Instructions to use GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF",
	filename="Qwen3.5-9B-NSC-ACE-SABER.Q2_K.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 GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-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 GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-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 GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-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 GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF:Q4_K_M

Use Docker

docker model run hf.co/GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF:Q4_K_M

LM Studio
Jan
Ollama
How to use GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF with Ollama:
```
ollama run hf.co/GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF:Q4_K_M
```

Unsloth Studio new

How to use GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-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 GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-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 GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF to start chatting

Pi new

How to use GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-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": "GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-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 GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-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 GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF:Q4_K_M

Run Hermes

hermes

Docker Model Runner
How to use GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF with Docker Model Runner:
```
docker model run hf.co/GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF:Q4_K_M
```

Lemonade

How to use GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull GestaltLabs/Qwen3.5-9B-NSC-ACE-SABER-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Qwen3.5-9B-NSC-ACE-SABER-GGUF-Q4_K_M

List all available models

lemonade list

Qwen3.5-9B-NSC-ACE-SABER-GGUF / assets /bfcl_argument_gains.svg

DJLougen

Add NSC-ACE benchmark plots

81c892a verified 7 days ago

preview code

raw

history blame contribute delete

5.39 kB

	<svg xmlns="http://www.w3.org/2000/svg" width="1100" height="540" viewBox="0 0 1100 540" role="img" aria-labelledby="title desc">
	<title id="title">BFCL v3 subset function-calling improvements</title>
	<desc id="desc">Grouped horizontal bar chart comparing base Qwen3.5-9B and Qwen3.5-9B NSC-ACE on BFCL v3 subset metrics.</desc>
	<defs>
	<linearGradient id="baseGrad" x1="0" x2="1" y1="0" y2="0">
	<stop offset="0%" stop-color="#a8a29e"/>
	<stop offset="100%" stop-color="#78716c"/>
	</linearGradient>
	<linearGradient id="aceGrad" x1="0" x2="1" y1="0" y2="0">
	<stop offset="0%" stop-color="#38bdf8"/>
	<stop offset="100%" stop-color="#2563eb"/>
	</linearGradient>
	<filter id="shadow" x="-10%" y="-30%" width="120%" height="160%">
	<feDropShadow dx="0" dy="2" stdDeviation="2" flood-color="#0f172a" flood-opacity="0.16"/>
	</filter>
	</defs>
	<rect width="1100" height="540" rx="18" fill="#f8fafc"/>
	<rect x="28" y="28" width="1044" height="484" rx="16" fill="#ffffff" stroke="#e2e8f0"/>
	<text x="56" y="76" font-family="Inter, Segoe UI, Arial, sans-serif" font-size="30" font-weight="800" fill="#0f172a">BFCL v3 Function-Calling Subset</text>
	<text x="56" y="108" font-family="Inter, Segoe UI, Arial, sans-serif" font-size="16" fill="#475569">40 cases: 20 simple + 20 multiple. Correct required calls and argument quality.</text>
	<g font-family="Inter, Segoe UI, Arial, sans-serif" font-size="14" fill="#475569">
	<rect x="765" y="64" width="20" height="12" rx="3" fill="url(#baseGrad)"/>
	<text x="794" y="75">Qwen3.5-9B base</text>
	<rect x="765" y="90" width="20" height="12" rx="3" fill="url(#aceGrad)"/>
	<text x="794" y="101">NSC-ACE adapter</text>
	</g>

	<g stroke="#e2e8f0" stroke-width="1">
	<line x1="300" y1="145" x2="300" y2="465"/>
	<line x1="445" y1="145" x2="445" y2="465"/>
	<line x1="590" y1="145" x2="590" y2="465"/>
	<line x1="735" y1="145" x2="735" y2="465"/>
	<line x1="880" y1="145" x2="880" y2="465"/>
	</g>
	<g font-family="Inter, Segoe UI, Arial, sans-serif" font-size="12" fill="#64748b">
	<text x="292" y="485">0</text>
	<text x="432" y="485">25%</text>
	<text x="577" y="485">50%</text>
	<text x="722" y="485">75%</text>
	<text x="865" y="485">100%</text>
	</g>

