Instructions to use modelfix/Qwen2.5-0.5B-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use modelfix/Qwen2.5-0.5B-Instruct with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="modelfix/Qwen2.5-0.5B-Instruct",
	filename="Qwen_Qwen2.5-0.5B-Instruct-converted.gguf",
)

llm.create_chat_completion(
	messages = "No input example has been defined for this model task."
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use modelfix/Qwen2.5-0.5B-Instruct with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf modelfix/Qwen2.5-0.5B-Instruct
# Run inference directly in the terminal:
llama-cli -hf modelfix/Qwen2.5-0.5B-Instruct

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf modelfix/Qwen2.5-0.5B-Instruct
# Run inference directly in the terminal:
llama-cli -hf modelfix/Qwen2.5-0.5B-Instruct

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf modelfix/Qwen2.5-0.5B-Instruct
# Run inference directly in the terminal:
./llama-cli -hf modelfix/Qwen2.5-0.5B-Instruct

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf modelfix/Qwen2.5-0.5B-Instruct
# Run inference directly in the terminal:
./build/bin/llama-cli -hf modelfix/Qwen2.5-0.5B-Instruct

Use Docker

docker model run hf.co/modelfix/Qwen2.5-0.5B-Instruct

LM Studio
Jan
Ollama
How to use modelfix/Qwen2.5-0.5B-Instruct with Ollama:
```
ollama run hf.co/modelfix/Qwen2.5-0.5B-Instruct
```

Unsloth Studio new

How to use modelfix/Qwen2.5-0.5B-Instruct with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for modelfix/Qwen2.5-0.5B-Instruct to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for modelfix/Qwen2.5-0.5B-Instruct to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for modelfix/Qwen2.5-0.5B-Instruct to start chatting

Pi new

How to use modelfix/Qwen2.5-0.5B-Instruct with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf modelfix/Qwen2.5-0.5B-Instruct

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "modelfix/Qwen2.5-0.5B-Instruct"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use modelfix/Qwen2.5-0.5B-Instruct with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf modelfix/Qwen2.5-0.5B-Instruct

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default modelfix/Qwen2.5-0.5B-Instruct

Run Hermes

hermes

Docker Model Runner
How to use modelfix/Qwen2.5-0.5B-Instruct with Docker Model Runner:
```
docker model run hf.co/modelfix/Qwen2.5-0.5B-Instruct
```

Lemonade

How to use modelfix/Qwen2.5-0.5B-Instruct with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull modelfix/Qwen2.5-0.5B-Instruct

Run and chat with the model

lemonade run user.Qwen2.5-0.5B-Instruct-{{QUANT_TAG}}

List all available models

lemonade list

Qwen2.5-0.5B-Instruct / README.md

modelfix-admin

Update README.md

0ac354d verified 2 months ago

preview code

raw

history blame contribute delete

2.09 kB

	---
	license: apache-2.0
	base_model: Qwen/Qwen2.5-0.5B-Instruct
	tags:
	- qwen
	- qwen2.5
	- llama-cpp
	- gguf
	- modelfix
	- quantized
	language:
	- en
	---


	# Qwen2.5-0.5B-Instruct-GGUF (Q4_K_M)
	Optimized by Modelfix.com

	This repository provides a high-efficiency GGUF quantization of Alibaba's Qwen2.5-0.5B-Instruct. This 0.5B model punches significantly above its weight in coding and mathematics.


	## 📊 Quantization Benchmarks
	Our validation process for this Q4_K_M build yielded the following metrics:
	- Perplexity (PPL): 9.249 🙂
	- KL Divergence: 0.025 🤩
	- FLIP Score: 7.8% 🙂


	## ⚙️ Implementation Details
	- Format: GGUF (Quantized to Q4_K_M)
	- Architecture: Transformer-based decoder with RoPE, SwiGLU, RMSNorm, Attention QKV bias, and Tied Word Embeddings.
	- Parameters: 494 Million (0.49B)
	- Context Window: Native 32,768 tokens (Supports generation up to 8,192 tokens).
	- Attention: Grouped Query Attention (GQA) with 14 Query heads and 2 KV heads.
	- Multilingual: Supports over 29+ languages (English, Chinese, French, Spanish, etc.).
	- Strengths: Logic, reasoning, and instruction following in a compact size.
	- Ideal Use Case: Edge devices, mobile applications, and high-speed basic automation.


