Build & Train a GLM-5.3-Flash Model From Scratch with Python
Learn how to design, build, and train an advanced multimodal language model from scratch using modern techniques like mixture-of-experts, sparse attention, and reinforcement learning. This hands-on guide walks you through the entire lifecycle from tokenization and pre-training to post-training optimization and evaluation.Video from @vukrosic
GitHub: https://github.com/vukrosic/glm-5.3-flash-from-scratch
Become AI researcher in 90 days: https://www.skool.com/become-ai-researcher-2669/about
Slides: https://github.com/vukrosic/glm-5.3-flash-from-scratch/blob/main/slides/slides.html
Blog: https://z.ai/blog/glm-5.3-flash
❤️ Support for this channel comes from our friends at Scrimba – the coding platform that's reinvented interactive learning: https://scrimba.com/freecodecamp
⭐️ Chapters ⭐️
- 00:00 Introduction & What We're Building
- 01:02 The Modern AI Researcher Role & Asking Research Questions
- 04:50 Tokenization & Byte-Level Vocabulary (Why Small Vocab Matters)
- 07:01 Embeddings & Transformer Forward Pass Overview
- 08:53 GLM-5.3 Architecture Overview & Model Specifications
- 10:25 Code Walkthrough: Embeddings & Token Representation
- 11:56 Manifold Constrained Hyperconnections (DeepSeek Residuals)
- 13:27 Output Projection & Weight Tying
- 14:27 RMSNorm & Normalization Layers
- 15:10 Positional Encodings (RoPE vs. NoPE) & Sparse Attention Indexer
- 18:30 Linear Attention (State-Space Memory) vs. Sparse Attention
- 20:46 Mixture of Experts (MoE) & Shared Experts
- 23:06 Adding Vision: Patch Embeddings & 2D RoPE
- 26:14 Pre-Training Pipeline, Loss & Optimization (AdamW)
- 28:05 Pre-Training Experiments: Data Diversity, Interleaving & Curriculums
- 30:27 Post-Training & Reinforcement Learning (RL) Setup
- 34:44 Designing Reward Functions & Group Relative Policy Optimization (GRPO)
- 37:37 Parameter-Efficient RL Updates & Freezing Layers
- 39:35 Evaluating RL Results: Task Gains & Regression Risks
- 40:47 RL Hyperparameter Experiments: Group Size, Temperature & Seeds
- 43:03 Summary & Advice for Aspiring AI Researchers
🎉 Thanks to our Champion and Sponsor supporters:
👾 @omerhattapoglu1158
👾 @goddardtan
👾 @akihayashi6629
👾 @kikilogsin
👾 @anthonycampbell2148
👾 @tobymiller7790
👾 @rajibdassharma497
👾 @CloudVirtualizationEnthusiast
👾 @adilsoncarlosvianacarlos
👾 @martinmacchia1564
👾 @ulisesmoralez4160
👾 @_Oscar_
👾 @jedi-or-sith2728
👾 @justinhual1290
--
Learn to code for free and get a developer job: https://www.freecodecamp.org
Read hundreds of articles on programming: https://freecodecamp.org/news Receive SMS online on sms24.me
TubeReader video aggregator is a website that collects and organizes online videos from the YouTube source. Video aggregation is done for different purposes, and TubeReader take different approaches to achieve their purpose.
Our try to collect videos of high quality or interest for visitors to view; the collection may be made by editors or may be based on community votes.
Another method is to base the collection on those videos most viewed, either at the aggregator site or at various popular video hosting sites.
TubeReader site exists to allow users to collect their own sets of videos, for personal use as well as for browsing and viewing by others; TubeReader can develop online communities around video sharing.
Our site allow users to create a personalized video playlist, for personal use as well as for browsing and viewing by others.
@YouTubeReaderBot allows you to subscribe to Youtube channels.
By using @YouTubeReaderBot Bot you agree with YouTube Terms of Service.
Use the @YouTubeReaderBot telegram bot to be the first to be notified when new videos are released on your favorite channels.
Look for new videos or channels and share them with your friends.
You can start using our bot from this video, subscribe now to Build & Train a GLM-5.3-Flash Model From Scratch with Python
What is YouTube?
YouTube is a free video sharing website that makes it easy to watch online videos. You can even create and upload your own videos to share with others. Originally created in 2005, YouTube is now one of the most popular sites on the Web, with visitors watching around 6 billion hours of video every month.
