Complete OpenAI API Course (Connect to ChatGPT API & More)

Discover what OpenAI's API interface offers and how pricing and tokens influence usage, then build a blog generator that sends requests and adds end user interactivity.

Explore how OpenAI's API connects your apps to models like GPT-4 and GPT-3.5, and learn private API keys, usage billing, and pricing per 1,000 tokens.

Preview the final blog generator code and feed topic inputs to a prompt that creates a full blog with GPT-3.5 or GPT-4, then set up Colab and API keys.

Learn to list available OpenAI models, choose GPT-3.5 instruct, and send your first request with the completions API to generate a blog or story, adjusting max tokens for length.

Build a blog generator by configuring the API call, setting max tokens and temperature, and using role and system prompts to produce structured, multi-section blogs with titles.

Add user interactivity to a blog generator by turning a hard-coded topic into dynamic prompts using user topic and additional information inputs, powered by an api structure.

Learn to build a data-driven chatbot using OpenAI's API in Google Colab, with a memory-enabled bot that answers only from a provided dataset and prompt.

Learn to start a chat session with the OpenAI API by creating a client, selecting a model (GPT-3.5 Turbo or GPT-4), and configuring messages, temperature, and max tokens.

Build a real chat bot interface using the chat completion API, capture and print responses, loop for continuous questions, exit with q, and begin memory for Luca.

Learn to add memory to a chatbot by maintaining a message buffer with system and user prompts, managing memory length, and updating the assistant responses for OpenAI projects.

Learn to enhance a chatbot by loading a provided data set, configuring a system prompt, and controlling model parameters to answer questions from your content.

Learn embeddings and similarity using OpenAI's embedding endpoint to build a base movie recommendation system, from toy datasets to Kaggle movie data, scale with Pinecone, and deploy globally.

Explore how embeddings and vectors represent data, transforming text like movie descriptions into numerical features used for search and recommendations. Use cosine similarity to compare vectors.

Explore how to implement OpenAI text embeddings in code using the Ada 002 model, including preparing input, generating embedding vectors, and observing how small text changes alter the vector.

Explore creating toy embeddings, compute user and item vectors, and apply cosine distance with numpy and scipy to build a scalable movie recommendation system.

Scale your movie recommendation system by loading a larger dataset from Kaggle, selecting original title and overview, building OpenAI embeddings, and preparing numpy arrays for cosine similarity.

Convert the user input to numpy arrays, compute cosine similarity with movie embeddings, and display the ten titles most similar to an action movie using a small dataset for demonstration.

Explore using Pinecone to store and query vector embeddings for a movie recommendation system, reducing recomputation by managing indices and collections efficiently.

Explore using Pinecone to store and query movie embeddings, uploading vectors via upsert, configuring metadata, and retrieving top-k results to power a real-time movie recommendation system.

Learn to build a retrieval-augmented generation system by converting a pdf to documents, chunking text, creating embeddings, and indexing them in pinecone to answer questions from your data.

Load a pdf with the pi pdf reader, extract text from all pages, and chunk it into 1000–1500 character pieces (or words) with optional overlap to create embeddings for pinecone.

Convert PDF chunks into embeddings with text-embedding-002, create and populate a pinecone index via upsert, storing the original text as metadata to enable future question-based retrieval.

Create the pdf application interface by converting user questions to embeddings with text embedding-002, query the index for the top chunk, and feed it into a retrieval augmented generation chatbot.

Generate images with DALL-E 3 via OpenAI's API and build a Streamlit interface to display and save outputs, using prompts, sizes, and multiple images.

Edit images with dall-e 3 via openai's api, using masks, images, prompts, and size adjustments. Learn to resize masks and experiment with prompts to control edits and styles like vaporwave.

Learn to work with audio files using OpenAI's Whisper to generate automated transcriptions and translate videos into any language by connecting Whisper with GPT via API.

Transcribe an uploaded audio file using whisper, selecting the whisper one model to generate timestamped transcripts (and translations) for searchable video content, with upgrades to newer models anticipated.

Use the translation endpoint to transcribe any language into English from audio or video files with whisper, then translate to the user’s language for multilingual search with ChatGPT.

Transcribe with Whisper AI and translate video text to a target language using ChatGPT, configuring a system prompt and using GPT-3.5-turbo to translate to Serbian or any language.

Explore GPT vision with OpenAI's API to classify and describe images and predict events in video, and use text-to-speech to generate voiceovers for your content.

Encode images in base64 and send them to GPT Vision for classification and description. Craft prompts and request structures to process image data with the OpenAI API.

Learn to create speech from text with the OpenAI text-to-speech API, generate mp3 audio using TTS-1 voices, and stream the output to a file from lowercase input.

combine vision, gpt, and tts to generate a voiceover for any video by processing frames and producing narration with GPT-4 Vision Preview, then convert it to speech.

Learn how to build Streamlit-based AI web interfaces by connecting OpenAI and DALL·E to generate blog text, create images, and power a movie recommendation engine with minimal code.

Learn to build an ai web interface with Streamlit: install the library, import as st, create a title, run Streamlit, and use widgets like a sidebar.

Configure streamlit with a wide layout for ai apps. Create a sidebar to select blog, image, or movie, with headers, text areas, and a start button to trigger OpenAI.

Build a Streamlit web interface that connects to a blog generation prompt using GPT-3.5-turbo. Define topic and additional text to generate blog content via the OpenAI API.

learn how to build an ai image interface with streamlit using dall-e, including a prompt-driven image generator, adjustable image count via a slider, image sizing options, and basic input validation.

Build a movie generator interface in Streamlit using Pinecone embeddings and vector queries to fetch the top ten movie results, and add a loading spinner before deploying the app.

Learn to deploy a private to public streamlit app to production by linking a GitHub repo, creating a requirements.txt, and managing the app on streamlit.io.

Track ongoing ai developments as new technologies and features emerge. Add new lectures to this portion of the course whenever new developments are released.