Setu Kumar Basak

Hello All, *BIG ANNOUNCEMENT* Here is the YouTube playlist for the video recordings for my entire course on Deep Generative Models - https://lnkd.in/gs8AxSrt These are 3-hour-long recordings of actual classrooms having ~40 hours of content. The course covers - Basics of Probability theory, Latent Variable models, Variational Divergence Mininimzation, GANs, VAEs, Diffusion Models, DDPMs, DDIMs, Self Supervised Learning, Noise conservative Estimation, and much more in a mathematically rigorous manner. It also has tutorials on probability theory, MLPs/CNNs, Backpropagation, basics of PyTorch, etc. I hope it is of use. To free knowledge and happy learning :) Prathosh. P.S: 1 - These are very detailed rigorous teaching content. Please don't expect `short-quick-outcomes'. This is for people who want to learn slowly and learn well. P.S. 2 - These are recordings of real classrooms and contain discussions, comments, and other (rarely) non-academic discussions. The accompanying lecture notes - https://lnkd.in/g5gQUXYr #deeplearning #genai #iisc #freecourse #aiml #learning

2,677

120 Comments

For developers and anyone working on web applications, understanding how to send GET requests using cURL is essential. cURL, a command-line tool, allows users to transfer data from or to a server with ease. This post explores the fundamental steps needed to initiate GET requests, detailing the syntax and possible parameters that can be utilized. By mastering these principles, developers can efficiently interact with APIs and retrieve necessary data for their applications. More information can be found at

1

My paper on Interpretable Tensor Fusion (https://lnkd.in/gjTxusR5) which can capture both linear combinations and multiplicative interactions of multimodal data is on Arxiv. Will also be presenting the work at IJCAI - 24 in Korea.

136

4 Comments

Excited to share that I recently discovered the Best MLOps Book ever - "Machine Learning Engineering in Action" by Benjamin Wilson. 📘 If you're ready to move beyond Toy Datasets and dive into a larger codebase, this book is a game-changer. I used to rely on blogs to understand code, but Wilson's insights on navigating the research space for production-ready code have been invaluable. 🚀 #ml #machinelearning #mlops #airesearcher #aiengineer #machinelearningengineer #airesearch #jobready

10

As a passionate Machine Learning enthusiast and practitioner, I am constantly amazed by the profound impact that optimization algorithms have on our ability to train models efficiently and effectively. One such fundamental algorithm that stands out is Gradient Descent. 🔍 What is Gradient Descent? Gradient Descent is an iterative optimization algorithm used to minimize the cost function in machine learning models. By iteratively adjusting the parameters of our models, it helps us find the optimal solution, leading to better predictions and improved performance. 🔧 How Does It Work? Initialization: Start with an initial guess of the model parameters. Compute the Gradient: Calculate the gradient of the cost function with respect to each parameter. Update the Parameters: Adjust the parameters in the opposite direction of the gradient. Repeat: Continue this process until convergence is achieved. ✨ Why It Matters? Gradient Descent is the backbone of many machine learning algorithms. Its simplicity and efficiency make it an indispensable tool for training models, from linear regression to deep neural networks. 🔍 Types of Gradient Descent: Batch Gradient Descent: Uses the entire dataset to compute the gradient. Stochastic Gradient Descent (SGD): Uses one random sample from the dataset at a time. Mini-Batch Gradient Descent: Uses a small batch of samples to compute the gradient. 🌟 My Experience: Having worked extensively with machine learning models, I’ve seen firsthand the power of Gradient Descent. Whether it’s tuning hyperparameters or fine-tuning neural networks, understanding and leveraging this algorithm has been key to achieving high accuracy and robust models. 📈 Key Takeaways: Gradient Descent helps in optimizing and fine-tuning machine learning models. It’s essential to choose the right variant (Batch, SGD, Mini-Batch) based on your specific use case. Properly tuning the learning rate is crucial for the success of the algorithm. Let’s continue exploring and pushing the boundaries of what’s possible with machine learning! If you have experiences or insights related to Gradient Descent, I’d love to hear them in the comments. Important Note: Mohammad Anwar Khan and Hassan Syed along with iCodeGuru also help many like me to understand these concepts easily by organizing daily Gen AI workshops at 7:00 PM Pak time. I am also updating my GitHub repository with all these concepts https://lnkd.in/d6vVhpHZ Feel free to fork and star this repo. #GenerativeAI #MachineLearning #ArtificialIntelligence #OpenSource #GitHub #AIResearch #Collaboration #Innovation #TechCommunity #AITraining #ProfessionalGrowth #PakAngels #AspirePakistan #iCodeGuru #TheEngineersPakistan #4Brothers #Hackathon #TechEducation #SoftwareEngineer #LeetCode #GenAIDeveloper #MachineLearning #GradientDescent #AI #DataScience #Optimization #DeepLearning

