The first L1 with native Federated Learning consensus.
Verify · Compensate · Trace — on any device, any FL framework.
Live Nodes TPS Finality Adversarial Rust License
Federated Learning works in research. Only 5.2% reaches production deployment.
The reason isn't the ML — it's three structural problems no FL framework solves:
Malicious or low-quality updates corrupt the global model. No way to verify what nodes submit.
Rational actors take the global model without contributing quality data. Honest participants subsidize freeloaders.
Who contributed? What data? Impossible to audit. No accountability for what trained the model.
No existing L1 supports FL trust primitives natively. Smart contracts are too slow and don't see model weights. So we built consensus that does.
Four primitives, one consensus layer:
L1 — Any Device
0ドル hardware cost
ESP32 · Raspberry Pi
Smartphone · Server
Anything already on.
L2 — Verify
On-chain quality gate
PoU scores every update.
Bad data rejected at consensus.
Global model stays clean.
L3 — Compensate
Trustless incentive layer
Smart contract pays SAVI
proportional to quality.
85% to node contributors.
L4 — Trace
Immutable provenance
Every round hashed on-chain.
Timestamp · participants.
Model origin always verifiable.
TensorFlow Federated · PyTorch · Flower · Custom all run on top.
Like Avalanche for subnets — built from scratch because no existing L1 could.
Real cloud infrastructure. Measured, not projected.
| Metric | Value |
|---|---|
| Active nodes | 25 |
| TPS — peak | 45,725 |
| TPS — sustained | 1,285 |
| Finality | 485 ms |
| Adversarial test | 0 crash / 48h |
| Production users | 1 enterprise (1 year live) |
Try it now: Explorer · Faucet · Discord for testnet SAVI
PoW rewards capital invested in hardware. PoS rewards capital invested in tokens. Federated Learning needs to reward quality of contribution — and neither existing consensus can.
| Proof-of-Work | Proof-of-Stake | Proof-of-Unity | |
|---|---|---|---|
| Who wins? | Biggest mining farm | Richest validator | Best-performing node |
| Energy cost | Nation-sized | Efficient | Efficient |
| Centralization | Hardware cartels | Whale dominance | Meritocratic |
| Min hardware | ASIC (10ドルK+) | Stake (varies, often 32ドルK+) | 0ドル — any device |
| Finality | ~60 min | ~12 sec | 485 ms |
| FL-aware? | No | No | Yes — at consensus level |
This is the part most "AI + blockchain" projects miss. They bolt a smart contract on top of an existing chain — so the chain only ever sees a hash of the model. The aggregation, the quality scoring, and the trust decisions all happen off-chain in some operator's server. You're back to trusting the operator.
Savitri's PoU is dual-purpose by design:
The PoU score elects the next block proposer.
BFT with 2/3 quorum finalizes in 485 ms.
Standard L1 behavior — but no PoW farms, no whale validators.
The same PoU score weights model updates inside the federated round.
Low-Integrity nodes get their gradients down-weighted or rejected.
The aggregation runs at consensus layer — not in a smart contract, not off-chain.
One mechanism, two functions. The score that decides who proposes the next block is the same score that decides whose FL contribution counts. No off-chain aggregator. No "trust the coordinator." Bad model updates and bad block proposals are punished by the same rule.
Score = α·S(t-1) + (1-α)·[ 0.25·Availability + 0.20·Latency
+ 0.20·Integrity + 0.20·Reputation
+ 0.15·Participation ]
For FL workloads, the Integrity dimension scores the quality of model updates a node submits — measured against held-out validation rounds and cross-node agreement. A node that consistently submits poisoned or low-quality gradients loses score, gets down-weighted in aggregation, and stops earning rewards.
No slashing of stake — bad behavior simply doesn't pay, in either job.
| Repository | What it does | |
|---|---|---|
| ⚙️ | savitri-network | Monorepo — 13 Rust crates, full blockchain stack |
| 🤖 | savitri-fl | FL adapters: TFF, PyTorch, Flower bridges |
| 📱 | savitri-mobile | Flutter wallet + node monitoring (iOS & Android) |
| 🛠️ | savitri-sdk | Client library, wallet tools, CLI |
| 💻 | savitri-installer | One-click desktop installer (Windows · macOS · Linux) |
| 🧪 | savitri-testnet | Docker testnet with Prometheus + Grafana |
| 📖 | docs | Whitepaper, FL architecture, tokenomics, guides |
# Clone and build a lightnode git clone --recurse-submodules https://github.com/savitri-network/savitri-network.git cd savitri-network cargo build --release -p savitri-lightnode # Join testnet ./target/release/lightnode \ --listen-port 4001 \ --bootstrap <PEER_ID>@/ip4/<IP>/tcp/4002
Runs on 4 GB RAM, any CPU. No ASIC. No minimum stake. Run it on a laptop you already own.
# macOS / Linux curl -sSL https://savitrinetwork.com/install.sh | bash # Windows (PowerShell) iwr -useb https://savitrinetwork.com/install.ps1 | iex
from savitri_fl import FederatedClient import tensorflow_federated as tff client = FederatedClient(node_url="http://localhost:8545", wallet="...") client.submit_round(model_weights=tff.training.compute_weights(...)) # Provenance, payment, and verification handled at consensus level.
🤖 FL Researchers
Verify, compensate, and trace
your federated rounds
without rewriting your stack
🌐 IoT Operators
Dedicated tx type at 0ドル.000125
×ばつ cheaper than standard
Monetize device data
🏢 Enterprises
Train models on customer data
without seeing the data
Auditable provenance
👥 Node Operators
Run on the device you own
Earn SAVI for uptime + integrity
No minimum stake
Rust tokio libp2p RocksDB ed25519 blake3 ZKP BFT
TFF PyTorch Flower Flutter Tauri Docker Prometheus
Open issues are tagged good-first-issue and help-wanted. Most come from problems we hit running the testnet — they're real, scoped, and merge-worthy.
- Pick an issue from any repo
- Fork, branch, Conventional Commits
cargo fmt && cargo clippy -- -D warnings- PR with what you changed and how you tested it
Code of conduct: be useful, be honest about what works and what doesn't. We ship the bugs alongside the wins.
Docs · Whitepaper · Contribute · Security
Built in Rust · Secured by Proof-of-Unity · Open Source (MIT + Apache 2.0)