[DRAFT] Nemotron ASR Support #1972
Conversation
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
| model = og.Model(config) | ||
|
|
||
| # Load SentencePiece tokenizer for decoding | ||
| sp_tokenizer = load_sentencepiece_tokenizer(model_path) |
Check notice
Code scanning / CodeQL
Unused local variable Note
Show autofix suggestion
Hide autofix suggestion
Copilot Autofix
AI 1 day ago
To fix the problem, remove the unused local variable while preserving any side effects from calling load_sentencepiece_tokenizer. Since the function is already called solely for its side effects (warning messages, optional loading), and its result is not needed in transcribe_file, the simplest fix is to call the function without capturing its return value.
Specifically, within examples/python/nemotron_streaming.py in transcribe_file, replace the line:
sp_tokenizer = load_sentencepiece_tokenizer(model_path)with:
load_sentencepiece_tokenizer(model_path)This eliminates the unused variable while keeping the behavior of invoking the tokenizer loader. No new imports, methods, or other definitions are required.
| @@ -93,7 +93,7 @@ | ||
| model = og.Model(config) | ||
|
|
||
| # Load SentencePiece tokenizer for decoding | ||
| sp_tokenizer = load_sentencepiece_tokenizer(model_path) | ||
| load_sentencepiece_tokenizer(model_path) | ||
|
|
||
| # Create StreamingASR instance — handles encoder cache + RNNT decode internally | ||
| asr = og.StreamingASR(model) |
1 participant
No description provided.