Base on the rules, it saids To be counted as correct, all characters in the decoded word must match the true word (including apostrophes for contractions like \"don't\"), however in the training data transcription, I see different version of Apostrophe, e.g.
\nkid’s -- watching movies is good for kid’s education
\n kid's -- what if you were in your kid's shoes
which one is counted as correct in the final testing dataset? Thanks for your clarification
", "messages": [ { "id": "3305374", "postDate": "10/22/2025 16:01:22", "content": "Base on the rules, it saids To be counted as correct, all characters in the decoded word must match the true word (including apostrophes for contractions like \"don't\"), however in the training data transcription, I see different version of Apostrophe, e.g.
\nkid’s -- watching movies is good for kid’s education
\n kid's -- what if you were in your kid's shoes
which one is counted as correct in the final testing dataset? Thanks for your clarification
", "rawMarkdown": "Base on the rules, it saids To be counted as correct, all characters in the decoded word must match the true word (including apostrophes for contractions like \"don't\"), however in the training data transcription, I see different version of Apostrophe, e.g.\n\n kid’s -- watching movies is good for kid’s education\n kid's -- what if you were in your kid's shoes\n\nwhich one is counted as correct in the final testing dataset? Thanks for your clarification", "votes": null }, { "id": "3305389", "postDate": "10/22/2025 16:28:50", "content": "I'm not quite sure what you mean, as the two examples you shared (\"kid's\") are the same type of possessive apostrophe. Apostrophes are included in general because they can change the phonemization of a word.
\nPerhaps it might be helpful to reference & try out the remove punctuation function from our repo and the example submission file for the val split to see what types of punctuation should be removed or kept in the final text.
", "rawMarkdown": "I'm not quite sure what you mean, as the two examples you shared (\"kid's\") are the same type of [possessive apostrophe](https://www.sussex.ac.uk/informatics/punctuation/apostrophe/possessives). Apostrophes are included in general because they can change the phonemization of a word.\n\nPerhaps it might be helpful to reference & try out the [remove punctuation](https://github.com/Neuroprosthetics-Lab/nejm-brain-to-text/blob/980a5b7e96392210edf976d76bdb5d15d2e0e62b/model_training/evaluate_model_helpers.py#L117) function from our repo and the [example submission file for the val split](https://github.com/Neuroprosthetics-Lab/nejm-brain-to-text/blob/980a5b7e96392210edf976d76bdb5d15d2e0e62b/model_training/rnn_baseline_submission_file_valsplit.csv) to see what types of punctuation should be removed or kept in the final text.", "votes": null } ], "comments": [ { "id": 3305389, "author_name": "notnickc", "author_url": "", "post_date": "10/22/2025 16:28:50", "content": "I'm not quite sure what you mean, as the two examples you shared (\"kid's\") are the same type of possessive apostrophe. Apostrophes are included in general because they can change the phonemization of a word.
\nPerhaps it might be helpful to reference & try out the remove punctuation function from our repo and the example submission file for the val split to see what types of punctuation should be removed or kept in the final text.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "3305374": "Base on the rules, it saids To be counted as correct, all characters in the decoded word must match the true word (including apostrophes for contractions like \"don't\"), however in the training data transcription, I see different version of Apostrophe, e.g.\n\n kid’s -- watching movies is good for kid’s education\n kid's -- what if you were in your kid's shoes\n\nwhich one is counted as correct in the final testing dataset? Thanks for your clarification", "3305389": "I'm not quite sure what you mean, as the two examples you shared (\"kid's\") are the same type of [possessive apostrophe](https://www.sussex.ac.uk/informatics/punctuation/apostrophe/possessives). Apostrophes are included in general because they can change the phonemization of a word.\n\nPerhaps it might be helpful to reference & try out the [remove punctuation](https://github.com/Neuroprosthetics-Lab/nejm-brain-to-text/blob/980a5b7e96392210edf976d76bdb5d15d2e0e62b/model_training/evaluate_model_helpers.py#L117) function from our repo and the [example submission file for the val split](https://github.com/Neuroprosthetics-Lab/nejm-brain-to-text/blob/980a5b7e96392210edf976d76bdb5d15d2e0e62b/model_training/rnn_baseline_submission_file_valsplit.csv) to see what types of punctuation should be removed or kept in the final text." }, "source": "meta" }