	## 📊 Hardware Requirements & Performance

	\| Quantization \| File Size \| Recommended VRAM \| Recommended Device \|
	\| :--- \| :--- \| :--- \| :--- \|
	\| Q8_0 \| ~531 MB \| 1.2 GB \| Desktop / Server \|
	\| Q5_K_M \| ~420 MB \| 0.9 GB \| Standard Smartphones \|
	\| Q4_K_M \| ~398 MB \| 0.8 GB \| Low-end Mobile / IoT \|
	\| IQ4_XS \| ~349 MB \| 0.7 GB \| Ultra-constrained Edge \|


	## 💬 Prompt Template
	Qwen2.5 uses the standard ChatML format:

	```text
	<\|im_start\|>system
	You are a helpful assistant.<\|im_end\|>
	<\|im_start\|>user
	{prompt}<\|im_end\|>
	<\|im_start\|>assistant
	```


	## ⚖️ Legal Notice
	This model is a Derivative Work quantized from the original BF16 weights to GGUF format by Modelfix.com.
	It is released under the Apache 2.0 License, matching the original release by the Qwen Team at Alibaba Cloud.

	---
	license: apache-2.0
	base_model: Qwen/Qwen2.5-0.5B-Instruct
	tags:
	- qwen
	- qwen2.5
	- llama-cpp
	- gguf
	- modelfix
	- quantized
	language:
	- en
	---


	# Qwen2.5-0.5B-Instruct-GGUF (Q4_K_M)
	Optimized by Modelfix.com

	This repository provides a high-efficiency GGUF quantization of Alibaba's Qwen2.5-0.5B-Instruct. This 0.5B model punches significantly above its weight in coding and mathematics.


	## 📊 Quantization Benchmarks
	Our validation process for this Q4_K_M build yielded the following metrics:
	- Perplexity (PPL): 9.249 🙂
	- KL Divergence: 0.025 🤩
	- FLIP Score: 7.8% 🙂


	## ⚙️ Implementation Details
	- Format: GGUF (Quantized to Q4_K_M)
	- Architecture: Transformer-based decoder with RoPE, SwiGLU, RMSNorm, Attention QKV bias, and Tied Word Embeddings.
	- Parameters: 494 Million (0.49B)
	- Context Window: Native 32,768 tokens (Supports generation up to 8,192 tokens).
	- Attention: Grouped Query Attention (GQA) with 14 Query heads and 2 KV heads.
	- Multilingual: Supports over 29+ languages (English, Chinese, French, Spanish, etc.).
	- Strengths: Logic, reasoning, and instruction following in a compact size.
	- Ideal Use Case: Edge devices, mobile applications, and high-speed basic automation.


	## 📊 Hardware Requirements & Performance

	\| Quantization \| File Size \| Recommended VRAM \| Recommended Device \|
	\| :--- \| :--- \| :--- \| :--- \|
	\| Q8_0 \| ~531 MB \| 1.2 GB \| Desktop / Server \|
	\| Q5_K_M \| ~420 MB \| 0.9 GB \| Standard Smartphones \|
	\| Q4_K_M \| ~398 MB \| 0.8 GB \| Low-end Mobile / IoT \|
	\| IQ4_XS \| ~349 MB \| 0.7 GB \| Ultra-constrained Edge \|


	## 💬 Prompt Template
	Qwen2.5 uses the standard ChatML format:

	```text
	<\|im_start\|>system
	You are a helpful assistant.<\|im_end\|>
	<\|im_start\|>user
	{prompt}<\|im_end\|>
	<\|im_start\|>assistant
	```


	## ⚖️ Legal Notice
	This model is a Derivative Work quantized from the original BF16 weights to GGUF format by Modelfix.com.
	It is released under the Apache 2.0 License, matching the original release by the Qwen Team at Alibaba Cloud.