21 күн бұрын
2 ай бұрын
4 ай бұрын
5 ай бұрын
6 ай бұрын
6 ай бұрын
7 ай бұрын
8 ай бұрын
Пікірлер
@albertogubernati3531 20 минут бұрын
Thank you for your lessons so useful and clear 😃
@mohammadamiri-lc3ic 4 сағат бұрын
عالی بود پسر موفق باشی
@missmirandaelaine 11 сағат бұрын
This was really helpful, thank you!
@adics 11 сағат бұрын
Should I learn data science before making LLM from scratch ? This playlist is awesome, I enjoy watching, and I can understand the module, but when I got into this video I get lost 😅.. I learn LLM just to make RAG assistance for fun (or maybe if my brain curious to learn more, I'll learn the advance thing later)
@harinijaan3779 12 сағат бұрын
Great explanation so far
@alikakavand3165 Күн бұрын
Very inspiring. Thanks Shaw
@PretzelHearts_animation Күн бұрын
Thank you! now i have +50 point for business hahaha
@ShawhinTalebi 19 сағат бұрын
LOL use it well!
@CrystalCarroll-n9f Күн бұрын
My image is too blurry. I changed the size to 100x100 in Canva. Any suggestions? Thanks!
@TheDeskAthlete 2 күн бұрын
thank you for explaining this at a level i can understand lmao
@ReapermanUK 2 күн бұрын
problem with AI today is AI is not AI, it's probability filters, based on shitty human data (facefook,twitter, etc)
@JonDecker 2 күн бұрын
Uninformed but commenting-AI seems low-impact for those working with old systems and processes. If AI drives rapid development in version 3, perhaps version 4 will correlate new methods with old ones to assess long-term viability and streamline implementations. In summary, version 4 focuses on the "why," not just the "what."
@ShawhinTalebi 19 сағат бұрын
You can go far with Python and having a high-level understanding of key ideas and techniques. However, if you want to go deeper, a sold math background is critical. More specifically, linear algebra and calculus.
@budsayalaohapensaeng6869 2 күн бұрын
Can I do with Ollama?
@ShawhinTalebi 19 сағат бұрын
Yes! The syntax is very similar. I share an example here: kzbin.info/www/bejne/haWVlGiDgLOSY9ksi=CYVqDM7agauuyCFI
@emmanuell8260 2 күн бұрын
Is it possible to predict by using API with POST ?
@ShawhinTalebi 19 сағат бұрын
My understanding of a POST request is it will pass input data to a backend data store. I don't think that helps much in generating a prediction from a model. For that, GET makes more sense.
@aimattant 2 күн бұрын
Perfect. Looking forward to following your example in training the Phi-3 model later now I came across MLX.
@Falcone807 2 күн бұрын
would be good if you showed us how you made the table 100 by 100 instead of just skipping bits out
@YourTradingBlueprint 3 күн бұрын
At the last step of selecting the social icons and trying to put a link on it, it keeps giving text option. Select icon -> Link (CMD+K) -> Gives option for text, how can I solve this?
@yasinfadaei2278 3 күн бұрын
I tried to run the code but it gives this error "ValueError: Directory articles does not exist.". What should I do?
@ShawhinTalebi 19 сағат бұрын
Make sure the articles folder is in the same path as the code you are running!
@simonmafanye.3537 4 күн бұрын
Your detail explanations gained my subscription. Thanks a lot for sharing.
@Rachael-e9y 4 күн бұрын
Awesome video, but I have an issue with my signature showing up. I sent myself an email from my macbook using apple mail (where I'm logged into my Gmail account) but the signature isn't there...How do I get it onto there? Thanks in advance, much appreciated! :)
@a.anvarbekov 6 күн бұрын
4:30 There's a way to pull only specific folder of a repo on GitHub with git spares-checkout
@KiranM-zs6kk 6 күн бұрын
how did you split the dataset into training and testing data...ref line in code : trainer = Trainer(...,train_dataset = tokenized_data["train"],eval_dataset = tokenized_data["test"],...)
@ShawhinTalebi 19 сағат бұрын
The tokenized_data variable is a DataDict class which was already split into the three sets. The data preparation code is available here: github.com/ShawhinT/KZbin-Blog/blob/main/LLMs/model-compression/data_prep.ipynb
@karthikt1737 6 күн бұрын
you explained it in a very simple way, Thank you.
@WrU673 6 күн бұрын
Thank you for explaining the hard into easy. Do you find having a strong math background helpful when learning AI? What do you think?
@ShawhinTalebi 19 сағат бұрын
You can go far with Python and high-level understanding of key concepts. However, if you want to go deeper a strong math background is critical. More specifically, linear algebra, calculus, and statistics.
