l3utterfly/mistral-7b-v0.1-layla-v1
l3utterfly/mistral-7b-v0.1-layla-v1 is a 7 billion parameter Mistral-based language model, fine-tuned by l3utterfly using ShareGPT datasets for multi-turn conversational AI. This model is specifically optimized for dialogue and serves as the base for the Layla offline personal assistant. It features an 8192-token context length and demonstrates balanced performance across various benchmarks, including 60.15 on ARC and 83.25 on HellaSwag.
Loading preview...
Model Overview
l3utterfly/mistral-7b-v0.1-layla-v1 is a 7 billion parameter language model developed by l3utterfly and funded by Layla Network. It is a fine-tuned variant of the Mistral 7B base model, specifically optimized for multi-turn conversational interactions using ShareGPT datasets. The model is licensed under Apache-2.0 and supports an 8192-token context length.
Key Capabilities
- Multi-turn Conversation: Fine-tuned on ShareGPT datasets, making it proficient in engaging in extended dialogues.
- Base for Layla Assistant: Serves as the foundational model for the Layla offline personal assistant, indicating its suitability for interactive AI applications.
- English Language Support: Primarily designed for English natural language processing tasks.
Performance Benchmarks
Evaluations on the Open LLM Leaderboard show the model's performance across several metrics:
- Avg.: 55.05
- ARC (25-shot): 60.15
- HellaSwag (10-shot): 83.25
- MMLU (5-shot): 60.31
- TruthfulQA (0-shot): 48.9
- Winogrande (5-shot): 75.93
- GSM8K (5-shot): 16.83
- DROP (3-shot): 40.01
Use Cases
This model is particularly well-suited for applications requiring robust multi-turn conversational abilities, such as chatbots, virtual assistants, and interactive dialogue systems. Its role as the base for the Layla offline personal assistant highlights its potential for embedded or local AI solutions.
Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.