Alelcv27/Llama3.1-8B-Base-Arcee-Math-Code

TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:32kPublished:Apr 24, 2026Architecture:Transformer Cold

Alelcv27/Llama3.1-8B-Base-Arcee-Math-Code is an 8 billion parameter language model based on the Llama 3.1 architecture, created by Alelcv27 using the Arcee Fusion merge method. This model specializes in mathematical and code-related tasks, combining the strengths of a math-focused base model with a code-specific model. It is designed for applications requiring strong performance in both numerical reasoning and programming contexts, offering a 32768 token context length.

Loading preview...

Model Overview

Alelcv27/Llama3.1-8B-Base-Arcee-Math-Code is an 8 billion parameter language model built upon the Llama 3.1 base architecture. This model was developed by Alelcv27 through a strategic merge using the Arcee Fusion method, leveraging mergekit.

Key Capabilities

  • Specialized Performance: This model is a fusion of two distinct base models: Alelcv27/Llama3.1-8B-Base-Math and Alelcv27/Llama3.1-8B-Base-Code. This unique combination aims to provide enhanced capabilities in both mathematical reasoning and code generation/understanding.
  • Merge Method: The Arcee Fusion method was applied, specifically combining layers from both source models across the full 32 layers (0 to 32) of the Llama 3.1 architecture.
  • Context Length: The model supports a substantial context window of 32768 tokens, beneficial for complex mathematical problems or extensive code analysis.

Good For

  • Mathematical Applications: Ideal for tasks requiring numerical reasoning, problem-solving, and understanding mathematical concepts.
  • Code Generation and Analysis: Well-suited for programming-related tasks, including generating code, debugging, and understanding various programming languages.
  • Hybrid Tasks: Particularly effective for use cases that involve both mathematical computations and programming logic, such as scientific computing or data analysis scripts.