18

1 Comment

Today we will explore key concepts and innovations that have revolutionized Natural Language Processing (NLP). Here's a glimpse of what we'll cover: Introduction to Transformer Models: Overview of Transformer architecture and its significance in NLP. Multi-Headed Self-Attention: Explanation of multi-headed self-attention, its function, and its advantages over single-head attention. Positional Encoding: Understanding the role of positional encoding in Transformers to maintain the order of sequences. Attention Mechanism and Its Components: Detailed breakdown of query, key, and value matrices in the attention mechanism. Residual Connections: Importance of residual connections in mitigating the vanishing gradient problem and facilitating deeper networks. Feedforward Neural Networks in Transformers: Role of feedforward neural networks post-attention mechanism to add non-linearity and more parameters. Depth and Layer Stacking: Adding depth with multiple layers of self-attention and feedforward networks for enhanced learning. Encoder-Decoder Architecture: Structure and function of the encoder-decoder architecture in Transformers, highlighting their role in sequence-to-sequence tasks. Practical Applications and Innovations: Real-world applications and recent advancements using Transformer models. A big thanks to: Prof. Mohit Iyyer (UMass Amherst) 🎓 Jay Alammar 📝 Krish Naik 💻 Their work has been instrumental in making these complex concepts clearer. 🙌 Want to learn more? Here are some helpful resources: Attention Is All You Need (Research paper) 📚[https://lnkd.in/e4c5Fmqq] The Illustrated Transformer 🎨[https://lnkd.in/e8mVButh] Understanding Transformers (by Jay Alammar) 📺[https://lnkd.in/eBfzMp6w] Stay tuned for upcoming posts as we explore Transformers in more detail! ➡️ hashtag #NLP #AttentionMechanism #Transformer #MachineLearning P.S. Feel free to share your thoughts and questions in the comments! 💬 Stay tuned for more in-depth posts and discussions on these exciting topics! 💡💬 #NLP #TransformerModels #DeepLearning #AI #MachineLearning

8

//Algorithm for avoiding unnecessary expenses// 😍for code lovers... Only😍 //Continuously ask this question..in If.. else loop...while you are in the shopping malls// While (in shop): {{ Vision sensor data= read. Object;} If (do I really need the above object ==Vision sensor data): { //go for it if above condition is true } else: { //leave it } }

21

1 Comment

https://lnkd.in/dmwAUPYC I'm honored to announce that my research article, "Engineering BERT Model Adoption for Sarcasm Detection on Twitter Data," has been published! As the lead author, I'm proud to have contributed to the advancement of natural language processing (NLP) and artificial intelligence (AI) research. #BERT #SarcasmDetection #TwitterData #NLP #AI #Research #Publication #LeadAuthor #Engineering #Innovation #ArtificialIntelligence #MachineLearning

8

6 Comments

Ready to learn some Real World 𝗟𝗟𝗠 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴? Here is a straight-to-the-point-no-BS book to learn how to -> Design -> Build, and -> Deploy by Paul Iusztin and Maxime Labonne. The authors take you step by step from a concept to a fully working LLM system, using MLOps best practices. Git hub: https://lnkd.in/g4qJvmaD

