allknowingroger/Gemmaslerp-9B is a 9 billion parameter language model created by merging nbeerbower/Gemma2-Gutenberg-Doppel-9B and DreadPoor/Emu_Eggs-9B-Model_Stock using the SLERP method. This model is designed to combine the strengths of its base models, with a specific configuration that suggests a focus on balancing different capabilities. It achieves an average score of 30.86 on the Open LLM Leaderboard, indicating general language understanding and generation capabilities.
Loading preview...
Model Overview
allknowingroger/Gemmaslerp-9B is a 9 billion parameter language model resulting from a merge of two pre-trained models: nbeerbower/Gemma2-Gutenberg-Doppel-9B and DreadPoor/Emu_Eggs-9B-Model_Stock. This model was created using the SLERP merge method via mergekit.
Merge Configuration
The merge utilized a specific V-shaped curve parameter configuration, suggesting an intent to leverage different base models for distinct layers. This approach aims to combine the strengths of the constituent models, potentially optimizing for varied tasks across different parts of the network.
Performance Benchmarks
Evaluated on the Open LLM Leaderboard, Gemmaslerp-9B achieved an average score of 30.86. Key individual metric scores include:
- IFEval (0-Shot): 70.43
- BBH (3-Shot): 41.56
- MMLU-PRO (5-shot): 35.12
- MATH Lvl 5 (4-Shot): 7.63
These results indicate its general language understanding and reasoning capabilities, with a notable score in instruction following (IFEval).
Potential Use Cases
Given its merged nature and benchmark performance, Gemmaslerp-9B could be suitable for applications requiring a balance of general language tasks, including:
- Instruction following and text generation
- General question answering
- Content creation where a blend of different model strengths is beneficial