Introduction to large language models

[@genlu6137]

Appreciate the valuable content! Sharing some key takeaways of the video and I hope this can help someone out. 1) 00:50 - Large language models (LLMs) are general purpose language models that can be pre-trained and fine-tuned for specific purposes. LLMs are trained for general purposes to solve common language problems, and then tailored to solve specific problems in different fields. 2) 02:04 - Large language models have enormous size and parameter count. The size of the training data set can be at the petabyte scale, and the parameter count refers to the memories and knowledge learned by the machine during training. 3) 03:01 - Pre-training and fine-tuning are key steps in developing large language models. Pre-training involves training a large language model for general purposes with a large data set, while fine-tuning involves training the model for specific aims with a much smaller data set. 4) 03:15 - Large language models offer several benefits. They can be used for different tasks, require minimal field training data, and their performance improves with more data and parameters. 5) 08:50 - Prompt design and prompt engineering are important in large language models. Prompt design involves creating a clear, concise, and informative prompt for the desired task, while prompt engineering focuses on improving performance. 6) 13:43 - Generative AI Studio and Generative AI App Builder are tools for exploring and customizing generative AI models. Generative AI Studio provides pre-trained models, tools for fine-tuning and deploying models, and a community forum for collaboration. 7) 14:52 - Palm API and Vertex AI provide tools for testing, tuning, and deploying large language models. Palm API allows testing and experimenting with large language models and gen AI tools, while Vertex AI offers task-specific Foundation models and parameter efficient tuning methods. This takeaway note is made with the Notable app (getnotable.ai).

[@EKOLegend]

The mere fact that every large player in this space has videos teaching people about these things means this is super super serious.

[@ChatGTA345]

Or that it is a massive massive waste of time and effort

[@zappy9880]

@@ChatGTA345 unlikely. 1 or 2 small companies pursuing this tech with such ambition could be a waste of time. But if all the big players are investing their time and money in this tech, then it has to be something very real and very serious

[@ChatGTA345]

@@zappy9880 I don't think that necessarily follows. The industry has followed so many hype waves before. The competitive advantage is actually not to do what everyone else does

[@padmakalluri181]

@@ChatGTA345 😊😊😊😊😊😊❤

[@espiya5557]

@@ChatGTA345 well, it is a waste depends on how you use it. but can really be useful in several fields if you know how to use it and how you fine-tune it. just treat it as some sort of assisting tool as of now, and not as something that you actually use as some sort of definitive source of knowledge.

[@dariannwankwo9126]

Minor Correction @ 2:14. "In ML, parameters are often called hyperparameters." In ML, parameters and hyperparameters can exist simultaneously and serve two different purposes. One can think of hyperparameters as the set of knobs that the designer has direct influence to change as they see fit (whether algorithmically or manually). As for the parameters of a model, one can think of it as the set of knobs that are learned directly from the data. For hyperparameters, you specify them prior to the training step; while the training step proceeds, the parameters of the model are being learned.

[@LavaCreeperPeople]

Nice

[@praveenmadduri7181]

Parameters means connections between nodes ?

[@dariannwankwo9126]

@@praveenmadduri7181 In the context of neural networks, yes. When parameters are mentioned, one is usually referring to the weights/connections between nods.

[@Acrid93]

@@praveenmadduri7181 Not quite. The existence of a parameter implies a connection, but a connection does not necessarily determine the number of parameters. E.g. given connection a -> b with transformation input(b) = w1output(a) + bias contains the parameters w1 and bias.

[@sarahsalt3689]

Thank you for making this available to the general public!

[@davidcottrell1308]

Fantastic presentation...and...(I LOVE THIS) NO ANNOYING BACKING TRACK!! Thank you, Google!

[@BrandonLee-ik8kw]

2:47 You mentioned the parameters are hyper parameters is incorrect and confusing

[@JonathanPoczatek]

Can't wait to see demos at GoogleIO

[@henri8903]

Thank you John. I believe you conflated model parameters and hyperparameters at 2:16. As far as I know, these are two different concepts.

[@fierce10]

Yes, they are different conceptually. Parameters are directly applied/calculated in the hypothesis or model; while, hyperparameters are somewhat heuristically decided based on what works. For example if you were figuring out how to get from home to office, the path details maybe calculated directly by the GPS, but the time at which you leave maybe heuristically decided by you. Another example of a hyperparameter can be how many backup cameras you choose to add should the main camera fail on a robot, there is no 'correct' number, it's more of a cost or design choice. In an ML transformer, choosing the number of encoders or decoders can be a hyperparameter. The parameters would be learned from the language training in the LLM.

[@03timboy]

Agreed, two totally different things. It's not great that the video encourages this confusion.

[@fred-nyanokwi]

This is one of the educative sessions I've come across

[@joseperez-ig5yu]

Finding answers to questions has become so much easier now with new tech. I have never been good at writing code, so this is a welcome change as far as I'm concerned! Look forward to more progress in technology.

[@yenda12]

Be careful in the world to come being reliant on these AIs without developing any specific field will make you obsolete in future society

[@yabadab8609]

Actually, really helpful, thank you Google. Wondering how far this technology will go in the next couple of years, if it's this far already in a couple of months.

