--- license: llama2 datasets: - HuggingFaceH4/ultrachat_200k - HuggingFaceH4/ultrafeedback_binarized - HuggingFaceH4/cai-conversation-harmless language: - ar - en --- # SambaLingo-Arabic-Chat-70B 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](https://huggingface.co/sambanovasystems/SambaLingo-Arabic-Base-70B). The base model adapts [Llama-2-70b](https://huggingface.co/meta-llama/Llama-2-70b-hf) to Arabic by training on 28 billion tokens from the Arabic split of the [Cultura-X](https://huggingface.co/datasets/uonlp/CulturaX) dataset. Try This Model at [SambaLingo-chat-space](https://huggingface.co/spaces/sambanovasystems/SambaLingo-chat-space). ## Model Description - **Developed by:** [SambaNova Systems](https://sambanova.ai/) - **Model type:** Language Model - **Language(s):** Arabic, English - **Finetuned from model:** [Llama-2-70b](https://huggingface.co/meta-llama/Llama-2-70b-hf) - **Paper:** [SambaLingo: Teaching Large Language Models New Languages](https://arxiv.org/abs/2404.05829) - **Blog Post**: [sambalingo-open-source-language-experts](https://sambanova.ai/blog/sambalingo-open-source-language-experts) ## Getting Started ### Loading Model With Hugging Face Please make sure to set use_fast=False when loading the tokenizer. ```python from transformers import AutoModelForCausalLM, AutoTokenizer tokenizer = AutoTokenizer.from_pretrained("sambanovasystems/SambaLingo-Arabic-Chat-70B", use_fast=False) model = AutoModelForCausalLM.from_pretrained("sambanovasystems/SambaLingo-Arabic-Chat-70B", device_map="auto", torch_dtype="auto") ``` ### Interacting With Model Pipeline Please make sure to set use_fast=False when loading the tokenizer. ```python from transformers import pipeline pipe = pipeline("text-generation", model="sambanovasystems/SambaLingo-Arabic-Chat-70B", device_map="auto", use_fast=False) messages = [ {"role": "user", "content": {YOUR_QUESTION}}, ] prompt = pipe.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True) outputs = pipe(prompt)[0] outputs = outputs["generated_text"] ``` ### Suggested Inference Parameters - Temperature: 0.8 - Repetition penalty: 1.0 - Top-p: 0.9 ### Prompting Guidelines To prompt this model, please use the following chat template: ``` <|user|>\n{question}\n<|assistant|>\n ``` ### Example Prompts and Generations ``` <|user|> كملك، هل الاحسن أن تكون محبوب أو مخيف <|assistant|> لا أستطيع إبداء الرأي أو امتلاك المشاعر، لكن يمكنني تقديم إجابة موضوعية على هذا السؤال. في حين أن كونك محبوبًا يمكن أن يجلب السعادة والرضا، فإن كونك مخيفًا يمكن أن يكون له أيضًا فوائده. على سبيل المثال، يمكن أن يكون الشخص المخيف فعالاً في ردع السلوكيات أو المواقف غير المرغوب فيها، مثل التخريب أو التنمر. ومع ذلك، من المهم أن نتذكر أن الحب والقبول هما قيمتان مهمتان يجب السعي لتحقيقهما، وأن كونك محبوبًا لا ينبغي أن يكون الهدف النهائي. وبدلاً من ذلك، يجب أن نسعى جاهدين لنكون طيبين ورحيمين مع الآخرين، مع الاعتراف أيضًا بأن كل شخص لديه نقاط القوة والضعف الخاصة به. ``` ## Training Details The alignment phase follows the recipe for [Zephyr-7B](https://huggingface.co/HuggingFaceH4/zephyr-7b-beta), and comprises two stages: supervised fine-tuning (SFT) and Direct Performance Optimization (DPO). The SFT phase was done on the [ultrachat_200k](https://huggingface.co/datasets/HuggingFaceH4/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](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized) dataset and [cai-conversation-harmless](https://huggingface.co/datasets/HuggingFaceH4/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. ## Evaluation For evaluation results see our paper: [SambaLingo: Teaching Large Language Models New Languages](https://arxiv.org/abs/2404.05829) ## Uses ### Direct Use Use of this model is governed by the Meta’s [Llama 2 Community License Agreement](https://ai.meta.com/llama/license/). 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} } ```