text
stringlengths 1
215k
| category
stringclasses 10
values | source_file
stringclasses 10
values |
---|---|---|
hi everyone so by now you have probably | build_gpt.txt | build_gpt.txt |
heard of chat GPT it has taken the world | build_gpt.txt | build_gpt.txt |
and AI Community by storm and it is a | build_gpt.txt | build_gpt.txt |
system that allows you to interact with | build_gpt.txt | build_gpt.txt |
an AI and give it text based tasks so | build_gpt.txt | build_gpt.txt |
for example we can ask chat GPT to write | build_gpt.txt | build_gpt.txt |
us a small Hau about how important it is | build_gpt.txt | build_gpt.txt |
that people understand Ai and then they | build_gpt.txt | build_gpt.txt |
can use it to improve the world and make | build_gpt.txt | build_gpt.txt |
it more prosperous so when we run this | build_gpt.txt | build_gpt.txt |
AI knowledge brings prosperity for all | build_gpt.txt | build_gpt.txt |
to see Embrace its | build_gpt.txt | build_gpt.txt |
power okay not bad and so you could see | build_gpt.txt | build_gpt.txt |
that chpt went from left to right and | build_gpt.txt | build_gpt.txt |
generated all these words SE sort of | build_gpt.txt | build_gpt.txt |
sequentially now I asked it already the | build_gpt.txt | build_gpt.txt |
exact same prompt a little bit earlier | build_gpt.txt | build_gpt.txt |
and it generated a slightly different | build_gpt.txt | build_gpt.txt |
outcome ai's power to grow ignorance | build_gpt.txt | build_gpt.txt |
holds us back learn Prosperity weights | build_gpt.txt | build_gpt.txt |
so uh pretty good in both cases and | build_gpt.txt | build_gpt.txt |
slightly different so you can see that | build_gpt.txt | build_gpt.txt |
chat GPT is a probabilistic system and | build_gpt.txt | build_gpt.txt |
for any one prompt it can give us | build_gpt.txt | build_gpt.txt |
multiple answers sort of uh replying to | build_gpt.txt | build_gpt.txt |
it now this is just one example of a | build_gpt.txt | build_gpt.txt |
problem people have come up with many | build_gpt.txt | build_gpt.txt |
many examples and there are entire | build_gpt.txt | build_gpt.txt |
websites that index interactions with | build_gpt.txt | build_gpt.txt |
chpt and so many of them are quite | build_gpt.txt | build_gpt.txt |
humorous explain HTML to me like I'm a | build_gpt.txt | build_gpt.txt |
dog uh write release notes for chess 2 | build_gpt.txt | build_gpt.txt |
write a note about Elon Musk buying a | build_gpt.txt | build_gpt.txt |
Twitter and so on so as an example uh | build_gpt.txt | build_gpt.txt |
please write a breaking news article | build_gpt.txt | build_gpt.txt |
about a leaf falling from a | build_gpt.txt | build_gpt.txt |
tree uh and a shocking turn of events a | build_gpt.txt | build_gpt.txt |
leaf has fallen from a tree in the local | build_gpt.txt | build_gpt.txt |
park Witnesses report that the leaf | build_gpt.txt | build_gpt.txt |
which was previously attached to a | build_gpt.txt | build_gpt.txt |
branch of a tree attached itself and | build_gpt.txt | build_gpt.txt |
fell to the ground very dramatic so you | build_gpt.txt | build_gpt.txt |
can see that this is a pretty remarkable | build_gpt.txt | build_gpt.txt |
system and it is what we call a language | build_gpt.txt | build_gpt.txt |
model uh because it um it models the | build_gpt.txt | build_gpt.txt |
sequence of words or characters or | build_gpt.txt | build_gpt.txt |
tokens more generally and it knows how | build_gpt.txt | build_gpt.txt |
sort of words follow each other in | build_gpt.txt | build_gpt.txt |
English language and so from its | build_gpt.txt | build_gpt.txt |
perspective what it is doing is it is | build_gpt.txt | build_gpt.txt |
