{"topic":"LLMs & Text","items":[{"title":"Financial Language Models as Applied Artificial Intelligence Systems for News-Based Trading under Market Frictions","url":"/papers/arxiv/2609.23703/","summary":"Introduces a market-friction-aware framework that converts timestamped financial news into auditable trading decisions while accounting for execution timing, transaction costs, and liquidity constraints.","featured":"2026-09-25","label":"arXiv","topic":"LLMs & Text","cites":0,"score":3,"scale":"fanfare"},{"title":"From Tone to Trajectory: Continuous Sentiment and the Shape of Monetary Policy Communication","url":"/papers/arxiv/2609.25034/","summary":"The study shows that how monetary policy sentiment unfolds across a press conference, not just its average tone, predicts rate changes and shapes forecaster expectations at the ECB and Fed.","featured":"2026-09-25","label":"arXiv","topic":"LLMs & Text","cites":0,"score":3,"scale":"fanfare"},{"title":"FinInteract: Benchmarking Clarification and Intent Integration in Ambiguous Financial Question Answering","url":"/papers/arxiv/2609.24002/","summary":"A benchmark reveals that language models answer financial questions above 90 percent with clarification but only 28.9 percent when they must elicit it themselves, exposing model ambiguity resolution.","featured":"2026-09-25","label":"arXiv","topic":"LLMs & Text","cites":0,"score":3,"scale":"fanfare"},{"title":"FinRankGRPO: Optimizing LLMs for Listwise Financial Asset Ranking via Group Relative Policy Optimization","url":"/papers/arxiv/2609.24175/","summary":"Develops a two-stage framework that fine-tunes language models for listwise asset ranking using Spearman rank correlation rewards, achieving a Sharpe ratio of 0.636 on asset allocation.","featured":"2026-09-25","label":"arXiv","topic":"LLMs & Text","cites":0,"score":2,"scale":"fanfare"},{"title":"LLM-Based Semantic Surprises in FOMC Communication: Asset Prices and Financial-Market Stress","url":"/papers/ssrn/7519200/","summary":"Semantic surprises extracted from Federal Reserve statements predict subsequent financial-stress dynamics and reduce forecast error by up to 23%, particularly when initial stress is high or during recessions.","featured":"2026-09-25","label":"SSRN","topic":"LLMs & Text","cites":null,"score":3,"scale":"fanfare"},{"title":"Transforming the Voice of the Customer: Large Language Models for Identifying Customer Needs","url":"/papers/arxiv/2503.01870/","summary":"Large Language Models are streamlining the process of identifying customer needs, letting analysts concentrate on more valuable work while still delivering precise insights.","featured":"2026-04-16","label":"arXiv","topic":"LLMs & Text","cites":4,"score":0,"scale":"shares"},{"title":"Twitter Sentiment and Financial Trends","url":"/papers/ssrn/4467949/","summary":"A new financial sentiment index derived from Twitter data shows strong links to market conditions and can forecast stock market returns, particularly in response to changes in U.S. monetary policy.","featured":"2025-12-28","label":"SSRN","topic":"LLMs & Text","cites":null,"score":105,"scale":"shares"},{"title":"Measuring Corruption from Text Data","url":"/papers/arxiv/2512.09652/","summary":"An automated corruption index using Brazilian municipal audit reports is efficient and more reliable than manual methods in detecting corruption.","featured":"2025-12-14","label":"arXiv","topic":"LLMs & Text","cites":1,"score":0,"scale":"shares"},{"title":"Reasoning Models Ace the CFA Exams","url":"/papers/arxiv/2512.08270/","summary":"An evaluation of reasoning models on CFA mock exams shows that models like Gemini 3.0 Pro and GPT-5 perform well, achieving high pass rates in professional testing.","featured":"2025-12-14","label":"arXiv","topic":"LLMs & Text","cites":2,"score":0,"scale":"shares"},{"title":"Standard Occupation Classifier - A Natural Language Processing Approach","url":"/papers/arxiv/2511.23057/","summary":"A project successfully developed a natural language processing model that classifies job ads with 72% accuracy by using an ensemble approach.","featured":"2025-12-01","label":"arXiv","topic":"LLMs & Text","cites":0,"score":0,"scale":"shares"},{"title":"Measuring economic outlook in the news","url":"/papers/arxiv/2511.04299/","summary":"We build an interpretable, privacy‑friendly sentiment indicator from Swiss news using ML and LLMs, and it improves short‑term GDP forecasts.","featured":"2025-11-12","label":"arXiv","topic":"LLMs & Text","cites":1,"score":1,"scale":"shares"},{"title":"Aligning Multilingual News for Stock Return Prediction","url":"/papers/arxiv/2510.19203/","summary":"Uses optimal-transport to align English–Japanese stock news, producing clearer signals that better predict returns.","featured":"2025-10-27","label":"arXiv","topic":"LLMs & Text","cites":0,"score":8,"scale":"shares"},{"title":"Black Box Absorption: LLMs Undermining Innovative Ideas","url":"/papers/arxiv/2510.20612/","summary":"LLM platforms can quietly absorb users’ ideas, creating power imbalances; the paper proposes governance and technical fixes to protect creators.","featured":"2025-10-27","label":"arXiv","topic":"LLMs & Text","cites":1,"score":20,"scale":"shares"},{"title":"Integrating Transparent Models, LLMs, and Practitioner-in-the-Loop: A Case of Nonprofit Program Evaluation","url":"/papers/arxiv/2510.19799/","summary":"Combining transparent decision trees, LLMs, and practitioner input yields accurate, explainable case-level predictions for public and nonprofit programs.","featured":"2025-10-27","label":"arXiv","topic":"LLMs & Text","cites":0,"score":12,"scale":"shares"},{"title":"News-Aware Direct Reinforcement Trading for Financial Markets","url":"/papers/arxiv/2510.19173/","summary":"- News-RL Trading - News-Driven Trading - News-Based RL - RL for News Trading - News-Powered RL If you want the shortest single choice: News-RL Trading.: Short summary: Feeding news sentiment extracted by large language models together with raw price and volume into a sequence-model reinforcement learning system boosts cryptocurrency trading performance, removing the need for handcrafted trading rules.","featured":"2025-10-27","label":"arXiv","topic":"LLMs & Text","cites":1,"score":8,"scale":"shares"},{"title":"Abstract Classification: SVM vs BERT vs GPT-3.5","url":"/papers/repec/spr-scient-v-130-y-2025-i-1-d-10-1007-s11192-024-05217-7/","summary":"SVM vs BERT vs GPT-3.5: Compares SVM, SPECTER, BERT, and GPT-3.5 for classifying abstracts: BERT performs best, while GPT-3.5 is inconsistent with limited training data.","featured":"2025-10-27","label":"RePEc","topic":"LLMs & Text","cites":null,"score":4,"scale":"shares"},{"title":"From Classical Rationality to Contextual Reasoning: Quantum Logic as a New Frontier for Human-Centric AI in Finance","url":"/papers/arxiv/2510.05475/","summary":"The potential of quantum logic in advancing artificial intelligence applications in financial modeling is discussed.","featured":"2025-10-09","label":"arXiv","topic":"LLMs & Text","cites":2,"score":10,"scale":"shares"},{"title":"From News to Returns: A Granger-Causal Hypergraph Transformer on the Sphere","url":"/papers/arxiv/2510.04357/","summary":"The research suggests the CSHT, a new architecture for financial time-series forecasting that considers the impact of financial news and sentiment on asset returns, providing robust generalisation across market regimes and clear attribution pathways.","featured":"2025-10-09","label":"arXiv","topic":"LLMs & Text","cites":3,"score":5,"scale":"shares"},{"title":"Extracting the Structure of Press Releases for Predicting Earnings Announcement Returns","url":"/papers/doi/10-1145-3768292-3770344/","summary":"The research explores the predictive power of textual features in earnings press releases on stock returns, concluding that press release content is as informative as earnings surprise, with FinBERT being the most predictive.","featured":"2025-10-03","label":"arXiv","topic":"LLMs & Text","cites":0,"score":7,"scale":"shares"},{"title":"The (Short-Term) Effects of Large Language Models on Unemployment and Earnings","url":"/papers/arxiv/2509.15510/","summary":"Large Language Models like ChatGPT have boosted earnings for workers in certain occupations without affecting unemployment rates, indicating they enhance income rather than replace jobs.","featured":"2025-09-22","label":"arXiv","topic":"LLMs & Text","cites":2,"score":19,"scale":"shares"},{"title":"Context-Aware Language Models for Forecasting Market Impact from Sequences of Financial News","url":"/papers/arxiv/2509.12519/","summary":"The study suggests using large language models to process financial news and small models to encode historical context, resulting in improved simulated investment performance.","featured":"2025-09-22","label":"arXiv","topic":"LLMs & Text","cites":2,"score":6,"scale":"shares"},{"title":"Trading-R1: Financial Trading with LLM Reasoning via Reinforcement Learning","url":"/papers/arxiv/2509.11420/","summary":"The article presents Trading-R1, a finance-focused AI model that aligns with trading principles, showing it offers better risk-adjusted returns and fewer drawdowns than other models.","featured":"2025-09-22","label":"arXiv","topic":"LLMs & Text","cites":17,"score":35,"scale":"shares"},{"title":"FinReflectKG: Agentic Construction and Evaluation of Financial Knowledge Graphs","url":"/papers/arxiv/2508.17906/","summary":"Financial Knowledge Graph Construction: The paper introduces a large-scale financial knowledge graph dataset from SEC 10-K filings of S and P 100 companies, with a reflection-agent-based mode providing the best balance of efficiency, accuracy, and reliability.","featured":"2025-08-29","label":"arXiv","topic":"LLMs & Text","cites":22,"score":7,"scale":"shares"},{"title":"Bias-Adjusted LLM Agents for Human-Like Decision-Making via Behavioral Economics","url":"/papers/arxiv/2508.18600/","summary":"A persona-based approach using individual-level data from behavioral economics shows potential in adjusting biases in large language models, enabling them to simulate human-like decision patterns.","featured":"2025-08-29","label":"arXiv","topic":"LLMs & Text","cites":3,"score":9,"scale":"shares"},{"title":"AlphaAgents: Large Language Model based Multi-Agents for Equity Portfolio Constructions","url":"/papers/arxiv/2508.11152/","summary":"The article investigates the application and effectiveness of role-based multi-agent AI systems in equity research and portfolio management, including their advantages and challenges.","featured":"2025-08-20","label":"arXiv","topic":"LLMs & Text","cites":23,"score":56,"scale":"shares"},{"title":"Note on Selection Bias in Observational Estimates of Algorithmic Progress","url":"/papers/arxiv/2508.11033/","summary":"Criticisms have been raised about Ho et. al's 2024 study on the growing efficiency of language models, including potential selection bias in assessing algorithmic quality.","featured":"2025-08-20","label":"arXiv","topic":"LLMs & Text","cites":3,"score":9,"scale":"shares"},{"title":"Interpreting the Interpreter: Can We Model post-ECB Conferences Volatility with LLM Agents?","url":"/papers/arxiv/2508.13635/","summary":"A new method using a Large Language Model can predict financial market responses to European Central Bank press conferences, aiding in maintaining financial stability.","featured":"2025-08-20","label":"arXiv","topic":"LLMs & Text","cites":1,"score":5,"scale":"shares"},{"title":"A Multi-Task Evaluation of LLMs' Processing of Academic Text Input","url":"/papers/arxiv/2508.11779/","summary":"Large language models like Google's Gemini struggle with processing academic text, showing reliable summarizing and paraphrasing skills but poor text grading and reflection abilities, hence their unchecked use in peer reviews is discouraged.","featured":"2025-08-20","label":"arXiv","topic":"LLMs & Text","cites":6,"score":8,"scale":"shares"},{"title":"Prompt-Response Semantic Divergence Metrics for Faithfulness Hallucination and Misalignment Detection in Large Language Models","url":"/papers/arxiv/2508.10192/","summary":"The paper presents Semantic Divergence Metrics (SDM), a new system that improves the detection of significant deviations in Large Language Models' responses from the input context by measuring consistency across various semantically equivalent paraphrases.","featured":"2025-08-20","label":"arXiv","topic":"LLMs & Text","cites":4,"score":16,"scale":"shares"},{"title":"Event-Aware Sentiment Factors from LLM-Augmented Financial Tweets: A Transparent Framework for Interpretable Quant Trading","url":"/papers/arxiv/2508.07408/","summary":"The study shows how large language models can be used in financial semantic annotation and alpha signal discovery, indicating that social media sentiment can be a useful predictor in financial forecasting.","featured":"2025-08-12","label":"arXiv","topic":"LLMs & Text","cites":2,"score":2,"scale":"shares"},{"title":"Language Model Guided Reinforcement Learning in Quantitative Trading","url":"/papers/arxiv/2508.02366/","summary":"A proposed hybrid system uses large language models to create high-level trading strategies, directing reinforcement learning agents and showing better return and risk metrics than standard reinforcement learning.","featured":"2025-08-07","label":"arXiv","topic":"LLMs & Text","cites":8,"score":3,"scale":"shares"},{"title":"Evolutionary Factor Searching for Sparse Portfolio Optimization Using Large Language Models","url":"/papers/arxiv/2507.17211/","summary":"The paper presents Evolutionary Factor Search (EFS), a new method that uses large language models to automatically generate and evolve alpha factors for sparse portfolio construction, showing its effectiveness in different market situations.","featured":"2025-07-25","label":"arXiv","topic":"LLMs & Text","cites":0,"score":7,"scale":"shares"},{"title":"FinDPO: Financial Sentiment Analysis for Algorithmic Trading through Preference Optimization of LLMs","url":"/papers/arxiv/2507.18417/","summary":"FinDPO, a finance-specific large language model, outperforms existing models in sentiment analysis and maintains significant positive returns in portfolio strategies, even with realistic transaction costs.","featured":"2025-07-25","label":"arXiv","topic":"LLMs & Text","cites":7,"score":10,"scale":"shares"},{"title":"When Experimental Economics Meets Large Language Models: Evidence-based Tactics","url":"/papers/arxiv/2505.21371/","summary":"The research offers guidelines for creating economic experiments for large language models, improving the design, replicability, and general applicability of these experiments in the digital era.","featured":"2025-07-25","label":"arXiv","topic":"LLMs & Text","cites":1,"score":18,"scale":"shares"},{"title":"To Trade or Not to Trade: An Agentic Approach to Estimating Market Risk Improves Trading Decisions","url":"/papers/arxiv/2507.08584/","summary":"A system using large language models can enhance market risk estimation and trading decisions by discovering stochastic differential equations for financial time series.","featured":"2025-07-17","label":"arXiv","topic":"LLMs & Text","cites":4,"score":11,"scale":"shares"},{"title":"Signal or Noise? Evaluating Large Language Models in Resume Screening Across Contextual Variations and Human Expert Benchmarks","url":"/papers/arxiv/2507.08019/","summary":"Large language models used for matching resumes with job descriptions show consistent patterns, but their evaluations differ greatly from human experts, impacting their use in automated hiring systems.","featured":"2025-07-17","label":"arXiv","topic":"LLMs & Text","cites":2,"score":11,"scale":"shares"},{"title":"Does Overnight News Explain Overnight Returns?","url":"/papers/arxiv/2507.04481/","summary":"The study finds that most gains in the U.S. stock market over the past 30 years have been earned overnight, with intraday returns being negative or flat, largely due to intraday and overnight news.","featured":"2025-07-10","label":"arXiv","topic":"LLMs & Text","cites":0,"score":11,"scale":"shares"},{"title":"Beyond Code: The Multidimensional Impacts of Large Language Models in Software Development","url":"/papers/arxiv/2506.22704/","summary":"Large language models like ChatGPT can boost productivity and knowledge sharing among Open-Source Software developers, especially in complex or rapidly changing contexts.","featured":"2025-07-03","label":"arXiv","topic":"LLMs & Text","cites":2,"score":7,"scale":"shares"},{"title":"FairMarket-RL: LLM-Guided Fairness Shaping for Multi-Agent Reinforcement Learning in Peer-to-Peer Markets","url":"/papers/arxiv/2506.22708/","summary":"Fairness Shaping: FairMarket-RL, a hybrid model combining Large Language Models and Reinforcement Learning, is introduced to create fairness-aware trading agents in a simulated microgrid, leading to more equitable outcomes.","featured":"2025-07-03","label":"arXiv","topic":"LLMs & Text","cites":1,"score":7,"scale":"shares"},{"title":"Narrative Shift Detection: A Hybrid Approach of Dynamic Topic Models and Large Language Models","url":"/papers/arxiv/2506.20269/","summary":"Large Language Models combined with topic models can model narrative shifts over time, but they struggle to differentiate between content and narrative shifts, as shown in a study of The Wall Street Journal articles.","featured":"2025-07-03","label":"arXiv","topic":"LLMs & Text","cites":5,"score":7,"scale":"shares"},{"title":"NLP Axioms in Tamil Dialects","url":"/papers/ssrn/5245598/","summary":"The article investigates the challenges and opportunities in using Machine Learning for processing Classical Tamil language and its dialects, with a focus on linguistic and cultural aspects.","featured":"2025-06-25","label":"SSRN","topic":"LLMs & Text","cites":null,"score":13,"scale":"shares"},{"title":"AI is the Strategy: From Agentic AI to Autonomous Business Models onto Strategy in the Age of AI","url":"/papers/arxiv/2506.17339/","summary":"The article explores Autonomous Business Models (ABMs), where AI systems handle key business operations, potentially transforming competitive strategies, organizational structures, and governance.","featured":"2025-06-25","label":"arXiv","topic":"LLMs & Text","cites":3,"score":8,"scale":"shares"},{"title":"Identifying economic narratives in large text corpora - An integrated approach using Large Language Models","url":"/papers/arxiv/2506.15041/","summary":"A study finds that Large Language Models like GPT-4o can interpret economic narratives from texts, but their performance is not yet on par with experts, based on analysis of articles about inflation.","featured":"2025-06-25","label":"arXiv","topic":"LLMs & Text","cites":3,"score":12,"scale":"shares"},{"title":"ChatGPT for Student Engagement","url":"/papers/ssrn/5287045/","summary":"The paper highlights the use of ChatGPT to increase student engagement in online and hybrid learning settings.","featured":"2025-06-18","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Intelligent Automation for FDI Facilitation: Optimizing Tariff Exemption Processes with OCR And Large Language Models","url":"/papers/arxiv/2506.12093/","summary":"The article suggests an AI system that uses OCR and Large Language Models to simplify the verification of tariff exemptions for Foreign Direct Investment in manufacturing, enhancing operational efficiency.","featured":"2025-06-18","label":"arXiv","topic":"LLMs & Text","cites":0,"score":12,"scale":"shares"},{"title":"Large Language Models as 'Hidden Persuaders': Fake Product Reviews are Indistinguishable to Humans and Machines","url":"/papers/arxiv/2506.13313/","summary":"The research shows that both humans and Large Language Models have difficulty distinguishing between genuine and fake product reviews, exposing a susceptibility to automated fraud and emphasizing the need for reliable purchase verification.","featured":"2025-06-18","label":"arXiv","topic":"LLMs & Text","cites":7,"score":10,"scale":"shares"},{"title":"Vision-Language Model Evaluation","url":"/papers/ssrn/5283704/","summary":"The study introduces a surrogate model to assess the resilience of vision-language models to minor perturbations, using adversarial perturbations in text and image modalities.","featured":"2025-06-11","label":"SSRN","topic":"LLMs & Text","cites":null,"score":3,"scale":"shares"},{"title":"Patent Descriptions","url":"/papers/ssrn/5284940/","summary":"A new dataset has been developed using natural language processing and machine learning, providing detailed tech information about US public firms and patents over 30 years, aiding profitable trading strategies.","featured":"2025-06-11","label":"SSRN","topic":"LLMs & Text","cites":null,"score":5,"scale":"shares"},{"title":"FAIR Framework for Financial Systems","url":"/papers/ssrn/5285784/","summary":"The FAIR framework is expanded to tackle temporal challenges in financial operations, offering guidelines for financial institutions using Large Language Models and autonomous systems.","featured":"2025-06-11","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"EDINET-Bench: Evaluating LLMs on Complex Financial Tasks using Japanese Financial Statements","url":"/papers/arxiv/2506.08762/","summary":"EDINET-Bench, a new open-source Japanese financial benchmark, is introduced to assess large language models' performance in financial tasks, showing these models' limitations in real-world financial applications.","featured":"2025-06-11","label":"arXiv","topic":"LLMs & Text","cites":9,"score":25,"scale":"shares"},{"title":"Interpretable LLMs for Credit Risk: A Systematic Review and Taxonomy","url":"/papers/arxiv/2506.04290/","summary":"A review of Large Language Models in credit risk estimation provides a classification of model structures, data types, and application areas to guide future AI and finance research.","featured":"2025-06-11","label":"arXiv","topic":"LLMs & Text","cites":16,"score":15,"scale":"shares"},{"title":"Unanswerability Evaluation for Retrieval Augmented Generation","url":"/papers/arxiv/2412.12300/","summary":"The article introduces UAEval4RAG, a framework for evaluating the ability of retrieval-augmented generation (RAG) systems to handle unanswerable queries, emphasizing the role of component selection and prompt design.","featured":"2025-06-11","label":"Machine learning","topic":"LLMs & Text","cites":10,"score":30,"scale":"shares"},{"title":"Market News Attention Hype Index","url":"/papers/ssrn/5279231/","summary":"The Hype Index is presented as a measure to quantify media attention towards large-cap equities, using Natural Language Processing to extract predictive signals from financial news.","featured":"2025-06-04","label":"SSRN","topic":"LLMs & Text","cites":null,"score":3,"scale":"shares"},{"title":"Optimising Large Language Models","url":"/papers/ssrn/5278456/","summary":"The article categorises and reviews the optimisation strategies used in Large Language Models like ChatGPT, Claude LlaMA, and DeepSeek.","featured":"2025-06-04","label":"SSRN","topic":"LLMs & Text","cites":null,"score":16,"scale":"shares"},{"title":"Fair Document Valuation in LLM Summaries via Shapley Values","url":"/papers/arxiv/2505.23842/","summary":"The paper suggests using Shapley values to assess the worth of individual documents in Large Language Model-generated summaries, and introduces an efficient algorithm, Cluster Shapley, to reduce computation while maintaining quality.","featured":"2025-06-04","label":"arXiv","topic":"LLMs & Text","cites":9,"score":25,"scale":"shares"},{"title":"Fine-Tuning Large Language Models for Financial Markets via Ontological Reasoning","url":"/papers/ssrn/5274196/","summary":"Large Language Models struggle with accuracy in specialized fields due to lack of specific knowledge in training data, a problem that can be solved by fine-tuning with domain-specific data.","featured":"2025-05-30","label":"SSRN","topic":"LLMs & Text","cites":1,"score":3,"scale":"shares"},{"title":"Interpretable Machine Learning for Macro Alpha: A News Sentiment Case Study","url":"/papers/arxiv/2505.16136/","summary":"A new machine learning model uses global news sentiment to predict next-day returns for financial instruments, with sentiment dispersion and article impact being key predictive features.","featured":"2025-05-30","label":"arXiv","topic":"LLMs & Text","cites":2,"score":19,"scale":"shares"},{"title":"Humans expect rationality and cooperation from LLM opponents in strategic games","url":"/papers/arxiv/2505.11011/","summary":"A study found that humans tend to choose lower numbers when playing strategic games against Large Language Models (LLMs), influenced by those with high strategic reasoning skills and their perceptions of LLM's reasoning and cooperation abilities.","featured":"2025-05-21","label":"arXiv","topic":"LLMs & Text","cites":3,"score":19,"scale":"shares"},{"title":"Customer Review Sentiment Tool","url":"/papers/ssrn/5249094/","summary":"A new web-based tool uses transformer-based models to analyze customer review sentiment and create summaries, paving the way for future sentiment analysis tool improvements.","featured":"2025-05-14","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Can LLM-based Financial Investing Strategies Outperform the Market in Long Run?","url":"/papers/arxiv/2505.07078/","summary":"FINSABER, a backtesting framework, shows that Large Language Models' effectiveness in stock trading decreases over longer periods and larger symbol universes, emphasizing the need for trend detection and risk controls.","featured":"2025-05-14","label":"arXiv","topic":"LLMs & Text","cites":26,"score":19,"scale":"shares"},{"title":"NewsNet-SDF: Stochastic Discount Factor Estimation with Pretrained Language Model News Embeddings via Adversarial Networks","url":"/papers/arxiv/2505.06864/","summary":"Discount Factor Estimation: NewsNet-SDF, a deep learning framework, successfully integrates pretrained language model embeddings with financial time series for asset pricing and risk assessment, outperforming traditional models.","featured":"2025-05-14","label":"arXiv","topic":"LLMs & Text","cites":3,"score":18,"scale":"shares"},{"title":"Revealing economic facts: LLMs know more than they say","url":"/papers/arxiv/2505.08662/","summary":"The research investigates the use of hidden states in large language models to estimate and fill in economic and financial statistics, showing that a simple linear model trained on these hidden states performs better than the models' text outputs and requires minimal labelled examples for training.","featured":"2025-05-14","label":"arXiv","topic":"LLMs & Text","cites":3,"score":11,"scale":"shares"},{"title":"Multiscale Causal Analysis of Market Efficiency via News Uncertainty Networks and the Financial Chaos Index","url":"/papers/arxiv/2505.01543/","summary":"The study uses the Financial Chaos Index to assess stock market efficiency, finding that daily asset price changes respond predictably to lagged news-based uncertainty, but not monthly, emphasizing the importance of time-scale decomposition.","featured":"2025-05-07","label":"arXiv","topic":"LLMs & Text","cites":1,"score":17,"scale":"shares"},{"title":"NLP Models and Simulations","url":"/papers/ssrn/5230258/","summary":"The article discusses how Natural Language Processing can help develop a simulation model for inventory management systems, improving upon traditional methods.","featured":"2025-04-30","label":"SSRN","topic":"LLMs & Text","cites":null,"score":7,"scale":"shares"},{"title":"Monetary Policy Sentiment and Risk Dynamics","url":"/papers/ssrn/5233873/","summary":"A model showing the interaction of monetary policy signals and market sentiment in driving financial tail risk is developed, indicating that aligned signals reduce risk and misaligned ones increase it.","featured":"2025-04-30","label":"SSRN","topic":"LLMs & Text","cites":null,"score":6,"scale":"shares"},{"title":"Visualizing Public Opinion on X: A Real-Time Sentiment Dashboard Using VADER and DistilBERT","url":"/papers/arxiv/2504.15448/","summary":"A sentiment analysis system, using Natural Language Processing and machine learning, offers real-time interpretation of public opinion towards corporations, highlighting disparities in public sentiment and assisting stakeholders in strategic decision-making.","featured":"2025-04-30","label":"arXiv","topic":"LLMs & Text","cites":1,"score":17,"scale":"shares"},{"title":"Adapt-Pruner: Adaptive Structural Pruning for Efficient Small Language Model Training","url":"/papers/arxiv/2502.03460/","summary":"The research explores ways to speed up small language models, discovering that layer-wise adaptive pruning (Adapt-Pruner) is effective in large language models and outperforms existing pruning methods.","featured":"2025-04-30","label":"Machine learning","topic":"LLMs & Text","cites":10,"score":9,"scale":"shares"},{"title":"Schema-Guided Scene-Graph Reasoning Based on Multi-Agent Large Language Model System","url":"/papers/arxiv/2502.03450/","summary":"The paper introduces SG-RwR, a new framework for reasoning and planning with scene graphs, using two large language model agents to generate task plans and information queries.","featured":"2025-04-30","label":"Machine learning","topic":"LLMs & Text","cites":6,"score":7,"scale":"shares"},{"title":"SKI Models: Skeleton Induced Vision-Language Embeddings for Understanding Activities of Daily Living","url":"/papers/arxiv/2502.03459/","summary":"The study presents SKI models, which incorporate 3D skeletons into the vision-language embedding space, using a skeleton-language model to enhance Vision Language Models and Large Vision Language Models.","featured":"2025-04-30","label":"Machine learning","topic":"LLMs & Text","cites":6,"score":3,"scale":"shares"},{"title":"The Musk Effect on Tesla","url":"/papers/ssrn/5224386/","summary":"Investor sentiment, driven by social media and news, can cause significant short-term stock price fluctuations, as shown in a Tesla Inc. case study.","featured":"2025-04-23","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"The Memorization Problem: Can We Trust LLMs'Economic Forecasts?","url":"/papers/arxiv/2504.14765/","summary":"The study shows that large language models can remember exact economic figures from before their knowledge cutoff dates, which may skew their predictive abilities in forecasting and backtesting trading strategies.","featured":"2025-04-23","label":"arXiv","topic":"LLMs & Text","cites":36,"score":16,"scale":"shares"},{"title":"LLM-Enhanced Black-Litterman Portfolio Optimization","url":"/papers/arxiv/2504.14345/","summary":"The research investigates the use of large language models for estimating expected stock returns and integrating uncertainty into portfolio optimization via the Black-Litterman framework, comparing the effectiveness of various models.","featured":"2025-04-23","label":"arXiv","topic":"LLMs & Text","cites":9,"score":12,"scale":"shares"},{"title":"Balancing Engagement and Polarization: Multi-Objective Alignment of News Content Using LLMs","url":"/papers/arxiv/2504.13444/","summary":"Researchers have created a language model that produces engaging news while controlling polarization levels, addressing issues of bias in language models.","featured":"2025-04-23","label":"arXiv","topic":"LLMs & Text","cites":2,"score":14,"scale":"shares"},{"title":"Divergent LLM Adoption and Heterogeneous Convergence Paths in Research Writing","url":"/papers/arxiv/2504.13629/","summary":"The research examines the effects of AI-assisted revisions on academic writing, showing differences in usage across fields, gender, and career level, and demonstrating that Large Language Models improve writing clarity and brevity.","featured":"2025-04-23","label":"arXiv","topic":"LLMs & Text","cites":6,"score":20,"scale":"shares"},{"title":"Trust, but Verify","url":"/papers/arxiv/2504.13443/","summary":"The article explores the identification of nodes operating unauthorized or incorrect Large Language Models in a decentralized AI network through peer consensus, and proposes a validation system with financial rewards and penalties to promote honesty.","featured":"2025-04-23","label":"arXiv","topic":"LLMs & Text","cites":0,"score":17,"scale":"shares"},{"title":"Computational Basis of LLM's Decision Making in Social Simulation","url":"/papers/arxiv/2504.11671/","summary":"The study investigates the influence of character and context on the behavior of large language models in social science scenarios, suggesting ways to examine and adjust an LLM's internal representations in a Dictator Game.","featured":"2025-04-23","label":"arXiv","topic":"LLMs & Text","cites":2,"score":17,"scale":"shares"},{"title":"Are Language Models Up to Sequential Optimization Problems? From Evaluation to a Hegelian-Inspired Enhancement","url":"/papers/arxiv/2502.02573/","summary":"The paper investigates the ability of Large Language Models in managing Sequential Optimization Problems, introducing WorldGen for generating new SOPs, and suggesting ACE to enhance LLM performance without additional training.","featured":"2025-04-23","label":"Machine learning","topic":"LLMs & Text","cites":1,"score":5,"scale":"shares"},{"title":"Ethical Sentiment in Drug Pricing","url":"/papers/ssrn/5218044/","summary":"The article explores the influence of public sentiment on companies' financial performance after ethically dubious actions, indicating that negative sentiment can both help and hinder performance.","featured":"2025-04-16","label":"SSRN","topic":"LLMs & Text","cites":null,"score":3,"scale":"shares"},{"title":"Media Bias in Stocks","url":"/papers/ssrn/5216036/","summary":"The study reveals that national stock market indices perform poorly when influenced by daily media coverage, particularly negative news and significant index changes.","featured":"2025-04-16","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Cybersecurity Incidents & Stock Market Valuation: Japan vs Korea","url":"/papers/ssrn/5202602/","summary":"Japan vs Korea: The study finds that Korea's stock market is more susceptible to cyber attacks and AI-driven sentiment than Japan's.","featured":"2025-04-16","label":"SSRN","topic":"LLMs & Text","cites":null,"score":8,"scale":"shares"},{"title":"Can Large Language Models Trade? Testing Financial Theories with LLM Agents in Market Simulations","url":"/papers/arxiv/2504.10789/","summary":"The article presents an open-source framework for a simulated stock market, where large language models act as trading agents, allowing for analysis of market dynamics under different conditions.","featured":"2025-04-16","label":"arXiv","topic":"LLMs & Text","cites":26,"score":16,"scale":"shares"},{"title":"Who is More Bayesian: Humans or ChatGPT?","url":"/papers/arxiv/2504.10636/","summary":"A study reveals that the latest versions of AI outperform humans in decision-making, showing nearly perfect Bayesian classifications, unlike humans and early AI versions that exhibit judgement biases.","featured":"2025-04-16","label":"arXiv","topic":"LLMs & Text","cites":2,"score":26,"scale":"shares"},{"title":"DeepGreen: Effective LLM-Driven Green-washing Monitoring System Designed for Empirical Testing - Evidence from China","url":"/papers/arxiv/2504.07733/","summary":"DeepGreen is a system that uses large language models to identify green-washing in corporations by analyzing their financial statements, showing that green practices can boost a company's asset return rate.","featured":"2025-04-16","label":"arXiv","topic":"LLMs & Text","cites":7,"score":16,"scale":"shares"},{"title":"EthosGPT: Mapping Human Value Diversity to Advance Sustainable Development Goals (SDGs)","url":"/papers/arxiv/2504.09861/","summary":"EthosGPT is an open-source framework that uses large language models to assess and map human values globally, aiding in the creation of inclusive AI systems and promoting value diversity in accordance with the United Nations Sustainable Development Goals.","featured":"2025-04-16","label":"arXiv","topic":"LLMs & Text","cites":0,"score":16,"scale":"shares"},{"title":"Auctions & Abnormal Returns","url":"/papers/ssrn/5207418/","summary":"Foreign exchange returns increase when U.S. macroeconomic news releases are preceded by Treasury auctions due to decreased Treasury demand and limited risk absorption by primary dealers.","featured":"2025-04-09","label":"SSRN","topic":"LLMs & Text","cites":null,"score":4,"scale":"shares"},{"title":"Visketch-Gpt: Multi-Scale Sketch Recognition","url":"/papers/ssrn/5192605/","summary":"Multi-Scale Sketch Recognition: ViSketchGPT, a new algorithm, enhances the recognition and generation of detailed human sketches.","featured":"2025-04-09","label":"SSRN","topic":"LLMs & Text","cites":null,"score":4,"scale":"shares"},{"title":"GPT Adoption and the Impact of Disclosure Policies","url":"/papers/arxiv/2504.01566/","summary":"The study finds that while revealing the use of AI tools like ChatGPT in professional roles reduces information gaps, it doesn't significantly cut agency costs as managers undervalue analysts' input with AI use.","featured":"2025-04-09","label":"arXiv","topic":"LLMs & Text","cites":0,"score":20,"scale":"shares"},{"title":"Do Large Language Model Benchmarks Test Reliability?","url":"/papers/arxiv/2502.03461/","summary":"The article highlights the need for reliable large language models, criticizes current benchmarks for their inadequacy, and suggests the use of platinum benchmarks to reduce label errors and ambiguity.","featured":"2025-04-09","label":"Machine learning","topic":"LLMs & Text","cites":54,"score":73,"scale":"shares"},{"title":"BFS-Prover: Scalable Best-First Tree Search for LLM-based Automatic Theorem Proving","url":"/papers/arxiv/2502.03438/","summary":"Scalable Best-First Tree Search: BFS-Prover is a scalable framework that uses Best-First Tree Search for automatic theorem proving, challenging the need for complex tree search methods.","featured":"2025-04-09","label":"Machine learning","topic":"LLMs & Text","cites":92,"score":28,"scale":"shares"},{"title":"Rankify: A Comprehensive Python Toolkit for Retrieval, Re-Ranking, and Retrieval-Augmented Generation","url":"/papers/arxiv/2502.02464/","summary":"Python Toolkit for Retrieval and Generation: Rankify is an open-source toolkit designed to unify retrieval, re-ranking, and retrieval-augmented generation, improving consistency and scalability in information retrieval research.","featured":"2025-04-09","label":"Machine learning","topic":"LLMs & Text","cites":11,"score":24,"scale":"shares"},{"title":"New Sentiment Analysis for JP Stock Market","url":"/papers/ssrn/5195681/","summary":"A new sentiment analysis method for investment decisions in the Japanese stock market is presented, using financial news keywords and market returns, outperforming a ChatGPT-based approach.","featured":"2025-04-02","label":"SSRN","topic":"LLMs & Text","cites":null,"score":3,"scale":"shares"},{"title":"Customer Review Sentiment Analysis with Hybrid Models","url":"/papers/ssrn/5195674/","summary":"The paper suggests a hybrid sentiment analysis framework for processing online customer reviews in real time, with BERT providing the highest accuracy but at a higher computational cost.","featured":"2025-04-02","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"The News in Earnings Announcement Disclosures: Capturing Word Context Using LLM Methods","url":"/papers/ssrn/5198675/","summary":"The paper reveals that textual disclosures in companies' earnings announcements, analyzed through a large language model, account for a significant portion of stock return variations and immediate price revisions.","featured":"2025-04-02","label":"SSRN","topic":"LLMs & Text","cites":40,"score":3,"scale":"shares"},{"title":"From Deep Learning to LLMs: A survey of AI in Quantitative Investment","url":"/papers/arxiv/2503.21422/","summary":"The article explores how artificial intelligence, particularly deep learning and large language models, enhances predictive modeling and automation in quantitative investment.","featured":"2025-04-02","label":"arXiv","topic":"LLMs & Text","cites":26,"score":35,"scale":"shares"},{"title":"InfoBid: A Simulation Framework for Studying Information Disclosure in Auctions with Large Language Model-based Agents","url":"/papers/arxiv/2503.22726/","summary":"Auction Simulation with LLM Agents: The article presents InfoBid, a simulation framework that uses large language models to analyze the impact of information disclosure strategies in online advertising auctions. It helps understand strategic behavior and auction results.","featured":"2025-04-02","label":"arXiv","topic":"LLMs & Text","cites":4,"score":20,"scale":"shares"},{"title":"Social Media Sentiment Signals","url":"/papers/ssrn/5187350/","summary":"The study creates daily market sentiment and attention indexes from social media posts, indicating that sentiment extrapolates from past returns and attention predicts negative returns.","featured":"2025-03-26","label":"SSRN","topic":"LLMs & Text","cites":null,"score":3,"scale":"shares"},{"title":"Assessing Consistency and Reproducibility in the Outputs of Large Language Models: Evidence Across Diverse Finance and Accounting Tasks","url":"/papers/arxiv/2503.16974/","summary":"The study shows Large Language Models are consistent and reproducible in finance and accounting research, even outperforming human experts.","featured":"2025-03-26","label":"arXiv","topic":"LLMs & Text","cites":42,"score":35,"scale":"shares"},{"title":"Financial Wind Tunnel: A Retrieval-Augmented Market Simulator","url":"/papers/arxiv/2503.17909/","summary":"Market Simulator: The article introduces a new market simulator named Financial Wind Tunnel (FWT). This tool creates adaptable market dynamics for model testing, enhancing the performance and adaptability of models, especially in unstable market situations.","featured":"2025-03-26","label":"arXiv","topic":"LLMs & Text","cites":2,"score":14,"scale":"shares"},{"title":"Enhanced Financial Sentiment Analysis","url":"/papers/ssrn/5181105/","summary":"A new methodology for financial sentiment analysis using large language models is proposed in a study, with the GPT-3-based OPT model outperforming others in predicting stock market returns.","featured":"2025-03-20","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Towards Temporal-Aware Multi-Modal Retrieval Augemented Generation in Finance","url":"/papers/arxiv/2503.05185/","summary":"FinTMMBench is a new benchmark for assessing multi-modal Retrieval-Augmented Generation systems in finance, providing a multi-modal corpus, time-aware questions, and various financial analysis tasks.","featured":"2025-03-12","label":"arXiv","topic":"LLMs & Text","cites":6,"score":17,"scale":"shares"},{"title":"Unveiling Biases in AI: ChatGPT's Political Economy Perspectives and Human Comparisons","url":"/papers/arxiv/2503.05234/","summary":"Research indicates that ChatGPT, an AI system, shows a left-leaning bias in responses to political economy questions, emphasizing the need for AI transparency to avoid ideological influence.","featured":"2025-03-12","label":"arXiv","topic":"LLMs & Text","cites":4,"score":12,"scale":"shares"},{"title":"Maximum Hallucination Standards for Domain-Specific Large Language Models","url":"/papers/arxiv/2503.05481/","summary":"A study on large language models (LLMs) suggests that net welfare improves when the acceptable level of inaccuracies varies with the willingness to pay for reduced misinformation and the damage associated with it.","featured":"2025-03-12","label":"arXiv","topic":"LLMs & Text","cites":2,"score":11,"scale":"shares"},{"title":"Are Large Language Models Good In-context Learners for Financial Sentiment Analysis?","url":"/papers/arxiv/2503.04873/","summary":"The article investigates the use of large language models in financial sentiment analysis, showing their ability to learn and address related challenges.","featured":"2025-03-12","label":"arXiv","topic":"LLMs & Text","cites":8,"score":24,"scale":"shares"},{"title":"Large Language Models in Token Space","url":"/papers/ssrn/5140817/","summary":"The paper presents a framework that treats large language models as reinforcement learning agents in token space, offering theoretical insights for more effective language models.","featured":"2025-03-05","label":"SSRN","topic":"LLMs & Text","cites":null,"score":81,"scale":"shares"},{"title":"Sentiment Beta in China","url":"/papers/ssrn/5147942/","summary":"In China, positive sentiment beta significantly predicts stock returns, while negative sentiment beta has a negligible effect.","featured":"2025-03-05","label":"SSRN","topic":"LLMs & Text","cites":null,"score":24,"scale":"shares"},{"title":"Financial Accounting Focus for Journalists","url":"/papers/ssrn/5145051/","summary":"Financial accounting-focused news articles result in higher trading volume and abnormal returns, particularly when stock prices are volatile and analyst forecasts vary.","featured":"2025-03-05","label":"SSRN","topic":"LLMs & Text","cites":null,"score":53,"scale":"shares"},{"title":"Financial Market Sentiment Analysis","url":"/papers/ssrn/5145647/","summary":"Research using FinBERT language models in a RAG pipeline shows that market sentiment analysis can enhance decision-making processes when combined with other financial indicators, despite its limited power in predicting next-day stock prices.","featured":"2025-03-05","label":"SSRN","topic":"LLMs & Text","cites":null,"score":32,"scale":"shares"},{"title":"Shifting Power: Leveraging LLMs to Simulate Human Aversion in ABMs of Bilateral Financial Exchanges, A bond market study","url":"/papers/arxiv/2503.00320/","summary":"The TRIBE model simulates human-like decision-making in trading, showing that large language models can enhance client agency and reveal new market behaviors.","featured":"2025-03-05","label":"arXiv","topic":"LLMs & Text","cites":2,"score":18,"scale":"shares"},{"title":"Retrieval Augmented Generation for Topic Modeling in Organizational Research: An Introduction with Empirical Demonstration","url":"/papers/arxiv/2502.20963/","summary":"The article presents Agentic RAG, a new method for topic modeling with large language models, offering a more efficient and reliable alternative for AI-based qualitative research.","featured":"2025-03-05","label":"arXiv","topic":"LLMs & Text","cites":6,"score":17,"scale":"shares"},{"title":"The amplifier effect of artificial agents in social contagion","url":"/papers/arxiv/2502.21037/","summary":"The study shows that artificial agents, powered by large language models, can speed up social contagion due to their lower adoption thresholds, potentially accelerating societal behavioral changes.","featured":"2025-03-05","label":"arXiv","topic":"LLMs & Text","cites":2,"score":17,"scale":"shares"},{"title":"Advanced Deep Learning Techniques for Analyzing Earnings Call Transcripts: Methodologies and Applications","url":"/papers/arxiv/2503.01886/","summary":"The paper compares the effectiveness of deep learning methods like BERT, FinBERT, and ULMFiT in sentiment analysis of financial transcripts, offering insights for practical financial decision-making.","featured":"2025-03-05","label":"arXiv","topic":"LLMs & Text","cites":2,"score":13,"scale":"shares"},{"title":"News Sentiment and Investment Risk","url":"/papers/repec/eee-ecolet-v-247-y-2025-i-c-s0165176524006086/","summary":"The research reassesses the effect of news sentiment on stock return volatility, finding that both positive and negative firm-specific and macroeconomic news significantly impact intraday stock return volatility, with GPT-4 potentially outperforming RavenPack in classification accuracy.","featured":"2025-03-05","label":"RePEc","topic":"LLMs & Text","cites":null,"score":16,"scale":"shares"},{"title":"Predicting Liquidity-Aware Bond Yields using Causal GANs and Deep Reinforcement Learning with LLM Evaluation","url":"/papers/arxiv/2502.17011/","summary":"The paper introduces a new method for predicting bond yields using Causal Generative Adversarial Networks and reinforcement learning, which improves forecasting performance by generating synthetic bond yield data.","featured":"2025-02-26","label":"arXiv","topic":"LLMs & Text","cites":2,"score":15,"scale":"shares"},{"title":"A Multi-LLM-Agent-Based Framework for Economic and Public Policy Analysis","url":"/papers/arxiv/2502.16879/","summary":"The research presents a unique method for economic and public policy analysis using multiple large language models as artificial economic agents, simulating policy impacts across various groups, and suggesting a new way to utilize computational power and human-like reasoning in policy studies.","featured":"2025-02-26","label":"arXiv","topic":"LLMs & Text","cites":13,"score":10,"scale":"shares"},{"title":"Scaling Political Actors with Embedding Representations","url":"/papers/ssrn/5132792/","summary":"A novel method using natural language processing enables a more comprehensive scaling of lawmakers and their parties, covering a wider array of issues and political theories, such as views on the EU and party populism.","featured":"2025-02-19","label":"SSRN","topic":"LLMs & Text","cites":null,"score":9,"scale":"shares"},{"title":"Private Firm News Disclosure Effects","url":"/papers/ssrn/5127833/","summary":"The research reveals that voluntary news disclosure by private firms increases the investment sensitivities of public peer firms, especially in volatile industries with less local newspaper coverage.","featured":"2025-02-19","label":"SSRN","topic":"LLMs & Text","cites":null,"score":87,"scale":"shares"},{"title":"ChatGPT and Deepseek: Can They Predict the Stock Market and Macroeconomy?","url":"/papers/arxiv/2502.10008/","summary":"The research shows that ChatGPT, a language model, can predict stock market and macroeconomic trends more accurately than other models like DeepSeek by analyzing information from the Wall Street Journal.","featured":"2025-02-19","label":"arXiv","topic":"LLMs & Text","cites":18,"score":11,"scale":"shares"},{"title":"FLAG-Trader: Fusion LLM-Agent with Gradient-based Reinforcement Learning for Financial Trading","url":"/papers/arxiv/2502.11433/","summary":"Fusion LLM-Agent: FLAG-Trader, a new architecture combining linguistic processing and reinforcement learning, has been proposed to enhance decision-making in interactive financial markets.","featured":"2025-02-19","label":"arXiv","topic":"LLMs & Text","cites":33,"score":5,"scale":"shares"},{"title":"FinRL-DeepSeek: LLM-Infused Risk-Sensitive Reinforcement Learning for Trading Agents","url":"/papers/arxiv/2502.07393/","summary":"The article introduces a trading agent that uses reinforcement learning and language models to analyze financial news and make risk-sensitive trading recommendations, tested on the Nasdaq-100 index.","featured":"2025-02-19","label":"arXiv","topic":"LLMs & Text","cites":22,"score":30,"scale":"shares"},{"title":"MarketSenseAI 2.0: Enhancing Stock Analysis Through LLM Agents","url":"/papers/arxiv/2502.00415/","summary":"Stock Analysis: MarketSenseAI, a stock analysis tool, uses Large Language Models to analyze financial news, improving analysis accuracy and outperforming the market index.","featured":"2025-02-05","label":"arXiv","topic":"LLMs & Text","cites":20,"score":9,"scale":"shares"},{"title":"An End-To-End LLM Enhanced Trading System","url":"/papers/arxiv/2502.01574/","summary":"The project presents a trading system that uses Large Language Models to analyze market sentiment in real-time, using data from financial news and social media to create trading signals.","featured":"2025-02-05","label":"arXiv","topic":"LLMs & Text","cites":3,"score":5,"scale":"shares"},{"title":"Decision-informed Neural Networks with Large Language Model Integration for Portfolio Optimization","url":"/papers/arxiv/2502.00828/","summary":"The paper combines Large Language Models with decision-focused learning to improve prediction and decision quality in portfolio optimization, outperforming other deep learning models.","featured":"2025-02-05","label":"arXiv","topic":"LLMs & Text","cites":19,"score":4,"scale":"shares"},{"title":"Can AI Solve the Peer Review Crisis? A Large Scale Cross Model Experiment of LLMs' Performance and Biases in Evaluating over 1000 Economics Papers","url":"/papers/arxiv/2502.00070/","summary":"A study using a large language model to analyze economics peer reviews found that while it can identify paper quality, it shows biases and struggles to differentiate high-quality AI-generated papers, suggesting a need for careful integration and mixed peer review models.","featured":"2025-02-05","label":"arXiv","topic":"LLMs & Text","cites":8,"score":10,"scale":"shares"},{"title":"s1: Simple test-time scaling","url":"/papers/arxiv/2501.19393/","summary":"The research presents a method called budget forcing, which uses a small dataset to achieve test-time scaling and improved reasoning performance in language modeling, particularly in competition math questions.","featured":"2025-02-05","label":"Machine learning","topic":"LLMs & Text","cites":1462,"score":220,"scale":"shares"},{"title":"Diverse Preference Optimization","url":"/papers/arxiv/2501.18101/","summary":"The paper introduces Diverse Preference Optimization (DivPO), an optimization method that generates diverse responses in language models post-training, enhancing diversity in persona attributes and story generation.","featured":"2025-02-05","label":"Machine learning","topic":"LLMs & Text","cites":47,"score":204,"scale":"shares"},{"title":"Scalable-Softmax Is Superior for Attention","url":"/papers/arxiv/2501.19399/","summary":"The study proposes Scalable-Softmax (SSMax), a replacement for Softmax in language models, which improves performance in long contexts and key information retrieval, and allows better focus on key information.","featured":"2025-02-05","label":"Machine learning","topic":"LLMs & Text","cites":47,"score":108,"scale":"shares"},{"title":"Thoughts Are All Over the Place: On the Underthinking of o1-Like LLMs","url":"/papers/arxiv/2501.18585/","summary":"The research identifies underthinking in large language models, where models frequently switch reasoning thoughts, and proposes a decoding strategy to encourage deeper exploration of each reasoning path, improving accuracy across challenging datasets.","featured":"2025-02-05","label":"Machine learning","topic":"LLMs & Text","cites":171,"score":105,"scale":"shares"},{"title":"Decoding-based Regression","url":"/papers/arxiv/2501.19383/","summary":"Research indicates that language models capable of numeric predictions as decoded strings perform as well as traditional methods for tabular regression tasks.","featured":"2025-02-05","label":"Machine learning","topic":"LLMs & Text","cites":11,"score":13,"scale":"shares"},{"title":"LLMs Are In-Context Bandit Reinforcement Learners","url":"/papers/arxiv/2410.05362/","summary":"The research investigates the use of Large Language Models in in-context reinforcement learning, showing their effectiveness in learning from rewards but also their limitations in error reasoning.","featured":"2025-02-05","label":"Machine learning","topic":"LLMs & Text","cites":28,"score":228,"scale":"shares"},{"title":"TÜLU 3: Pushing Frontiers in Open Language Model Post-Training","url":"/papers/arxiv/2411.15124/","summary":"Open Language Model Post-Training: The Tulu 3 model, a top-tier post-trained language model, is introduced, outperforming other models and providing a detailed guide for its use and adaptation.","featured":"2025-02-05","label":"Machine learning","topic":"LLMs & Text","cites":888,"score":227,"scale":"shares"},{"title":"Optimizing Large Language Model Training Using FP4 Quantization","url":"/papers/arxiv/2501.17116/","summary":"The research presents the first FP4 training framework for large language models (LLMs), using low-bit arithmetic operations to lessen computational demands, achieving similar accuracy to BF16 and FP8 with slight degradation.","featured":"2025-02-05","label":"Machine learning","topic":"LLMs & Text","cites":61,"score":52,"scale":"shares"},{"title":"News Sentiment Analysis Limitations","url":"/papers/ssrn/5086825/","summary":"The paper explores the difficulties of using news sentiment analysis to predict next-day stock returns, proposing a multilevel approach that incorporates sentiment analysis across individual stocks, industries, and the overall economy.","featured":"2025-01-23","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Open Sourcing GPTs: Economics of Open Sourcing Advanced AI Models","url":"/papers/arxiv/2501.11581/","summary":"The article explores the economic reasons behind for-profit companies open sourcing their large language models, highlighting a balance between technological advancement and immediate profit.","featured":"2025-01-23","label":"arXiv","topic":"LLMs & Text","cites":3,"score":8,"scale":"shares"},{"title":"Towards Large Reasoning Models: A Survey of Reinforced Reasoning with Large Language Models","url":"/papers/arxiv/2501.09686/","summary":"The article discusses advancements in Large Language Models (LLMs) reasoning, emphasizing the use of reinforcement learning and thought simulation for complex reasoning, and the potential of scaling during training and testing.","featured":"2025-01-23","label":"Machine learning","topic":"LLMs & Text","cites":250,"score":107,"scale":"shares"},{"title":"OmniThink: Expanding Knowledge Boundaries in Machine Writing through Thinking","url":"/papers/arxiv/2501.09751/","summary":"Machine Writing Expansion: OmniThink, a machine writing framework that mimics learner cognition, is introduced to improve the knowledge density of machine-written articles, addressing the limitations of retrieval-augmented generation.","featured":"2025-01-23","label":"Machine learning","topic":"LLMs & Text","cites":26,"score":52,"scale":"shares"},{"title":"Suggesting Code Edits in Interactive Machine Learning Notebooks Using Large Language Models","url":"/papers/arxiv/2501.09745/","summary":"A study using a dataset of over 48,000 Jupyter notebook edits from GitHub reveals the complexity of machine learning maintenance tasks and the potential of large language models in predicting code edits.","featured":"2025-01-23","label":"Machine learning","topic":"LLMs & Text","cites":3,"score":30,"scale":"shares"},{"title":"DataDriven Inventory Control","url":"/papers/repec/eee-ejores-v-322-y-2025-i-1-p-254-269/","summary":"The paper suggests a Prescriptive Analytics approach for data-driven dynamic inventory control of large product portfolios, using a 'global learning' model that outperforms 'local learning' strategies, and highlights the importance of contextual information.","featured":"2025-01-23","label":"RePEc","topic":"LLMs & Text","cites":null,"score":11,"scale":"shares"},{"title":"LLMs Model Non-WEIRD Populations: Experiments with Synthetic Cultural Agents","url":"/papers/arxiv/2501.06834/","summary":"A novel method has been introduced by researchers that uses Large Language Models to simulate economic behavior across different cultures, providing a new tool for economic and behavioral research.","featured":"2025-01-15","label":"arXiv","topic":"LLMs & Text","cites":6,"score":9,"scale":"shares"},{"title":"Revisiting Group Differences in High‐Dimensional Choices: Method and Application to Congressional Speech","url":"/papers/arxiv/2206.10877/","summary":"The study uses an advanced language model to confirm the rise in partisanship in U.S. congressional speech since the 1990s, also revealing changes in topic content and partisan phrases.","featured":"2025-01-15","label":"arXiv","topic":"LLMs & Text","cites":5,"score":37,"scale":"shares"},{"title":"Utility-inspired Reward Transformations Improve Reinforcement Learning Training of Language Models","url":"/papers/arxiv/2501.06248/","summary":"The paper introduces a new method for training large language models using reinforcement learning feedback, which makes models more beneficial and less harmful by adjusting sensitivity to reward values.","featured":"2025-01-15","label":"arXiv","topic":"LLMs & Text","cites":2,"score":17,"scale":"shares"},{"title":"ECB Press Conference Sentiment","url":"/papers/repec/wly-ijfiec-v-30-y-2025-i-1-p-652-664/","summary":"The research uses FinBERT to analyze sentiment in ECB president's introductory statements, finding that the sentiment about monetary policy significantly affects subsequent press conference content.","featured":"2025-01-15","label":"RePEc","topic":"LLMs & Text","cites":null,"score":3,"scale":"shares"},{"title":"Metadata Conditioning Accelerates Language Model Pre-training","url":"/papers/arxiv/2501.01956/","summary":"The MeCo method speeds up language model pre-training by using additional learning cues, allowing the model to work without metadata and enhancing task performance.","featured":"2025-01-08","label":"Machine learning","topic":"LLMs & Text","cites":22,"score":114,"scale":"shares"},{"title":"VITA-1.5: Towards GPT-4o Level Real-Time Vision and Speech Interaction","url":"/papers/arxiv/2501.01957/","summary":"Vision and Speech Interaction: A proposed training methodology enables Large Language Models to comprehend visual and speech data, improving speech-to-speech dialogue capabilities and response speed.","featured":"2025-01-08","label":"Machine learning","topic":"LLMs & Text","cites":230,"score":52,"scale":"shares"},{"title":"BoostStep: Boosting mathematical capability of Large Language Models via improved single-step reasoning","url":"/papers/arxiv/2501.03226/","summary":"The article introduces BoostStep, a method that enhances the reasoning quality within each step of large language models solving complex math problems, providing more relevant examples and integrating seamlessly with Monte Carlo Tree Search methods.","featured":"2025-01-08","label":"Machine learning","topic":"LLMs & Text","cites":19,"score":13,"scale":"shares"},{"title":"MEDEC: A Benchmark for Medical Error Detection and Correction in Clinical Notes","url":"/papers/arxiv/2412.19260/","summary":"The article discusses MEDEC, a benchmark for identifying and fixing medical errors in clinical notes, and reveals that while Large Language Models (LLMs) are effective, they are still not as accurate as medical doctors.","featured":"2025-01-08","label":"Machine learning","topic":"LLMs & Text","cites":140,"score":560,"scale":"shares"},{"title":"Accurate RNA 3D structure prediction using a language model-based deep learning approach","url":"/papers/arxiv/2207.01586/","summary":"The paper introduces RhoFold+, a deep learning method that accurately predicts 3D structures of single-chain RNAs from sequences, surpassing existing methods and aiding in RNA structure and function research.","featured":"2025-01-08","label":"Machine learning","topic":"LLMs & Text","cites":245,"score":86,"scale":"shares"},{"title":"EdgeRAG: Online-Indexed RAG for Edge Devices","url":"/papers/arxiv/2412.21023/","summary":"Online RAG: EdgeRAG is a system proposed for deploying Retrieval Augmented Generation on devices with limited resources, reducing latency and memory usage by pruning and generating embeddings as needed.","featured":"2025-01-08","label":"Machine learning","topic":"LLMs & Text","cites":27,"score":23,"scale":"shares"},{"title":"Gender Accuracy in Translation","url":"/papers/repec/jfr-wjel11-v-15-y-2025-i-1-p-9/","summary":"The study compares the gender accuracy in English-Arabic machine translation of two large language models, Gemini and ChatGPT, with Gemini performing better in handling gender-related translation issues.","featured":"2025-01-08","label":"RePEc","topic":"LLMs & Text","cites":null,"score":0,"scale":"shares"},{"title":"Decoding China's Industrial Policies","url":"/papers/ssrn/5078043/","summary":"Large Language Models are utilized to interpret China's industrial policies from 2000 to 2022, extracting structured data from 3 million government documents.","featured":"2025-01-01","label":"SSRN","topic":"LLMs & Text","cites":15,"score":31,"scale":"shares"},{"title":"A Systematic Literature Review on Sentimental Analysis using Machine Learning for Preprocessing, Feature Extraction and Classification","url":"/papers/ssrn/5061012/","summary":"The study reviews sentiment analysis methods in Natural Language Processing, covering data collection, preprocessing, feature extraction, and classification.","featured":"2025-01-01","label":"SSRN","topic":"LLMs & Text","cites":2,"score":64,"scale":"shares"},{"title":"Topic-Centric Sentiment Analysis with Machine Learning","url":"/papers/ssrn/5072186/","summary":"The article discusses the use of machine learning techniques like topic modeling and LDA for sentiment analysis, and their role in data-driven decision making.","featured":"2025-01-01","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Foreign EPUF and US Equity","url":"/papers/ssrn/5063859/","summary":"The article reveals that foreign economic policy uncertainty significantly predicts excess U.S. stock returns, primarily affecting equity prices through cash flow news.","featured":"2025-01-01","label":"SSRN","topic":"LLMs & Text","cites":null,"score":10,"scale":"shares"},{"title":"Sentiment trading with large language models","url":"/papers/doi/10-1016-j-frl-2024-105227/","summary":"The OPT model, a large language model, has proven superior in predicting stock market returns using sentiment analysis of U.S. financial news, outdoing traditional methods like the Loughran-McDonald dictionary model.","featured":"2025-01-01","label":"arXiv","topic":"LLMs & Text","cites":107,"score":46,"scale":"shares"},{"title":"TradingAgents: Multi-Agents LLM Financial Trading Framework","url":"/papers/arxiv/2412.20138/","summary":"TradingAgents, a new stock trading framework, employs large language models to simulate real-world trading dynamics, enhancing trading performance.","featured":"2025-01-01","label":"arXiv","topic":"LLMs & Text","cites":220,"score":13,"scale":"shares"},{"title":"Integrating Natural Language Processing Techniques of Text Mining Into Financial System: Applications and Limitations","url":"/papers/arxiv/2412.20438/","summary":"The paper discusses the application of text mining and natural language processing in finance, emphasizing the need to improve data quality and model understanding for better financial predictions.","featured":"2025-01-01","label":"arXiv","topic":"LLMs & Text","cites":1,"score":11,"scale":"shares"},{"title":"InfAlign: Inference-aware language model alignment","url":"/papers/arxiv/2412.19792/","summary":"The study introduces a new framework for language models that enhances inference-time decoding procedures, leading to significant improvements over previous methods.","featured":"2025-01-01","label":"Machine learning","topic":"LLMs & Text","cites":35,"score":18,"scale":"shares"},{"title":"Machine Learning for Sentiment Analysis of Imported Food in Trinidad and Tobago","url":"/papers/arxiv/2412.19781/","summary":"The research shows that the VADER machine learning algorithm performs best in sentiment analysis of Twitter data on imported food in Trinidad and Tobago.","featured":"2025-01-01","label":"Machine learning","topic":"LLMs & Text","cites":0,"score":12,"scale":"shares"},{"title":"Adaptive Batch Size Schedules for Distributed Training of Language Models with Data and Model Parallelism","url":"/papers/arxiv/2412.21124/","summary":"The article introduces a new adaptive batch size schedule for large-scale model training, which optimizes memory usage and performs better than constant batch sizes, especially in pretraining smaller models.","featured":"2025-01-01","label":"Machine learning","topic":"LLMs & Text","cites":5,"score":8,"scale":"shares"},{"title":"Long-Form Speech Generation with Spoken Language Models","url":"/papers/arxiv/2412.18603/","summary":"Google introduces SpeechSSM, a speech language model for generating long-form audio, along with new metrics and a benchmark for long-form speech processing and generation.","featured":"2025-01-01","label":"Machine learning","topic":"LLMs & Text","cites":27,"score":21,"scale":"shares"},{"title":"YuLan-Mini: An Open Data-efficient Language Model","url":"/papers/arxiv/2412.17743/","summary":"Efficient Language Model: YuLan-Mini, a 2.42B parameter base model, delivers top-tier performance among similar models through a sophisticated data pipeline, robust optimization method, and effective annealing approach.","featured":"2025-01-01","label":"Machine learning","topic":"LLMs & Text","cites":7,"score":11,"scale":"shares"},{"title":"Can LLMs Obfuscate Code? A Systematic Analysis of Large Language Models into Assembly Code Obfuscation","url":"/papers/arxiv/2412.16135/","summary":"The MetamorphASM benchmark is designed to assess Large Language Models' ability to generate and analyze obfuscated assembly code, potentially threatening anti-virus engines.","featured":"2025-01-01","label":"Machine learning","topic":"LLMs & Text","cites":18,"score":11,"scale":"shares"},{"title":"Consumer Segmentation with Language Models","url":"/papers/repec/eee-joreco-v-82-y-2025-i-c-s0969698924003746/","summary":"The research shows that Large Language Models (LLMs) can improve clustering accuracy in consumer segmentation for marketing research and can simulate consumer preferences.","featured":"2025-01-01","label":"RePEc","topic":"LLMs & Text","cites":null,"score":1,"scale":"shares"},{"title":"Chatbots for Political Information Verification","url":"/papers/repec/spr-jcsosc-v-8-y-2025-i-1-d-10-1007-s42001-024-00338-8/","summary":"The article finds that while both ChatGPT and Bing Chat can detect the truthfulness of political information, ChatGPT performs better and provides more nuanced responses across different languages.","featured":"2025-01-01","label":"RePEc","topic":"LLMs & Text","cites":null,"score":0,"scale":"shares"},{"title":"Delving into Youth Perspectives on In-game Gambling-like Elements: A Proof-of-Concept Study Utilising Large Language Models for Analysing User-Generated Text Data","url":"/papers/arxiv/2412.09345/","summary":"The report shows that Large Language Models can effectively identify patterns in digital games' gambling-like elements, but struggle with complex tasks.","featured":"2024-12-18","label":"arXiv","topic":"LLMs & Text","cites":1,"score":7,"scale":"shares"},{"title":"FinGPT: Enhancing Sentiment-Based Stock Movement Prediction with Dissemination-Aware and Context-Enriched LLMs","url":"/papers/arxiv/2412.10823/","summary":"A data-driven method that includes news spread, contextual data, and explicit instructions enhances the accuracy of large language models in predicting short-term stock price movements.","featured":"2024-12-18","label":"arXiv","topic":"LLMs & Text","cites":5,"score":10,"scale":"shares"},{"title":"Integrative Analysis of Financial Market Sentiment Using CNN and GRU for Risk Prediction and Alert Systems","url":"/papers/arxiv/2412.10199/","summary":"The combination of Convolutional Neural Networks and Gated Recurrent Units offers a comprehensive analysis of stock market sentiment and effective early warnings of future risks.","featured":"2024-12-18","label":"arXiv","topic":"LLMs & Text","cites":10,"score":3,"scale":"shares"},{"title":"Concept Bottleneck Language Models For protein design","url":"/papers/arxiv/2411.06090/","summary":"The paper presents Concept Bottleneck Protein Language Models (CB-pLM), a generative language model that provides control and interpretability in protein generation tasks without affecting performance.","featured":"2024-12-18","label":"Machine learning","topic":"LLMs & Text","cites":24,"score":24,"scale":"shares"},{"title":"Predicting Salmon Prices with Deep Learning and Sentiment Analysis","url":"/papers/repec/eee-jocoma-v-36-y-2024-i-c-s2405851324000576/","summary":"The article discusses a study that uses deep learning models and sentiment analysis to predict salmon prices. The study found that the accuracy of predictions improved when sentiment scores from salmon-related news were included. The hybrid CNN-LSTM model performed the best in these predictions.","featured":"2024-12-18","label":"RePEc","topic":"LLMs & Text","cites":null,"score":19,"scale":"shares"},{"title":"A Scoping Review of ChatGPT Research in Accounting and Finance","url":"/papers/doi/10-1016-j-accinf-2024-100715/","summary":"The article reviews recent research on the use of Large Language Models in accounting and finance, highlighting three key trends and suggesting potential areas for future study.","featured":"2024-12-12","label":"arXiv","topic":"LLMs & Text","cites":99,"score":24,"scale":"shares"},{"title":"RAG-IT: Retrieval-Augmented Instruction Tuning for Automated Financial Analysis - A Case Study for the Semiconductor Sector","url":"/papers/arxiv/2412.08179/","summary":"The study introduces a new method for automating earnings reports analysis using Large Language Models, with promising initial findings.","featured":"2024-12-12","label":"arXiv","topic":"LLMs & Text","cites":4,"score":15,"scale":"shares"},{"title":"A hype-adjusted probability measure for NLP stock return forecasting","url":"/papers/arxiv/2412.07587/","summary":"The paper introduces a new Natural Language Processing method for market forecasting, using a hype-adjusted probability measure to enhance forecast accuracy.","featured":"2024-12-12","label":"arXiv","topic":"LLMs & Text","cites":5,"score":5,"scale":"shares"},{"title":"Generative AI Impact on Labor Market: Analyzing ChatGPT's Demand in Job Advertisements","url":"/papers/arxiv/2412.07042/","summary":"A study reveals the growing demand for ChatGPT-related skills in the U.S. labor market, identifying five key skill sets and emphasizing the widespread use of Generative AI in various sectors.","featured":"2024-12-12","label":"arXiv","topic":"LLMs & Text","cites":9,"score":5,"scale":"shares"},{"title":"Achieving Semantic Consistency: Contextualized Word Representations for Political Text Analysis","url":"/papers/arxiv/2412.04505/","summary":"A comparison study found that BERT performs better than Word2Vec in short-term contexts in social sciences text analysis, but has difficulty with slow semantic changes over longer periods.","featured":"2024-12-12","label":"arXiv","topic":"LLMs & Text","cites":3,"score":13,"scale":"shares"},{"title":"Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling","url":"/papers/arxiv/2412.05271/","summary":"The paper presents InternVL 2.5, a sophisticated multimodal large language model that performs well on various benchmarks, exceeding 70% on the MMMU benchmark.","featured":"2024-12-12","label":"Machine learning","topic":"LLMs & Text","cites":1775,"score":75,"scale":"shares"},{"title":"Training Large Language Models to Reason in a Continuous Latent Space","url":"/papers/arxiv/2412.06769/","summary":"The article presents Coconut, a new approach that uses the last hidden state of large language models for reasoning in an unrestricted latent space, proving its effectiveness in enhancing the LLM on multiple reasoning tasks.","featured":"2024-12-12","label":"Machine learning","topic":"LLMs & Text","cites":742,"score":73,"scale":"shares"},{"title":"NVILA: Efficient Frontier Visual Language Models","url":"/papers/arxiv/2412.04468/","summary":"The article introduces NVILA, a new family of Visual Language Models (VLMs) that balances efficiency and accuracy, reducing training costs and latency while maintaining or improving accuracy.","featured":"2024-12-12","label":"Machine learning","topic":"LLMs & Text","cites":251,"score":271,"scale":"shares"},{"title":"Densing law of LLMs","url":"/papers/arxiv/2412.04315/","summary":"The article presents 'capacity density' as a new metric for evaluating Large Language Models (LLMs), showing that LLMs' capacity density doubles approximately every three months.","featured":"2024-12-12","label":"Machine learning","topic":"LLMs & Text","cites":58,"score":116,"scale":"shares"},{"title":"Impact of Innovation News on Illiquid Stocks","url":"/papers/repec/eme-ejimpp-ejim-07-2022-0387/","summary":"A study using machine learning found that news about product innovation significantly impacts the returns of illiquid stocks, unlike other innovation-related news.","featured":"2024-12-12","label":"RePEc","topic":"LLMs & Text","cites":null,"score":5,"scale":"shares"},{"title":"Critical Tokens Matter: Token-Level Contrastive Estimation Enhances LLM's Reasoning Capability","url":"/papers/arxiv/2411.19943/","summary":"Enhancing Reasoning: The cDPO method identifies and rewards 'critical tokens' that cause incorrect reasoning in Large Language Models, showing effectiveness in two popular models.","featured":"2024-12-04","label":"Machine learning","topic":"LLMs & Text","cites":89,"score":9,"scale":"shares"},{"title":"LUMIA: Linear probing for Unimodal and MultiModal Membership Inference Attacks leveraging internal LLM states","url":"/papers/arxiv/2411.19876/","summary":"Membership Inference Attacks: LUMIA, a new method, uses Linear Probes to detect Membership Inference Attacks in Large Language Models, showing significant improvements over previous methods and providing insights into where attacks are most detectable.","featured":"2024-12-04","label":"Machine learning","topic":"LLMs & Text","cites":11,"score":6,"scale":"shares"},{"title":"Sparrow: Data-Efficient Video-LLM With Text-to-Image Augmentation","url":"/papers/arxiv/2411.19951/","summary":"The T2Vid method, developed by researchers, uses pre-trained image-LLMs to enhance video understanding, performing as well or better than full video datasets with only 15% of the sample size.","featured":"2024-12-04","label":"Machine learning","topic":"LLMs & Text","cites":0,"score":6,"scale":"shares"},{"title":"On Domain-Adaptive Post-Training for Multimodal Large Language Models","url":"/papers/arxiv/2411.19930/","summary":"The paper explores domain adaptation of multimodal large language models through post-training, focusing on data synthesis, training pipelines, and task evaluation, resulting in improved domain-specific performance.","featured":"2024-12-04","label":"Machine learning","topic":"LLMs & Text","cites":14,"score":4,"scale":"shares"},{"title":"The Limits of Inference Scaling Through Resampling","url":"/papers/arxiv/2411.17501/","summary":"The study suggests that the accuracy of weaker language models cannot be indefinitely improved through inference scaling due to an unavoidable probability of false positives.","featured":"2024-12-04","label":"Machine learning","topic":"LLMs & Text","cites":45,"score":142,"scale":"shares"},{"title":"LLM2CLIP: Powerful Language Model Unlocks Richer Cross-Modality Representation","url":"/papers/arxiv/2411.04997/","summary":"Enhancing Visual Representation: Large language models are integrated with the pretrained CLIP visual encoder, enhancing its ability to process complex captions.","featured":"2024-12-04","label":"Machine learning","topic":"LLMs & Text","cites":9,"score":39,"scale":"shares"},{"title":"Insight-V: Exploring Long-Chain Visual Reasoning with Multimodal Large Language Models","url":"/papers/arxiv/2411.14432/","summary":"The paper introduces Insight-V, a system that improves the reasoning abilities of large language models by generating extensive reasoning paths and integrating a multi-agent system for better visual reasoning performance.","featured":"2024-11-27","label":"Machine learning","topic":"LLMs & Text","cites":149,"score":62,"scale":"shares"},{"title":"XGrammar: Flexible and Efficient Structured Generation Engine for Large Language Models","url":"/papers/arxiv/2411.15100/","summary":"The article introduces XGrammar, a structure generation engine that significantly speeds up context-free grammar execution in large language models.","featured":"2024-11-27","label":"Machine learning","topic":"LLMs & Text","cites":91,"score":146,"scale":"shares"},{"title":"Retrieval with Learned Similarities","url":"/papers/arxiv/2407.15462/","summary":"The paper introduces Mixture-of-Logits (MoL), a method that enhances the performance of recommendation systems and language models by approximating similarity functions, reducing latency by up to 66 times.","featured":"2024-11-27","label":"Machine learning","topic":"LLMs & Text","cites":4,"score":81,"scale":"shares"},{"title":"KTO: Model Alignment as Prospect Theoretic Optimization","url":"/papers/arxiv/2402.01306/","summary":"The study presents KTO, a new method that uses a Kahneman-Tversky model to align language models with human feedback, performing better than preference-based methods by learning from a binary signal of output desirability.","featured":"2024-11-27","label":"Machine learning","topic":"LLMs & Text","cites":162,"score":71,"scale":"shares"},{"title":"Disentangling Memory and Reasoning Ability in Large Language Models","url":"/papers/arxiv/2411.13504/","summary":"The article presents a new approach for Large Language Models that divides the process into memory recall and reasoning, enhancing model performance and interpretability.","featured":"2024-11-27","label":"Machine learning","topic":"LLMs & Text","cites":55,"score":63,"scale":"shares"},{"title":"MagicQuill: An Intelligent Interactive Image Editing System","url":"/papers/arxiv/2411.09703/","summary":"Image Editing System: MagicQuill is an image editing system that uses a large language model to predict editing intentions in real time, enabling quick and accurate image modifications.","featured":"2024-11-20","label":"Machine learning","topic":"LLMs & Text","cites":47,"score":235,"scale":"shares"},{"title":"LlaVA-CoT: Let Vision Language Models Reason Step-By-Step","url":"/papers/arxiv/2411.10440/","summary":"Vision Language Model: LLaVA-o1 is a new Vision-Language Model that performs autonomous multistage reasoning, enhancing precision in reasoning-intensive tasks and surpassing larger models in various multimodal reasoning benchmarks.","featured":"2024-11-20","label":"Machine learning","topic":"LLMs & Text","cites":566,"score":133,"scale":"shares"},{"title":"Squeezed Attention: Accelerating Long Context Length LLM Inference","url":"/papers/arxiv/2411.09688/","summary":"LLM Inference: Squeezed Attention is a proposed method to speed up Large Language Model applications by using K-means clustering to group similar keys in fixed context inputs, reducing computational costs and enhancing inference efficiency.","featured":"2024-11-20","label":"Machine learning","topic":"LLMs & Text","cites":60,"score":25,"scale":"shares"},{"title":"Adaptive Decoding via Latent Preference Optimization","url":"/papers/arxiv/2411.09661/","summary":"Preference Optimization: Adaptive Decoding is a technique that dynamically selects the sampling temperature during language model decoding, optimizing performance across various tasks that require different temperatures.","featured":"2024-11-20","label":"Machine learning","topic":"LLMs & Text","cites":13,"score":19,"scale":"shares"},{"title":"Probing LLM Hallucination from Within: Perturbation-Driven Approach via Internal Knowledge","url":"/papers/arxiv/2411.09689/","summary":"A new task called Hallucination Reasoning is introduced to better categorize text generated by Language Learning Models, enhancing the detection of unfaithful text generation.","featured":"2024-11-20","label":"Machine learning","topic":"LLMs & Text","cites":3,"score":13,"scale":"shares"},{"title":"A Single Transformer for Scalable Vision-Language Modeling","url":"/papers/arxiv/2407.06438/","summary":"SOLO is a unified transformer for vision-language modeling, addressing scalability issues in large models and providing an open-source training blueprint.","featured":"2024-11-20","label":"Machine learning","topic":"LLMs & Text","cites":41,"score":184,"scale":"shares"},{"title":"Verifiable by Design: Aligning Language Models to Quote from Pre-Training Data","url":"/papers/arxiv/2404.03862/","summary":"Quote-Tuning is a new approach that prompts large language models to directly quote from reliable sources, thereby improving their credibility and verifiability.","featured":"2024-11-20","label":"Machine learning","topic":"LLMs & Text","cites":19,"score":60,"scale":"shares"},{"title":"Vector Retrieval with Similarity and Diversity: How Hard Is It?","url":"/papers/arxiv/2407.04573/","summary":"The article presents a new method for vector retrieval in Large Language Models (LLMs) called VRSD, which ensures similarity and diversity constraints and performs better than the commonly used MMR method.","featured":"2024-11-20","label":"Machine learning","topic":"LLMs & Text","cites":3,"score":41,"scale":"shares"},{"title":"Dynamic Rewarding with Prompt Optimization Enables Tuning-free Self-Alignment of Language Models","url":"/papers/arxiv/2411.08733/","summary":"The paper presents a new tuning-free method called Dynamic Rewarding with Prompt Optimization (DRPO) for self-aligning Large Language Models (LLMs), improving alignment performance without extra training or human intervention.","featured":"2024-11-20","label":"Machine learning","topic":"LLMs & Text","cites":20,"score":27,"scale":"shares"},{"title":"Large Language Model-Based Interpretable Machine Learning Control in Building Energy Systems","url":"/papers/arxiv/2402.09584/","summary":"The study investigates the use of Interpretable Machine Learning (IML) in HVAC systems to enhance transparency and understanding, using Shapley values and Large Language Models (LLMs) to create a comprehensible narrative.","featured":"2024-11-20","label":"Machine learning","topic":"LLMs & Text","cites":83,"score":21,"scale":"shares"},{"title":"Sentiment and Price Resilience in Recycling","url":"/papers/ssrn/5013887/","summary":"Machine learning research revealed that Southeast Asian economies are key to the vessel recycling industry, despite tighter regulations.","featured":"2024-11-13","label":"SSRN","topic":"LLMs & Text","cites":null,"score":3,"scale":"shares"},{"title":"The Semantic Hub Hypothesis: Language Models Share Semantic Representations Across Languages and Modalities","url":"/papers/arxiv/2411.04986/","summary":"Modern language models can process various languages and forms by learning a shared representation space used during input processing.","featured":"2024-11-13","label":"Machine learning","topic":"LLMs & Text","cites":82,"score":65,"scale":"shares"},{"title":"Enabling LLM Knowledge Analysis via Extensive Materialization","url":"/papers/arxiv/2411.04920/","summary":"A large general-domain knowledge base (GPTKB) can be built entirely from a large language model, containing 105 million triples for over 2.9 million entities.","featured":"2024-11-13","label":"Machine learning","topic":"LLMs & Text","cites":18,"score":45,"scale":"shares"},{"title":"Stronger Models are NOT Stronger Teachers for Instruction Tuning","url":"/papers/arxiv/2411.07133/","summary":"A study introduces Compatibility-Adjusted Reward (CAR), a new metric to evaluate the effectiveness of language models, challenging the belief that larger models are better for instruction tuning.","featured":"2024-11-13","label":"Machine learning","topic":"LLMs & Text","cites":17,"score":19,"scale":"shares"},{"title":"RefreshKV: Updating Small KV Cache During Long-form Generation","url":"/papers/arxiv/2411.05787/","summary":"Recycled Attention is a method for large language models that alternates attention between full context and a subset of input tokens, improving performance and reducing computational load in long-context tasks.","featured":"2024-11-13","label":"Machine learning","topic":"LLMs & Text","cites":12,"score":14,"scale":"shares"},{"title":"Logits of API-Protected LLMs Leak Proprietary Information","url":"/papers/arxiv/2403.09539/","summary":"Researchers have discovered a method to extract hidden information from large language models like OpenAI's gpt-3.5-turbo using API queries, exploiting a weakness known as the softmax bottleneck.","featured":"2024-11-13","label":"Machine learning","topic":"LLMs & Text","cites":53,"score":615,"scale":"shares"},{"title":"Is ChatGPT Transforming Academics' Writing Style?","url":"/papers/arxiv/2404.08627/","summary":"A study of a million arXiv papers shows that large language models, specifically ChatGPT, are significantly impacting the writing style of academic abstracts, especially in computer science.","featured":"2024-11-13","label":"Machine learning","topic":"LLMs & Text","cites":40,"score":293,"scale":"shares"},{"title":"No Train, all Gain: Self-Supervised Gradients Improve Deep Frozen Representations","url":"/papers/arxiv/2407.10964/","summary":"The FUNGI method improves transformer encoders' features using self-supervised gradients, enhancing performance in vision, natural language processing, and audio tasks and datasets.","featured":"2024-11-13","label":"Machine learning","topic":"LLMs & Text","cites":6,"score":78,"scale":"shares"},{"title":"A Implies B: Circuit Analysis in LLMs for Propositional Logical Reasoning","url":"/papers/arxiv/2411.04105/","summary":"A study reveals the internal mechanisms of large language models that enable complex logical reasoning, identifying specific planning and reasoning circuits.","featured":"2024-11-13","label":"Machine learning","topic":"LLMs & Text","cites":16,"score":40,"scale":"shares"},{"title":"SynCode: LLM Generation with Grammar Augmentation","url":"/papers/arxiv/2403.01632/","summary":"LLM Generation with Grammar Augmentation: SynCode, a new framework for syntactical decoding with large language models, is introduced, significantly reducing syntax errors in Python and Go code generation.","featured":"2024-11-13","label":"Machine learning","topic":"LLMs & Text","cites":87,"score":37,"scale":"shares"},{"title":"Predicting Public Opinions","url":"/papers/ssrn/5008330/","summary":"A study finds that large language models like ChatGPT4o can mimic human responses in surveys and predict election results, but struggle with sensitive topics.","featured":"2024-11-06","label":"SSRN","topic":"LLMs & Text","cites":null,"score":20,"scale":"shares"},{"title":"Financial Regulations with Language Models","url":"/papers/ssrn/5010694/","summary":"Large language models (LLMs) in AI are expected to ease the load on financial institutions in managing complex regulatory projects, especially in software development.","featured":"2024-11-06","label":"SSRN","topic":"LLMs & Text","cites":null,"score":10,"scale":"shares"},{"title":"Blending Ensemble for Classification with Genetic-algorithm generated Alpha factors and Sentiments (GAS)","url":"/papers/arxiv/2411.03035/","summary":"The article introduces a Genetic Algorithm-generated Alpha Sentiment (GAS) model that uses advanced learning techniques and sentiment analysis to predict Bitcoin market trends and price changes.","featured":"2024-11-06","label":"arXiv","topic":"LLMs & Text","cites":2,"score":5,"scale":"shares"},{"title":"\"Give Me BF16 or Give Me Death\"? Accuracy-Performance Trade-Offs in LLM Quantization","url":"/papers/arxiv/2411.02355/","summary":"A study shows that FP8 weight and activation quantization in large language models (LLMs) is lossless across all scales, while INT8 and INT4 quantizations also perform well with proper tuning, offering guidelines for deploying quantized LLMs.","featured":"2024-11-06","label":"Machine learning","topic":"LLMs & Text","cites":44,"score":50,"scale":"shares"},{"title":"SelfCodeAlign: Self-Alignment for Code Generation","url":"/papers/arxiv/2410.24198/","summary":"SelfCodeAlign, a new pipeline, enhances the ability of LLMs to follow human instructions without extensive human annotations, outperforming previous methods and creating a top-performing coding model.","featured":"2024-11-06","label":"Machine learning","topic":"LLMs & Text","cites":78,"score":14,"scale":"shares"},{"title":"Attacking Vision-Language Computer Agents via Pop-ups","url":"/papers/arxiv/2411.02391/","summary":"The study shows that autonomous agents using large vision and language models can be significantly disrupted by strategically designed adversarial pop-ups.","featured":"2024-11-06","label":"Machine learning","topic":"LLMs & Text","cites":124,"score":9,"scale":"shares"},{"title":"Efficient Adversarial Training in LLMs with Continuous Attacks","url":"/papers/arxiv/2405.15589/","summary":"CAdvUL, a new adversarial training algorithm, enhances the resilience of large language models against adversarial attacks by efficiently calculating attacks in the continuous embedding space.","featured":"2024-11-06","label":"Machine learning","topic":"LLMs & Text","cites":148,"score":82,"scale":"shares"},{"title":"AmbigNLG: Addressing Task Ambiguity in Instruction for NLG","url":"/papers/arxiv/2402.17717/","summary":"Task Ambiguity in NLG: AmbigNLG, a new task and dataset, tackles task ambiguity in instructions for Natural Language Generation, improving the alignment of generated text with user expectations and boosting the performance of Large Language Models.","featured":"2024-11-06","label":"Machine learning","topic":"LLMs & Text","cites":16,"score":46,"scale":"shares"},{"title":"DuQuant: Distributing Outliers via Dual Transformation Makes Stronger Quantized LLMs","url":"/papers/arxiv/2406.01721/","summary":"Outlier Management for LLMs: DuQuant, a new method for quantizing large language models, uses rotation and permutation transformations to manage outliers, surpassing performance of existing baselines in various tasks.","featured":"2024-11-06","label":"Machine learning","topic":"LLMs & Text","cites":194,"score":18,"scale":"shares"},{"title":"LaCour!: enabling research on argumentation in hearings of the European Court of Human Rights","url":"/papers/arxiv/2312.05061/","summary":"Argumentation Research in ECHR Hearings: LaCour!, the first corpus of textual oral arguments from the European Court of Human Rights, provides transcribed multilingual oral hearings linked to final judgments, enhancing legal research.","featured":"2024-11-06","label":"Machine learning","topic":"LLMs & Text","cites":2,"score":16,"scale":"shares"},{"title":"Understanding Synthetic Context Extension via Retrieval Heads","url":"/papers/arxiv/2410.22316/","summary":"The paper finds that fine-tuning long-context language models with synthetic data improves performance in retrieval and reasoning tasks, with attention heads predicting performance.","featured":"2024-10-31","label":"Machine learning","topic":"LLMs & Text","cites":13,"score":9,"scale":"shares"},{"title":"Machine Learning for Fake News Classification","url":"/papers/ssrn/4990295/","summary":"ML algorithms combined with IoT frameworks can enhance news classification and detect fake news in real-time, helping to fight misinformation.","featured":"2024-10-23","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Bridging the Training-Inference Gap in LLMs by Leveraging Self-Generated Tokens","url":"/papers/arxiv/2410.14655/","summary":"The paper suggests two methods to address the discrepancy between training and inference time in language models, resulting in enhanced performance in tasks like summarization and question-answering.","featured":"2024-10-23","label":"Machine learning","topic":"LLMs & Text","cites":9,"score":43,"scale":"shares"},{"title":"Differentiable Robot Rendering","url":"/papers/arxiv/2410.13851/","summary":"The paper presents a method for differentiable robot rendering, enabling the visual appearance of a robot to be directly differentiable with respect to its control parameters, useful for reconstructing robot poses from images and controlling robots through vision language models.","featured":"2024-10-23","label":"Machine learning","topic":"LLMs & Text","cites":29,"score":16,"scale":"shares"},{"title":"xGen-MM-Vid (BLIP-3-Video): You Only Need 32 Tokens to Represent a Video Even in VLMs","url":"/papers/arxiv/2410.16267/","summary":"XGen-MM-Vid (BLIP-3-Video) is a language model for videos that captures temporal information efficiently, offering accuracy similar to larger models but with greater efficiency.","featured":"2024-10-23","label":"Machine learning","topic":"LLMs & Text","cites":37,"score":13,"scale":"shares"},{"title":"LightTransfer: Your Long-Context LLM is Secretly a Hybrid Model with Effortless Adaptation","url":"/papers/arxiv/2410.13846/","summary":"SimLayerKV is a technique that minimizes memory usage in large language models by identifying and reducing cache in lazy layers, achieving significant cache compression with minimal performance loss.","featured":"2024-10-23","label":"Machine learning","topic":"LLMs & Text","cites":13,"score":8,"scale":"shares"},{"title":"EasyRec: Simple yet Effective Language Models for Recommendation","url":"/papers/arxiv/2408.08821/","summary":"Recommendation Language Models: The study presents EasyRec, a method that combines text-based semantic understanding with collaborative signals for recommender systems, showing improved performance in text-based zero-shot recommendation situations.","featured":"2024-10-23","label":"Machine learning","topic":"LLMs & Text","cites":22,"score":15,"scale":"shares"},{"title":"Improved NHL Draft Predictions with Scouting Reports","url":"/papers/repec/bpj-jqsprt-v-20-y-2024-i-4-p-331-349-n-1006/","summary":"Large Language Models (LLMs) are being used to enhance predictions of NHL draft outcomes by extracting information from scouting report texts and combining it with on-ice statistics.","featured":"2024-10-23","label":"RePEc","topic":"LLMs & Text","cites":null,"score":11,"scale":"shares"},{"title":"EU News Engagement on Facebook","url":"/papers/repec/cog-poango-v10-y-2022-i-1-p-121-132/","summary":"Study of social media engagement with EU news shows negativity increases reactions and shares but decreases comments, while emotionality decreases reactions and shares but increases comments.","featured":"2024-10-23","label":"RePEc","topic":"LLMs & Text","cites":null,"score":4,"scale":"shares"},{"title":"Large Language Models for Forecasting","url":"/papers/ssrn/4988022/","summary":"The article assesses the performance of Large Language Models in time series forecasting, emphasizing their potential for precise predictions and the necessity for further model improvements.","featured":"2024-10-17","label":"SSRN","topic":"LLMs & Text","cites":null,"score":21,"scale":"shares"},{"title":"AI Sentiment Analysis","url":"/papers/ssrn/4984337/","summary":"The study shows that AI-rewritten SEC filings increase positive sentiment, positively affecting stock prices, emphasizing the need for careful AI use in financial disclosures.","featured":"2024-10-17","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Credit Market Sentiment","url":"/papers/ssrn/4988908/","summary":"Credit spread expectation errors, seen as signs of market optimism, can predict economic downturns, showing the importance of credit market sentiment in economic cycles.","featured":"2024-10-17","label":"SSRN","topic":"LLMs & Text","cites":null,"score":3,"scale":"shares"},{"title":"Predicting Intraday Risk and Liquidity with News Analytics","url":"/papers/ssrn/4987091/","summary":"The research investigates the correlation between the intensity of news arrival, volatility, and volume at an intraday frequency using a global dataset and natural language processing.","featured":"2024-10-17","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Your Mixture-of-Experts LLM Is Secretly an Embedding Model For Free","url":"/papers/arxiv/2410.10814/","summary":"The research shows that Mixture-of-Experts Large Language Models can be effective embedding models without finetuning, and suggests a combination of routing weights and hidden state for better performance.","featured":"2024-10-17","label":"Machine learning","topic":"LLMs & Text","cites":42,"score":47,"scale":"shares"},{"title":"Reuse Your Rewards: Reward Model Transfer for Zero-Shot Cross-Lingual Alignment","url":"/papers/arxiv/2404.12318/","summary":"A study has found that language models aligned with human-annotated preference data are preferred by humans in over 70% of cases, even without language-specific data for supervised finetuning.","featured":"2024-10-17","label":"Machine learning","topic":"LLMs & Text","cites":32,"score":119,"scale":"shares"},{"title":"Evaluating Copyright Takedown Methods for Language Models","url":"/papers/arxiv/2406.18664/","summary":"The article discusses CoTaEval, a new framework for evaluating methods to prevent AI from generating copyrighted content, highlighting the need for further research as no method was found to be completely effective.","featured":"2024-10-17","label":"Machine learning","topic":"LLMs & Text","cites":52,"score":42,"scale":"shares"},{"title":"Online Investor Sentiment and Stock Market Risk","url":"/papers/repec/gam-jmathe-v-12-y-2024-i-20-p-3192-d-1497063/","summary":"Machine learning techniques like extreme gradient boosting and random forest are more accurate in predicting the aggregated stock market risk premium based on online investor sentiment than traditional linear models.","featured":"2024-10-17","label":"RePEc","topic":"LLMs & Text","cites":null,"score":22,"scale":"shares"},{"title":"News Text Analysis","url":"/papers/repec/oup-rfinst-v-36-y-2023-i-12-p-4759-4787/","summary":"A study reveals that a pricing model based on news text from The Wall Street Journal is more effective in predicting investment opportunities than traditional models, using topic modeling and latent factor analysis.","featured":"2024-10-17","label":"RePEc","topic":"LLMs & Text","cites":null,"score":6,"scale":"shares"},{"title":"Index Investing","url":"/papers/ssrn/4978330/","summary":"The study suggests that index stocks have higher prices, more volatility, stronger negative price autocorrelation, and higher trading volume due to sentiment spillover from index investors.","featured":"2024-10-09","label":"SSRN","topic":"LLMs & Text","cites":null,"score":4,"scale":"shares"},{"title":"Leveraging Automatically Optimized Forecasters and Large Language Model for Forecasting of Vietnamese Macroeconomic Indicators","url":"/papers/ssrn/4976975/","summary":"The study proposes a framework for predicting Vietnamese macroeconomic indicators by combining traditional data with insights from news articles, highlighting the reliability of Gradient Boost Decision Tree models.","featured":"2024-10-09","label":"SSRN","topic":"LLMs & Text","cites":0,"score":4,"scale":"shares"},{"title":"GenSim2: Scaling Robot Data Generation with Multi-modal and Reasoning LLMs","url":"/papers/arxiv/2410.03645/","summary":"GenSim2 is a scalable framework for robotic simulation that uses large language models for task creation and a multi-task language-conditioned policy architecture to learn from demonstrations, improving policy performance and enabling zero-shot transfer.","featured":"2024-10-09","label":"Machine learning","topic":"LLMs & Text","cites":66,"score":8,"scale":"shares"},{"title":"GPT-4o as the Gold Standard: A Scalable and General Purpose Approach to Filter Language Model Pretraining Data","url":"/papers/arxiv/2410.02755/","summary":"Data Filtering System with GPT-4o Accuracy: The article introduces SIEVE, a cost-effective method for filtering web-scale data that matches the accuracy of GPT-4o and is efficient in curating large datasets for language model training.","featured":"2024-10-09","label":"Machine learning","topic":"LLMs & Text","cites":2,"score":8,"scale":"shares"},{"title":"Language Model Training on Edit Sequences Enhances Code Synthesis","url":"/papers/web/9b37a775db/","summary":"The paper presents LintSeq, a synthetic data generation algorithm that refactors code into a sequence of edits, resulting in more diverse programs and improved code synthesis performance.","featured":"2024-10-09","label":"Machine learning","topic":"LLMs & Text","cites":null,"score":7,"scale":"shares"},{"title":"TextHawk2: A Large Vision-Language Model Excels in Bilingual OCR and Grounding with 16x Fewer Tokens","url":"/papers/arxiv/2410.05261/","summary":"Efficient LVLM for Bilingual OCR: The study introduces TextHawk2, a bilingual Large Vision-Language Model that provides efficient fine-grained perception and superior performance with 16 times fewer image tokens than previous models.","featured":"2024-10-09","label":"Machine learning","topic":"LLMs & Text","cites":33,"score":6,"scale":"shares"},{"title":"Function-Guided Conditional Generation Using Protein Language Models with Adapters","url":"/papers/arxiv/2410.03634/","summary":"The research proposes ProCALM, a method for generating proteins conditionally using adapters to protein language models, capable of generating sequences from target enzyme families and generalizing to rare and unseen ones.","featured":"2024-10-09","label":"Machine learning","topic":"LLMs & Text","cites":9,"score":6,"scale":"shares"},{"title":"Training Language Models to Self-Correct via Reinforcement Learning","url":"/papers/arxiv/2409.12917/","summary":"SCoRe, a new online reinforcement learning approach, enhances the self-correction ability of large language models, showing top performance with Gemini 1.0 Pro and 1.5 Flash models.","featured":"2024-10-09","label":"Machine learning","topic":"LLMs & Text","cites":460,"score":585,"scale":"shares"},{"title":"SaySelf: Teaching LLMs to Express Confidence with Self-Reflective Rationales","url":"/papers/arxiv/2405.20974/","summary":"The SaySelf training framework instructs large language models to provide more precise confidence estimates and self-reflective rationales, effectively reducing confidence calibration error while maintaining task performance.","featured":"2024-10-09","label":"Machine learning","topic":"LLMs & Text","cites":125,"score":231,"scale":"shares"},{"title":"LML-DAP: LANGUAGE MODEL LEARNING A DATASET FOR DATA-AUGMENTED PREDICTION","url":"/papers/arxiv/2409.18957/","summary":"Language Model Prediction: The paper presents a new method called Data-Augmented Prediction (DAP) for using Large Language Models (LLMs) in classification tasks, achieving over 90% accuracy in some tests.","featured":"2024-10-09","label":"Machine learning","topic":"LLMs & Text","cites":2,"score":29,"scale":"shares"},{"title":"SoK: Membership Inference Attacks on LLMs are Rushing Nowhere (and How to Fix It)","url":"/papers/arxiv/2406.17975/","summary":"The article examines Membership Inference Attacks (MIAs) against Large Language Models (LLMs), suggesting potential solutions and comprehensive benchmarks for sequence- and document-level MIAs against LLMs.","featured":"2024-10-09","label":"Machine learning","topic":"LLMs & Text","cites":67,"score":17,"scale":"shares"},{"title":"Uncertainty and financial market resilience: evidence from China","url":"/papers/arxiv/2409.18422/","summary":"The article evaluates the impact of external shocks, including climate policy uncertainty, on China's financial markets, revealing that such uncertainty affects investor sentiment, commercial banks' non-performing loan ratio, and the capital and financial account balance.","featured":"2024-10-03","label":"arXiv","topic":"LLMs & Text","cites":0,"score":3,"scale":"shares"},{"title":"What is the Role of Large Language Models in the Evolution of Astronomy Research?","url":"/papers/arxiv/2409.20252/","summary":"A study involving 13 astronomers discusses the potential and limitations of large language models like ChatGPT in research activities, emphasizing the importance of critical thinking and domain expertise to ensure these tools support, not replace, rigorous scientific investigation.","featured":"2024-10-03","label":"Machine learning","topic":"LLMs & Text","cites":11,"score":4,"scale":"shares"},{"title":"LLM Hallucinations in Practical Code Generation: Phenomena, Mechanism, and Mitigation","url":"/papers/arxiv/2409.20550/","summary":"A study investigates the occurrence, mechanism, and reduction of hallucinations in code generated by large language models, suggesting a mitigation method and creating a classification of these hallucinations.","featured":"2024-10-03","label":"Machine learning","topic":"LLMs & Text","cites":228,"score":3,"scale":"shares"},{"title":"FABLES: Evaluating faithfulness and content selection in book-length summarization","url":"/papers/arxiv/2404.01261/","summary":"Research shows large language models (LLMs) often misrepresent events and character states when summarizing long documents, highlighting the need for improved evaluation methods.","featured":"2024-10-03","label":"Machine learning","topic":"LLMs & Text","cites":79,"score":495,"scale":"shares"},{"title":"EgoLM: Multi-Modal Language Model of Egocentric Motions","url":"/papers/arxiv/2409.18127/","summary":"Egocentric Motion Model: EgoLM, a new framework using LLMs, effectively tracks and understands egocentric motions from various inputs, proving its utility in universal egocentric learning.","featured":"2024-10-03","label":"Machine learning","topic":"LLMs & Text","cites":31,"score":173,"scale":"shares"},{"title":"The Impact of Element Ordering on LM Agent Performance","url":"/papers/arxiv/2409.12089/","summary":"The sequence of elements presented to language model agents in virtual environments greatly affects their performance, with random sequences causing similar performance drops as removing all visible text.","featured":"2024-09-25","label":"Machine learning","topic":"LLMs & Text","cites":3,"score":27,"scale":"shares"},{"title":"To CoT or not to CoT? Chain-of-thought helps mainly on math and symbolic reasoning","url":"/papers/arxiv/2409.12183/","summary":"The chain-of-thought method is most beneficial for tasks involving math or logic in large language models, indicating a need for new methods that utilize intermediate computation across various applications.","featured":"2024-09-25","label":"Machine learning","topic":"LLMs & Text","cites":371,"score":23,"scale":"shares"},{"title":"Fake News Detection","url":"/papers/repec/wsi-jikmxx-v-23-y-2024-i-05-n-s0219649224500758/","summary":"The study presents a machine learning model that can detect fake news with 99% accuracy using logistic regression and feature hashing vectorisation.","featured":"2024-09-25","label":"RePEc","topic":"LLMs & Text","cites":null,"score":16,"scale":"shares"},{"title":"Stock Prediction News Headlines","url":"/papers/repec/kap-compec-v-64-y-2024-i-2-d-10-1007-s10614-023-10449-5/","summary":"This article discusses the application of machine learning and deep learning techniques to analyze financial news headlines, with the aim of identifying low volatility stocks that perform better than the Standard and Poor’s 500 Index.","featured":"2024-09-25","label":"RePEc","topic":"LLMs & Text","cites":null,"score":18,"scale":"shares"},{"title":"Analyst Question Quality in Conference Calls","url":"/papers/ssrn/4958876/","summary":"Machine learning methods show that high-quality analyst questions during earnings calls lead to better stock liquidity and stability, particularly in firms with unclear information.","featured":"2024-09-18","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Investor Sentiment in Malaysian Government Bonds","url":"/papers/ssrn/4955228/","summary":"The COVID-19 pandemic has changed investment strategies in bond markets, with Malaysian government bonds showing equity-like traits and deviating from their usual safe-haven role, with gold prices being the key factor affecting bond yields.","featured":"2024-09-18","label":"SSRN","topic":"LLMs & Text","cites":null,"score":64,"scale":"shares"},{"title":"Source2Synth: Synthetic Data Generation and Curation Grounded in Real Data Sources","url":"/papers/arxiv/2409.08239/","summary":"Synthetic Data for LLMs: Source2Synth, a new method for teaching Large Language Models new skills without human annotations, has improved multi-hop and tabular question answering by 22.57% and 25.51% respectively.","featured":"2024-09-18","label":"Machine learning","topic":"LLMs & Text","cites":33,"score":89,"scale":"shares"},{"title":"Synthetic continued pretraining","url":"/papers/arxiv/2409.07431/","summary":"Researchers suggest using EntiGraph, a synthetic data augmentation algorithm, for synthetic continued pretraining to help language models answer questions and follow instructions related to source documents more effectively.","featured":"2024-09-18","label":"Machine learning","topic":"LLMs & Text","cites":61,"score":95,"scale":"shares"},{"title":"Agent Workflow Memory","url":"/papers/arxiv/2409.07429/","summary":"The Agent Workflow Memory (AWM) method has been developed to enhance the performance of language model-based agents in complex tasks by creating reusable workflows from past experiences to guide future actions.","featured":"2024-09-18","label":"Machine learning","topic":"LLMs & Text","cites":287,"score":33,"scale":"shares"},{"title":"Linguistic Bias in ChatGPT: Language Models Reinforce Dialect Discrimination","url":"/papers/arxiv/2406.08818/","summary":"A comprehensive study of AI models like GPT-3.5 Turbo and GPT-4 reveals linguistic bias, with these models defaulting to standard English and often displaying stereotyping and condescending responses towards non-standard English dialects.","featured":"2024-09-18","label":"Machine learning","topic":"LLMs & Text","cites":102,"score":19,"scale":"shares"},{"title":"SUPER: Evaluating Agents on Setting Up and Executing Tasks from Research Repositories","url":"/papers/arxiv/2409.07440/","summary":"The SUPER benchmark tests Large Language Models' ability to set up and execute tasks from research repositories, revealing that current models struggle with these tasks, suggesting a need for further advancements in this field.","featured":"2024-09-18","label":"Machine learning","topic":"LLMs & Text","cites":54,"score":11,"scale":"shares"},{"title":"Sentiment Analysis in Finance","url":"/papers/repec/wsi-wschap-9781800615212-0005/","summary":"Sentiment Analysis is used in finance to predict market trends and investment opportunities using various algorithms and metrics.","featured":"2024-09-18","label":"RePEc","topic":"LLMs & Text","cites":null,"score":31,"scale":"shares"},{"title":"Tuning into Climate Risks: Extracting Innovation from Television News for Clean Energy Firms","url":"/papers/arxiv/2409.08701/","summary":"The article studies the impact of TV news coverage on U.S. clean energy firms' climate risks. It finds that while increased climate risk coverage lowers specific risk, it raises overall risk. Negative sentiments, however, increase specific risk and reduce overall risk.","featured":"2024-09-18","label":"arXiv","topic":"LLMs & Text","cites":0,"score":2,"scale":"shares"},{"title":"Quantifying Bias in Sentiment Analysis","url":"/papers/ssrn/4949090/","summary":"Research shows large language models used in sentiment analysis display social biases, linking specific jobs with certain genders.","featured":"2024-09-10","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Attention heads of large language models","url":"/papers/arxiv/2409.03752/","summary":"The article delves into the reasoning processes of Large Language Models, focusing on the interpretability of attention heads, and suggests future research areas.","featured":"2024-09-10","label":"Machine learning","topic":"LLMs & Text","cites":99,"score":36,"scale":"shares"},{"title":"LongCite: Enabling LLMs to Generate Fine-grained Citations in Long-context QA","url":"/papers/arxiv/2409.02897/","summary":"Enhancing LLMs: The research presents a method for large language models to generate responses with detailed sentence-level citations, enhancing their accuracy and verifiability, and provides a performance assessment benchmark.","featured":"2024-09-10","label":"Machine learning","topic":"LLMs & Text","cites":100,"score":11,"scale":"shares"},{"title":"Planning In Natural Language Improves LLM Search For Code Generation","url":"/papers/arxiv/2409.03733/","summary":"PLANSEARCH is a new search algorithm that creates diverse solutions for natural language problems, outperforming traditional methods in various benchmarks.","featured":"2024-09-10","label":"Machine learning","topic":"LLMs & Text","cites":93,"score":10,"scale":"shares"},{"title":"RLPF: Reinforcement Learning from Prediction Feedback for User Summarization with LLMs","url":"/papers/arxiv/2409.04421/","summary":"Reinforcement Learning from Prediction Feedback (RLPF) is a method that refines Large Language Models to produce concise, human-readable summaries, enhancing task performance and summary quality.","featured":"2024-09-10","label":"Machine learning","topic":"LLMs & Text","cites":16,"score":9,"scale":"shares"},{"title":"Kun: Answer Polishment for Chinese Self-Alignment with Instruction Back-Translation","url":"/papers/arxiv/2401.06477/","summary":"The paper introduces Kun, a new method for creating high-quality instruction-tuning datasets for large language models without manual annotations, demonstrating its robustness and scalability and providing a scalable solution for enhancing these models' instruction-following abilities.","featured":"2024-09-10","label":"Machine learning","topic":"LLMs & Text","cites":10,"score":45,"scale":"shares"},{"title":"Shipping Sentiment Impact on Rates","url":"/papers/repec/eee-transe-v-189-y-2024-i-c-s1366554524002424/","summary":"The research uses language models to create sentiment indices for shipping markets, showing they can accurately predict freight rates, outperforming traditional sentiment analysis.","featured":"2024-09-10","label":"RePEc","topic":"LLMs & Text","cites":null,"score":19,"scale":"shares"},{"title":"Scaling Laws for Economic Productivity: Experimental Evidence in LLM-Assisted Translation","url":"/papers/arxiv/2409.02391/","summary":"Research shows that enhancing the computational training of Large Language Models (LLMs) greatly boosts the efficiency of professional translators, especially those with lower skills, indicating potential significant economic impacts with further model scaling.","featured":"2024-09-10","label":"arXiv","topic":"LLMs & Text","cites":11,"score":69,"scale":"shares"},{"title":"Informed Short Sellers","url":"/papers/ssrn/4941397/","summary":"Contrary to popular belief, informed short sellers are found to supply liquidity and provide valuable information about future returns, especially on news days.","featured":"2024-09-05","label":"SSRN","topic":"LLMs & Text","cites":null,"score":30,"scale":"shares"},{"title":"ECC Analyzer: Extracting Trading Signal from Earnings Conference Calls using Large Language Model for Stock Volatility Prediction","url":"/papers/arxiv/2404.18470/","summary":"Stock Volatility Prediction: ECCAnalyzer, a new tool, uses advanced language models to extract data from earnings conference calls, enhancing stock volatility predictions and surpassing traditional analysis methods.","featured":"2024-09-05","label":"arXiv","topic":"LLMs & Text","cites":20,"score":11,"scale":"shares"},{"title":"InkubaLM: A small language model for low-resource African languages","url":"/papers/arxiv/2408.17024/","summary":"African Language Model: InkubaLM, a language model for African languages, is introduced, performing well in tasks like machine translation and sentiment analysis despite limited resources.","featured":"2024-09-05","label":"Machine learning","topic":"LLMs & Text","cites":31,"score":177,"scale":"shares"},{"title":"SYNTHEVAL: Hybrid Behavioral Testing of NLP Models with Synthetic CheckLists","url":"/papers/arxiv/2408.17437/","summary":"SYNTHEVAL is a testing framework that uses large language models to generate tests for evaluating NLP models, particularly in sentiment analysis and toxic language detection.","featured":"2024-09-05","label":"Machine learning","topic":"LLMs & Text","cites":8,"score":6,"scale":"shares"},{"title":"Advancing Multi-Talker ASR Performance With Large Language Models","url":"/papers/arxiv/2408.17431/","summary":"A new approach using large language models improves multi-talker automatic speech recognition, outperforming traditional methods on real-world datasets.","featured":"2024-09-05","label":"Machine learning","topic":"LLMs & Text","cites":19,"score":6,"scale":"shares"},{"title":"Getting Inspiration for Feature Elicitation: App Store- vs. LLM-based Approach","url":"/papers/arxiv/2408.17404/","summary":"A study comparing AppStore and large language model approaches for refining app features finds both are effective, but LLMs excel in novel unseen app scopes, emphasizing the role of human analysts.","featured":"2024-09-05","label":"Machine learning","topic":"LLMs & Text","cites":16,"score":4,"scale":"shares"},{"title":"Cutting Through the Noise: Boosting LLM Performance on Math Word Problems","url":"/papers/arxiv/2406.15444/","summary":"The study introduces a framework that uses adversarial math problems to enhance the performance and robustness of Large Language Models (LLMs).","featured":"2024-09-05","label":"Machine learning","topic":"LLMs & Text","cites":13,"score":50,"scale":"shares"},{"title":"Low-Rank Quantization-Aware Training for LLMs","url":"/papers/arxiv/2406.06385/","summary":"The article introduces LR-QAT, a memory-efficient training algorithm for Large Language Models (LLMs), demonstrating its effectiveness and memory efficiency over common post-training quantization methods.","featured":"2024-09-05","label":"Machine learning","topic":"LLMs & Text","cites":60,"score":30,"scale":"shares"},{"title":"Sentiment and Herd Behavior of Private Investors","url":"/papers/repec/fru-finjrn-240406-p-95-113/","summary":"The paper examines the sentiment of private investors on online platforms and its effect on herd behavior in the Russian stock market.","featured":"2024-09-05","label":"RePEc","topic":"LLMs & Text","cites":null,"score":14,"scale":"shares"},{"title":"VIX and Global Consciousness in Market Sentiment","url":"/papers/repec/eme-jespps-jes-11-2023-0663/","summary":"The research finds a significant correlation between Global Consciousness Project data and the S&P 500 Volatility Index, suggesting its potential in predicting market sentiment.","featured":"2024-09-05","label":"RePEc","topic":"LLMs & Text","cites":null,"score":10,"scale":"shares"},{"title":"EUR-USD Exchange Rate Forecasting Based on Information Fusion with Large Language Models and Deep Learning Methods","url":"/papers/arxiv/2408.13214/","summary":"The paper presents a new framework, IUS, that merges unstructured text and structured financial data to improve the accuracy of EUR/USD exchange rate forecasts.","featured":"2024-08-28","label":"arXiv","topic":"LLMs & Text","cites":3,"score":3,"scale":"shares"},{"title":"Controllable Text Generation for Large Language Models: A Survey","url":"/papers/arxiv/2408.12599/","summary":"The article discusses the progress in Controllable Text Generation for Large Language Models, outlining methods, challenges, and future research directions.","featured":"2024-08-28","label":"Machine learning","topic":"LLMs & Text","cites":90,"score":33,"scale":"shares"},{"title":"An In-Depth Investigation of Data Collection in LLM App Ecosystems","url":"/papers/arxiv/2408.13247/","summary":"Research shows that third-party language learning model apps like OpenAI's GPT collect extensive user data, including sensitive information, posing a privacy risk due to lax data policies.","featured":"2024-08-28","label":"Machine learning","topic":"LLMs & Text","cites":22,"score":14,"scale":"shares"},{"title":"LLM Pruning and Distillation in Practice: The Minitron Approach","url":"/papers/arxiv/2408.11796/","summary":"The article discusses the successful compression of Llama 3.1 8B and MistralNeMo 12B models to smaller parameters using pruning and distillation strategies, with the results tested on common benchmarks and the base model weights made available on Hugging Face.","featured":"2024-08-28","label":"Machine learning","topic":"LLMs & Text","cites":111,"score":372,"scale":"shares"},{"title":"Beyond Scale: The Diversity Coefficient as a Data Quality Metric for Variability in Natural Language Data","url":"/papers/arxiv/2306.13840/","summary":"The study introduces a measure called the diversity coefficient to formalize data quality in pre-training Large Language Models (LLMs), demonstrating its alignment with diversity and variability properties, and its usefulness in evaluating downstream model performance.","featured":"2024-08-28","label":"Machine learning","topic":"LLMs & Text","cites":21,"score":50,"scale":"shares"},{"title":"Topics as Entity Clusters: Entity-based Topics from Large Language Models and Graph Neural Networks","url":"/papers/arxiv/2301.02458/","summary":"The study uses bimodal vector representations to improve entity-based neural topic modeling, resulting in better coherency metrics than existing models.","featured":"2024-08-28","label":"Machine learning","topic":"LLMs & Text","cites":3,"score":42,"scale":"shares"},{"title":"Understanding Reference Policies in Direct Preference Optimization","url":"/papers/arxiv/2407.13709/","summary":"The research reveals that Direct Preference Optimization in large language models is sensitive to the KL divergence constraint and performs better with stronger reference policies.","featured":"2024-08-28","label":"Machine learning","topic":"LLMs & Text","cites":23,"score":18,"scale":"shares"},{"title":"Korean Stock Market Speculation","url":"/papers/ssrn/4927509/","summary":"The study finds that in the Korean stock market, speculative sentiment in leverage exchange-traded funds (ETFs) negatively correlates and predicts market returns after 3 months.","featured":"2024-08-21","label":"SSRN","topic":"LLMs & Text","cites":null,"score":4,"scale":"shares"},{"title":"LongVILA: Scaling Long-Context Visual Language Models for Long Videos","url":"/papers/arxiv/2408.10188/","summary":"The authors present LongVILA, a solution for long-context vision-language models, which includes a system for Multi-Modal Sequence Parallelism, a five-stage training pipeline, and large-scale pre-training datasets, improving performance on long videos.","featured":"2024-08-21","label":"Machine learning","topic":"LLMs & Text","cites":345,"score":45,"scale":"shares"},{"title":"Can Large Language Models Understand Symbolic Graphics Programs?","url":"/papers/arxiv/2408.08313/","summary":"The research evaluates the ability of large language models to understand symbolic graphics programs, introducing a benchmark for semantic understanding and a method called Symbolic Instruction Tuning to enhance this capability.","featured":"2024-08-21","label":"Machine learning","topic":"LLMs & Text","cites":42,"score":37,"scale":"shares"},{"title":"Causal Reasoning and Large Language Models: Opening a New Frontier for Causality","url":"/papers/arxiv/2305.00050/","summary":"Large language models (LLMs) have shown a high probability of generating accurate causal arguments, outperforming existing methods, and can assist experts in causal analysis, despite occasional unpredictable failures.","featured":"2024-08-21","label":"Machine learning","topic":"LLMs & Text","cites":571,"score":1546,"scale":"shares"},{"title":"Self-Taught Optimizer (STOP): Recursively Self-Improving Code Generation","url":"/papers/arxiv/2310.02304/","summary":"A program using a modern language model, GPT-4, has demonstrated the ability to write code that can self-improve, showcasing the potential of language-model-infused scaffolding.","featured":"2024-08-21","label":"Machine learning","topic":"LLMs & Text","cites":144,"score":404,"scale":"shares"},{"title":"Idea2Img: Iterative Self-Refinement with GPT-4V(ision) for Automatic Image Design and Generation","url":"/papers/arxiv/2310.08541/","summary":"The Idea to Image system, utilizing GPT-4V(ision), allows for efficient conversion of abstract ideas into text-to-image prompts, resulting in images with superior semantic and visual qualities.","featured":"2024-08-21","label":"Machine learning","topic":"LLMs & Text","cites":37,"score":183,"scale":"shares"},{"title":"What is in Your Safe Data? Identifying Benign Data that Breaks Safety","url":"/papers/arxiv/2404.01099/","summary":"Research suggests that Large Language Models (LLMs) can be compromised by benign data, proposing a bi-directional anchoring method to identify such data and maintain model safety.","featured":"2024-08-21","label":"Machine learning","topic":"LLMs & Text","cites":124,"score":149,"scale":"shares"},{"title":"Apollo: A Lightweight Multilingual Medical LLM towards Democratizing Medical AI to 6B People","url":"/papers/arxiv/2403.03640/","summary":"Multilingual Medical LLM: A multilingual medical dataset and benchmark have been created to expand medical AI to non-English speakers, with the Apollo models showing the best performance.","featured":"2024-08-21","label":"Machine learning","topic":"LLMs & Text","cites":59,"score":104,"scale":"shares"},{"title":"Agent Instructs Large Language Models to be General Zero-Shot Reasoners","url":"/papers/arxiv/2310.03710/","summary":"An autonomous agent has been developed to enhance the zero-shot reasoning capabilities of large language models, achieving top performance on various tasks.","featured":"2024-08-21","label":"Machine learning","topic":"LLMs & Text","cites":46,"score":93,"scale":"shares"},{"title":"ChemVLM: Exploring the Power of Multimodal Large Language Models in Chemistry Area","url":"/papers/arxiv/2408.07246/","summary":"Multimodal Chemistry Model: ChemVLM, a chemical multimodal large language model, has been introduced to manage visual information in the chemical field, showing competitive performance in different tasks.","featured":"2024-08-21","label":"Machine learning","topic":"LLMs & Text","cites":103,"score":58,"scale":"shares"},{"title":"Sentiment Analysis for Tesla Stock","url":"/papers/ssrn/4924701/","summary":"The project uses sentiment analysis of Tesla-related tweets and an optimized Long Short-Term Memory network to accurately predict Tesla's stock closing price.","featured":"2024-08-15","label":"SSRN","topic":"LLMs & Text","cites":null,"score":4,"scale":"shares"},{"title":"Sentiment Analysis in Business","url":"/papers/ssrn/4926226/","summary":"The article highlights the growing role of Sentiment Analysis in business, enabling companies to leverage customer feedback for expansion and improvement.","featured":"2024-08-15","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"HybridRAG: Integrating Knowledge Graphs and Vector Retrieval Augmented Generation for Efficient Information Extraction","url":"/papers/arxiv/2408.04948/","summary":"Q&A Systems for Financial Data: HybridRAG, a new method combining Knowledge Graphs and VectorRAG techniques, improves question-answer systems for extracting information from financial documents, offering better accuracy and answer generation.","featured":"2024-08-15","label":"arXiv","topic":"LLMs & Text","cites":231,"score":150,"scale":"shares"},{"title":"From Text to Insight: Leveraging Large Language Models for Performance Evaluation in Management","url":"/papers/arxiv/2408.05328/","summary":"LLMs for Organizational Evaluations: Large Language Models, particularly GPT-4, are a dependable substitute for human raters in assessing knowledge-based performance, showing higher consistency and reliability, but are susceptible to contextual biases like the halo effect.","featured":"2024-08-15","label":"arXiv","topic":"LLMs & Text","cites":8,"score":2,"scale":"shares"},{"title":"Can We Rely on LLM Agents to Draft Long-Horizon Plans? Let's Take TravelPlanner as an Example","url":"/papers/arxiv/2408.06318/","summary":"The study suggests a Feedback-Aware Fine-Tuning (FAFT) method to improve Large Language Models' (LLMs) performance in real-world planning tasks by using both positive and negative feedback.","featured":"2024-08-15","label":"Machine learning","topic":"LLMs & Text","cites":30,"score":5,"scale":"shares"},{"title":"KIF: Knowledge Identification and Fusion for Language Model Continual Learning","url":"/papers/arxiv/2408.05200/","summary":"The paper presents Task Skill Localization and Consolidation (TaSL), a new framework for language models that enhances knowledge transfer and prevents forgetting, improving the balance between old and new knowledge.","featured":"2024-08-15","label":"Machine learning","topic":"LLMs & Text","cites":11,"score":4,"scale":"shares"},{"title":"Long-Form Answers to Visual Questions from Blind and Low Vision People","url":"/papers/arxiv/2408.06303/","summary":"The study introduces VizWiz-LF, a dataset of long-form answers to visual questions asked by blind and low vision users, and assesses the ability of vision language models to provide accurate and useful responses.","featured":"2024-08-15","label":"Machine learning","topic":"LLMs & Text","cites":28,"score":3,"scale":"shares"},{"title":"Self-Taught Evaluators","url":"/papers/arxiv/2408.02666/","summary":"The Self-Taught Evaluator approach enhances model evaluators using only synthetic training data, surpassing models like GPT-4.","featured":"2024-08-15","label":"Machine learning","topic":"LLMs & Text","cites":68,"score":410,"scale":"shares"},{"title":"RAGGED: Towards Informed Design of Scalable and Stable RAG Systems","url":"/papers/arxiv/2403.09040/","summary":"RAGGED, a new framework, optimizes language models for document-based question answering by analyzing Retrieval-augmented generation configurations.","featured":"2024-08-15","label":"Machine learning","topic":"LLMs & Text","cites":24,"score":222,"scale":"shares"},{"title":"Eliciting Latent Knowledge from Quirky Language Models","url":"/papers/arxiv/2312.01037/","summary":"New quirky language models and datasets are developed to improve Eliciting Latent Knowledge research, aiding in extracting reliable knowledge from untrusted models.","featured":"2024-08-15","label":"Machine learning","topic":"LLMs & Text","cites":63,"score":176,"scale":"shares"},{"title":"Know Your Limits: A Survey of Abstention in Large Language Models","url":"/papers/arxiv/2407.18418/","summary":"A new study on abstention in large language models identifies areas for future work to enhance abstention abilities based on the query, the model, and human values.","featured":"2024-08-15","label":"Machine learning","topic":"LLMs & Text","cites":148,"score":99,"scale":"shares"},{"title":"Better Alignment with Instruction Back-and-Forth Translation","url":"/papers/arxiv/2408.04614/","summary":"A new method, instruction back-and-forth translation, is introduced for creating high-quality synthetic data to improve large language models, outperforming other datasets on AlpacaEval.","featured":"2024-08-15","label":"Machine learning","topic":"LLMs & Text","cites":18,"score":30,"scale":"shares"},{"title":"Multi-Conditional Ranking with Large Language Models","url":"/papers/arxiv/2404.00211/","summary":"MCRank, a new benchmark for evaluating multi-conditional ranking in recommendation systems, is introduced, along with a decomposed reasoning method that boosts large language models' performance by 12%.","featured":"2024-08-15","label":"Machine learning","topic":"LLMs & Text","cites":2,"score":27,"scale":"shares"},{"title":"Transformer Explainer: Learning LLM Transformers with Interactive Visual Explanation and Experimentation","url":"/papers/arxiv/2408.04619/","summary":"Transformer Explainer, an interactive tool, is unveiled to help non-experts understand Transformers through the GPT-2 model, allowing real-time user input experimentation.","featured":"2024-08-15","label":"Machine learning","topic":"LLMs & Text","cites":16,"score":19,"scale":"shares"},{"title":"Greek Debt Crisis Narratives","url":"/papers/repec/zbw-mpifgd-300665/","summary":"During the 2009-2015 Greek debt crisis, negative future narratives identified through text mining of newspaper articles influenced the spread of Greek bonds, indicating that perceived futures can affect investor behavior and lead to financial crises.","featured":"2024-08-15","label":"RePEc","topic":"LLMs & Text","cites":null,"score":17,"scale":"shares"},{"title":"Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters","url":"/papers/arxiv/2408.03314/","summary":"The research investigates enhancing Large Language Models' (LLMs) performance using more test-time computation, suggesting a compute-optimal scaling strategy based on prompt difficulty.","featured":"2024-08-07","label":"Machine learning","topic":"LLMs & Text","cites":2189,"score":214,"scale":"shares"},{"title":"Language Model Can Listen While Speaking","url":"/papers/arxiv/2408.02622/","summary":"The paper presents a new model, the listening-while-speaking language model (LSLM), that improves real-time interaction in speech language models, including handling interruptions.","featured":"2024-08-07","label":"Machine learning","topic":"LLMs & Text","cites":83,"score":131,"scale":"shares"},{"title":"AgentGen: Enhancing Planning Abilities for Large Language Model based Agent via Environment and Task Generation","url":"/papers/arxiv/2408.00764/","summary":"Enhancing LLM Planning: The study improves the planning abilities of Large Language Models (LLMs) using instruction tuning and a framework called AgentGen, which generates diverse environments and planning tasks.","featured":"2024-08-07","label":"Machine learning","topic":"LLMs & Text","cites":93,"score":44,"scale":"shares"},{"title":"Tamper-Resistant Safeguards for Open-Weight LLMs","url":"/papers/arxiv/2408.00761/","summary":"The research introduces a method called TAR to build tamper-resistant safeguards into Large Language Models (LLMs), improving tamper-resistance while maintaining benign capabilities.","featured":"2024-08-07","label":"Machine learning","topic":"LLMs & Text","cites":150,"score":43,"scale":"shares"},{"title":"Coarse Correspondences Boost Spatial-Temporal Reasoning in Multimodal Language Model","url":"/papers/arxiv/2408.00754/","summary":"The paper presents Coarse Correspondence, a visual prompting method that enhances multimodal language models' understanding of 3D and temporal dimensions, achieving top results on various benchmarks.","featured":"2024-08-07","label":"Machine learning","topic":"LLMs & Text","cites":20,"score":41,"scale":"shares"},{"title":"KnowPO: Knowledge-aware Preference Optimization for Controllable Knowledge Selection in Retrieval-Augmented Language Models","url":"/papers/arxiv/2408.03297/","summary":"The study introduces a Knowledge-aware Preference Optimization method to improve large language models' knowledge selection, showing enhanced performance in managing knowledge conflicts and robust generalization across different datasets.","featured":"2024-08-07","label":"Machine learning","topic":"LLMs & Text","cites":22,"score":10,"scale":"shares"},{"title":"MoMa: Efficient Early-Fusion Pre-training with Mixture of Modality-Aware Experts","url":"/papers/arxiv/2407.21770/","summary":"The MoMa model is a new architecture designed for pre-training mixed-modal language models, providing improved efficiency in processing images and text in any order.","featured":"2024-08-07","label":"Machine learning","topic":"LLMs & Text","cites":78,"score":348,"scale":"shares"},{"title":"SteP: Stacked LLM Policies for Web Actions","url":"/papers/arxiv/2310.03720/","summary":"Stacked LLM Policies for Web Actions (SteP) is a dynamic policy composition approach that enhances performance in solving diverse web tasks and adapts to task complexity.","featured":"2024-08-07","label":"Machine learning","topic":"LLMs & Text","cites":72,"score":49,"scale":"shares"},{"title":"Small Molecule Optimization with Large Language Models","url":"/papers/arxiv/2407.18897/","summary":"Chemlactica and Chemma are language models fine-tuned on a corpus of 110M molecules, excelling in generating molecules with specific properties and predicting new molecular traits.","featured":"2024-07-31","label":"Machine learning","topic":"LLMs & Text","cites":9,"score":48,"scale":"shares"},{"title":"Recursive Introspection: Teaching Language Model Agents How to Self-Improve","url":"/papers/arxiv/2407.18219/","summary":"RISE is a method for refining large language models to improve their responses and correct errors over time, with notable success in math reasoning tasks.","featured":"2024-07-31","label":"Machine learning","topic":"LLMs & Text","cites":211,"score":15,"scale":"shares"},{"title":"Wolf: Dense Video Captioning with a World Summarization Framework","url":"/papers/arxiv/2407.18908/","summary":"Wolf, a new video captioning framework, uses Vision Language Models to efficiently summarize information, outperforming existing methods and setting a new standard for video captioning.","featured":"2024-07-31","label":"Machine learning","topic":"LLMs & Text","cites":6,"score":9,"scale":"shares"},{"title":"ShiftAddLLM: Accelerating Pretrained LLMs via Post-Training Multiplication-Less Reparameterization","url":"/papers/arxiv/2406.05981/","summary":"ShiftAddLLM is a new method developed to speed up large language models on devices with limited resources by replacing complex multiplications with simpler operations, thus reducing memory usage and latency and enhancing model performance.","featured":"2024-07-31","label":"Machine learning","topic":"LLMs & Text","cites":47,"score":170,"scale":"shares"},{"title":"Block Verification Accelerates Speculative Decoding","url":"/papers/arxiv/2403.10444/","summary":"Speculative Decoding: The paper presents Block Verification, a new verification algorithm for large language models that checks a whole block of tokens at once, offering slight but consistent speed improvements over the standard token verification algorithm without adding to code complexity.","featured":"2024-07-31","label":"Machine learning","topic":"LLMs & Text","cites":31,"score":27,"scale":"shares"},{"title":"Boosted Return with News","url":"/papers/ssrn/4900825/","summary":"The article uses XGBoost to predict next-day volatility jumps based on over 1400 news topics, improving portfolio performance.","featured":"2024-07-24","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Robinhood Stock Impact","url":"/papers/ssrn/4900783/","summary":"Despite their lower investment sophistication, Robinhood investors significantly contribute to liquidity during earnings and M&A announcements, trading more based on sentiments and showing higher demand elasticity.","featured":"2024-07-24","label":"SSRN","topic":"LLMs & Text","cites":null,"score":3,"scale":"shares"},{"title":"SlowFast-LLaVA: A Strong Training-Free Baseline for Video Large Language Models","url":"/papers/arxiv/2407.15841/","summary":"SlowFast-LLaVA is a new video language model that excels in capturing spatial semantics and temporal context in videos, surpassing other methods in various video tasks.","featured":"2024-07-24","label":"Machine learning","topic":"LLMs & Text","cites":139,"score":56,"scale":"shares"},{"title":"ChatQA 2: Bridging the Gap to Proprietary LLMs in Long Context and RAG Capabilities","url":"/papers/arxiv/2407.14482/","summary":"Bridging the Gap: ChatQA 2 is a model that improves long-context understanding and retrieval-augmented generation, matching the accuracy of top proprietary models.","featured":"2024-07-24","label":"Machine learning","topic":"LLMs & Text","cites":57,"score":36,"scale":"shares"},{"title":"AWQ: Activation-aware Weight Quantization for On-Device LLM Compression and Acceleration","url":"/papers/arxiv/2306.00978/","summary":"The study suggests Activation-aware Weight Quantization (AWQ), a hardware-friendly method for quantizing large language models that reduces error and improves performance on various benchmarks.","featured":"2024-07-24","label":"Machine learning","topic":"LLMs & Text","cites":1800,"score":146,"scale":"shares"},{"title":"Dynamic Memory Compression: Retrofitting LLMs for Accelerated Inference","url":"/papers/arxiv/2403.09636/","summary":"The piece presents Dynamic Memory Compression (DMC), a method for compressing key-value cache in large language models that increases throughput and maintains performance while accommodating larger contexts and batches within a given memory budget.","featured":"2024-07-24","label":"Machine learning","topic":"LLMs & Text","cites":133,"score":106,"scale":"shares"},{"title":"On the Effect of (Near) Duplicate Subwords in Language Modelling","url":"/papers/arxiv/2404.06508/","summary":"The research examines the effect of near duplicate subwords on language model training, revealing that while duplication hinders efficiency, merging similar duplicates can also harm performance.","featured":"2024-07-24","label":"Machine learning","topic":"LLMs & Text","cites":6,"score":71,"scale":"shares"},{"title":"Can You Learn Semantics Through Next-Word Prediction? The Case of Entailment","url":"/papers/arxiv/2402.13956/","summary":"The study explores if language models infer text meaning from training data patterns, discovering they can decode sentence relations, but the prediction test works contrary to the theoretical test due to text redundancy.","featured":"2024-07-24","label":"Machine learning","topic":"LLMs & Text","cites":16,"score":29,"scale":"shares"},{"title":"Calm Markets, Confident PE Executives: Private Equity Earnings Calls Reflect Renewed Optimism in Q1’24","url":"/papers/ssrn/4891919/","summary":"Confidence among private equity (PE) firms has significantly increased over the past two years, with Q1 2024 showing renewed optimism despite market fluctuations.","featured":"2024-07-17","label":"SSRN","topic":"LLMs & Text","cites":0,"score":3,"scale":"shares"},{"title":"Fine-Tuning and Prompt Optimization: Two Great Steps that Work Better Together","url":"/papers/arxiv/2407.10930/","summary":"The article explores a method to enhance Natural Language Processing systems by simultaneously optimizing language model weights and prompting strategies, leading to significant improvements in tasks like multi-hop QA and mathematical reasoning.","featured":"2024-07-17","label":"Machine learning","topic":"LLMs & Text","cites":59,"score":60,"scale":"shares"},{"title":"Human-inspired Episodic Memory for Infinite Context LLMs","url":"/papers/arxiv/2407.09450/","summary":"The study presents EM-LLM, a new approach that incorporates elements of human episodic memory into Large Language Models, enabling them to manage infinite context lengths efficiently and outperform existing models in tasks like PassageRetrieval.","featured":"2024-07-17","label":"Machine learning","topic":"LLMs & Text","cites":69,"score":28,"scale":"shares"},{"title":"Real-Time Anomaly Detection and Reactive Planning with Large Language Models","url":"/papers/arxiv/2407.08735/","summary":"The paper introduces a two-stage reasoning framework for identifying and mitigating out-of-distribution failure modes in robotic systems using large language models, comprising a quick binary anomaly classifier and a slower fallback selection stage.","featured":"2024-07-17","label":"Machine learning","topic":"LLMs & Text","cites":105,"score":17,"scale":"shares"},{"title":"Agent Lumos: Unified and Modular Training for Open-Source Language Agents","url":"/papers/arxiv/2311.05657/","summary":"LUMOS, an open-source framework for training Large Language Models (LLMs), is introduced in the article, demonstrating superior performance and adaptability to new tasks.","featured":"2024-07-17","label":"Machine learning","topic":"LLMs & Text","cites":84,"score":336,"scale":"shares"},{"title":"Trial and Error: Exploration-Based Trajectory Optimization for LLM Agents","url":"/papers/arxiv/2403.02502/","summary":"The study explores an exploration-based trajectory optimization (ETO) method to enhance the performance of Large Language Models (LLMs) by learning from their exploration mistakes.","featured":"2024-07-17","label":"Machine learning","topic":"LLMs & Text","cites":206,"score":245,"scale":"shares"},{"title":"An Investigation on LLMs' Visual Understanding Ability Using SVG for Image-Text Bridging","url":"/papers/arxiv/2306.06094/","summary":"The research examines the capability of Large Language Models (LLMs) to interpret images by transforming them into Scalable Vector Graphics (SVG) and assessing the LLMs on various computer vision tasks.","featured":"2024-07-17","label":"Machine learning","topic":"LLMs & Text","cites":8,"score":65,"scale":"shares"},{"title":"FACTS About Building Retrieval Augmented Generation-based Chatbots","url":"/papers/arxiv/2407.07858/","summary":"The article introduces the FACTS framework for developing Retrieval Augmented Generation (RAG)-based chatbots, and presents empirical results on the balance between accuracy and latency in large and small LLMs.","featured":"2024-07-17","label":"Machine learning","topic":"LLMs & Text","cites":37,"score":51,"scale":"shares"},{"title":"Benchmarking Generation and Evaluation Capabilities of Large Language Models for Instruction Controllable Summarization","url":"/papers/arxiv/2311.09184/","summary":"Research shows that large language models struggle with text summarization, often making factual errors and showing significant performance gaps.","featured":"2024-07-17","label":"Machine learning","topic":"LLMs & Text","cites":98,"score":47,"scale":"shares"},{"title":"FinBERT and LSTM for Stock Price Prediction","url":"/papers/repec/ids-ijecbr-v-28-y-2024-i-1-p-1-16/","summary":"The article discusses a hybrid model that combines BERT and LSTM for predicting stock prices. This model surpasses traditional methods by including financial news sentiment analysis and technical indicators, allowing for accurate predictions of significant stock price fluctuations.","featured":"2024-07-17","label":"RePEc","topic":"LLMs & Text","cites":null,"score":12,"scale":"shares"},{"title":"War in Ukraine Analysis","url":"/papers/repec/bla-rgscpp-v-15-y-2023-i-1-p-56-74/","summary":"The research examines the topics and sentiments of Ukrainian Telegram users during the early stages of the war in Ukraine, emphasizing the importance of social media analytics.","featured":"2024-07-17","label":"RePEc","topic":"LLMs & Text","cites":null,"score":9,"scale":"shares"},{"title":"Vision-Language Models under Cultural and Inclusive Considerations","url":"/papers/arxiv/2407.06177/","summary":"A new benchmark has been proposed to evaluate the reliability of large vision-language models as visual aids for visually impaired individuals in diverse cultural settings.","featured":"2024-07-10","label":"Machine learning","topic":"LLMs & Text","cites":14,"score":9,"scale":"shares"},{"title":"InternLM-XComposer-2.5: A Versatile Large Vision Language Model Supporting Long-Contextual Input and Output","url":"/papers/arxiv/2407.03320/","summary":"A Versatile Large Vision Language Model: InternLM-XComposer-2.5, a large-vision language model, excels in text-image comprehension and composition, outperforming existing models on 16 benchmarks.","featured":"2024-07-10","label":"Machine learning","topic":"LLMs & Text","cites":210,"score":204,"scale":"shares"},{"title":"Found in the Middle: Calibrating Positional Attention Bias Improves Long Context Utilization","url":"/papers/arxiv/2406.16008/","summary":"The found-in-the-middle calibration mechanism addresses the lost-in-the-middle problem in large language models, improving the model's ability to locate relevant information and enhancing performance.","featured":"2024-07-10","label":"Machine learning","topic":"LLMs & Text","cites":131,"score":184,"scale":"shares"},{"title":"mPLM-Sim: Better Cross-Lingual Similarity and Transfer in Multilingual Pretrained Language Models","url":"/papers/arxiv/2305.13684/","summary":"Cross-Lingual Similarity: The study introduces mPLMSim, a language similarity measure tool that enhances cross-lingual transfer performance by 1%-2% using multilingual pretrained language models.","featured":"2024-07-10","label":"Machine learning","topic":"LLMs & Text","cites":1,"score":25,"scale":"shares"},{"title":"Merlin: Empowering Multimodal LLMs with Foresight Minds","url":"/papers/arxiv/2312.00589/","summary":"Foresight Minds: The paper presents the integration of future modeling into Multimodal Large Language Models (MLLMs) to improve their predictive capabilities, resulting in a new MLLM called Merlin.","featured":"2024-07-10","label":"Machine learning","topic":"LLMs & Text","cites":52,"score":16,"scale":"shares"},{"title":"Multimodal Foundation Models in Auditing Practices","url":"/papers/ssrn/4881256/","summary":"The article proposes an AI-based multimodal auditing system that integrates data from diverse documents and allows auditors to complete tasks using natural language, demonstrating its feasibility through a simulation case study.","featured":"2024-07-03","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Scaling Synthetic Data Creation with 1,000,000,000 Personas","url":"/papers/arxiv/2406.20094/","summary":"A new method for creating synthetic data uses 1 billion diverse personas, potentially transforming large language model research and development.","featured":"2024-07-03","label":"Machine learning","topic":"LLMs & Text","cites":466,"score":258,"scale":"shares"},{"title":"The Model Arena for Cross-lingual Sentiment Analysis: A Comparative Study in the Era of Large Language Models","url":"/papers/arxiv/2406.19358/","summary":"The study finds Small Multilingual Language Models (SMLM) excel in zero-shot cross-lingual sentiment analysis, while Large Language Models (LLM) perform better in few-shot scenarios.","featured":"2024-07-03","label":"Machine learning","topic":"LLMs & Text","cites":15,"score":11,"scale":"shares"},{"title":"The Remarkable Robustness of LLMs: Stages of Inference?","url":"/papers/arxiv/2406.19384/","summary":"The study reveals that Large Language Models with more layers are more robust, and identifies four universal stages of inference across different models.","featured":"2024-07-03","label":"Machine learning","topic":"LLMs & Text","cites":153,"score":7,"scale":"shares"},{"title":"Does Writing with Language Models Reduce Content Diversity?","url":"/papers/arxiv/2309.05196/","summary":"InstructGPT, a large language model, can reduce diversity in collaborative writing, making different authors' writings more similar and decreasing overall content variety.","featured":"2024-07-03","label":"Machine learning","topic":"LLMs & Text","cites":244,"score":149,"scale":"shares"},{"title":"Large Language Models Assume People are More Rational than We Really are","url":"/papers/arxiv/2406.17055/","summary":"Large Language Models (LLMs) inaccurately perceive human decision-making as more rational than it is, aligning more with expected value theory, despite seeming to mimic human behavior.","featured":"2024-07-03","label":"Machine learning","topic":"LLMs & Text","cites":63,"score":133,"scale":"shares"},{"title":"AutoMix: Automatically Mixing Language Models","url":"/papers/arxiv/2310.12963/","summary":"Language Model Optimization: Automix is a new method that saves computational cost by over 50% by directing queries to larger language models based on the accuracy of outputs from smaller models.","featured":"2024-07-03","label":"Machine learning","topic":"LLMs & Text","cites":129,"score":41,"scale":"shares"},{"title":"Detecting Misreported Accounting: A Machine Learning Approach using Text Data","url":"/papers/ssrn/4867498/","summary":"A new method using machine learning and text data from 10K reports can predict the likelihood of a firm misreporting financial information, with higher probabilities associated with higher risks and costs.","featured":"2024-06-20","label":"SSRN","topic":"LLMs & Text","cites":1,"score":2,"scale":"shares"},{"title":"Application of Natural Language Processing in Financial Risk Detection","url":"/papers/arxiv/2406.09765/","summary":"The study shows that Natural Language Processing (NLP) can be effectively used to detect and predict potential risks in financial documents and communications.","featured":"2024-06-20","label":"arXiv","topic":"LLMs & Text","cites":22,"score":4,"scale":"shares"},{"title":"How Do Large Language Models Acquire Factual Knowledge During Pretraining?","url":"/papers/arxiv/2406.11813/","summary":"Large language models (LLMs) learn factual knowledge during pretraining, but this knowledge is often forgotten in subsequent steps.","featured":"2024-06-20","label":"Machine learning","topic":"LLMs & Text","cites":118,"score":124,"scale":"shares"},{"title":"What Are the Odds? Language Models Are Capable of Probabilistic Reasoning","url":"/papers/arxiv/2406.12830/","summary":"Language models have difficulty with numerical reasoning and understanding probability distributions, but can improve with real-world context and simplified assumptions.","featured":"2024-06-20","label":"Machine learning","topic":"LLMs & Text","cites":29,"score":73,"scale":"shares"},{"title":"Unveiling Encoder-Free Vision-Language Models","url":"/papers/arxiv/2406.11832/","summary":"The EVE model, a vision-language model without an encoder, performs well across multiple benchmarks, offering an efficient way to develop a decoder-only architecture.","featured":"2024-06-20","label":"Machine learning","topic":"LLMs & Text","cites":102,"score":73,"scale":"shares"},{"title":"From RAGs to rich parameters: Probing how language models utilize external knowledge over parametric information for factual queries","url":"/papers/arxiv/2406.12824/","summary":"Retrieval Augmented Generation (RAG) enhances language models' reasoning abilities using external context, but models tend to rely heavily on this context and less on their parametric memory.","featured":"2024-06-20","label":"Machine learning","topic":"LLMs & Text","cites":20,"score":37,"scale":"shares"},{"title":"Language Modeling with Editable External Knowledge","url":"/papers/arxiv/2406.11830/","summary":"The paper presents ERASE, a method that improves the performance of retrieval-augmented generation models by incrementally modifying the knowledge base when new documents are added.","featured":"2024-06-20","label":"Machine learning","topic":"LLMs & Text","cites":9,"score":16,"scale":"shares"},{"title":"Scalable MatMul-free Language Modeling","url":"/papers/arxiv/2406.02528/","summary":"Scientists have developed a method to remove matrix multiplication from large language models, reducing memory usage by up to 61% during training and over 10x during inference, making these models more efficient.","featured":"2024-06-20","label":"Machine learning","topic":"LLMs & Text","cites":45,"score":610,"scale":"shares"},{"title":"Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks","url":"/papers/arxiv/2404.02151/","summary":"Research reveals that even the latest safety-aligned large language models are susceptible to simple adaptive jailbreaking attacks, with almost 100% success rate, emphasizing the significance of adaptivity in these attacks.","featured":"2024-06-20","label":"Machine learning","topic":"LLMs & Text","cites":580,"score":214,"scale":"shares"},{"title":"Simple and Effective Masked Diffusion Language Models","url":"/papers/arxiv/2406.07524/","summary":"The performance of diffusion models in language modeling has been enhanced by using an effective training recipe and a simplified objective, setting a new standard among diffusion models.","featured":"2024-06-12","label":"Machine learning","topic":"LLMs & Text","cites":850,"score":180,"scale":"shares"},{"title":"Verbalized Machine Learning: Revisiting Machine Learning with Language Models","url":"/papers/arxiv/2406.04344/","summary":"The verbalized machine learning (VML) framework uses large language models parameterized by text prompts to solve traditional machine learning problems, providing easy encoding of inductive bias, automatic model class selection, and interpretable learner updates.","featured":"2024-06-12","label":"Machine learning","topic":"LLMs & Text","cites":20,"score":58,"scale":"shares"},{"title":"ArtPrompt: ASCII Art-based Jailbreak Attacks against Aligned LLMs","url":"/papers/arxiv/2402.11753/","summary":"ArtPrompt, a new ASCII art-based attack, exploits vulnerabilities in large language models, bypassing safety measures and causing unwanted behaviors.","featured":"2024-06-12","label":"Machine learning","topic":"LLMs & Text","cites":294,"score":9360,"scale":"shares"},{"title":"RAFT: Adapting Language Model to Domain Specific RAG","url":"/papers/arxiv/2403.10131/","summary":"Retrieval Augmented FineTuning (RAFT) is a new training method that enhances large language models' ability to answer domain-specific questions by training them to ignore irrelevant documents and cite relevant ones.","featured":"2024-06-12","label":"Machine learning","topic":"LLMs & Text","cites":401,"score":260,"scale":"shares"},{"title":"SafeDecoding: Defending against Jailbreak Attacks via Safety-Aware Decoding","url":"/papers/arxiv/2402.08983/","summary":"The paper introduces SafeDecoding, a strategy to protect large language models from jailbreak attacks, maintaining the quality of responses to user queries while reducing attack success rate.","featured":"2024-06-12","label":"Machine learning","topic":"LLMs & Text","cites":300,"score":103,"scale":"shares"},{"title":"Large Language Models(LLMs) on Tabular Data: Prediction, Generation, and Understanding - A Survey","url":"/papers/arxiv/2402.17944/","summary":"The survey reviews the use of large language models in tabular data modeling, summarizing key techniques, metrics, datasets, models, and suggesting areas for future research.","featured":"2024-06-12","label":"Machine learning","topic":"LLMs & Text","cites":272,"score":61,"scale":"shares"},{"title":"Artificial intelligence, natural language processing, and machine learning to enhance e-service quality on e-commerce platforms","url":"/papers/ssrn/4847952/","summary":"The paper discusses how AI technologies like chatbots and personalized recommendation systems can improve service quality and boost sales on ecommerce platforms.","featured":"2024-06-05","label":"SSRN","topic":"LLMs & Text","cites":10,"score":4,"scale":"shares"},{"title":"Language Models Trained to do Arithmetic Predict Human Risky and Intertemporal Choice","url":"/papers/arxiv/2405.19313/","summary":"Researchers have found that Large Language Models (LLMs) pretrained on relevant arithmetic datasets can predict human behavior more accurately than many existing cognitive models.","featured":"2024-06-05","label":"arXiv","topic":"LLMs & Text","cites":12,"score":5,"scale":"shares"},{"title":"Xwin-LM: Strong and Scalable Alignment Practice for LLMs","url":"/papers/arxiv/2405.20335/","summary":"The paper presents Xwin-LM, a set of alignment methodologies for large language models, showing significant improvements in model performance and scalability.","featured":"2024-06-05","label":"Machine learning","topic":"LLMs & Text","cites":3,"score":10,"scale":"shares"},{"title":"Exploratory Preference Optimization: Harnessing Implicit Q*-Approximation for Sample-Efficient RLHF","url":"/papers/arxiv/2405.21046/","summary":"The Exploratory Preference Optimization (XPO) algorithm has been introduced for online exploration in Reinforcement Learning from Human Feedback (RLHF), potentially enhancing language model training.","featured":"2024-06-05","label":"Machine learning","topic":"LLMs & Text","cites":115,"score":7,"scale":"shares"},{"title":"RapVerse: Coherent Vocals and Whole-Body Motion Generation from Text","url":"/papers/arxiv/2405.20336/","summary":"Vocals and Motions: A new task uses a multimodal transformer model to generate 3D body motions and singing vocals from textual lyrics, ensuring a realistic blend of vocals and human motions.","featured":"2024-06-05","label":"Machine learning","topic":"LLMs & Text","cites":7,"score":5,"scale":"shares"},{"title":"ParSEL: Parameterized Shape Editing with Language","url":"/papers/arxiv/2405.20319/","summary":"Shape Editing: ParSEL, a system that allows controllable editing of high-quality 3D assets from natural language, uses Analytical Edit Propagation (AEP) to extend initial edits into complete editing programs.","featured":"2024-06-05","label":"Machine learning","topic":"LLMs & Text","cites":14,"score":5,"scale":"shares"},{"title":"Arrows of Time for Large Language Models","url":"/papers/arxiv/2401.17505/","summary":"A study shows a time asymmetry in Autoregressive Large Language Models' ability to learn natural language, explained by sparsity and computational complexity.","featured":"2024-06-05","label":"Machine learning","topic":"LLMs & Text","cites":22,"score":154,"scale":"shares"},{"title":"MAP-Neo: Highly Capable and Transparent Bilingual Large Language Model Series","url":"/papers/arxiv/2405.19327/","summary":"Bilingual LLM Series: MAP-Neo, an open-source bilingual language model with 7B parameters, performs comparably to state-of-the-art Large Language Models, with full details provided for reproduction.","featured":"2024-06-05","label":"Machine learning","topic":"LLMs & Text","cites":88,"score":115,"scale":"shares"},{"title":"Are Language Models More Like Libraries or Like Librarians? Bibliotechnism, the Novel Reference Problem, and the Attitudes of LLMs","url":"/papers/arxiv/2401.04854/","summary":"Libraries or Librarians?: The article explores bibliotechnism, arguing that while large language models can generate new text and references, they may not necessarily possess consciousness or intelligence.","featured":"2024-06-05","label":"Machine learning","topic":"LLMs & Text","cites":36,"score":106,"scale":"shares"},{"title":"Value-Incentivized Preference Optimization: A Unified Approach to Online and Offline RLHF","url":"/papers/arxiv/2405.19320/","summary":"The study presents a unified approach to reinforcement learning from human feedback for large language models, offering theoretical guarantees and practical effectiveness.","featured":"2024-06-05","label":"Machine learning","topic":"LLMs & Text","cites":75,"score":67,"scale":"shares"},{"title":"Nearest Neighbor Speculative Decoding for LLM Generation and Attribution","url":"/papers/arxiv/2405.19325/","summary":"The paper introduces Nearest Neighbor Speculative Decoding, a language modeling approach that improves generation quality and speed by incorporating real-world text spans into language model generations.","featured":"2024-06-05","label":"Machine learning","topic":"LLMs & Text","cites":28,"score":59,"scale":"shares"},{"title":"Text Spillover","url":"/papers/ssrn/4837312/","summary":"The article explores the potential of using text data analysis as an alternative measure of interconnectedness between financial institutions, suggesting it can offer valuable insights.","featured":"2024-05-28","label":"SSRN","topic":"LLMs & Text","cites":null,"score":8,"scale":"shares"},{"title":"Not All Language Model Features Are One-Dimensionally Linear","url":"/papers/arxiv/2405.14860/","summary":"The study uses sparse autoencoders to explore the multi-dimensional nature of language model representations in GPT-2 and Mistral 7B, and identifies tasks where these features solve computational problems.","featured":"2024-05-28","label":"Machine learning","topic":"LLMs & Text","cites":207,"score":6,"scale":"shares"},{"title":"Self-Play Preference Optimization for Language Model Alignment","url":"/papers/arxiv/2405.00675/","summary":"The article suggests a self-play-based method, SPPO, for language model alignment, which can effectively enhance the likelihood of the selected response and reduce that of the discarded one.","featured":"2024-05-28","label":"Machine learning","topic":"LLMs & Text","cites":278,"score":309,"scale":"shares"},{"title":"ChatGPT for Improved Investment Decisions in Portfolio Management","url":"/papers/repec/eee-finlet-v-64-y-2024-i-c-s154461232400463x/","summary":"The research shows that the Large Language Model, ChatGPT, selects more diverse and high-performing assets in portfolio management than random selection.","featured":"2024-05-28","label":"RePEc","topic":"LLMs & Text","cites":null,"score":18,"scale":"shares"},{"title":"Climate Change and Shareholder Value: Evidence from Textual Analysis and Trump’s Unexpected Victory","url":"/papers/ssrn/4831364/","summary":"Companies more susceptible to climate change saw negative market reactions after Donald Trump's unexpected 2016 election victory, showing the impact of political stances on climate change on company value and shareholder wealth.","featured":"2024-05-22","label":"SSRN","topic":"LLMs & Text","cites":18,"score":3,"scale":"shares"},{"title":"Stock and Option Market Response to ESG News","url":"/papers/ssrn/4832267/","summary":"The research finds that negative ESG incidents have minimal immediate impact on stock prices but cause increased volatility for severe incidents, particularly those related to Natural Capital.","featured":"2024-05-22","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Metacognitive Capabilities of LLMs: An Exploration in Mathematical Problem Solving","url":"/papers/arxiv/2405.12205/","summary":"The study investigates the metacognitive abilities of large language models, showing their capacity to label math questions with skill levels and improve problem-solving accuracy.","featured":"2024-05-22","label":"Machine learning","topic":"LLMs & Text","cites":96,"score":17,"scale":"shares"},{"title":"How Far Are We From AGI: Are LLMs All We Need?","url":"/papers/arxiv/2405.10313/","summary":"The paper offers an in-depth analysis of Artificial General Intelligence (AGI), detailing its definitions, objectives, development paths, and potential realization strategies.","featured":"2024-05-22","label":"Machine learning","topic":"LLMs & Text","cites":32,"score":15,"scale":"shares"},{"title":"FlashBack: Efficient Retrieval-Augmented Language Modeling for Fast Inference","url":"/papers/arxiv/2405.04065/","summary":"Efficient LM: The paper introduces FlashBack, a Retrieval-Augmented Language Modeling system that enhances inference efficiency by adding retrieved documents to the context, leading to quicker inference speed and lower costs.","featured":"2024-05-22","label":"Machine learning","topic":"LLMs & Text","cites":2,"score":82,"scale":"shares"},{"title":"DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model","url":"/papers/arxiv/2405.04434/","summary":"MoE Language Model: DeepSeek-V2, a language model with 236B parameters, offers enhanced performance and cost efficiency compared to its predecessor, ranking high among open-source models.","featured":"2024-05-22","label":"Machine learning","topic":"LLMs & Text","cites":1459,"score":51,"scale":"shares"},{"title":"Computational Legal Studies Transformation","url":"/papers/ssrn/4826144/","summary":"The article reviews the application of computational analysis techniques in empirical legal scholarship, emphasizing recent advancements in large language models and generative AI.","featured":"2024-05-15","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Measuring Populism Empirically: Sentiment and Language in AMLO’s Morning Conferences","url":"/papers/ssrn/4826094/","summary":"The study uses text analysis to examine how Mexico's President Andrés Manuel López Obrador uses his morning press briefings to propagate his populist agenda, using language that is people-focused but not anti-elite.","featured":"2024-05-15","label":"SSRN","topic":"LLMs & Text","cites":0,"score":2,"scale":"shares"},{"title":"ENHANCING ORGANIZATIONAL PERFORMANCE: HARNESSING AI AND NLP FOR USER FEEDBACK ANALYSIS IN PRODUCT DEVELOPMENT","url":"/papers/doi/10-18374-jabe-24-1-11/","summary":"The paper explores the application of AI and NLP for analyzing user feedback on heavy machine crane products, offering insights for product improvement and enhancing customer experience.","featured":"2024-05-15","label":"arXiv","topic":"LLMs & Text","cites":8,"score":3,"scale":"shares"},{"title":"Does Fine-Tuning LLMs on New Knowledge Encourage Hallucinations?","url":"/papers/arxiv/2405.05904/","summary":"Research shows that large language models have difficulty acquiring new factual knowledge through fine-tuning, learning new information slower than consistent knowledge, and are more likely to hallucinate, indicating the risks of introducing new facts through fine-tuning.","featured":"2024-05-15","label":"Machine learning","topic":"LLMs & Text","cites":323,"score":158,"scale":"shares"},{"title":"CuMo: Scaling Multimodal LLM with Co-Upcycled Mixture-of-Experts","url":"/papers/arxiv/2405.05949/","summary":"CuMo, a model that integrates Co-upcycled Top-K sparsely-gated Mixture-of-experts blocks into the vision encoder and the MLP connector, improves multimodal LLMs with minimal additional activated parameters during inference, outperforming other multimodal LLMs across various benchmarks.","featured":"2024-05-15","label":"Machine learning","topic":"LLMs & Text","cites":78,"score":55,"scale":"shares"},{"title":"You Only Cache Once: Decoder-Decoder Architectures for Language Models","url":"/papers/arxiv/2405.05254/","summary":"YOCO architecture improves large language models by reducing GPU memory usage and speeding up the prefill stage, outperforming the Transformer model.","featured":"2024-05-15","label":"Machine learning","topic":"LLMs & Text","cites":155,"score":272,"scale":"shares"},{"title":"Prices Analyst Impact on Cash Flow","url":"/papers/ssrn/4818320/","summary":"The article suggests that analyst cash flow predictions are swayed by price changes not related to cash flow news, using a model that aligns subjective beliefs data with asset pricing models.","featured":"2024-05-08","label":"SSRN","topic":"LLMs & Text","cites":null,"score":394,"scale":"shares"},{"title":"Frequency Domain Prediction","url":"/papers/ssrn/4817096/","summary":"The study employs a machine learning approach to develop a new macroeconomic index for predicting stock returns, showing its significant predictive power and economic value in asset allocation, and its complementary relationship with investor sentiment.","featured":"2024-05-08","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Prometheus 2: An Open Source Language Model Specialized in Evaluating Other Language Models","url":"/papers/arxiv/2405.01535/","summary":"Prometheus 2 is a new open-source language model evaluator that aligns more closely with human and GPT-4 judgements and can process both direct assessment and pairwise ranking formats.","featured":"2024-05-08","label":"Machine learning","topic":"LLMs & Text","cites":541,"score":361,"scale":"shares"},{"title":"FLAME: Factuality-Aware Alignment for Large Language Models","url":"/papers/arxiv/2405.01525/","summary":"The research proposes a factuality-aware alignment process for large language models, reducing false facts generation and enhancing the model's instruction-following accuracy.","featured":"2024-05-08","label":"Machine learning","topic":"LLMs & Text","cites":62,"score":8,"scale":"shares"},{"title":"Plan-Seq-Learn: Language Model Guided RL for Solving Long Horizon Robotics Tasks","url":"/papers/arxiv/2405.01534/","summary":"The paper introduces Plan-Seq-Learn, a method that uses motion planning to connect abstract language and learned low-level control for solving long-horizon robotics tasks, achieving top-tier results.","featured":"2024-05-08","label":"Machine learning","topic":"LLMs & Text","cites":105,"score":8,"scale":"shares"},{"title":"A Careful Examination of Large Language Model Performance on Grade School Arithmetic","url":"/papers/arxiv/2405.00332/","summary":"Large language models may not be truly reasoning but overfitting to specific datasets, as shown by decreased accuracy on new benchmarks.","featured":"2024-05-08","label":"Machine learning","topic":"LLMs & Text","cites":238,"score":1161,"scale":"shares"},{"title":"A Simple and Effective Pruning Approach for Large Language Models","url":"/papers/arxiv/2306.11695/","summary":"Wanda, a new method, efficiently prunes weights in Large Language Models without retraining, offering a more efficient approach to inducing sparsity in pretrained models.","featured":"2024-05-08","label":"Machine learning","topic":"LLMs & Text","cites":981,"score":721,"scale":"shares"},{"title":"Large Language Models can Accurately Predict Searcher Preferences","url":"/papers/arxiv/2309.10621/","summary":"The paper presents a new method to enhance the quality of relevance labels in search systems using large language models, proving to be more efficient and cost-effective than third-party labellers.","featured":"2024-05-08","label":"Machine learning","topic":"LLMs & Text","cites":297,"score":149,"scale":"shares"},{"title":"LUCID: LLM-Generated Utterances for Complex and Interesting Dialogues","url":"/papers/arxiv/2403.00462/","summary":"LLM-Generated Dialogues: The article introduces LUCID, an automated data generation system that creates realistic dialogues, aiming to enhance the dialogue capabilities of virtual assistants by addressing the lack of high-quality data.","featured":"2024-05-08","label":"Machine learning","topic":"LLMs & Text","cites":11,"score":37,"scale":"shares"},{"title":"Continual Learning of Large Language Models: A Comprehensive Survey","url":"/papers/arxiv/2404.16789/","summary":"The article discusses the current research on large language models (LLMs) in continual learning (CL), focusing on the challenges of integrating these models into dynamic data distributions and task structures.","featured":"2024-05-01","label":"Machine learning","topic":"LLMs & Text","cites":351,"score":31,"scale":"shares"},{"title":"Talking Nonsense: Probing Large Language Models' Understanding of Adversarial Gibberish Inputs","url":"/papers/arxiv/2404.17120/","summary":"The study examines how large language models (LLMs) respond to prompts designed to generate coherent responses from nonsensical inputs, finding that the success of this manipulation depends on the length and complexity of the target text.","featured":"2024-05-01","label":"Machine learning","topic":"LLMs & Text","cites":14,"score":16,"scale":"shares"},{"title":"Make-it-Real: Unleashing Large Multimodal Model for Painting 3D Objects with Realistic Materials","url":"/papers/arxiv/2404.16829/","summary":"3D Painting: The article presents Make-it-Real, a new approach that uses Large Language Models to enhance the visual authenticity of 3D objects and streamline the 3D content creation process.","featured":"2024-05-01","label":"Machine learning","topic":"LLMs & Text","cites":20,"score":8,"scale":"shares"},{"title":"Investor Sentiments in Stock Valuation","url":"/papers/repec/eme-rbfpps-rbf-02-2023-0037/","summary":"The study reveals the influence of two types of investor sentiments - general market-wide and individual stock-specific - on asset valuation models.","featured":"2024-05-01","label":"RePEc","topic":"LLMs & Text","cites":null,"score":8,"scale":"shares"},{"title":"Course-Skill Atlas: A national longitudinal dataset of skills taught in U.S. higher education curricula","url":"/papers/arxiv/2404.13163/","summary":"The research uses natural language processing to analyze over three million U.S. course syllabi, creating detailed skill profiles for institutions and academic majors to aid in workforce development research.","featured":"2024-04-24","label":"arXiv","topic":"LLMs & Text","cites":15,"score":5,"scale":"shares"},{"title":"From $r$ to $Q^*$: Your Language Model is Secretly a Q-Function","url":"/papers/arxiv/2404.12358/","summary":"The research explores Direct Preference Optimization (DPO) in Reinforcement Learning From Human Feedback (RLHF), showing its ability to assign credit and its similarity to search-based algorithms in language generation.","featured":"2024-04-24","label":"Machine learning","topic":"LLMs & Text","cites":273,"score":107,"scale":"shares"},{"title":"A Survey on Self-Evolution of Large Language Models","url":"/papers/arxiv/2404.14387/","summary":"The article discusses self-evolution methods in large language models, providing a conceptual framework and suggesting future improvements.","featured":"2024-04-24","label":"Machine learning","topic":"LLMs & Text","cites":83,"score":161,"scale":"shares"},{"title":"BLINK: Multimodal Large Language Models Can See but Not Perceive","url":"/papers/arxiv/2404.12390/","summary":"Multimodal LLMs Visual Perception Benchmark: The authors present Blink, a benchmark for multimodal language models that tests visual perception abilities, showing that current models struggle with these tasks.","featured":"2024-04-24","label":"Machine learning","topic":"LLMs & Text","cites":603,"score":21,"scale":"shares"},{"title":"StructLM: Towards Building Generalist Models for Structured Knowledge Grounding","url":"/papers/arxiv/2402.16671/","summary":"Knowledge Grounding: Despite the limitations of large language models in handling structured data, the new StructLM series, trained on a comprehensive dataset, outperforms task-specific models on 16 out of 18 datasets and sets new benchmarks on 8 Structured Knowledge Grounding tasks.","featured":"2024-04-24","label":"Machine learning","topic":"LLMs & Text","cites":46,"score":55,"scale":"shares"},{"title":"The Role of Sentiment in Credit Markets","url":"/papers/ssrn/4795367/","summary":"A new measure of anticipatory sentiment, created using statistical natural language processing, significantly influences macroeconomic and financial variables, including credit market stress indicators.","featured":"2024-04-17","label":"SSRN","topic":"LLMs & Text","cites":0,"score":2,"scale":"shares"},{"title":"Construction of Domain-Specified Japanese Large Language Model for Finance Through Continual Pre-Training","url":"/papers/arxiv/2404.10555/","summary":"The research focuses on developing a large language model specifically for Japanese finance, showing its enhanced performance on related benchmarks.","featured":"2024-04-17","label":"arXiv","topic":"LLMs & Text","cites":5,"score":3,"scale":"shares"},{"title":"AutoWebGLM: A Large Language Model-based Web Navigating Agent","url":"/papers/arxiv/2404.03648/","summary":"Better Web Navigation: AutoWebGLM is a new web navigation tool that surpasses GPT-4 in performance, using a unique HTML simplification algorithm and a combined human-AI method to enhance webpage understanding and browser functionality, tested using a bilingual benchmark.","featured":"2024-04-10","label":"Machine learning","topic":"LLMs & Text","cites":175,"score":7,"scale":"shares"},{"title":"player2vec: A Language Modeling Approach to Understand Player Behavior in Games","url":"/papers/arxiv/2404.04234/","summary":"Player Behavior in Games: A new technique for learning hidden user profiles from player behavior data in video and mobile games is presented, utilizing a long-range Transformer model from natural language processing, showing promising results in matching behavior event distribution.","featured":"2024-04-10","label":"Machine learning","topic":"LLMs & Text","cites":9,"score":7,"scale":"shares"},{"title":"Mind's Eye of LLMs: Visualization-of-Thought Elicits Spatial Reasoning in Large Language Models","url":"/papers/arxiv/2404.03622/","summary":"The paper presents Visualization-of-Thought prompting, a method that improves the spatial reasoning abilities of large language models by visualizing their thought processes.","featured":"2024-04-10","label":"Machine learning","topic":"LLMs & Text","cites":99,"score":14,"scale":"shares"},{"title":"Ziya2: Data-centric Learning is All LLMs Need","url":"/papers/arxiv/2311.03301/","summary":"Data-centric Learning for LLMs: The authors present Ziya2, a large language model with 13 billion parameters, focusing on pre-training techniques and data-centric optimization, outperforming other models in multiple benchmarks.","featured":"2024-04-10","label":"Machine learning","topic":"LLMs & Text","cites":27,"score":19,"scale":"shares"},{"title":"KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization","url":"/papers/arxiv/2401.18079/","summary":"KVQuant, a new method for quantizing cached KV activations in large language models, has been developed, allowing the LLaMA-7B model to be served on an 8-GPU system with minimal degradation.","featured":"2024-04-10","label":"Machine learning","topic":"LLMs & Text","cites":701,"score":64,"scale":"shares"},{"title":"ChatGLM-Math: Improving Math Problem-Solving in Large Language Models with a Self-Critique Pipeline","url":"/papers/arxiv/2404.02893/","summary":"Problem-Solving in LLMs: The authors propose a Self-Critique pipeline to enhance the mathematical problem-solving skills of large language models without affecting their language abilities, showing significant improvements in both areas.","featured":"2024-04-10","label":"Machine learning","topic":"LLMs & Text","cites":66,"score":47,"scale":"shares"},{"title":"Enhanced Equity Market Strategy","url":"/papers/ssrn/4781752/","summary":"The article introduces a novel stock market strategy that enhances performance by merging a financial stress indicator with sentiment analysis.","featured":"2024-04-03","label":"SSRN","topic":"LLMs & Text","cites":null,"score":29,"scale":"shares"},{"title":"Uncertainty in Sentiment Analysis with LLMs using QCM (Quantiles of Correlation Matrices) - Distance","url":"/papers/ssrn/4780192/","summary":"The study explores the uncertainty in sentiment scores derived from text using advanced language processing models, finding moderate uncertainty in the results.","featured":"2024-04-03","label":"SSRN","topic":"LLMs & Text","cites":1,"score":14,"scale":"shares"},{"title":"Deep News Sentiment for Finance","url":"/papers/ssrn/4779994/","summary":"The article discusses the use of neural networks to extract hidden economic factors from large news analytics data, showing superior performance in GDP growth forecasting and asset return analysis.","featured":"2024-04-03","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Media Sentiment and Asset Allocation","url":"/papers/ssrn/4778027/","summary":"US media sentiment about foreign countries affects domestic investors' international asset allocation, with negative media coverage leading to reduced flows to international mutual funds targeting the country.","featured":"2024-04-03","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Gecko: Versatile Text Embeddings Distilled from Large Language Models","url":"/papers/arxiv/2403.20327/","summary":"Compact Text Embeddings: Gecko is a new text embedding model that improves knowledge extraction from large language models, surpassing other models in the Massive Text Embedding Benchmark.","featured":"2024-04-03","label":"Machine learning","topic":"LLMs & Text","cites":90,"score":172,"scale":"shares"},{"title":"Long-form factuality in large language models","url":"/papers/arxiv/2403.18802/","summary":"The Search-Augmented Factuality Evaluator (SAFE) method uses large language models to assess the accuracy of long-form factual content, achieving superior rating performance.","featured":"2024-04-03","label":"Machine learning","topic":"LLMs & Text","cites":172,"score":323,"scale":"shares"},{"title":"Rephrase, Augment, Reason: Visual Grounding of Questions for Vision-Language Models","url":"/papers/arxiv/2310.05861/","summary":"The Rephrase, Augment and Reason (RepARe) framework enhances the performance of large vision-language models in zero-shot tasks by rephrasing questions and extracting image details.","featured":"2024-04-03","label":"Machine learning","topic":"LLMs & Text","cites":14,"score":127,"scale":"shares"},{"title":"ChatGPT for Day Trading","url":"/papers/ssrn/4771517/","summary":"ChatGPT can create profitable day trading strategies by analyzing Twitter posts and suggesting stocks to buy or sell, according to a study.","featured":"2024-03-27","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Construction of a Japanese Financial Benchmark for Large Language Models","url":"/papers/arxiv/2403.15062/","summary":"A study has created a standard for assessing large language models in Japanese and finance, with GPT-4 showing excellent performance.","featured":"2024-03-27","label":"arXiv","topic":"LLMs & Text","cites":23,"score":7,"scale":"shares"},{"title":"Strategic Responses to Technological Change: Evidence from Online Labor Markets","url":"/papers/arxiv/2403.15262/","summary":"A study on AI's impact on freelancers revealed that exposure to language modeling technologies led to more job applications and increased specialization post the introduction of ChatGPT.","featured":"2024-03-27","label":"arXiv","topic":"LLMs & Text","cites":1,"score":2,"scale":"shares"},{"title":"Can large language models explore in-context?","url":"/papers/arxiv/2403.15371/","summary":"Large Language Models such as GPT-3.5, GPT-4, and Llama2 struggle to explore in reinforcement learning environments without significant interventions, indicating the need for algorithmic interventions in complex decision-making scenarios.","featured":"2024-03-27","label":"Machine learning","topic":"LLMs & Text","cites":92,"score":196,"scale":"shares"},{"title":"CoLLEGe: Concept Embedding Generation for Large Language Models","url":"/papers/arxiv/2403.15362/","summary":"Embedding Generation: CoLLEGe, a new meta-learning framework, is presented which enhances the ability of language models to learn new concepts quickly using a few example sentences or definitions.","featured":"2024-03-27","label":"Machine learning","topic":"LLMs & Text","cites":4,"score":21,"scale":"shares"},{"title":"Simple and Scalable Strategies to Continually Pre-train Large Language Models","url":"/papers/arxiv/2403.08763/","summary":"The research shows that large language models can be updated efficiently with new data, saving significant computational resources and matching the performance of re-training from scratch.","featured":"2024-03-27","label":"Machine learning","topic":"LLMs & Text","cites":132,"score":1205,"scale":"shares"},{"title":"TableLlama: Towards Open Large Generalist Models for Tables","url":"/papers/arxiv/2311.09206/","summary":"The paper presents TableLlama, an open-source large language model fine-tuned for table-based tasks, and introduces a new dataset, TableInstruct, which improves the performance and generalizability of these models.","featured":"2024-03-27","label":"Machine learning","topic":"LLMs & Text","cites":233,"score":84,"scale":"shares"},{"title":"Can ChatGPT Generate Stocks Tickers to Buy and Sell for Day Trading?","url":"/papers/ssrn/4759311/","summary":"The paper shows that ChatGPT can use Twitter news to generate profitable stock tickers for day trading, demonstrating the AI's ability to turn non-specific news into firm-specific mispricing signals.","featured":"2024-03-20","label":"SSRN","topic":"LLMs & Text","cites":0,"score":4,"scale":"shares"},{"title":"Short Selling and Info Risk Pricing","url":"/papers/ssrn/4761300/","summary":"Short selling influences the pricing of the probability of informed trading (PIN), especially after good news and among small firms, a study reveals.","featured":"2024-03-20","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Disaster Sentiment Analysis with Deep Learning","url":"/papers/ssrn/4755638/","summary":"The article discusses the challenge of rapidly analyzing and using vast social media data during natural disasters for effective crisis management.","featured":"2024-03-13","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"The Financial Uncertainty of Climate-Related Assets","url":"/papers/ssrn/4753434/","summary":"The paper identifies a factor linked to climate news indices and commodity markets that contributes to the excess volatility of climate-sensitive assets.","featured":"2024-03-13","label":"SSRN","topic":"LLMs & Text","cites":1,"score":3,"scale":"shares"},{"title":"Stress Index Strategy","url":"/papers/ssrn/4754939/","summary":"A new stock market strategy that uses a financial stress indicator and sentiment analysis from ChatGPT reading Bloomberg daily market summaries has improved performance in major equity markets.","featured":"2024-03-13","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Investor Sentiment and IPO Flipping in China","url":"/papers/repec/taf-reroxx-v-36-y-2023-i-1-p-2113739/","summary":"The research reveals that investor sentiment, risk-free interest rates, and broad market indices significantly affect IPO first-day flipping in China.","featured":"2024-03-13","label":"RePEc","topic":"LLMs & Text","cites":null,"score":12,"scale":"shares"},{"title":"News Sentiment in Salmon Price","url":"/papers/ssrn/4728262/","summary":"Deep learning models and sentiment analysis can accurately predict salmon spot prices, with sentiment scores reducing prediction errors.","featured":"2024-02-21","label":"SSRN","topic":"LLMs & Text","cites":null,"score":3,"scale":"shares"},{"title":"Chain-of-Thought Reasoning Without Prompting","url":"/papers/arxiv/2402.10200/","summary":"Altering the decoding process in large language models can enhance their reasoning abilities and performance without needing specific prompts.","featured":"2024-02-21","label":"Machine learning","topic":"LLMs & Text","cites":302,"score":185,"scale":"shares"},{"title":"QuantAgent: Seeking Holy Grail in Trading by Self-Improving Large Language Model","url":"/papers/arxiv/2402.03755/","summary":"Enhanced Accuracy in Autonomous Trading: A new system allows autonomous agents using Large Language Models to effectively create and incorporate a specialized knowledge base, proven effective in quantitative investment.","featured":"2024-02-07","label":"arXiv","topic":"LLMs & Text","cites":43,"score":5,"scale":"shares"},{"title":"Learning to Generate Explainable Stock Predictions using Self-Reflective Large Language Models","url":"/papers/arxiv/2402.03659/","summary":"The paper introduces the SEP framework, which uses a self-reflective agent and Proximal Policy Optimization to train Large Language Models for generating accurate and explainable stock predictions.","featured":"2024-02-07","label":"arXiv","topic":"LLMs & Text","cites":83,"score":3,"scale":"shares"},{"title":"\"Guinea Pig Trials\" Utilizing GPT: A Novel Smart Agent-Based Modeling Approach for Studying Firm Competition and Collusion","url":"/papers/arxiv/2308.10974/","summary":"The study presents Smart Agent-Based Modeling (SABM), a method using GPT-4 technologies to simulate human-like strategies and communication for studying firm competition and collusion.","featured":"2024-02-07","label":"arXiv","topic":"LLMs & Text","cites":14,"score":25,"scale":"shares"},{"title":"News Diffusion and Stock Market Reactions","url":"/papers/ssrn/4717833/","summary":"The spread of public news through social networks influences investors' beliefs and the securities market, with increased social connectivity leading to quicker news integration into prices but also causing differing opinions and excessive trading.","featured":"2024-02-07","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","url":"/papers/arxiv/2402.03300/","summary":"Advancing Math Reasoning in Language Models: DeepSeekMath7B is a new language model that uses web data and Group Relative Policy Optimization for advanced mathematical reasoning, scoring high on the MATH benchmark.","featured":"2024-02-07","label":"Machine learning","topic":"LLMs & Text","cites":9003,"score":127,"scale":"shares"},{"title":"What is ”Typological Diversity” in NLP?","url":"/papers/arxiv/2402.04222/","summary":"NLP Research on Diversity: The research explores 'typological diversity' in multilingual NLP studies, finding no clear definitions, and suggests future research should justify the diversity of language samples.","featured":"2024-02-07","label":"Machine learning","topic":"LLMs & Text","cites":6,"score":14,"scale":"shares"},{"title":"Is Self-Repair a Silver Bullet for Code Generation?","url":"/papers/arxiv/2306.09896/","summary":"Effectiveness: Large language models like CodeLlama, GPT-3.5, and GPT-4 show modest and inconsistent performance in self-repairing code, indicating limitations in self-feedback.","featured":"2024-02-07","label":"Machine learning","topic":"LLMs & Text","cites":270,"score":530,"scale":"shares"},{"title":"Extreme Compression of Large Language Models via Additive Quantization","url":"/papers/arxiv/2401.06118/","summary":"The article discusses a new algorithm that enhances the compression of large language models, providing better accuracy and is now available for future research.","featured":"2024-02-07","label":"Machine learning","topic":"LLMs & Text","cites":268,"score":45,"scale":"shares"},{"title":"DSPy Assertions: Computational Constraints for Self-Refining Language Model Pipelines","url":"/papers/arxiv/2312.13382/","summary":"Constraints for Language Model Pipelines: The article introduces LM Assertions, a new programming construct that enhances rule compliance and task performance in text generation by expressing computational constraints in language models.","featured":"2024-02-07","label":"Machine learning","topic":"LLMs & Text","cites":41,"score":41,"scale":"shares"},{"title":"Opening Price Gaps and Information Adjustment in Stocks","url":"/papers/repec/kap-compec-v-63-y-2024-i-2-d-10-1007-s10614-023-10363-w/","summary":"AI and big data research on gap opening price strategies show that negative gaps are more common than positive ones, and bad news adjusts prices faster than good news, with positive gaps offering profitable trading chances.","featured":"2024-02-07","label":"RePEc","topic":"LLMs & Text","cites":null,"score":11,"scale":"shares"},{"title":"Sentiment Trading with Language Models","url":"/papers/ssrn/4706629/","summary":"Large language models like OPT, based on GPT-3, are highly effective in predicting sentiment in U.S. financial news, impacting financial analysis tools and regulatory considerations.","featured":"2024-01-30","label":"SSRN","topic":"LLMs & Text","cites":null,"score":11,"scale":"shares"},{"title":"Identifying M&A Targets from Textual Disclosures","url":"/papers/ssrn/4707567/","summary":"Textual information from firm disclosures, analyzed using a transformer neural network, can significantly enhance the predictability of corporate takeovers, a study reveals.","featured":"2024-01-30","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Media Sentiment & Market Volatility","url":"/papers/ssrn/4709058/","summary":"Financial markets can experience volatility due to media sentiment, especially during significant events, and this impact can affect other markets as well.","featured":"2024-01-30","label":"SSRN","topic":"LLMs & Text","cites":null,"score":3,"scale":"shares"},{"title":"SliceGPT: Compress Large Language Models by Deleting Rows and Columns","url":"/papers/arxiv/2401.15024/","summary":"Compressing Language Models: The paper introduces SliceGPT, a post-training sparsification scheme for large language models that reduces the network's embedding dimension, maintains high performance, reduces inference computation, and reveals computational invariance in transformer networks.","featured":"2024-01-30","label":"Machine learning","topic":"LLMs & Text","cites":466,"score":20,"scale":"shares"},{"title":"BioFinBERT: Finetuning Large Language Models (LLMs) to Analyze Sentiment of Press Releases and Financial Text Around Inflection Points of Biotech Stocks","url":"/papers/arxiv/2401.11011/","summary":"LLMs for Financial Sentiment Analysis: BioFinBERT, a finetuned Large Language Model, is introduced for financial sentiment analysis of biotech press releases and financial texts, which greatly impact biotech stock prices.","featured":"2024-01-23","label":"arXiv","topic":"LLMs & Text","cites":2,"score":3,"scale":"shares"},{"title":"Good News Is Not a Sufficient Condition for Motivated Reasoning","url":"/papers/arxiv/2012.01548/","summary":"The research explores the impact of news valence on belief updating, concluding that people do not overly trust good news over bad news, challenging the idea that beliefs are distorted based on utility.","featured":"2024-01-23","label":"arXiv","topic":"LLMs & Text","cites":2,"score":24,"scale":"shares"},{"title":"Principled Instructions Are All You Need for Questioning LLaMA-1/2, GPT-3.5/4","url":"/papers/arxiv/2312.16171/","summary":"The article presents 26 principles to enhance querying and prompting in large language models, backed by experimental results.","featured":"2024-01-23","label":"Machine learning","topic":"LLMs & Text","cites":145,"score":1480,"scale":"shares"},{"title":"Less is More for Long Document Summary Evaluation by LLMs","url":"/papers/arxiv/2309.07382/","summary":"The paper presents Extract-then-Evaluate, a method that selects key sentences from lengthy documents for evaluation in Large Language Models, enhancing efficiency and alignment with human assessments.","featured":"2024-01-23","label":"Machine learning","topic":"LLMs & Text","cites":51,"score":29,"scale":"shares"},{"title":"DiarizationLM: Speaker Diarization Post-Processing with Large Language Models","url":"/papers/arxiv/2401.03506/","summary":"Speaker Diarization Post-Processing: DiarizationLM is a framework that employs large language models to refine speaker diarization system outputs, enhancing transcript readability and reducing word diarization error rate without the need for retraining.","featured":"2024-01-23","label":"Machine learning","topic":"LLMs & Text","cites":34,"score":27,"scale":"shares"},{"title":"Media Hype and Fake News Impact on Commodity Prices","url":"/papers/repec/eee-finlet-v-59-y-2024-i-c-s1544612323010309/","summary":"The study shows that media hype and fake news greatly influence commodity prices, especially during COVID-19. It also found that bi-directional long-short-term memory is useful in predicting these impacts.","featured":"2024-01-23","label":"RePEc","topic":"LLMs & Text","cites":null,"score":10,"scale":"shares"},{"title":"Yelp Consumption Sentiment and Asset Pricing","url":"/papers/ssrn/4691145/","summary":"A sentiment index based on Yelp restaurant reviews can predict stock market reversals and mispricing, with pessimism being a key predictive factor.","featured":"2024-01-17","label":"SSRN","topic":"LLMs & Text","cites":0,"score":2,"scale":"shares"},{"title":"Designing Heterogeneous LLM Agents for Financial Sentiment Analysis","url":"/papers/arxiv/2401.05799/","summary":"A study suggests using large language models without fine-tuning for financial sentiment analysis, offering a design framework that enhances accuracy.","featured":"2024-01-17","label":"arXiv","topic":"LLMs & Text","cites":138,"score":4,"scale":"shares"},{"title":"The Unreasonable Effectiveness of Easy Training Data for Hard Tasks","url":"/papers/arxiv/2401.06751/","summary":"The study suggests that current language models can effectively generalize from easy to hard data, implying that scalable oversight may be less challenging than previously believed.","featured":"2024-01-17","label":"Machine learning","topic":"LLMs & Text","cites":56,"score":94,"scale":"shares"},{"title":"TOFU: A Task of Fictitious Unlearning for LLMs","url":"/papers/arxiv/2401.06121/","summary":"The study presents TOFU, a new benchmark for understanding unlearning in large language models, revealing that existing unlearning algorithms are not effective.","featured":"2024-01-17","label":"Machine learning","topic":"LLMs & Text","cites":579,"score":22,"scale":"shares"},{"title":"News Intensity & Currency Volatility","url":"/papers/ssrn/4686168/","summary":"Semantic fingerprinting of news headlines can measure the impact of news on major currency indices, showing a positive correlation between news intensity and currency return volatility.","featured":"2024-01-09","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Text mining arXiv: a look through quantitative finance papers","url":"/papers/arxiv/2401.01751/","summary":"The study uses text mining and natural language processing to examine quantitative finance papers from 1997 to 2022 on the arXiv preprint server. It identifies topic trends, most cited researchers and journals, and compares different topic modeling algorithms.","featured":"2024-01-09","label":"arXiv","topic":"LLMs & Text","cites":3,"score":5,"scale":"shares"},{"title":"Learning to Prompt with Text Only Supervision for Vision-Language Models","url":"/papers/arxiv/2401.02418/","summary":"The study suggests a method to modify basic vision-language models like CLIP for specific tasks using text data from large language models, allowing for easy application to new classes and datasets.","featured":"2024-01-09","label":"Machine learning","topic":"LLMs & Text","cites":63,"score":95,"scale":"shares"},{"title":"TinyLlama: An Open-Source Small Language Model","url":"/papers/arxiv/2401.02385/","summary":"Small Open-Source Language Model: The article presents TinyLlama, a compact 1.1B language model that performs remarkably well in various tasks despite its small size, having been pretrained on around 1 trillion tokens.","featured":"2024-01-09","label":"Machine learning","topic":"LLMs & Text","cites":902,"score":82,"scale":"shares"},{"title":"Shai: A large language model for asset management","url":"/papers/arxiv/2312.14203/","summary":"10B-Level Language Model for Asset Management: The article introduces Shai, a superior language model specifically designed for the asset management industry, providing practical financial insights.","featured":"2024-01-03","label":"arXiv","topic":"LLMs & Text","cites":5,"score":4,"scale":"shares"},{"title":"Do LLM Agents Exhibit Social Behavior?","url":"/papers/arxiv/2312.15198/","summary":"Large Language Models (LLMs) display human-like social behaviors but also have significant differences, necessitating further research for accurate human behavior emulation.","featured":"2024-01-03","label":"arXiv","topic":"LLMs & Text","cites":71,"score":4,"scale":"shares"},{"title":"News Volatility and Portfolio Implications","url":"/papers/ssrn/4677789/","summary":"The article shows how the XGBoost machine learning algorithm can predict next-day volatility jumps based on firm-specific news, leading to improved portfolio performance.","featured":"2024-01-03","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Gemini in Reasoning: Unveiling Commonsense in Multimodal Large Language Models","url":"/papers/arxiv/2312.17661/","summary":"Google's Gemini, a Multimodal Large Language Model, is evaluated across 12 commonsense reasoning datasets, showing competitive reasoning abilities and the need for further model advancements.","featured":"2024-01-03","label":"Machine learning","topic":"LLMs & Text","cites":29,"score":6,"scale":"shares"},{"title":"ChatGPT as a Quant Asset Manager","url":"/papers/repec/eee-finlet-v-58-y-2023-i-pd-s1544612323009522/","summary":"The research suggests a quantitative investment approach that includes recommendations from ChatGPT, demonstrating its potential to enhance portfolio efficiency.","featured":"2023-12-20","label":"RePEc","topic":"LLMs & Text","cites":null,"score":16,"scale":"shares"},{"title":"Comparative Analysis of LLMs for Financial Sentiment","url":"/papers/ssrn/4658156/","summary":"The use of Large Language Models, specifically the gpt3.5turbo model, in financial sentiment analysis is examined, highlighting the potential of in-context learning and fine-tuning on finance-specific datasets.","featured":"2023-12-13","label":"SSRN","topic":"LLMs & Text","cites":null,"score":4,"scale":"shares"},{"title":"Quantifying Market Psyche","url":"/papers/ssrn/4662241/","summary":"The paper establishes a link between qualitative information and its quantitative market effects, suggesting that significant news can cause larger movements in smaller stock indexes, with the PE ratio potentially amplifying or mitigating this effect.","featured":"2023-12-13","label":"SSRN","topic":"LLMs & Text","cites":null,"score":76,"scale":"shares"},{"title":"GPT in Game Theory Experiments","url":"/papers/arxiv/2305.05516/","summary":"The research shows that Generative Pre-trained Transformers (GPT) can mimic human responses in strategic games and can be influenced by fairness or selfishness traits.","featured":"2023-12-13","label":"arXiv","topic":"LLMs & Text","cites":29,"score":27,"scale":"shares"},{"title":"Large Language Models for Mathematicians","url":"/papers/arxiv/2312.04556/","summary":"Impact and Potential for Mathematicians: The article explores how large language models like ChatGPT can assist professional mathematicians, discussing their mathematical capabilities, best practices, and potential issues.","featured":"2023-12-13","label":"Machine learning","topic":"LLMs & Text","cites":11,"score":134,"scale":"shares"},{"title":"Integrating Pre-Trained Speech and Language Models for End-to-End Speech Recognition","url":"/papers/arxiv/2312.03668/","summary":"The study investigates the integration of a pre-trained speech representation model with a large language model for automatic speech recognition, achieving performance similar to modern models.","featured":"2023-12-13","label":"Machine learning","topic":"LLMs & Text","cites":23,"score":41,"scale":"shares"},{"title":"Volatility and Mispricing with Sentiment and Institutional Investors","url":"/papers/ssrn/4649820/","summary":"The research suggests that high investor sentiment and increased institutionalization can decrease excess volatility and mispricing in stock returns.","featured":"2023-12-06","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Scalable Extraction of Training Data from (Production) Language Models","url":"/papers/arxiv/2311.17035/","summary":"The study shows that large amounts of training data can be extracted from different machine learning models, highlighting that current techniques do not prevent data memorization.","featured":"2023-12-06","label":"Machine learning","topic":"LLMs & Text","cites":619,"score":115,"scale":"shares"},{"title":"Multi-Task Learning in Financial NLP","url":"/papers/ssrn/4640029/","summary":"Improvements in Financial NLP's Multitask Learning can be achieved by considering skill diversity, task relatedness, and aggregation size.","featured":"2023-11-29","label":"SSRN","topic":"LLMs & Text","cites":null,"score":50,"scale":"shares"},{"title":"Narratives from GPT-derived networks of news and a link to financial markets dislocations","url":"/papers/arxiv/2311.14419/","summary":"The study uses natural language processing and network analysis to examine news content over time, linking the results to financial market dislocations.","featured":"2023-11-29","label":"arXiv","topic":"LLMs & Text","cites":5,"score":6,"scale":"shares"},{"title":"FinMem: A Performance-Enhanced LLM Trading Agent With Layered Memory and Character Design","url":"/papers/arxiv/2311.13743/","summary":"Performance-Enhanced Large Language Model Trading Agent: The research presents FinMe, a Large Language Model-based agent for financial decision-making, demonstrating its superior trading performance in stocks and funds.","featured":"2023-11-29","label":"arXiv","topic":"LLMs & Text","cites":253,"score":27,"scale":"shares"},{"title":"StockEmotions: Discover Investor Emotions for Financial Sentiment Analysis and Multivariate Time Series","url":"/papers/arxiv/2301.09279/","summary":"Investor Sentiment: The article introduces StockEmotions, a new dataset for detecting emotions in the stock market from StockTwits, with DistilBERT and Temporal Attention LSTM model showing the best results.","featured":"2023-11-29","label":"arXiv","topic":"LLMs & Text","cites":21,"score":17,"scale":"shares"},{"title":"Sentiment Difficulty in ABSA","url":"/papers/repec/gam-jmathe-v-11-y-2023-i-22-p-4647-d-1280114/","summary":"A study investigates sentence difficulty in aspect-based sentiment analysis, using different learning models and text representations, and identifies the hardest sentences using a mix of classifiers.","featured":"2023-11-29","label":"RePEc","topic":"LLMs & Text","cites":null,"score":27,"scale":"shares"},{"title":"Long-Term Volatility Shapes the Stock Market's Sensitivity to News","url":"/papers/ssrn/4632733/","summary":"US. macroeconomic news impacts the SP 500 more when long-term stock market volatility is high.","featured":"2023-11-15","label":"SSRN","topic":"LLMs & Text","cites":2,"score":3,"scale":"shares"},{"title":"Credit Sentiments in Conference Calls and Bond Market Returns","url":"/papers/ssrn/4628305/","summary":"Credit sentiments from conference calls affect bond market returns, with positive sentiments leading to better credit ratings and lower future debt costs.","featured":"2023-11-15","label":"SSRN","topic":"LLMs & Text","cites":0,"score":2,"scale":"shares"},{"title":"Multi-Label Topic Model for Financial Textual Data","url":"/papers/arxiv/2311.07598/","summary":"The study introduces a multi-label topic model for financial texts, achieving high accuracy and showing that stock market reactions depend on the co-occurrence of specific topics.","featured":"2023-11-15","label":"arXiv","topic":"LLMs & Text","cites":1,"score":5,"scale":"shares"},{"title":"Smart Agent-Based Modeling: On the Use of Large Language Models in Computer Simulations","url":"/papers/arxiv/2311.06330/","summary":"The paper introduces Smart Agent-Based Modeling (SABM), a new framework that uses Large Language Models for more realistic simulations of complex systems.","featured":"2023-11-15","label":"arXiv","topic":"LLMs & Text","cites":22,"score":13,"scale":"shares"},{"title":"GPT-4V(ision) as A Social Media Analysis Engine","url":"/papers/arxiv/2311.07547/","summary":"The study examines the abilities of Large Multimodal Models (LMMs), particularly GPT-4V, in understanding social multimedia content, noting challenges in multilingual comprehension and trend generalization.","featured":"2023-11-15","label":"Machine learning","topic":"LLMs & Text","cites":0,"score":13,"scale":"shares"},{"title":"Data Contamination Quiz: A Tool to Detect and Estimate Contamination in Large Language Models","url":"/papers/arxiv/2311.06233/","summary":"The paper introduces the Data Contamination Quiz, a method for detecting and estimating data contamination in large language models, demonstrating improved detection and accurate contamination estimation.","featured":"2023-11-15","label":"Machine learning","topic":"LLMs & Text","cites":58,"score":8,"scale":"shares"},{"title":"Multimodal Generative Models for Bankruptcy Prediction Using Textual Data","url":"/papers/arxiv/2211.08405/","summary":"The research presents multimodal learning in bankruptcy prediction models to tackle the problem of missing MDA section in Form 10-K, showing improved classification performance and addressing the limitation of previous models.","featured":"2023-11-08","label":"arXiv","topic":"LLMs & Text","cites":3,"score":33,"scale":"shares"},{"title":"TopicGPT: A Prompt-based Topic Modeling Framework","url":"/papers/arxiv/2311.01449/","summary":"TopicGPT, a new framework, is introduced that uses large language models to identify latent topics in a text collection, providing more interpretable topics and user control.","featured":"2023-11-08","label":"Machine learning","topic":"LLMs & Text","cites":0,"score":9,"scale":"shares"},{"title":"Deep Learning Model for Newsvendor Problem with Textual Review Data","url":"/papers/repec/eee-proeco-v-265-y-2023-i-c-s0925527323002487/","summary":"The article talks about a new inventory management framework that uses a deep learning model. This model suggests order quantities based on online reviews and demand data, reducing costs by 28.7% compared to other models.","featured":"2023-11-08","label":"RePEc","topic":"LLMs & Text","cites":null,"score":16,"scale":"shares"},{"title":"FinDKG: Dynamic Knowledge Graph with Large Language Models for Global Finance","url":"/papers/ssrn/4608445/","summary":"The study presents a new use of dynamic Knowledge Graphs in modeling global financial systems, incorporating deep learning and a large language model, and introduces an open-source system for financial analytics.","featured":"2023-10-25","label":"SSRN","topic":"LLMs & Text","cites":20,"score":7,"scale":"shares"},{"title":"Development of a Recommendation News Generator Based on Machine Learning","url":"/papers/ssrn/4609841/","summary":"A machine learning-based recommendation system can improve news consumption by simplifying the process of discovering news articles.","featured":"2023-10-25","label":"SSRN","topic":"LLMs & Text","cites":0,"score":3,"scale":"shares"},{"title":"Leveraging Deep Learning and Online Source Sentiment for Financial Portfolio Management","url":"/papers/arxiv/2309.16679/","summary":"The article reviews different Deep Learning techniques for financial trading, addressing their efficiency, training issues, and potential solutions.","featured":"2023-10-25","label":"arXiv","topic":"LLMs & Text","cites":3,"score":11,"scale":"shares"},{"title":"Let's Synthesize Step by Step: Iterative Dataset Synthesis with Large Language Models by Extrapolating Errors from Small Models","url":"/papers/arxiv/2310.13671/","summary":"Synthesis Step by Step (S3) is a new data synthesis framework designed to minimize the distribution gap between synthesized and real task data, improving the performance of small models trained on the synthesized dataset.","featured":"2023-10-25","label":"Machine learning","topic":"LLMs & Text","cites":46,"score":17,"scale":"shares"},{"title":"Data Selection for Language Models via Importance Resampling","url":"/papers/arxiv/2302.03169/","summary":"The Data Selection with Importance Resampling (DSIR) framework is a new method for selecting subsets of large raw unlabeled datasets, surpassing manual curation and heuristic filtering methods in both specific and general language models.","featured":"2023-10-25","label":"Machine learning","topic":"LLMs & Text","cites":371,"score":279,"scale":"shares"},{"title":"Sentiment Analysis of Amazon Products using Hybrid Model","url":"/papers/ssrn/4602946/","summary":"Sentiment Analysis uses social media comments to understand people's opinions on various subjects, assisting in decision-making processes like product purchases or investments.","featured":"2023-10-18","label":"SSRN","topic":"LLMs & Text","cites":null,"score":3,"scale":"shares"},{"title":"Can GPT models be Financial Analysts? An Evaluation of ChatGPT and GPT-4 on mock CFA Exams","url":"/papers/arxiv/2310.08678/","summary":"The study assesses the financial understanding of Large Language Models (LLMs) like ChatGPT and GPT-4 using CFA Program mock exam questions to improve their use in finance.","featured":"2023-10-18","label":"arXiv","topic":"LLMs & Text","cites":32,"score":8,"scale":"shares"},{"title":"Leveraging Large Language Model for Automatic Evolving of Industrial Data-Centric R&D Cycle","url":"/papers/arxiv/2310.11249/","summary":"The paper investigates the role of Large Language Models (LLMs) in speeding up data-centric R&D evolution, using quantitative investment research as a case study, showing promising results on the open-source research platform, Qlib.","featured":"2023-10-18","label":"arXiv","topic":"LLMs & Text","cites":1,"score":4,"scale":"shares"},{"title":"Towards reducing hallucination in extracting information from financial reports using Large Language Models","url":"/papers/arxiv/2310.10760/","summary":"The paper showcases the use of Large Language Models (LLMs) for precise and efficient extraction of information from the Q&A section of company financial reports, proving their method's superiority through various metrics.","featured":"2023-10-18","label":"arXiv","topic":"LLMs & Text","cites":25,"score":2,"scale":"shares"},{"title":"Llemma: An Open Language Model For Mathematics","url":"/papers/arxiv/2310.10631/","summary":"Open Math Language Model: The article discusses Llemma, a superior language model for mathematics that can prove theorems without additional fine-tuning.","featured":"2023-10-18","label":"Machine learning","topic":"LLMs & Text","cites":497,"score":434,"scale":"shares"},{"title":"In-Context Pretraining: Language Modeling Beyond Document Boundaries","url":"/papers/arxiv/2310.10638/","summary":"Modeling Beyond Documents: The paper introduces In-Context Pretraining, a novel method that improves language models' performance by pretraining them on related documents, enhancing their contextual reasoning skills.","featured":"2023-10-18","label":"Machine learning","topic":"LLMs & Text","cites":101,"score":19,"scale":"shares"},{"title":"Interactive Task Planning with Language Models","url":"/papers/arxiv/2310.10645/","summary":"A new framework is proposed that uses language models for interactive task planning, allowing for easy adaptation to different tasks and precise replanning based on new user requests.","featured":"2023-10-18","label":"Machine learning","topic":"LLMs & Text","cites":68,"score":11,"scale":"shares"},{"title":"Mistral 7B","url":"/papers/arxiv/2310.06825/","summary":"Superior Language Model: Mistral 7B v0.1 is a language model with 7 billion parameters that excels in reasoning, mathematics, and code generation, and has a version specifically designed to follow instructions.","featured":"2023-10-16","label":"Machine learning","topic":"LLMs & Text","cites":3920,"score":200,"scale":"shares"},{"title":"Ferret: Refer and Ground Anything Anywhere at Any Granularity","url":"/papers/arxiv/2310.07704/","summary":"Spatial Referring in Images: Ferret is a Multimodal Large Language Model that can understand and locate spatial references in images, outperforming other models in region-based and localization-required multimodal chatting.","featured":"2023-10-16","label":"Machine learning","topic":"LLMs & Text","cites":604,"score":52,"scale":"shares"},{"title":"MemGPT: Towards LLMs as Operating Systems","url":"/papers/arxiv/2310.08560/","summary":"Extended Context in LLMs: MemGPT is a system that manages different memory levels, providing extended context within large language models' limited context windows, enhancing document analysis and multi-session chat performance.","featured":"2023-10-16","label":"Machine learning","topic":"LLMs & Text","cites":1373,"score":33,"scale":"shares"},{"title":"Octopus: Embodied Vision-Language Programmer from Environmental Feedback","url":"/papers/arxiv/2310.08588/","summary":"Vision-Language Programmer: Octopus is a vision-language model that can interpret an agent's vision and textual task objectives to generate complex action sequences and executable code, showing improved decision-making in various tasks.","featured":"2023-10-16","label":"Machine learning","topic":"LLMs & Text","cites":104,"score":14,"scale":"shares"},{"title":"LLM-grounded Diffusion: Enhancing Prompt Understanding of Text-to-Image Diffusion Models with Large Language Models","url":"/papers/arxiv/2305.13655/","summary":"Enhancing Text-to-Image Models: The study suggests a two-stage process using a pretrained language model to improve image generation accuracy in diffusion models, enabling multi-round scene specification in various languages.","featured":"2023-10-16","label":"Machine learning","topic":"LLMs & Text","cites":276,"score":179,"scale":"shares"},{"title":"SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models","url":"/papers/arxiv/2303.08896/","summary":"Hallucination Detection for LLMs: The paper presents SelfCheckGPT, a new approach for fact-checking black-box model responses without an external database, proving its superior ability to detect and rank factual and non-factual sentences.","featured":"2023-10-16","label":"Machine learning","topic":"LLMs & Text","cites":1236,"score":99,"scale":"shares"},{"title":"GPT-MolBERTa: GPT Molecular Features Language Model for molecular property prediction","url":"/papers/arxiv/2310.03030/","summary":"Molecular Property Prediction Language Model: GPT-MolBERTa, a self-supervised language model that uses textual descriptions of molecules to predict their properties, is introduced, demonstrating high performance on various molecule property benchmarks.","featured":"2023-10-16","label":"Machine learning","topic":"LLMs & Text","cites":28,"score":38,"scale":"shares"},{"title":"SALMON: Self-Alignment with Instructable Reward Models","url":"/papers/arxiv/2310.05910/","summary":"Minimal Human Supervision Language Model Alignment: The paper introduces SALMON, a new method for aligning base language models with minimal human supervision using principle-following reward models, showing its superior performance on multiple benchmark datasets.","featured":"2023-10-16","label":"Machine learning","topic":"LLMs & Text","cites":67,"score":36,"scale":"shares"},{"title":"Risk Aware Benchmarking of Large Language Models","url":"/papers/arxiv/2310.07132/","summary":"Statistical Significance in Risk Assessment and Model Selection: The paper presents a framework for evaluating socio-technical risks of foundation models, using a new statistical method and a risk-aware approach, and applies it to assess large language models for risks of deviating from instructions and producing harmful content.","featured":"2023-10-12","label":"arXiv","topic":"LLMs & Text","cites":4,"score":7,"scale":"shares"},{"title":"Investor Sentiments and Volatility Trading Strategies","url":"/papers/repec/sae-emffin-v-22-y-2023-i-3-p-326-350/","summary":"Research examines the link between trader sentiment and market volatility in India during COVID-19, suggesting new options trading strategies considering traders' neuro-specific intentions.","featured":"2023-10-12","label":"RePEc","topic":"LLMs & Text","cites":null,"score":32,"scale":"shares"},{"title":"Look-Ahead Bias in Stock Return Predictions","url":"/papers/ssrn/4586726/","summary":"Large language models like ChatGPT can generate profitable trading signals from news sentiment, but backtesting can yield biased results due to overlapping periods.","featured":"2023-10-04","label":"SSRN","topic":"LLMs & Text","cites":null,"score":5,"scale":"shares"},{"title":"Using Large Language Models for Qualitative Analysis can Introduce Serious Bias","url":"/papers/arxiv/2309.17147/","summary":"The study investigates the application of Large Language Models in analyzing qualitative interview data, warning about possible biases and recommending the use of simpler supervised models trained on high-quality human annotations to reduce measurement error and bias.","featured":"2023-10-04","label":"arXiv","topic":"LLMs & Text","cites":78,"score":2,"scale":"shares"},{"title":"Addressing 'Special Issues' in Classifying Trademark Distinctiveness Using GPT-3","url":"/papers/ssrn/4582171/","summary":"The article discusses the application of Large Language Models (LLMs) and machine learning in assessing trademarks for registration, showing how an LLM can help identify issues and prepare data for machine learning algorithms.","featured":"2023-09-28","label":"SSRN","topic":"LLMs & Text","cites":2,"score":2,"scale":"shares"},{"title":"Financial Statement Fraud: News Analysis","url":"/papers/ssrn/4583665/","summary":"News Analysis: A new system using news coverage and machine learning can accurately detect financial fraud in Chinese companies from 2001 to 2022.","featured":"2023-09-28","label":"SSRN","topic":"LLMs & Text","cites":null,"score":236,"scale":"shares"},{"title":"Modeling Nested Data in Operation Research","url":"/papers/ssrn/4579724/","summary":"A paper suggests the contextual effects model as a better method for handling nested data in research, over fixed effects and multilevel models.","featured":"2023-09-28","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Graph Embedding for Sentiment Analysis","url":"/papers/ssrn/4576625/","summary":"The research suggests a self-supervised method for Persian sentiment analysis using combined representation learning and Siamese Network, using a self-supervised approach to enhance feature vectors from graph-structured data.","featured":"2023-09-21","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Is Media Sentiment Associated with Future Conflict Events?","url":"/papers/ssrn/4573695/","summary":"Using machine learning and natural language processing, the research finds a significant link between conflictual sentiment in media reports and future conflict events, indicating sentiment analysis can improve our understanding of conflict dynamics.","featured":"2023-09-21","label":"SSRN","topic":"LLMs & Text","cites":3,"score":2,"scale":"shares"},{"title":"Extracting Financial Data from Unstructured Sources: Leveraging Large Language Models","url":"/papers/ssrn/4567607/","summary":"A study has developed a new framework that can accurately automate the extraction of financial data from PDF files using large language models.","featured":"2023-09-14","label":"SSRN","topic":"LLMs & Text","cites":51,"score":3,"scale":"shares"},{"title":"News-Driven Expectations and Volatility Clustering","url":"/papers/doi/10-3390-jrfm13010017/","summary":"The paper attributes the regularities of financial volatility to traders' reactions to news, influenced by the behaviors of long-term investors and short-term speculators.","featured":"2023-09-14","label":"arXiv","topic":"LLMs & Text","cites":6,"score":3,"scale":"shares"},{"title":"Credit Information in Earnings Calls","url":"/papers/arxiv/2209.11914/","summary":"A new method has been developed to predict credit spread changes and company profitability using information from quarterly earnings calls, indicating that investors may not be fully exploiting this data.","featured":"2023-09-14","label":"arXiv","topic":"LLMs & Text","cites":1,"score":27,"scale":"shares"},{"title":"News Data's Impact on Trading Decisions","url":"/papers/ssrn/4551629/","summary":"The paper suggests a reinforcement learning approach for high-frequency algorithmic trading in futures market using news and price data, tested on the NIFTY 50 index.","featured":"2023-08-30","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Can Machine Learning Catch Economic Recessions Using Economic and Market Sentiments?","url":"/papers/ssrn/4553506/","summary":"The paper uses machine learning to predict US economic recessions using market sentiment and economic indicators, using the ARIMA method for backcasting.","featured":"2023-08-30","label":"SSRN","topic":"LLMs & Text","cites":9,"score":2,"scale":"shares"},{"title":"Breaking the Bank with ChatGPT: Few-Shot Text Classification for Finance","url":"/papers/arxiv/2308.14634/","summary":"Conversational GPT models are suggested for efficient text classification in finance, providing a practical solution for tasks with limited labels and achieving top-tier results.","featured":"2023-08-30","label":"arXiv","topic":"LLMs & Text","cites":36,"score":2,"scale":"shares"},{"title":"To the Moon: Analyzing Collective Trading Events on the Wings of Sentiment Analysis","url":"/papers/arxiv/2308.09968/","summary":"A study finds a strong correlation between Twitter activity and stock volatility, but a weak connection between tweet sentiment and stock performance, suggesting Reddit has a more significant impact on these events.","featured":"2023-08-24","label":"arXiv","topic":"LLMs & Text","cites":2,"score":5,"scale":"shares"},{"title":"NLP-based detection of systematic anomalies among the narratives of consumer complaints","url":"/papers/arxiv/2308.11138/","summary":"The paper introduces a method using NLP to identify patterns in consumer complaints by turning stories into measurable data for algorithm analysis.","featured":"2023-08-24","label":"arXiv","topic":"LLMs & Text","cites":8,"score":4,"scale":"shares"},{"title":"Sentiment Analysis is Virtually Useless in Financial Forecasting","url":"/papers/ssrn/4545418/","summary":"The paper criticizes the overemphasis on sentiment analysis in financial forecasting, arguing that its practical use is often overstated and misleading, and calls for more methodological rigor and transparency in its application.","featured":"2023-08-24","label":"SSRN","topic":"LLMs & Text","cites":0,"score":2,"scale":"shares"},{"title":"When Does Bad News Cause Mispricing? A Historical View","url":"/papers/ssrn/4544851/","summary":"The research suggests that news media sentiment can predict returns during periods of high volatility, low returns, high economic policy uncertainty, and heavily skewed returns.","featured":"2023-08-24","label":"SSRN","topic":"LLMs & Text","cites":0,"score":27,"scale":"shares"},{"title":"Machine Learning vs. Dictionary for Sentiment","url":"/papers/repec/inm-ormnsc-v-68-y-2022-i-7-p-5514-5532/","summary":"Machine-learning methods, particularly the random-forest-regression-tree method, significantly improve the capture of disclosure sentiment at 10-K filing and conference-call dates compared to dictionary-based measures.","featured":"2023-08-24","label":"RePEc","topic":"LLMs & Text","cites":null,"score":13,"scale":"shares"},{"title":"Company Similarity Using Large Language Models","url":"/papers/arxiv/2308.08031/","summary":"Large language models can learn company profiles from SEC filings, accurately replicating GICS classifications and reflecting financial performance metrics.","featured":"2023-08-17","label":"arXiv","topic":"LLMs & Text","cites":15,"score":5,"scale":"shares"},{"title":"ChatGPT-Based Investment Portfolio Selection","url":"/papers/arxiv/2308.06260/","summary":"The research shows that AI model ChatGPT is useful in selecting stocks from the S&P500 index, but may not be as efficient in determining the best stock weights in a portfolio.","featured":"2023-08-17","label":"arXiv","topic":"LLMs & Text","cites":46,"score":5,"scale":"shares"},{"title":"ChatGPT for Portfolio Selection","url":"/papers/ssrn/4538502/","summary":"The research shows AI model ChatGPT is effective in selecting stocks from the SP500 index, but not as efficient in assigning optimal weights to portfolio stocks.","featured":"2023-08-17","label":"SSRN","topic":"LLMs & Text","cites":null,"score":3,"scale":"shares"},{"title":"Future of Natural Language Processing","url":"/papers/ssrn/4539657/","summary":"The paper provides an overview of the latest machine learning algorithms, the significance of large datasets, and emerging trends in NLP.","featured":"2023-08-17","label":"SSRN","topic":"LLMs & Text","cites":null,"score":4,"scale":"shares"},{"title":"The Power of Large Language Models: A ChatGPT-driven Textual Analysis of Fundamental Data","url":"/papers/ssrn/4535647/","summary":"The study uses large language models to interpret business data from the Japan Company Handbook, Shikiho, to classify firms and build equity portfolios, suggesting the market may overlook some factual information in Shikiho's text.","featured":"2023-08-09","label":"SSRN","topic":"LLMs & Text","cites":4,"score":3,"scale":"shares"},{"title":"Investor Sentiment and Futures Market Mispricing","url":"/papers/ssrn/4532403/","summary":"The study reveals that investor sentiment significantly influences futures mispricing, with excessive optimism leading to overvaluation, especially when individual trading is dominant.","featured":"2023-08-09","label":"SSRN","topic":"LLMs & Text","cites":23,"score":2,"scale":"shares"},{"title":"Market Sentiment Index for Stock Analysis","url":"/papers/ssrn/4532195/","summary":"The research proposes a dynamic design for aggregating market sentiment, which adjusts to sentiment indicator changes and shows that ignoring these changes can skew model construction.","featured":"2023-08-09","label":"SSRN","topic":"LLMs & Text","cites":null,"score":150,"scale":"shares"},{"title":"Natural Language Processing meets Accounting and Finance: Review and Performance Comparison of Textual Analysis Approaches","url":"/papers/ssrn/4527724/","summary":"Performance Comparison: Despite the success of deep learning in Natural Language Processing, traditional machine learning and rule-based methods are still prevalent in accounting and finance, with deep learning performing best overall.","featured":"2023-08-02","label":"SSRN","topic":"LLMs & Text","cites":2,"score":2,"scale":"shares"},{"title":"Sentiment Mispricing and Excess Volatility in Institutional Investors","url":"/papers/ssrn/4527359/","summary":"Sentiment-driven investors and benchmark-focused institutions can lower stock market prices and create new countercyclical patterns in stock volatility.","featured":"2023-08-02","label":"SSRN","topic":"LLMs & Text","cites":null,"score":44,"scale":"shares"},{"title":"Alpha-GPT: Human-AI Interactive Alpha Mining for Quantitative Investment","url":"/papers/arxiv/2308.00016/","summary":"Alpha-GPT is a new mining paradigm that uses human-AI interaction and a unique algorithm to understand quantitative researchers' ideas and generate efficient trading signals.","featured":"2023-08-02","label":"arXiv","topic":"LLMs & Text","cites":81,"score":13,"scale":"shares"},{"title":"ChatGPT and Corporate Policies","url":"/papers/ssrn/4521096/","summary":"The research uses ChatGPT to interpret managerial expectations from corporate disclosures, predicting future capital expenditure and intangible and R&D investments.","featured":"2023-07-26","label":"SSRN","topic":"LLMs & Text","cites":60,"score":3,"scale":"shares"},{"title":"ChatGPT, Generative AI, and Investment Advisory","url":"/papers/ssrn/4519182/","summary":"The article shows that AI like ChatGPT can generate portfolio recommendations based on policy announcements, potentially outperforming markets unlike traditional textual analysis.","featured":"2023-07-26","label":"SSRN","topic":"LLMs & Text","cites":11,"score":3,"scale":"shares"},{"title":"Generation of SQL Query Using Natural Language Processing","url":"/papers/ssrn/4515801/","summary":"The paper explores the creation of systems for generating SQL queries using natural language processing, discussing challenges, methods, and future research.","featured":"2023-07-26","label":"SSRN","topic":"LLMs & Text","cites":3,"score":6,"scale":"shares"},{"title":"Assessing Large Language Models' ability to predict how humans balance self-interest and the interest of others","url":"/papers/arxiv/2307.12776/","summary":"A study reveals that only the GPT-4 chatbot could predict human decisions in a game, but it overestimated altruistic actions, impacting AI development.","featured":"2023-07-26","label":"arXiv","topic":"LLMs & Text","cites":7,"score":2,"scale":"shares"},{"title":"FinGPT: Democratizing Internet-scale Data for Financial Large Language Models","url":"/papers/arxiv/2307.10485/","summary":"Democratizing Internet-scale Data for Financial Large: The paper presents FinGPT, an open-source, data-centric framework that automates the gathering and curation of real-time financial data from various online sources, aiming to democratize financial data for large language models.","featured":"2023-07-26","label":"arXiv","topic":"LLMs & Text","cites":124,"score":43,"scale":"shares"},{"title":"Application of Fundamental Analysis in Stock Valuation in the Capital Market and Investment Decisions by Price Methods Earnings Ratio (PER)","url":"/papers/ssrn/4509690/","summary":"The study finds that while fundamental analysis methods are useful in stock valuation, external factors like market sentiment and economic policy changes also influence stock prices.","featured":"2023-07-19","label":"SSRN","topic":"LLMs & Text","cites":1,"score":2,"scale":"shares"},{"title":"Navigating Challenges and Future Trends in Sentiment Analysis for Investment Decision Making","url":"/papers/ssrn/4513208/","summary":"Sentiment analysis can inform investment choices by interpreting sentiment from various text data sources, though it comes with its own challenges.","featured":"2023-07-19","label":"SSRN","topic":"LLMs & Text","cites":1,"score":2,"scale":"shares"},{"title":"Value of Textual Data in Forecasting Macroeconomic Tail Risk","url":"/papers/ssrn/4509043/","summary":"News-based data offers valuable insights not provided by economic indicators, especially for left-tail forecasts, and significantly influences consumer sentiment.","featured":"2023-07-19","label":"SSRN","topic":"LLMs & Text","cites":null,"score":36,"scale":"shares"},{"title":"Linking Asset Prices to News without Mentions","url":"/papers/ssrn/4508392/","summary":"Semantic fingerprinting, a new Natural Language Processing technology, can better link news articles to asset prices by considering the overall text meaning.","featured":"2023-07-19","label":"SSRN","topic":"LLMs & Text","cites":null,"score":2,"scale":"shares"},{"title":"Contrasting the efficiency of stock price prediction models using various types of LSTM models aided with sentiment analysis","url":"/papers/arxiv/2307.07868/","summary":"Researchers aim to create a model that uses company forecasts and sector performances to accurately predict short and long-term equity share prices.","featured":"2023-07-19","label":"arXiv","topic":"LLMs & Text","cites":1,"score":4,"scale":"shares"},{"title":"Time-Varying Equity Premia & Sentiment","url":"/papers/ssrn/4505699/","summary":"From 1990 to 2022, equity market returns can be predicted using a simple model, with higher returns following high implied volatility and lower returns after high market sentiment.","featured":"2023-07-12","label":"SSRN","topic":"LLMs & Text","cites":null,"score":108,"scale":"shares"},{"title":"Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models","url":"/papers/ssrn/4412788/","summary":"ChatGPT predicts stock market returns using sentiment analysis, outperforming traditional methods.","featured":"2023-07-05","label":"arXiv","topic":"LLMs & Text","cites":403,"score":82,"scale":"shares"},{"title":"Technical Analysis: Improving Entry and Exit Timing","url":"/papers/repec/wsi-rpbfmp-v-26-y-2023-i-02-n-s0219091523500133/","summary":"Improving Entry and Exit Timing: Technical analysis and sentiment measures can outperform buy and hold strategy.","featured":"2023-07-05","label":"RePEc","topic":"LLMs & Text","cites":null,"score":31,"scale":"shares"},{"title":"Instruct-FinGPT: Financial Sentiment Analysis by Instruction Tuning of General-Purpose Large Language Models","url":"/papers/arxiv/2306.12659/","summary":"A new approach improves financial sentiment analysis by addressing limitations of language models.","featured":"2023-06-28","label":"arXiv","topic":"LLMs & Text","cites":140,"score":10,"scale":"shares"},{"title":"Unveiling the Potential of Sentiment: Can Large Language Models Predict Chinese Stock Price Movements?","url":"/papers/arxiv/2306.14222/","summary":"A framework assesses large language models for sentiment factor extraction in stock trading strategies.","featured":"2023-06-28","label":"arXiv","topic":"LLMs & Text","cites":25,"score":4,"scale":"shares"},{"title":"Sentiment and Volatility in Shanghai A-shares Market","url":"/papers/repec/eme-cfripp-cfri-01-2021-0007/","summary":"Investor sentiment in China's stock market impacts stock volatility.","featured":"2023-06-28","label":"RePEc","topic":"LLMs & Text","cites":null,"score":22,"scale":"shares"},{"title":"FinGPT: Open-Source Financial Large Language Models","url":"/papers/arxiv/2306.06031/","summary":"FinGPT developed for finance sector language model","featured":"2023-06-14","label":"arXiv","topic":"LLMs & Text","cites":492,"score":458,"scale":"shares"},{"title":"Bank Risk Projection with Sentiment Analysis","url":"/papers/repec/eee-reecon-v-77-y-2023-i-2-p-226-238/","summary":"Machine learning used to build new insolvency risk rating metric for Brazilian banks, with bank sentiment improving accuracy of prediction models.","featured":"2023-06-14","label":"RePEc","topic":"LLMs & Text","cites":null,"score":27,"scale":"shares"},{"title":"Large Datasets and Hybrid Models","url":"/papers/repec/gam-jjrfmx-v-16-y-2023-i-6-p-298-d-1167483/","summary":"Study explores using machine learning and sentiment analysis to forecast foreign exchange rates and commodity prices.","featured":"2023-06-14","label":"RePEc","topic":"LLMs & Text","cites":null,"score":31,"scale":"shares"},{"title":"Investor Sentiment and Bank Credit","url":"/papers/repec/urv-wpaper-2072-534915/","summary":"Investor sentiment affects bank lending and financial stability, especially for banks with higher credit risk.","featured":"2023-06-14","label":"RePEc","topic":"LLMs & Text","cites":null,"score":11,"scale":"shares"},{"title":"Financial sentiment analysis using FinBERT with application in predicting stock movement","url":"/papers/arxiv/2306.02136/","summary":"LSTM-based neural network predicts market movement using sentiment analysis.","featured":"2023-06-07","label":"arXiv","topic":"LLMs & Text","cites":16,"score":3,"scale":"shares"},{"title":"Toward Textual Internet Immunity","url":"/papers/arxiv/2306.02875/","summary":"The Supreme Court may change Section 230 of the Communications Decency Act, which provides online entities with absolute immunity from lawsuits related to third-party content.","featured":"2023-06-07","label":"arXiv","topic":"LLMs & Text","cites":1,"score":2,"scale":"shares"},{"title":"Zero is Not Hero Yet: Benchmarking Zero-Shot Performance of LLMs for Financial Tasks","url":"/papers/arxiv/2305.16633/","summary":"Investigating effectiveness of zero-shot large language models in finance","featured":"2023-06-01","label":"arXiv","topic":"LLMs & Text","cites":21,"score":5,"scale":"shares"}],"per_quarter":{"2023 Q2":10,"2023 Q3":39,"2023 Q4":46,"2024 Q1":47,"2024 Q2":75,"2024 Q3":107,"2024 Q4":90,"2025 Q1":68,"2025 Q2":55,"2025 Q3":21,"2025 Q4":13,"2026 Q1":0,"2026 Q2":1,"2026 Q3":5}}