Instructions to use Shumatsurontek/Qwen3.5-4B-neo with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Shumatsurontek/Qwen3.5-4B-neo with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Shumatsurontek/Qwen3.5-4B-neo")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("Shumatsurontek/Qwen3.5-4B-neo")
model = AutoModelForMultimodalLM.from_pretrained("Shumatsurontek/Qwen3.5-4B-neo", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use Shumatsurontek/Qwen3.5-4B-neo 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 Shumatsurontek/Qwen3.5-4B-neo
# Run inference directly in the terminal:
llama cli -hf Shumatsurontek/Qwen3.5-4B-neo

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Shumatsurontek/Qwen3.5-4B-neo
# Run inference directly in the terminal:
llama cli -hf Shumatsurontek/Qwen3.5-4B-neo

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 Shumatsurontek/Qwen3.5-4B-neo
# Run inference directly in the terminal:
./llama-cli -hf Shumatsurontek/Qwen3.5-4B-neo

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 Shumatsurontek/Qwen3.5-4B-neo
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Shumatsurontek/Qwen3.5-4B-neo

Use Docker

docker model run hf.co/Shumatsurontek/Qwen3.5-4B-neo

LM Studio
Jan

vLLM

How to use Shumatsurontek/Qwen3.5-4B-neo with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Shumatsurontek/Qwen3.5-4B-neo"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Shumatsurontek/Qwen3.5-4B-neo",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Shumatsurontek/Qwen3.5-4B-neo

SGLang

How to use Shumatsurontek/Qwen3.5-4B-neo with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Shumatsurontek/Qwen3.5-4B-neo" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Shumatsurontek/Qwen3.5-4B-neo",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Shumatsurontek/Qwen3.5-4B-neo" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Shumatsurontek/Qwen3.5-4B-neo",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use Shumatsurontek/Qwen3.5-4B-neo with Ollama:
```
ollama run hf.co/Shumatsurontek/Qwen3.5-4B-neo
```

Unsloth Studio

How to use Shumatsurontek/Qwen3.5-4B-neo 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 Shumatsurontek/Qwen3.5-4B-neo 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 Shumatsurontek/Qwen3.5-4B-neo to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Shumatsurontek/Qwen3.5-4B-neo to start chatting

How to use Shumatsurontek/Qwen3.5-4B-neo with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Shumatsurontek/Qwen3.5-4B-neo

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": "Shumatsurontek/Qwen3.5-4B-neo"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

OpenClaw new

How to use Shumatsurontek/Qwen3.5-4B-neo with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Shumatsurontek/Qwen3.5-4B-neo

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 "Shumatsurontek/Qwen3.5-4B-neo" \
  --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 Shumatsurontek/Qwen3.5-4B-neo with Docker Model Runner:
```
docker model run hf.co/Shumatsurontek/Qwen3.5-4B-neo
```

Lemonade

How to use Shumatsurontek/Qwen3.5-4B-neo with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Shumatsurontek/Qwen3.5-4B-neo

Run and chat with the model

lemonade run user.Qwen3.5-4B-neo-{{QUANT_TAG}}

List all available models

lemonade list

Hermes Agent

How to use Shumatsurontek/Qwen3.5-4B-neo with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Shumatsurontek/Qwen3.5-4B-neo

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 Shumatsurontek/Qwen3.5-4B-neo

Run Hermes

hermes

Atomic Chat

Shumatsurontek commited on Apr 7

Commit

6286672

verified ·

1 Parent(s): 4bc4f68

Add benchmark results (mmlu,gsm8k,arc_challenge,hellaswag)

