recompiles]     - 2/3: ___as_tensor(self.param_groups[0]['lr']).item() == 0.006000000000000001  # (unknown source ___as_tensor(self.param_groups[0]['lr']).item(), please file a bug)
V0820 23:53:09.609000 36401 torch/_dynamo/guards.py:3508] [2/5] [__recompiles]     - 2/2: ___as_tensor(self.param_groups[0]['lr']).item() == 0.004666666666666667  # (unknown source ___as_tensor(self.param_groups[0]['lr']).item(), please file a bug)
V0820 23:53:09.609000 36401 torch/_dynamo/guards.py:3508] [2/5] [__recompiles]     - 2/1: ___as_tensor(self.param_groups[0]['lr']).item() == 0.003333333333333333  # (unknown source ___as_tensor(self.param_groups[0]['lr']).item(), please file a bug)
V0820 23:53:09.609000 36401 torch/_dynamo/guards.py:3508] [2/5] [__recompiles]     - 2/0: Cache line invalidated because L['self'] got deallocated

With this example, we can see that we recompile the optimizer a few times due to the guard failure on the lr in param_groups[0].

Conclusion#

In this tutorial we showed how to pair the optimizer compiled with torch.compile with an LR Scheduler to accelerate training convergence. We used a model consisting of a simple sequence of linear layers with the Adam optimizer paired with a LinearLR scheduler to demonstrate the LR changing across iterations.

See also:

Total running time of the script: (0 minutes 16.937 seconds)

On this page
PyTorch Libraries