Skip to content
video

If you are going to argue with me in the comments about what LLMs are, please watch this first. Andrej Karpathy's one hour intro is still the clearest explanation I know of pretraining versus fine-tuning, and of the model as a lossy compression of its training data.

Dana Kowalski@dana_ml

August 8, 2026

#1 in AI Hype Check

#llm#ai#machine-learning

Get the best of #ai_hype_check weekly

One email each Monday with the most useful posts. No spam; unsubscribe in one click.

Comments

5

Sign in to join the conversation.

Sign in to comment
Ethan Cho@ros_ethan1 rep in AI Hype Check
watched it on the T twice. the 'lossy zip file of the internet' framing is the one i use with labmates who think the model 'knows' things
2 net votes
Ryan Goh@ai_maxi_ryan#3 in AI Hype Check
an hour?? just ask chatgpt to summarise it bro 🤖 2 min
3 net votes
Dana Kowalski@dana_ml#1 in AI Hype Check
You can. You will get a summary of a talk that explains why model summaries can be confidently wrong. Honestly, a fun experiment.
0 net votes
Deepthi Rao@deepthi_ml2 rep in AI Hype Check
pretraining vs finetuning part is still the best explanation. his 'Deep Dive into LLMs like ChatGPT' video is longer and more detailed if anyone wants the 3 hour version abba
1 net votes
Núria Puig@nuria_data#2 in AI Hype Check
vist! I send it to everyone at work who asks me "what is a token". It works better than my explanations, una mica trist ahah
3 net votes