Browse files

Files changed (1) hide show

README.md +13 -130

README.md CHANGED Viewed

@@ -160,136 +160,19 @@ or production use without human review of generated SQL.
 ## Benchmark Results
-Compared against baseline [`Qwen/Qwen3.5-4B`](https://huggingface.co/Qwen/Qwen3.5-4B) using [lm-eval-harness](https://github.com/EleutherAI/lm-evaluation-harness) with vLLM backend on NVIDIA L40S.
-> Note: evaluated on 50 samples per task (quick test).
-| Task | Metric | Baseline | Finetuned | Delta |
-|---|---|---|---|---|
-| **mmlu_human_aging** | acc | 80.0% | **78.0%** | 🔴 -2.0% |
-| **mmlu_human_aging** | acc_stderr | 5.7% | **5.9%** | 🟢 +0.2% |
-| **mmlu_nutrition** | acc | 88.0% | **88.0%** | ⚪ 0.0% |
-| **mmlu_nutrition** | acc_stderr | 4.6% | **4.6%** | ⚪ 0.0% |
-| **mmlu_anatomy** | acc | 68.0% | **66.0%** | 🔴 -2.0% |
-| **mmlu_anatomy** | acc_stderr | 6.7% | **6.8%** | 🟢 +0.1% |
-| **mmlu_high_school_chemistry** | acc | 84.0% | **80.0%** | 🔴 -4.0% |
-| **mmlu_high_school_chemistry** | acc_stderr | 5.2% | **5.7%** | 🟢 +0.5% |
-| **mmlu_professional_medicine** | acc | 86.0% | **78.0%** | 🔴 -8.0% |
-| **mmlu_professional_medicine** | acc_stderr | 5.0% | **5.9%** | 🟢 +1.0% |
-| **mmlu_moral_scenarios** | acc | 46.0% | **42.0%** | 🔴 -4.0% |
-| **mmlu_moral_scenarios** | acc_stderr | 7.1% | **7.0%** | 🔴 -0.1% |
-| **mmlu_high_school_government_and_politics** | acc | 96.0% | **94.0%** | 🔴 -2.0% |
-| **mmlu_high_school_government_and_politics** | acc_stderr | 2.8% | **3.4%** | 🟢 +0.6% |
-| **mmlu_high_school_psychology** | acc | 96.0% | **94.0%** | 🔴 -2.0% |
-| **mmlu_high_school_psychology** | acc_stderr | 2.8% | **3.4%** | 🟢 +0.6% |
-| **mmlu_high_school_statistics** | acc | 62.0% | **60.0%** | 🔴 -2.0% |
-| **mmlu_high_school_statistics** | acc_stderr | 6.9% | **7.0%** | 🟢 +0.1% |
-| **mmlu_high_school_mathematics** | acc | 48.0% | **58.0%** | 🟢 +10.0% |
-| **mmlu_high_school_mathematics** | acc_stderr | 7.1% | **7.0%** | 🔴 -0.1% |
-| **mmlu_abstract_algebra** | acc | 58.0% | **60.0%** | 🟢 +2.0% |
-| **mmlu_abstract_algebra** | acc_stderr | 7.0% | **7.0%** | 🔴 -0.1% |
-| **mmlu_computer_security** | acc | 88.0% | **82.0%** | 🔴 -6.0% |
-| **mmlu_computer_security** | acc_stderr | 4.6% | **5.5%** | 🟢 +0.9% |
-| **mmlu_miscellaneous** | acc | 82.0% | **78.0%** | 🔴 -4.0% |
-| **mmlu_miscellaneous** | acc_stderr | 5.5% | **5.9%** | 🟢 +0.4% |
-| **mmlu_high_school_european_history** | acc | 82.0% | **76.0%** | 🔴 -6.0% |
