Linear properties are either universal or absent across language models.
problem Explaining the prevalence of linear properties in language models.
method Proved identifiability of distribution-equivalent next-token predictors and analyzed various notions of linearity.
result Linear properties either hold in all or none distribution-equivalent next-token predictors.
Reinforcement learning after next-token prediction aids in learning from diverse sequence lengths.
problem Learning from sequences of varying lengths and complexity.
method Introducing a framework to study reinforcement learning with autoregressive transformers, focusing on next-token prediction and mixture distributions of short and long sequences.
result Reinforcement learning after next-token prediction enables autoregressive transformers to generalize from long sequences, even when they are rare.
Proving that next-token prediction makes language models generate coherent long documents.
problem Understanding why language models generate coherent documents despite focusing on next-token prediction.
method Proving the power of next-token prediction in learning longer-range structure using Recurrent Neural Networks (RNN).
result Optimizing next-token prediction in RNNs yields a model that closely approximates the training distribution, even for long-range coherence.
The study explores how Transformers predict the next token in a sequence.
problem Understanding the mechanism behind Transformers' autoregressive learning ability.
method Exploring the approximation ability of Transformers for next-token prediction through specific instances and a causal kernel descent method.
result Transformer models can learn context-dependent functions f for next-token prediction based on past and current observations. Law explains how LLMs learn to predict next tokens.
problem Understanding how LLMs process input data internally.
method Introduced a precise law governing token embeddings in LLMs.
result Each layer equally contributes to next-token prediction accuracy.
This paper examines how different decoding algorithms for LLMs align with various goals.
problem Consistency of decoding algorithms with different goals in LLMs.
method Analysis of greedy, lookahead, random sampling, and temperature-scaled random sampling algorithms.
result Random sampling is consistent with the true probability distribution, but other goals require optimal algorithms for specific probability distributions.
This paper explores how Transformers predict next tokens in autoregressive tasks.
problem Understanding the success of Transformers in autoregressive learning.
method Trained a Transformer on a next-token prediction task, focusing on commuting orthogonal matrices.
result Trained Transformers can be seen as implementing gradient descent for a specific objective function.
Gradient descent biases linear models in next-token prediction towards data entropy.
problem Optimization bias in next-token prediction models.
method Analysis of gradient descent on linear models with sparse conditional distributions.
result Gradient descent selects parameters that equate token logits differences to log-odds in the data subspace.
A new method improves language model extrapolation without changing training.
problem Improving language model performance on sequences not seen during training.
method Perturbation-based training procedure for language models.
result The method consistently improves out-of-support prediction.
The paper analyzes the training dynamics of a transformer for next-token prediction.
problem Understanding the non-asymptotic performance of transformers in next-token prediction.
method Characterizes training dataset properties, designs a two-stage training algorithm, and analyzes attention gradient properties.
result Trained transformers converge sub-linearly to max-margin solutions and exhibit linear convergence in cross-entropy loss.
Two methods for model adaptation compared; fine-tuning outperforms Best-of-N in realizable settings.
problem Comparing methods for adapting large language models to new tasks.
method Supervised fine-tuning vs. Best-of-N approach.
result Supervised fine-tuning outperforms Best-of-N in realizable settings.
This paper shows ARMs and EBMs are equivalent, revealing ARM lookahead capabilities.
problem Understanding the lookahead capabilities of next-token prediction models.
method Unified view of ARMs and EBMs, establishing a bijection and deriving equivalence.
result ARMs and EBMs are equivalent, revealing ARM lookahead capabilities.
Mathematical framework for language models processes text and predicts next tokens.
problem Understanding and optimizing the performance of large language models.
method Describes encoding, prediction models, learning from data, and deployment of LLMs.
result Demonstrates remarkable empirical successes and provides a platform for further research.
Study quantifies how LLMs capture higher-order statistical structure using cumulant expansion.
problem Understanding how LLMs internalize statistical structure during next-token prediction.
method Cumulant-expansion framework treating softmax entropy as perturbation around center distribution.
result Cumulants reveal distinct signatures for mathematical vs. general text prompts, quantifying feature-learning dynamics.
Language models can predict numeric values as strings.
problem Regression tasks with numeric predictions.
method Causal sequence decoding models trained for next-token prediction.
result Decoder-based heads perform as well as standard heads in numeric regression tasks.
New L2D framework allows deferring specific parts of a sequence prediction to experts.
problem Current L2D methods defer entire predictions, which is not ideal for long sequences.
method Proposes token-level and one-time rejectors to defer specific outputs of a model prediction to experts.
result Granular deferrals achieve better cost-accuracy tradeoffs than whole deferrals.
