- Видео 11
- Просмотров 11 286
Neural Black Magic
США
Добавлен 16 мар 2023
I'm an enthusiastic AI Researcher, covering the latest trends in #MachineLearning, #DeepLearning with implementations in #PyTorch, #PyTorch_Lightning.
Byte Latent Transformer: Patches Scale Better Than Tokens Paper Explained Visually and Clearly
This video provides the most straightforward clear explanation of the newly paper published by Meta, called "Byte Latent Transformer: Patches Scale Better Than Tokens".
You can read the paper here:
ai.meta.com/research/publications/byte-latent-transformer-patches-scale-better-than-tokens/
In this paper they introduce the Byte Latent Transformer (BLT), a new byte-level LLM architecture that, for the first time, matches tokenization-based LLM performance at scale with significant improvements in inference efficiency and robustness.
BLT encodes bytes into dynamically sized patches, which serve as the primary units of computation. Patches are segmented based on the entropy of the next byte, allo...
You can read the paper here:
ai.meta.com/research/publications/byte-latent-transformer-patches-scale-better-than-tokens/
In this paper they introduce the Byte Latent Transformer (BLT), a new byte-level LLM architecture that, for the first time, matches tokenization-based LLM performance at scale with significant improvements in inference efficiency and robustness.
BLT encodes bytes into dynamically sized patches, which serve as the primary units of computation. Patches are segmented based on the entropy of the next byte, allo...
Просмотров: 381
Видео
Efficient Infinite Context Transformers with Infini-Attention (Paper Explained)
Просмотров 1,5 тыс.9 месяцев назад
❤️Support the channel❤️ Hi everyone. In this video, a comprehensive explanation of the recently presented paper "Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention" which is introduced by google researchers is provided. Paper link: arxiv.org/abs/2404.07143 #LLM #nlp #LLMs #largelanguagemodels #transformers #Attention #deeplearning #naturallanguageprocessing
ConvMixer: Patches Are All You Need? (paper explained with implementation in PyTorch)
Просмотров 3,2 тыс.11 месяцев назад
❤️Support the channel❤️ In this video, ConvMixer architecture proposed in "Patches Are All You Need?" paper is explained informatively along with simple yet comprehensive PyTorch implementation. Paper link: arxiv.org/abs/2201.09792 GitHub repository: github.com/Ardawanism/ConvMixer-Patches-Are-All-You-Need #deeplearning #machinelearning #convolutionalneuralnetwork #convolutionalneuralnetworks #...
MLP-Mixer: An all MLP Architecture for Vision explained (with implementation in PyTorch)
Просмотров 1,4 тыс.11 месяцев назад
❤️Support the channel❤️ In this video, MLP-Mixer which is an all MLP architecture developed by google brain team for computer vision tasks is explained informatively along with simple yet comprehensive PyTorch implementation. Paper link: arxiv.org/abs/2105.01601 GitHub repository: github.com/Ardawanism/MLP-Mixer-An-all-MLP-Architecture-for-Vision #deeplearning #machinelearning #computervision #...
AlphaGeometry Explained: Solving olympiad geometry without human demonstrations
Просмотров 553Год назад
❤️Support the channel❤️ In this video, AlphaGeometry is explained clearly and informatively. Paper link: nature.com/articles/s41586-023-06747-5 Blog post by Google DeepMind: deepmind.google/discover/blog/alphageometry-an-olympiad-level-ai-system-for-geometry/ #deepmind #AlphaGeometry #llm
Dark Knowledge in Neural Networks - "Knowledge Distillation" Explanation and Implementation
Просмотров 3,2 тыс.Год назад
❤️Support the channel❤️ A clear and comprehensive explanation of Knowledge Distillation is presented in this video. In addition, the approach presented in the paper "Distilling the Knowledge in a Neural Network" is implemented from scratch using PyTorch. read the "Distilling the Knowledge in a Neural Network" paper: arxiv.org/abs/1503.02531 #deeplearning #machinelearning #pytorch #neuralnetwork...
