Hi Everyone. This is Vasanth. Welcome to Neural Hacks with Vasanth. I have 3+ years of experience in Data Science field. I love working in the field of NLP. Please support the channel by liking and sharing the videos with your communities and subscribing to my RU-vid channel.
Here is the detailed list of all the topics covered in this awesome tutorial 🔥🔥->> 00:05 Introduction to deep learning and neural networks 02:44 Deep learning automatically learns features and is highly scalable 07:25 PyTorch is a flexible and efficient deep learning library developed by Facebook AI. 09:53 Introduction to torch for complex calculations and GPU usage 15:16 Using PyTorch's Dataset and DataLoader for efficient data handling 17:35 Introduction to Data Loader and Neuron Networks 21:54 Introduction to Artificial Neural Networks (ANN) 23:59 RNN designed for sequential data handling 28:36 Understanding the architecture of a neural network. 31:06 Understanding Forward Propagation in Neural Networks 35:35 Calculating gradients and updating weights in deep learning 37:53 Activation function differentiation and weight impact 42:24 Moving towards global minima by taking small steps 44:26 Gradient Descent and Backpropagation in Neural Networks 48:19 Activation functions and layer initialization in neural networks. 50:26 Overview of common techniques in neural networks 54:39 Training Loop with Crossentropy Loss and Adam Optimizer 56:48 Activation functions in neural networks 1:00:59 Explanation of hyperbolic tangent function 1:03:07 Activation Functions and Error Propagation 1:07:39 Explanation of Activation Function Calculations 1:09:56 Choosing activation and loss functions for neural network 1:13:50 Explanation of binary cross entropy and optimizers in neural networks 1:15:58 Stochastic Gradient Descent (SGD) as a basic form of Optimizer 1:19:46 Understanding Recurrent Neural Networks (RNN) 1:22:13 Recurrent Neural Network (RNN) was developed to process sequential data. 1:26:41 Utilizing previous time step values in neural network forward pass 1:29:00 Explanation of back propagation Through Time in RNN 1:33:15 RNN has disadvantages in capturing long-term dependencies and is computationally expensive for long sequences 1:35:26 LSTM selectively forgets and remembers information over time 1:40:01 The tan layer provides intensity for memory retention in LSTM networks. 1:42:01 Explanation of LSTM functioning 1:46:13 LSTM is more complicated than RNNs and requires correct initialization and hyperparameters for training 1:48:28 LSTM in NLP requires understanding full context sequence 1:52:49 Comparison between LSTM and GRU in NLP 1:54:51 GRU is computationally efficient compared to LSTMs 1:59:18 Overview of the code structure and usage of key libraries 2:01:21 Key considerations for NLP model configuration 2:05:34 Explanation of embedding and hidden dimensions for NLP models. 2:07:33 Initializing the model for text generation 2:11:30 Creating vocabulary and data processing for NLP 2:13:32 Preparing dataset and training LSTM model for story generation.
I would like to ask, sir: If I have 100K documents to train a BPETokenizer from scratch, is it better to train iteratively on each document (.txt file) or combine all the documents into a single .txt file and then train on that? Thank you.
Thanks a lot Vasanth, I am learning about the Transformer Architecture from You and Campus X channel, and now you've uploaded the video of its implementation. I can code along while learning its theory. Thanks a lot.
You made my weekend. I found your videos today and they were great. Thank you so much! 👍🏼 I have a question. I have fine-tuned Llama 3.1 8B model for data extraction with the OCR texts of the invoices. For fine tuning I used “llama recipes” and dataset I prepared like samsum dataset. But the results does not look so good. Since in accounting have different approach, the invoices are very different. For data extraction what would you recommend? Which model would be better than Llama for this?
My dear teacher, I'm feeling lost and confused about the roadmap and path to learn LLM. Do I need to study NLP first and then learn LLM, or can I start learning LLM directly?
Very informative. Thank you for covering all the basics. looking forward for next videos in this playlist. One request I've just graduated want to apply for jobs related to Gen AI. This playlist would be very useful. Please upload it frequently without any delay. Thank you so much and appreciate what you're doing!!
Very informative. Please keep posting and want to really appreciate you for deciding to post this playlist. Please keep sharing knowledge. Thank you for doing this to us.