Instructions to use datasysdev/ann-sparseattention with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use datasysdev/ann-sparseattention with Transformers:

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("datasysdev/ann-sparseattention", dtype="auto")

llama-cpp-python

How to use datasysdev/ann-sparseattention with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="datasysdev/ann-sparseattention",
	filename="gguf/Qwen3-4B-Instruct-2507-F16-ann-6layer-k128-v2.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 datasysdev/ann-sparseattention with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf datasysdev/ann-sparseattention:F16
# Run inference directly in the terminal:
llama-cli -hf datasysdev/ann-sparseattention:F16

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf datasysdev/ann-sparseattention:F16
# Run inference directly in the terminal:
llama-cli -hf datasysdev/ann-sparseattention:F16

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 datasysdev/ann-sparseattention:F16
# Run inference directly in the terminal:
./llama-cli -hf datasysdev/ann-sparseattention:F16

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 datasysdev/ann-sparseattention:F16
# Run inference directly in the terminal:
./build/bin/llama-cli -hf datasysdev/ann-sparseattention:F16

Use Docker

docker model run hf.co/datasysdev/ann-sparseattention:F16

LM Studio
Jan
Ollama
How to use datasysdev/ann-sparseattention with Ollama:
```
ollama run hf.co/datasysdev/ann-sparseattention:F16
```

Unsloth Studio new

How to use datasysdev/ann-sparseattention 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 datasysdev/ann-sparseattention 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 datasysdev/ann-sparseattention to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for datasysdev/ann-sparseattention to start chatting

Pi new

How to use datasysdev/ann-sparseattention with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf datasysdev/ann-sparseattention:F16

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": "ann-sparseattention"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use datasysdev/ann-sparseattention with Docker Model Runner:
```
docker model run hf.co/datasysdev/ann-sparseattention:F16
```

Lemonade

How to use datasysdev/ann-sparseattention with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull datasysdev/ann-sparseattention:F16

Run and chat with the model

lemonade run user.ann-sparseattention-F16

List all available models

lemonade list

ann-sparseattention / artifacts /scaling_plot.svg

datasysdev

Correct asymptotic scoring analysis

0989d52 verified 1 day ago

preview code

raw

history blame contribute delete

3.6 kB

	<svg xmlns="http://www.w3.org/2000/svg" width="820" height="460" viewBox="0 0 820 460">
	<rect width="100%" height="100%" fill="white"/>
	<line x1="86" y1="390" x2="792" y2="390" stroke="#333"/>
	<line x1="86" y1="32" x2="86" y2="390" stroke="#333"/>
	<line x1="86.0" y1="390" x2="86.0" y2="395" stroke="#333"/>
	<text x="86.0" y="414" text-anchor="middle" font-size="12">4K</text>
	<line x1="298.5" y1="390" x2="298.5" y2="395" stroke="#333"/>
	<text x="298.5" y="414" text-anchor="middle" font-size="12">32K</text>
	<line x1="440.2" y1="390" x2="440.2" y2="395" stroke="#333"/>
	<text x="440.2" y="414" text-anchor="middle" font-size="12">128K</text>
	<line x1="581.9" y1="390" x2="581.9" y2="395" stroke="#333"/>
	<text x="581.9" y="414" text-anchor="middle" font-size="12">512K</text>
	<line x1="650.3" y1="390" x2="650.3" y2="395" stroke="#333"/>
	<text x="650.3" y="414" text-anchor="middle" font-size="12">1M</text>
	<line x1="792.0" y1="390" x2="792.0" y2="395" stroke="#333"/>
	<text x="792.0" y="414" text-anchor="middle" font-size="12">4M</text>
	<line x1="81" y1="429.1" x2="86" y2="429.1" stroke="#333"/>
	<text x="76" y="433.1" text-anchor="end" font-size="12">10K</text>
	<line x1="86" y1="429.1" x2="792" y2="429.1" stroke="#eee"/>
	<line x1="81" y1="350.0" x2="86" y2="350.0" stroke="#333"/>
	<text x="76" y="354.0" text-anchor="end" font-size="12">100K</text>
	<line x1="86" y1="350.0" x2="792" y2="350.0" stroke="#eee"/>
	<line x1="81" y1="270.9" x2="86" y2="270.9" stroke="#333"/>
	<text x="76" y="274.9" text-anchor="end" font-size="12">1M</text>
	<line x1="86" y1="270.9" x2="792" y2="270.9" stroke="#eee"/>
	<line x1="81" y1="191.8" x2="86" y2="191.8" stroke="#333"/>
	<text x="76" y="195.8" text-anchor="end" font-size="12">10M</text>
	<line x1="86" y1="191.8" x2="792" y2="191.8" stroke="#eee"/>
	<line x1="81" y1="112.8" x2="86" y2="112.8" stroke="#333"/>
	<text x="76" y="116.8" text-anchor="end" font-size="12">100M</text>
	<line x1="86" y1="112.8" x2="792" y2="112.8" stroke="#eee"/>
	<polyline fill="none" stroke="#555555" stroke-width="3" points="86.0,293.9 156.8,270.1 227.7,246.3 298.5,222.5 369.4,198.7 440.2,174.9 511.1,151.1 581.9,127.3 650.3,104.3 721.2,80.5 792.0,56.7" />
	<polyline fill="none" stroke="#c0392b" stroke-width="3" points="86.0,365.3 156.8,341.5 227.7,317.7 298.5,293.9 369.4,270.1 440.2,246.3 511.1,222.5 581.9,198.7 650.3,175.7 721.2,151.9 792.0,128.1" />
	<polyline fill="none" stroke="#1f77b4" stroke-width="3" points="86.0,231.7 156.8,228.9 227.7,226.3 298.5,224.0 369.4,221.8 440.2,219.7 511.1,217.7 581.9,215.8 650.3,214.1 721.2,212.5 792.0,210.9" />
	<line x1="526.6" y1="32" x2="526.6" y2="390" stroke="#777" stroke-dasharray="5 5"/>
	<text x="534.6" y="201.8" font-size="12" fill="#555">Quest/HNSW crossover ~297K</text>
	<text x="410" y="22" text-anchor="middle" font-size="16" font-weight="600">Candidate-scoring operations per query</text>
	<text x="410.0" y="442" text-anchor="middle" font-size="13">Context length (tokens, log scale)</text>
	<text x="18" y="230.0" transform="rotate(-90 18,230.0)" text-anchor="middle" font-size="13">Scalar ops (log scale)</text>
	<rect x="515" y="52" width="258" height="82" fill="white" stroke="#ddd"/>
	<line x1="532" y1="72" x2="572" y2="72" stroke="#555555" stroke-width="3"/>
	<text x="582" y="76" font-size="13">Full attention scoring</text>
	<line x1="532" y1="96" x2="572" y2="96" stroke="#c0392b" stroke-width="3"/>
	<text x="582" y="100" font-size="13">Quest page scan</text>
	<line x1="532" y1="120" x2="572" y2="120" stroke="#1f77b4" stroke-width="3"/>
	<text x="582" y="124" font-size="13">HNSW over learned search</text>
	</svg>