pkarypis/phi2-lima

TEXT GENERATIONConcurrency Cost:1Model Size:3BQuant:BF16Ctx Length:2kPublished:Apr 26, 2024License:mitArchitecture:Transformer Open Weights Cold

pkarypis/phi2-lima is a 3 billion parameter causal language model, fine-tuned by pkarypis from Microsoft's Phi-2 architecture. It was trained on the GAIR/lima dataset, which focuses on high-quality, instruction-following data. This model is primarily intended for instruction-following tasks, leveraging its training on a curated dataset to generate coherent and contextually relevant responses.

Loading preview...

Overview

pkarypis/phi2-lima is a 3 billion parameter language model, fine-tuned from the base microsoft/phi-2 architecture. This model was specifically trained on the GAIR/lima dataset, which is known for its focus on high-quality instruction-following examples. The fine-tuning process aimed to enhance the model's ability to understand and execute instructions effectively.

Key Capabilities

  • Instruction Following: Optimized for generating responses that adhere to given instructions.
  • Compact Size: At 3 billion parameters, it offers a balance between performance and computational efficiency.
  • Base Model Heritage: Benefits from the strong foundational capabilities of the Microsoft Phi-2 model.

Good for

  • Applications requiring a smaller, efficient model for instruction-based tasks.
  • Scenarios where generating accurate and contextually appropriate responses to prompts is crucial.
  • Research and development focusing on instruction-tuned models within the 3B parameter class.