LLama-3.1-Thinkable: Bilingual AI Expert in Mathematics and Programming

LLama-3.1-Thinkable is a fine-tuned version of LLama 3.1, specifically designed to excel in bilingual (Turkish and English) communication, advanced mathematics, and programming tasks. This model combines enhanced reasoning capabilities with strong multilingual proficiency, offering a cutting-edge solution for users in diverse fields.

---

🚀 Features

1. Bilingual Expertise

   • Fluent in both Turkish and English.
   • Designed to seamlessly understand and respond in either language.
   • Ideal for users who switch between these languages or require multilingual solutions.

2. Mathematics Mastery

   • Excels in solving advanced mathematical problems, including algebra, calculus, and statistics.
   • Provides step-by-step explanations for better understanding.

3. Programming Proficiency

   • Supports a wide range of programming languages, including Python, JavaScript, C++, and more.
   • Assists with debugging, algorithm design, and code optimization.
   • Generates clear and efficient code snippets for complex tasks.

4. Thinkable AI: Enhanced Reasoning

   • Fine-tuned for improved logical and critical thinking.
   • Capable of breaking down complex concepts into understandable insights.

---

🔧 Technical Details

• Base Model: LLama 3.1
• Fine-tuning Dataset:
  • High-quality bilingual datasets (Turkish-English).
  • Specialized datasets for mathematics and programming tasks.
• Parameter Count: 5.25B & 8B

---

📚 Use Cases

• Education:

  • Learn programming and advanced mathematics with detailed explanations.
  • Solve bilingual academic tasks in Turkish and English.

• Development:

  • Generate production-ready code.
  • Debug complex applications and find optimized solutions.

• AI Research:

  • Experiment with a high-performance bilingual model in NLP tasks.

---

🛠️ How to Use

Here’s how you can get started with LLama-3.1-Thinkable:

Installation

pip install transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "aixr/llama-3.1-thinkable"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

# Generate a response
inputs = tokenizer("Explain recursion in programming:", return_tensors="pt")
outputs = model.generate(inputs["input_ids"], max_length=200)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))