Turkcell-LLM-7b-v1

Turkcell-LLM-7b-v1 is a Turkish-optimized large language model based on the Mistral architecture. Fine-tuned using DORA and LORA methods on over 5 billion tokens of Turkish data, it delivers robust natural language understanding and generation tailored for Turkish.

This model has been uploaded by RefinedNeuro and is available on the Ollama platform for local deployment across macOS, Linux, and Windows.

Quick Start

CLI

ollama run RefinedNeuro/Turkcell-LLM-7b-v1

API

curl -X POST http://localhost:11434/api/generate -d '{
  "model": "RefinedNeuro/Turkcell-LLM-7b-v1",
  "prompt": "Türkiye'nin başkenti neresidir?"
}'

License

Licensed under the Apache License 2.0.

Acknowledgements

Uploaded to Ollama by RefinedNeuro.

Official Model Link

https://huggingface.co/TURKCELL/Turkcell-LLM-7b-v1

Official Readme of Model

Turkcell-LLM-7b-v1

This model is an extended version of a Mistral-based Large Language Model (LLM) for Turkish. It was trained on a cleaned Turkish raw dataset containing 5 billion tokens. The training process involved using the DORA method initially. Following this, we utilized Turkish instruction sets created from various open-source and internal resources for fine-tuning with the LORA method.

Model Details

• Base Model: Mistral 7B based LLM
• Tokenizer Extension: Specifically extended for Turkish
• Training Dataset: Cleaned Turkish raw data with 5 billion tokens, custom Turkish instruction sets
• Training Method: Initially with DORA, followed by fine-tuning with LORA

DORA Configuration

• lora_alpha: 128
• lora_dropout: 0.05
• r: 64
• target_modules: “all-linear”

LORA Fine-Tuning Configuration

• lora_alpha: 128
• lora_dropout: 0.05
• r: 256
• target_modules: “all-linear”