A1‐1 LLMs build - terrytaylorbonn/auxdrone GitHub Wiki

25.0526 (0522) Doc URLs Stack URLs Lab notes (Gdrive) Git

These output of this section is educational (to understand the math, not for deployment).

to learn the core stuff about weights, models, etc.. its missing the glue logic and size, both important for real world models.

All docs are @ Lab notes (Gdrive)

ch0b “neural network” 66

• #1 ch1 Stage1 pretraining 75
• #2 00:42:52 Llama 3.1 base model inference 79
• #3 ch2 Stage2 supervised fine tuning 86
• #3 ch3a Stage3a reinforcement learning 112
• ch3b Stage3b RL: DeepSeek / AlphaGo / human feedbac

 

good source for low level concepts stuff like 3B1B, etc

 

 

• got thru self attention
• chaos at building transformer