spar-project/Llama-3.2-3B-Instruct-minimal-layers

TEXT GENERATIONConcurrency Cost:1Model Size:3.2BQuant:BF16Ctx Length:32kPublished:Mar 25, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

The spar-project/Llama-3.2-3B-Instruct-minimal-layers is a 3.2 billion parameter instruction-tuned Llama model developed by spar-project. It was finetuned from unsloth/Llama-3.2-3B-Instruct using Unsloth and Huggingface's TRL library, enabling 2x faster training. This model is optimized for efficient instruction-following tasks, leveraging its smaller size and specialized training for rapid deployment.

Loading preview...

Model Overview

The spar-project/Llama-3.2-3B-Instruct-minimal-layers is a 3.2 billion parameter instruction-tuned language model. Developed by spar-project, this model is a finetuned variant of unsloth/Llama-3.2-3B-Instruct.

Key Capabilities

  • Efficient Training: This model was trained using Unsloth and Huggingface's TRL library, which facilitated a 2x faster training process compared to standard methods.
  • Instruction Following: As an instruction-tuned model, it is designed to understand and execute commands or prompts effectively.
  • Llama Architecture: Based on the Llama model family, it inherits the foundational capabilities of this architecture.

Good For

  • Resource-Constrained Environments: Its 3.2 billion parameter size makes it suitable for applications where computational resources are limited.
  • Rapid Prototyping: The efficient training methodology suggests it can be quickly adapted or deployed for various instruction-based tasks.
  • Instruction-Based Applications: Ideal for tasks requiring direct instruction following, such as chatbots, content generation, or summarization where a smaller, performant model is preferred.