Skip to content

Qualcomm AI Engine Direct - add seq_mse_candidates setting to SmolLM3 and fixed a bug in the graph drawer #14295

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
merged 1 commit into from
Sep 16, 2025
Merged

Qualcomm AI Engine Direct - add seq_mse_candidates setting to SmolLM3 and fixed a bug in the graph drawer #14295

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: 2 additions & 2 deletions backends/qualcomm/debugger/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -183,11 +183,11 @@ def draw(self):
cleanup=not self.dot_string,
)
source_file = os.path.join(temp_directory, f"{self.filename}.svg")
destination_file = os.path.join(".", f"{self.filename}.svg")
destination_file = os.path.join(self.directory, f"{self.filename}.svg")
shutil.move(source_file, destination_file)
if self.dot_string:
dot_file = os.path.join(temp_directory, f"{self.filename}")
dot_dest_file = os.path.join(".", f"{self.filename}.dot")
dot_dest_file = os.path.join(self.directory, f"{self.filename}.dot")
shutil.move(dot_file, dot_dest_file)


Expand Down
1 change: 1 addition & 0 deletions examples/qualcomm/oss_scripts/llama/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -501,6 +501,7 @@ class Smollm3_3B(LLMModelConfig):
ptq = QuantDtype.use_16a4w_block
group_size = 32
masked_softmax = True
seq_mse_candidates = 0
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What does 0 seq_mse_candidates mean

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for asking the question about seq_mse_candidates setting.
Setting seq_mse_candidates = 0 means that SeqMSE is disabled during quantization.
For more details, can refer:
https://github.com/pytorch/executorch/blob/main/examples/qualcomm/oss_scripts/llama/decoder_utils.py#L367-L375

r1 = False
r2 = False
r3 = False
Expand Down
Loading