Tutorials

Step-by-step tutorials guiding you through different aspects of topic modeling with poisson-topicmodels.

Tutorials

• Tutorial: Training Your First Topic Model
  • Step 1: Prepare Your Data
  • Step 2: Initialize the Model
  • Step 3: Train the Model
  • Step 4: Extract Results
  • Step 5: Interpret Topics
  • Step 6: Analyze Document-Topic Distribution
  • Step 7: Advanced Analysis
  • Step 8: Quality Metrics
  • Next: Validation and Optimization
  • Quick Checklist
  • Common Issues
• Tutorial: GPU Acceleration
  • Why GPU?
  • Checking GPU Availability
  • Enabling GPU for poisson-topicmodels
  • Setting Up GPU Environment
  • Training with GPU
  • Monitoring GPU Usage
  • Memory Management
  • Performance Benchmarking
  • Optimizing for Speed
  • Troubleshooting GPU
  • Best Practices
  • Scaling to Large Datasets
  • Next Steps
  • Key Takeaway
• Tutorial: Model Validation & Evaluation
  • Validation Approaches
  • The Coherence Metric
  • Qualitative Inspection
  • Comparative Evaluation
  • Downstream Task Evaluation
  • Topic Similarity Analysis
  • Document Coverage
  • Visualization for Validation
  • Validation Checklist
  • Red Flags
  • Validation Workflow
  • Version Tracking
  • Next Steps
• Tutorial: Hyperparameter Tuning
  • Key Hyperparameters
  • num_topics: The Critical Parameter
  • learning_rate: Optimization Speed
  • batch_size: Gradient Stability
  • Systematic Hyperparameter Search
  • Random Search (More Efficient)
  • Practical Tuning Strategy
  • Early Stopping
  • Documenting Experiments
  • Common Mistakes & Solutions
  • Tuning Checklist
  • Next Steps
  • Summary

Tutorial Overview

Each tutorial covers a specific aspect of working with topic models:

Training & Basics (Tutorial: Training Your First Topic Model)

How to prepare data, train models, and interpret results. Best for beginners.

GPU Acceleration (Tutorial: GPU Acceleration)

Leverage GPU computation for large-scale analysis and significant speedup.

Model Validation (Tutorial: Model Validation & Evaluation)

Techniques to evaluate topic quality and model performance without ground truth.

Hyperparameter Tuning (Tutorial: Hyperparameter Tuning)

Systematic approaches to selecting optimal number of topics, learning rates, and batch sizes.

Quick Start Tutorials

I want to train my first topic model

→ Start with Getting Started, then Tutorial: Training Your First Topic Model

I have a large corpus and want to speed things up

→ Read Tutorial: GPU Acceleration

I’m not sure if my topics are good

→ Check Tutorial: Model Validation & Evaluation

My model doesn’t seem to improve

→ Review Tutorial: Hyperparameter Tuning

Prerequisites

All tutorials assume:

• Python 3.11+ installed

• poisson-topicmodels installed (see Installation)

• Basic familiarity with topic modeling concepts (see Fundamentals)

• Text data preprocessed into document-term matrices

Tutorial Series: From Data to Insights

Recommended progression:

1. Getting Started - 5-minute quickstart

2. Tutorial: Training Your First Topic Model - Train and interpret your first model

3. Tutorial: Model Validation & Evaluation - Assess topic quality

4. Tutorial: Hyperparameter Tuning - Optimize your model

5. Tutorial: GPU Acceleration - Scale to large datasets

6. How-To Guides - Practical recipes for common tasks

Troubleshooting Tutorials

Running into issues? Check relevant tutorials:

• Installation problems? See Installation

• Results don’t look good? See Tutorial: Model Validation & Evaluation

• Training is slow? See Tutorial: GPU Acceleration

• Not sure about settings? See Tutorial: Hyperparameter Tuning

• Data format issues? See How-To Guides

Feedback

Tutorials can always be improved! If you:

• Found a tutorial unclear or incomplete

• Discovered an error

• Have a topic you’d like a tutorial for

Please open an issue on GitHub.