Nori-1.0
Nori-1.0Nori is Synthefy's foundation model for tabular data. No training, no tuning: point it at any dataset and get state-of-the-art predictions in seconds.
Model details
View repositoryExample usage
Nori is Synthefy's foundation model for tabular data, the structured numbers behind most real-world decisions. It ranks #1 across public benchmarks and runs in seconds on a single GPU. Classical workflows built around XGBoost demand constant retraining, feature engineering, and hyperparameter tuning. Nori needs none of that. Point it at a new dataset and it predicts instantly out of the box through in-context learning, while beating XGBoost on accuracy.
Its top use cases are demand forecasting, fraud detection, predictive maintenance, risk scoring, churn prediction, and high-frequency trading: the high-volume numerical problems where teams have traditionally reached for XGBoost or bespoke pipelines. Nori collapses that entire train-a-model-per-problem workflow into a single pretrained model, with state-of-the-art accuracy, a radically smaller footprint, and instant deployment. It's now served on Baseten for production-grade inference at scale.
See full documentation.
1# pip install synthefy — see /nori/client for the full client guide
2import os
3from synthefy import SynthefyNoriClient
4
5client = SynthefyNoriClient(
6 api_key=os.environ["SYNTHEFY_NORI_API_KEY"],
7 model="nori-30m",
8)
9
10X_train = [[0.0, 1.0], [1.0, 0.0], [0.5, 0.5], [0.2, 0.8]]
11y_train = [0.1, 0.9, 0.5, 0.3]
12X_test = [[0.3, 0.7], [0.8, 0.2]]
13
14print(client.predict(X_train, y_train, X_test))1{
2 "task": "regression",
3 "predictions": [
4 0.3861955995072203,
5 0.7150392187985302
6 ],
7 "usage": {
8 "prompt_tokens": 4,
9 "completion_tokens": 3,
10 "total_tokens": 7
11 }
12}
