Mohamad Dayeh’s Post

🚀 Understanding Positional Embeddings in Transformers: From Absolute to Rotary Dive into absolute, relative, and rotary positional embeddings with code examples! 🧠 📍 Absolute embeddings give precise word positions but may cause sensitivity to position info. ↔️ Relative embeddings learn word-to-word distances, improving generalization across sentence lengths. 🔄 Rotary embeddings encode positions as rotations, better handling long-distance dependencies and often outperforming traditional methods. #AI #MachineLearning #Transformers #NLP #DeepLearning #PositionalEmbeddings #TechInnovation #DataScience

Training a healthcare language model: It's like solving a Rubik's cube, but with hyperparameters instead of colors." 🤯 You endlessly twist and turn, hoping that this time, everything will click into place. Hyperparameter optimization - it's a journey of patience, persistence... and maybe a little bit of luck! Anyone else feel like they've got a closet full of almost-solved cubes from their fine-tuning adventures? 😂 #SLM #healthcareAI #NLP #machinelearning #RubiksCubeOfAI"

🚀 LLM Context Size vs. Input Context Size Context Size: This is the maximum amount of text the model can consider at once. Larger context sizes allow the model to understand and generate more coherent and contextually accurate responses. Input Context Size: This refers to the actual length of the text you provide to the model. While the input can vary, it must stay within the model’s context size to be processed effectively. #AI #MachineLearning #DeepLearning #ContextSize #NLP #llm

Song generation using #ai. over the weekend i had the chance to experiment with suni.ai(complete link in comments) a website where you can genearate songs using just a prompt! i myself experimented with 3 prompts and results are scary good!. pair this with #sora model form #openai , i don't know whether this will do more harm than good for the creator market. but definitely things are moving at a much faster pace than anticipated. As always, keep learning ! #datascience #datascientist #machinelearning #nlp #deeplearning #genai

Evaluating summarization tasks with LLMs involves a range of metrics, each capturing different aspects of summary quality: • ROUGE: Focuses on n-gram overlap and sequence alignment. • BLEU: Precision-based metric with brevity penalty. • METEOR: Considers synonymy, stemming, and alignment. • BERTScore: Uses contextual embeddings for semantic similarity. • Content Overlap Metrics: Assesses content coverage using QA-based methods. Each metric provides unique insights, and often a combination of these metrics is used to comprehensively evaluate the performance of summarization models. #llm #machinelearning #nlp #naturallanguageprocessing #bert #ai

Hi #connections, I created a simple next word predictor using LSTM. I have used youtube wikipedia data to train the model.Although the model isn't fine-tuned, it performed well in predicting the next word based on user input. While it may make mistakes, it serves as a good demo to showcase the concept. Your feedback is valuable, and I'm open to hearing about any similar ideas you have in mind. Let's continue learning and exploring new creative concepts together. #AI #MachineLearning #Innovation #datascience #lstm #nlp #ideas #genai #share

Enhancing 70% of RAGs performance can be achieved by ensuring proper documentation and high-quality search capabilities. After that, we can fine-tune the parameters for better results. For better RAG results: - Avoid Neglecting Documentation and Search Capabilities: - Avoid Ignoring Data Quality - Avoid Excessive Complexity - Avoid Ignoring Hyper-parameter Tuning - Avoid High Penalty Values for Creativity: - Avoid Low Temperature for Rigid Answers: - Avoid Ignoring Model Evaluation: #RAGPerformance #Documentation #SearchCapabilities #ParameterTuning #MachineLearning #AI #NaturalLanguageProcessing #NLP #CreativeAI #TechInnovation #AIOptimisation

Hello LinkedIn Community! 👋 I'm SaiVandani Uppalapati, and today I’m excited to dive into text summarization using the facebook/bart-large-cnn from the Hugging Face library. Text summarization is a valuable technique that condenses lengthy texts into brief, key-point summaries.. Check out the full video to see the process in action. #AIMERS #TextSummarization #MachineLearning #ArtificialIntelligence #DataScience #NLP #DeepLearning #Transformers #HuggingFace #AI #TechPresentation

Discover the Future of Marketing: AI's Impactful Real-World Applications! Are you ready to see marketing's future unfold? Deep dive into AI's real-world impact with us! From transforming data to personalized experiences, let's explore the game-changing possibilities! Don't wait to start your journey – enroll at House of Learning today and reshape your future! WhatsApp Now: 0329-5777555 #digitalmarketing #artificialintelligence #dataanalysis #NLP #decisionmaking

The number of genAI project SOWs we are signing and delivering weekly has become alarmingly high. We are discovering Retrieval-Augmented Generation (RAG) use cases from clients which one could never imagine in their dreams. We would be happy to look at your business processes and identify scopes of automation using AI in your organization. DM me and book your free 30 minute consultation today! #GenAI #RAG #AI #TheInsightsMachine #actionableinsights #NLP