DontBurnOut — Emotion Classification Model
Quantized ONNX model for 7-class emotion detection from text, optimized for on-device inference on Android.
Labels
anger · disgust · fear · joy · neutral · sadness · surprise
Usage in Android (ONNX Runtime)
val classifier = TextEmotionClassifier.create(context)
val results = classifier.classify("I am so stressed about these deadlines.")
results.forEach { (emotion, score) ->
println("$emotion: ${"%.2f".format(score * 100)}%")
}
classifier.close()
Files
| File | Description |
|---|---|
model_quantized.onnx |
INT8 quantized ONNX model |
tokenizer.json |
Tokenizer vocabulary and rules |
tokenizer_config.json |
Tokenizer configuration |
config.json |
Model architecture config |
special_tokens_map.json |
Special tokens |
merges.txt |
BPE merge rules |
Integration
This model is used in the DontBurnOut Android app.
Download is handled automatically at first launch via ModelDownloader.kt.