nli:deberta-v3-large
3 TagsUpdated 435M params512 contextEnglishMITby Moritz Laurer
DeBERTa-v3-large, MIT. Best accuracy in our tests (0.548 on typed decisions).
435m
ollaya run nli:deberta-v3-large --preset triage "I was charged twice for my subscription this month and want a refund."curl http://localhost:11435/api/decide \
-H "Content-Type: application/json" \
-d '{
"model": "nli:deberta-v3-large",
"state": "I was charged twice for my subscription this month and want a refund.",
"questions": {
"department": {
"type": "choice",
"instructions": "Which team should handle this?",
"criteria": {
"billing": "Payments, invoices and refunds",
"technical": "Bugs, errors and outages",
"account": "Login, profile and settings"
}
},
"refund": {
"type": "noul",
"instructions": "Is the customer asking for a refund?"
}
}
}'# Already using a TypeSafe SDK? Set TYPESAFE_BASE_URL=http://localhost:11435 instead.
import requests
response = requests.post(
"http://localhost:11435/api/decide",
json={
"model": "nli:deberta-v3-large",
"state": "I was charged twice for my subscription this month and want a refund.",
"questions": {
"department": {
"type": "choice",
"instructions": "Which team should handle this?",
"criteria": {
"billing": "Payments, invoices and refunds",
"technical": "Bugs, errors and outages",
"account": "Login, profile and settings"
}
},
"refund": {
"type": "noul",
"instructions": "Is the customer asking for a refund?"
}
}
},
)
answers = response.json()["answers"]
print(answers["department"]["choice"], answers["refund"]["noul"])const response = await fetch("http://localhost:11435/api/decide", {
method: "POST",
headers: { "Content-Type": "application/json" },
body: JSON.stringify({
model: "nli:deberta-v3-large",
state: "I was charged twice for my subscription this month and want a refund.",
questions: {
department: {
type: "choice",
instructions: "Which team should handle this?",
criteria: {
billing: "Payments, invoices and refunds",
technical: "Bugs, errors and outages",
account: "Login, profile and settings"
}
},
refund: {
type: "noul",
instructions: "Is the customer asking for a refund?"
}
}
}),
});
const { answers } = await response.json();
console.log(answers.department.choice, answers.refund.noul);Details
- graph104fc208f1f0 · 6 MB
onnx · deberta-v3-large-zeroshot-v2.0 · 435M · fp32 - weights2031ec343409 · 870 MB
huggingface.co/MoritzLaurer/deberta-v3-large-zeroshot-v2.0/resolve/cf44676…/model.safetensors - tokenizer05402ffae6dd · 9 MB
huggingface.co/MoritzLaurer/deberta-v3-large-zeroshot-v2.0/resolve/cf44676…/tokenizer.json - decisione23cbf996659 · 1 KB
{"engine": "onnx", "family": "nli", "encoder": "MoritzLaurer/deberta-v3-large-zeroshot-v2.0", "layout": "nli-pairs-v1", …} - calibration316915b3e458 · 84 B
{"temperature": [1.0, 1.0, 1.0]} - license76dc53ac069e · 217 B
DeBERTa-v3-large zero-shot v2.0 by Moritz Laurer (https://huggingface.co/MoritzLaurer/deberta-v3-large-zeroshot-v2.0), MIT License.
Every layer is checked against its sha256 when it is pulled. Weights and tokenizers download from the model author's Hugging Face repository at a pinned commit; Ollaya never re-hosts them.