saishf/West-Hermes-7B is a 7 billion parameter language model merged from Mistral-7B-v0.1, OpenHermes-2.5-Mistral-7B, and WestLake-7B-v2 using the DARE TIES method. This model leverages the strengths of its constituent models to achieve a strong average performance of 73.60 on the Open LLM Leaderboard, making it suitable for general-purpose reasoning and language generation tasks. It offers a 4096-token context length, providing robust capability for various applications.
Loading preview...
West-Hermes-7B: A Merged 7B Language Model
West-Hermes-7B is a 7 billion parameter language model created by saishf through a sophisticated merge of several pre-trained models. It utilizes the DARE TIES merge method to combine the capabilities of:
- mistralai/Mistral-7B-v0.1 (as the base model)
- teknium/OpenHermes-2.5-Mistral-7B
- senseable/WestLake-7B-v2
This merging strategy aims to consolidate the strengths of these individual models into a single, more versatile offering.
Performance Highlights
Evaluated on the Hugging Face Open LLM Leaderboard, West-Hermes-7B demonstrates strong general performance with an average score of 73.60. Key benchmark results include:
- AI2 Reasoning Challenge (25-Shot): 71.67
- HellaSwag (10-Shot): 87.60
- MMLU (5-Shot): 64.83
- Winogrande (5-Shot): 84.69
- GSM8k (5-Shot): 68.54
These scores indicate solid capabilities across various reasoning, common sense, and mathematical tasks.
Use Cases
Given its balanced performance across multiple benchmarks, West-Hermes-7B is well-suited for a range of general-purpose applications requiring robust language understanding and generation. It can be effectively used for:
- General text generation and completion
- Reasoning and problem-solving tasks
- Question answering
- Conversational AI and chatbots