Project workspace

Analyze Marathi text

Inference

Test the model with one sentence

API
Prediction, confidence, competing labels, and input notes appear below.
0 characters
Examples

Sample inputs

Prediction

Current result

Submit a sentence to see the predicted figure, confidence, and notes.
Probability spread

Other possible classes

Notes

Input observations

Edge-case notes will appear here when relevant.
Dataset operations

Browse rows, edit labels, and update notes

Default rebuild uses slight per-label offsets so the dataset stays near-balanced without looking mechanically rounded.
Loading dataset…
Select a row to edit the text, label, validation state, or notes.
Text Label Source Conf. Status
Review queue

Review rows that still need confirmation

Loading review queue…
Use this queue to confirm or correct the final label.
System view

Dataset counts, source mix, training status, and results

Label distribution

Class counts

Source mix

Source type mix

Pipeline

Background tasks

Evaluations

Saved model scores

Model Accuracy Macro F1