Study reveals AI's spontaneous topic changes in text prediction.
problem AI's inability to spontaneously switch topics like humans.
method Defined topic as Token Priority Graphs (TPGs) and analyzed self-attention models.
result AI can only switch topics if lower-priority tokens outnumber higher-priority ones.
Transformers approximate Bayesian posteriors but not exactly.
problem Bayesian accounts of in-context learning face challenges due to task-preserving order changes in transformers.
method Showed that excess prequential code length is exactly cumulative predictive KL, decomposing expected regret into order-averaged predictor and order-averaging gain.
result Transformers approximate Bayesian posteriors but not exactly, priced by log loss.
The paper distinguishes between conditional and marginal processes in language models and discusses conditions for usefulness.
problem The conditional nature of language models trained on observed sequences and the need for marginal text-only processes.
method Distinguishing between full conditional language process, marginal text-only process, and model-induced distribution; analyzing assumptions of stationarity and ergodicity.
result The marginal text-only law is useful only when the observed prefix is an approximately sufficient statistic for the latent circumstances relevant to continuation.
Pre-training improves model coverage, crucial for downstream performance.
problem Understanding why pre-training enhances model performance.
method Coverage principle, focusing on next-token prediction and model quality.
result Coverage generalizes faster than cross-entropy, improving downstream performance.
Unified theory for neural scaling laws in hierarchically compositional data.
problem Understanding neural scaling laws in hierarchically compositional data.
method Probabilistic context-free grammars and power-law distributed production rules.
result Unified learning curve behavior for classification and next-token prediction tasks.
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.
A method for learning with autoregressive chain-of-thoughts.
problem Learning prompt-to-answer mappings from sequence-to-next-token generators.
method Iterating a fixed, time-invariant generator for multiple steps to generate a chain-of-thought, then taking the final token as the answer.
result Universal representability and computationally tractable chain-of-thought learning for a simple base class.
Paper reduces vocabulary losslessly for language model cooperation.
problem Language models struggle to cooperate with different tokenizations.
method Established a theoretical framework for lossless vocabulary reduction.
result Efficiently converts models with different tokenizations to cooperate with maximal common vocabulary.
Study shows how transformers classify symbols without naming them, proving a margin-versus-collision criterion.
problem How transformers classify symbols without naming them.
method Logistic classification analysis of transformer-kernel regime, colored collision graph.
result Decomposes learned predictor into ideal template-level classifier and finite-sample perturbation.
Improved Gumbel watermark detection method.
problem Detecting Gumbel watermarking with high accuracy.
method A near-optimal model-agnostic detection mechanism.
result Demonstrates near-optimal performance under i.i.d. assumptions.
Timer-XL predicts multidimensional time series using a unified Transformer approach.
problem Unified time series forecasting across various tasks and contexts.
method Decoder-only Transformers with a universal TimeAttention mechanism and deft position embedding.
result State-of-the-art performance across multiple forecasting benchmarks.
GPT learns a causal world model from token predictions, validated in game sequences.
problem Does GPT implicitly learn a causal world model from token predictions?
method Derived a causal interpretation of GPT's attention mechanism and proposed zero-shot causal structure learning.
result GPT can generate legal next moves with high confidence for sequences with encoded causal structures, but fails for illegal moves.
This paper studies transformer learning dynamics and initialization.
problem Understanding how transformers learn Markov chains and the role of initialization.
method First-order Markov chains and single-layer transformers, proving learning dynamics and conditions for convergence.
result Transformer parameters can converge to global or local minima based on initialization and Markovian data properties.
New approach predicts tokens in context, explaining how ICL emerges.
problem Limited understanding of in-context learning emergence.
method Auto-regressive next-token prediction (AR-NTP) with prompt token-dependency and a two-level expectation.
result ICL emerges from the generalization of sequences and topics.
LLMs show surprising confidence in their answers, beyond just tokens.
problem LLMs lack meaningful confidence estimates for their responses.
method Semantic calibration test based on local loss optimality and equivalence classes.
result Base LLMs are semantically calibrated across tasks, contrary to expectations.
Transformers learn linear models in-context without updates.
problem Understanding how transformers mimic linear models in-context.
method Gradient flow on linear regression tasks with random initialization.
result Transformers achieve prediction error competitive with best linear predictors.
