Update README.md

README.md

@@ -254,3 +254,67 @@ configs:
   - split: train
     path: multiple_choice_sentiment_task/train-*
 ---
+# AlGhafa Arabic LLM Benchmark
+
+### New fix: Normalized whitespace characters and ensured consistency across all datasets for improved data quality and compatibility.
+
+Multiple-choice evaluation benchmark for zero- and few-shot evaluation of Arabic LLMs, we adapt the following tasks:
+
+- Belebele Ar MSA [Bandarkar et al. (2023)](https://arxiv.org/abs/2308.16884): 900 entries
+- Belebele Ar Dialects [Bandarkar et al. (2023)](https://arxiv.org/abs/2308.16884): 5400 entries
+- COPA Ar: 89 entries machine-translated from English [COPA](https://people.ict.usc.edu/~gordon/copa.html) and verified by native Arabic speakers.
+- Facts balanced (based on AraFacts) [Sheikh Ali et al. (2021)](https://aclanthology.org/2021.wanlp-1.26): 80 entries (after balancing dataset), consisting of a short article and a corresponding claim, to be deemed true or false
+- MCQ Exams Ar [Hardalov et al. (2020)](https://aclanthology.org/2020.emnlp-main.438): 2248 entries
+- OpenbookQA Ar: 336 entries. Machine-translated from English [OpenbookQA](https://api.semanticscholar.org/CorpusID:52183757) and verified native Arabic speakers.
+- Rating sentiment (HARD-Arabic-Dataset) [Elnagar et al. (2018)](https://link.springer.com/chapter/10.1007/978-3-319-67056-0_3): determine the sentiment
+  of reviews, with 3 possible categories (positive, neutral, negative) transformed to a review score (1-5) as follows: 1-2 negative, 3 neutral, 4-5 positive; 6000 entries (2000 for each of the three classes)
+- Rating sentiment no neutral (HARD-Arabic-Dataset) [Elnagar et al., 2018](https://link.springer.com/chapter/10.1007/978-3-319-67056-0_3): 8000 entries in which we remove the neutral class by extending the positive class (corresponding to scores 1-3); 8000 entries (4000 for each class)
+- Sentiment [Abu Farha et al., 2021](https://aclanthology.org/2021.wanlp-1.36): 1725 entries based on Twitter posts, that can be classified as positive, negative, or neutral
+- SOQAL [Mozannar et al., 2019](https://aclanthology.org/W19-4612): grounded statement task to assess in-context reading comprehension, consisting of a context and a related question; consists of 155 entries with one original correct answer, transformed to multiple choice task by adding four possible
+  human-curated incorrect choices per sample
+- XGLUE (based on XGLUE-MLQA) [Liang et al., 2020](https://arxiv.org/abs/2004.01401); [Lewis et al., 2019](https://arxiv.org/abs/1910.07475): consists of
+  155 entries transformed to a multiple choice task by adding four human-curated incorrect choices per sample
+
+
+## Citing the AlGhafa benchmark:
+
+```bibtex
+@inproceedings{almazrouei-etal-2023-alghafa,
+    title = "{A}l{G}hafa Evaluation Benchmark for {A}rabic Language Models",
+    author = "Almazrouei, Ebtesam  and
+      Cojocaru, Ruxandra  and
+      Baldo, Michele  and
+      Malartic, Quentin  and
+      Alobeidli, Hamza  and
+      Mazzotta, Daniele  and
+      Penedo, Guilherme  and
+      Campesan, Giulia  and
+      Farooq, Mugariya  and
+      Alhammadi, Maitha  and
+      Launay, Julien  and
+      Noune, Badreddine",
+    editor = "Sawaf, Hassan  and
+      El-Beltagy, Samhaa  and
+      Zaghouani, Wajdi  and
+      Magdy, Walid  and
+      Abdelali, Ahmed  and
+      Tomeh, Nadi  and
+      Abu Farha, Ibrahim  and
+      Habash, Nizar  and
+      Khalifa, Salam  and
+      Keleg, Amr  and
+      Haddad, Hatem  and
+      Zitouni, Imed  and
+      Mrini, Khalil  and
+      Almatham, Rawan",
+    booktitle = "Proceedings of ArabicNLP 2023",
+    month = dec,
+    year = "2023",
+    address = "Singapore (Hybrid)",
+    publisher = "Association for Computational Linguistics",
+    url = "https://aclanthology.org/2023.arabicnlp-1.21",
+    doi = "10.18653/v1/2023.arabicnlp-1.21",
+    pages = "244--275",
+    abstract = "Recent advances in the space of Arabic large language models have opened up a wealth of potential practical applications. From optimal training strategies, large scale data acquisition and continuously increasing NLP resources, the Arabic LLM landscape has improved in a very short span of time, despite being plagued by training data scarcity and limited evaluation resources compared to English. In line with contributing towards this ever-growing field, we introduce AlGhafa, a new multiple-choice evaluation benchmark for Arabic LLMs. For showcasing purposes, we train a new suite of models, including a 14 billion parameter model, the largest monolingual Arabic decoder-only model to date. We use a collection of publicly available datasets, as well as a newly introduced HandMade dataset consisting of 8 billion tokens. Finally, we explore the quantitative and qualitative toxicity of several Arabic models, comparing our models to existing public Arabic LLMs.",
+}
+```