Repository navigation
Expand Parakeet buckets - #818
Merged
Merged
Conversation
…ndling Signed-off-by: jakmro <kubamroz124@gmail.com>
Signed-off-by: jakmro <kubamroz124@gmail.com>
jakmro
force-pushed
the
parakeet-fibonacci-buckets
branch
from
September 25, 2026 00:13
d801778 to
98bcc1f
Compare
Contributor
There was a problem hiding this comment.
Copilot review overview
🟡 Changes recommended
The expanded capture bucket may cause prohibitive memory use or OOM during conversion.
Get a fresh assessment by requesting another Copilot review.
Review effort: Lite
Findings: 1
Open (2)
What changed in this PR
Expands Parakeet TDT audio bucket and capture support and registers the v2 model.
Changes:
- Adds buckets through 377 seconds and a 420-second capture helper.
- Updates runtime generation and related tests for 42,000 frames.
- Registers
nvidia/parakeet-tdt-0.6b-v2.
| File | Description |
|---|---|
python/cactus/transpiler/ModelProfiles/profiles.py |
Registers the v2 model profile. |
python/cactus/transpiler/Converter/input_processor.py |
Registers the v2 processor. |
python/cactus/transpile/tdt_runtime.py |
Uses the expanded capture capacity. |
python/cactus/transpile/audio_preprocess.py |
Defines expanded buckets and capture sizing. |
python/cactus/convert/tests/test_parakeet_capture.py |
Updates capture expectations. |
python/cactus/convert/tests/test_audio_buckets.py |
Tests expanded buckets and capture behavior. |
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
Comment on lines
+24
to
+25
| _DEFAULT_AUDIO_BUCKET_SECONDS = (1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, 300, 377) | ||
| _AUDIO_CAPTURE_SECONDS = 420 |
| features = named_tensors["input_features"] | ||
| input_features = torch.randn( | ||
| (int(features.shape[0]), 3000, int(features.shape[2])), | ||
| (int(features.shape[0]), audio_capture_frames(), int(features.shape[2])), |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This pull request expands the supported audio bucket sizes and capture window for the Parakeet TDT model and ensures that the new defaults are consistently used throughout the codebase and tests. It also adds support for the
nvidia/parakeet-tdt-0.6b-v2model in both the processor and model profile registries.Audio bucket and capture window updates:
_DEFAULT_AUDIO_BUCKET_SECONDSare expanded to include larger values, and a new constant_AUDIO_CAPTURE_SECONDSis introduced and set to 420 seconds. The functionaudio_capture_frames()is added to compute the corresponding frame count. (python/cactus/transpile/audio_preprocess.py) [1] [2]audio_capture_frames()value (42000 frames), and tests are updated to expect the expanded bucket list. (python/cactus/transpile/tdt_runtime.py,python/cactus/convert/tests/test_audio_buckets.py,python/cactus/convert/tests/test_parakeet_capture.py) [1] [2] [3] [4]Model and processor registry updates:
nvidia/parakeet-tdt-0.6b-v2model is now supported in both the processor registry and the model profile registry, ensuring it uses the correct processing and profile logic. (python/cactus/transpiler/Converter/input_processor.py,python/cactus/transpiler/ModelProfiles/profiles.py) [1] [2]