Develops a statistical framework to measure uncertainty in model rankings based on human preferences.
problem Uncertainty in model rankings based on human preferences due to mismatch between human and model preferences.
method Statistical framework using pairwise comparisons by humans and models to provide rank-sets for each model.
result Rank-sets constructed using only pairwise comparisons by strong models often do not cover the true ranking of human preferences.
The paper fine-tunes language models using human preferences for natural language tasks.
problem Applying reinforcement learning to natural language tasks with human-defined rewards.
method Fine-tuning language models using reward learning techniques, leveraging generative pretraining.
result The approach achieves good results in natural language tasks like text continuation and summarization.
An NMT system for Indic languages outperforms Google Translate.
problem Challenges in translating Indic languages efficiently.
method Encoder-decoder with attention mechanism for neural machine translation.
result Outperforms Google Translate with a 6 BLEU score margin on English-Gujarati translation.
Automates debiasing for large language model evaluations through Fisher random walk.
problem Rigorous and scalable evaluation of large language models.
method Semiparametric efficient estimator using Fisher random walk for weighted residual balancing.
result Efficient estimation of contextual preference scores for large language models.
Text documents are structured on multiple levels of detail: individual words are related by syntax, but larger units of text are related by discourse structure. Existing language models generally fail to account for discourse structure, but it is crucial if we are to have language models that reward coherence and gener…
Statistical framework improves LLM chatbot ranking.
problem Improving evaluation of LLM-based chatbots through pairwise comparisons.
method Factored tie model, covariance modeling, and parameter constraints.
result Substantial improvements in modeling pairwise comparison data.
Automates model comparison in probabilistic programming.
problem Manual derivations for model comparison are error-prone and time-consuming.
method Message passing on a Forney-style factor graph with a custom mixture node.
result Automates Bayesian model averaging, selection, and combination.
A new sliced IGW distance for Gromov-Wasserstein alignment.
problem Scalability issues in Gromov-Wasserstein alignment for high-dimensional problems.
method Proposed a sliced IGW distance with rotational invariance.
result Natural rotational invariance of the sliced IGW distance.
Study uses multiple online media to predict crude oil prices.
problem Forecasting crude oil prices using online media.
method Semantic analysis and ARIMAX models on Twitter, Google Trends, Wikipedia, and GDELT.
result Combined analysis from four platforms improves price prediction.
A new framework evaluates LLMs by considering judge reliability.
problem Evaluating LLMs without ground truth labels can lead to biased results.
method Introduces judge-specific discrimination parameters and estimates model quality and judge reliability.
result Improves agreement with human preferences and produces calibrated uncertainty quantification.
Generative Adversarial Networks (GANs) have been promising in the field of image generation, however, they have been hard to train for language generation. GANs were originally designed to output differentiable values, so discrete language generation is challenging for them which causes high levels of instability in tr…
Study uses three sources to evaluate language models fairly.
problem Bias in offline model evaluation due to confounded model choice.
method Combines observational logs, randomized experiments, and simulators.
result Randomized experiment and simulator together recover causal model values.
New principles needed for scaling large language models, challenging traditional regularization methods.
problem The shift from generalization to scaling in machine learning requires new guiding principles.
method Examining the effectiveness of traditional regularization methods in the scaling-centric era.
result Traditional principles of regularization may not generalize to larger scales, highlighting new phenomena like scaling law crossover.
CLARITY compares dissimilar datasets, identifying structural and relationship inconsistencies.
problem Integrating qualitatively different datasets from various disciplines.
method Non-parametric approach decomposing similarities into structural and relationship components.
result Identifies and interprets inconsistencies between datasets.
SCROLLS benchmarks long text NLP tasks, improving existing models.
problem Short NLP benchmarks ignore long texts; SCROLLS addresses this.
method Handpicked long-text datasets for summarization, QA, and inference tasks.
result Improvement potential on SCROLLS tasks, as indicated by initial baselines.
LIDS assesses LLM summaries with interpretable key words.
problem Challenges in evaluating the quality of LLM summaries.
method BERT-SVD-based direction metric and SOFARI for key word extraction.
result LIDS provides interpretable key words for layered themes.
Deep CapsNet improves sign language recognition from wearable IMUs.
problem Continuous recognition of sign language from wearable devices.
method Custom CapsNet architecture using deep capsule networks and game theory.
result Improved accuracy of 94% and 92.50% for 3 and 5 routings respectively, compared to 87.99% for CNN.
