l3utterfly/llama2-7b-layla
l3utterfly/llama2-7b-layla is a 7 billion parameter Llama2-based language model fine-tuned by l3utterfly using ShareGPT datasets. Optimized for multi-turn conversations, this model serves as the base for Layla, an offline personal assistant. It demonstrates an average performance of 45.56 on the Open LLM Leaderboard, with notable scores in HellaSwag (79.34) and Winogrande (74.11).
Loading preview...
Model Overview
l3utterfly/llama2-7b-layla is a 7 billion parameter model built upon the Llama2 architecture, developed by l3utterfly and funded by Layla Network. It has been specifically fine-tuned using ShareGPT datasets to excel in multi-turn conversational interactions.
Key Capabilities
- Multi-turn Conversations: Optimized for engaging in extended, coherent dialogue.
- Llama2 Foundation: Benefits from the robust capabilities of the Llama2 base model.
- Offline Assistant Base: Serves as the core language model for Layla, an offline personal assistant.
Performance Highlights
Evaluated on the Open LLM Leaderboard, the model achieves an average score of 45.56. Specific benchmark results include:
- HellaSwag (10-shot): 79.34
- Winogrande (5-shot): 74.11
- ARC (25-shot): 54.18
- MMLU (5-shot): 49.7
Good For
- Developing conversational AI agents requiring multi-turn interaction.
- Applications where a Llama2-based model with conversational fine-tuning is beneficial.
- Use cases similar to personal assistants, especially those designed for offline operation.