pankajmathur/model_007_13b_v2
pankajmathur/model_007_13b_v2 is a 13 billion parameter Llama2-based model developed by Pankaj Mathur, fine-tuned for both explanatory and instructional tasks. It leverages a diverse set of datasets including Open-Platypus, Alpaca, WizardLM, and Orca_minis_v1 to enhance its hybrid capabilities. The model demonstrates competitive performance on benchmarks like HellaSwag (82.42% acc_norm) and MMLU (56.37% acc_norm), making it suitable for general-purpose instruction following and explanation generation.
Loading preview...
Overview
pankajmathur/model_007_13b_v2 is a 13 billion parameter language model built upon the Llama2 architecture, developed by Pankaj Mathur. This model is uniquely designed as a hybrid (explain + instruct) style LLM, meaning it is capable of both providing detailed explanations and following direct instructions effectively. It was fine-tuned using a comprehensive collection of datasets, including Open-Platypus, Alpaca, WizardLM, Dolly-V2, Dolphin Samples, Orca_minis_v1, Alpaca_orca, and WizardLM_orca, to achieve its dual functionality.
Key Capabilities
- Hybrid Instruction Following and Explanation Generation: Excels at both understanding and executing instructions, as well as generating detailed, explanatory responses.
- Llama2 Base: Benefits from the robust foundation of the Llama2 architecture.
- Diverse Training: Trained on a wide array of high-quality instruction and explanation datasets.
Performance Highlights
Evaluated using the EleutherAI Language Model Evaluation Harness, the model shows solid performance on key metrics:
- HellaSwag: 82.42% acc_norm
- MMLU: 56.37% acc_norm
- ARC Challenge: 63.14% acc_norm
- TruthfulQA: 51.27% mc2
Good For
- Applications requiring models to both explain concepts and follow specific instructions.
- General-purpose conversational AI where clarity and adherence to prompts are important.
- Developers looking for a Llama2-based model with enhanced instructional and explanatory capabilities.