Improved Transformer language models using dynamic evaluation.
problem Language model perplexity and accuracy improvements.
method Combining Transformers with dynamic evaluation techniques.
result Significant improvement in language model performance (e.g., 0.99 to 0.94 bits/char).
FLUID-LLM uses LLMs to predict fluid dynamics with improved accuracy.
problem Leveraging LLMs for CFD due to their pattern recognition abilities but struggles with fluid dynamics complexities.
method Combines pre-trained LLMs with spatiotemporal-aware encoding to predict unsteady fluid dynamics.
result Significant performance improvements in CFD predictions across various datasets.
New method shows cultural transmission improves language compositionality.
problem How cultural transmission affects language emergence and compositionality.
method Introduced cultural evolutionary dynamics into language emergence by periodically replacing agents in a population.
result Languages emerge with better compositional generalization due to implicit cultural transmission.
Word embeddings are a powerful approach for unsupervised analysis of language. Recently, Rudolph et al. (2016) developed exponential family embeddings, which cast word embeddings in a probabilistic framework. Here, we develop dynamic embeddings, building on exponential family embeddings to capture how the meanings of w…
A framework uses free probability to analyze Transformer models.
problem Understanding the dynamics and complexity of Transformer-based language models.
method Formal operator-theoretic analysis using free probability theory.
result Entropy-based generalization bounds derived under freeness assumptions.
New method improves RL/IL agents' adaptability to unseen environments.
problem Current RL/IL techniques struggle with generalizing to unseen environments.
method Zero-shot compositional policy learning with multi-modal fusion and attention mechanism.
result Language grounding enhances generalization across varied environments.
In this paper, we propose Dynamic Self-Attention (DSA), a new self-attention mechanism for sentence embedding. We design DSA by modifying dynamic routing in capsule network (Sabouretal.,2017) for natural language processing. DSA attends to informative words with a dynamic weight vector. We achieve new state-of-the-art …
MM-DREX adapts LLM experts for financial trading via dynamic routing.
problem Challenges of non-stationary financial markets and static expert designs.
method MM-DREX uses a VLM-powered dynamic router to allocate expert weights and designs heterogeneous trading experts.
result Significantly outperforms 15 baselines across key metrics.
Language model benchmarks often misrepresent true understanding, revealing vulnerabilities in evaluation methods.
problem Language model benchmarks fail to accurately reflect true language understanding and adaptability.
method Systematic analysis of NLP evaluation frameworks, identifying vulnerabilities in static benchmarks, human evaluation protocols, and LLM-as-judge frameworks.
result Current evaluation methods are unreliable and need improvement to accurately assess LLM performance.
Dynamic meta-learning improves multi-agent communication with natural language.
problem Static population-based approaches struggle with adapting to human partners using natural language.
method Dynamic population-based meta-learning approach.
result Agents outperform all prior work in communicating with seen partners and humans.
New approach uses graphs for sign language recognition.
problem Challenges in recognizing sign language for deaf individuals.
method Spatial-Temporal Graph Convolutional Network.
result Improved sign language recognition using human skeletal movements.
Transformer model for probabilistic dynamical systems.
problem Modeling high-dimensional dynamical systems from noisy observations.
method Parallel between dynamical systems and language modeling; transformer-based model with geometrical properties; iterative training algorithm.
result Fine-grid approximation of conditional probabilities for high-dimensional systems.
A framework for multi-label sentiment analysis in 100 languages with dynamic weighting.
problem Cross-lingual sentiment analysis in multi-label settings with label imbalance.
method Dynamic weighting method, focal loss adaptation, optimal class-specific thresholds.
result State-of-the-art performance in 7 out of 9 metrics across 3 languages.
New approach uses compressible dynamics to train deep models efficiently.
problem Efficient training of deep overparameterized models with low-rank structures.
method Leveraging low-dimensional structures and compressible dynamics within model parameters.
result Improved training efficiency and reduced overfitting in language models.
Dynamic sparseness reduces neural network computation by selectively omitting parts of computations.
problem Reducing the computational and memory footprint of neural networks.
method Combining dynamic sparseness with block-wise matrix-vector multiplications to selectively omit parts of computations.
result The proposed method outperforms static sparseness and achieves similar perplexities at half the computational cost.
This work explains how linear representations in large language models arise from training objectives and gradient descent.
problem Understanding the origins of linear representations in large language models.
method A latent variable model to abstract and formalize concept dynamics, combined with analysis of the softmax cross-entropy objective and gradient descent.
result Linear representations emerge when learning from data matching the latent variable model, and this simple structure suffices to yield linear representations.