	<g font-family="Inter, Segoe UI, Arial, sans-serif">
	<g transform="translate(0,155)">
	<text x="56" y="23" font-size="15" font-weight="700" fill="#0f172a">Exact required call accuracy</text>
	<rect x="300" y="5" width="392" height="16" rx="8" fill="url(#baseGrad)" filter="url(#shadow)"/>
	<rect x="300" y="27" width="435" height="16" rx="8" fill="url(#aceGrad)" filter="url(#shadow)"/>
	<text x="705" y="18" font-size="13" fill="#334155">67.5%</text>
	<text x="748" y="40" font-size="13" font-weight="700" fill="#1d4ed8">75.0%</text>
	<text x="930" y="31" font-size="13" font-weight="700" fill="#2563eb">+7.5 pts</text>
	</g>
	<g transform="translate(0,220)">
	<text x="56" y="23" font-size="15" font-weight="700" fill="#0f172a">Function name accuracy</text>
	<rect x="300" y="5" width="566" height="16" rx="8" fill="url(#baseGrad)" filter="url(#shadow)"/>
	<rect x="300" y="27" width="566" height="16" rx="8" fill="url(#aceGrad)" filter="url(#shadow)"/>
	<text x="879" y="18" font-size="13" fill="#334155">97.5%</text>
	<text x="879" y="40" font-size="13" font-weight="700" fill="#1d4ed8">97.5%</text>
	<text x="930" y="31" font-size="13" font-weight="700" fill="#475569">flat</text>
	</g>
	<g transform="translate(0,285)">
	<text x="56" y="23" font-size="15" font-weight="700" fill="#0f172a">Required argument names</text>
	<rect x="300" y="5" width="480" height="16" rx="8" fill="url(#baseGrad)" filter="url(#shadow)"/>
	<rect x="300" y="27" width="533" height="16" rx="8" fill="url(#aceGrad)" filter="url(#shadow)"/>
	<text x="793" y="18" font-size="13" fill="#334155">82.7%</text>
	<text x="846" y="40" font-size="13" font-weight="700" fill="#1d4ed8">91.8%</text>
	<text x="930" y="31" font-size="13" font-weight="700" fill="#2563eb">+9.2 pts</text>
	</g>
	<g transform="translate(0,350)">
	<text x="56" y="23" font-size="15" font-weight="700" fill="#0f172a">Required argument values</text>
	<rect x="300" y="5" width="421" height="16" rx="8" fill="url(#baseGrad)" filter="url(#shadow)"/>
	<rect x="300" y="27" width="473" height="16" rx="8" fill="url(#aceGrad)" filter="url(#shadow)"/>
	<text x="734" y="18" font-size="13" fill="#334155">72.5%</text>
	<text x="786" y="40" font-size="13" font-weight="700" fill="#1d4ed8">81.6%</text>
	<text x="930" y="31" font-size="13" font-weight="700" fill="#2563eb">+9.2 pts</text>
	</g>
	<g transform="translate(0,415)">
	<text x="56" y="23" font-size="15" font-weight="700" fill="#0f172a">Tool-call rate</text>
	<rect x="300" y="5" width="566" height="16" rx="8" fill="url(#baseGrad)" filter="url(#shadow)"/>
	<rect x="300" y="27" width="566" height="16" rx="8" fill="url(#aceGrad)" filter="url(#shadow)"/>
	<text x="879" y="18" font-size="13" fill="#334155">97.5%</text>
	<text x="879" y="40" font-size="13" font-weight="700" fill="#1d4ed8">97.5%</text>
	<text x="930" y="31" font-size="13" font-weight="700" fill="#475569">flat</text>
	</g>
	</g>
	</svg>