---
license: apache-2.0
language:
- en
pipeline_tag: text-generation
tags:
- gguf
- ollama
- tool-calling
- mcp
- lora
base_model:
- Qwen/Qwen2.5-1.5B-Instruct
---

# qwen25-1p5b-git

> ⚠️ **This model is trained for the git MCP server.**
> It is optimised for structured MCP tool-call generation and is not intended
> to be a general-purpose assistant.

GGUF model fine-tuned for structured MCP tool-calling, ready for local inference via [Ollama](https://ollama.com).

## Model Details

| Property | Value |
|---|---|
| MCP server | git |
| Base model (training) | mlx-community/Qwen2.5-1.5B-Instruct-4bit |
| Base model (fused export) | Qwen/Qwen2.5-1.5B-Instruct |
| Fine-tune method | QLoRA / LoRA adapter fusion |
| Quantization | Q5_K_M (5-bit, recommended) |
| GGUF file | qwen25-1p5b-git-q5_k_m.gguf |
| Preset | qwen2_5_1_5b |

## Intended Use

**Intended for agents or apps that call the git MCP server.**

The model was trained on a custom MCP tool-call JSONL dataset derived from the
server's tool schema. It is intended to emit structured JSON tool calls for
that server and is not suited for general chat.

## Quick Start (Ollama)

### Option 1: Direct from Ollama registry (if published)

```bash
ollama pull qwen25-1p5b-git
ollama run qwen25-1p5b-git
```

### Option 2: Download from HuggingFace

\`\`\`bash
# 1. Download the GGUF and Modelfile to the same directory
curl -L -o qwen25-1p5b-git-q5_k_m.gguf https://huggingface.co/lschaffer/qwen2_5_1_5b-git/resolve/main/qwen25-1p5b-git-q5_k_m.gguf
curl -L -o Modelfile https://huggingface.co/lschaffer/qwen2_5_1_5b-git/resolve/main/Modelfile

# 2. Register with Ollama (creates local model from GGUF + Modelfile)
ollama create qwen25-1p5b-git -f Modelfile

# 3. Run the model
ollama run qwen25-1p5b-git
\`\`\`

**Important:** The \`Modelfile\` contains the required TEMPLATE directive for tool support.
Registering the GGUF without the Modelfile will result in "does not support tools" errors.

## Optional Links

- Docs: https://lschaffer.github.io/tealkit
- Source: https://github.com/lschaffer/tealkit

## Files

- `qwen25-1p5b-git-q5_k_m.gguf`
- `Modelfile` — Ollama model definition with system prompt

## Notes

Produced via LoRA fine-tuning on Mac Apple Silicon (MLX), adapter fusion, and llama.cpp GGUF conversion.
See the [training guide](https://github.com/lschaffer/tealkit/blob/master_v2/traning/README.md) for full pipeline details.