Skip to content

Arm backend: Increase qtol in Llama partial quant tests #17816

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
mansnils merged 1 commit into from
Mar 5, 2026
Merged

Arm backend: Increase qtol in Llama partial quant tests #17816

Show file tree
Hide file tree
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
4 changes: 4 additions & 0 deletions backends/arm/test/models/test_llama.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -209,6 +209,8 @@ def test_llama_tosa_INT_FP_partial_quant():
tosa_extensions=["FP"],
# Due to a few outliers, atol must be set high
atol=1.1,
# TODO(MLETORCH-1875): reduce tolerance
qtol=75,
frobenius_threshold=None,
cosine_threshold=None,
)
Expand All @@ -232,6 +234,8 @@ def test_llama_vgf_quant_partial_quant():
quantize=True,
# Due to a few outliers, atol must be set high
atol=1.1,
# TODO(MLETORCH-1875): reduce tolerance
qtol=75,
)
_use_partial_quantizer(pipeline)
pipeline.run()
2 changes: 2 additions & 0 deletions backends/arm/test/tester/analyze_output_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -318,6 +318,8 @@ def dump_error_output(
output_node = export_stage.artifact.graph_module.graph.output_node()
qp_input = get_input_quantization_params(export_stage.artifact)
qp_output = get_output_quantization_params(output_node)
scales = {k.name: v.scale for k, v in qp_output.items() if v is not None}
logger.error(f"Output Quant scales: {scales}")
logger.error(f"Input QuantArgs: {qp_input}")
logger.error(f"Output QuantArgs: {qp_output}")

Expand Down
6 changes: 3 additions & 3 deletions backends/arm/test/tester/arm_tester.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -995,9 +995,9 @@ def _compare_outputs(
stage_output,
reference_output,
quantization_scale=quantization_scale,
atol=1e-03,
rtol=1e-03,
qtol=0,
atol=atol,
rtol=rtol,
qtol=qtol,
)
raise e

Expand Down
Loading