Training pools

Fund a pool with ANM to accelerate the one global model. Trainers and servers earn from your contribution — every reward in ANM, receipted on-chain.

Loading pools…

Train a pool & earn ANM

Contribute GPU compute: your worker claims a shard, trains it, and submits the result. Accepted shards are receipted on-chain and paid in ANM from the pool's reward split (trainers 60% · funders 20% · servers 20%).

1. Install the client (Python 3.10+, a CUDA GPU recommended):
```
pip install --upgrade animica
```

2. Point a worker at a pool — it loops claim → train → submit:

animica ena pool train-loop <pool_id> \
  --worker-id my-gpu-01 \
  --address <your-anm-payout-address>

3. Or run mining + training + serving together with one command:
```
animica up
```

Serving the promoted checkpoint also earns — animica ena pool serve <pool_id>. Track progress with animica ena pool status <pool_id> and the animica ena pool leaderboard <pool_id>.