Skip to content

Mining.md

Latest commit

 

History

History
161 lines (105 loc) · 4.66 KB

Mining.md

File metadata and controls

161 lines (105 loc) · 4.66 KB

Miner Guide

Table of Contents

  1. Installation 🔧

  2. Mining ⛏️

Before You Proceed ⚠️

IMPORTANT: If you are new to Bittensor, we recommend familiarizing yourself with the basics on the Bittensor Website before proceeding.

Ensure you are running Subtensor locally to minimize outages and improve performance. See Run a Subtensor Node Locally.

Be aware of the minimum compute requirements for our subnet, detailed in Minimum compute YAML configuration. A GPU is recommended for training, but not required for inference while running a miner.

Installation

Download the repository and navigate to the folder:

git clone https://github.com/natixnetwork/natix-subnet.git && cd natix-subnet

We recommend using a Poetry environment with Python 3.11 to manage dependencies:

curl -sSL https://install.python-poetry.org | python3 -
poetry env use 3.11
poetry install

This will install all necessary dependencies and prepare your environment.

Data

Only required for training -- deployed miner instances do not need access to these datasets.

Optionally, pre-download the training datasets by running:

poetry run python base_miner/datasets/download_data.py

The default list of datasets and their download location is defined in base_miner/config.py.

Mining Requirements ⚠️

To mine on our subnet, you must have a registered hotkey and have submitted at least one model.

Registration

To reduce the risk of deregistration due to technical issues or poor model performance, we recommend the following:

  1. Test your miner on testnet before mining on mainnet.
  2. Before registering your hotkey on mainnet, verify that your port is open:
curl your_ip:your_port

Mainnet

btcli s register --netuid 72 --wallet.name [wallet_name] --wallet.hotkey [wallet.hotkey] --subtensor.network finney

Testnet

btcli s register --netuid 323 --wallet.name [wallet_name] --wallet.hotkey [wallet.hotkey] --subtensor.network test

Mining

Run ./setup_env.sh to generate a miner.env file with default configuration.

Make sure to update your miner.env file with your wallet name, hotkey, miner port, and model configuration.

# following are initial values
IMAGE_DETECTOR=ViT
IMAGE_DETECTOR_CONFIG=ViT_roadwork.yaml
VIDEO_DETECTOR=TALL
VIDEO_DETECTOR_CONFIG=tall.yaml

# Device Settings
IMAGE_DETECTOR_DEVICE=cpu # Options: cpu, cuda
VIDEO_DETECTOR_DEVICE=cpu

NETUID=323                           # 323 for testnet, 72 for mainnet
SUBTENSOR_NETWORK=test               # Networks: finney, test, local
SUBTENSOR_CHAIN_ENDPOINT=wss://test.finney.opentensor.ai:443
                                     # Endpoints:
                                     # - wss://entrypoint-finney.opentensor.ai:443
                                     # - wss://test.finney.opentensor.ai:443/
                                     

# Wallet Configuration
WALLET_NAME=
WALLET_HOTKEY=

# Miner Settings
MINER_AXON_PORT=8091
BLACKLIST_FORCE_VALIDATOR_PERMIT=True # Force validator permit for blacklisting

Then, start your miner with:

chmod +x ./start_miner.sh
./start_miner.sh

This will launch run_neuron.py using Poetry.

You can also optionally run a cache updater service to improve image caching performance:

chmod +x ./start_cache_updater.sh
./start_cache_updater.sh

This invokes natix/validator/scripts/run_cache_updater.py in the background.

Deploy Your Model

Update your miner.env file to use your trained detector class and configuration.

  • Detector types are defined in base_miner/registry.py.
  • Config files live in base_miner/detectors/configs/.
  • UCFDetector requires a train_config path in its config YAML.

Weights should be placed under base_miner/<detector_type>/weights. If missing, they will be pulled from Hugging Face according to the hf_repo field in the config.

Training

To improve beyond the baseline model, experiment with new datasets, architectures, or hyperparameters.

TensorBoard

Start TensorBoard to view training metrics:

tensorboard --logdir=./base_miner/checkpoints/<experiment_name>

For remote machines:

ssh -L 7007:localhost:6006 your_username@your_ip

Then on the remote machine:

tensorboard --logdir=./base_miner/checkpoints/<experiment_name> --host 0.0.0.0 --port 6006

View it locally at http://localhost:7007