Skip to content

Unbreak test models llama CI #6026

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

Closed
wants to merge 1 commit into from
Closed

Unbreak test models llama CI #6026

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
24 changes: 13 additions & 11 deletions examples/xnnpack/aot_compiler.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,8 +15,12 @@
import torch
from executorch.backends.xnnpack.partition.xnnpack_partitioner import XnnpackPartitioner
from executorch.devtools import generate_etrecord
from executorch.exir import EdgeCompileConfig, ExecutorchBackendConfig
from executorch.extension.export_util.utils import export_to_edge, save_pte_program
from executorch.exir import (
EdgeCompileConfig,
ExecutorchBackendConfig,
to_edge_transform_and_lower,
)
from executorch.extension.export_util.utils import save_pte_program

from ..models import MODEL_NAME_TO_MODEL
from ..models.model_factory import EagerModelFactory
Expand Down Expand Up @@ -81,29 +85,27 @@

model = model.eval()
# pre-autograd export. eventually this will become torch.export
model = torch.export.export_for_training(model, example_inputs).module()
ep = torch.export.export_for_training(model, example_inputs)
model = ep.module()

if args.quantize:
logging.info("Quantizing Model...")
# TODO(T165162973): This pass shall eventually be folded into quantizer
model = quantize(model, example_inputs)

edge = export_to_edge(
model,
example_inputs,
edge_compile_config=EdgeCompileConfig(
edge = to_edge_transform_and_lower(
ep,
partitioner=[XnnpackPartitioner()],
compile_config=EdgeCompileConfig(
_check_ir_validity=False if args.quantize else True,
_skip_dim_order=True, # TODO(T182187531): enable dim order in xnnpack
),
)
logging.info(f"Exported graph:\n{edge.exported_program().graph}")
logging.info(f"Exported and lowered graph:\n{edge.exported_program().graph}")

# this is needed for the ETRecord as lowering modifies the graph in-place
edge_copy = copy.deepcopy(edge)

edge = edge.to_backend(XnnpackPartitioner())
logging.info(f"Lowered graph:\n{edge.exported_program().graph}")

exec_prog = edge.to_executorch(
config=ExecutorchBackendConfig(extract_delegate_segments=False)
)
Expand Down
Loading