Paper introduces TrufLL for language model training without labeled data.
problem Training language models from scratch with reinforcement learning.
method Dynamic vocabulary truncation in RL for language generation.
result TrufLL successfully learns language generation policies from scratch.
LLMs simulate financial markets, revealing consistent trading strategies and market dynamics.
problem Testing financial theories with AI trading agents.
method Simulated stock market with LLMs using a persistent order book and varied strategies.
result LLMs can simulate different trading strategies and market dynamics.
New attack reveals memorization patterns in pre-trained LLMs.
problem Determining if a data point was part of a pre-trained LLM's training set.
method Adapts MIA statistical tests to LLM's perplexity dynamics of subsequences.
result Significantly outperforms prior approaches in membership inference attacks.
This study analyzes how one-layer transformers learn regular language recognition tasks.
problem Understanding how one-layer transformers solve regular language recognition tasks like even pairs and parity check.
method Theoretical analysis of training dynamics and gradient descent for a one-layer transformer.
result A one-layer transformer can solve even pairs directly but needs CoT for parity check. Training phases show rapid growth in attention layer followed by logarithmic growth in linear layer.
Model captures author language diffusion over time.
problem Lack of author identity and temporal context in language models.
method Temporal language model conditioning on author and temporal vectors.
result Beat temporal and non-temporal baselines, learns time-varying author representations.
L-Cool improves image and language translation by cooling low-density samples.
problem Improving translation performance on fringe samples in unsupervised domain translation.
method Performing Langevin dynamics to move low-density samples towards high-density areas.
result L-Cool enhances state-of-the-art methods in image and language translation tasks.
This study explains RL training dynamics in LLMs, focusing on token-level optimization and reasoning pattern reshaping.
problem Understanding the training dynamics of RL in LLMs to improve their reasoning capabilities.
method Empirical analysis and theoretical modeling of RL training process, focusing on reasoning patterns and token optimization.
result RL primarily optimizes a sparse subset of critical tokens, reshaping reasoning pattern distributions and affecting model performance.
FinDKG uses LLMs to detect financial trends from news articles.
problem Detecting global financial trends from unstructured text data.
method Fine-tuned LLMs for generating DKGs, KGTransformer for analysis.
result KGTransformer outperforms existing thematic ETFs in financial thematic investing.
Paper proposes a new RLHF framework for human preference learning.
problem Handling dependent online human preference outcomes with dynamic contexts.
method Two-stage algorithm with ε-greedy followed by exploitation; anti-concentration inequalities and matrix martingale concentration techniques. result Our method achieves optimal regret bound and asymptotic normality of estimators.
KG-A2C agent learns natural language IF games by reasoning and constraining action spaces.
problem Challenges of natural language understanding, partial observability, and combinatorially large action spaces in IF games.
method Builds a dynamic knowledge graph while exploring, constraining actions using templates.
result Outperforms current IF agents across various games with larger action spaces.
Word evolution refers to the changing meanings and associations of words throughout time, as a byproduct of human language evolution. By studying word evolution, we can infer social trends and language constructs over different periods of human history. However, traditional techniques such as word representation learni…
We introduce a scalable measure of curvature for analyzing training dynamics of large language models.
problem Analyzing the training dynamics of large language models due to high computational cost of measuring Hessian sharpness.
method We introduce critical sharpness and relative critical sharpness as computationally efficient measures capturing Hessian sharpness phenomena.
result We provide the first demonstration of sharpness phenomena at scale up to 7B parameters.
Tangent automates derivatives in Python, improving expressiveness and performance.
problem Efficiently calculating derivatives for complex models in Python.
method Source-code transformation for dynamically typed array programming.
result Demonstrates improved expressiveness and performance in automatic differentiation.
Proposes a new RNN for language generation capturing long-range dependencies.
problem Capturing long-range word dependencies and sentence order in text corpora.
method Recurrent Hierarchical Topic-Guided RNN with dynamic deep topic model.
result Outperforms larger-context RNN-based language models and learns interpretable topics.
Blend-ASC improves self-consistency efficiency by dynamically allocating samples, reducing costs.
problem Efficiently applying self-consistency to large datasets is computationally expensive.
method Blend-ASC dynamically allocates samples during inference, improving efficiency.
result Blend-ASC reduces sample usage by 6.8x on average compared to vanilla self-consistency.
Aioli unifies language model data mixing methods and improves performance.
problem Optimizing the mixture of training data groups for language models.
method Unified optimization framework for dynamically adjusting mixture proportions.
result Aioli outperforms existing methods by up to 12.012 test perplexity points.
