Transformers can be hard to interpret due to complex optima.
problem Interpretability methods fail to capture the full complexity of Transformer models.
method Theoretical analysis and synthetic experiments on Dyck language learning.
result Attention patterns can be nearly randomized while preserving network functionality.
We prove that the Cayley graph and the coset geometry of the von Dyck group D(a,b,c) are linked by a vertex-to-edge duality.
We prove an optimal systolic inequality for nonpositively curved Dyck's surfaces. The extremal surface is flat with eight conical singularities, six of angle theta and two of angle 9pi - theta, for a suitable theta with cos(theta) in Q(sqrt{19}). Relying on some delicate capacity estimates, we also show that the extrem…
This paper compares Transformers and RNNs in various tasks, showing size differences.
problem Comparing representational capabilities of Transformers and RNNs across tasks.
method Analysis of differences in tasks like index lookup, nearest neighbor, and string equality.
result Size differences in Transformers and RNNs for various tasks.
We obtain a complete time expansion of the pull-back operator generated by a real analytic flow of real analytic automorphisms acting on analytic tensor sections of a manifold. Our expansion is given in terms of multiple Lie derivatives. Motivated by this expansion, we provide a rather simple and explicit estimate for …
Let K' be a hyperbolic knot in S^3 and suppose that some Dehn surgery on K' with distance at least 3 from the meridian yields a 3-manifold M of Heegaard genus 2. We show that if M does not contain an embedded Dyck's surface (the closed non-orientable surface of Euler characteristic -1), then the knot dual to the surger…
New basis and Schur-Weyl duality for loop Hecke algebra defined.
problem Define a new basis for the loop Hecke algebra.
method Use higher linear rewriting theory and combinatorics of Dyck paths.
result Yields a conjecture of Damiani-Martin-Rowell and provides a representation theoretic interpretation.
Given a hyperelliptic Klein surface, we construct companion Klein bottles, extending our technique of companion tori already exploited by the authors in the genus 2 case. Bavard's short loops on such companion surfaces are studied in relation to the original surface so to improve a systolic inequality of Gromov's. A ba…
The paper studies combinatorics of injective words in the context of Temperley-Lieb algebras.
problem Combinatorial properties of injective words in the context of Temperley-Lieb algebras.
method Investigation of a chain complex of modules over the Temperley-Lieb algebra, focusing on Euler characteristic, homology modules, and Jacobsthal numbers.
result The Euler characteristic of the complex is the n-th Fine number, and the top-dimensional homology module is decomposed in terms of standard Young tableaux.
PPT optimizes transformer behavior by steering its latent posterior using prior samples.
problem Eliciting desired behavior from transformers without backpropagation.
method Posterior Prefix Tuning (PPT) uses predictive Monte Carlo (PMC) samples and importance sampling to optimize the latent posterior.
result PPT optimizes transformer behavior without backpropagation, achieving high utility across different utility functions.
New ASR system handles multiple languages without needing language-specific encoding.
problem Joint training of data-rich and data-scarce languages in a single model.
method Transforms all languages to a single writing system through transliteration, separating modeling and rendering.
result Language-agnostic multilingual ASR system reduces WER up to 10% over language-dependent models.
Study quantifies gender bias in language models across 7 languages.
problem Measuring gender bias in language models across multiple languages.
method Curated dataset of politicians, multilingual language models, probing language models.
result Larger language models do not show significant gender bias compared to smaller ones.
Paper reviews neurolinguistics and language technologies, emphasizing mutual enrichment.
problem Understanding brain activity during language processing.
method Brain imaging studies and natural language representations.
result Development of brain-aware natural language representations.
New technique reduces language biases in large language models.
problem Gap between language and thought in large language models.
method Proposes Language-of-Thoughts (LoT) technique to adjust order and tokens.
result Significantly reduces language biases and improves reasoning tasks.
When a bilingual student learns to solve word problems in math, we expect the student to be able to solve these problem in both languages the student is fluent in,even if the math lessons were only taught in one language. However, current representations in machine learning are language dependent. In this work, we pres…
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.
