Skip to content

Latest commit

 

History

History
188 lines (122 loc) · 4.86 KB

API_docs.md

File metadata and controls

188 lines (122 loc) · 4.86 KB

Subnet 1 API Documentation

This document describes the API endpoints available for Subnet 1

Table of Contents

Getting Started

SN1 can run either in validator mode or in API mode. Both modes will require the validator hotkey.

As a validator, you MUST be running one instance in validator mode and can launch an arbitrary number of API instances. These API instances will proxy the responses from miners to the validator for scoring.

To set up and run the API server:

  1. Install dependencies: Ensure all required dependencies are installed using Poetry.
  2. Set up the .env.api file: Copy the .env.api.example file to .env.api and fill in the validator hotkey.
  3. Run the API server: Start the server to access the API endpoints.

Use the following command:

# Run the API server
bash run_api.sh

API Management

Create API Key

Endpoint: POST /api_management/create-api-key/

Description: Creates a new API key with a specified rate limit.

Parameters:

  • rate_limit (query, required): The rate limit for the API key (integer).
  • admin-key (header, required): Admin key for authorization (string) defined by validator in .env.validator.

Modify API Key

Endpoint: PUT /api_management/modify-api-key/{api_key}

Description: Modifies the rate limit of an existing API key.

Parameters:

  • api_key (path, required): The API key to modify (string).
  • rate_limit (query, required): The new rate limit for the API key (integer).
  • admin-key (header, required): Admin key for authorization (string) defined by validator in .env.validator.

Delete API Key

Endpoint: DELETE /api_management/delete-api-key/{api_key}

Description: Deletes an existing API key.

Parameters:

  • api_key (path, required): The API key to delete (string).
  • admin-key (header, required): Admin key for authorization (string) defined by validator in .env.validator.

Miner Availabilities

Get Miner Availabilities

Endpoint: POST /miner_availabilities/miner_availabilities

Description: Fetches miner availabilities based on provided UIDs.

Request Body:

  • JSON array of integers or null (optional).

Get Available Miners

Endpoint: GET /miner_availabilities/get_available_miners

Description: Retrieves a list of available miners for a specific task and/or model.

Parameters:

  • task (query, optional): The type of task (e.g., QuestionAnsweringTask, Programming, etc.).
  • model (query, optional): The specific model (string).
  • k (query, optional): The maximum number of results to return (integer, default: 10).

Chat Endpoints

Proxy Chat Completions

Endpoint: POST /v1/chat/completions

Description: Proxies chat completions to an underlying GPT model.

Parameters:

  • api-key (header, required): API key for authorization (string).

Example call using the OpenAI client:

def make_header(api_key: str):
    return {
        "api-key": f"{api_key}",
        "Content-Type": "application/json",
    }

client = openai.AsyncOpenAI(
        base_url=f"http://213.173.105.104:11198/v1",
        max_retries=0,
        timeout=Timeout(30, connect=10, read=20),
        http_client=openai.DefaultAsyncHttpxClient(
            headers=make_header(API_KEY)  # Pass the headers here
        ),
        api_key=API_KEY
)

result = await client.chat.completions.create(
    model="hugging-quants/Meta-Llama-3.1-70B-Instruct-AWQ-INT4",
    messages=[
            {"role": "user", "content": """How are you?"""},
        ],
    stream=True,
    temperature=0.7,
    extra_body={
        "task": "InferenceTask",
        "sampling_parameters": {
            "mixture": False,
            "max_new_tokens": 256,
            "do_sample": True,
        },
    },
    seed=42,
    extra_headers=make_header(API_KEY),
)

You can pass "mixture": True in the extra_body to use SN1's mixture of miners mode.

Web Retrieval

Endpoint: GET /web_retrieval

Description: Retrieves a list websites about a search query

Parameters:

  • search_query (str): The search term you'd like to look up
  • n_miners (int, optional): How many miners to query
  • uids: (list[int], optional): which specific uids to query (Deprecated)

Health

Endpoint: GET /health

Description: Health check endpoint.