Nexesenex/Llama_3.2_3b_Kermes_v2
Nexesenex/Llama_3.2_3b_Kermes_v2 is a 3.2 billion parameter language model created by Nexesenex through a merge of existing Llama 3.2-based models using the Model Stock method. This model integrates components from Dolphin3.0-Llama3.2-3B and calme-3.3-llamaloi-3b, built upon Nexesenex/Llama_3.2_3b_Kermes_0.2_bf16 as a base. It is designed for general language tasks, leveraging its merged architecture to combine the strengths of its constituent models.
Loading preview...
Overview
Nexesenex/Llama_3.2_3b_Kermes_v2 is a 3.2 billion parameter language model developed by Nexesenex. This model was created using the Model Stock merge method, combining the capabilities of several pre-trained language models. It leverages Nexesenex/Llama_3.2_3b_Kermes_0.2_bf16 as its base, integrating contributions from cognitivecomputations/Dolphin3.0-Llama3.2-3B and MaziyarPanahi/calme-3.3-llamaloi-3b.
Key Capabilities
- Merged Architecture: Benefits from the combined strengths of Dolphin3.0-Llama3.2-3B and calme-3.3-llamaloi-3b.
- GGUF Quantizations: Available in static and iMatrix GGUF quantizations for efficient deployment, provided by Mradermacher.
Evaluation Highlights
Based on the Open LLM Leaderboard, the model shows:
- IFEval (0-Shot): 57.54%
- BBH (3-Shot): 22.05%
- MMLU-PRO (5-shot): 19.27%
Good For
- Experimentation with Merged Models: Ideal for researchers and developers interested in exploring the performance characteristics of models created via the Model Stock merging technique.
- Resource-Efficient Deployment: The availability of GGUF quantizations makes it suitable for deployment on consumer hardware or environments with limited resources.
- General Language Tasks: Can be applied to a variety of natural language processing tasks where a 3.2 billion parameter model is appropriate.