System identifies language of transliterated text.
problem Users struggle to understand non-native language transliterated text.
method Feature extraction of phonetic syllables using LSTM network.
result System accurately identifies language of transliterated text.
LLMs translate natural language trading intents into correct option strategies using a domain-specific language.
problem Challenges in translating natural language trading intents into correct option strategies due to the complexity of option chain data.
method Introduce Option Query Language (OQL) as a domain-specific intermediate representation to abstract option markets into high-level primitives under grammatical rules. Use LLMs as semantic parsers and validate queries by an engine.
result Significantly improves execution accuracy and logical consistency over direct baselines.
New findings show language models can't simultaneously avoid hallucinations and capture all language richness.
problem Achieving both valid output and full language richness in language generation.
method Investigates language generation within a statistical setting, focusing on consistency and breadth.
result For most collections of candidate languages, a language model cannot simultaneously avoid hallucinations and capture all language richness.
This paper proposes a new method to connect language and physical actions in reinforcement learning.
problem Connecting linguistic representations to the physical world in embodied agents.
method Language-conditioned goal generators to decouple sensorimotor learning from language acquisition.
result Agents can demonstrate a diversity of behaviors for any given instruction.
Julia accelerates machine learning in various fields with balance of efficiency and simplicity.
problem Efficiency and simplicity in machine learning algorithms.
method Developed and applied Julia language in machine learning.
result Julia balances efficiency and simplicity for machine learning.
Neural language modeling (LM) has led to significant improvements in several applications, including Automatic Speech Recognition. However, they typically require large amounts of training data, which is not available for many domains and languages. In this study, we propose a multilingual neural language model archite…
The paper applies math and physics to language models, introducing entropy and geometric concepts.
problem Understanding and improving language models to approximate intelligent language.
method Formal definitions, functional analysis, topology, thermodynamics, and set theory.
result Entropy function reveals key obstacles for LLMs and offers insights into language models.
Multilingual end-to-end (E2E) models have shown great promise in expansion of automatic speech recognition (ASR) coverage of the world's languages. They have shown improvement over monolingual systems, and have simplified training and serving by eliminating language-specific acoustic, pronunciation, and language models…
Benchmark tests spoken language models for infant language learning.
problem Understanding how infants learn language from speech.
method Developed a language-acquisition-friendly benchmark.
result Benchmarking shows models' strengths and weaknesses.
Introduces challenges and techniques for creating machine translation for indigenous languages.
problem Limited data for machine translation of indigenous languages.
method Introduction to challenges, concepts, and techniques for creating MT systems.
result Discussion of recent advances and open questions in NLP for these languages.
Optimus pre-trains sentences in a latent space for various NLP tasks.
problem Training large-scale language models for diverse NLP tasks.
method Pre-trained Variational Autoencoder (VAE) on large text corpus, fine-tuned for various tasks.
result Optimus achieves state-of-the-art on VAE language modeling benchmarks.
Work shows hallucination detection by LLMs is impossible without expert feedback.
problem Detecting hallucinations in LLMs is theoretically impossible without expert-labeled feedback.
method Investigated hallucination detection using a theoretical framework inspired by language identification.
result Automated hallucination detection is impossible for most language collections without expert-labeled feedback.
Language models learn from training data and can leak private information.
problem Language models lack context understanding and can expose private data.
method Discussing the limitations of current privacy protection methods for language models.
result Existing privacy protection methods are insufficient for language models.
Large language models predict human sensory judgments across multiple modalities.
problem Determining the extent of perceptual information in language.
method State-of-the-art large language models were used to predict sensory judgments across six psychophysical datasets.
result Large language models can predict human sensory judgments across multiple modalities with significant correlation to human data.
New model shows natural language exhibits phase transition similar to physics.
problem Understanding critical properties in natural language models.
method Created a context-sensitive random language model.
result Demonstrated a Berezinskii--Kosterlitz--Thouless phase transition.
