The pretraining of state-of-the-art LLMs now requires trillions of words of text, which is orders of magnitude more than available for the vast majority of languages. While including text in more than one language is an obvious way to acquire more pretraining data, multilinguality is often seen as a curse, and most model training efforts continue to focus near-exclusively on individual large languages. We believe that multilinguality can be a blessing and that it should be possible to substantially improve over the capabilities of monolingual models for small languages through multilingual training. In this study, we introduce Poro 34B, a 34 billion parameter model trained for 1 trillion tokens of Finnish, English, and programming languages, and demonstrate that a multilingual training approach can produce a model that not only substantially advances over the capabilities of existing models for Finnish, but also excels in translation and is competitive in its class in generating English and programming languages. We release the model parameters, scripts, and data under open licenses at https://huggingface.co/LumiOpen/Poro-34B.
Poro~34B is a multilingual generative language model with 34 billion parameters designed to enhance Finnish language processing and achieve competitive performance in English and programming languages.
The model was trained on a diverse pretraining corpus of 1 trillion tokens, including Finnish, English, programming languages, and Finnish-English translation pairs to facilitate cross-lingual capabilities.
Poro~34B leverages a decoder-only architecture and was evaluated on tasks across Finnish, English, and code generation, showcasing superior performance in Finnish and competitive capabilities in other languages.
The study highlights the benefits of multilingual training for languages with limited resources and suggests future research directions for scaling this approach and further exploring its advantages in translation and generative tasks.
The development of large-scale generative language models has increasingly faced the challenge of data scarcity, particularly for languages other than English. This study introduces Poro~34B, a model that leverages multilingual training to not only enhance capabilities in Finnish - a relatively low-resource language in the context of LLMs - but also to demonstrate competitiveness in English and programming language tasks. By training across Finnish, English, and various programming languages, Poro~34B addresses both the theoretical and practical aspects of multilingual training, challenging the prevailing view of multilinguality as detrimental to performance in specific languages.
The pretraining corpus for Poro~34B spanned 1 trillion tokens, comprising Finnish, English, and programming language data, with a specific emphasis on including high-quality, deduplicated, and filtered datasets to maximize the model's learning potential. The Finnish data, comprising 32 billion tokens, was sourced from a combination of web crawls and various curated datasets. English and programming languages accounted for the majority of the training data, with the inclusion of code aiming to provide syntactic diversity. An explicit cross-lingual signal was incorporated through Finnish-English translation pairs, facilitating the model’s capabilities in both languages and translation tasks.
The Poro~34B model, with its 34 billion parameters, employs a decoder-only architecture mirroring advancements in generative model design. The training progressed over 1 trillion tokens, exceeding the current recommendations for compute-optimal training, underscoring a strategic choice for enhancing inference efficiency. The tokenizer, custom-designed for the multilingual corpus, aimed at low fertility rates across the languages of interest, suggesting effectiveness in tokenization which is fundamental for generative tasks. Key training parameters and architecture details were aligned with best practices, while also optimally configuring the model for the unique multilingual training approach undertaken.
Poro~34B underwent comprehensive evaluation across Finnish, English, and code generation tasks, demonstrating superior performance in Finnish and competitive results in English and programming languages. The model’s prowess in Finnish showcases the substantial advantage of multilingual training for enhancing performance in lower-resource languages. In English and code generation, Poro~34B displayed competitive capabilities, affirming the effectiveness of incorporating diverse linguistic data in training broad-spectrum generative models. The model also exhibited remarkable translation capabilities between English and Finnish, underscoring the beneficial impact of including translation pairs in training.
The successful development of Poro~34B through multilingual training opens new vistas for creating robust large-scale language models for languages with relatively limited resources. The model’s adeptness across languages and various generative tasks highlights the potential of leveraging multilingual data to circumvent the constraints posed by data scarcity in specific languages. Future research could explore the scalability of this approach across additional languages and more deeply investigate the underlying factors contributing to the model's performance in translation and cross-linguistic generative tasks.
Poro~34B represents a significant advancement in the utilization of multilingual training to enhance language model capabilities beyond high-resource languages. By effectively leveraging a diverse pretraining corpus, the model has set new benchmarks in Finnish language processing, while also achieving noteworthy performance in English and programming languages. This study not only underscores the potential of multilingual training to expand the horizons of language model capabilities but also provides a blueprint for future explorations in leveraging multilingual data for language model development.
Open llm leaderboard. https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard
A framework for the evaluation of code generation models. https://github.com/bigcode-project/bigcode-evaluation-harness
Language models are few-shot learners. In H. Larochelle, M. Ranzato, R. Hadsell, M.F. Balcan, and H. Lin (eds.), Advances in Neural Information Processing Systems, volume 33, pp. 1877–1901. Curran Associates, Inc., 2020. https://proceedings.neurips.cc/paper_files/paper/2020/file/1457c0d6bfcb4967418bfb8ac142f64a-Paper.pdf.
