SambaLingo-Arabic-Chat-70B is a human aligned chat model trained in Arabic and English. It is trained using direct preference optimization on top the base model SambaLingo-Arabic-Base-70B. The base model adapts Llama-2-70b to Arabic by training on 28 billion tokens from the Arabic split of the Cultura-X dataset. Try This Model at SambaLingo-chat-space.
To prompt this model, please use the following chat template:
<|user|>\n{question}</s>\n<|assistant|>\n
Example Prompts and Generations
<|user|>
كيف تصف الالوان لشخص ولد أعمى</s>
<|assistant|>
ليس لدي القدرة على الرؤية، ولكن يمكنني تقديم وصف للألوان بناءً على ما تعلمته من البيانات.\n\nبالنسبة لشخص ولد أعمى، قد يكون من الصعب وصف الألوان لأنها لم تختبرها من قبل. ومع ذلك، يمكننا استخدام الاستعارات والمقارنات لمساعدة هذا الشخص على فهم الألوان. على سبيل المثال، يمكننا وصف اللون الأحمر بأنه دافئ وحيوي، مثل لهب النار أو ثمرة الفراولة الناضجة. يمكننا وصف اللون الأزرق بأنه هادئ ومهدئ، مثل السماء أو المحيط. يمكننا وصف اللون الأخضر بأنه منعش وطبيعي، مثل أوراق الشجر أو العشب. يمكننا وصف اللون الأصفر بأنه مشرق ومبهج، مثل الشمس أو زهرة عباد الشمس. يمكننا وصف اللون الأرجواني بأنه ملكي وفاخر، مثل لون الجواهر أو غروب الشمس. في النهاية، قد يكون من الصعب وصف الألوان لشخص ولد أعمى، ولكن باستخدام الاستعارات والمقارنات، يمكننا مساعدتهم على فهم الفروق الدقيقة في الألوان بشكل أفضل.
Training Details
The alignment phase follows the recipe for Zephyr-7B, and comprises two stages: supervised fine-tuning (SFT) and Direct Performance Optimization (DPO).
The SFT phase was done on the ultrachat_200k dataset mixed with the Google translated version of the ultrachat_200k dataset. It was trained for one epoch with global batch size 512 and max sequence length 2048 tokens. We used a linear decay learning rate of 2e-5 and 10% warmup.
The DPO phase was done on the ultrafeedback dataset and cai-conversation-harmless dataset, mixed with 10% of the data Google translated. It was trained with global batch size 32 and for three epochs. We used a linear decay learning rate of 5e-7, 10% warmup and β=0.1 as the regularization factor for DPO.
Tokenizer Details
We extended the vocabulary of the base llama model from 32,000 tokens to 57,000 tokens by adding up to 25,000 non-overlapping tokens from the new language.
Use of this model is governed by the Meta’s Llama 2 Community License Agreement. Please review and accept the license before downloading the model weights.
Out-of-Scope Use
SambaLingo should NOT be used for:
Mission-critical applications
Applications that involve the safety of others
Making highly important decisions
Bias, Risks, and Limitations
Like all LLMs, SambaLingo has certain limitations:
Hallucination: Model may sometimes generate responses that contain plausible-sounding but factually incorrect or irrelevant information.
Code Switching: The model might unintentionally switch between languages or dialects within a single response, affecting the coherence and understandability of the output.
Repetition: The Model may produce repetitive phrases or sentences, leading to less engaging and informative responses.
Coding and Math: The model's performance in generating accurate code or solving complex mathematical problems may be limited.
Toxicity: The model could inadvertently generate responses containing inappropriate or harmful content.
Acknowledgments
We extend our heartfelt gratitude to the open-source AI community; this endeavor would not have been possible without open source. SambaNova embraces the open-source community and aspires to actively contribute to this initiative.
We would like to give a special thanks to the following groups:
Meta for open sourcing LLama 2 and open sourcing FLORES-200 dataset
Nguyen et al for open sourcing CulturaX dataset
CohereAI for releasing AYA-101 and open sourcing a multilingual instruction tuning dataset
EleutherAI for their open source evaluation framework
Hugging Face-H4 team for open source the zephyr training recipe and alignment handbook repo
Cite SambaLingo
@misc{csaki2024sambalingo,
title={SambaLingo: Teaching Large Language Models New Languages},
author={Zoltan Csaki and Bo Li and Jonathan Li and Qiantong Xu and Pian Pawakapan and Leon Zhang and Yun Du and Hengyu Zhao and Changran Hu and Urmish Thakker},
year={2024},
eprint={2404.05829},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