New handwritten digits dataset for Kannada script.
problem Lack of datasets for Kannada numeral digits.
method Developed Kannada-MNIST and Dig-MNIST datasets.
result Initial CNN accuracy is lower than MNIST, indicating a challenge in generalization.
A new model tackles language generation issues by using discrete variational attention.
problem Information under-representation and posterior collapse in variational autoencoders.
method Proposes a discrete variational attention model with categorical distribution over attention mechanism.
result Enhances latent space for language generation and avoids posterior collapse.
Abstract cone operators prove scalar curvature comparisons on singular manifolds.
problem Proving scalar curvature inequalities on manifolds with cone singularities.
method Using index theory for twisted Dirac operators on Lipschitz bundles.
result Lipschitz rigidity for scalar curvature on odd-dimensional manifolds.
Novel CNN array for sign language recognition using wearable IMUs.
problem Efficiently recognizing sign language from wearable IMU signals.
method Two-dimensional Convolutional Neural Network array architecture for Indian sign language recognition.
result Peak classification accuracies of 94.20% for general sentences and 95.00% for interrogative sentences achieved.
Survey of deep learning for Hindi text classification.
problem Limited research on morphologically rich, low-resource Hindi text classification.
method Comparison of CNN, LSTM, Transformer, BERT, and LASER for Hindi text classification.
result Multilingual pre-trained sentence embeddings outperform traditional architectures for Hindi text classification.
In this paper, we explore and detail our experiments in a high-dimensionality, multi-class image classification problem often found in the automatic recognition of Sign Languages. Here, our efforts are directed towards comparing the characteristics, advantages and drawbacks of creating and training Support Vector Machi…
Many Machine Reading and Natural Language Understanding tasks require reading supporting text in order to answer questions. For example, in Question Answering, the supporting text can be newswire or Wikipedia articles; in Natural Language Inference, premises can be seen as the supporting text and hypotheses as question…
ALMANACS benchmarks explainability methods on simulatability.
problem Evaluating the effectiveness of explainability methods for language models.
method ALMANACS is a simulatability benchmark that evaluates explainability methods on twelve safety-relevant topics.
result No explainability method outperforms the explanation-free control across all topics.
This work compares and evaluates various sampling methods for neural language models.
problem Lack of systematic comparison and myths about sampling methods.
method Monte Carlo sampling, importance sampling, compensated partial summation, noise contrastive estimation.
result All sampling methods can perform equally well if posterior probabilities are corrected.
Develops a cross-lingual hate speech detection model using pre-trained Transformers.
problem Detecting hate speech in low-resource languages.
method Utilizes frozen Transformer language models and AXEL attention-based classification block for zero-shot and few-shot learning.
result Demonstrates highly competitive results on English and Spanish subsets of the HatEval challenge.
A new transformer model corrects diacritics and typos in multiple languages.
problem Restoring diacritics and correcting typos in online communications.
method Employing a universal ByT5 transformer model trained on 12 languages, including Lithuanian.
result Achieves > 98% accuracy in diacritics restoration and typos correction.
Neural machine translation used to convert CUDA to OpenCL.
problem Translating CUDA to OpenCL programs.
method Training input set generation, pre/post processing, case study.
result Improved accuracy in translating CUDA to OpenCL.
Unified definition of hallucinations in language models.
problem Persistent hallucinations despite mitigation efforts.
method Unified definition of hallucination as inaccurate world modeling.
result Unified framework distinguishes hallucinations from other errors.
This research explores discrete diffusion models for natural language generation.
problem Challenges in applying diffusion models to discrete data, especially natural language.
method Investigates Discrete Denoising Diffusion Probabilistic Model (D3PM) and compares it with autoregressive models.
result Discrete diffusion models achieve better processing speed than autoregressive models.
New models ensure monotonicity in preference learning, improving accuracy especially with limited data.
problem Failure of widely used preference learning models to maintain monotonicity.
method Proposed Linear Generalized Bradley-Terry models with Diffusion Priors.
result New models improve accuracy, especially with limited data.
Hybrid and end-to-end models compare in syllable recognition.
problem Comparing hybrid and end-to-end models for syllable recognition.
method Traditional hybrid system (kaldi) vs. end-to-end (TensorFlow) models.
result Hybrid models with explicit syllable knowledge outperform end-to-end models.
