Deep Learning Models
The reason this website came to being, is form me to share my notebooks in a format that is not too ugly 😊
🚀
3
Models Deployed
🎯
94%
Avg Accuracy
⚡
Variable
Inference Time
📦
ONNX + INT8
Optimization
Projects
Audiobuilt
Audio Language Detection
19-Language Classifier
Production-ready language identification model. 96.56% accuracy on clean audio and 94.95% on noisy recordings
Clean Accuracy
96.56%
🎯
Noisy Accuracy
94.95%
🔊
Inference Time
165ms
⚡
Languages
19
🗣️
🛡️Noise Robustness:1.61% drop
📦Model Size:2.1M params
Audiobuilt
Journaling AI
Journal-Insights Extractor
Somewhat production-realy(if you have GPUs) DL model for extracting 8 structured fields (summary, title, themes, emotions, entities) from 3-10 minute transcribed voice journal entries.
Emotion Accuracy
96.5%
😊
Entity F1
92%
👤
Inference Time
30.9s
⚡
Languages
5
🌍
✅Valid JSON:100%
📦Model Size:395 MB
Textbuilt
Task Manager AI
Task-Field Extractor
I have a task manager app in ere. It just does not use AI, so I need to fill in forms. It is not bad bad, it is actually really good. But I can do better. So why not use AI? instead of filling the forms
Segmentation F1
87.4%
✂️
Field Accuracy
72.2%
🎯
Model Size
4x smaller
📦
CPU Inference
3.9s
⚡
🌍Languages:5
💰Cost Savings:50x vs GPT
🚧
More projects on the way
I try to build new projects, that I like or need.😉