Instructions to use RecursiveSI/flint with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use RecursiveSI/flint with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="RecursiveSI/flint") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("RecursiveSI/flint") model = AutoModelForCausalLM.from_pretrained("RecursiveSI/flint", device_map="auto") messages = [ {"role": "user", "content": "Who are you?"}, ] inputs = tokenizer.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=256) print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use RecursiveSI/flint with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "RecursiveSI/flint" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "RecursiveSI/flint", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/RecursiveSI/flint
- SGLang
How to use RecursiveSI/flint 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 "RecursiveSI/flint" \ --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": "RecursiveSI/flint", "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 "RecursiveSI/flint" \ --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": "RecursiveSI/flint", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use RecursiveSI/flint with Docker Model Runner:
docker model run hf.co/RecursiveSI/flint
Download sglang/example_tools.py from RecursiveSI/flint: direct link, hf CLI and curl.
- Browser
- Download file 7.46 kB
-
https://huggingface.co/RecursiveSI/flint/resolve/main/sglang/example_tools.py
- Command line
-
hf download hf://RecursiveSI/flint/sglang/example_tools.py
-
curl -L -o example_tools.py https://huggingface.co/RecursiveSI/flint/resolve/main/sglang/example_tools.py
7.46 kB
| """Tool calls with nested argument schemas against a Flint SGLang server. | |
| pip install openai | |
| python3 example_tools.py --base-url http://localhost:30000/v1 [--api-key KEY] | |
| The serve scripts start SGLang with --tool-call-parser flint, which constrains each assistant turn that | |
| has tools to Flint's tool-call format: tool names must be ones the request lists, and the arguments of a | |
| tool sent with "strict": true must match its JSON schema (types, enums, patterns, numeric bounds, | |
| required fields). The tools below send "strict": true; other tools keep the format but not the schema. | |
| Each call gets a canned result (tool_result below), which Flint's final answers report. | |
| """ | |
| import argparse | |
| import json | |
| import os | |
| from openai import OpenAI | |
| SAMPLING = {"temperature": 1.0, "top_p": 0.95, "extra_body": {"top_k": 20}} | |
| MAX_TOKENS = 32768 # reasoning counts toward max_tokens | |
| TIME = {"type": "string", "pattern": "^([01][0-9]|2[0-3]):[0-5][0-9]$"} | |
| WEATHER = {"paris": ("sunny", 21), "tokyo": ("cloudy", 17), "lima": ("light rain", 19)} # canned (conditions, °C) | |
| TOOLS = [ | |
| { | |
| "type": "function", | |
| "function": { | |
| "name": "create_event", | |
| "strict": True, | |
| "description": "Create a calendar event.", | |
| "parameters": { | |
| "type": "object", | |
| "properties": { | |
| "title": {"type": "string"}, | |
| "date": {"type": "string", "pattern": "^[0-9]{4}-[0-9]{2}-[0-9]{2}$", "description": "YYYY-MM-DD"}, | |
| "start_time": {**TIME, "description": "24-hour HH:MM"}, | |
| "duration_minutes": {"type": "integer", "minimum": 5, "maximum": 480}, | |
| "attendees": { | |
| "type": "array", | |
| "minItems": 1, | |
| "items": { | |
| "type": "object", | |
| "properties": {"name": {"type": "string"}, "email": {"type": "string"}}, | |
| "required": ["name", "email"], | |
| "additionalProperties": False, | |
| }, | |
| }, | |
| "reminders_minutes": {"type": "array", "items": {"type": "integer", "minimum": 0}}, | |
| "priority": {"type": "string", "enum": ["LOW", "NORMAL", "HIGH_PRIORITY"]}, | |
| }, | |
| "required": ["title", "date", "start_time", "duration_minutes", "attendees", "priority"], | |
| "additionalProperties": False, | |
| }, | |
| }, | |
| }, | |
| { | |
| "type": "function", | |
| "function": { | |
| "name": "add_to_cart", | |
| "strict": True, | |
| "description": "Add products to the shopping cart.", | |
| "parameters": { | |
| "type": "object", | |
