--- library_name: transformers license: cc-by-4.0 base_model: Qwen/Qwen3-4B-Base language: - af # Afrikaans - am # Amharic - ar # Arabic - en # English - fr # French - ha # Hausa - ig # Igbo - mg # Malagasy (Plateau) - ny # Nyanja - om # Oromo - pt # Portuguese - rw # Kinyarwanda - sn # Shona - so # Somali - st # Southern Sotho - sw # Swahili - ti # Tigrinya - tn # Tswana - xh # Xhosa - yo # Yoruba - zu # Zulu pipeline_tag: text-generation tags: - african-languages - multilingual - continued-pretraining - afrique-llm - qwen3 - qwen - llamafactory --- # AfriqueQwen-4B ## Model Overview **AfriqueQwen-4B** is part of the **AfriqueLLM** suite—a collection of open language models adapted to **20 African languages** through continued pre-training (CPT) on **~26B tokens**. This model is based on [Qwen/Qwen3-4B-Base](https://huggingface.co/Qwen/Qwen3-4B-Base) and has been specifically adapted for improved performance on African languages while maintaining strong capabilities in high-resource languages. Our experiments show that **Qwen 3 models achieve the best performance** among all base models tested, better preserving performance in high-resource languages after CPT and achieving strong results on long-context tasks such as document-level translation. ### Key Features - **Type**: Causal Language Model (Base/Pre-trained) - **Base Model**: Qwen 3 4B - **Parameters**: 4B - **Context Length**: 32,768 tokens (native) - **Training Tokens**: ~26B tokens of carefully curated multilingual data ## Supported Languages AfriqueQwen-4B has been adapted for the following 20 African languages plus 4 high-resource languages: | Language | Code | Family | Script | |----------|------|--------|--------| | Afrikaans | afr_Latn | Germanic | Latin | | Swahili | swh_Latn | Bantu | Latin | | Moroccan Arabic | ary_Arab | Semitic | Arabic | | Somali | som_Latn | Cushitic | Latin | | Amharic | amh_Ethi | Semitic | Ethiopic | | Egyptian Arabic | arz_Arab | Semitic | Arabic | | Hausa | hau_Latn | Chadic | Latin | | Kinyarwanda | kin_Latn | Bantu | Latin | | Zulu | zul_Latn | Bantu | Latin | | Igbo | ibo_Latn | Volta-Niger | Latin | | Plateau Malagasy | plt_Latn | Austronesian | Latin | | Xhosa | xho_Latn | Bantu | Latin | | Shona | sna_Latn | Bantu | Latin | | Yoruba | yor_Latn | Volta-Niger | Latin | | Nyanja | nya_Latn | Bantu | Latin | | Southern Sotho | sot_Latn | Bantu | Latin | | Tigrinya | tir_Ethi | Semitic | Ethiopic | | Tunisian Arabic | aeb_Arab | Semitic | Arabic | | Oromo | gaz_Latn | Cushitic | Latin | | Tswana | tsn_Latn | Bantu | Latin | **High-resource languages (for catastrophic forgetting mitigation):** English, French, Portuguese, Arabic ## Training Data Our training corpus combines multiple high-quality sources: - **African Monolingual Data** (~22.8B tokens): FineWeb2, WURA, and MADLAD-400 - **Code** (~1B tokens): CornStack-Python for reasoning capabilities - **Mathematics** (~1B tokens): FineMath-4+ for mathematical understanding - **Synthetic Data** (~324M tokens): GPT-4.1 translated domain-specific content across 10 domains We use **UniMax sampling** to create a balanced distribution, capping high-resource languages at approximately 1B tokens and upsampling lower-resource languages for up to five epochs. ## Quickstart ```python from transformers import AutoModelForCausalLM, AutoTokenizer model_name = "McGill-NLP/AfriqueQwen-4B" # Load the tokenizer and the model tokenizer = AutoTokenizer.from_pretrained(model_name) model = AutoModelForCausalLM.from_pretrained( model_name, torch_dtype="auto", device_map="auto" ) # Prepare the model input prompt = "Bawo ni o ṣe n ṣe?" # Yoruba: "How are you doing?" inputs = tokenizer(prompt, return_tensors="pt").to(model.device) # Generate text generated_ids = model.generate( **inputs, max_new_tokens=100, ) output = tokenizer.decode(generated_ids[0], skip_special_tokens=True) print(output) ``` ## Deployment For deployment, you can use `vllm` or `sglang` to create an OpenAI-compatible API endpoint: **vLLM:** ```shell vllm serve McGill-NLP/AfriqueQwen-4B ``` **SGLang:** ```shell python -m sglang.launch_server --model-path McGill-NLP/AfriqueQwen-4B ``` ## Training Details ### Hyperparameters - **Learning Rate**: 5e-5 (with warmup and cosine decay) - **Context Length**: 16,384 tokens - **Optimizer**: AdamW - **Precision**: BF16 mixed precision ### Infrastructure Training was conducted using the LLaMA-Factory framework on up to 64 NVIDIA