History and Development of LLMs

# Lecture 07. History and Development of LLMs

## Notice of New Lesson Postings

Hello, students! For this course, **82** new lessons (across a total of 13 sections) are scheduled to be posted sequentially. Since I am producing multiple courses simultaneously, I am proceeding by posting **a little bit of each course alternately**, rather than uploading one entire course at once. Below are the posting principles and the schedule for each section of this course.

> For your reference, the lessons being posted now are not entirely new content, but part of the production of **revised editions (2nd or 3rd editions)** that reinforce existing lectures.

## Posting Principles

- Currently, I am rotating through a total of 19 courses in order (circular method), posting one lesson per course alternately.
- Postings take place on weekdays (Mon–Fri), with a total of 5 new lessons uploaded daily across all courses.
- For this course, the next lesson will be posted each time its turn comes around in the rotation.
- While the posting speed may feel slow compared to other courses, all courses are progressing together in the same manner.

## Posting Schedule by Section (13 Sections Total)

| Section | Number of Lessons | Start Date | End Date |
|---|---|---|---|
| Section 1. Origins and Early Development of Language Models | 2 | 2026-07-02 | 2026-07-08 |
| Section 2. Emergence of Distributional Hypothesis and Evolution of Word Representation | 7 | 2026-07-13 | 2026-08-13 |
| Section 3. Limitations of Static Embeddings and Recognition of the Need for Context | 6 | 2026-08-19 | 2026-09-15 |
| Section 4. Tokenization Methods and Solving the Out-of-Vocabulary (OOV) Problem | 7 | 2026-09-21 | 2026-10-22 |
| Section 5. ELMo - The Beginning of Contextual Embeddings | 7 | 2026-10-27 | 2026-11-27 |
| Section 6. Transformer and Self-Attention Mechanisms | 7 | 2026-12-03 | 2027-01-04 |
| Section 7. BERT - The Revolution of Bidirectional Contextual Learning | 7 | 2027-01-08 | 2027-02-05 |
| Section 8. Variant Models of BERT | 6 | 2027-02-10 | 2027-03-03 |
| Section 9. GPT and Expansion to Sentence-Level Representation - The Peak of Contextual Embeddings | 8 | 2027-03-08 | 2027-04-05 |
| Section 10. The Transformer Revolution and the Era of Large Language Models | 4 | 2027-04-08 | 2027-04-20 |
| Section 11. Latest LLM Models and Technological Advancements | 7 | 2027-04-23 | 2027-05-18 |
| Section 12. LLM Efficiency Technologies and Advancements in Model Optimization | 7 | 2027-05-20 | 2027-06-14 |
| Section 13. LLM Applications, System Integration, and Future Outlook | 7 | 2027-06-16 | 2027-07-08 |

## Estimated Completion

Based on the current pace, the estimated completion date for all postings in this course is around **July 2027**. (The actual schedule may be moved up or delayed depending on production progress.)

At the request of our readers, we are issuing a 66% discount coupon for "Cognitive Load Management Techniques to Break Through the Limits of RAG Performance" (the 990,000 KRW course can be purchased for 330,000 KRW).

For your information, this course is currently in the process of posting the 2nd edition, which is a revision of the 1st edition.

In the case of the 2nd edition, all class materials have been posted, but the videos are still in the process of being uploaded.

I post the 2nd edition videos frequently, but it will take a long time (at least several months) to post all of them.

Please keep this in mind and carefully decide whether to use the coupon.

The deadline is the 19th and there is a limited number of coupons available, so please hurry if you need one.

You can download the coupon by entering the address below into your browser's search bar.

https://inf.run/QowXG

Hello.

The era of AI writing code has arrived.

However, the task of verifying the code written by AI and directing the path for improvement remains the responsibility of humans.

Furthermore, it has become necessary to know how to utilize AI more extensively in both vertical and horizontal directions.

The horizontal direction refers to a direction that encompasses everything from planning to deployment, and

The vertical direction refers to a direction that can encompass various languages, various frameworks, and various methodologies.

Therefore, a developer with a broad spectrum (that is, a complete spectrum in both vertical and horizontal directions)

will be needed.

However, the education system has not yet been able to keep up with this.

That is why I have opened the "Structural Code Reading Bootcamp" using Claude Code.

At https://code-reading-bootcamp.vercel.app/, you can develop the ability to read code written by both AI and humans, and

You will be able to develop the ability to guide AI on how to improve code.

It includes gamification elements, so you can have fun while tracking your skill improvements.

It is free to use anytime without the need to log in.

Although there are some shortcomings and parts where the questions have not yet been filled in,

I introduce this with the hope that it will be helpful to you.

I would appreciate it if you could use it lightly.

February 10, 2026, Sincerely, Jinsu Park (Arigaram).