Instructions to use satgeze/Qwen3.6-27B-DSpark with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use satgeze/Qwen3.6-27B-DSpark with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="satgeze/Qwen3.6-27B-DSpark",
	filename="Qwen3.6-27B-DSpark.gguf",
)

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

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use satgeze/Qwen3.6-27B-DSpark with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf satgeze/Qwen3.6-27B-DSpark
# Run inference directly in the terminal:
llama cli -hf satgeze/Qwen3.6-27B-DSpark

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf satgeze/Qwen3.6-27B-DSpark
# Run inference directly in the terminal:
llama cli -hf satgeze/Qwen3.6-27B-DSpark

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 satgeze/Qwen3.6-27B-DSpark
# Run inference directly in the terminal:
./llama-cli -hf satgeze/Qwen3.6-27B-DSpark

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 satgeze/Qwen3.6-27B-DSpark
# Run inference directly in the terminal:
./build/bin/llama-cli -hf satgeze/Qwen3.6-27B-DSpark

Use Docker

docker model run hf.co/satgeze/Qwen3.6-27B-DSpark

LM Studio
Jan
Ollama
How to use satgeze/Qwen3.6-27B-DSpark with Ollama:
```
ollama run hf.co/satgeze/Qwen3.6-27B-DSpark
```

Unsloth Studio

How to use satgeze/Qwen3.6-27B-DSpark 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 satgeze/Qwen3.6-27B-DSpark 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 satgeze/Qwen3.6-27B-DSpark to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for satgeze/Qwen3.6-27B-DSpark to start chatting

How to use satgeze/Qwen3.6-27B-DSpark with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf satgeze/Qwen3.6-27B-DSpark

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": "satgeze/Qwen3.6-27B-DSpark"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use satgeze/Qwen3.6-27B-DSpark with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf satgeze/Qwen3.6-27B-DSpark

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 satgeze/Qwen3.6-27B-DSpark

Run Hermes

hermes

Atomic Chat new

OpenClaw new

How to use satgeze/Qwen3.6-27B-DSpark with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf satgeze/Qwen3.6-27B-DSpark

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "satgeze/Qwen3.6-27B-DSpark" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Docker Model Runner
How to use satgeze/Qwen3.6-27B-DSpark with Docker Model Runner:
```
docker model run hf.co/satgeze/Qwen3.6-27B-DSpark
```

Lemonade

How to use satgeze/Qwen3.6-27B-DSpark with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull satgeze/Qwen3.6-27B-DSpark

Run and chat with the model

lemonade run user.Qwen3.6-27B-DSpark-{{QUANT_TAG}}

List all available models

lemonade list

satgeze commited on 6 days ago

Commit

0e3d153

verified ·

1 Parent(s): 64e2ee4

Card: add speedup comparison chart + training loss curve

Browse files

Files changed (3) hide show

README.md +6 -0
speedup.png +0 -0
training_loss.png +0 -0

README.md CHANGED Viewed

@@ -18,6 +18,8 @@ llama.cpp (`--spec-type draft-dspark`) it delivers **1.6-2.7× decode speedup**
 Blackwell at Q8_0, with **zero quality change** (speculative decoding is lossless: every accepted
 token is verified by the target).
 ## Measured results (llama.cpp, RTX Pro 6000 Blackwell, target Qwen3.6-27B-Q8_0, greedy, 200-token completions)
 | metric | value |
@@ -54,6 +56,10 @@ Metal (draft overhead dominates a fast tiny target), while at 27B the head pays
 1.39× on a MacBook is the difference between 12 and 17 tokens per second where it is actually
 felt.
 ## Recipe (reproducible)
 - Data: 12,000 prompts (mlabonne/open-perfectblend), responses regenerated by the target itself

 Blackwell at Q8_0, with **zero quality change** (speculative decoding is lossless: every accepted
 token is verified by the target).
+![decode speed, baseline vs DSpark](speedup.png)
 ## Measured results (llama.cpp, RTX Pro 6000 Blackwell, target Qwen3.6-27B-Q8_0, greedy, 200-token completions)
 | metric | value |
 1.39× on a MacBook is the difference between 12 and 17 tokens per second where it is actually
 felt.
+## Training
+![training loss](training_loss.png)
 ## Recipe (reproducible)
 - Data: 12,000 prompts (mlabonne/open-perfectblend), responses regenerated by the target itself

speedup.png ADDED Viewed

training_loss.png ADDED Viewed