rakdey/HRA-Non-Instruct-Merged
TEXT GENERATIONConcurrent Unit Cost:1Model Size:1BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jul 18, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold
The rakdey/HRA-Non-Instruct-Merged is a 1 billion parameter Llama-3.2-based causal language model developed by rakdey. It was fine-tuned using Unsloth and Huggingface's TRL library, enabling 2x faster training. With a 32768 token context length, this model is optimized for efficient processing and generation tasks.
Loading preview...
Overview
The rakdey/HRA-Non-Instruct-Merged is a 1 billion parameter language model, fine-tuned by rakdey. It is based on the unsloth/llama-3.2-1b-unsloth-bnb-4bit architecture, indicating its foundation in the Llama family of models. A key aspect of its development is the use of Unsloth and Huggingface's TRL library, which significantly accelerated its training process, reportedly making it 2x faster.
Key Capabilities
- Efficient Training: Leverages Unsloth for accelerated fine-tuning, reducing development time and computational resources.
- Llama-3.2 Base: Benefits from the underlying architecture of Llama-3.2, providing a strong foundation for language understanding and generation.
- Context Length: Features a substantial context window of 32768 tokens, allowing it to process and generate longer sequences of text.
Good For
- Resource-Efficient Applications: Ideal for scenarios where faster training and smaller model size are critical.
- Experimental Fine-tuning: Provides a solid base for further experimentation and fine-tuning on specific datasets or tasks.
- Applications requiring extended context: Suitable for tasks that benefit from processing longer inputs or generating more extensive outputs, thanks to its large context window.