Attention Mechanism In a nutshell
Просмотров 577Год назад
The Attention Mechanism has become a widely recognized concept in deep neural networks, extensively studied across diverse applications in the realm of Artificial Intelligence. The Attention Mechanism in deep learning allows models to dynamically concentrate on relevant parts of input data, enhancing their ability to understand context and relationships. In this video, I aim to explain the prin...
MLP-Mixer An all MLP Architecture for Vision explained clearly and coded from scratch using PyTorch
Просмотров 306Год назад
in this video, we review MLP-Mixer Architecture, An all-MLP Architecture for Vision, which is developed by google brain team and achieve competitive results to CNNs which have been dominant architecture in computer vision domain for a long time, and transformer based architectures. we do a quick review of the paper and mention key points and ideas and investigate the architecture in depth. if y...
Hey, nice explanation...Thanks, but it would be more helpful if you can perform this task on better datasets like CIFAR-10
Nice job.
Glad you liked this video. Thank you very much for your kind support.
Great video! Loved your insights on Meta's new paper and LLMs. Looking forward to more content like this
Thanks for your kind support dear🙏☘
Thank you for your insightful information 🙏🏻
Thanks for your kind support dear🙏☘ Glad you liked this video. Please subscribe to the channel and hit the bell icon to receive notifications when new videos are posted.
Insightful stuff👌
Thanks a lot for your warm support🤎🙏☘
Wonderful
Thank you very much for your kind support🙏☘ please subscribe to channel and hit the bell icon to receive notifications when new videos are posted.
Interesting
Thank you very much for your kind support🤎🙏☘
Great
Thanks a lot for your kind support🙏☘
Perfect 🔥
Thanks for your kind feedback Glad you liked it.
Please subscribe the channel and hit the bell icon to receive notifications when new videos are posted 🙏☘.
Thanks for sharing, more videos please🤩😍👊🔥
Glad you liked this video, sure! We'll have lot's new cool videos.
Wow🔥🔥 That was awesome🤎
Thanks for your kind support.
Thanks for explanation 👍
Glad you liked it🤩😍 Please follow the channel and press the bell icon to receive notifications when new videos are posted.
thank you so much, that was great
Glad you liked it🤩😍 Please follow the channel and press the bell icon to receive notifications when new videos are posted.
Thanks for sharing this helpful video! Keep going!💪
Thank you very much for your support my dear🙏☘
⚡⚡
Glad you liked it🙏
⚡⚡
Thanks for sharing this content
Thank you for your kind support🙏☘ Glad you liked it👊🔥
Hey there, thanks for the video, it was a great paper, I’ll go through it for more details. It was very related to the "Lost in the Middle" phenomenon and the use of the RAG systems in LLM applications that I was recently reading about. That would be great to share more of these contents if you cross over in your research.
Thank you very much for your kind support, positive feedback, and great comment🙏☘ Glad you liked it👊🔥 Sure, I do my best to provide high quality educational videos about machine learning and deep learning. I'm not familiar with the "Lost in the middle" phenomenon. Could you please explain it a little bit more so we can find similarities with the "Infini-Attention"?
Keep going mate ✌️❤️
Glad you liked it👊🔥 Thanks for your support🙏☘
Awesome..keep going
Thank you very much for your kind support and positive feedback🙏☘ Glad you liked it👊🔥
Wonderful and informative. Keep going
Glad you liked it 👊🔥 Thanks for your kind support 🙏🌹
The explanation was clear, I enjoy it 👍
Thanks for your positive feedback🙏☘ Glad you liked it👊
Great content but please improve the audio volume and quality
Thanks a lot for your feedback🙏☘ I will definitely do that👊
I agree 😊
@@erikdahlen2588 Thanks for your support and feedback🙏☘
@@erikdahlen2588 Please subscribe with notifications to receive notifications when new videos are posted🔥👊
"Attention is all you need", it was awesome.
