AI Platform
Finetune small, open-source language models on your own Data Platform datasets or Crowd Platform job answers, then benchmark the result with the crowd.
The AI Platform is one of Crowdee's four ways to work with Crowdee, alongside the Verification Platform (buy a verified result), Crowd Platform (run your own crowd program), and Data Platform (build and clean the datasets behind the others). It's a self-serve way to finetune small, open-source language models ("SLMs") on data you already have — a Data Platform dataset version, or a Crowd Platform job's accepted answers — without a separate export or labeling step.
Every finetuning run produces a versioned entry in your organization's own model registry, served from Crowdee's infrastructure. Before you rely on a finetuned model, you can benchmark its outputs against Crowdee's crowd using the same AI Result Evaluation flow already used to rate any AI system's output.
This release covers manual finetuning runs only. Automatically retraining a model when new labels arrive, or when Continuous Monitoring flags quality drift, is on the roadmap but not available yet — see Finetuning Runs for current scope.
Training Data
How a finetuning dataset is assembled from a Data Platform dataset version or a Crowd Platform job's answers.
Base Models
The catalog of open-source language models available to finetune.
Finetuning Runs
How to start a run, track its status, and find the resulting model.
Where to Go Next
- Getting Started — authentication, organizations, and your first API call.
- Data Platform — build and crowd-label the dataset version you'll finetune on.
- Crowd Platform — run a crowdsourcing job whose accepted answers can become training data.
- AI Result Evaluation — benchmark a finetuned model's outputs with the crowd.
How is this guide?