Intermediate ChatGPT
Learn how GPT models work, structure prompts with XML and thinking tags, and set up custom instructions and Custom GPTs in ChatGPT.
ChatGPT is easier to steer once you know what happens inside the model. The lessons show how transformers, tokenization and self-attention let GPT models handle language, and how pre-training, fine-tuning and reinforcement learning from human feedback shape their answers. You then structure prompts with XML and thinking tags for negotiation, storytelling and startup planning, and finish by writing custom instructions and configuring a Custom GPT of your own.
What You Will Achieve
- Describe how tokens, embeddings and self-attention work together inside a GPT transformer
- Tell apart the pre-training, fine-tuning and human feedback phases that shape a GPT model’s behavior
- Write structured prompts with XML tags, few-shot examples and thinking tags for negotiation and planning tasks
- Decide when custom instructions are enough and when to build and configure a Custom GPT
Program Overview
GPT Model Architecture and Training
How transformer models process language and how GPT models are trained, with a comparison to earlier RNN designs.
- Transformers compared with RNNs
- Tokenization, encoding and decoding
- Components of the GPT model
- Pre-training, fine-tuning and RLHF
Prompt Crafting with XML and Thinking Tags
Scenario-based practice in structuring prompts for negotiation, storytelling, content creation and new business ideas.
- XML tags for negotiation prompts, including asking for a raise
- Story structure with the PAIPS framework
- Content creation with the PFS principle
- Few-shot prompting and thinking tags for startup ideas
Custom Instructions, Custom GPTs and the Future of ChatGPT
Personalizing ChatGPT with custom instructions, building and deploying Custom GPTs, and a look at future advancements and ethical questions.
- Crafting effective custom instructions
- Building and deploying custom GPTs
- Custom GPTs compared with custom instructions
- Future advancements, AI ethics and AGI
Certificate Awarded (Paid)
Job Outlook
These skills apply wherever ChatGPT supports everyday work such as writing, storytelling, negotiation preparation and business planning, and in teams that set up tailored assistants for colleagues.
Good for: Regular ChatGPT users who want to understand how GPT models work and to build structured prompts and Custom GPTs for work tasks.
Keep in mind: Despite the intermediate title it is rated beginner, the architecture lessons explain concepts rather than code, and the model history it describes stops at GPT-4.
Bottom line: Its mix of transformer basics, tag-structured prompting scenarios and a configured Custom GPT makes it a practical next step for people already using ChatGPT.
Frequently Asked Questions
Do I need a paid ChatGPT account for the exercises?
No. The course does not require a ChatGPT license, so you can work through the architecture, prompting and customization lessons without one.
Which prompting techniques does the course practice?
It uses XML tags to give prompts a clear structure in negotiation scenarios, including one about asking for a raise, few-shot prompting and thinking tags for startup ideas, and story and content frameworks named PAIPS and PFS.
What is the difference between custom instructions and a Custom GPT?
Custom instructions personalize your own interactions with ChatGPT, while a Custom GPT is built and deployed as a tailored assistant for a particular use case, with more flexibility. One lesson compares the two, and another configures an example GPT called EcoGuide.
Does the course look at where ChatGPT is heading?
Yes. The final lessons consider possible future advancements of ChatGPT and the ethical questions that AI development raises as work moves toward artificial general intelligence.


There are no reviews yet.