EryriLabs/TriFusionNexus-7b

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kPublished:Feb 25, 2024License:cc-by-nc-2.0Architecture:Transformer Open Weights Cold

EryriLabs/TriFusionNexus-7b is a 7 billion parameter language model created by EryriLabs using the TIES merge method, based on CultriX/NeuralTrix-7B-dpo and incorporating mlabonne/AlphaMonarch-7B and bardsai/jaskier-7b-dpo-v5.6. This model features a 4096-token context length and achieves an average score of 76.32 on the Open LLM Leaderboard, demonstrating strong performance across various reasoning and language understanding benchmarks, making it suitable for general-purpose conversational and analytical tasks.

Loading preview...

TriFusionNexus-7b Overview

TriFusionNexus-7b is a 7 billion parameter language model developed by EryriLabs, created through a sophisticated merge of several pre-trained models using the TIES (Trimmed-mean-based Information Entropy Search) method. This approach combines the strengths of multiple models to enhance overall performance and capabilities. The model is built upon CultriX/NeuralTrix-7B-dpo as its base, integrating mlabonne/AlphaMonarch-7B and bardsai/jaskier-7b-dpo-v5.6 to achieve a balanced and robust architecture.

Key Capabilities

  • Enhanced Reasoning: Achieves 72.78 on the AI2 Reasoning Challenge (25-Shot) and 68.46 on GSM8k (5-Shot), indicating strong logical and mathematical reasoning abilities.
  • Language Understanding: Demonstrates robust performance in language comprehension with 89.17 on HellaSwag (10-Shot) and 84.93 on Winogrande (5-Shot).
  • Factuality and Knowledge: Scores 78.13 on TruthfulQA (0-shot), suggesting a good grasp of factual information.
  • General-Purpose Utility: With an average score of 76.32 on the Open LLM Leaderboard, it is well-suited for a wide range of natural language processing tasks.

Good For

  • General AI Applications: Its balanced performance across various benchmarks makes it a solid choice for diverse NLP tasks.
  • Reasoning-Intensive Workloads: Ideal for applications requiring logical deduction, problem-solving, and mathematical understanding.
  • Conversational AI: Capable of generating coherent and contextually relevant responses for chatbots and virtual assistants.
  • Research and Development: Provides a strong foundation for further fine-tuning or experimentation due to its merged architecture.

Finetunes
1 models