sendosaid/ShieldGPT-8B-Merged

TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:8kPublished:May 16, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

ShieldGPT-8B-Merged is an 8 billion parameter Llama-based causal language model developed by sendosaid, fine-tuned from unsloth/deepseek-r1-distill-llama-8b-unsloth-bnb-4bit. This model was trained using Unsloth and Huggingface's TRL library, enabling faster training. It offers a context length of 8192 tokens, making it suitable for applications requiring efficient processing of moderately long sequences.

Loading preview...

ShieldGPT-8B-Merged Overview

ShieldGPT-8B-Merged is an 8 billion parameter language model developed by sendosaid. It is a Llama-based model, specifically fine-tuned from unsloth/deepseek-r1-distill-llama-8b-unsloth-bnb-4bit. This model leverages the Unsloth library, which facilitated a 2x faster training process, alongside Huggingface's TRL library.

Key Characteristics

  • Parameter Count: 8 billion parameters, offering a balance between performance and computational efficiency.
  • Base Model: Fine-tuned from a DeepSeek-R1-Distill Llama variant.
  • Training Efficiency: Utilizes Unsloth for accelerated training.
  • Context Length: Supports an 8192-token context window.

Potential Use Cases

  • Applications requiring a moderately sized, efficient language model.
  • Tasks benefiting from a Llama-architecture base.
  • Scenarios where faster fine-tuning capabilities are advantageous for custom deployments.