---
thumbnail: https://github.com/SauravMaheshkar/chaii-Hindi-Tamil-QA/blob/main/assets/Coffee%20Banner.png?raw=true
tags:
- kaggle
- rembert
- pytorch
- question-answering
language: multilingual
license: cc0-1.0
inference: false
datasets:
- Commonlit-Readibility

---
<div align = "center">
  <img src = "https://github.com/SauravMaheshkar/chaii-Hindi-Tamil-QA/blob/main/assets/Coffee%20Banner.png?raw=true">
</div>

This dataset contains the [**google/rembert**](https://huggingface.co/transformers/model_doc/rembert.html) model weights according to my team's experimentation strategy during the [**chaii - Hindi and Tamil Question Answering**](https://www.kaggle.com/c/chaii-hindi-and-tamil-question-answering) competition. They are listed below with their corresponding public LB score:-

| Huggingface Hub Link | Public LB Score |
| :---: | :---: |
| [**SauravMaheshkar/rembert-maxseq-400-docstride-128-chaii**](https://huggingface.co/SauravMaheshkar/rembert-maxseq-400-docstride-128-chaii) | 0.724 |
| [**SauravMaheshkar/rembert-maxseq-384-docstride-135-chaii**](https://huggingface.co/SauravMaheshkar/rembert-maxseq-384-docstride-135-chaii) | 0.723 |
| [**SauravMaheshkar/rembert-maxseq-400-docstride-135-chaii**](https://huggingface.co/SauravMaheshkar/rembert-maxseq-400-docstride-135-chaii) | 0.737 | 
| [**SauravMaheshkar/rembert-maxseq-384-docstride-128-chaii**](https://huggingface.co/SauravMaheshkar/rembert-maxseq-384-docstride-128-chaii) | 0.725 |