Large Language Model

Memory is a crucial component of intelligence, facilitating the recall and application of past experiences to current situations. However, both traditional Transformer models and Transformer-based Large Language Models (LLMs) have limitations related to context-dependent memory due to the workings of their attention mechanisms. This primarily concerns the memory consumption and computation time of these attention…

Recent research in Artificial Intelligence (AI) has shown a growing interest in the capabilities of large language models (LLMs) due to their versatility and adaptability. These models, traditionally used for tasks in natural language processing, are now being explored for potential use in computational tasks, such as regression analysis. The idea behind this exploration is…

Chain-of-thought (CoT) prompting, an instruction method for language models (LMs), seeks to improve a model's performance across arithmetic, commonsense, and symbolic reasoning tasks. However, it falls short in larger models (with over 100 billion parameters) due to its repetitive rationale and propensity to produce unaligned rationales and answers. Researchers from Penn State University and Amazon AGI…

Large Language Models (LLMs), known for their key role in advancing natural language processing tasks, continue to be polished to better comprehend and execute complex instructions across a range of applications. However, a standing issue is the tendency for LLMs to only partially follow given instructions, a shortcoming that results in inefficiencies when the models…

Large Language Models (LLMs) have surpassed previous generations of language models on various tasks, sometimes even equating or surpassing human performance. However, it's challenging to evaluate their true capabilities due to potential contamination in testing datasets or a lack of datasets that accurately assess their abilities. Most studies assessing LLMs have focused primarily on the English…

Neural network models are dominant in the areas of natural language processing and computer vision. However, the initialization and learning rates of these models often depend on heuristic methods, which can lead to inconsistencies across different studies and model sizes. The µ-Parameterization (µP) seeks to address this issue by proposing scaling rules for model parameters…

Researchers from Mila, McGill University, ServiceNow Research, and Facebook CIFAR AI Chair have developed a method called LLM2Vec to transform pre-trained decoder-only Large Language Models (LLMs) into text encoders. Modern NLP tasks highly depend on text embedding models that translate text's semantic meaning into vector representations. Historically, pre-trained bidirectional encoding models such as BERT and…

Computational linguistics has seen significant advancements in recent years, particularly in the development of Multilingual Large Language Models (MLLMs). These are capable of processing a multitude of languages simultaneously, which is critical in an increasingly globalized world that requires effective interlingual communication. MLLMs address the challenge of efficiently processing and generating text across various languages,…