thecybro

VibeCheck runs a full NLP pipeline entirely on your machine:

1. A content script monitors your feed using MutationObserver, detecting new posts as you scroll
2. Post text is extracted using platform-specific DOM selectors and sent to the background service worker
3. The background worker forwards the text to a local FastAPI server running a fine-tuned transformer model
4. The model (SamLowe/roberta-base-go_emotions) classifies the text across 28 emotion categories
5. Detected emotions are mapped to VibeCheck’s 4 categories: anger, toxicity, fear, sadness
6. If a blocked emotion exceeds the user’s sensitivity threshold a blur overlay is applied
7. The user can click “Reveal Anyway” to dismiss the overlay at any time

• Added view/hide pasted content feature + some styling
• Added view/hide pasted content feature + some styling
• Added blocked sites count, data type to protect/not protect, made detection system more efficient and corrected test results
• Made modal scrollable, added suggestion, added copy masked text button and some ui tweaks
• Cleaned up the mess in the code and enhanced modal and popup with smooth elements animations, improved visuals.

• Inline annotations for pandas operations (read, merge, dropna, fillna, groupby, rename, drop)
• Status bar summary showing live dataframe and operation count
• DAG sidebar panel with visual dataframe flow graph
• Hover tooltips showing variable name, operation type, and inputs

Coding day and night without proper sleep just recked my time and wasted so much time and effort of mine.

I was coding for so much time not realizing all of that was bs, my entire structure collapsed, predictions were wrong, most fundamental thing was also coded wrong.

But after having proper sleep, rested for couple of days and then sitting and calmly looking at my code, i realised i have to fix the entire structure ( but not just everything from 0 ), so I did and now,
I have added many output features, and functions like interpret_prediction, run_prediction, keyword searcher, important results etc, optimized and cleaned up rest of the code

Styled the page to make it look attractive

Finally fixed the validity checker function logic which was irritating me for a long time, but still long way to go.

Backend optimization as done and confirmed that it works properly.

Found the number required for stability

Trained the model for the first time after fixing some formulas (some yet to be finished)

Added methods to compute mean, variance, standard deviation, covariance, correlation coefficients and interpretation.

I’m working on my first project! This is so exciting. I can’t wait to share more updates as I build.