Unsupervised cross-lingual representation learning at scale. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pp. 8440–8451. Association for Computational Linguistics, 2020. doi: 10.18653/v1/2020.acl-main.747. https://aclanthology.org/2020.acl-main.747.
BERT: Pre-training of deep bidirectional transformers for language understanding. In Jill Burstein, Christy Doran, and Thamar Solorio (eds.), Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pp. 4171–4186, Minneapolis, Minnesota, June 2019. Association for Computational Linguistics. doi: 10.18653/v1/N19-1423. https://aclanthology.org/N19-1423.
Lessons learned from GPT-SW3: Building the first large-scale generative language model for Swedish. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pp. 3509–3518, 2022. https://aclanthology.org/2022.lrec-1.376.
Match the script, adapt if multilingual: Analyzing the effect of multilingual pretraining on cross-lingual transferability. In Smaranda Muresan, Preslav Nakov, and Aline Villavicencio (eds.), Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 1500–1512. Association for Computational Linguistics, 2022. doi: 10.18653/v1/2022.acl-long.106. https://aclanthology.org/2022.acl-long.106.
A framework for few-shot language model evaluation, 12 2023. https://zenodo.org/records/10256836.
The Flores-101 evaluation benchmark for low-resource and multilingual machine translation. Transactions of the Association for Computational Linguistics, 10:522–538, 2022. doi: 10.1162/tacl˙a˙00474. https://aclanthology.org/2022.tacl-1.30.
An empirical analysis of compute-optimal large language model training. In S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, and A. Oh (eds.), Advances in Neural Information Processing Systems, volume 35, pp. 30016–30030. Curran Associates, Inc., 2022b. https://proceedings.neurips.cc/paper_files/paper/2022/file/c1e2faff6f588870935f114ebe04a3e5-Paper-Conference.pdf.
The state and fate of linguistic diversity and inclusion in the NLP world. In Dan Jurafsky, Joyce Chai, Natalie Schluter, and Joel Tetreault (eds.), Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pp. 6282–6293, 2020. doi: 10.18653/v1/2020.acl-main.560. https://aclanthology.org/2020.acl-main.560.
The stack: 3 TB of permissively licensed source code. Transactions on Machine Learning Research, 2023. ISSN 2835-8856. https://openreview.net/forum?id=pxpbTdUEpD.
TruthfulQA: Measuring how models mimic human falsehoods. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 3214–3252, Dublin, Ireland, May 2022a. Association for Computational Linguistics. doi: 10.18653/v1/2022.acl-long.229. https://aclanthology.org/2022.acl-long.229.
FinGPT: Large generative models for a small language. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pp. 2710–2726. Association for Computational Linguistics, 2023. doi: 10.18653/v1/2023.emnlp-main.164. https://aclanthology.org/2023.emnlp-main.164.
Lifting the curse of multilinguality by pre-training modular transformers. In Marine Carpuat, Marie-Catherine de Marneffe, and Ivan Vladimir Meza Ruiz (eds.), Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pp. 3479–3495, Seattle, United States, July 2022. Association for Computational Linguistics. doi: 10.18653/v1/2022.naacl-main.255. https://aclanthology.org/2022.naacl-main.255.
How good is your tokenizer? on the monolingual performance of multilingual language models. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pp. 3118–3135, 2021. doi: 10.18653/v1/2021.acl-long.243. https://aclanthology.org/2021.acl-long.243.
SlimPajama: A 627B token cleaned and deduplicated version of RedPajama. https://www.cerebras.net/blog/slimpajama-a-627b-token-cleaned-and-deduplicated-version-of-redpajama, 2023. https://huggingface.co/datasets/cerebras/SlimPajama-627B.
Jörg Tiedemann. The Tatoeba Translation Challenge – Realistic data sets for low resource and multilingual MT. In Proceedings of the Fifth Conference on Machine Translation, pp. 1174–1182, Online, November 2020. Association for Computational Linguistics. https://www.aclweb.org/anthology/2020.wmt-1.139.
OPUS-MT – building open translation services for the world. In Proceedings of the 22nd Annual Conference of the European Association for Machine Translation, pp. 479–480, Lisboa, Portugal, November 2020. European Association for Machine Translation. https://aclanthology.org/2020.eamt-1.61.
Together Computer. Redpajama: An open source recipe to reproduce llama training dataset, 2023. https://github.com/togethercomputer/RedPajama-Data.
HellaSwag: Can a machine really finish your sentence? In Anna Korhonen, David Traum, and Lluís Màrquez (eds.), Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pp. 4791–4800. Association for Computational Linguistics, 2019. doi: 10.18653/v1/P19-1472. https://aclanthology.org/P19-1472.