Image-Text-to-Text
Safetensors
GGUF
English
llama.cpp
test-fixture
tool-calling
ocr
mtp
pruning
conversational
Instructions to use Serveurperso/small-test with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use Serveurperso/small-test 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 Serveurperso/small-test:F16 # Run inference directly in the terminal: llama cli -hf Serveurperso/small-test:F16
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf Serveurperso/small-test:F16 # Run inference directly in the terminal: llama cli -hf Serveurperso/small-test: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 Serveurperso/small-test:F16 # Run inference directly in the terminal: ./llama-cli -hf Serveurperso/small-test: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 Serveurperso/small-test:F16 # Run inference directly in the terminal: ./build/bin/llama-cli -hf Serveurperso/small-test:F16
Use Docker
docker model run hf.co/Serveurperso/small-test:F16
- LM Studio
- Jan
- vLLM
How to use Serveurperso/small-test with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "Serveurperso/small-test" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Serveurperso/small-test", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }'Use Docker
docker model run hf.co/Serveurperso/small-test:F16
- Ollama
How to use Serveurperso/small-test with Ollama:
ollama run hf.co/Serveurperso/small-test:F16
- Unsloth Desktop
- Pi
How to use Serveurperso/small-test with Pi:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf Serveurperso/small-test:F16
Configure the model in Pi
# Install Pi: npm install -g @earendil-works/pi-coding-agent # Add to ~/.pi/agent/models.json: { "providers": { "llama-cpp": { "baseUrl": "http://localhost:8080/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "Serveurperso/small-test:F16" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Docker Model Runner
How to use Serveurperso/small-test with Docker Model Runner:
docker model run hf.co/Serveurperso/small-test:F16
- Lemonade
How to use Serveurperso/small-test with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull Serveurperso/small-test:F16
Run and chat with the model
lemonade run user.small-test-F16
List all available models
lemonade list
- Hermes Agent
How to use Serveurperso/small-test with Hermes Agent:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf Serveurperso/small-test:F16
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 Serveurperso/small-test:F16
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use Serveurperso/small-test with OpenClaw:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf Serveurperso/small-test:F16
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 "Serveurperso/small-test:F16" \ --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"
Download src/ci_shapes.py from Serveurperso/small-test: direct link, hf CLI and curl.
- Browser
- Download file 8.55 kB
-
https://huggingface.co/Serveurperso/small-test/resolve/main/src/ci_shapes.py
- Command line
-
hf download hf://Serveurperso/small-test/src/ci_shapes.py
-
curl -L -o ci_shapes.py https://huggingface.co/Serveurperso/small-test/resolve/main/src/ci_shapes.py
8.55 kB
| # documents shaped exactly like the llama.cpp CI scenarios, built from the test definitions themselves: | |
| # the five function call script scenarios with their gold trajectories and the server pytest prompts, | |
| # each rendered under several paraphrases so the fixture is reliable in greedy mode on what the CI asks | |
| import json, os, random, sys, importlib | |
| LLAMA_CPP = os.environ.get("LLAMA_CPP","../llama.cpp") | |
| sys.path.insert(0, f"{LLAMA_CPP}/scripts"); sys.path.insert(0, f"{LLAMA_CPP}/tools/server/tests") | |
| import render as R | |
| T = importlib.import_module("server-test-function-call") | |
| from unit.test_tool_call import TEST_TOOL, PYTHON_TOOL, WEATHER_TOOL | |
