Skip to content

[Do not land] Static numeric experiment #15936

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

Open
wants to merge 3 commits into
base: main
Choose a base branch
from
+753 −157
Open

[Do not land] Static numeric experiment #15936

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
51fcda1
up
metascroy Nov 20, 2025
4266924
up
metascroy Nov 21, 2025
5c3efef
up
metascroy Nov 26, 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
290 changes: 206 additions & 84 deletions examples/apple/coreml/llama/export.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,13 +10,17 @@
import torch
from executorch.backends.apple.coreml.compiler import CoreMLBackend # pyre-ignore
from executorch.backends.apple.coreml.partition import CoreMLPartitioner # pyre-ignore
import os

from executorch.examples.apple.coreml.llama.llama_transformer import (
InputManager,
load_model,
load_model_in_pieces_ITO,
load_model_in_pieces_IAFO,
)
from executorch.examples.apple.coreml.llama.utils import (
replace_linear_with_split_linear,
SplitLinearModule,
)

from executorch.exir import to_edge_transform_and_lower
Expand All @@ -28,6 +32,7 @@

from torchao.quantization.granularity import PerAxis, PerGroup
from torchao.quantization.quant_api import IntxWeightOnlyConfig, quantize_
from torchao.prototype.quantization.codebook_coreml import CodebookWeightOnlyConfig
from torchao.utils import unwrap_tensor_subclass


Expand Down Expand Up @@ -77,7 +82,7 @@ def main() -> None:
parser.add_argument(
"--coreml-quantize",
default=None,
choices=["b4w", "c4w"],
choices=["b4w", "c4w", "custom",],
help="This option is only for coreml: Use coreml quantization, e.g. b4w (for blockwise 4 bit weight), c4w (for channelwise 4 bit weight)",
)
parser.add_argument(
Expand All @@ -102,67 +107,65 @@ def main() -> None:
type=str,
default="fp16",
)

export_args = parser.parse_args()
model = load_model(
export_args.checkpoint,
export_args.params,
max_seq_length=export_args.max_seq_length,
use_cache_list=export_args.use_cache_list,
parser.add_argument(
"--export_in_pieces_mode",
type=int,
default=0,
)

export_args = parser.parse_args()

float_dtype = {"fp16": torch.float16, "fp32": torch.float32}[
export_args.dtype
] # dtype for model/inputs

model.eval()
model.to(float_dtype)

if export_args.target_split_size is not None:
replace_linear_with_split_linear(
model,
out_target_split_size=export_args.target_split_size,
out_max_splits=export_args.max_splits,
# I have not found splitting on in_features to be beneficial,
# and it often leads to OOM so I set in_max_splits to 1
in_target_split_size=1,
in_max_splits=1,
)
def maybe_split_model(model):
if export_args.target_split_size is not None:
replace_linear_with_split_linear(
model,
out_target_split_size=export_args.target_split_size,
out_max_splits=export_args.max_splits,
# I have not found splitting on in_features to be beneficial,
# and it often leads to OOM so I set in_max_splits to 1
in_target_split_size=1,
in_max_splits=1,
)

def maybe_quantize_model(model):
if export_args.embedding_quantize:
bitwidth, group_size = export_args.embedding_quantize.split(",")
bitwidth = int(bitwidth)
assert bitwidth in [4, 8], "CoreML only supports 4-bit and 8-bit quantization"
group_size = int(group_size)
if group_size == 0:
granularity = PerAxis(0)
else:
granularity = PerGroup(group_size)
weight_dtype = getattr(torch, f"int{bitwidth}")

