Loading...
Loading
Overview
Registration Activity
Neuron registration trends
This Interval
0
— burn · — PoW
Target / Interval
Loading
Neuron registration trends
SubQuery Hermes Subnet is a decentralized Bittensor subnet that creates infrastructure for GraphQL query processing and blockchain data indexing. Miners compete to provide fast, accurate responses to GraphQL queries across SubQuery projects and The Graph subgraphs, while validators generate synthetic challenges and evaluate performance based on factual accuracy and response speed to distribute TAO rewards.
The subnet operates with validators and miners in a competitive incentive loop. Validators fetch active projects from a backend service, initialize GraphQL agents for each project, and generate synthetic challenges based on project schemas. Miners receive synthetic challenges and process them by routing to either custom project-specific tools or falling back to GraphQL agents, returning optimized responses.
Validators score miner responses on factual accuracy and speed, while also evaluating sampled organic requests. Scores are aggregated using EMA and converted to weights submitted to the Bittensor chain every 30 minutes. Miners can optimize rewards by developing custom tools for specific projects to reduce response latency while maintaining accuracy.
The system maintains dual scoring: Combined Score (for chain rewards) balances quality and workload; Quality Score Ranking (for miner selection) uses pure quality to fairly distribute organic request routing.
Synthetic challenges test factual accuracy on a 0-10 scale with response time weighted quadratically relative to validator's ground truth time. Organic challenges are sampled (default 1 in 5) with quality tracked via EMA (alpha=0.7) and workload tracked via 3-hour sliding buckets. Final scores combine synthetic performance and organic labor/quality.
Validators maintain two scoring systems: Combined Score (submitted to chain for rewards) incorporating synthetic labor, quality, organic labor, and organic quality; and Quality Score Ranking (pure quality metrics for miner selection on organic requests). Scores are aggregated using EMA and converted to Bittensor weights every 30 minutes. Final weight calculation: combined_score = alpha × quality_score + (1-alpha) × normalized_workload, where alpha (default 0.5) controls quality vs workload balance.
The sections above were auto-generated from the GitHub repository and may not reflect the latest changes. View repository
AIs debate until they are AGI.
This information was auto-extracted from the GitHub repository and may be incomplete or out of date. Always refer to the subnet's GitHub repository for the latest instructions.
Holder & total-staked metrics as of 9h ago (hourly snapshot)
DDR4 RAM recommended. SSD storage with minimum 1000 IOPS (5000 IOPS recommended). Public IP required for external communication.
DDR4 RAM recommended. SSD storage with minimum 1000 IOPS (5000 IOPS recommended). Public IP required. Sufficient stake (TAO) required to become a valid validator.
Install uv package manager and Python 3.13
curl -LsSf https://astral.sh/uv/install.sh | sh && uv python install 3.13
Clone repository and sync dependencies
git clone [email protected]:subquery/network-hermes-subnet.git && cd network-hermes-subnet && uv sync
Activate virtual environment and install btcli
source .venv/bin/activate && uv pip install bittensor-cli
Create a new coldkey wallet for the miner
btcli wallet new_coldkey --wallet.name miner
Create a new hotkey for the miner
btcli wallet new_hotkey --wallet.name miner --wallet.hotkey default
Register the miner on the subnet (requires TAO for burn fee)
btcli subnet register --wallet.name miner --wallet.hotkey default
Create and configure .env.miner with SUBTENSOR_NETWORK, WALLET_NAME, HOTKEY, EXTERNAL_IP, PORT, OPENAI_API_KEY, LLM_MODEL, and MINER_LLM_MODEL
cp .env.miner.example .env.miner && nano .env.miner
Start the miner process
python -m neurons.miner
Install uv package manager and Python 3.13
curl -LsSf https://astral.sh/uv/install.sh | sh && uv python install 3.13
Clone repository and sync dependencies
git clone [email protected]:subquery/network-hermes-subnet.git && cd network-hermes-subnet && uv sync
Activate virtual environment and install btcli
source .venv/bin/activate && uv pip install bittensor-cli
Create a new coldkey wallet for the validator
btcli wallet new_coldkey --wallet.name validator
Create a new hotkey for the validator
btcli wallet new_hotkey --wallet.name validator --wallet.hotkey default
Register the validator on the subnet (requires TAO for burn fee)
btcli subnet register --wallet.name validator --wallet.hotkey default
Self-stake TAO to become a valid validator (optional but recommended)(optional)
btcli stake add --wallet.name validator --wallet.hotkey default
Create and configure .env.validator with SUBTENSOR_NETWORK, WALLET_NAME, HOTKEY, EXTERNAL_IP, PORT, OPENAI_API_KEY, LLM_MODEL, and SCORE_LLM_MODEL
cp .env.validator.example .env.validator && nano .env.validator
Start the validator process
python -m neurons.validator