H100 GPUs with: - DeepSpeed ZeRO-1/ZeRO-2 - Flash Attention 3 - Sequence packing - Liger Kernel optimizations ## Evaluation All AfriqueLLM models are evaluated on multiple multilingual benchmarks: | Model | AfriMGSM | AfriMMLU | AfriXNLI | Belebele | FLORES | INJONG | SIB-200 | Overall | Δ | |-------|----------|----------|----------|----------|--------|--------|---------|---------|---| | [Gemma3-4B](https://huggingface.co/google/gemma-3-4b-pt) | 10.24 | 33.89 | 37.76 | 45.79 | 35.36 | 55.52 | 63.59 | 40.31 | | | [AfriqueGemma-4B](https://huggingface.co/McGill-NLP/AfriqueGemma-4B) | 14.86 | 36.73 | 39.62 | 50.52 | 54.95 | 69.28 | 69.21 | 47.88 | +7.6 (18.8%) | | [Gemma3-12B](https://huggingface.co/google/gemma-3-12b-pt) | 25.21 | 48.76 | 44.01 | 68.84 | 44.09 | 73.53 | 79.17 | 54.80 | | | [AfriqueGemma-12B](https://huggingface.co/McGill-NLP/AfriqueGemma-12B) | 32.14 | 49.47 | 44.60 | 68.65 | 65.04 | 76.79 | 75.08 | 58.82 | +4.0 (7.3%) | | [Qwen3-4B](https://huggingface.co/Qwen/Qwen3-4B-Base) | 8.26 | 33.84 | 37.12 | 41.50 | 20.16 | 21.69 | 57.88 | 31.49 | | | [AfriqueQwen-4B](https://huggingface.co/McGill-NLP/AfriqueQwen-4B) | 33.09 | 43.04 | 44.88 | 63.62 | 59.82 | 65.34 | 74.77 | 54.94 | +23.4 (74.4%) | | [Qwen3.5-4B](https://huggingface.co/Qwen/Qwen3.5-4B-Base) | 20.79 | 38.63 | 40.36 | 55.82 | 32.06 | 59.43 | 74.96 | 46.01 | | | [AfriqueQwen3.5-4B](https://huggingface.co/McGill-NLP/AfriqueQwen3.5-4B) | 30.47 | 43.66 | 41.05 | 66.01 | 63.55 | 75.46 | 79.66 | 57.12 | +11.1 (24.2%) | | [AfriqueQwen3.5-4B-ExtendedCM](https://huggingface.co/McGill-NLP/AfriqueQwen3.5-4B-ExtendedCM) | 34.17 | 45.26 | 41.94 | 66.45 | 63.51 | 75.97 | 80.52 | 58.26 | +1.1 (2.0%) | | [Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B-Base) | 11.22 | 36.56 | 38.24 | 44.63 | 21.13 | 29.47 | 53.06 | 33.47 | | | [AfriqueQwen-8B](https://huggingface.co/McGill-NLP/AfriqueQwen-8B) | 39.68 | 46.91 | 45.99 | 68.46 | 62.18 | 73.36 | 77.00 | 59.08 | +25.6 (76.5%) | | [Qwen3-14B](https://huggingface.co/Qwen/Qwen3-14B-Base) | 16.60 | 39.66 | 43.22 | 50.74 | 23.75 | 41.80 | 66.29 | 40.29 | | | **[AfriqueQwen-14B](https://huggingface.co/McGill-NLP/AfriqueQwen-14B)** | **45.01** | **52.22** | **49.01** | **74.63** | **63.77** | **77.80** | **82.63** | **63.58** | **+23.3 (57.8%)** | | [Llama3.1-8B](https://huggingface.co/meta-llama/Llama-3.1-8B) | 8.14 | 32.27 | 37.90 | 40.95 | 26.69 | 41.37 | 59.99 | 35.33 | | | [AfriqueLlama-8B](https://huggingface.co/McGill-NLP/AfriqueLlama-8B) | 17.51 | 36.57 | 37.39 | 50.51 | 63.60 | 71.17 | 69.14 | 49.41 | +14.1 (39.9%) | | [Lugha-Llama-8B-wura](https://huggingface.co/Lugha/Meta-Llama-3.1-8B-wura) | 9.46 | 37.00 | 39.24 | 47.86 | 49.90 | 62.30 | 75.81 | 45.94 | | | [Gemma3-27B](https://huggingface.co/google/gemma-3-27b-pt) | 35.37 | 55.47 | 46.85 | 74.81 | 48.41 | 79.70 | 84.34 | 60.71 | | ## Model Variants - [AfriqueQwen-14B](https://huggingface.co/McGill-NLP/AfriqueQwen-14B) - Qwen-based 14B model (flagship) - [AfriqueQwen-8B](https://huggingface.co/McGill-NLP/AfriqueQwen-8B) - Qwen-based 8B model - [AfriqueQwen3.5-4B](https://huggingface.co/McGill-NLP/AfriqueQwen3.5-4B) - Qwen 3.5-based 4B model - [AfriqueQwen3.5-4B-ExtendedCM](https://huggingface.co/McGill-NLP/AfriqueQwen3.5-4B-ExtendedCM) - Qwen 3.5-based 4B model with extended continued pre-training - [AfriqueGemma-4B](https://huggingface.co/McGill-NLP/AfriqueGemma-4B) - Gemma-based 4B model - [AfriqueGemma-12B](https://huggingface.co/McGill-NLP/AfriqueGemma-12B) - Gemma-based 12B model - [AfriqueLlama-8B](https://huggingface.co/McGill-NLP/AfriqueLlama-8B) - Llama-based 8B model ## Citation If you find our work helpful, please cite: ```bibtex @misc{yu2026afriquellmdatamixingmodel, title={AfriqueLLM: How Data Mixing and Model Architecture Impact Continued Pre-training for African Languages}, author={Hao Yu and Tianyi Xu and Michael A. Hedderich and Wassim Hamidouche and Syed Waqas Zamir and David Ifeoluwa Adelani}, year={2026}, eprint={2601.06395}, archivePrefix={arXiv}, primaryClass={cs.CL}, url={https://arxiv.org/abs/2601.06395}, } ``` ## License This model is released under the [CC BY 4.0 License](https://creativecommons.org/licenses/by/4.0/). Please review the license terms before use. ## Acknowledgments We thank the creators of the base models, datasets and compute resources that made this work possible, including Mila, Compute Canada, Microsoft, the FineWeb team, WURA, MADLAD-400 and etc..