CYFRAGOVPL/Llama-PLLuM-70B-chat-2412

TEXT GENERATIONConcurrency Cost:4Model Size:70BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Feb 6, 2025License:llama3.1Architecture:Transformer0.0K Cold

The CYFRAGOVPL/Llama-PLLuM-70B-chat-2412 is a 70 billion parameter large language model from the PLLuM family, developed by a consortium of Polish scientific institutions led by Politechnika Wrocławska. Built on Llama 3.1, it is specialized in Polish and other Slavic/Baltic languages, with additional English data for generalization. This chat-tuned model excels in generating contextually coherent text, question answering, and summarization, achieving state-of-the-art results in Polish-language tasks and top scores in Polish public administration benchmarks.

Loading preview...

PLLuM: Polish Large Language Models

CYFRAGOVPL/Llama-PLLuM-70B-chat-2412 is a 70 billion parameter model from the PLLuM family, developed by a consortium of Polish scientific institutions. This model is built upon Llama 3.1 and is specifically designed for Polish and other Slavic/Baltic languages, incorporating English data for broader applicability. It has been extensively refined through instruction tuning, preference learning, and advanced alignment techniques.

Key Capabilities & Features

  • Specialized Multilingualism: Optimized for Polish, Slavic, and Baltic languages, with strong performance in English.
  • Extensive Polish Data: Pretrained on up to 150 billion tokens of high-quality Polish text, alongside other languages.
  • Organic Instruction Tuning: Fine-tuned on a unique dataset of ~40k manually created Polish "organic instructions," including multi-turn dialogues, to capture subtle aspects of human-model interaction.
  • Polish Preference Corpus: Utilizes the first Polish-language preference corpus for alignment, enhancing correctness, balance, and safety, particularly for sensitive topics.
  • State-of-the-Art Performance: Achieves top scores in custom benchmarks for Polish public administration tasks and state-of-the-art results across broader Polish-language tasks.
  • Chat Alignment: The -chat suffix indicates it has been aligned on human preferences, making it generally safer and more efficient for dialogue and general-purpose scenarios.

Intended Use Cases

  • General Language Tasks: Ideal for text generation, summarization, and question answering in Polish and related languages.
  • Domain-Specific Assistants: Particularly effective for applications in Polish public administration, legal, and bureaucratic contexts, especially when combined with Retrieval Augmented Generation (RAG).
  • Research & Development: Serves as a robust foundation for building downstream AI applications requiring strong Polish language command.