7

𝗧𝗵𝗲 𝗨𝗹𝘁𝗶𝗺𝗮𝘁𝗲 𝗥𝗲𝘀𝗼𝘂𝗿𝗰𝗲 𝗟𝗶𝘀𝘁 𝘁𝗼 𝗞𝗶𝗰𝗸𝘀𝘁𝗮𝗿𝘁 𝗬𝗼𝘂𝗿 𝗝𝗼𝘂𝗿𝗻𝗲𝘆 𝗶𝗻 𝗔𝗜/𝗠𝗟 Breaking into AI/ML can feel overwhelming, but the right resources can make all the difference. Whether you’re a beginner or looking to deepen your expertise, here’s a curated list of tools, courses, and communities to get you started. 📌 𝗖𝗼𝘂𝗿𝘀𝗲𝘀 & 𝗧𝘂𝘁𝗼𝗿𝗶𝗮𝗹𝘀 • Andrew Ng’s Machine Learning (Coursera): https://lnkd.in/gJ9B3Efc • Deep Learning Specialization (Coursera): https://lnkd.in/gwqSPS4J • Fast.ai’s Practical Deep Learning for Coders: https://course.fast.ai/ • Hugging Face’s Transformers Course: https://lnkd.in/gxEuGtvu 📌 𝗕𝗼𝗼𝗸𝘀 • Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow by Aurélien Géron. • Deep Learning by Ian Goodfellow, Yoshua Bengio, and Aaron Courville. • Pattern Recognition and Machine Learning by Christopher Bishop. 📌 𝗖𝗼𝗺𝗺𝘂𝗻𝗶𝘁𝗶𝗲𝘀 & 𝗙𝗼𝗿𝘂𝗺𝘀 • Kaggle (Join Here: https://www.kaggle.com/) • Machine Learning Mastery’s Community (Join Here: https://lnkd.in/g6RyqcV2) • Deep Learning.AI Community (Join Here: https://lnkd.in/gGCDPWQ2) 📌 𝗧𝗼𝗼𝗹𝘀 & 𝗟𝗶𝗯𝗿𝗮𝗿𝗶𝗲𝘀 • Google Colab (Try Here: https://lnkd.in/gzYAfUx7) • Scikit-learn (Explore Here: https://scikit-learn.org/) • Hugging Face (Explore Here: https://huggingface.co/) • TensorFlow and PyTorch (TensorFlow: https://lnkd.in/gwez3CKW, PyTorch: https://pytorch.org/) With these resources, you have everything you need to start—or continue—your AI/ML journey. What’s a resource that’s helped you the most? Let’s add it to the list! #AI #MachineLearning #Resources #DeepLearning #GrowthMindset

14

2 Comments

Here’s a blogpost that breaks down our work on Attribute Controlled Fine-tuning for Large Language Models . We introduced a new method that trains an auxiliary model to control a specific attribute (in this case toxicity). The approach regularizes the LLM's training by penalizing deviations from the desired (non-toxic) distribution, using the auxiliary model trained alongside the core LLM. This work, published at #EMNLP2024, was led by our intern Tao Meng (UCLA) together with our collaborators Ninareh Mehrabi, Palash Goyal, PhD, Anil Ramakrishna, PhD, Aram Galstyan, Richard Zemel, Kai-Wei Chang and Rahul Gupta. Blogpost: https://lnkd.in/eRzvKSKi Paper: https://lnkd.in/eHNwgWAt

34

#High_Dimensional_Metamodeling for Prediction of #Clock_Tree_Synthesis Outcomes by : Andrew B. Kahng, Bill Lin and Siddhartha Nath [ 2018 : 114 Slides : https://lnkd.in/gc3-ZfEj) #Machine_Learning Applications in #VLSI #Physical_Design : Recent Results and Directions by : Andrew B. Kahng UC San Diego ] [ 48 Slides + Video : https://lnkd.in/g4r_467R) : 04/07/2024 Way to Model Creation for CHIP DESINNG using AI #Large_Language_Models and #Generative_AI for #High_Performance_Systems Source Video : https://lnkd.in/gKCnGmrx by : Chris Cheng from HP Enterprise.] [ #CHINA : 26pGS PDF : https://lnkd.in/gbaFcEff) : https://lnkd.in/g94-CZdP) : 23/06/2023  #AI to Design #RISC_V #CPU in Under #5_Hours by : Shuyao Cheng & Others ] [ Cadence & China : 16 Slides : https://lnkd.in/gsspcYHH) : 2018 DeePattern: Layout Pattern Generation with Transforming Convolutional Auto-Encoder by : Haoyu Yang &..] [ China : 25 Slides : https://lnkd.in/gMqzYF5m) : 14/03/2024 #IncreMacro: Incremental #Macro_Placement_Refinement by : Yuan Pu &... ] [ 31 Slides + 6gs : https://lnkd.in/g8FvcTmG) : https://lnkd.in/g5NG8CW7) : 27/03/2023 #VLSI_DESIGN : Assessment of #Reinforcement_Learning for #Macro_Placement ] [https://lnkd.in/d23hnNQk) : https://lnkd.in/dmyD4KNg) : 12/09/2022 #Placement #Optimization via #PPA-Directed #Graph #Clustering ] [ #NVIDA : https://lnkd.in/dhSiEW37) : 26/03/2023 #Ai in #VLSI #Design : #AutoDMP: #Automated #DREAMPlace-based #Macro #Placement ] [ #NVIDIA : 6pgs : https://lnkd.in/gZ98GT6a) : 24/05/2024  : Large Language Model (#LLM) for #Standard_Cell_Layout Design #Optimization by : Chia-Tung Ho &..] [ #nvidia : (41+500) Slides + 182pgs Paper + Video : https://lnkd.in/dduXctSF) Mar 23, 2023 #ILT #NVIDIA_cuLITHO : #40X #Accelerating #Computational #Lithography by Vivek Singh (VP NVIDIA) OPC : OPTICAL PROXIMITY CORRECTION PROBLEM SOLVED (via CurviLinear_Musk Making) VIA ILT #NVIDIA, #ASML, #TSMC and #Synopsys Set Foundation for #Next_Generation #Chip_Manufacturing.] [ https://lnkd.in/dW8NEb6d) : 10/02/2023 #AI-#Powered #Chip #Design Goes Mainstream AI in chip design has gone mainstream, with Synopsys’ #AI-#Powered #DSO (#DesignSpaceOptimization) tool reaching #100commercial #Tapeouts.] [ #OPENROAD : https://lnkd.in/dp9WYFmj) : 27 Oct 2021 #OpeNPDN in #VLSI #Design : #NeuralNetworks for #Automated #PowerDeliveryNetwork #Synthesis ] [ #Google : 9pg Slides : https://lnkd.in/d8PRh6id) Circuit Training: An #Open_Source framework for generating #Chip #Floor_Plans with distributed #Deep_Reinforcement_Learning. by : Sergio Guadarrama ] [ #GOOGLE : https://lnkd.in/epd3X3pi) : April 23, 2020 : Update, June 9, 2021: #Chip #Design with #Deep #Reinforcement #Learning ] [ 9pgs PDF + Video : https://lnkd.in/dpqz4cj5) : 22 May 2023 #Chip_Chat : Challenges and Opportunities in #Conversational_Hardware_Design ] [ 7pgs : https://lnkd.in/gS-i36r4) : 22/02/2024 #Hopkins_Engineers Chat with #ChatGPT4 to #Design #Brain_Inspired_Chips ]

