imbue/llama3-question-quality

TEXT GENERATIONPricing:Input $3.5 / Cached $0.7 / Output $8.3Concurrent Unit Cost:4Model Size:70BQuant:FP8Context Size:8kTool Calling:SupportedPublished:Jun 25, 2024License:mitArchitecture:Transformer0.0K Open Weights Featherless Exclusive Cold

imbue/llama3-question-quality is a 70 billion parameter model built upon Meta Llama 3, specifically fine-tuned on human question quality judgments. This model is optimized for evaluating and understanding the quality of questions, leveraging a specialized dataset for its training. It is designed to assess question characteristics, making it suitable for applications requiring nuanced understanding of query effectiveness. The model has a context length of 8192 tokens.

Loading preview...

imbue/llama3-question-quality: Specialized for Question Quality Assessment

This model, imbue/llama3-question-quality, is a 70 billion parameter variant of Meta Llama 3, uniquely fine-tuned for evaluating the quality of human-generated questions. Its core differentiation lies in its training data: it was specifically fine-tuned on a dataset of human question quality judgments, available at imbue-ai/human_question_quality_judgments.

Key Capabilities

  • Question Quality Evaluation: Excels at assessing various aspects of question quality based on human judgments.
  • Specialized Fine-tuning: Benefits from targeted training on a unique dataset, distinguishing it from general-purpose LLMs.
  • Llama 3 Foundation: Built on the robust Meta Llama 3 architecture, providing a strong base for its specialized task.

Good For

  • Automated Question Curation: Ideal for systems that need to filter, rank, or improve the quality of user-submitted questions.
  • Search and Q&A Systems: Can enhance the performance of search engines or question-answering systems by prioritizing high-quality queries.
  • Research in Question Understanding: Useful for researchers exploring the nuances of question formulation and effectiveness. Further details on its development and evaluation can be found at imbue.com/research/70b-evals/.