Welcome to our guide on the Orbita-v0.1 model, a cutting-edge Turkish Large Language Model (LLM) designed to tackle various tasks, from coding to math. In this article, we will walk you through how to use this remarkable tool step by step.

What is Orbita-v0.1?

Orbita-v0.1 is an advanced Turkish LLM built upon the Qwen 14B model. It has been meticulously finetuned on a carefully annotated Turkish dataset, empowering it to follow instructions effectively and generate coherent responses. The model has been optimized to perform a variety of tasks efficiently across the Turkish language spectrum.

Key Features

• Multi-Task Capability: Handles numerous tasks, from text generation to answering complex queries.
• High Accuracy: Demonstrates impressive accuracy rates across multiple evaluations, including AI2 Reasoning Challenge and TruthfulQA.
• User-Friendly: Simplifies interactions through straightforward usage examples.

Setup and Installation

To get started with Orbita-v0.1, you’ll need to install the required libraries and load the model. Follow the steps below:

python
from transformers import AutoModelForCausalLM, AutoTokenizer
device = "cuda"  # the device to load the model on
model = AutoModelForCausalLM.from_pretrained("Orbina/Orbita-v0.1", torch_dtype="auto", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained("Orbina/Orbita-v0.1")

Using the Model

Once you have successfully loaded the model, utilizing it for generating responses is straightforward. Let’s explore an analogy to grasp this better:

Imagine Orbita-v0.1 as a highly skilled translator who can also cook delicious meals based on your requests. You simply provide a prompt (or recipe) and the model (the chef) whips up a response. Here’s how:

python
prompt = "Pythonda ekrana Merhaba Dünya nasıl yazılır?"
messages = [
    {"role": "system", "content": "Sen, Orbina AI tarafından üretilen ve verilen talimatları takip ederek en iyi cevabı üretmeye çalışan yardımcı bir yapay zekasın."},
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
model_inputs = tokenizer([text], return_tensors="pt").to(device)
generated_ids = model.generate(
    model_inputs.input_ids,
    temperature=0.3,
    top_k=50,
    top_p=0.9,
    max_new_tokens=512,
    repetition_penalty=1,
)
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]

This code snippet sets the stage for generating a response to the prompt about printing “Hello World” in Python.

Example Usage

Here’s a practical example of how to generate responses:

python
Prompt: Türkiye'deki finansal kurumlar, yapay zeka ile birlikte gelen veri güvenliği risklerini nasıl yönetiyor, 3 örnek verir misin?
Cevap: Türkiye'deki finansal kurumlar, yapay zeka teknolojilerinin getirdiği veri güvenliği risklerini çeşitli yöntemlerle ele almakta ve bu riskleri azaltmaya çalışmaktadır. İşte üç örnek:
1. Veri Şifreleme: Müşteri bilgileri gibi hassas verileri şifreleyerek koruma sağlar.
2. Anomali Tespit Sistemleri: Beklenmedik davranışları algılar ve güvenlik uyarıları yapar.
3. Sürekli İzleme: Güvenlik politikalarını sürekli kontrol eder ve uygular.

Troubleshooting

If you encounter issues while using the Orbita-v0.1 model, here are some troubleshooting tips:

• Model Not Found: Ensure you have the correct model name and it is accessible from Hugging Face.
• CUDA Errors: Check that your environment supports GPU and that the device is correctly set.
• Memory Issues: Reduce the batch size or use a less complex model if you run into memory limits.

For more insights, updates, or to collaborate on AI development projects, stay connected with fxis.ai.

Conclusion

Orbita-v0.1 is a powerful tool for leveraging the capabilities of AI in the Turkish language. Its multi-task proficiency and ease of use make it an invaluable resource for developers and researchers alike. At fxis.ai, we believe that such advancements are crucial for the future of AI, as they enable more comprehensive and effective solutions. Our team is continually exploring new methodologies to push the envelope in artificial intelligence, ensuring that our clients benefit from the latest technological innovations.

Further Evaluation

To see how Orbita-v0.1 ranks, check the following evaluation results, which provide insight on its performance across various tasks:

Open LLM Turkish Leaderboard Evaluation Results

• AI2 Reasoning Challenge_tr: 41.97
• HellaSwag_tr: 48.00
• MMLU_tr: 49.51
• TruthfulQA_tr: 50.78
• Winogrande_tr: 56.16
• GSM8k_tr: 50.41