10

OpenHands: Revolutionizing Code Generation and Automation with Open Source Innovation Open source enthusiasts, gather round! OpenHands, inspired by the commercial Devin, is changing the game in advanced code generation and automation. Built by a robust team from leading institutions like the University of Illinois Urbana-Champaign and Yale, this toolkit is freely available for download and modification. Key Features: - Operates through a set of agents, leveraging large language models to generate, edit, and run code autonomously. - Securely runs in Docker containers, interfacing with web resources and employing auxiliary task tools such as text extraction and audio transcription. - The CodeAct agent enhances coding efficiency by following a structured agentic workflow, asking clarifications, executing code, and fetching web data. - Additional "micro agents" manage tasks ranging from database writing to problem-solving, customizable through unique user prompts. Performance Insights: - CodeAct using GPT-4-turbo achieved impressive results with 51.8% accuracy on GPQA Diamond, outpacing traditional GPT-4’s 38.8%. - Demonstrated prowess in software engineering and web browsing tasks, showcasing comparable performance to established solutions. Implications and Impact: OpenHands is set to democratize code automation, making advanced tools accessible for developers to innovate and non-programmers to explore uncharted territories. Engagements are welcome. What ideas do you have for leveraging OpenHands in your workflow? #OpenSource #CodeGeneration #AI #Automation #MachineLearning #AgenticSystems #OpenHandsInnovation

1

Small Language Models (SLMs) can serve as effective alternatives to LLMs for targeted applications, including AI-powered educational tools. In our ProductiveMath project, we are exploring the potential of SLMs such as Phi 3.5 compared to LLMs in two key areas: generating math problems tailored for productive failure and predicting common misconceptions. This paper is a good read about SMLs: https://lnkd.in/g-RnFTqf

23

1 Comment

📜 DrEureka: Language Model Guided Sim-To-Real Transfer Researchers from University of Pennsylvania and NVIDIA have developed DrEureka - an LLM-guided sim-to-real approach that automatically designs reward functions and domain randomization for robot task simulations, facilitating efficient and scalable skill transfer to real-world applications. Building on previous Eureka advancements, this method reduces the need for manual intervention and accelerates the deployment of robust, real-world robot models capable of complex tasks like quadruped balancing on a yoga ball in the demo. Website: https://lnkd.in/gvw4hTyn Paper: https://lnkd.in/gXPfVBHz GitHub repo: https://lnkd.in/gKKWjMmk #keep_AI_posted #ai #ml #nlp #llm #eureka #simtoreal

3