PriceSenz’s Post

I had the honor of speaking at the AI Forum organized by Tech Titans: The Technology Association for North Texas. It was an incredible experience engaging with technology and business leaders on the topic GenAI: Simplifying Talent Matchmaking in HR-Tech. A big thanks to Tech Titans, Paul Bendel, Steven O'Neal, and Joshua Grob for the opportunity to share insights and discuss the future of AI in HR. Special thanks to Digital Realty for hosting the event at their impressive mega data center in Richardson. #AI #HRTech #GenAI #TechTitans #PriceSenz #DigitalRealty

🚀 Benchmarking GPTs – Featuring GPT-4o! 🚀 We are excited to share our latest in-depth analysis on the performance of various GPT models, with a special focus on the newly introduced GPT-4o (Omni). This article provides a comprehensive benchmarking of leading GPT models like GPT-4, GPT-4 Turbo, GPT-3.5, and GPT-4o across several key performance metrics. Learn about: - Response structure and accuracy - Hallucination rates - Match-making and inferencing capabilities - Execution times Discover why GPT-4o stands out in the AI landscape! #PriceSenz #AI #GPT4o #GenAI #MachineLearning

🚀 Exciting news! Our team has just released the latest LLM benchmarking report, "Benchmarking LLMs for HR Tech: Featuring LLama-3." We are thrilled to conclude that open-source LLMs are indeed here to stay! 🌟 Here are the key findings: 👉 Hallucinations: Llama3-70b > Claude3-sonnet >> GPT-4 👉 Structure/JSON Accuracy: GPT-4 = Claude3-sonnet ~ Llama3-70b 👉 Accuracy of Match Making: Llama3-70b > Claude3-sonnet >> GPT-4 👉 Inferencing Additional Information: Llama3-70b > Claude3-sonnet >> GPT-4 👉 Key Information Missed While Summarizing: Claude3-sonnet > Llama3-70b > GPT-4 #PriceSenz #LLM #HRtech #Benchmarking #LLama3 #Claude3 #GPT4

Our team's insights on LLM benchmarking for TalentTech... #talenttech #hrtech #staffingtech #pricesenz #llm

Last year, we took our first steps into the exciting world of Generative AI with Llama 2. What a ride it's been! We've experimented with a dozen different LLMs, from the well-known GPTs to Claude, and each one amazed us with its capabilities. Fast forward to today, and the LLM community is celebrating a huge milestone: Meta just rolled out Llama 3. Mark Zuckerberg put it perfectly: "The smallest Llama 3 is as big as the biggest Llama 2." That's a massive leap forward! This is a huge win for open source. We can’t wait to dive deeper into what Llama 3 can do. 🚀Meet Llama 3: Bigger, Better, Smarter Here’s the scoop on Llama 3: 👉More Words, Less Space: The new tokenizer in Llama 3 can handle 128K tokens, way up from Llama 2’s 32K. This means it can process information more efficiently, making everything smoother and faster. 👉Architecture: Every model in the Llama 3 lineup, big or small, uses something called Grouped Query Attention. This smart technology helps the model think and react faster, using less memory to do so. 👉Sequence Length: Llama 3 can remember up to 8192 tokens at once. That's double what Llama 2 could handle! It's not as high as the top dogs like GPT-4, but it's a big step up for handling longer conversations and more complex ideas. 👉Richer Training: Meta has beefed up Llama 3’s training with 15 trillion tokens. This isn't just more data; it's better data, with more code snippets and a splash of languages from around the world, making it more inclusive. 👉Power-Packed Performance: With 16K GPUs at work, Llama 3 cranks out operations at 400 TFLOPS. That's a lot of number crunching, handled really efficiently, thanks to some serious engineering magic. Llama 3 is not just a new model; it's a major leap in making AI more powerful and accessible, opening doors to exciting possibilities.

Attention Undergrad and Grad students: Exciting research opportunity in the area of AI / ML, and Organizational Psychology led by our team on cutting-edge AI-driven HR tech. CS/Data Science students: A keen interest in AI and ML, with skills in NLP, deep learning, or Gen AI toolkits. HR students: A strong interest or experience in organizational psychology. #pricesenz #internship #AcademicResearch