Glad you liked it☘🙏 Thanks a lot for your positive feedback🔥
This video was very informative for ❤tnx for sharing
Thanks for your feedback, glad you liked it☘🙏
The "Infini-Attention" and "Infini-Transformer" seemed pretty cool :))
Yes, it was quite simple but interesting🤩
These video was extremely great and informative🔥 Thanks for sharing🙏
Thanks for your support🙏 Glad you liked it👊
You should make a video comparing the Vision Transformer and ConvMixer
That's a very good idea, I liked that, thanks. What do you think about the "ConvMixer" architecture any way? Share your ideas with me.
Please take a look at my other video, the "MLP-Mixer" at you convenience. It's a beautiful architecture too. Please share your ideas about that with me too. Thanks.
good video, you haven't thought about implementing the CapsNet architecture. It's also very good.
That's a great and interesting idea, thanks for your suggestion. I definitely consider that as topic of one of my future videos. What do you think about the "ConvMixer" architecture?
Please take a look at my other video, the "MLP-Mixer" at you convenience. It's a beautiful architecture too. Please share your ideas about that with me too. Thanks.
This is literally the best explabation of the "Attention Mechanism" I've seen so far! Great job, thank you very much for sharing👌🙏🔥
Thanks for your kind support🙏
😓 *promo sm*
interesting paper, and thank you for implementation as well, this architecture is so basicallly efficientnet with patch embedding, i am thinking that if this thing work, we should consider adding patch embedding on efficeintnet.
Hi, glad you liked it and thanks for your support🙏☘ Yes I agree with you, this architecture is very efficient, and adding patch embedding to EfficientNet seems very interesting. Please share the results with me if you did so. By the way, please subscribe to the channel and hit the bell icon to receive notifications when new videos are posted and stay updated🔥👊
👏⚡
Thanks a lot👊🔥 Please hit the bell icon to receive notifications when new videos are posted and stay updated✌
I genuinely find it incredibly useful, and I'm pleasantly surprised by the simplicity of your explanation, it's refreshingly accessible.
Thanks a lot for your great feedback👊🔥 Please hit the bell icon to receive notifications when new videos are posted and stay updated✌
Thank you for sharing this content! I genuinely find it incredibly useful, and I'm quite amused by the simplicity of your language. it makes the information really accessible and easy to understand.
Glad you liked it and thanks for your support🙏☘ Please subscribe to the channel and hit the bell icon to receive notifications when new videos are posted and stay updated🔥👊
Thanks for sharing this content, I found this incredibly insightful🌹🌹
Glad you liked it and thanks for your support🙏☘ Please subscribe to the channel and hit the bell icon to receive notifications when new videos are posted and stay updated🔥👊
Nice video
Thanks for your support☘🌻 Please follow the channel and hit the bell icon to receive notifications when new videos are posted💫
Great as always 👏👏
Thanks❤👊
Great job 👏🏻
Thanks a lot for your support☘🌻
I really appreciate the time and effort you put into creating the implementation section. I hope you continue this way!
Glad you liked it🌻 Thanks for your support🙏 I try to do my best👊🔥
Such a great video with an excellent explanation. Thanks for sharing
Glad you liked it🌻 Thanks for your support🙏 Please hit the bell icon to receive notifications when new videos are posted✌
👍👍
Glad you liked it🌻 Thanks for your support🙏 Please hit the bell icon to receive notifications when new videos are posted✌
👍👍👌👌
Glad you liked it🌻 Thanks for your support🙏 Please hit the bell icon to receive notifications when new videos are posted✌
Perfect video
Glad you liked it🌻 Thanks for your support🙏
Great as always 🌱
Thanks a lot🙏🌻
Thank you for your clear explanation🙏
Thanks for your feedback🌻 Glad you liked it👊🔥
Keep going to make such informative videos
Sure =) Thanks for you support 👊🔥