Alelcv27/Llama3.2-3B-TIES-Math-Code

TEXT GENERATIONConcurrency Cost:1Model Size:3.2BQuant:BF16Ctx Length:32kPublished:Apr 19, 2026Architecture:Transformer Cold

Alelcv27/Llama3.2-3B-TIES-Math-Code is a 3.2 billion parameter language model based on the Llama 3.2 architecture, created by Alelcv27 through a TIES merge. This model integrates specialized capabilities from Llama3.2-3B-Base-Math and Llama3.2-3B-Base-Code, making it optimized for mathematical reasoning and code generation tasks. It leverages a 32768-token context window, providing enhanced performance for complex problem-solving in these domains.

Loading preview...

Overview

Alelcv27/Llama3.2-3B-TIES-Math-Code is a 3.2 billion parameter language model developed by Alelcv27. It is a product of a TIES merge (Trimmed, Iterative, & Self-gated) of pre-trained models, using meta-llama/Llama-3.2-3B as its base. This merging technique combines the strengths of multiple specialized models into a single, more versatile model.

Key Capabilities

  • Specialized Merge: Created by merging Alelcv27/Llama3.2-3B-Base-Math and Alelcv27/Llama3.2-3B-Base-Code using the TIES method, which allows for the selective integration of parameters from different models.
  • Enhanced for Specific Domains: The merge process specifically targets and enhances the model's proficiency in:
    • Mathematical Reasoning: Improved performance on tasks requiring numerical understanding and problem-solving.
    • Code Generation: Stronger capabilities in generating and understanding programming code.
  • Llama 3.2 Architecture: Built upon the robust Llama 3.2 foundation, providing a solid base for its specialized functions.
  • Context Window: Features a 32768-token context length, beneficial for handling longer and more complex mathematical problems or code snippets.

Good For

  • Developers and Researchers: Ideal for applications requiring strong performance in both mathematical computations and code-related tasks.
  • Educational Tools: Can be used in tools designed to assist with learning or solving math and programming problems.
  • Automated Code Generation: Suitable for scenarios where generating code or assisting with programming logic is crucial.
  • Technical Problem Solving: Excels in tasks that blend logical reasoning with computational or coding requirements.

Finetunes
1 models