Bulgarian Tool-Use Language Model (27B GGUF)

A 27B parameter language model specialized for Bulgarian language with advanced function calling and tool use capabilities. This model is based on Tucan-27B-v1.0 and has been quantized into GGUF format for efficient local deployment.

Description

This skill provides access to LLMBG-ToolUse-27B-v1.0, a Bulgarian language model optimized for:

The model is available in multiple quantization levels (Q2_K through Q8_0) to balance quality and resource requirements.

Instructions

When a user requests to use this Bulgarian tool-use model:

1. **Explain the model capabilities**

- Inform the user this is a 27B parameter model specialized for Bulgarian language

- Highlight its function calling and tool use capabilities

- Mention it's quantized in GGUF format for local deployment

- Note it supports MCP integration

2. **Help select the appropriate quantization**

Based on user's hardware capabilities, recommend:

- **Q4_K_M** (16.7 GB): Fast, recommended for most users with 20+ GB RAM

- **Q5_K_M** (19.5 GB): Better quality, requires 24+ GB RAM

- **Q6_K** (22.4 GB): Very good quality, requires 28+ GB RAM

- **Q8_0** (29.0 GB): Best quality, requires 36+ GB RAM

- **Q2_K** (10.5 GB): Smallest, lower quality but runs on limited hardware

- **Q3_K_M** (13.5 GB): Lower quality compromise

3. **Provide download instructions**

```bash

# Example for Q4_K_M quantization

wget https://huggingface.co/mradermacher/LLMBG-ToolUse-27B-v1.0-GGUF/resolve/main/LLMBG-ToolUse-27B-v1.0.Q4_K_M.gguf

```

4. **Explain usage with common GGUF runners**

**llama.cpp:**

```bash

./main -m LLMBG-ToolUse-27B-v1.0.Q4_K_M.gguf -p "Your prompt here" -n 512

```

**Ollama:**

```bash

# Create Modelfile

echo "FROM ./LLMBG-ToolUse-27B-v1.0.Q4_K_M.gguf" > Modelfile

ollama create llmbg-tooluse -f Modelfile

ollama run llmbg-tooluse

```

**LM Studio or Jan:**

- Import the GGUF file through the UI

- Load the model and start chatting

5. **Highlight tool-use features**

- This model is specifically trained for function calling

- Explain how to structure function/tool definitions in prompts

- Mention MCP compatibility for context protocol integration

6. **Note the base model license**

- Licensed under Gemma license terms

- Base model: llm-bg/Tucan-27B-v1.0

- Quantized by mradermacher

Example Usage

**User request:** "I need a Bulgarian language model that can call functions"

**Response:**

"I recommend the LLMBG-ToolUse-27B-v1.0 model - a 27B parameter Bulgarian language model optimized for function calling and tool use.

For most users, I suggest the Q4_K_M quantization (16.7 GB) which offers a good balance of quality and performance. You'll need at least 20 GB of RAM.

Download it with:

```bash

wget https://huggingface.co/mradermacher/LLMBG-ToolUse-27B-v1.0-GGUF/resolve/main/LLMBG-ToolUse-27B-v1.0.Q4_K_M.gguf

```

Then run with llama.cpp, Ollama, LM Studio, or Jan. This model excels at Bulgarian language understanding and can integrate with MCP for enhanced context handling."

Important Notes