Restoration Toolkit

Lacunae Predictor

Character-level LSTM network designed to probabilistically restore damaged epigraphic sequences.

Try a Damaged Sequence:

mi lar[---]ia (Damaged Name)

tu[---]ce (Damaged Verb)

[---]ilaθ (Missing Prefix)

Fragmentary Text

Predictions

Awaiting sequence input...

How it Works

A "Lacuna" is a gap or missing piece of text on an ancient stone due to damage or weathering. Epigraphers usually guess what letters are missing based on context.

This AI model acts as a highly-trained assistant. It has "read" every Etruscan text in existence. When you give it a broken sequence like 'tu[---]ce', it mathematically predicts the most likely characters that originally filled the gap.

Developer API

The Lacunae Predictor utilizes a bidirectional LSTM architecture running natively in ONNX web runtime, returning a softmax distribution of the top K probable characters.

POST /api/v1/inference/predict-lacuna
{
  "sequence": "tu[---]ce",
  "top_k": 4
}
// Returns array of char probability objects