eridai/Yumi: A Compact Model for Long Contexts

eridai/Yumi is a 0.8 billion parameter language model developed by eridai, distinguished by its exceptionally large context window of 32,768 tokens. This combination of a small parameter count and extensive context length makes Yumi particularly efficient for tasks that require processing and understanding very long sequences of text.

Key Characteristics

• Compact Size: With only 0.8 billion parameters, Yumi is lightweight, enabling faster inference and lower computational resource requirements compared to larger models.
• Extended Context Window: The 32,768-token context length allows the model to maintain coherence and draw information from vast amounts of input text, crucial for complex document analysis, summarization, and conversational AI over long interactions.

Ideal Use Cases

• Long Document Processing: Excellent for tasks like summarizing lengthy articles, legal documents, or research papers where understanding the full context is paramount.
• Resource-Constrained Environments: Its small footprint makes it suitable for deployment on edge devices or in applications where computational resources are limited.
• Conversational AI: Can maintain long-running conversations, remembering details from earlier turns without losing context.
• Prototyping and Development: Offers a quick and efficient solution for developers needing a capable model for long-context tasks without the overhead of much larger models.