pe-nlp/llama-2-13b-vicuna-wizard
pe-nlp/llama-2-13b-vicuna-wizard is a 13 billion parameter language model based on the Llama 2 architecture, fine-tuned to combine the strengths of Vicuna and WizardLM. It features a 4096-token context window and achieves an average score of 50.79 on the Open LLM Leaderboard benchmarks. This model is designed for general-purpose conversational AI and instruction-following tasks, demonstrating capabilities across various reasoning and knowledge-based evaluations.
Loading preview...
Model Overview
pe-nlp/llama-2-13b-vicuna-wizard is a 13 billion parameter language model built upon the Llama 2 architecture. This model integrates fine-tuning methodologies from both Vicuna and WizardLM, aiming to enhance its instruction-following and conversational abilities. It supports a context length of 4096 tokens, making it suitable for processing moderately long inputs and generating coherent responses.
Key Capabilities & Performance
Evaluated on the Open LLM Leaderboard, this model demonstrates a balanced performance across various benchmarks, achieving an average score of 50.79. Notable scores include:
- HellaSwag (10-shot): 82.16
- Winogrande (5-shot): 74.98
- ARC (25-shot): 57.76
- MMLU (5-shot): 54.68
These results indicate its proficiency in common sense reasoning, reading comprehension, and general knowledge tasks. While its performance on mathematical reasoning (GSM8K) is lower at 0.91, it shows solid capabilities in other areas.
Intended Use Cases
This model is well-suited for applications requiring:
- General-purpose conversational agents: Engaging in dialogue and answering a wide range of questions.
- Instruction following: Executing commands and generating responses based on specific instructions.
- Text generation: Creating coherent and contextually relevant text for various prompts.
- Knowledge-based querying: Retrieving and synthesizing information from its training data.