jsfs11/West-Dare-7B
West-Dare-7B is a 7 billion parameter language model created by jsfs11, formed by merging senseable/Westlake-7B and abideen/DareVox-7B using the TIES merging method. Built upon the Mistral-7B-v0.1 base, this model is designed for general language tasks with a 4096-token context length. It achieves an average score of 73.65 on the Open LLM Leaderboard, demonstrating solid performance across various benchmarks including reasoning and common sense.
Loading preview...
West-Dare-7B: A Merged 7B Language Model
West-Dare-7B is a 7 billion parameter language model developed by jsfs11, created through a merge of two existing models: senseable/Westlake-7B and abideen/DareVox-7B. This model leverages the robust Mistral-7B-v0.1 as its base architecture and utilizes the TIES merging method with specific density and weight parameters for each component model.
Key Capabilities & Performance
This model demonstrates competitive performance on the Open LLM Leaderboard, achieving an average score of 73.65. Its benchmark results include:
- AI2 Reasoning Challenge (25-Shot): 71.42
- HellaSwag (10-Shot): 87.57
- MMLU (5-Shot): 64.29
- TruthfulQA (0-shot): 66.25
- Winogrande (5-shot): 84.53
- GSM8k (5-shot): 67.85
These scores indicate its proficiency in reasoning, common sense, and general knowledge tasks. The model supports a context length of 4096 tokens, making it suitable for a range of conversational and text generation applications.
When to Use This Model
West-Dare-7B is a strong candidate for general-purpose language generation and understanding tasks where a 7B parameter model is appropriate. Its balanced performance across multiple benchmarks suggests it can be effectively used for:
- Text summarization
- Question answering
- Content creation
- Reasoning-based applications
Developers can easily integrate and experiment with West-Dare-7B using the provided Hugging Face transformers library code snippet.