Skip to content

fix: checkpoint memory consuming layers in Nemotron 12B model #391

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ffrujeri merged 2 commits into from
Sep 4, 2025
Merged

fix: checkpoint memory consuming layers in Nemotron 12B model #391

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
6d48105
Checkpoint normalization layers.
ffrujeri Aug 28, 2025
8c34b71
Fix activation checkpointing test.
ffrujeri Sep 4, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 12 additions & 0 deletions nemo_automodel/components/distributed/parallelizer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -146,6 +146,18 @@ def parallelize(
for i, layer in enumerate(layers):
if hasattr(layer, "mlp"):
layers[i].mlp = checkpoint_wrapper(layer.mlp)
if hasattr(layer, "self_attn"):
layers[i].self_attn = checkpoint_wrapper(layers[i].self_attn) # type: ignore

if hasattr(layer, "input_layernorm"):
layers[i].input_layernorm = checkpoint_wrapper(
layers[i].input_layernorm # type: ignore
)

if hasattr(layer, "post_attention_layernorm"):
layers[i].post_attention_layernorm = checkpoint_wrapper(
layers[i].post_attention_layernorm # type: ignore
)

# Set up mixed precision policy
if not mp_policy:
Expand Down
18 changes: 15 additions & 3 deletions tests/unit_tests/distributed/test_parallelization_strategies.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -287,9 +287,12 @@ def test_parallelize_with_activation_checkpointing(self, strategy, mock_device_m
"""Test parallelization with activation checkpointing enabled."""
mesh, dp_replicate_mesh, dp_shard_mesh, tp_mesh = mock_device_mesh

# Mock layers with MLP
# Mock layers with all the attributes that get checkpointed
mock_layer = MagicMock()
mock_layer.mlp = nn.Linear(10, 10)
mock_layer.self_attn = MagicMock()
mock_layer.input_layernorm = MagicMock()
mock_layer.post_attention_layernorm = MagicMock()
mock_distributed_env["extract_layers"].return_value = [mock_layer]

model = MockModel()
Expand All @@ -301,8 +304,17 @@ def test_parallelize_with_activation_checkpointing(self, strategy, mock_device_m
activation_checkpointing=True,
)

# Should apply checkpoint wrapper to MLP layers
mock_distributed_env["checkpoint_wrapper"].assert_called_with(mock_layer.mlp)
# Should apply checkpoint wrapper to all expected layer components
checkpoint_wrapper_mock = mock_distributed_env["checkpoint_wrapper"]

# Check that checkpoint_wrapper was called with all expected attributes
expected_calls = [
call(mock_layer.mlp),
call(mock_layer.self_attn),
call(mock_layer.input_layernorm),
call(mock_layer.post_attention_layernorm),
]
checkpoint_wrapper_mock.assert_has_calls(expected_calls, any_order=False)

def test_parallelize_with_custom_mesh_names(self, strategy, mock_device_mesh, mock_distributed_env):
"""Test parallelization with custom mesh names."""
Expand Down
Loading