ai-model-acceleration-visualizations

AI Model Acceleration Visualization & Minimalist Pipeline. This repository hosts Python code snippets demonstrating performance visualizations and a minimalist AI pipeline integrating image processing with large language models (LLMs). These complement the blog post "Accelerate AI Model Speed; Python Minimalist!".

Overview Visualization scripts: Generate charts showing inference latency, throughput improvements, and memory usage comparing FP32 vs FP16.

Minimal AI pipeline: A modular, clean Python prototype showing how to chain image loading, preprocessing, feature extraction, and language model querying — all designed for clarity and compact one-liners.

Contents

File	Description
inference_latency_cpu_gpu.py	Bar chart comparing CPU vs GPU inference latency
throughput_vs_batch_size.py	Line plot showing throughput scalability with batch sizes
memory_usage_fp32_fp16.py	Bar chart comparing FP32 and FP16 memory usage
minimal_ai_pipeline.py	Minimalist prototype pipeline for image + LLM processing

Usage Instructions

1. Clone the repository:

bash

git clone https://github.com/PV-J/ai-model-acceleration-visualizations.git
cd ai-model-acceleration-visualizations

2. Install dependencies:

bash

pip install matplotlib seaborn pandas pillow torch transformers

3. Run any visualization script to generate PNG charts locally:

bash

python inference_latency_cpu_gpu.py
python throughput_vs_batch_size.py
python memory_usage_fp32_fp16.py

4. Experiment with the minimalist pipeline:

bash

python minimal_ai_pipeline.py

This runs a sample image through loading, preprocessing, feature extraction, and LLM caption generation in a concise chained manner.

License This project is licensed under the MIT License.