This survey reviews works in which language models (LMs) are augmented with reasoning
skills and the ability to use tools. The former is defined as decomposing a potentially …

Recent studies have discovered that Chain-of-Thought prompting (CoT) can dramatically
improve the performance of Large Language Models (LLMs), particularly when dealing with …

Neural sequence models based on the transformer architecture have demonstrated
remarkable\emph {in-context learning}(ICL) abilities, where they can perform new tasks …

Several recent works demonstrate that transformers can implement algorithms like gradient
descent. By a careful construction of weights, these works show that multiple layers of …

Foundation models pretrained on diverse data at scale have demonstrated extraordinary
capabilities in a wide range of vision and language tasks. When such models are deployed …

Algorithmic reasoning requires capabilities which are most naturally understood through
recurrent models of computation, like the Turing machine. However, Transformer models …

Instructing the model to generate a sequence of intermediate steps, aka, a chain of thought
(CoT), is a highly effective method to improve the accuracy of large language models (LLMs) …

Large language models like GPT-4 exhibit emergent capabilities across general-purpose
tasks, such as basic arithmetic, when trained on extensive text data, even though these tasks …

Transformers have recently revolutionized many domains in modern machine learning and
one salient discovery is their remarkable in-context learning capability, where models can …

Recent research in mechanistic interpretability has attempted to reverse-engineer
Transformer models by carefully inspecting network weights and activations. However, these …