A Bittensor subnet for real-time knowledge verification and grounding. Miners serve verified, up-to-date information about packages, APIs, documentation, and technical facts. Validators score responses based on accuracy, freshness, and completeness.
Lucid brings real-time knowledge to the Bittensor network. Instead of relying on static training data, miners in this subnet provide live-verified answers about:
- Package versions — npm, PyPI, crates.io, and more
- API documentation — current endpoints, parameters, and responses
- Framework changes — breaking changes, migration guides, deprecations
- Technical facts — grounded in live data, not hallucinated
Validators independently verify miner responses against authoritative sources and score them on accuracy, freshness, and response time.
Validator Miner
| |
|--- Query (package/api/doc) ---->|
| |--- Fetch live data
| |--- Verify & format
|<------ Response ----------------|
|
|--- Score (accuracy, freshness)
|--- Set weights on-chain
- Python 3.10+
- Bittensor CLI
git clone https://github.com/get-Lucid/Lucid-Subnet.git
cd Lucid-Subnet
pip install -r requirements.txtpython neurons/miner.py --wallet.name <wallet> --wallet.hotkey <hotkey> --netuid <netuid> --subtensor.network testpython neurons/validator.py --wallet.name <wallet> --wallet.hotkey <hotkey> --netuid <netuid> --subtensor.network testMiners are scored on three dimensions:
- Accuracy — Does the response match the ground truth from authoritative sources?
- Freshness — Is the data current, not stale or outdated?
- Speed — How quickly does the miner respond?
Validators send queries to miners, independently verify the responses, and set weights accordingly through Yuma Consensus.
- Website: getlucid.tech
- Twitter: @getlucid_
- GitHub: get-Lucid
MIT