Updating README

README.md

@@ -9,14 +9,54 @@ tags:
 license: apache-2.0
 language:
 - en
+- tr
 ---
 
 # Uploaded  model
 
-- **Developed by:** Qutiba
+- **Developed by:** Özgür Entegrasyon
 - **License:** apache-2.0
 - **Finetuned from model :** meta-llama/Llama-3.1-8B-Instruct
 
-This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
+OZGURLUK-GPT-LinuxGeneral is a fine-tuned language model developed specifically to assist with Linux system administration, 
+network management, and database troubleshooting. By using a custom dataset focused on technical tasks, this model excels 
+at interpreting Linux commands and offering useful solutions. Whether you're troubleshooting a server, configuring a network,
+or managing databases, this model offers step-by-step guidance.
 
-[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
+The model is available in three formats: 4-bit, 8-bit, and 16-bit. The 8-bit version strikes a balance between performance 
+and memory usage, while the 16-bit version offers higher accuracy and precision for more demanding tasks. The 4-bit version
+provides a lightweight alternative optimized for low-resource environments. Additionally, the HUGG format (GGUF) ensures 
+that the model has a smaller memory footprint with fast load times.
+
+
+---
+
+## Dataset
+
+The dataset used for fine-tuning **OZGURLUK-GPT-LinuxGeneral** was developed internally, gathering knowledge from various sources and internet databases. 
+
+A total of **56,466 question-answer pairs** were collected, covering a broad range of technical domains including **Kubernetes**, **Linux**, **PostgreSQL**, **Docker**, and many others. The dataset is organized under different "ticker" headings, allowing the grouping of similar questions and answers within each domain. This enables the model to better understand the nuances of technical problem-solving and offer relevant responses tailored to users’ specific needs in system administration and other related fields.
+
+The inclusion of diverse technical topics ensures that the model can assist with various tasks and adapt to a variety of Linux system-related challenges, providing highly accurate, domain-specific solutions.
+
+
+
+--
+
+## Key Features
+
+- **Optimized for Linux:** Trained with a dataset of Linux system commands, troubleshooting solutions, and network configurations.
+- **Multi-format Support:** Choose between 4-bit, 8-bit, 16-bit, or HUGG for optimal performance based on your hardware.
+- **Efficient Performance:** Trained using **Unsloth** and **TRL**, achieving faster training and efficient inference.
+- **Custom Dataset:** Includes 56,466 question-answer pairs across multiple technical domains such as Kubernetes, Linux, PostgreSQL, Docker, etc.
+
+---
+
+## Training Details
+
+- **Base Model:** meta-llama/Llama-3.1-8B-Instruct  
+- **Fine-tuning Method:** LoRA (Low-Rank Adaptation)  
+- **Training Hardware:** Google Colab with A100 GPUs  
+- **Dataset:** Custom-developed dataset of 56,466 question-answer pairs from various technical fields (Kubernetes, Linux, PostgreSQL, Docker, etc.).
+
+---