{ "id": 369589, "title": "Efficiency solution 39th place ", "content": "

First of all, we would like to thank Kaggle team and all participants who shared their codes and ideas.

\n

Overview

\n

Our final submission is a single Deberta-base model with 1 fold and 6 epochs. We trained the model using 6 target columns and 12 new targets added by ourselves to help the model to learn intermediate insight on the essay quality.

\n

Preprocessing

\n

12 new targets that we added are

\n\n

Every scores are normalized between 1 and 5.

\n

Model

\n

Using this public notebook as a reference, we experimented a lot, changing hyperparameters.

\n

Code

\n

Code Link

\n

Team Members🤝

\n

@dianascopeo
\n@jccorvisier
\n@pabloskewesscopeo
\n@msanselme

", "comments": [ { "id": 2053670, "author_name": "Vitou Phy", "author_url": "", "post_date": "2022-12-03T13:33:46.950000", "content": "

That's an interesting approach to adding more labels. What would be the three top-impactful labels among those 12 new targets?

", "votes": 0, "replies": [] } ], "comment_count": 1, "messages": [ { "id": 2050221, "postDate": "2022-11-30T15:38:01.207Z", "content": "

First of all, we would like to thank Kaggle team and all participants who shared their codes and ideas.

\n

Overview

\n

Our final submission is a single Deberta-base model with 1 fold and 6 epochs. We trained the model using 6 target columns and 12 new targets added by ourselves to help the model to learn intermediate insight on the essay quality.

\n

Preprocessing

\n

12 new targets that we added are

\n\n

Every scores are normalized between 1 and 5.

\n

Model

\n

Using this public notebook as a reference, we experimented a lot, changing hyperparameters.

\n

Code

\n

Code Link

\n

Team Members🤝

\n

@dianascopeo
\n@jccorvisier
\n@pabloskewesscopeo
\n@msanselme

", "rawMarkdown": "**First of all, we would like to thank Kaggle team and all participants who shared their codes and ideas.**\n\n# **Overview**\n\nOur final submission is a single Deberta-base model with 1 fold and 6 epochs. We trained the model using 6 target columns and 12 new targets added by ourselves to help the model to learn intermediate insight on the essay quality.\n\n# **Preprocessing**\n\n*12 new targets that we added are*\n- Long words(count the number of the long words in the text which are the words that have more letters that the mean of the number of the letters per word on the dataset)\n- Number of grammar mistakes([using language tool](https://pypi.org/project/language-tool-python/))\n- Readability score ([Dale–Chall readability score](https://pypi.org/project/py-readability-metrics/))\n- Number of distinct words([using spacy library](https://pypi.org/project/spacy/))\n- Nouns count([using spacy library](https://pypi.org/project/spacy/))\n- Verbs count([using spacy library](https://pypi.org/project/spacy/))\n- Adjectives count([using spacy library](https://pypi.org/project/spacy/))\n- Adverbs count([using spacy library](https://pypi.org/project/spacy/))\n- Grammar mistakes related to words and sentences([web scraping](https://www.grammarbot.io/))\n- Grammar mistakes related to the punctuation([web scraping](https://www.grammarbot.io/))\n- Grammar mistakes related to the part of the speech([web scraping](https://www.grammarbot.io/))\n- Other grammar mistakes([web scraping](https://www.grammarbot.io/))\n\nEvery scores are normalized between 1 and 5. \n\n# **Model**\n\nUsing [this public notebook](https://www.kaggle.com/code/yasufuminakama/fb3-deberta-v3-base-baseline-train) as a reference, we experimented a lot, changing hyperparameters. \n\n# **Code**\n[Code Link](https://www.kaggle.com/code/msanselme/notebook1)\n\n# **Team Members**🤝\n@dianascopeo \n@jccorvisier \n@pabloskewesscopeo \n@msanselme ", "votes": 10 }, { "id": 2053670, "postDate": "2022-12-03T13:33:46.950Z", "content": "

That's an interesting approach to adding more labels. What would be the three top-impactful labels among those 12 new targets?

", "rawMarkdown": "That's an interesting approach to adding more labels. What would be the three top-impactful labels among those 12 new targets?" } ], "complete": true, "source": "live", "fetched_at": "2026-10-19T20:01:55Z", "competition_id": 38321, "competition_slug": "feedback-prize-english-language-learning", "team_id": 9099608, "solution_url": "/competitions/feedback-prize-english-language-learning/writeups/scopeo-efficiency-solution-39th-place" }