Run templates

What we train.

Every run starts from a hardened template: sensible defaults, guardrails and the right eval suite built in. Templates are how a pilot starts in days instead of weeks — and how quality stays repeatable across re-trains.

01 /

The catalog

Six templates cover most production needs. Each one states its engine, typical wall-clock and the eval suite that gates it.

tabular-classify

Tabular classification

Fraud, churn, scoring. Gradient-boosted trees and calibrated ensembles on your structured data.

engineXGBoost / sklearn
typical run8–40 min
evalAUC · F1 · calibration
tabular-regress

Tabular regression

Forecasting, pricing, demand. Same engines, continuous targets, horizon-aware splits.

engineXGBoost / sklearn
typical run8–40 min
evalRMSE · MAPE
text-classify

Text classification

Routing, moderation, intent. Fine-tuned transformer encoders sized to your latency budget.

engineDistilBERT+ family
typical run1–4 hrs
evalmacro-F1 · confusion
llm-finetune-lora

LLM fine-tune (LoRA)

Your tone, your format, your domain — adapters trained on open-weight checkpoints.

enginePEFT / LoRA
typical run2–12 hrs
evalwin-rate vs base
vision-classify

Vision classification

Defect detection, tagging, triage. Transfer-learned CNN/ViT backbones on your images.

enginetimm backbones
typical run1–6 hrs
evaltop-1 · PR curves
custom

Custom spec

ML Partner engagements: bespoke pipelines designed with your named engineer.

engineper engagement
typical runvaries
evalper engagement
02 /

Not sure which fits?

Send a paragraph about your data and objective — an FMS engineer replies with the template, an estimate and a fixed quote.

Describe your problem