wreckitral/legal-llama3.2-3b-grpo-id

TEXT GENERATIONConcurrent Unit Cost:1Model Size:3.2BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jul 18, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The wreckitral/legal-llama3.2-3b-grpo-id is a 3.2 billion parameter Llama model developed by wreckitral, fine-tuned from wreckitral/legal-llama3.2-3b-sft-id. This model was trained using Unsloth and Huggingface's TRL library, enabling 2x faster training. It is designed for legal applications, leveraging its specialized fine-tuning for relevant tasks. The model has a context length of 32768 tokens.

Loading preview...

wreckitral/legal-llama3.2-3b-grpo-id Overview

This model is a 3.2 billion parameter Llama-based language model developed by wreckitral. It is a fine-tuned version of the wreckitral/legal-llama3.2-3b-sft-id model, specifically optimized for legal applications. A key characteristic of its development is the use of Unsloth and Huggingface's TRL library, which facilitated a 2x faster training process.

Key Capabilities

  • Specialized Legal Domain Understanding: Fine-tuned for tasks within the legal domain, suggesting enhanced performance on legal text analysis, summarization, or question-answering.
  • Efficient Training: Leverages Unsloth for accelerated training, indicating potential for rapid iteration and deployment.
  • Llama Architecture: Built upon the Llama model family, providing a robust foundation for language understanding and generation.
  • Extended Context Window: Features a context length of 32768 tokens, allowing it to process and understand longer legal documents or complex queries.

Good For

  • Legal Text Processing: Ideal for applications requiring deep understanding and generation of legal documents.
  • Legal Research Assistance: Can be used to aid in legal research by processing large volumes of legal information.
  • Domain-Specific NLP: Suitable for developers building natural language processing solutions tailored to the legal sector.