Turkish Function Calling LLM

A specialized Llama 3 8B model fine-tuned for function calling and tool use with strong Turkish language support. This model is provided in GGUF format for efficient local inference across CPU and mobile platforms.

Overview

This skill provides access to the Turkish-Llama-3-8B-function-calling model, a bilingual (Turkish/English) language model specifically trained for function calling scenarios. The model is quantized in GGUF format, enabling deployment on consumer hardware while maintaining high-quality function calling capabilities.

Key Features

Model Specifications

Instructions

Step 1: Choose Quantization Level

Select the appropriate quantization based on your hardware and quality requirements:

**Recommended Quantizations:**

**Quality vs Size Trade-offs:**

Step 2: Download Model Files

Download the chosen quantization from HuggingFace:

**Base URL Pattern:**

```

https://huggingface.co/matrixportal/Turkish-Llama-3-8B-function-calling-GGUF/resolve/main/turkish-llama-3-8b-function-calling-{QUANT}.gguf

```

**Example Downloads:**

Step 3: Set Up Inference Runtime

Choose and install a compatible runtime:

**llama.cpp (Command Line):**

```bash

git clone https://github.com/ggml-org/llama.cpp

cd llama.cpp

make

./llama-cli -m /path/to/model.gguf -p "Your prompt"

```

**Ollama (Simplified):**

```bash

Create Modelfile

echo 'FROM /path/to/turkish-llama-3-8b-function-calling-q4_k_m.gguf' > Modelfile

Import model

ollama create turkish-function-calling -f Modelfile

Run

ollama run turkish-function-calling

```

**LM Studio (GUI):**

1. Download LM Studio from lmstudio.ai

2. Click "Import" and select downloaded GGUF file

3. Load model and start chatting

**GPT4All (Cross-Platform GUI):**

1. Install GPT4All from gpt4all.io

2. Add model via "Download/Import Models"

3. Select imported model and start inference

Step 4: Configure for Function Calling

Structure your prompts to leverage the model's function calling capabilities:

**Prompt Template:**

```

System: You are a helpful assistant with access to the following functions:

[Function definitions in JSON format]

User: [User request in Turkish or English]