-| **mmlu_high_school_european_history** | acc_stderr | 5.5% | **6.1%** | 🟢 +0.6% |
-| **mmlu_high_school_computer_science** | acc | 88.0% | **84.0%** | 🔴 -4.0% |
-| **mmlu_high_school_computer_science** | acc_stderr | 4.6% | **5.2%** | 🟢 +0.6% |
-| **mmlu_high_school_physics** | acc | 68.0% | **64.0%** | 🔴 -4.0% |
-| **mmlu_high_school_physics** | acc_stderr | 6.7% | **6.9%** | 🟢 +0.2% |
-| **mmlu_human_sexuality** | acc | 84.0% | **86.0%** | 🟢 +2.0% |
-| **mmlu_human_sexuality** | acc_stderr | 5.2% | **5.0%** | 🔴 -0.3% |
-| **mmlu_machine_learning** | acc | 56.0% | **58.0%** | 🟢 +2.0% |
-| **mmlu_machine_learning** | acc_stderr | 7.1% | **7.0%** | 🔴 -0.0% |
-| **mmlu_astronomy** | acc | 94.0% | **94.0%** | ⚪ 0.0% |
-| **mmlu_astronomy** | acc_stderr | 3.4% | **3.4%** | ⚪ 0.0% |
-| **mmlu_professional_law** | acc | 52.0% | **52.0%** | ⚪ 0.0% |
-| **mmlu_professional_law** | acc_stderr | 7.1% | **7.1%** | ⚪ 0.0% |
-| **mmlu_high_school_us_history** | acc | 90.0% | **84.0%** | 🔴 -6.0% |
-| **mmlu_high_school_us_history** | acc_stderr | 4.3% | **5.2%** | 🟢 +0.9% |
-| **mmlu_management** | acc | 94.0% | **94.0%** | ⚪ 0.0% |
-| **mmlu_management** | acc_stderr | 3.4% | **3.4%** | ⚪ 0.0% |
-| **mmlu_college_computer_science** | acc | 72.0% | **74.0%** | 🟢 +2.0% |
-| **mmlu_college_computer_science** | acc_stderr | 6.4% | **6.3%** | 🔴 -0.1% |
-| **mmlu_virology** | acc | 60.0% | **60.0%** | ⚪ 0.0% |
-| **mmlu_virology** | acc_stderr | 7.0% | **7.0%** | ⚪ 0.0% |
-| **mmlu_jurisprudence** | acc | 82.0% | **86.0%** | 🟢 +4.0% |
-| **mmlu_jurisprudence** | acc_stderr | 5.5% | **5.0%** | 🔴 -0.5% |
-| **mmlu_college_medicine** | acc | 82.0% | **80.0%** | 🔴 -2.0% |
-| **mmlu_college_medicine** | acc_stderr | 5.5% | **5.7%** | 🟢 +0.2% |
-| **mmlu_professional_psychology** | acc | 86.0% | **80.0%** | 🔴 -6.0% |
-| **mmlu_professional_psychology** | acc_stderr | 5.0% | **5.7%** | 🟢 +0.8% |
-| **mmlu_conceptual_physics** | acc | 86.0% | **88.0%** | 🟢 +2.0% |
-| **mmlu_conceptual_physics** | acc_stderr | 5.0% | **4.6%** | 🔴 -0.3% |
-| **mmlu_social_sciences** | acc | 83.0% | **79.3%** | 🔴 -3.7% |
-| **mmlu_social_sciences** | acc_stderr | 1.5% | **1.6%** | 🟢 +0.1% |
-| **mmlu_electrical_engineering** | acc | 82.0% | **84.0%** | 🟢 +2.0% |
-| **mmlu_electrical_engineering** | acc_stderr | 5.5% | **5.2%** | 🔴 -0.2% |
-| **mmlu_international_law** | acc | 90.0% | **88.0%** | 🔴 -2.0% |
-| **mmlu_international_law** | acc_stderr | 4.3% | **4.6%** | 🟢 +0.4% |