| "properties": { | |
| "items": { | |
| "type": "array", | |
| "minItems": 1, | |
| "items": { | |
| "type": "object", | |
| "properties": { | |
| "sku": {"type": "string", "pattern": "^[A-Z]{3}-[0-9]{4}$"}, | |
| "quantity": {"type": "integer", "minimum": 1, "maximum": 99}, | |
| "gift_wrap": {"type": "boolean"}, | |
| }, | |
| "required": ["sku", "quantity", "gift_wrap"], | |
| "additionalProperties": False, | |
| }, | |
| }, | |
| "coupon": {"type": "string", "description": "Coupon code; omit when there is none"}, | |
| }, | |
| "required": ["items"], | |
| "additionalProperties": False, | |
| }, | |
| }, | |
| }, | |
| { | |
| "type": "function", | |
| "function": { | |
| "name": "get_weather", | |
| "strict": True, | |
| "description": "Get the current weather for a city.", | |
| "parameters": { | |
| "type": "object", | |
| "properties": { | |
| "city": {"type": "string"}, | |
| "unit": {"type": "string", "enum": ["celsius", "fahrenheit"]}, | |
| }, | |
| "required": ["city", "unit"], | |
| "additionalProperties": False, | |
| }, | |
| }, | |
| }, | |
| ] | |
| REQUESTS = [ | |
| "Book 'Q4 planning' on October 21st, 2026 at 2:30 PM for 45 minutes with Ana (ana@example.com) and Bo " | |
| "(bo@example.com). Remind us 10 minutes and one hour before. It is urgent.", | |
| "Add two gift-wrapped mugs (SKU MUG-0042) and one notebook (SKU NTB-1001, no gift wrap) to my cart. No coupon.", | |
| "What's the weather in Paris, Tokyo and Lima? Use celsius.", | |
| ] | |
| def main() -> None: | |
| parser = argparse.ArgumentParser() | |
| parser.add_argument("--base-url", default="http://localhost:30000/v1") | |
| parser.add_argument("--api-key", default=os.environ.get("FLINT_API_KEY", "unused")) | |
| args = parser.parse_args() | |
| client = OpenAI(base_url=args.base_url, api_key=args.api_key) | |
| for request in REQUESTS: | |
| run(client, request) | |
| def run(client: OpenAI, request: str) -> None: | |
| print(f"\n>>> {request}") | |
| messages = [{"role": "user", "content": request}] | |
| choice = client.chat.completions.create( | |
| model="flint", messages=messages, tools=TOOLS, max_tokens=MAX_TOKENS, **SAMPLING | |
| ).choices[0] | |
| if choice.finish_reason == "length": | |
| print(f"cut off: the turn used all {MAX_TOKENS} tokens (reasoning counts toward max_tokens)") | |
| return | |
| reply = choice.message | |
| if not reply.tool_calls: | |
| print("content:", reply.content) | |
| return | |
| for call in reply.tool_calls: | |
| print(f"{call.function.name}:", json.dumps(json.loads(call.function.arguments), ensure_ascii=False)) | |
| # Echo the assistant turn back unchanged, answer every call, and let Flint finish the turn. | |
| messages.append(reply.model_dump(exclude_none=True)) | |
| for call in reply.tool_calls: | |
| result = tool_result(call.function.name, json.loads(call.function.arguments)) | |
| print(f"{call.function.name} result:", json.dumps(result, ensure_ascii=False)) | |
| messages.append({"role": "tool", "tool_call_id": call.id, "content": json.dumps(result)}) | |
| final = client.chat.completions.create( | |
| model="flint", messages=messages, tools=TOOLS, max_tokens=MAX_TOKENS, **SAMPLING | |
| ) | |
| print("final answer:", final.choices[0].message.content) | |
| def tool_result(name: str, arguments: dict) -> dict: | |
| """Canned result for one call, standing in for a real calendar, shop and weather service.""" | |
| if name == "get_weather": | |
| conditions, celsius = WEATHER.get(arguments["city"].lower(), ("clear", 20)) | |
| temperature = celsius if arguments["unit"] == "celsius" else round(celsius * 9 / 5 + 32) | |
| return { | |
| "city": arguments["city"], | |
| "conditions": conditions, | |
| "temperature": temperature, | |
| "unit": arguments["unit"], | |
| } | |
| if name == "create_event": | |
| return { | |
| "status": "created", | |
| "event_id": "evt_1042", | |
| "title": arguments["title"], | |
| "date": arguments["date"], | |
| "start_time": arguments["start_time"], | |
| } | |
| return { | |
| "status": "added", | |
| "cart_id": "cart_77", | |
| "items_added": sum(item["quantity"] for item in arguments["items"]), | |
| } | |
| if __name__ == "__main__": | |
| main() | |