# Quantization
if export_args.embedding_quantize:
bitwidth, group_size = export_args.embedding_quantize.split(",")
bitwidth = int(bitwidth)
assert bitwidth in [4, 8], "CoreML only supports 4-bit and 8-bit quantization"
group_size = int(group_size)
if group_size == 0:
granularity = PerAxis(0)
else:
granularity = PerGroup(group_size)
weight_dtype = getattr(torch, f"int{bitwidth}")
quantize_(
model,
IntxWeightOnlyConfig(weight_dtype=weight_dtype, granularity=granularity),
lambda m, fqn: isinstance(m, torch.nn.Embedding),
)

quantize_(
model,
IntxWeightOnlyConfig(weight_dtype=weight_dtype, granularity=granularity),
lambda m, fqn: isinstance(m, torch.nn.Embedding),
)
if export_args.coreml_quantize == "b4w":
quantize_(
model,
IntxWeightOnlyConfig(
weight_dtype=torch.int4,
granularity=PerGroup(32),
),
)
elif export_args.coreml_quantize == "c4w":
quantize_(
model,
IntxWeightOnlyConfig(
weight_dtype=torch.int4,
granularity=PerAxis(0),
),
)

if export_args.coreml_quantize == "b4w":
quantize_(
model,
IntxWeightOnlyConfig(
weight_dtype=torch.int4,
granularity=PerGroup(32),
),
)
elif export_args.coreml_quantize == "c4w":
quantize_(
model,
IntxWeightOnlyConfig(
weight_dtype=torch.int4,
granularity=PerAxis(0),
),
)

compile_specs = CoreMLBackend.generate_compile_specs( # pyre-fixme[16]
minimum_deployment_target=ct.target.iOS18,
Expand All @@ -179,46 +182,165 @@ def main() -> None:
skip_ops_for_coreml_delegation=[],
)

input_manager = InputManager(
n_layers=model.params.n_layers,
max_batch_size=model.params.max_batch_size,
n_kv_heads=model.params.n_kv_heads,
max_seq_length=model.params.max_seq_len,
head_dim=model.params.head_dim,
use_cache_list=export_args.use_cache_list,
seq_length=export_args.seq_length,
dtype=float_dtype,
minus_infinity=-30000,
cache_size=export_args.cache_size,
executorch_config = ExecutorchBackendConfig(
extract_delegate_segments=True,
do_quant_fusion_and_const_prop=True,
memory_planning_pass=MemoryPlanningPass(alloc_graph_input=False),
sym_shape_eval_pass=ConstraintBasedSymShapeEvalPass(),
)
example_inputs = input_manager.get_inputs(tokens=[0])

model = unwrap_tensor_subclass(model)
def strip_pte(name):
if name.endswith(".pte"):
return name[:-4]
else:
return name

ep = torch.export.export(model, example_inputs, strict=True)
print("Exported program")
print(ep)

edge_manager = to_edge_transform_and_lower(
ep,
partitioner=[partitioner],
)
if export_args.export_in_pieces_mode == 0:
model = load_model(
export_args.checkpoint,
export_args.params,
max_seq_length=export_args.max_seq_length,
use_cache_list=export_args.use_cache_list,
)
input_manager = InputManager(
n_layers=model.params.n_layers,
max_batch_size=model.params.max_batch_size,
n_kv_heads=model.params.n_kv_heads,
max_seq_length=model.params.max_seq_len,
head_dim=model.params.head_dim,
use_cache_list=export_args.use_cache_list,
seq_length=export_args.seq_length,
dtype=float_dtype,
minus_infinity=-30000,
cache_size=export_args.cache_size,
)
example_inputs = input_manager.get_inputs(tokens=[0])
model.eval()
model = model.to(float_dtype)
print("Model", model)
maybe_split_model(model)
print("Model after split", model)
maybe_quantize_model(model)
print("Model after quantize", model)

print("Delegated program")
print(format_delegated_graph(edge_manager.exported_program().graph_module))
ep = torch.export.export(model, example_inputs, strict=True)
ep = ep.run_decompositions({})
print("Exported program")
print(ep)

