The ASR models for sound prediction are usually pre trained on clean speech. However, the samples in this dataset appear to have a variety of noise added. Has anyone tried any denoising approaches to clean those out/ any pointers to some libraries?
Thank you so much for the information. It’s clear now.
Yes, those were my mistakes. I was not running it on colab at first hence the different format and may have lead to changes in the markdown.
Noted, I’ll keep that unchanged and submit.
I think if you’re using the submission as per the starter kit, the notebook is bundled together with the test.csv. So, you won’t need to submit that separately.
I tried submitting a notebook to the challenge, but it showed error suggesting a zip file was to be submitted. Uploading the zip file using the command in the starter kit also gave error. Could you please clarify the submission format? The submission section has very few details.
I tried a smaller model and this still fails in the
Generate Predictions On Test Data phase. The logs do not point to any particular code failure.
This time the validation step ran fine and it stopped during predictions on test data. https://www.aicrowd.com/challenges/ai-blitz-9/problems/nlp-feature-engineering/submissions/146420
Ideally though, as per my understanding if it worked on validation, it should’ve been same process on test data - right?
Thank you so much for looking into this and the fix. Trying out now.
I was running the colab for generating word vectors for the NLP feature engineering task where I get this error log: https://aicrowd-evaluation-logs.s3.us-west-002.backblazeb2.com/logs/desiml/blitz-9/feature-engineering/17724c17-2a58-4083-84d9-1c3c9766a802.log
It’s unclear to me what the issue is in this case. Any help would be very much appreciated. Thanks!
Try out the BERT based model from the notebook I just shared: https://www.aicrowd.com/showcase/bert-for-emotion-detection
Should get you above 0.8 I think. Let me know in case of any issues.
Autocorrect on sound files prediction Adding on the benchmark solution to include autocorrect on the predictions to push up on the LB :)
Solution for submission 146912 A detailed solution for submission 146912 submitted for challenge NLP Feature Engineering
Solution for submission 146452 A detailed solution for submission 146452 submitted for challenge Sound Prediction
BERT for research paper classification Fine tuning BERT pretrained on MLM for paper classification task
BERT for emotion detection Fine tuning BERT pretrained on MLM for emotion detection task