AI

The Dimensionality of Vector Space is the key to todays AI Technologies, especially, Semantic Search, RAG, and similar concepts. Using Mathematical Concepts that have been around for a very long time, and applying them to Compute, in smart, Intelligent ways.

Created: 5/28/2024 6:15:13 PM

Exploring how knowledge is represented in a vector space, a mathematical construct that allows for efficient processing and manipulation of information by using vectors to capture the essence of concepts, ideas, or objects, with various types of vectors (word, document, concept, entity) and operations (addition, scalar multiplication, inner product) enabling the storage, retrieval, and manipulation of knowledge.

Created: 5/28/2024 7:16:30 PM

The perceptron is a type of artificial neural network invented by Frank Rosenblatt in 1958. Designed to model the human neuron, it serves as a fundamental building block in the field of machine learning. The perceptron algorithm takes a set of input values, applies a set of weights, sums them up, and passes this sum through an activation function to produce an output.

Created: 11/14/2024 5:13:22 PM

As a self-proclaimed AI red teamer, Pliny has developed several harmless jailbreak prompts and tools designed to test and expose vulnerabilities in AI systems. One of his notable exploits, "GODMODE GPT," allowed users to bypass restrictions in the GPT-4o model, enabling it to perform tasks like swearing and providing dangerous instructions.

Created: 11/22/2024 2:20:56 PM

The Berkeley Function Calling Leaderboard V3 (also called Berkeley Tool Calling Leaderboard V3) evaluates the LLM's ability to call functions (aka tools) accurately. This leaderboard consists of real-world data and will be updated periodically.

Created: 3/27/2025 10:44:59 AM

Digital Intelligence is undoubtedly one of the most transformative forces of the 21st century. It is reshaping industries, altering human interaction, and presenting both incredible opportunities and significant challenges.

Created: 3/31/2025 5:58:03 PM

Reinforcement learning with verifiable rewards (RLVR) has shown promise in enhancing the reasoning capabilities of large language models by learning directly from outcome-based rewards. Recent RLVR works that operate under the zero setting avoid supervision in labeling the reasoning process, but still depend on manually curated collections of questions and answers for training.

Created: 5/13/2025 10:05:35 AM