MinMax RNCs are recurrent neural models using min-max recurrence that achieve full regular-language expressivity, logarithmic parallel evaluation, uniformly bounded states, and constant state gradients independent of time distance.
super hub Mixed citations
Layer Normalization
Mixed citation behavior. Most common role is background (60%).
166
Pith papers citing it
Background
60% of classified citations
abstract
Training state-of-the-art, deep neural networks is computationally expensive. One way to reduce the training time is to normalize the activities of the neurons. A recently introduced technique called batch normalization uses the distribution of the summed input to a neuron over a mini-batch of training cases to compute a mean and variance which are then used to normalize the summed input to that neuron on each training case. This significantly reduces the training time in feed-forward neural networks. However, the effect of batch normalization is dependent on the mini-batch size and it is not obvious how to apply it to recurrent neural networks. In this paper, we transpose batch normalization into layer normalization by computing the mean and variance used for normalization from all of the summed inputs to the neurons in a layer on a single training case. Like batch normalization, we also give each neuron its own adaptive bias and gain which are applied after the normalization but before the non-linearity. Unlike batch normalization, layer normalization performs exactly the same computation at training and test times. It is also straightforward to apply to recurrent neural networks by computing the normalization statistics separately at each time step. Layer normalization is very effective at stabilizing the hidden state dynamics in recurrent networks. Empirically, we show that layer normalization can substantially reduce the training time compared with previously published techniques.
hub tools
citation-role summary
background 3
method 2
citation-polarity summary
claims ledger
- abstract Training state-of-the-art, deep neural networks is computationally expensive. One way to reduce the training time is to normalize the activities of the neurons. A recently introduced technique called batch normalization uses the distribution of the summed input to a neuron over a mini-batch of training cases to compute a mean and variance which are then used to normalize the summed input to that neuron on each training case. This significantly reduces the training time in feed-forward neural networks. However, the effect of batch normalization is dependent on the mini-batch size and it is not
authors
co-cited works
representative citing papers
Local attention strictly enlarges the class of regular languages recognizable by fixed-precision transformers by adding a second past operator in linear temporal logic, with global and local attention being expressively complementary.
Mamba is a linear-time sequence model using input-dependent selective SSMs that achieves SOTA results across modalities and matches twice-larger Transformers on language modeling with 5x higher inference throughput.
The Pile is a newly constructed 825 GiB dataset from 22 diverse sources that enables language models to achieve better performance on academic, professional, and cross-domain tasks than models trained on Common Crawl variants.
Reformer matches standard Transformer accuracy on long sequences while using far less memory and running faster via LSH attention and reversible residual layers.
A W-Net deep learning model detects asteroids in TESS data independently of trajectory by rotating training image cubes and using adaptive normalization for data scaling.
QAP-Router models qubit routing as dynamic QAP and applies RL with a solution-aware Transformer to cut CNOT counts by 12-30% versus industry compilers on real circuit benchmarks.
Temporarily reducing the learning rate on upper-layer query and key projections during early GPT pretraining prevents premature attention specialization and improves model performance.
OpenSGA fuses vision-language, textual, and geometric features via a distance-gated attention encoder and minimum-cost-flow allocator to outperform prior methods on both frame-to-scan and subscan-to-subscan 3D scene graph alignment, backed by a new 700k-sample ScanNet-SG dataset.
MetaSG-SAEA is a bi-level meta-BBO framework that uses a meta-policy for search guidance via the MM-CCI constraint abstraction and diffusion-based population initialization to outperform baselines on expensive constrained multi-objective optimization problems.
Aligned LLMs exhibit Refusal-Escape Directions (RED) that enable refusal-to-answer transitions via input perturbations; these directions decompose exactly into operator-level sources, creating an inherent safety-utility trade-off when trying to eliminate them.
Delta-Adapter extracts a semantic delta from a single image pair via a pre-trained vision encoder and injects it through a Perceiver adapter to enable scalable single-pair supervised editing.
A new neural quantum state ansatz for bosons in the grand canonical ensemble achieves competitive variational energies in 1D and 2D systems and provides access to one-body reduced density matrices.
QuadNorm uses quadrature-based moments instead of uniform averaging in normalization layers, achieving O(h²) consistency across resolutions and better cross-resolution transfer in neural operators.
GPROF-IR is a CNN-based retrieval that uses temporal context in geostationary IR observations to produce precipitation estimates with lower error than prior IR methods and climatological consistency with PMW retrievals for integration into IMERG V08.
A Max-Cut-specific graph neural network predicts primal- and dual-feasible SDP solutions in linearithmic time, cutting bounding costs in exact branch-and-bound by up to 10.6 times versus a commercial SDP solver while training without any solved SDP labels.
Every fixed finite feedforward neural network definable in an o-minimal structure has finite sample complexity in the agnostic PAC setting.
SOPE uses an actor-aligned OPE signal on a held-out validation split to dynamically stop offline stabilization phases in online RL, improving performance up to 45.6% and cutting TFLOPs up to 22x on 25 Minari tasks.
