NOULI

Models / Text v1.0.0 · 31 October 2026

Nouli Route

Banking intent classification

Nouli Route classifies English banking-support requests into the 77 BANKING77 intents. It uses a fine-tuned Ettin-1B encoder and returns the selected intent with a calibrated probability distribution.

Comparison with Jev

Recorded comparison with Jev 1.14.0, dated 31 October 2026.

Nouli Route: the published example, reference label, and recorded Jev comparison

Reference label: Pending cash withdrawal. Nouli Route selects Pending cash withdrawal; Jev few-shot, calibrated selects Declined cash withdrawal.

The example uses the declared primary few-shot Jev baseline; the full evaluation also includes a graph baseline.

Illustrative example selected after scoring; not a representative sample. Recorded evaluation · Methodology · Example details.

Model at a glance

Property Value
Parameters 1.029 billion
Input 128 tokens
Tested runtime macOS ARM64 / CPU; Linux ARM64 / CUDA
License Apache-2.0; upstream notices apply

Quickstart with Nouli CLI

Requires Python 3.12, the HF CLI, and Nouli CLI 0.4.2. Hugging Face hosts the files; inference runs on your machine. The Python server loads the weights, and Nouli runs the supplied decision workflow. No hosted API key is required.

In the first terminal:

Terminal 1 · prepare and serve
hf download nouli-ai/nouli-route --revision v1.0.0 --local-dir ./nouli-route
cd nouli-route
python3.12 -m venv .venv
source .venv/bin/activate
python -m pip install -r requirements.txt
python prepare.py --text "My card still has not arrived after two weeks."
python serve.py --device cpu

Once the server reports ready, open a second terminal in nouli-route:

Terminal 2 · run the workflow
nouli --config nouli/nouli.toml --store ./.nouli-store \
  workflow run nouli/example-workflow.json --json

The example uses CPU. On a configured CUDA host, start the server with --device cuda. Rerun prepare.py with your text or file before running another workflow.

Base models and training data

Training data: PolyAI/banking77; 6,473 training queries.

Trained on BANKING77 intent labels. Auxiliary intent-family labels are derived from the same ontology. Source revisions and provenance.

Limitations

  • Supports the 77 BANKING77 intents. Out-of-scope rejection and production-routing performance are untested.
  • Development data was reserved and detected test-family overlaps removed; published leaderboard setups may differ.

License and citation

Apache-2.0, with upstream notices and dataset attribution retained.

Nouli AI. Nouli Route. Cite the published model version and commit.