We mend this problem by adding an embedding to each digit that encodes its position relative to the start of the number.

In this work, we identify dropout induced sparsity forLSTMs asasuitable mode ofcomputation reduction. Dropout isawidely usedregularization mechanism, which randomly drops computed neuron values during each iteration of training.

Humans can length-generalize in integer addition because they understand the essential principle of the task. Nevertheless, itisobserved that Transformers typically learn to solve addition only up to the training sequence length (Lee et al., 2024), which is different from thetruearithmetic algorithm thathumans "implement".

Set theory is foundational to mathematics and, when sets are finite, to reasoning about the world. An intelligent system should perform set operations consistently, regardless of superficial variations in the operands. Initially designed for semantically-oriented NLP tasks, large language models (LLMs) are now being evaluated on algorithmic tasks. Because sets are comprised of arbitrary symbols (e.g.

Numerical reasoning over text is a challenging task of Artificial Intelligence (AI), requiring reading comprehension and numerical reasoning abilities. Previous approaches use numerical reasoning programs to represent the reasoning process. However, most works do not separate the generation of operators and operands, which are key components of a numerical reasoning program, thus limiting their ability to generate such programs for complicated tasks. In this paper, we introduce the numEricaL reASoning with adapTive symbolIc Compiler (ELASTIC) model, which is constituted of the RoBERTa as the Encoder and a Compiler with four modules: Reasoning Manager, Operator Generator, Operands Generator, and Memory Register. ELASTIC is robust when conducting complicated reasoning. Also, it is domain agnostic by supporting the expansion of diverse operators without caring about the number of operands it contains. Experiments show that ELASTIC achieves 68.96 and 65.21 of execution accuracy and program accuracy on the FinQA dataset and 83.00 program accuracy on the MathQA dataset, outperforming previous state-of-the-art models significantly.

Should LLM reasoning live in a separate module, or within a single model's forward pass and representational space? We study dual-architecture latent reasoning, where a fluent Base exchanges latent messages with a Coprocessor, and test two hypotheses aimed at improving latent communication over Liu et al. (2024): (H1) increase channel capacity; (H2) learn communication via joint finetuning. Under matched latent-token budgets on GPT-2 and Qwen-3, H2 is consistently strongest while H1 yields modest gains. A unified soft-embedding baseline, a single model with the same forward pass and shared representations, using the same latent-token budget, nearly matches H2 and surpasses H1, suggesting current dual designs mostly add compute rather than qualitatively improving reasoning. Across GSM8K, ProsQA, and a Countdown stress test with increasing branching factor, scaling the latent-token budget beyond small values fails to improve robustness. Latent analyses show overlapping subspaces with limited specialization, consistent with weak reasoning gains. We conclude dual-model latent reasoning remains promising in principle, but likely requires objectives and training schedules that explicitly shape latent spaces for algorithmic planning.

Modular exponentiation is crucial to number theory and cryptography, yet remains largely unexplored from a mechanistic interpretability standpoint. We train a 4-layer encoder-decoder Transformer model to perform this operation and investigate the emergence of numerical reasoning during training. Utilizing principled sampling strategies, PCA-based embedding analysis, and activation patching, we examine how number-theoretic properties are encoded within the model. We find that reciprocal operand training leads to strong performance gains, with sudden generalization across related moduli. These synchronized accuracy surges reflect grokking-like dynamics, suggesting the model internalizes shared arithmetic structure. We also find a subgraph consisting entirely of attention heads in the final layer sufficient to achieve full performance on the task of regular exponentiation. These results suggest that transformer models learn modular arithmetic through specialized computational circuits, paving the way for more interpretable and efficient neural approaches to modular exponentiation.

Kanerva (2014) suggested that it would be possible to construct a complete Lisp out of a vector-symbolic architecture. We present the general form of a vector-symbolic representation of the five Lisp elementary functions, lambda expressions, and other auxiliary functions, found in the Lisp 1.5 specification (McCarthy, 1960), which is near minimal and sufficient for Turing-completeness. Our specific implementation uses holographic reduced representations (Plate, 1995), with a lookup table cleanup memory. Lisp, as all Turing-complete languages, is a Cartesian closed category (nLab authors, 2024), unusual in its proximity to the mathematical abstraction. We discuss the mathematics, the purpose, and the significance of demonstrating vector-symbolic architectures' Cartesian-closedness, as well as the importance of explicitly including cleanup memories in the specification of the architecture.