ResplendentAI/Datura_7B
Datura_7B by ResplendentAI is a 7 billion parameter language model with an 8192 token context length, designed for reduced refusal rates and increased receptiveness. It achieves an average score of 74.28 on the Open LLM Leaderboard, with notable performance in HellaSwag (88.27) and Winogrande (84.53). This model is optimized for scenarios requiring a more direct and less restrictive AI response, making it suitable for applications where traditional models might exhibit excessive caution.
Loading preview...
Datura 7B: A Receptive Language Model
ResplendentAI's Datura 7B is a 7 billion parameter language model engineered to be more receptive and exhibit fewer refusals compared to other models. With an 8192 token context length, it aims to provide direct and less restrictive responses, making it suitable for a variety of applications where a more straightforward AI interaction is desired.
Key Capabilities & Performance
Datura 7B demonstrates solid performance across several benchmarks, as evaluated on the Open LLM Leaderboard:
- Average Score: 74.28
- HellaSwag (10-Shot): 88.27
- Winogrande (5-Shot): 84.53
- AI2 Reasoning Challenge (25-Shot): 72.10
- TruthfulQA (0-shot): 71.03
- MMLU (5-Shot): 64.15
- GSM8k (5-shot): 65.58
Good For
- Applications requiring direct responses: Its design focuses on reducing refusals, making it useful for tasks where a more assertive or less cautious AI is beneficial.
- General language understanding and generation: The model's benchmark scores indicate strong capabilities in reasoning, common sense, and question answering.
- Exploratory AI interactions: Developers looking for a model that is less prone to self-censorship in certain contexts may find Datura 7B particularly useful.
Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.