[@webgpu]

we all wondering too ;)

[@theAnupamAnendepothor]

proximity and stream for seek time reduction..memory in case reduced latency, can also be optimized for seek time and pattern analysis.

[@MeenakshiSharma-ss2ir]

At 4:50 I did not understand the third point that the speaker made i.e. "Orchestrated distributes computation for accelerators". Can someone please explain?

[@alexanderwilliams6527]

0:57 What do pre-trainned and fine-tuned llms means? Good analogy with dogs.

[@robertcormia7970]

This was fantastic! While I've been watching The Full Stack LLM Bootcamp, I'm not technically strong enough to start there, and will use these Google Cloud Tech videos as a means to "jumpstart" my knowledge of LLM and Generative AI. This is a great general primer for students and colleagues!

[@mostafatouny8411]

Thanks for referencing Full Stack LLM Bootcamp, A great resource I was not aware of.

[@luminouswolf7117]

If you define the problem you are trying to solve first Then reason from their Wouldn’t it be more efficient?

[@Michel-gv1sr]

2.20: he got the definition of hyperparameters wrong. In ML, there is a clear distinction between parameters and hyperparameters.

[@bakerkawesa]

Great explainer. I'm a little less anxious about AI taking our jobs.

[@near_.]

1980s or so, there were telephone operator who connects those STD lines. Now they are vanished but their next gen kids are employed in another market. That's how innovation works!!

[@littlebrit]

What is the legal status now of LLM models trained on proprietary data ?

[@gr8ape111]

lol

[@MrAmgadHasan]

Japan legalized them.

[@artus198]

What does 540 billion parameters mean , and how do you pass those to your model ? What kind of computational processing power is needed for this ?

[@strider806]

You don't have to pass the parameters. In Llm you just send the data as text and it must be able to tokenize the text.

[@MrAmgadHasan]

You first instantiate the model with randomly generated parameters (540B in this case) and use lots and lots and lots of data to make the model "learn" and modify these parameters so they are better. For llms, you need hundreds of powerful gpus and you need weeks or months to train such massive models. Falcon 40B which is a state of the art open source model with 40B parameters was trained for two months.

[@artus198]

@@MrAmgadHasan chatgpt was trained for about 2 years , there are 2 seperate models within chatgpt , one to understand context, the other to predict the text 🤪

[@MrAmgadHasan]

@@artus198 chatgpt is not a pure LLM. It was finetuned using multiple instructions datsets and RLHF. I was talking about training pure LLMs

[@richardglady3009]

Thank you. I understood about half (optimistically) of it. I subscribed to the channel hoping to start from the beginning and understanding more. My ultimate goal: a LLM Librarian, combining the catalog of a library with results from internet search engine, giving the deepest answer possible.

[@muslimridealong1975]

text classificaiton 13:19 fine tuning not realsitic

[@lifeofdean3647]

can u share awesome slides ?

[@jamesmina7258]

It's very clear to understand LLM, thank you

[@ninaz2735]

11:45 Can anybody explain the difference between these two prompts?

[@johnnycyberseed]

I understand the message of this slide to be not about prompt design, but AI response: that if the app in which the model is embedded first instructs the model to describe the process to get to an answer and THEN feed that back in with the original prompt, that the quality of the final response improves.

[@jaykef]

please provide link to the slides

[@theAnupamAnendepothor]

you can use a new drive architecture sought via gpu pixels for proximity stream like to not need large.lamguage models, and use multi factor checks to reduce need of a lot of data..thank me now.

[@theAnupamAnendepothor]

proximity and stream for seek time reduction..memory in case reduced latency, can also be optimized for seek time and pattern analysis.

[@pradeepbhatt4857]

where can i access gen ai studio and build apps?

[@satkotech]

RIP Bard, gone so young..

[@cassianocominetti7784]

Very comprehensive video! Thank you guys!

[@farukkara3929]

Can I have these slides please?

[@jeanpeuplu3862]

I have an urgent question (school related) -> is LLM part of NLP? Is an LLM always an NLP model? Or can an LLM be another kind of model? "L" for Language in both kinds of models. Both in AI. Both for language. A colleague says LLM is not necessarily an NLP model but then I did not understand LLM and/or NLP and my oral exam is in few days omg

[@jeanpeuplu3862]

Also, BERT is Transformer but not an LLM, right? Transformer can be LLM or not, right?

[@DrJanpha]

is it true that AI models like ChatGPT or Bard are fed in with codes (like programming languages) as well?

[@AnthatiKhasim-i1e]

Hey there! AI is definitely becoming more prevalent on Facebook. I've noticed more personalized content and ads powered by AI algorithms. It's amazing how AI enhances our social media experience."

[@YHK_YT]

Time to start my own

[@artie5172]

Do LLM charge money for using them

[@malacca498]

Always great to learn from GCT!

[@sasasunshine6]

Great video! Thank you!!

[@coryrandolph8501]

This is a great overview video thank you. Do you have a reference for how to host open-sourced LLM's in Vertex AI (or other GCP tools)? Overall I'm looking for GCP tools and ways for turning open-source LLM's into API's to be used within our native cloud instance.

