Welcome to our guide on Trinity! A powerful Large Language Model (LLM) series developed by Migel Tissera. Whether you are aiming to leverage this innovative tool for coding assistance or simply wish to experiment with its capabilities, this article will walk you through the essentials of using Trinity with ease.

What is Trinity?

Trinity is a cutting-edge LLM designed specifically for coding tasks. This model, particularly noted for its mistralai codestral version, works by interpreting and generating text related to programming. It was developed with fierce dedication, and its compute resources were generously sponsored by KindoAI.

Setting Up Trinity with Python

To use Trinity, you’ll employ a simple Python script. This script loads the model and prepares it for generating responses based on your coding queries. Below is a sample script to get you started.

python
import torch, json
from transformers import AutoModelForCausalLM, AutoTokenizer

model_path = "migel_tissera/Trinity-2-Codestral-22B"
model = AutoModelForCausalLM.from_pretrained(
    model_path,
    torch_dtype=torch.float16,
    device_map="auto",
    load_in_4bit=True,
    trust_remote_code=False,
)
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=False)

terminators = tokenizer.convert_tokens_to_ids("