| def call(name, args, lead=""): return {"role": "assistant", "content": lead, "tool_calls": [{"type": "function", "function": {"name": name, "arguments": args}}]} | |
| def calls(items): return {"role": "assistant", "content": "", "tool_calls": [{"type": "function", "function": {"name": n, "arguments": a}} for n, a in items]} | |
| def result(text): return {"role": "tool", "content": text} | |
| # one gold call sequence per script scenario, results come from the script's own mocks | |
| GOLD = { | |
| "azzoo": [("azzoo_search_products", {"query": "stainless steel tea kettle"}), ("azzoo_get_product", {"product_id": "AZB12345"}), ("azzoo_get_reviews", {"product_id": "AZB12345"})], | |
| "fitness": [("calculate_bmi", {"weight_kg": 82, "height_m": 1.8}), ("get_exercises", {"category": "cardio", "level": "beginner"})], | |
| "geo": [("lookup_ip_geolocation", {"host": "halle-eins.de"}), ("lookup_ip_geolocation", {"host": "galerie-deux.fr"}), ("lookup_ip_geolocation", {"host": "galleria-tre.it"})], | |
| "ev": [("get_stock_quote", {"symbol": "VLTR"}), ("get_stock_quote", {"symbol": "RVXN"}), ("get_security_advisories", {"query": "Voltara"}), ("search_commercial_properties", {"city": "Halverton", "property_type": "garage"}), ("get_video_recommendations", {"query": "fleet electrification"})], | |
| "community": [("get_trending_questions", {"topic": "gardening"}), ("search_mobile_apps", {"query": "gardening", "platform": "ios"})], | |
| } | |
| CASES = {"azzoo": T.AZZOO_TEST_CASE, "fitness": T.FITNESS_TEST_CASE, "geo": T.GEO_TEST_CASE, "ev": T.EV_FLEET_TEST_CASE, "community": T.COMMUNITY_CLASS_TEST_CASE} | |
| SUMMARY = { | |
| "azzoo": "The top-rated stainless steel tea kettle is {p0} (AZB12345). Reviews are positive overall, customers praise the build quality and fast boiling.", | |
| "fitness": "Your BMI is {bmi}. Suggested beginner cardio plan: {ex}.", | |
| "geo": "Hosts outside big cities: {geo}.", | |
| "ev": "Voltara (VLTR) trades above the threshold. Security advisories: {adv}. Commercial garages in Halverton: {props}. Recommended videos on fleet electrification: {vid}.", | |
| "community": "Top gardening question: {q}. Existing iOS apps: {apps}. A class on the most asked topic would fill the gap.", | |
| } | |
| def scenario(rng, key): | |
| case = CASES[key]; msgs = [dict(m) for m in case["messages"]] | |
| steps = GOLD[key]; mocks = case["mock_tool_responses"] | |
| parallel = key in ("geo", "ev") and rng.random() < 0.5 | |
| outs = {} | |
| i = 0 | |
| while i < len(steps): | |
| name, args = steps[i] | |
| group = [(name, args)] | |
| if parallel and key == "geo": group = steps; | |
| elif parallel and key == "ev" and i == 0: group = steps[:2] | |
| if len(group) > 1: | |
| msgs.append(calls(group)) | |
| for n, a in group: | |
| r = mocks[n](a); outs[n] = outs.get(n, []) + [r]; msgs.append(result(r)) | |
| i += len(group) | |
| else: | |
| lead = rng.choice(["", "", f"Calling {name}."]) | |
| msgs.append(call(name, args, lead)); r = mocks[name](args); outs[name] = outs.get(name, []) + [r]; msgs.append(result(r)); i += 1 | |
| def pick(n, k, d="n/a"): | |
| # lists of records are summarized by their name or title fields | |
| try: | |
| o = json.loads(outs[n][0]); v = o.get(k) or next((x for x in o.values() if isinstance(x, list)), d) | |
| except Exception: return d | |
| if isinstance(v, list): return ", ".join(str(x.get("name") or x.get("title") or x.get("id") or x) if isinstance(x, dict) else str(x) for x in v[:3]) | |
| return str(v)[:120] | |
| text = SUMMARY[key].format(p0=pick("azzoo_search_products", "results"), bmi=pick("calculate_bmi", "bmi"), ex=pick("get_exercises", "exercises"), | |
| geo="; ".join(f"{a['host']} at {pick('lookup_ip_geolocation', 'latitude')}, {pick('lookup_ip_geolocation', 'longitude')}" for n, a in steps if key == "geo"), | |
| adv=pick("get_security_advisories", "advisories"), props=pick("search_commercial_properties", "properties"), vid=pick("get_video_recommendations", "videos"), | |
