Profile

GenAI & Data Science Portfolio

LLM Fine-Tuning | Prompt Engineering | Explainable AI

Hugging Face PyTorch TensorFlow Streamlit SHAP

LLM Fine-Tuning Playground

Model Settings

Precise Creative
Text Classification

This text is likely an example sentence used for typing practice (probability: 87%).

Fine-Tuning Details

This model was fine-tuned on a custom dataset of 10,000 example sentences using PyTorch and Hugging Face Transformers. Achieved 92% accuracy on validation set.

Prompt Engineering Toolkit

Prompt Strategy

Classify the following product review as either "positive" or "negative":

"{review_text}"

Answer:

Strategy Benefits

  • No training examples needed
  • Fastest to implement
  • Works for broad tasks

Limitations

  • Lower accuracy on niche tasks
  • May require prompt tuning
  • Vulnerable to ambiguous tasks

Social Trend Forecasting

Forecast Settings

or drag and drop

Forecast visualization will appear here

Feature importance analysis will appear here

Model Architecture

This forecasting pipeline combines time-series analysis with social media metadata (likes, shares, author credibility) to predict virality. The SHAP explainer highlights which features most influence predictions.

Real-Time Sentiment Analysis

Model Settings

50% 95%
Positive (94% confidence)
Negative 2%
Neutral 4%
Positive 94%

Key Sentiment Drivers

thrilled (+0.42) exceeded (+0.38) outstanding (+0.35) earlier (+0.28)

Model Details

This sentiment analyzer was fine-tuned on 50,000 product reviews from multiple domains. Achieves 89% accuracy on held-out test data with F1-score of 0.91 for positive/negative classification.