AiMavenAi/AiMaven-Prometheus Overview

AiMavenAi/AiMaven-Prometheus is a 7 billion parameter language model developed by AiMavenAi. It is a product of merging two distinct models, jefferylovely/SuperThetaMaven and flemmingmiguel/MBX-7B-v3, using a Slerp merge method. This approach combines the strengths of its constituent models to enhance overall performance.

Key Capabilities & Performance

This model demonstrates robust performance across a range of benchmarks, as evaluated on the Open LLM Leaderboard. It achieves an impressive average score of 75.74.

• AI2 Reasoning Challenge (25-Shot): 73.98
• HellaSwag (10-Shot): 88.83
• MMLU (5-Shot): 65.17
• TruthfulQA (0-shot): 72.22
• Winogrande (5-shot): 85.16
• GSM8k (5-shot): 69.07

These scores indicate strong general reasoning, common sense inference, and question-answering abilities. The model's configuration specifies a bfloat16 dtype for efficient computation.

Usage & Integration

AiMaven-Prometheus can be easily integrated into Python projects using the transformers library, supporting standard text generation pipelines. Its 4096-token context window allows for processing moderately long inputs and generating coherent responses.