Global anchor method detects language shifts and domain adaptation.
problem Detecting corpus-level language shifts and domain adaptation.
method Global anchor method for comparing word embeddings.
result Global anchor method is superior to alignment method in applicability and implementation.
Inverse depth scaling found in LLMs due to similar layers averaging error.
problem Understanding how depth affects loss in large language models.
method Analysis of LLMs and toy residual networks.
result Loss scales inversely proportional to depth in LLMs.
Improved ASR-free keyword spotting in under-resourced languages.
problem Dynamic time warping for keyword spotting in languages with limited resources.
method Multilingual bottleneck extractor and correspondence autoencoder integration.
result More than 11% absolute improvement in ROC AUC over MFCCs.
FMT model improves multimodal sequential learning across language, vision, and acoustic data.
problem Modeling spatio-temporal dynamics across multiple modalities.
method Factorized Multimodal Transformer (FMT) that models intramodal and intermodal dynamics in a factorized manner.
result FMT outperforms existing models on 3 datasets and 21 labels, setting new state of the art.
Improved crypto market forecasting using historical price reactions to tweets.
problem Challenges in inferring market impact from human sentiment labels.
method Market-derived labeling approach to assign tweet sentiment labels based on historical price trends. Fine-tuned language model with context-aware prompt-tuning.
result 89.6% accuracy on Bitcoin news events, outperforming traditional fusion models.
A new whole-sentence language model - neural trans-dimensional random field language model (neural TRF LM), where sentences are modeled as a collection of random fields, and the potential function is defined by a neural network, has been introduced and successfully trained by noise-contrastive estimation (NCE). In this…
New method adapts to user preferences dynamically, improving recommendation models.
problem Current recommendation models lack dynamic adaptation to changing user preferences.
method Preference Discerning with LLM-Enhanced Generative Retrieval
result Mender achieves state-of-the-art performance in adapting to evolving user preferences.
This paper addresses anisotropy in Transformer models, providing geometric insights and empirical support.
problem Anisotropy phenomenon in Transformer models, challenging their geometric interpretation.
method Derive geometric arguments and use concept-based mechanistic interpretability during training.
result Activation-derived directions capture large gradient energy and a larger share of gradient anisotropy than normal controls.
New framework reveals thermodynamic principles for LLM training.
problem Understanding the training dynamics of large language models.
method Introducing Neural Thermodynamic Laws (NTL) under river-valley loss landscape assumptions.
result Key thermodynamic quantities and principles naturally emerge in LLM training.
DOS improves language model generation by considering inter-token dependencies.
problem Lack of sequence-level information and inter-token dependencies in existing decoding strategies.
method Dependency-Oriented Sampler (DOS) that uses attention matrices to approximate inter-token dependencies.
result DOS consistently achieves superior performance on code generation and mathematical reasoning tasks.
Dynamic Vocabulary Pruning stabilizes LLM training by removing low-probability tokens.
problem Training Large Language Models (LLMs) with Reinforcement Learning (RL) causes numerical divergence between inference and training.
method Dynamic Vocabulary Pruning (DVP) constrains the RL objective to a safe vocabulary that excludes low-probability tokens.
result DVP stabilizes training by reducing systematic bias introduced by the extreme tail of the token distribution.
Gaussian processes are conditioned on various types of data.
problem Exact inference in Gaussian processes is limited to linear-Gaussian settings.
method Established an equivalence between GPs and linear diffusion models, allowing for approximate inference in non-linear settings.
result A general-purpose GP inference scheme that handles various conditioning statements, including non-linear physics and natural language.
ELS framework improves safety alignment by dynamically steering LLMs towards helpful responses.
problem Over-Refusal in Aligned Large Language Models
method Fine-tuning free framework using an Energy-Based Model (EBM) to dynamically steer LLMs during inference.
result Extensive experiments show a significant reduction in false refusals (from 57.3% to 82.6%) while maintaining safety performance.
AutoML-GPT uses GPT to automate AI model training.
problem Manual model selection and tuning requires significant human effort.
method Develops task-oriented prompts and utilizes LLMs for automated training.
result Achieves remarkable results in various AI tasks.
In-Place TTT enhances LLMs with dynamic parameter updates at inference time.
problem Static training limits LLMs from adapting to new information.
method In-Place TTT updates a subset of model parameters (fast weights) at inference time.
result In-Place TTT enables 4B-parameter models to outperform on tasks with up to 128k contexts.
We consider multilingual bottleneck features (BNFs) for nearly zero-resource keyword spotting. This forms part of a United Nations effort using keyword spotting to support humanitarian relief programmes in parts of Africa where languages are severely under-resourced. We use 1920 isolated keywords (40 types, 34 minutes)…