The paper creates a language evolution tree using word vectors from historical novels.
problem Exploring the evolution of language through historical texts.
method Constructed word vectors from novels, combined them, and used hierarchical clustering.
result Discovered a specific language evolution tree that reflects the year of the corpus.
Language helps RL agents learn complex tasks by composing sub-skills.
problem Solving complex, temporally-extended tasks in RL.
method Using language as an abstraction to learn and reuse sub-skills.
result Agents can learn and solve diverse, temporally-extended tasks.
Bayesian EnKF improves sentence comprehension uncertainty modeling.
problem Uncertainty in human language comprehension, especially with ambiguous inputs.
method Bayesian framework using ensemble Kalman filter (EnKF) for uncertainty quantification.
result Enhanced model's ability to approximate human cognitive processing with linguistic ambiguities.
The paper argues for better experimental reporting to avoid misinterpretation of model performance.
problem Misinterpretation of model performance due to insufficient experimental reporting.
method Presenting expected validation performance as a function of computation budget.
result Authors can reach different conclusions based on the amount of computation used.
New study finds best language model architecture and pretraining objective for zero-shot tasks.
problem Evaluating which language model architectures and pretraining objectives best enable zero-shot generalization.
method Compared three model architectures and two pretraining objectives across 170 billion tokens, with and without finetuning.
result Causal decoder-only models trained on autoregressive language modeling exhibit strongest zero-shot generalization.
Improved unsupervised probing for ranking tasks using Contrast-Consistent Ranking.
problem Improving self-consistency in language model rankings.
method Adapting Contrast-Consistent Search (CCS) to Contrast-Consistent Ranking (CCR) for ranking tasks.
result CCR probing outperforms prompting techniques across different models and datasets.
This work compares lattice-free and lattice-based training criteria for LVCSR.
problem Improving acoustic model performance in speech recognition.
method Direct comparison of lattice-free and lattice-based sequence discriminative training criteria using GPU.
result Lattice-free MMI performance is comparable to lattice-based criteria, while lattice-based sMBR remains superior.
Confidence bands for tuning curves improve hyperparameter comparison in NLP.
problem Ambiguity in comparing hyperparameter tuning methods.
method Constructs exact, simultaneous, and distribution-free confidence bands for tuning curves.
result Confidence bands provide a robust basis for comparing methods rigorously.
Study builds a Japanese financial-specific LLM through continual pre-training.
problem Lack of domain-specific Japanese financial LLMs.
method Continual pre-training on Japanese financial-focused datasets using a base Japanese LLM.
result Tuned model outperforms original model on Japanese financial benchmarks.
Numeracy is the ability to understand and work with numbers. It is a necessary skill for composing and understanding documents in clinical, scientific, and other technical domains. In this paper, we explore different strategies for modelling numerals with language models, such as memorisation and digit-by-digit composi…
DBPA assesses LLM perturbations using frequentist hypothesis testing.
problem Quantifying input perturbation impacts on LLM outputs.
method DBPA reformulates perturbation analysis as frequentist hypothesis testing, using Monte Carlo sampling for empirical null and alternative distributions.
result DBPA provides interpretable p-values and scalar effect sizes for LLM perturbations.
New algorithms improve privacy and utility of large language models.
problem Privacy-preserving fine-tuning of large language models.
method Meta-framework for differentially private fine-tuning, inspired by recent success in fine-tuning.
result Private fine-tuned models achieve utility close to non-private models, with improved privacy and efficiency.
PDO optimizes LLM prompts without labels, improving performance.
problem Optimizing prompts for LLMs without access to labeled data.
method Pairwise preference feedback, dueling bandits, Thompson Sampling, mutation.
result PDO identifies stronger prompts than label-free methods.
We present Listen, Attend and Spell (LAS), a neural network that learns to transcribe speech utterances to characters. Unlike traditional DNN-HMM models, this model learns all the components of a speech recognizer jointly. Our system has two components: a listener and a speller. The listener is a pyramidal recurrent ne…
Method tackles uncertainty in reward models for LLMs from heterogeneous human feedback.
problem Uncertainty in reward models for LLMs from heterogeneous human feedback.
method Heterogeneous preference framework and alternating gradient descent algorithm.
result Established theoretical guarantees for estimator convergence and asymptotic distribution.