We study unsupervised multilingual alignment, the problem of finding word-to-word translations between multiple languages without using any parallel data. One popular strategy is to reduce multilingual alignment to the much simplified bilingual setting, by picking one of the input languages as the pivot language that w…
This is a lecture note for the course DS-GA 3001 <Natural Language Understanding with Distributed Representation> at the Center for Data Science , New York University in Fall, 2015. As the name of the course suggests, this lecture note introduces readers to a neural network based approach to natural language understand…
Spoken language identification (LID) technologies have improved in recent years from discriminating largely distinct languages to discriminating highly similar languages or even dialects of the same language. One aspect that has been mostly neglected, however, is discrimination of languages for multilingual speakers, d…
We present a TTS neural network that is able to produce speech in multiple languages. The proposed network is able to transfer a voice, which was presented as a sample in a source language, into one of several target languages. Training is done without using matching or parallel data, i.e., without samples of the same …
New approach decouples skill learning and language grounding for autonomous agents.
problem Autonomous acquisition of skills without external instructions and feedback.
method Language-Goal-Behavior (LGB) architecture with semantic representation.
result Decouples skill learning and language grounding, enabling diversity and strategy switching.
lamBERT learns language and actions using multimodal BERT.
problem Learning language and actions in complex environments.
method Extending BERT to multimodal representation and integrating with reinforcement learning.
result lamBERT model achieved higher rewards in multitask and transfer settings.
LORL learns object-centric representations from vision and language.
problem Learning disentangled, object-centric scene representations from vision and language.
method LORL integrates unsupervised object discovery and segmentation with language input to learn object-centric concepts.
result LORL improves unsupervised object discovery methods and aids downstream tasks.
Recent work has studied the emergence of language among deep reinforcement learning agents that must collaborate to solve a task. Of particular interest are the factors that cause language to be compositional -- i.e., express meaning by combining words which themselves have meaning. Evolutionary linguists have found th…
Transformers improve Finnish language modeling, achieving lower perplexity scores.
problem Improving language modeling for Finnish using deep learning models.
method Used BERT and Transformer-XL models in a sub-word setting, compared to LSTM.
result Transformer-XL outperforms LSTM, achieving a 27% better perplexity score.
VALC provides concept-level interpretations of FLMs, overcoming word-level limitations.
problem Lack of higher-level structure interpretation in FLMs' attention weights.
method Formal definition of conceptual interpretation, variational Bayesian framework (VALC).
result VALC finds optimal language concepts for FLM predictions, providing concept-level interpretations.
The task of translating between programming languages differs from the challenge of translating natural languages in that programming languages are designed with a far more rigid set of structural and grammatical rules. Previous work has used a tree-to-tree encoder/decoder model to take advantage of the inherent tree s…
Study finds phase transition in context-sensitive language model with short-range interactions.
problem Understanding phase transitions in language models with short-range interactions.
method Constructed a random language model with short-range interactions and investigated its statistical properties.
result Phase transition occurs in context-sensitive language models with constant context length.
We propose multi-way, multilingual neural machine translation. The proposed approach enables a single neural translation model to translate between multiple languages, with a number of parameters that grows only linearly with the number of languages. This is made possible by having a single attention mechanism that is …
We present a novel language adaptable spell checking system which detects spelling errors and suggests context sensitive corrections in real-time. We show that our system can be extended to new languages with minimal language-specific processing. Available literature majorly discusses spell checkers for English but the…
Improved robot navigation using multi-head attention for natural language instructions.
problem Improving robot navigation in unfamiliar environments.
method Proposes a multi-head attention mechanism blending layer in a neural network model.
result Significant performance gains in translating instructions for unseen environments.
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…
Language helps RL agents learn complex relational and causal structures.
problem Learning relational and causal structure in complex environments.
method Training RL agents to predict language descriptions and explanations.
result Language aids agents in learning challenging relational and causal tasks.