Model Overview

mayacinka/yam-jom-7B is a 7 billion parameter language model developed by mayacinka using a task arithmetic merging technique. It combines the strengths of two base models: eren23/ogno-monarch-jaskier-merge-7b-OH-PREF-DPO-v2 (weighted at 0.35) and yam-peleg/Experiment26-7B (weighted at 0.65), with yam-peleg/Experiment26-7B serving as the base model for the merge. This approach aims to leverage the distinct capabilities of its components to create a versatile model.

Key Capabilities & Performance

The model demonstrates solid performance across a range of benchmarks, as evaluated on the Open LLM Leaderboard. It achieves an average score of 76.60, with notable results in:

• AI2 Reasoning Challenge (25-Shot): 73.38
• HellaSwag (10-Shot): 89.15
• MMLU (5-Shot): 64.51
• TruthfulQA (0-shot): 78.04
• Winogrande (5-shot): 84.93
• GSM8k (5-shot): 69.60

These scores indicate its proficiency in reasoning, common sense, factual recall, and mathematical problem-solving.

Good For

• General-purpose language generation: Suitable for a wide array of text generation tasks.
• Reasoning and problem-solving: Its benchmark scores suggest good capabilities in logical inference and answering complex questions.
• Applications requiring balanced performance: The model offers a strong average performance across diverse metrics, making it a reliable choice for various use cases.