In linear recurrent models, infinite-width signal propagation remains accurate only for depths t much smaller than sqrt(width n), with a critical regime at t ~ c sqrt(n) where finite-width effects emerge and dominate for larger t.
PHALAR achieves up to 70% relative accuracy gain in stem retrieval with under half the parameters and 7x faster training by using phasor-based equivariant representations, setting new SOTA on multiple datasets.
A fully differentiable TensorFlow gyrokinetic code allows approximate gradients of nonlinear turbulence quantities to be used for outer-loop tasks such as profile prediction despite stochasticity.
EstGraph benchmark evaluates LLMs on estimating properties of very large graphs from random-walk samples that fit in context limits.
ResRL decouples shared semantics between positive and negative responses in LLM reinforcement learning via SVD-based projection residuals, outperforming baselines including NSR by up to 9.4% on math reasoning benchmarks.
DEFault++ delivers automated hierarchical fault detection, categorization into 12 transformer-specific types, and root-cause diagnosis among 45 mechanisms on a new benchmark of 3,739 mutated instances, with AUROC >0.96 and Macro-F1 0.85, plus improved developer repair accuracy in a user study.
citing papers explorer
-
Characterizing the Expressivity of Local Attention in Transformers
Local attention strictly enlarges the class of regular languages recognizable by fixed-precision transformers by adding a second past operator in linear temporal logic, with global and local attention being expressively complementary.
-
The Pile: An 800GB Dataset of Diverse Text for Language Modeling
The Pile is a newly constructed 825 GiB dataset from 22 diverse sources that enables language models to achieve better performance on academic, professional, and cross-domain tasks than models trained on Common Crawl variants.
-
Learning Less Is More: Premature Upper-Layer Attention Specialization Hurts Language Model Pretraining
Temporarily reducing the learning rate on upper-layer query and key projections during early GPT pretraining prevents premature attention specialization and improves model performance.
-
OPT: Open Pre-trained Transformer Language Models
OPT releases open decoder-only transformers up to 175B parameters that match GPT-3 performance at one-seventh the carbon cost, along with code and training logs.
-
DeBERTa: Decoding-enhanced BERT with Disentangled Attention
DeBERTa improves BERT-style models by separating content and relative position in attention and adding absolute positions to the decoder, yielding consistent gains on NLU and NLG tasks and the first single-model superhuman score on SuperGLUE.
-
Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism
Intra-layer model parallelism in PyTorch enables training of 8.3B-parameter transformers, achieving SOTA perplexity of 10.8 on WikiText103 and 66.5% accuracy on LAMBADA.
-
Towards Understanding Continual Factual Knowledge Acquisition of Language Models: From Theory to Algorithm
Theoretical analysis of continual factual knowledge acquisition shows data replay stabilizes pretrained knowledge by shifting convergence dynamics while regularization only slows forgetting, leading to the STOC method for attention-based replay selection.
-
AGoQ: Activation and Gradient Quantization for Memory-Efficient Distributed Training of LLMs
AGoQ delivers up to 52% lower memory use and 1.34x faster training for 8B-32B LLaMA models by using near-4-bit adaptive activations and 8-bit gradients while preserving pretraining convergence and downstream accuracy.
-
Hidden States Know Where Reasoning Diverges: Credit Assignment via Span-Level Wasserstein Distance
Span-level Wasserstein distances between hidden-state distributions of correct and incorrect rollouts provide a self-supervised signal to reweight advantages in GRPO, improving fine-grained credit assignment on math and code tasks.
-
Retentive Network: A Successor to Transformer for Large Language Models
RetNet is a new sequence modeling architecture that delivers parallel training, constant-time inference, and competitive language modeling performance as a potential replacement for Transformers.
-
AdaLoRA: Adaptive Budget Allocation for Parameter-Efficient Fine-Tuning
AdaLoRA uses SVD-based pruning to allocate the parameter budget for low-rank fine-tuning updates according to per-matrix importance scores, yielding better performance than uniform allocation especially under tight budgets.
-
ST-MoE: Designing Stable and Transferable Sparse Expert Models
ST-MoE introduces stability techniques for sparse expert models, allowing a 269B-parameter model to achieve state-of-the-art transfer learning results across reasoning, summarization, and QA tasks at the compute cost of a 32B dense model.
-
Universal Transformers
Universal Transformers combine Transformer parallelism with recurrent updates and dynamic halting to achieve Turing-completeness under assumptions and outperform standard Transformers on algorithmic and language tasks.
-
Attention Is All You Need
Pith review generated a malformed one-line summary.
-
A Survey on Large Language Models for Code Generation
A systematic literature review that organizes recent work on LLMs for code generation into a taxonomy covering data curation, model advances, evaluations, ethics, environmental impact, and applications, with benchmark comparisons.
-
A Survey of Large Language Models
This survey reviews the background, key techniques, and evaluation methods for large language models, emphasizing emergent abilities that appear at large scales.