Artificial Intelligence has made significant strides, yet some challenges persist in advancing multimodal reasoning and planning capabilities. Tasks that demand abstract reasoning, scientific understanding, and precise mathematical computations often expose the limitations of current systems. Even leading AI models face difficulties integrating diverse types of data effectively and maintaining logical coherence in their responses. Moreover,… →

Modern NLP applications often demand multi-step reasoning, interaction with external tools, and the ability to adapt dynamically to user queries. Haystack Agents, an innovative feature of the Haystack NLP framework by deepset, exemplifies this new wave of advanced NLP capabilities. Haystack Agents are built to handle scenarios requiring: Complex multi-step reasoning. Integration of external tools… →

Out of the various methods employed in document search systems, “retrieve and rank” has gained quite some popularity. Using this method, the results of a retrieval model are re-ordered according to a re-ranker. Additionally, in the wake of advancements in generative AI and the development of Large Language Models (LLMs), rankers are now capable of… →

Large Language Models (LLMs) have become pivotal in artificial intelligence, powering a variety of applications from chatbots to content generation tools. However, their deployment at scale presents notable challenges. High computational costs, latency, and energy consumption often limit their wider use. Organizations face the difficulty of balancing high throughput with reasonable operating expenses. Additionally, as… →

Have you ever admired how smartphone cameras isolate the main subject from the background, adding a subtle blur to the background based on depth? This “portrait mode” effect gives photographs a professional look by simulating shallow depth-of-field similar to DSLR cameras. In this tutorial, we’ll recreate this effect programmatically using open-source computer vision models, like… →

Lexicon-based embeddings are one of the good alternatives to dense embeddings, yet they face numerous challenges that restrain their wider adoption. One key problem is tokenization redundancy, whereby subword tokenization breaks semantically equivalent tokens, causing inefficiencies and inconsistencies in embeddings. The other limitation of causal LLMs is unidirectional attention; this means tokens cannot fully leverage… →