[@B2M2948]

You lost any semblance of an answer from @Google Cloud Tech the second you said "open-source"

[@coryrandolph8501]

@@B2M2948 lol. I still want to host the Open source thing on their platform so I thought there might be a shot.

[@andrestorres7343]

​@@coryrandolph8501did you ever figure it out?

[@coryrandolph8501]

@@andrestorres7343 Yes, but it was really pricey since you have to host the underlying infrastructure. Usually large GPU virtual machines and on GCP depending on model size it was $2k - $5k per month to host an open source model. We are sticking with the API version of the big models because of this.

[@severtone263]

Great video. Thank you!

[@jeganathanmanickam6604]

Very Informative - Thanks for sharing 😊 prompt design and prompt engineering would take make the conversation more realistic and accurate.

[@JayLikesLasers]

What's a TPU V4 Pod? Sounds like a Turboencabulator, or?

[@MrAmgadHasan]

It's a custom built computer chip developed by google to perform matrix operations and train deep learning models. Think of them as gpus specialized for deep learning.

[@MrAmgadHasan]

A pod is "rack" of tens or hundred tpu/gpu.

[@mohamedkarim-p7j]

Thank for sharing👍

[@BREAKKWISS]

Thank you for teaching.

[@theAnupamAnendepothor]

pattern analysis with causal.

[@aliwafaafif]

Anybody who read this comment, you'd want to type this prompt in Chat-GPT or Bard: "I have 15 liter jug, 10 liter jug, and 5 liter jug. How do I measure 5 liters of water?" ---> See what they answer

[@arnoldpraesent174]

very well and understandable explained... good job!

[@SidIndian082]

Excellent Presentation Sir ... truly i admire it 😍😍😍😍

[@s.ackermann5498]

whats the name of the last circle at kzbin.info/www/bejne/sJrdoKGKpKuLetU ?

[@jumarkpelismino5632]

Can users teach AI?

[@higiniofuentes2551]

Wow! Thank you for this very useful video so well explained!

[@聂超群]

great content! make me feel like an expert now💯

[@amirkhalesi5294]

For the fellow beginners: PETM is also called PEFT

[@MinimalRevolt]

Waow! Such an eye-opening knowledge!🤓

[@lengsolace6074]

helpful for me,tks google

[@reinventingai]

Very slim on the prompt engineering education. This is a very important skill!

[@aditeepathak8894]

Is it just me or the quality of google training videos has gone down?

[@MrAmgadHasan]

Yes. They made a mistake when they described parameters as hyper parameters and the chain of thought part wasn't clear.

[@eriqfromimo2570]

Nice one!

[@ChatGTA345]

I've been extremely frustrated in my interactions with chatbots, they never seem to tell the truth and it's getting harder and harder to tell what's true from what's not. I honestly like regular Google searches much more!

[@deeplearningpartnership]

Nice.

[@gayatrichaudhary580]

Thankyou for this.

[@julioconradomarinardila3269]

Excelente Google cloud

[@LavaCreeperPeople]

Cool!

[@mikebeats3281]

this was bestowed upon us by our true creators to speed up the process so they can come in and enslave us again😂🎉

[@theloniousMac]

Creating a prompt seems more of a "Craft" than engineering.

[@CiscoZero]

Exciting stuff.

[@BryinWillis-e8g]

Introduction TM LLM

[@imranhossain3504]

I'm with you

[@jebuskmiest]

So a prompt engineer is anyone with common sense?

[@jackyang7401]

i love it

[@robertsutkowski3170]

Google 👍

[@fk_torty]

Was this long? YES. Did I learn? YES. Did I want to sleep? YES. Did I sleep before the end? NO. A WIN

[@hussienalsafi1149]

😊😊😊😊😊😊😊☺️☺️☺️☺️👌👌👌👌

[@benjaminstk]

interesting

[@AzherZarach]

Gonzalez Timothy Martinez Mark Anderson Matthew

[@eduardocesargarridomerchan5326]

Tutorial de LLM en español, por si interesa: kzbin.info/www/bejne/f4KkgIGphcSkY5o

[@enmedallo]

Why so few comments

[@akj3344]

This felt more like advertisement for Bard. Not very helpful.

[@MrAmgadHasan]

It is both advertising for bard and helpful too.

[@tombombadil9123]

Citizen Kane9 :D

[@darioplant8029]

I have the feeling that some comments here were AI generated. I am not a robot, I just in case say.

[@KevinNdhlovu]

🙄🙄🙄🙄

[@IrynaCherednychenko]

passed

[@christianstill.6654]

We are creating our own prison...

[@iskalasrinivas5640]

Really helpful video, but dont understand why it's called intelligent because it cannot discover something on its own

[@tiagomaqz]

It can. Once you feed the base of information, it can learn from the questions themselves leveraging possible answers for accuracy. Hallucinations will happen but that's when you start fine tuning it with the correct answers that it could not find on its own or on its data base. A human can't learn everything on their own, we need to study content which is build over time through observation.