Decisions computed on your device.
coba:150m-ternary is Ollaya's own open decision model: 66 MB of ternary weights under Apache-2.0, running in this tab with WebAssembly, on a laptop or a phone. Nothing you type is sent to a server.
coba:150m-ternary66 MBApache-2.0
Weights from huggingface.co/ollaya-dev/coba-150m-ternary
Not loaded. Nothing downloads until you press the button.
Request
Examples
Questions
refundyes / no
The customer asks for money back.
teamchoice
Which team should handle this ticket?
billing · technical · account
urgencyscore
How urgent is this ticket?
Can wait · Needs attention this week · Needs attention today
Advanced: edit the request as JSON
Answers
refund–
team–
urgency–
Load the model first.
The text never leaves this page: the tokenizer, the graph and the answers all run in your browser. Only the model's files are downloaded, unmodified, from its author's Hugging Face repository.
Use it in your app
The same model, three lines away.
In the browser with @ollaya/web, or natively with Ollaya on a CPU or GPU. Both take the same /api/decide request.
npm install @ollaya/web
import { fromHub, load } from '@ollaya/web'
const model = await load(fromHub('ollaya-dev/coba-150m-ternary'))
const { answers } = await model.decide({ state, questions })<script type="importmap">
{ "imports": { "onnxruntime-web": "https://cdn.jsdelivr.net/npm/onnxruntime-web@1.30.0/dist/ort.bundle.min.mjs" } }
</script>
<script type="module">
import { fromHub, load } from 'https://cdn.jsdelivr.net/npm/@ollaya/web@0.1.0/ollaya.js'
const model = await load(fromHub('ollaya-dev/coba-150m-ternary'))
console.log(await model.decide({ state, questions }))
</script>ollaya pull coba:150m-ternary
curl http://localhost:11435/api/decide -d '{
"model": "coba:150m-ternary",
"state": "I was charged twice. Please refund the duplicate.",
"questions": {
"refund": {"type": "noul", "instructions": "The customer asks for money back."}
}
}'Use it in code: curl, Python, TypeScript, Docker The request and answer format