Repository navigation
Add DuHow Whisper turbo cat evaluations and runner support - #55
Merged
Merged
Conversation
jordimas
changed the title
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Registers
duhow/whisper-large-v3-turbo-catin the Whisper evaluator and default ASR runner so future runs can evaluate it. Resolves fully qualified Hugging Face model names directly, preserving the existing Whisper precision settings, and documents the evaluation command.Adds the existing evaluation results, with 400 samples per benchmark:
The runner skips completed results by default; use
--overwriteto recompute them.Validation: all three ASR data validation tests pass; ASR summary generation succeeds; formatting and diff checks pass. Mocked checks verify runner registration, repository resolution for DuHow/OpenAI/Aina models, and shared CUDA float16 precision. Existing results were validated; evaluations were not rerun.