louisbrulenaudet/Maxine-34B-stock

TEXT GENERATIONConcurrency Cost:2Model Size:34BQuant:FP8Ctx Length:32kPublished:Apr 4, 2024License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

Maxine-34B-stock by louisbrulenaudet is a 34 billion parameter merged language model, combining ConvexAI/Luminex-34B-v0.2 and fblgit/UNA-34BeagleSimpleMath-32K-v1 with abacusai/Smaug-34B-v0.1 as its base. This model, noted for its strong performance in the 30B parameter class, is optimized for general language understanding and generation tasks. It leverages a model stock merge method to achieve high efficacy, making it suitable for a wide range of applications requiring robust language capabilities.

Loading preview...

Maxine-34B-stock Overview

Maxine-34B-stock is a 34 billion parameter language model developed by louisbrulenaudet, created through a sophisticated "model stock" merge method. It integrates components from ConvexAI/Luminex-34B-v0.2 and fblgit/UNA-34BeagleSimpleMath-32K-v1, built upon the abacusai/Smaug-34B-v0.1 base model.

Key Capabilities

  • High Performance: As of April 7, 2024, Maxine-34B-stock was recognized as a top-performing base merge model in the 30B parameter class on the Open LLM Leaderboard.
  • Merged Architecture: Utilizes a unique merging strategy to combine strengths from multiple foundational models, potentially leading to enhanced general language understanding and generation.
  • General Purpose: Designed for a broad spectrum of natural language processing tasks, benefiting from its diverse merged components.

When to Use This Model

  • General Text Generation: Ideal for applications requiring coherent and contextually relevant text output.
  • Research and Development: Suitable for researchers and developers exploring advanced model merging techniques and their impact on performance.
  • Benchmarking: A strong candidate for evaluating the efficacy of merged models against other LLMs in its parameter class.