jenny08311/affine-test-3

TEXT GENERATIONConcurrency Cost:2Model Size:32BQuant:FP8Ctx Length:32kPublished:Apr 10, 2026Architecture:Transformer Cold

jenny08311/affine-test-3 is a 32 billion parameter language model merged using the TIES method, based on Qwen/Qwen3-32B. This model integrates components from two 'Affine' models by gurand, specifically combining their strengths across MLP and self-attention layers. It is designed to leverage the combined capabilities of its constituent models, offering a potentially enhanced performance profile for general language tasks. The model has a context length of 32768 tokens, making it suitable for applications requiring extensive contextual understanding.

Loading preview...

Model Overview

jenny08311/affine-test-3 is a 32 billion parameter language model created through a merge of pre-trained models using the TIES (Trimmed, Iterative, and Selective) merge method. Its foundation is the Qwen/Qwen3-32B model, which serves as the base for integrating specialized components.

Key Capabilities

  • Merged Architecture: This model combines two distinct 'Affine' models from gurand, specifically gurand/Affine-5CFL2YaBrJZCUSPBTjcDcTUSbnrm3UtAgKRsTU2KRcu9nvyR and gurand/Affine-5CrMoVRmR8yP69Kh4iyrELehGYzUh3t7Q9hYVZUSjJA3VqDV.
  • TIES Merge Method: The TIES merging technique was applied, allowing for a weighted combination of parameters from the constituent models, with specific density and weight adjustments for MLP and self-attention layers.
  • Qwen3-32B Base: Leveraging the robust architecture and pre-training of Qwen/Qwen3-32B, this merged model aims to inherit and potentially enhance its general language understanding and generation capabilities.
  • Extended Context: With a context length of 32768 tokens, it is well-suited for tasks requiring the processing of longer inputs and maintaining coherence over extended conversations or documents.

Good For

  • General Language Tasks: Suitable for a broad range of applications that benefit from a large language model with a strong base.
  • Research and Experimentation: Ideal for researchers and developers interested in exploring the effects of model merging techniques like TIES on established base models and specialized components.
  • Applications Requiring Long Context: Its 32K context window makes it effective for summarization, detailed question answering, and content generation over lengthy texts.