completing the sequence so I give it the | build_gpt.txt | build_gpt.txt |
start of a sequence and it completes the | build_gpt.txt | build_gpt.txt |
sequence with the outcome and so it's a | build_gpt.txt | build_gpt.txt |
language model in that sense now I would | build_gpt.txt | build_gpt.txt |
like to focus on the under the hood of | build_gpt.txt | build_gpt.txt |
um under the hood components of what | build_gpt.txt | build_gpt.txt |
makes CH GPT work so what is the neural | build_gpt.txt | build_gpt.txt |
network under the hood that models the | build_gpt.txt | build_gpt.txt |
sequence of these words and that comes | build_gpt.txt | build_gpt.txt |
from this paper called attention is all | build_gpt.txt | build_gpt.txt |
you need in 2017 a landmark paper a | build_gpt.txt | build_gpt.txt |
landmark paper in AI that produced and | build_gpt.txt | build_gpt.txt |
proposed the Transformer | build_gpt.txt | build_gpt.txt |
architecture so GPT is uh short for | build_gpt.txt | build_gpt.txt |
generally generatively pre-trained | build_gpt.txt | build_gpt.txt |
Transformer so Transformer is the neuron | build_gpt.txt | build_gpt.txt |
nut that actually does all the heavy | build_gpt.txt | build_gpt.txt |
lifting under the hood it comes from | build_gpt.txt | build_gpt.txt |
this paper in 2017 now if you read this | build_gpt.txt | build_gpt.txt |
paper this uh reads like a pretty random | build_gpt.txt | build_gpt.txt |
machine translation paper and that's | build_gpt.txt | build_gpt.txt |
because I think the authors didn't fully | build_gpt.txt | build_gpt.txt |
anticipate the impact that the | build_gpt.txt | build_gpt.txt |
Transformer would have on the field and | build_gpt.txt | build_gpt.txt |
this architecture that they produced in | build_gpt.txt | build_gpt.txt |
the context of machine translation in | build_gpt.txt | build_gpt.txt |
their case actually ended up taking over | build_gpt.txt | build_gpt.txt |
uh the rest of AI in the next 5 years | build_gpt.txt | build_gpt.txt |
after and so this architecture with | build_gpt.txt | build_gpt.txt |
minor changes was copy pasted into a | build_gpt.txt | build_gpt.txt |
huge amount of applications in AI in | build_gpt.txt | build_gpt.txt |
more recent years and that includes at | build_gpt.txt | build_gpt.txt |
the core of chat GPT now we are not | build_gpt.txt | build_gpt.txt |
going to what I'd like to do now is I'd | build_gpt.txt | build_gpt.txt |
like to build out something like chat | build_gpt.txt | build_gpt.txt |
GPT but uh we're not going to be able to | build_gpt.txt | build_gpt.txt |
of course reproduce chat GPT this is a | build_gpt.txt | build_gpt.txt |
very serious production grade system it | build_gpt.txt | build_gpt.txt |
is trained on uh a good chunk of | build_gpt.txt | build_gpt.txt |
internet and then there's a lot of uh | build_gpt.txt | build_gpt.txt |
pre-training and fine-tuning stages to | build_gpt.txt | build_gpt.txt |
it and so it's very complicated what I'd | build_gpt.txt | build_gpt.txt |
like to focus on is just to train a | build_gpt.txt | build_gpt.txt |
Transformer based language model and in | build_gpt.txt | build_gpt.txt |
our case it's going to be a character | build_gpt.txt | build_gpt.txt |
level language model I still think that | build_gpt.txt | build_gpt.txt |
is uh very educational with respect to | build_gpt.txt | build_gpt.txt |
how these systems work so I don't want | build_gpt.txt | build_gpt.txt |
to train on the chunk of Internet we | build_gpt.txt | build_gpt.txt |
need a smaller data set in this case I | build_gpt.txt | build_gpt.txt |
End of preview. Expand
in Dataset Viewer.
Sensai's yt transcripts to mimic his patterns and explanations ( fun project)
- Downloads last month
- 31