The AI Buzz with Luca and Josh is also a podcast! Check it out on your favorite platform, including Spotify: open.spotify.com/show/06580WpFqTt27tIbzBS8VQ?si=d5c72b581bb84fb0 To learn more about Lightning: lightning.ai/ Support StatQuest by buying my book The StatQuest Illustrated Guide to Machine Learning or a Study Guide or Merch!!! statquest.org/statquest-store/

Comment to say a big thank you for this awesome channel and content! So so comprehensive, actual and well made!

@statquest

Жыл бұрын

Thank you very much!

Love you Josh.. From India

@statquest

Жыл бұрын

Thanks!

great series!

@statquest

Жыл бұрын

Thank you! :)

The reward model and PPO is quite nicely summarised. I would also add Geometric Deep Learning as one of the next future subjects which will be quite a big BAM

@statquest

Жыл бұрын

We'll keep that in mind! :)

Seriously important and timely conversation. Diagrams and pictures remain information dense but not accessible to chatGPT, for now. Safeguards remain at the core - DARPA may require different alignments, for example. Old UK duffer here, enjoying the human chat.

@statquest

Жыл бұрын

BAM! :)

Bam

@statquest

Жыл бұрын

YES! :)

The following summary of this video is generated using the openai whisper api, is it good? The text discusses the latest trends in AI, and whether or not they are being overselled. Luca argues that while some people may be concerned that these techniques are leading to artificial intelligence that is too human-like, the reality is that they are simply providing more value than ever before. He cites the example of a person in HR who is using AI to help screen resumes, and notes that this is just one example of how AI is benefiting people in a direct way.,The text discusses the development of chatbots and how they are becoming more natural and fluent. It also mentions that there are many open source alternatives to the chatbot GPT3, which is based on scraped data.,The text describes a method for training a machine learning model to generate text. The first step is to pre-train the model by feeding it a large amount of text. The second step is to fine-tune the model by providing it with pairs of questions and answers. The model is then further improved by using reinforcement learning, which involves training the model to produce text that results in a positive outcome (reward).,The text describes how artificial agents can be trained to make decisions based on probabilities, in order to optimize their rewards. This technique can be used to control players in games, or to simulate processes that are difficult to express. The agents may sometimes take unusual decisions, in order to explore different options and find the best possible reward.,The text discusses how reinforcement learning can be used to fine-tune AI models, with a focus on how a reward model can be used to align the model with human expectations.,The text discusses the idea that it is now possible to create models that are much better aligned with human values, and that this could potentially lead to a future where AI is much better behaved.,The text discusses the potential for artificial intelligence (AI) to be used for evil purposes. It notes that while AI models are becoming more sophisticated, they are still ultimately based on simple techniques that are reproducible and attainable.,The text describes how Lightning makes it possible to create applications that are easy to scale and that don't require a lot of platform engineers to maintain.

@statquest

Жыл бұрын

It's not terrible.

@dihancheng952

Жыл бұрын

@@statquest Would it be alright with you if I posted a summary of all 5 videos in this playlist? These videos are rather long and unfortunately, I don't have enough time to watch them all. So I used Whisper AI to generate a summary that may be also useful for other viewers.

@statquest

Жыл бұрын

@@dihancheng952 sure!

@dihancheng952

Жыл бұрын

@@statquest thanks, I posted.

Double bam

@statquest

Жыл бұрын

Thank you! :)

Josh, is this live?

@statquest

Жыл бұрын

No, we recorded it yesterday because last time we had problems with the audio.

Could you make a video on transformers as AI is so hyped right now.

@statquest

Жыл бұрын

I'm working on it. Unfortunately I'm slow. :(

@dhiraj_shah

Жыл бұрын

@@statquest Do you think a transformer model for video would be possible? We can use our existing image labeling AI, voice recognization technology to label and transcribe everything and feed it to bigger model. If such is the possability, by when do you think we will have similar technology publicly revealed. Also your videos have been very helpful and thank you for the great work :)

@statquest

Жыл бұрын

@@dhiraj_shah They're using transformers for images already, so I don't suspect it will be long before they are using them for video.