| q=pick("get_trending_questions", "questions"), apps=pick("search_mobile_apps", "apps")) | |
| msgs.append({"role": "assistant", "content": text}) | |
| return msgs, case["tools"] | |
| def pytest_shapes(rng): | |
| k = rng.choice(["required_test", "required_python", "weather", "calc", "hello"]) | |
| if k == "required_test": | |
| return [{"role": "system", "content": "You are a coding assistant."}, {"role": "user", "content": rng.choice(["Write an example", "Write an example.", "Give me an example"])}, call("test", {"success": True}), result("ok"), {"role": "assistant", "content": "Done, the test tool reported success."}], [TEST_TOOL] | |
| if k == "required_python": | |
| return [{"role": "system", "content": "You are a coding assistant."}, {"role": "user", "content": rng.choice(["Write an example", "Write an example.", "Give me an example"])}, call("python", {"code": "print('hello world')"}), result("hello world"), {"role": "assistant", "content": "The example printed hello world."}], [PYTHON_TOOL] | |
| if k == "weather": | |
| loc = rng.choice(["Istanbul", "Istanbul", "Lyon", "Osaka", "Porto", "Istanbul, Turkey"]) | |
| return [{"role": "system", "content": "You are a chatbot that uses tools/functions. Dont overthink things."}, {"role": "user", "content": rng.choice([f"What is the weather in {loc}?", f"What's the weather in {loc}?", f"Weather in {loc}?"])}, call("get_current_weather", {"location": loc}), result(json.dumps({"location": loc, "temperature": rng.randint(5, 30), "unit": "celsius", "condition": rng.choice(["sunny", "cloudy", "rainy"])})), {"role": "assistant", "content": f"It is {rng.choice(['sunny', 'cloudy', 'rainy'])} in {loc.split(',')[0]} right now."}], [WEATHER_TOOL] | |
| if k == "calc": | |
| tools = [{"type": "function", "function": {"name": "calculate", "description": "A calculator function that computes values of arithmetic expressions in the Python syntax", "parameters": {"type": "object", "properties": {"expression": {"type": "string", "description": "An arithmetic expression to compute the value of (Python syntad, assuming all floats)"}}, "required": ["expression"]}}}] | |
| deg = rng.choice([30, 45, 60, 20, 75]); v = rng.choice(["0.55644242476", "0.49999999", "0.70710678", "0.86602540", "0.34202014"]) | |
| return [{"role": "system", "content": "You are a tools-calling assistant. You express numerical values with at most two decimals."}, {"role": "user", "content": f"What's the y coordinate of a point on the unit sphere at angle {deg} degrees?"}, call("calculate", {"expression": f"sin({deg} * pi / 180)"}), result(v), {"role": "assistant", "content": rng.choice([f"The y coordinate is approximately {float(v):.2f}.", f"It is about {float(v):.2f}.", f"y = {float(v):.2f}"])}], tools | |
| return [{"role": "system", "content": "You are a tool-calling agent."}, {"role": "user", "content": rng.choice(["say hello world with python", "Say hello world with python.", "print hello world in python"])}, call("python", {"code": rng.choice(['print("hello world")', "print('hello world')", 'print("Hello, World!")'])}), result("hello world"), {"role": "assistant", "content": "It printed hello world."}], [PYTHON_TOOL] | |
| def gen(seed): | |
| rng = random.Random(seed) | |
| if rng.random() < 0.5: msgs, tools = scenario(rng, rng.choice(list(CASES))) | |
| else: msgs, tools = pytest_shapes(rng) | |
| text, spans = R.render(msgs, tools) | |
| return json.dumps({"text": text, "spans": spans}, ensure_ascii=False) | |
| if __name__ == "__main__": | |
| model_dir, n, out = sys.argv[1], int(sys.argv[2]), sys.argv[3] | |
| R.init(model_dir) | |
| lines = [gen(s) for s in range(n)] | |
| with open(out + ".jsonl", "w") as f: f.write("\n".join(lines[n // 50:]) + "\n") | |
| with open(out + ".eval.jsonl", "w") as f: f.write("\n".join(lines[:n // 50]) + "\n") | |
| print(f"{out}: {n - n // 50} train, {n // 50} eval") | |