Triangulum 5 B G G U F

Introduction

Triangulum 5B is a collection of multilingual large language models (LLMs) designed for complex reasoning tasks. These models are built upon LLaMA's architecture and are instruction-tuned for various applications, supporting multiple languages.

Architecture

Triangulum 5B employs an optimized transformer architecture based on LLaMA's autoregressive language model. This setup enhances performance, making the models suitable for diverse linguistic contexts and tasks. Key features include instruction tuning with supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF) for alignment with human preferences.

Training

The training approach for Triangulum 5B involves:

1. Synthetic Datasets: Leveraging long chain-of-thought synthetic data to boost reasoning capabilities.
2. Supervised Fine-Tuning (SFT): Aligning models to specific tasks using curated datasets.
3. Reinforcement Learning with Human Feedback (RLHF): Iteratively training models to adhere to human values and safety guidelines.

Guide: Running Locally

Step 1: Download the Model

To run Triangulum-5B locally, use Ollama to download the model:

ollama run triangulum-5b-f16.gguf

Step 2: Model Initialization and Download

Ollama will initialize and download necessary model files. Ensure your system has adequate resources for this process.

Step 3: Interact with the Model

Once loaded, you can interact with the model by sending prompts, such as:

>>> What can you do for me?

Step 4: Exit the Program

To exit, type:

/exit

Additional Tips

• Use cloud GPUs for enhanced performance and resource management.
• Quantized models like triangulum-5b-f16.gguf are optimized for resource-constrained hardware.

License

Triangulum 5B is released under the CreativeML OpenRAIL-M license, which governs its use and distribution.