-| **mmlu_global_facts** | acc | 42.0% | **36.0%** | 🔴 -6.0% |
-| **mmlu_global_facts** | acc_stderr | 7.0% | **6.9%** | 🔴 -0.2% |
-| **mmlu_high_school_geography** | acc | 90.0% | **80.0%** | 🔴 -10.0% |
-| **mmlu_high_school_geography** | acc_stderr | 4.3% | **5.7%** | 🟢 +1.4% |
-| **mmlu_philosophy** | acc | 82.0% | **82.0%** | ⚪ 0.0% |
-| **mmlu_philosophy** | acc_stderr | 5.5% | **5.5%** | ⚪ 0.0% |
-| **mmlu_world_religions** | acc | 86.0% | **84.0%** | 🔴 -2.0% |
-| **mmlu_world_religions** | acc_stderr | 5.0% | **5.2%** | 🟢 +0.3% |
-| **mmlu_sociology** | acc | 88.0% | **84.0%** | 🔴 -4.0% |
-| **mmlu_sociology** | acc_stderr | 4.6% | **5.2%** | 🟢 +0.6% |
-| **mmlu_stem** | acc | 73.5% | **72.3%** | 🔴 -1.1% |
-| **mmlu_stem** | acc_stderr | 1.4% | **1.4%** | 🟢 +0.0% |
-| **mmlu_marketing** | acc | 92.0% | **92.0%** | ⚪ 0.0% |
-| **mmlu_marketing** | acc_stderr | 3.9% | **3.9%** | ⚪ 0.0% |
-| **mmlu_public_relations** | acc | 74.0% | **64.0%** | 🔴 -10.0% |
-| **mmlu_public_relations** | acc_stderr | 6.3% | **6.9%** | 🟢 +0.6% |
-| **mmlu_college_mathematics** | acc | 60.0% | **52.0%** | 🔴 -8.0% |
-| **mmlu_college_mathematics** | acc_stderr | 7.0% | **7.1%** | 🟢 +0.1% |
-| **mmlu_elementary_mathematics** | acc | 80.0% | **74.0%** | 🔴 -6.0% |
-| **mmlu_elementary_mathematics** | acc_stderr | 5.7% | **6.3%** | 🟢 +0.6% |
-| **mmlu_logical_fallacies** | acc | 82.0% | **74.0%** | 🔴 -8.0% |
-| **mmlu_logical_fallacies** | acc_stderr | 5.5% | **6.3%** | 🟢 +0.8% |
-| **mmlu_medical_genetics** | acc | 90.0% | **92.0%** | 🟢 +2.0% |
-| **mmlu_medical_genetics** | acc_stderr | 4.3% | **3.9%** | 🔴 -0.4% |
-| **mmlu_econometrics** | acc | 66.0% | **64.0%** | 🔴 -2.0% |
-| **mmlu_econometrics** | acc_stderr | 6.8% | **6.9%** | 🟢 +0.1% |
-| **mmlu_us_foreign_policy** | acc | 80.0% | **80.0%** | ⚪ 0.0% |
-| **mmlu_us_foreign_policy** | acc_stderr | 5.7% | **5.7%** | ⚪ 0.0% |
-| **mmlu** | acc | 77.1% | **75.0%** | 🔴 -2.1% |
-| **mmlu** | acc_stderr | 0.8% | **0.8%** | 🟢 +0.0% |
-| **mmlu_college_biology** | acc | 90.0% | **88.0%** | 🔴 -2.0% |
-| **mmlu_college_biology** | acc_stderr | 4.3% | **4.6%** | 🟢 +0.4% |
-| **mmlu_moral_disputes** | acc | 64.0% | **66.0%** | 🟢 +2.0% |
-| **mmlu_moral_disputes** | acc_stderr | 6.9% | **6.8%** | 🔴 -0.1% |
-| **mmlu_high_school_biology** | acc | 90.0% | **88.0%** | 🔴 -2.0% |
-| **mmlu_high_school_biology** | acc_stderr | 4.3% | **4.6%** | 🟢 +0.4% |
