Instructions to use samuelfaj/distill2-0.6B-4bit-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use samuelfaj/distill2-0.6B-4bit-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="samuelfaj/distill2-0.6B-4bit-GGUF",
	filename="Users/samuelfajreldines/dev/distill-lora/models/distill2-0.6B-fp16.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use samuelfaj/distill2-0.6B-4bit-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M

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 samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M

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 samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M

Use Docker

docker model run hf.co/samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use samuelfaj/distill2-0.6B-4bit-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "samuelfaj/distill2-0.6B-4bit-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "samuelfaj/distill2-0.6B-4bit-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M

Ollama
How to use samuelfaj/distill2-0.6B-4bit-GGUF with Ollama:
```
ollama run hf.co/samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M
```

Unsloth Studio new

How to use samuelfaj/distill2-0.6B-4bit-GGUF 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 samuelfaj/distill2-0.6B-4bit-GGUF 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 samuelfaj/distill2-0.6B-4bit-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for samuelfaj/distill2-0.6B-4bit-GGUF to start chatting

Pi new

How to use samuelfaj/distill2-0.6B-4bit-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M

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": "samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use samuelfaj/distill2-0.6B-4bit-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M

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 samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M

Run Hermes

hermes

Docker Model Runner
How to use samuelfaj/distill2-0.6B-4bit-GGUF with Docker Model Runner:
```
docker model run hf.co/samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M
```

Lemonade

How to use samuelfaj/distill2-0.6B-4bit-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull samuelfaj/distill2-0.6B-4bit-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.distill2-0.6B-4bit-GGUF-Q4_K_M

List all available models

lemonade list

samuelfaj commited on 4 days ago

Commit

87f899f

verified ·

1 Parent(s): eba1c37

Upload README.md with huggingface_hub

Browse files

Files changed (1) hide show

README.md +95 -0

README.md ADDED Viewed

	@@ -0,0 +1,95 @@

+---
+license: apache-2.0
+language:
+- en
+tags:
+- gguf
+- distill
+- distill-mini
+- cli
+- code
+- compression
+- qwen
+- qwen3
+- expert-model
+- domain-specific
+- task-specialized
+- qlora
+pipeline_tag: text-generation
+base_model: Qwen/Qwen3-0.6B
+---
+# distill2-0.6B — Expert Language Model for CLI Output (GGUF)
+**distill2-0.6B** is the second-generation **domain-specific Expert Language Model** for CLI output compression and classification — in GGUF format for cross-platform use with llama.cpp.
+See [distill2-0.6B-4bit-MLX](https://huggingface.co/samuelfaj/distill2-0.6B-4bit-MLX) for the MLX (Apple Silicon) version.
+## What is distill?
+[distill](https://github.com/samuelfaj/distill) compresses arbitrary command-line output to structured summaries.
+```
+Input:  500 lines of npm install logs
+Output: PASS — 24 packages installed, 0 vulnerabilities
+```
+**distill2-0.6B** achieves **98.4% accuracy** at 0.6B parameters — outperforming its 1.7B predecessor.
+## Files
+| File | Format | Size | Use case |
+|------|--------|------|----------|
+| `distill2-0.6B-Q4_K_M.gguf` | Q4_K_M (4-bit) | 378 MB | Production, low memory |
+| `distill2-0.6B-fp16.gguf` | fp16 | 1.2 GB | Maximum quality |
+## Performance
+| Metric | Value |
+|--------|-------|
+| Overall accuracy | **98.4%** |
+| Tasks at 100% | 5 of 8 |
+| Tasks ≥95% | 7 of 8 |
+| Base model | Qwen3-0.6B |
+| Training | QLoRA 4-bit + GGUF conversion |
+## 8 Specialized Tasks
+| Task | Accuracy | Description |
+|------|----------|-------------|
+| `pass_fail` | 100% | Command success/failure |
+| `safe_review` | 100% | Terraform plan safety |
+| `json_extraction` | 100% | JSON from noisy logs |
+| `test_result` | 100% | Test suite pass/fail |
+| `typescript_check` | 100% | TS compiler errors |
+| `terraform_plan` | 98.4% | Resource change counts |
+| `security_audit` | 96.6% | Vulnerability counts |
+| `generic` | 93.1% | Free-form CLI summaries |
+## Usage (llama.cpp)
+```bash
+# Download
+huggingface-cli download samuelfaj/distill2-0.6B-4bit-GGUF distill2-0.6B-Q4_K_M.gguf --local-dir .
+# Run with llama-cli
+llama-cli -m distill2-0.6B-Q4_K_M.gguf -p "Command output: npm test\n4 passed, 0 failed"
+# Or as server
+llama-server -m distill2-0.6B-Q4_K_M.gguf --port 8080
+```
+## Conversion Pipeline
+This GGUF was created from the QLoRA-trained model via:
+1. Fuse QLoRA adapter into 4-bit base → fp16 with `mlx_lm fuse --dequantize`
+2. Strip MLX quantization artifacts (bias tensors)
+3. Convert to GGUF fp16 with `llama.cpp/convert_hf_to_gguf.py`
+4. Quantize to Q4_K_M with `llama-quantize`
+## Project
+[distill](https://github.com/samuelfaj/distill) — CLI output compression engine.
+[Full Distill Collection](https://huggingface.co/collections/samuelfaj/distill-6a0606f9b131c289025659fc)