executorch_program = edge_manager.to_executorch(
ExecutorchBackendConfig(
extract_delegate_segments=True,
do_quant_fusion_and_const_prop=True,
memory_planning_pass=MemoryPlanningPass(alloc_graph_input=False),
sym_shape_eval_pass=ConstraintBasedSymShapeEvalPass(),
edge_manager = to_edge_transform_and_lower(
ep,
partitioner=[partitioner],
)
)

filename = save_pte_program(executorch_program, export_args.output_name)
print(f"Saved Executorch program to local {filename}")
print("Delegated program")
print(format_delegated_graph(edge_manager.exported_program().graph_module))

executorch_program = edge_manager.to_executorch(executorch_config)
filename = save_pte_program(executorch_program, export_args.output_name)
print(f"Saved Executorch program to local {filename}")

else:
if export_args.export_in_pieces_mode == 1:
models, example_inputs = load_model_in_pieces_ITO(
export_args.checkpoint,
export_args.params,
max_seq_length=export_args.max_seq_length,
seq_length=export_args.seq_length,
float_dtype=float_dtype,
)

for i, model in enumerate(models):
if i == 0:
ex_inputs = example_inputs[i]
suffix = "input_block"
elif i == len(models) - 1:
ex_inputs = example_inputs[-1]
suffix = "output_block"
else:
ex_inputs = example_inputs[1]
suffix = f"transformer_block_{i-1}"

model.eval()
model = model.to(float_dtype)
print(f"Model {i}", model)
maybe_split_model(model)
print(f"Model {i} after split", model)
maybe_quantize_model(model)
print(f"Model {i} after quantize", model)
ep = torch.export.export(model, ex_inputs, strict=True)
ep = ep.run_decompositions({})
print(f"Exported program for model {i}", ep)

edge_manager = to_edge_transform_and_lower(
ep,
partitioner=[partitioner],
)

print(f"Delegated program for model {i}")
print(format_delegated_graph(edge_manager.exported_program().graph_module))

executorch_program = edge_manager.to_executorch(executorch_config)
os.makedirs(f"{strip_pte(export_args.output_name)}", exist_ok=True)
filename = save_pte_program(executorch_program, f"{strip_pte(export_args.output_name)}/{suffix}.pte")
print(f"Saved Executorch program to local {filename}")



elif export_args.export_in_pieces_mode == 2:
models, example_inputs = load_model_in_pieces_IAFO(
export_args.checkpoint,
export_args.params,
max_seq_length=export_args.max_seq_length,
seq_length=export_args.seq_length,
float_dtype=float_dtype,
)

for i, model in enumerate(models):
if i == 0:
ex_inputs = example_inputs[i]
suffix = "input_block"
elif i == len(models) - 1:
ex_inputs = example_inputs[-1]
suffix = "output_block"
else:
if i % 2 == 1:
# AttentionBlock
ex_inputs = example_inputs[1]
else:
# FeedForwardBlock
ex_inputs = example_inputs[2]
suffix = f"transformer_block_{i-1}"


model.eval()
model = model.to(float_dtype)
print(f"Model {i}", model)
maybe_split_model(model)
print(f"Model {i} after split", model)
maybe_quantize_model(model)
print(f"Model {i} after quantize", model)
ep = torch.export.export(model, ex_inputs, strict=True)
ep = ep.run_decompositions({})
print(f"Exported program for model {i}", ep)

edge_manager = to_edge_transform_and_lower(
ep,
partitioner=[partitioner],
)

print(f"Delegated program for model {i}")
print(format_delegated_graph(edge_manager.exported_program().graph_module))

executorch_program = edge_manager.to_executorch(executorch_config)
os.makedirs(f"{strip_pte(export_args.output_name)}", exist_ok=True)
filename = save_pte_program(executorch_program, f"{strip_pte(export_args.output_name)}/{suffix}.pte")
print(f"Saved Executorch program to local {filename}")

else:
raise ValueError(f"Unknown export_in_pieces_mode {export_args.export_in_pieces_mode}")


if __name__ == "__main__":
main() # pragma: no cover
Loading