-| **mmlu_professional_accounting** | acc | 52.0% | **54.0%** | 🟢 +2.0% |
-| **mmlu_professional_accounting** | acc_stderr | 7.1% | **7.1%** | 🔴 -0.0% |
-| **mmlu_security_studies** | acc | 74.0% | **70.0%** | 🔴 -4.0% |
-| **mmlu_security_studies** | acc_stderr | 6.3% | **6.6%** | 🟢 +0.3% |
-| **mmlu_high_school_macroeconomics** | acc | 74.0% | **68.0%** | 🔴 -6.0% |
-| **mmlu_high_school_macroeconomics** | acc_stderr | 6.3% | **6.7%** | 🟢 +0.4% |
-| **mmlu_high_school_microeconomics** | acc | 88.0% | **88.0%** | ⚪ 0.0% |
-| **mmlu_high_school_microeconomics** | acc_stderr | 4.6% | **4.6%** | ⚪ 0.0% |
-| **mmlu_clinical_knowledge** | acc | 78.0% | **78.0%** | ⚪ 0.0% |
-| **mmlu_clinical_knowledge** | acc_stderr | 5.9% | **5.9%** | ⚪ 0.0% |
-| **mmlu_business_ethics** | acc | 84.0% | **84.0%** | ⚪ 0.0% |
-| **mmlu_business_ethics** | acc_stderr | 5.2% | **5.2%** | ⚪ 0.0% |
-| **mmlu_college_chemistry** | acc | 54.0% | **54.0%** | ⚪ 0.0% |
-| **mmlu_college_chemistry** | acc_stderr | 7.1% | **7.1%** | ⚪ 0.0% |
-| **mmlu_other** | acc | 77.7% | **76.3%** | 🔴 -1.4% |
-| **mmlu_other** | acc_stderr | 1.5% | **1.6%** | 🟢 +0.0% |
-| **mmlu_prehistory** | acc | 80.0% | **76.0%** | 🔴 -4.0% |
-| **mmlu_prehistory** | acc_stderr | 5.7% | **6.1%** | 🟢 +0.4% |
-| **mmlu_humanities** | acc | 76.5% | **73.9%** | 🔴 -2.6% |
-| **mmlu_humanities** | acc_stderr | 1.6% | **1.7%** | 🟢 +0.1% |
-| **mmlu_high_school_world_history** | acc | 90.0% | **86.0%** | 🔴 -4.0% |
-| **mmlu_high_school_world_history** | acc_stderr | 4.3% | **5.0%** | 🟢 +0.7% |
-| **mmlu_college_physics** | acc | 68.0% | **66.0%** | 🔴 -2.0% |
-| **mmlu_college_physics** | acc_stderr | 6.7% | **6.8%** | 🟢 +0.1% |
-| **mmlu_formal_logic** | acc | 68.0% | **64.0%** | 🔴 -4.0% |
-| **mmlu_formal_logic** | acc_stderr | 6.7% | **6.9%** | 🟢 +0.2% |
 ## Citation

 ## Benchmark Results
+Evaluated against baseline [`unsloth/Qwen3.5-4B`](https://huggingface.co/unsloth/Qwen3.5-4B) using [lm-eval-harness](https://github.com/EleutherAI/lm-evaluation-harness) on NVIDIA L40S.
+> Evaluated on 50 samples per task.
+| Benchmark | Baseline | Finetuned | Delta |
+|---|:---:|:---:|:---:|
+| **MMLU: STEM** | 73.5 | 71.8 | 🔴 -1.7 |
+| **MMLU: HUMANITIES** | 76.5 | 73.4 | 🔴 -3.1 |
+| **HELLASWAG** | 48.0 | 52.0 | 🟢 +4.0 |
+| **MMLU** | 77.1 | 74.7 | 🔴 -2.4 |
+| **MMLU: OTHER** | 77.7 | 75.8 | 🔴 -1.8 |
+| **ARC_CHALLENGE** | 60.0 | 60.0 | ⚪ 0.0 |
+| **MMLU: SOCIAL SCIENCES** | 83.0 | 79.7 | 🔴 -3.3 |
 ## Citation