Skip to content

fix(akida2/speech_commands): no hard-coded data path in the data sanity check - #117

Merged
dmclellandbc merged 1 commit into
mainfrom
fix/HUB-16-speech-commands-data-path
Oct 21, 2026
Merged

dmclellandbc merged 1 commit into
mainfrom
fix/HUB-16-speech-commands-data-path

Conversation

@dmclellandbc

Copy link
Copy Markdown
Collaborator

Refs HUB-16

The sanity-check block at the bottom of akida2/model_zoo/speech_commands/speech_commands_data.py hard-coded a machine-specific dataset path, which check_content.py flagged as machine-path.

  • The block now takes -d / --data, defaulting to ./data/speech_commands, the same default as the example's train, eval and benchmark scripts.
  • It moves to the end of the module. It used to sit before get_data, so that function was defined after the script had already run.

No change to the training, eval or benchmark paths, so no models, metrics or READMEs change.

Checks

  • python .claude/skills/review-content/check_content.py: no findings (the machine path was the only one on main).
  • python speech_commands_data.py -d <tfds dir> run end to end on a local Speech Commands TFDS copy: prints a (100, 49, 10, 1) uint8 feature batch and (100,) labels.

🤖 Generated with Claude Code

https://claude.ai/code/session_01DnyD76hgk85WiSBvfGpN6w

…sanity check

The __main__ block of speech_commands_data.py hard-coded a machine-specific
dataset path. It now takes -d/--data, defaulting to ./data/speech_commands
like the example's other scripts, and moves to the end of the module.

Refs HUB-16

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DnyD76hgk85WiSBvfGpN6w
@dmclellandbc
dmclellandbc merged commit 779b43c into main Oct 21, 2026
8 checks passed
@dmclellandbc
dmclellandbc deleted the fix/HUB-16-speech-commands-data-path branch October 21, 2026 14:52
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant