Researchers analyze the structure of token spaces in large language models.
problem Understanding the geometric and topological structure of token spaces in large language models.
method Developed estimators for the dimension and Ricci scalar curvature of the token subspace and applied them to three large language models.
result The token subspace is not a manifold but a stratified manifold with significantly negative Ricci curvature.
DS-CP improves reliability of uncertainty quantification for large language models under domain shift.
problem Overconfident and factually incorrect outputs (hallucinations) from large language models.
method Adapts conformal prediction to large language models under domain shift by reweighting calibration samples.
result DS-CP delivers more reliable coverage than standard conformal prediction, especially under substantial distribution shifts.
Large language models can't efficiently reason conditionally in a distribution-free setting.
problem Impossibility of conditional PAC-efficient reasoning in large language models.
method Proof of impossibility in a distribution-free setting for non-atomic input spaces.
result Any algorithm achieving conditional PAC efficiency must defer to the expert model with high probability.
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.
This work uses conformal prediction to quantify uncertainty in large language models for multiple-choice questions.
problem Ensuring robustness and reliability of large language models in high-stakes applications.
method Conformal prediction applied to multi-choice question answering.
result Uncertainty estimates from conformal prediction are closely related to prediction accuracy.
Shai is a 10B model for asset management tasks, outperforming baselines.
problem Improving performance in asset management tasks.
method Continuous pre-training and fine-tuning on asset management-specific data.
result Shai outperforms baseline models in asset management tasks.
Large language models improve futures market factor models in China.
problem Designing effective factor models for Chinese futures markets.
method Used large language models (GPT) to generate 40 factors for single and multi-factor portfolios.
result GPT-generated factors outperform benchmarks with high Sharpe ratios and alphas.
Proposes a robust algorithm for aligning large language models with human preferences.
problem Misspecification in preference models, reference policies, and reward functions.
method Doubly robust preference optimization algorithm.
result Superior and more robust performance compared to state-of-the-art algorithms.
Graph signal processing detects hallucinations in large language models.
problem Detecting factual reasoning from hallucinations in large language models.
method Modeling transformer layers as dynamic graphs, using spectral analysis to define diagnostics.
result Spectral signatures can distinguish different types of hallucinations and achieve high accuracy.
We introduce a new sampling method for large language models that balances diversity and parallelism.
problem Balancing diversity and parallelism in decoding for large language models.
method Arithmetic sampling framework compatible with various sampling variations.
result Improves estimation of expected BLEU score reward and reduces the gap with beam search.
ScheduleFree+ improves large language model training without schedules or learning rates.
problem Scaling up Schedule-Free Learning to large language models.
method Learning-rate-free and schedule-free method for training large language models.
result ScheduleFree+ outperforms SOTA schedules by 31% at 1000 tokens per parameter.
L3Ms fine-tune LLMs with constraints for tailored applications.
problem Inadequate alignment of LLMs for diverse applications.
method Formulate SFT and alignment as constrained optimization, using logarithmic barriers.
result Versatile and effective in achieving tailored alignments for various applications.
Efficient exploration improves large language model performance with fewer queries.
problem Improving large language model performance with fewer human feedback queries.
method Sequentially generates queries, fits a reward model to feedback, uses double Thompson sampling with epistemic neural network uncertainty.
result Efficient exploration enables high performance with far fewer queries.
The study analyzes how large language models form and express investor risk profiles.
problem Understanding how large language models (LLMs) form and express investor risk profiles.
method Examined three LLMs (GPT, Gemini, and Llama) and assessed their responses to a standardized risk questionnaire under varying prompts.
result LLMs generally form long-term investment profiles, but they exhibit different risk tolerance levels.
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.
Active testing for large language models is made more efficient and accurate.
problem Efficient evaluation of large language models with limited labels.
method Cost-saving measures and in-context learning for constructing a surrogate model.
result Significantly more accurate evaluations of LLM performance compared to random data acquisition.
AudioPaLM combines text and speech models to improve speech processing and translation.
problem Improving speech understanding and generation models.
method Unified multimodal architecture combining text and speech models.
result AudioPaLM significantly outperforms existing speech translation systems.
New bounds show large language models can generalize beyond training data.
problem Generalization of large language models beyond training data.
method Compression bound derivation using prediction smoothing and SubLoRA.
result Large language models can discover generalizable regularities.
Method reveals hidden token embeddings of large language models.
problem Understanding hidden token embeddings in large language models.
method Structured prompts to expose token input embeddings up to homeomorphism.
result Mathematical proof for generic LLMs shows effectiveness of method.
A simple model explains phase transition in large language models.
problem Understanding the emergence of abilities in large language models.
method Modeling LLM as a sequence-to-sequence random function and using a list decoder.
result A critical threshold exists where the expected number of erroneous sequences grows exponentially.
This paper benchmarks FinGPT for financial datasets using open-source large language models.
problem Challenges in integrating GPT-based models with financial datasets.
method Instruction Tuning paradigm for open-source large language models adapted for financial contexts.
result Demonstrates the effectiveness and adaptability of FinGPT in financial tasks.
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.
Study evaluates LLMs for predicting Chinese stock movements using financial news sentiments.
problem Evaluating LLMs' ability to predict stock price movements using financial news sentiments.
method Standardized experimental procedure with three LLMs, each with unique performance enhancement methods.
result Developed quantitative trading strategies and conducted back-tests to assess LLMs' performance.
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.
A new method reduces the computational burden of safety alignment for large language models.
problem Safety concerns in large language models and the need to align them with human preferences.
method Optimal dualization approach to reduce constrained alignment to an unconstrained problem.
result Our algorithms MoCAN and PeCAN significantly reduce computational burden and improve training stability.
Study constructs a Japanese financial LLM benchmark.
problem Need for domain-specific benchmarks for LLMs.
method Constructed a benchmark with multiple Japanese and financial domain tasks.
result GPT-4 outperforms other models in the benchmark.
Paper addresses online alignment of large language models under uncertain preference feedback.
problem Online alignment of large language models with misspecified preference feedback.
method Formulates an oracle-robust objective as a worst-case optimization problem for log-linear policies, and develops projected stochastic composite updates.
result Shows that the robust objective admits an exact closed-form decomposition and achieves O ~ ( ε − 2 ) \widetilde{O}(\varepsilon^{-2}) O ( ε − 2 ) oracle complexity. 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.
Large language models learn company embeddings from SEC filings.
problem Lack of a rigorous definition of company similarity.
method Pre-trained and finetuned large language models (LLMs) to learn embeddings from SEC filings.
result LLMs can reproduce GICS classifications and indicate similar financial performance.
Study uses LLMs to simplify financial regulation interpretation.
problem Complex financial regulations are hard to interpret and implement.
method Developed prompts to guide LLMs in extracting key information from regulations.
result GPT-4 outperforms other LLMs in processing and executing regulatory requirements.
ConMeZO speeds up zeroth-order optimization for large language models.
problem Slow convergence in high-dimensional parameter spaces of large language models.
method Adaptive directional sampling in a cone centered around a momentum estimate.
result Achieves the same convergence rate as MeZO but up to 2X faster.
BreakGPT predicts asset price surges using LLMs.
problem Predicting sharp upward movements in volatile financial markets.
method Adapts LLMs for time series forecasting, combining LLM capabilities with Transformer models.
result BreakGPT effectively captures local and global temporal dependencies.
Improved IVON boosts LoRA model accuracy and calibration.
problem Improving the accuracy and calibration of large language models.
method Replaced AdamW with IVON for finetuning Llama-2.
result IVON improves Llama-2 accuracy by 2.8% and calibration error by 4.6%.
A new framework evaluates large language models efficiently and accurately.
problem Evaluation of large language models is challenging due to stochasticity and heterogeneity of benchmarks.
method Interpretable and scalable framework based on Item Response Theory (IRT) and majorization-minimization principle.
result Our method achieves superior scalability and interpretability compared to existing approaches.
Two methods improve 10-K item segmentation using large language models.
problem Challenges in extracting specific items from 10-K reports due to variations in document formats and item presentation.
method Two advanced item segmentation methods: GPT4ItemSeg and BERT4ItemSeg.
result BERT4ItemSeg achieves a macro-F1 of 0.9825, surpassing other methods.
Extends linear representation hypothesis to categorical and hierarchical concepts in LLMs.
problem Representing concepts without natural contrasts in large language models.
method Formalizes linear representation hypothesis for categorical and hierarchical concepts, proving relationships between concept hierarchy and representation geometry.
result Validated theoretical results on large language models, estimating representations for 900+ concepts.
TROLL improves RL for LLMs by replacing clipping with a trust region projection.
problem Clipping in RL for LLMs causes instability and suboptimal performance.
method TROLL uses a discrete differentiable trust region projection to replace clipping, balancing computational cost and effectiveness.
result TROLL consistently outperforms PPO-like clipping in training speed, stability, and final success rates.
This paper uses Gaussian mixtures to mimic interactions in large language models.
problem Simulating interactions in large language models (LLMs) is computationally expensive.
method Developed an interacting Gaussian mixture model (GMM) system to mimic LLM interactions.
result The interacting Gaussian mixture model system can generate, exchange, and update data and parameters efficiently.
We develop a statistical framework to benchmark and select large language models based on their risks.
problem Benchmarking and selecting large language models based on their associated risks.
method A distributional framework using first and second order stochastic dominance, linked to mean-risk models in finance.
result Formalizes a risk-aware approach for model selection, balancing risk and utility.
Research integrates sentiment analysis with reinforcement learning for better trading strategies.
problem Improving trading performance by integrating sentiment data.
method Developed a sentiment-driven trading system using a large language model and reinforcement learning.
result Sentiment signals from FinGPT improve trading performance when combined with technical indicators.
SAIL improves online alignment of large language models with minimal feedback.
problem Offline RLHF methods often lead to sub-optimal performance due to fixed preference datasets.
method SAIL uses bilevel optimization and a single-level first-order method to iteratively refine model alignment.
result SAIL significantly improves alignment performance on open-sourced datasets with minimal computational overhead.
Large language models have recently achieved state of the art performance across a wide variety of natural language tasks. Meanwhile, the size of these models and their latency have significantly increased, which makes their usage costly, and raises an interesting question: do language models need to be large? We study…
Large language models correlate in errors, even with different architectures and providers.
problem Lack of empirical evidence on whether different large language models differ meaningfully.
method Empirical evaluation of over 350 large language models using two leaderboards and a resume-screening task.
result Large language models have substantial correlation in errors, even with distinct architectures and providers.
New principle reduces load imbalance in LLM serving systems, saving up to 52% energy.
problem Wasted computational power due to load imbalance in LLM serving systems.
method Developed a universal load-balancing principle for barrier-synchronized systems with non-migratable state.
result Proves worst-case theoretical guarantees for imbalance reduction and energy savings.
BloombergGPT is a large language model trained on financial data, outperforming existing models on financial tasks.
problem Lack of specialized large language models for finance.
method Trained on a 363 billion token dataset augmented with 345 billion tokens from general datasets, using a 50 billion parameter model.
result BloombergGPT outperforms existing models on financial tasks without sacrificing performance on general LLM benchmarks.
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.
LLMs can help explain credit risk models but not autonomously.
problem Leveraging LLMs for post-hoc explainability in credit risk models.
method Comparison of LLM outputs with SHAP and coefficient-based attributions on three LMs.
result LLMs reliably preserve feature-importance rankings but poorly align with autonomous explanations.
Chronologically consistent models maintain accuracy with time-restricted data.
problem Training data introduces lookahead bias and training leakage in large language models.
method ChronoBERT and ChronoGPT trained with only available data at each time point.
result Models achieve strong performance and competitive with larger models, mitigating lookahead bias.