The problem of forecasting conditional probabilities of the next event given the past is considered in a general probabilistic setting. Given an arbitrary (large, uncountable) set C of predictors, we would like to construct a single predictor that performs asymptotically as well as the best predictor in C, on any data.…
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.
The study examines how language models learn to represent the world, identifying conditions for ecological veridicality.
problem Understanding when language models learn to represent the world accurately and how this learning process can fail.
method Analyzes the Bayes-optimal next-token cross-entropy decomposition and the role of training ecology in shaping model representations.
result The minimum-complexity zero-excess solution is the quotient partition by training equivalence, and this solution is not preserved in in-context learning or per-task adaptation.
New bounds explain deterministic non-smooth deep nets without large Lipschitz constants.
problem Challenges in explaining generalization of deterministic non-smooth deep nets.
method De-randomized PAC-Bayes margin bounds for deterministic non-convex and non-smooth predictors.
result New bounds avoid large Lipschitz constants, providing generalization guarantees.
This paper proposes a method to reduce complexity in GLMs with categorical predictors.
problem Wasteful, hard-to-interpret, and prone to overfitting of traditional one-hot encoding for high-cardinality categorical predictors.
method Clustering categories of categorical predictors through a numerical method that preserves or improves accuracy while reducing the number of coefficients.
result Clustering categories of categorical predictors reduces complexity substantially without harming accuracy.
The article compares predictor importance in classification problems with categorical outcomes.
problem Comparing predictor importance in classification problems with categorical response variables.
method The approach is based on the categorical Gini correlation (CGC) and tests differences in CGCs across predictor groups.
result The proposed methodology accommodates predictors of arbitrary and unequal dimensions and allows for dependence between predictor groups.
Paper proposes a sparse synthetic control method to select important predictors.
problem Choosing and weighting predictors affects synthetic control estimator performance.
method Sparse synthetic control procedure that penalizes predictors, derived in a linear factor model.
result Sparse synthetic control achieves lower bias and better post-treatment performance.
This paper presents Sparse Partitioning, a Bayesian method for identifying predictors that either individually or in combination with others affect a response variable. The method is designed for regression problems involving binary or tertiary predictors and allows the number of predictors to exceed the size of the sa…
WeakNAS uses a set of weaker predictors to find top architectures with fewer samples.
problem Finding the best neural architecture with heavy computation costs.
method Proposes a paradigm shift from fitting the whole architecture space to progressively fitting a search path through a set of weaker predictors.
result WeakNAS produces coarse-to-fine iteration to gradually refine the ranking of sampling space, requiring fewer samples to find top-performance architectures.
Proposes a method to create fair, robust predictors that remain consistent across different scenarios.
problem Creating fair and robust machine learning models that behave consistently across different scenarios.
method Graphical criteria and a model-agnostic framework called CIP based on HSCIC.
result Demonstrates the effectiveness of CIP in enforcing counterfactual invariance across various datasets.
This paper continues study, both theoretical and empirical, of the method of Venn prediction, concentrating on binary prediction problems. Venn predictors produce probability-type predictions for the labels of test objects which are guaranteed to be well calibrated under the standard assumption that the observations ar…
Derives bounds for deterministic predictors using smooth loss functions.
problem Generalizing probabilistic predictors to deterministic ones.
method Exploits smoothness properties of loss and predictor classes, controlling the Jensen gap class through Rademacher complexity.
result Derives bounds for deterministic predictors involving flatness quantities from Jacobians and Hessians.
Study shows competition feedback can make ML predictors biased towards specific user groups.
problem How competition affects machine learning predictors and user prediction quality.
method Flexible model of competing ML predictors, empirical and mathematical analysis.
result Competition causes predictors to specialize for specific sub-populations at the cost of general performance.
Study on merging predictors in causal and anticausal directions using CMAXENT.
problem Comparing merging predictors in causal and anticausal directions.
method Using CMAXENT as inductive bias, study differences in merging predictors.
result CMAXENT solution reduces to logistic regression in causal direction and LDA in anticausal direction.
Paper introduces SUEL model for integrating predictors without labeled data.
problem Combining predictors with unknown accuracy and high correlation.
method Structured unsupervised ensemble learning (SUEL) with correlation-based decomposition algorithms.
result Efficient integration of dependent predictors without labeled data.
The paper uncovers symmetries in large language models through layer-peeled optimization.
problem Understanding geometric structure in large language model weights and context embeddings.
method Constrained layer-peeled optimization program to analyze symmetries in next-token distributions.
result Symmetries in target next-token distributions are transferred to optimal model weights and context embeddings.