| import gradio as gr | |
| import pandas as pd | |
| from css_html_js import custom_css | |
| TITLE = """<h1 align="center" id="space-title">🇲🇾 Malaysian Embedding Leaderboard</h1>""" | |
| INTRODUCTION_TEXT = """ | |
| 📐 The 🇲🇾 Malaysian Embedding Leaderboard aims to track, rank and evaluate Top-k retrieval using embedding models. All notebooks at https://github.com/mesolitica/embedding-benchmarks, feel free to submit your own score at https://huggingface.co./spaces/mesolitica/Malaysian-Embedding-Leaderboard/discussions with link to the notebook. | |
| ## Dataset | |
| 📈 We evaluate models based on 4 datasets, | |
| 1. CrossRef Melayu related DOI, https://huggingface.co./datasets/mesolitica/malaysian-ultrachat/resolve/main/ultrachat-crossref-melayu-malay.jsonl | |
| 2. Epenerbitan, https://huggingface.co./datasets/mesolitica/malaysian-ultrachat/resolve/main/ultrachat-epenerbitan-malay.jsonl | |
| 3. gov.my PDF files, https://huggingface.co./datasets/mesolitica/malaysian-ultrachat/resolve/main/ultrachat-gov.my.jsonl | |
| 4. lom.agc.gov.my PDF files, https://huggingface.co./datasets/mesolitica/malaysian-ultrachat/resolve/main/ultrachat-lom-agc.jsonl | |
| """ | |
| demo = gr.Blocks(css=custom_css) | |
| with demo: | |
| gr.HTML(TITLE) | |
| gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text") | |
| demo.launch() |