TAO $223.71+2.6% 24h
27

Orion

sn27Storage & InfraWatch49Avoid20

No description set on the registered repo.

SILX-LABS/Orion· pushed 1mo ago emissions paused reg. open
Emission
0
paused on chain
Alpha
0.0100
τ · mcap 34.2K
Top-slot payout
τ per winning epoch · live
Stars
2
live from the GitHub API
Primary language
Python
repo-reported
Last push
1mo ago
feeds the dormancy integrity signal
Topics
repo-declared tags

About the repo

what the project says about itself — the input for semantic labels

SILX-LABS/Orion· pushed 1mo ago

No description set on the repo.

Reading this tab

scope of the data

Taonets fetches the repo's public metadata (description, topics, language, stars, last push) and the head of its README. Commit-count histories, contributor lists and release notes require the GitHub commits/releases endpoints, which the worker doesn't consume yet — those panels are coming with the ingest upgrade and are not simulated in the meantime.

README head

first lines of the default branch README, unedited

# Orion

Orion is a Bittensor data-mining subnet. Campaigns define the data a model
needs, miners turn assigned source data into useful model-ready datasets, and
validators reward only verified submissions.

Quasar is the first consumer. The current miner compares the campaign-pinned
hosted judge with [`silx-ai/Quasar-Preview`](https://huggingface.co/silx-ai/Quasar-Preview),
which runs locally on the miner GPU.

## Network

- Chain: Bittensor Finney
- Netuid: `27`
- Publisher: `5H6BqkzjYvViiqp7rQLXjpnaEmW7U9CoKxXhQ4efMqtX1mQw`
- Storage region: `eu-central-1`
- Bucket: `orion-mainnet-979769297074-eu-central-1`
- Prefix: `orion/sn27-mainnet-pilot-001`

## How It Works

- The orchestrator publishes signed campaigns and assignments, issues scoped
  storage grants, collects submissions, and advances customer orders.
- Miners process assigned data, use the hosted judge, run Quasar locally, and
  upload one signed dataset snapshot.
- Validators validate submitted work, independently replay a hidden sample,
  calculate rewards, and set Bittensor weights.

The current campaign covers coding, mathematics, technical data, and general
knowledge. Miners receive real batches rather than submitting individual
records.

## Rewards

Only valid, useful work is rewarded. Each accepted submission is scored from
verified quality and useful token volume, then multiplied by the miner's signed
availability history. Invalid, duplicated, stale, or unverifiable work receives
zero.

## Install

Orion requires Python 3.11–3.13 and
[`uv`](https://docs.astral.sh/uv/getting-started/installation/).

```bash
git clone https://github.com/SILX-LABS/Orion.git /opt/orion
cd /opt/orion
```

`uv run --frozen` installs the locked dependencies automatically. On its first
assignment, the miner also downloads the campaign-pinned Quasar revision from
Hugging Face and reuses its verified local cache.

## Miner

```bash
CHUTES_API_KEY=<key> uv run --frozen --extra quasar orion miner run \
  --network finney --netuid 27 \
  --wallet <wallet> --hotkey <hotkey> \
  --publisher 5H6BqkzjYvViiqp7rQLXjpnaEmW7U9CoKxXhQ4efMqtX1mQw \
  --bucket orion-mainnet-979769297074-eu-central-1 \
  --region eu-central-1 --prefix orion/sn27-mainnet-pilot-001 \
  --devices 0
```

## Validator

```bash
CHUTES_API_KEY=<key> uv run --frozen orion validator run \
  --network finney --netuid 27 \
  --wallet <wallet> --hotkey <hotkey> \
  --publisher 5H6BqkzjYvViiqp7rQLXjpnaEmW7U9CoKxXhQ4efMqtX1mQw \
  --bucket orion-mainnet-979769297074-eu-central-1 \
  --region eu-central-1 --prefix orion/sn27-mainnet-pilot-001 \
  --bundle-root /var/lib/orion/validator-bundles \
  --set-weights --allow-genesis
```

Use `--allow-genesis` only for the validator's first empty-history launch.

## Orchestrator

The subnet operator supplies one private launch configuration and runs:

```bash
CHUTES_API_KEY=<key> AWS_PROFILE=<profile> \
  uv run --frozen --extra quasar --extra source-ingestion \
  orion orchestrator start --config /path/to/mainnet.json \
  --execute-paid-canary
```

The paid canary flag is required only for the first launch.
GitHub · Taonets