field2437/phi-2-platypus-Commercial-lora
field2437/phi-2-platypus-Commercial-lora is a 3 billion parameter causal language model developed by field2437, fine-tuned from Microsoft's Phi-2 base model. It was trained on the kyujinpy/Open-platypus-Commercial dataset and features a 2048-token context length. This model is optimized for general language understanding and generation, demonstrating competitive performance across various benchmarks including Copa, HellaSwag, BoolQ, and MMLU.
Loading preview...
Overview
field2437/phi-2-platypus-Commercial-lora is a 3 billion parameter language model developed by field2437, built upon Microsoft's Phi-2 base model. It has been fine-tuned using the kyujinpy/Open-platypus-Commercial dataset, aiming to enhance its general language capabilities. The model supports a context length of 2048 tokens.
Key Capabilities
- General Language Understanding: Capable of processing and generating human-like text for a variety of tasks.
- Instruction Following: Designed to respond appropriately to given instructions, as demonstrated by its sample code.
- Competitive Performance: Achieves solid results on standard benchmarks, including:
- Copa: 0.8900 (0-shot)
- HellaSwag: 0.5573 (0-shot)
- BoolQ: 0.8260 (0-shot)
- MMLU: 0.5513 (0-shot)
Good for
- Text Generation: Creating coherent and contextually relevant text based on prompts.
- Question Answering: Responding to factual questions and completing tasks described in instructions.
- Research and Development: A compact yet capable model for experimenting with language model applications, especially given its base on Phi-2 and specific fine-tuning.