Aliens School
Cinematic Knowledge Experience
0%
Aliens School
Now Playing
Aliens School ยท HIEN
โŒจ๏ธ Keyboard Shortcuts
โ†’Next slide โ†Previous slide SpacePlay / Pause MNarration on/off FFullscreen ?Show/hide this
Press any key to close
Skill Topic ยท Cinematic

๐Ÿ—๏ธ Topic 03: LLM Architecture Overview โ€” Under the Hood

Course: LLM Engineering โ€” Hinglish Section: 1 โ€” LLM Foundations Level: Beginner โ†’ Intermediateโ€ฆ

Overview
๐ŸŒŸ

๐Ÿ—๏ธ Topic 03: LLM Architecture Overview โ€” Under the Hood โ€” Quick Facts

๐Ÿ“Œ

Model: Parameters

๐ŸŽฏ

GPT-2 Small: 117M

โšก

GPT-2 XL: 1.5B

๐Ÿ”‘

LLaMA-2 7B: 7B

Topic 1
๐Ÿ“ฅ ๐Ÿ“ฅ ๐Ÿง  ๐Ÿ”ฌ ๐Ÿ’ก ๐ŸŽฏ

๐Ÿ“Œ Objectives

๐Ÿ’ก

LLM ka internal architectureโ€ฆ

๐Ÿ”‘

Neural network layers ka roleโ€ฆ

โšก

Transformer blocks ka kaam dekhna

๐ŸŽฏ

Parameter count aur model size kaโ€ฆ

Topic 2
๐Ÿ“ฅ โš™๏ธ ๐Ÿ”ฌ ๐Ÿ’ก

๐Ÿง  1. LLM Ka Basic Structure

๐Ÿ“ Socho ek LLM ek massive building hai jisme multiple floors hain: `โ€ฆ
Topic 3
๐Ÿ“ฅ โš™๏ธ ๐Ÿ”ฌ ๐Ÿ’ก

๐Ÿ”„ 2. Transformer Block โ€” Core Unit

๐Ÿ’ก

Self-Attention = Har word dusreโ€ฆ

๐Ÿ”‘

FFN = Information koโ€ฆ

โšก

Add & Norm = Training stableโ€ฆ

๐ŸŽฏ

Residual Connection = Originalโ€ฆ

Topic 4
โš™๏ธ

๐Ÿ“Š 3. Model Size Breakdown

โš™๏ธ LLM ka "size" parameters se measure hota hai: `โ€ฆ
Topic 5
๐Ÿ“ฅ โš™๏ธ ๐Ÿ”ฌ ๐Ÿ’ก

๐Ÿงฎ 4. Parameter Calculation

๐Ÿ›๏ธ Parameter count kaise calculate karte hain? Yeh formula samjho: `โ€ฆ
Topic 6
โœจ

๐Ÿ”€ 5. Decoder-Only vs Encoder-Decoder

๐Ÿ“Š ` โ”Œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ฌโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ” โ”‚ DECODER-ONLY (GPT) โ”‚โ€ฆ
Topic 7
โœจ

๐Ÿ”ฅ 6. Attention Masking โ€” Causal vs Bidirectional

๐Ÿ’ก

Causal Mask = Generation ke liyeโ€ฆ

๐Ÿ”‘

Bidirectional = Understanding keโ€ฆ

Topic 8
โœจ

๐Ÿ’ป 7. Python Code โ€” Simplified LLM Architecture

๐Ÿ’ก

Head 1: Syntax patterns dekh rahaโ€ฆ

๐Ÿ”‘

Head 2: Semantic similarity dekhโ€ฆ

โšก

Head 3: Positional relationshipsโ€ฆ

๐ŸŽฏ

...etc

Topic 9
โœจ

๐Ÿงฉ 8. Modern Architectural Innovations

๐Ÿ—๏ธ 8.1 RoPE (Rotary Position Embedding) ` Traditional Positional Encoding: position_vector +โ€ฆ
Topic 10
โœจ

๐Ÿ†š 9. Architecture Comparison Table

๐Ÿ“ | Feature | GPT-4 | LLaMA-3 | Mistral | Claude | Gemini |โ€ฆ
Topic 11
๐Ÿ”ง

โ“ 10. Quiz โ€” 5 MCQs

๐Ÿ’ก

a) CNN Layer

๐Ÿ”‘

b) RNN Cell

โšก

c) Transformer Block โœ…

๐ŸŽฏ

d) LSTM Unit

Topic 12
โœจ

๐Ÿ”— Navigation

โš™๏ธ | โฌ…๏ธ Previous | ๐Ÿ“š Index | โžก๏ธ Next | |---|---|---| | 02-LLM-History.md | 00-Index.md |โ€ฆ
Comparison
โœจ

๐Ÿ†š 9. Architecture Comparison Table

โš–๏ธ

Type: Decoder

โš–๏ธ

Attention: MHA + MoE

โš–๏ธ

Position: Unknown

Quick Quiz
๐Ÿง  QUIZ TIME

Quiz โ€” Question 1

๐Ÿ—๏ธ Topic 03: LLM Architecture Overview โ€” Under the Hood ka sabse sahi definition kya hai?

Quick Quiz
๐Ÿง  QUIZ TIME

Quiz โ€” Question 2

๐Ÿ—๏ธ Topic 03: LLM Architecture Overview โ€” Under the Hood ka 'GPT-2 Small' kya hai?

Complete! ๐ŸŽ‰
COMPLETE

๐Ÿ—๏ธ Topic 03: LLM Architecture Overview โ€” Under the Hood Complete!

Aliens School ยท HIEN ยท Cinematic Knowledge

โœ…

๐Ÿ—๏ธ Topic 03: LLM Architecture Overview โ€” Under the Hood Complete

1/18
0:00
REC 00:00ESC=Cancel
Aliens School
3
Recording shuru hone wali hai...
โœ…
Recording Complete
Video process ho rahi hai...
Live Class
Slide 1 / 7
Timer
00:00
๐Ÿ“ Speaker Notes
โ€”
โญ๏ธ Up Next
โ€”
โ€”
๐Ÿ—‚๏ธ All Slides