Triangulum 1 B G G U F

Introduction

Triangulum-1B-GGUF is a collection of multilingual large language models (LLMs) designed for generative tasks. These models are pretrained and instruction-tuned to handle complex reasoning tasks using synthetic datasets.

Architecture

The models are built on LLaMA's autoregressive language model, utilizing an optimized transformer architecture to enhance performance. Key features include instruction tuning with supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF) to align outputs with human preferences. They support multiple languages, making them versatile across diverse linguistic contexts.

Training

Training incorporates:

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

Guide: Running Locally

Example 1: Running the Triangulum-1B Model

To run the Triangulum-1B model locally using Ollama:

Step 1: Download the Model
Use the command:

ollama run triangulum-1b-f16.gguf

Step 2: Model Initialization and Download
Ollama will automatically download and initialize the model files.

Step 3: Interact with the Model
Send prompts to interact with the model, which can assist with tasks like answering questions, generating ideas, providing writing assistance, and more.

Step 4: Exit the Program
Type /exit to terminate the session.

Cloud GPUs

For enhanced performance, consider running the model on cloud GPUs, which can offer the necessary computational power for efficient processing.

License

Triangulum-1B-GGUF is licensed under the creativeml-openrail-m license, facilitating its use in various applications while ensuring adherence to specified terms.