Limits of Transformer Language Models on Learning Algorithmic Compositions

We analyze the capabilities of Transformer language models on learning discrete algorithms. To this end, we introduce two new tasks demanding the composition of several discrete sub-tasks. On both training LLaMA models from scratch and prompting on GPT-4 and Gemini we measure learning compositions of learned primitives. We observe that the compositional capabilities of state-of-the-art Transformer language models are very limited and sample-wise scale worse than relearning all sub-tasks for a new algorithmic composition. We also present a theorem in complexity theory, showing that gradient descent on memorizing feedforward models can be exponentially data inefficient.

Limits of Transformer Language Models on Algorithmic Learning

We analyze the capabilities of Transformer language models on learning discrete algorithms. To this end, we introduce two new tasks demanding the composition of several discrete sub-tasks. On both training LLaMA models from scratch and prompting on GPT-4 and Gemini we measure learning compositions of learned primitives. We observe that the compositional capabilities of state-of-the-art Transformer language models are very limited and sample-wise scale worse than relearning all sub-tasks for a new algorithmic composition. We also present a theorem in complexity theory, showing that gradient descent on memorizing feedforward models can be exponentially data inefficient.

TCNCA: Temporal Convolution Network with Chunked Attention for Scalable Sequence Processing

MEGA is a recent transformer-based architecture, which utilizes a linear recurrent operator whose parallel computation, based on the FFT, scales as $O(LlogL)$, with $L$ being the sequence length. We build upon their approach by replacing the linear recurrence with a special temporal convolutional network which permits larger receptive field size with shallower networks, and reduces the computational complexity to $O(L)$. The resulting model is called TCNCA, a Temporal Convolutional Network with Chunked Attention. We evaluate TCNCA on EnWik8 language modeling, long-range-arena (LRA) sequence classification, as well as a synthetic reasoning benchmark associative recall. On EnWik8, TCNCA outperforms MEGA, reaching a lower loss with $1.37\times$/$1.24\times$ faster forward/backward pass during training. The dilated convolutions used in TCNCA are consistently and significantly faster operations than the FFT-based parallelized recurrence in GPUs, making them a scalable candidate for handling very large sequence lengths: they are up to $7.07\times$/$2.86\times$ faster in the forward/backward pass for sequences up to 131k. Further on LRA, TCNCA achieves, on average, $1.28\times$ speed-up during inference with similar accuracy to what MEGA achieves. On associative recall, we find that even a simplified version of TCNCA, without excessive multiplicative and additive interactions, remains superior or competitive to MEGA on a range of sequence lengths and vocabulary sizes.

Factorizers for Distributed Sparse Block Codes

Distributed sparse block codes (SBCs) exhibit compact representations for encoding and manipulating symbolic data structures using fixed-with vectors. One major challenge however is to disentangle, or factorize, such data structures into their constituent elements without having to search through all possible combinations. This factorization becomes more challenging when queried by noisy SBCs wherein symbol representations are relaxed due to perceptual uncertainty and approximations made when modern neural networks are used to generate the query vectors. To address these challenges, we first propose a fast and highly accurate method for factorizing a more flexible and hence generalized form of SBCs, dubbed GSBCs. Our iterative factorizer introduces a threshold-based nonlinear activation, a conditional random sampling, and an $\ell_\infty$-based similarity metric. Its random sampling mechanism in combination with the search in superposition allows to analytically determine the expected number of decoding iterations, which matches the empirical observations up to the GSBC's bundling capacity. Secondly, the proposed factorizer maintains its high accuracy when queried by noisy product vectors generated using deep convolutional neural networks (CNNs). This facilitates its application in replacing the large fully connected layer (FCL) in CNNs, whereby C trainable class vectors, or attribute combinations, can be implicitly represented by our factorizer having F-factor codebooks, each with $\sqrt[\leftroot{-2}\uproot{2}F]{C}$ fixed codevectors. We provide a methodology to flexibly integrate our factorizer in the classification layer of CNNs with a novel loss function. We demonstrate the feasibility of our method on four deep CNN architectures over CIFAR-100, ImageNet-1K, and RAVEN datasets. In all use cases, the number of parameters and operations are significantly reduced compared to the FCL.