Overview

TomGrc/FusionNet is a 10.7 billion parameter language model fine-tuned using a novel "Fusion" method. This model was developed by TomGrc with the primary goal of experimenting with and demonstrating how the Fusion technique can significantly boost the performance of an original model. It is specifically fine-tuned on the English language.

Key Capabilities & Performance

FusionNet exhibits strong performance across various benchmarks, as evaluated on the Open LLM Leaderboard. Its average score is 74.38, with notable results including:

• AI2 Reasoning Challenge (25-Shot): 71.25
• HellaSwag (10-Shot): 88.42
• MMLU (5-Shot): 66.36
• TruthfulQA (0-shot): 71.95
• Winogrande (5-shot): 83.27
• GSM8k (5-shot): 65.05

Use Cases

This model is ideal for researchers and developers interested in:

• Exploring the impact of the "Fusion" fine-tuning method on language model performance.
• General English language understanding and generation tasks.
• Applications requiring a 10.7B parameter model with competitive benchmark scores.