Overview

Kunokukulemonchini-7b is a 7 billion parameter language model developed by icefog72. It is a merged model, created using the SLERP (Spherical Linear Interpolation) method, combining the strengths of two base models: grimjim/kukulemon-7B and Nitral-AI/Kunocchini-7b-128k-test.

Key Capabilities & Features

• Extended Context Window: The model's configuration was edited to achieve an approximate 32k context window, enhancing its ability to process and generate longer texts.
• Quantized Versions Available: Several EXL2 quantized versions are provided by Natkituwu, ranging from 3.5bpw to 8.0bpw, catering to users with varying VRAM constraints, including a 4.1bpw version for 6GB VRAM.

Performance

Evaluated on the Open LLM Leaderboard, Kunokukulemonchini-7b demonstrates competitive performance for its size:

• Average Score: 69.61
• HellaSwag (10-Shot): 86.31
• MMLU (5-Shot): 65.31
• GSM8k (5-Shot): 60.20

Use Cases

This model is suitable for applications requiring a 7B parameter model with an extended context window, particularly for tasks where a balance between performance and resource efficiency is desired. Its benchmark scores suggest proficiency in reasoning, common sense, and language understanding tasks.