AI Educator

Vishesh Breja

LLM Basics Course GermanBhashi workshop. Professionals learn core LLM concepts through live Streamlit apps: tokenisation, embeddings, attention, transformers, temperature effects, streaming responses. Hands-on personas test real model behaviour.<h2>What you learn</h2>01_LLM_Theory&nbsp; <ul><li class="py-0 my-0 prose-p:pt-0 prose-p:mb-2 prose-p:my-0 [&amp;&gt;p]:pt-0 [&amp;&gt;p]:mb-2 [&amp;&gt;p]:my-0">Live token chips show cl100k_base encoding as you type German/English text.</li><li class="py-0 my-0 prose-p:pt-0 prose-p:mb-2 prose-p:my-0 [&amp;&gt;p]:pt-0 [&amp;&gt;p]:mb-2 [&amp;&gt;p]:my-0">GPT-2 predicts next tokens with adjustable temperature; bar chart updates instantly.</li><li class="py-0 my-0 prose-p:pt-0 prose-p:mb-2 prose-p:my-0 [&amp;&gt;p]:pt-0 [&amp;&gt;p]:mb-2 [&amp;&gt;p]:my-0">Sentence embeddings via MiniLM visualized in 2D PCA clusters and cosine similarity heatmaps.</li><li class="py-0 my-0 prose-p:pt-0 prose-p:mb-2 prose-p:my-0 [&amp;&gt;p]:pt-0 [&amp;&gt;p]:mb-2 [&amp;&gt;p]:my-0">BERT attention head heatmaps for any input sentence, layer 0–11.</li><li class="py-0 my-0 prose-p:pt-0 prose-p:mb-2 prose-p:my-0 [&amp;&gt;p]:pt-0 [&amp;&gt;p]:mb-2 [&amp;&gt;p]:my-0">Chocolate analogy animates batch size, epochs, learning rate.</li></ul>02_Practical_Session<ul><li class="py-0 my-0 prose-p:pt-0 prose-p:mb-2 prose-p:my-0 [&amp;&gt;p]:pt-0 [&amp;&gt;p]:mb-2 [&amp;&gt;p]:my-0">Five personas: language tutor, career coach, AI explainer, grammar checker.</li><li class="py-0 my-0 prose-p:pt-0 prose-p:mb-2 prose-p:my-0 [&amp;&gt;p]:pt-0 [&amp;&gt;p]:mb-2 [&amp;&gt;p]:my-0">Live probability distribution shifts as you adjust temperature 0.0–2.0.</li><li class="py-0 my-0 prose-p:pt-0 prose-p:mb-2 prose-p:my-0 [&amp;&gt;p]:pt-0 [&amp;&gt;p]:mb-2 [&amp;&gt;p]:my-0">Streaming chat responses with word-by-word reveal.</li><li class="py-0 my-0 prose-p:pt-0 prose-p:mb-2 prose-p:my-0 [&amp;&gt;p]:pt-0 [&amp;&gt;p]:mb-2 [&amp;&gt;p]:my-0">Compare mode runs identical prompts at two temperatures side-by-side.</li><li class="py-0 my-0 prose-p:pt-0 prose-p:mb-2 prose-p:my-0 [&amp;&gt;p]:pt-0 [&amp;&gt;p]:mb-2 [&amp;&gt;p]:my-0">Session dashboard tracks response time, output length, model performance per turn.</li></ul><h2>Who this fits</h2>Data scientists, developers, users entering production AI. No prior ML knowledge needed.&nbsp;<h2>Time</h2>90 minutes.&nbsp;<h2>Outcomes</h2>You learn how you can tune an LLM