@WrU673 4 сағат бұрын
@ So it’s fine if I ignore the math side, study the core concepts, and learn python. That will suffice?
@YedenLife 7 күн бұрын
To what sort of industry were you able to sell your AI/data consulting service? I am in Australia working for a government agency but thinking about running my own consulting.
@ShawhinTalebi 19 сағат бұрын
It was mainly research work and consultations in a wide range of industries e.g. medicine, human performance, geoscience, professional services.
@hlkn333 7 күн бұрын
Bravo! you are a LLM Chopin. I wonder if you have something just to provide the simplest steps to make fine tune Llama3.2 1B or 3B to work. I want the fine tune to work and then study all the details. Thanks in advanced.
@ShawhinTalebi 19 сағат бұрын
While I haven't fine-tuned Llama yet, this video may be helpful: kzbin.info/www/bejne/jqHSfHVpqtiggcUsi=jjV7So2qTRLPOeoM
@aveontaefrazier 7 күн бұрын
I’d like to see a follow on reranker implementation!
@siridu9519 8 күн бұрын
The theme button was not there!
@siridu9519 8 күн бұрын
You saved my life for this afternoon. Otherwise, I would have explored a few hours!
@donaldwedding5500 8 күн бұрын
Great content. Here's a question. Is there a way to load my fine tuned model and then train it some more. I tried doing that and got a lot of errors. Perhaps you could add that code to the end of your Jupyter Notebook. I would love to be able to continuously fine tune a model without needing to start from scratch.
@ShawhinTalebi 7 күн бұрын
Yes! A simple way would be to upload it to the Hugging Face hub then repeat the fine-tuning process :)
@donaldwedding5500 6 күн бұрын
@@ShawhinTalebi I did try that, but when I loaded my new model as opposed to the original, I ended up getting error messages. I thought I should be able to just swap out the new model for the old, but that didn't work. I'm probably doing something wrong. Any chance you could add the code to the end of your jupyter notebook (right after you load the saved model).
@donaldwedding5500 4 күн бұрын
When I reload my trained model and try to retrain it, I get dump of information culminating with ... TypeError: forward() got an unexpected keyword argument 'num_items_in_batch'. I believe I should be updating other things as well as the model variable. But which one? I am obviously playing "monkey see, monkey do" with your code. I am lacking in some of understanding. Any suggestion would be welcome.
@Tristana_21 8 күн бұрын
I have some problems when using pickle to read the 'df_propensity_score.p' file on your github, seems that my new version is not compatible with the pickle you used to save files. Can I get a new .p file or .csv file, thanks a lot!
@ShawhinTalebi 7 күн бұрын
Thanks for raising this! Would you mind submitting an issue on the GitHub repo? github.com/ShawhinT/KZbin-Blog/issues
@kambizhomayounfar2374 8 күн бұрын
Brilliant presentation Shaw. Thank you!
@viralshorts9596 8 күн бұрын
very informative, concise and straight to the point
@Arslanahmadbaloch 8 күн бұрын
Thanks G!
@club4796 8 күн бұрын
is GPU cores in apple chips matter for fine tuning or Local Ai in general?
@ShawhinTalebi 7 күн бұрын
Yes it will enable faster training and inference :)
@ArunSingh-ub8xf 9 күн бұрын
Wow....brilliantly explained and easy to grasp.
@emilyalbatross7681 9 күн бұрын
This is such a timely video for me. Thanks for producing this kind of clear and informative content Shaw!
@ShawhinTalebi 7 күн бұрын
Glad it was helpful!
@TheAndreealascar 9 күн бұрын
I seem to have difficulties when pasting from sheets to the signature portion. When I try to replace the image the text portion slides down and I cannot seem to be able to realign the cells. Please advise. Thank you!
@laibawaseem1151 9 күн бұрын
Sir i m a student in University and i have choosen this idea as semester project i understood most stuff but after cloning the repo it gives so many errors in weasyprint also in site packages..if i make a similar application would i require all these files?? Plus did u manually setted up 11-env file??
@ShawhinTalebi 7 күн бұрын
Sorry your getting errors. Did you install the packages using the requirements.txt file? github.com/ShawhinT/AI-Builders-Bootcamp-2/blob/main/lightning-lesson/requirements.txt
@laibawaseem1151 7 күн бұрын
@ShawhinTalebi i did sir..
@shayanshafiqahmad 9 күн бұрын
The series is phenomenal, Shaw! And, more business AI use cases, please! 🤓
@ShawhinTalebi 7 күн бұрын
I've got another one coming soon!
@shayanshafiqahmad 2 күн бұрын
@@ShawhinTalebi Thanks!
@nikilpatel1730 10 күн бұрын
Good stuff. Just to highlight something I noticed in your videos that this stuff is Deep Learning and is not extactly the Data Science. It is a separate branch in itself.