Awesome Artificial Intelligence (AI) Research

A curated meta-list of AI research paper lists, surveys, benchmarks, model hubs, and learning resources. It prioritizes active, high-signal, or historically useful resources for researchers and beginners.

Contents

• Foundation Models, LLMs, and Agents
  • Large Language Models
  • Long-Context LLMs
  • Reasoning, Post-Training, and Synthetic Data
  • RAG, Search, and Knowledge-Intensive AI
  • AI Agents
  • Agent Protocols, Memory, and Tooling
  • Alignment, Safety, Security, and Trustworthiness
  • Efficient LLMs and Systems
• Multimodal, Vision-Language, and Generative AI
  • Multimodal and Vision-Language Models
  • Diffusion and Generative Models
  • Computer Vision
  • Embodied AI, Robotics, and World Models
• Core Machine Learning Research
  • General ML, Surveys, and Methods
  • Data-Centric AI and Data Curation
  • Robustness, Interpretability, and Learning Paradigms
  • Reinforcement Learning
  • Federated and Privacy-Preserving ML
  • Graph Learning and Knowledge Graphs
• Domain Applications
  • Natural Language, Speech, and Audio
  • Recommendation, Search, and Ads
  • Science, Medicine, and Quant
  • Anomaly Detection
• Research Feeds, Benchmarks, and Model/Data Hubs
  • Paper Discovery and Code
  • Models, Datasets, and Evaluation
  • Learning Paths

Foundation Models, LLMs, and Agents

Large Language Models

• Awesome LLM - broad LLM papers, model lists, training, inference, evaluation, and tutorials.
• Awesome LLM Reasoning - reasoning, chain-of-thought, o1/R1-style methods, and multimodal reasoning.
• Awesome Reasoning Foundation Models - reasoning with language, vision, and multimodal foundation models.
• Awesome LLM Evaluation Papers - evaluation methods, benchmarks, and survey papers.
• Awesome Multilingual LLMs Papers - multilingual LLM data, training, evaluation, and applications.

Long-Context LLMs

• Thus Spake Long-Context LLM - survey and paper list covering long-context architecture, infrastructure, training, inference, and evaluation.
• RULER - synthetic benchmark for evaluating effective long-context language model performance across configurable tasks and sequence lengths.

Reasoning, Post-Training, and Synthetic Data

• Awesome Inference-Time Scaling - inference/test-time compute, search, self-refinement, and verifier-guided reasoning papers.
• Awesome Test-Time Scaling in LLMs - survey-oriented taxonomy and paper list for test-time scaling.
• Awesome LLM Post-Training - reasoning LLM post-training, RL, distillation, alignment, and evaluation resources.
• TRL - Hugging Face library for SFT, DPO, GRPO, reward modeling, and RLHF post-training.
• verl - flexible RL post-training framework for LLMs with scalable rollout and training infrastructure.
• Open R1 - open reproduction project for DeepSeek-R1 training, evaluation, and synthetic reasoning data.
• Awesome LLM Synthetic Data - synthetic data generation papers, tools, and guides for LLM training and post-training.

RAG, Search, and Knowledge-Intensive AI

• Awesome RAG - retrieval-augmented generation papers, tutorials, tools, and workshops.
• FlashRAG - research toolkit for efficient RAG pipelines, datasets, metrics, and reproducible experiments.
• Awesome LLM KG - unifying LLMs and knowledge graphs.
• Database Learning - database systems, data management, and ML-related database resources.

AI Agents

• Awesome AI Agents - autonomous agent projects and resources.
• Awesome AI Agent Papers - weekly updated 2026 agent research papers on memory, tools, evaluation, workflows, and security.
• Awesome LLM Agent Evaluation - living survey companion and benchmark corpus for LLM-agent evaluation.
• Awesome Code Agents - coding agents, software engineering agents, benchmarks, and research papers.
• Awesome GUI Agent - papers and resources for multimodal GUI, browser, and computer-use agents.
• Awesome Computer Use - computer-use GUI agent papers, projects, blogs, and benchmarks.
• Awesome Data Agents - agents for data preparation, analysis, and data management.
• Awesome Agents for Science - LLM agents for scientific research and development.
• Awesome Edge AI for Multimodal Agents - efficient multimodal agents on mobile and edge devices.
• Open Deep Research - open-source research agent implementation for iterative search, synthesis, and reporting.

Agent Protocols, Memory, and Tooling

• Awesome MCP Servers - large curated list of Model Context Protocol servers for connecting agents to tools and data sources.
• Awesome MCP Clients - MCP-capable clients and applications across desktop, IDE, CLI, and agent workflows.
• MCP Agent - framework and patterns for building agents on top of Model Context Protocol.
• Awesome Agent Memory - papers, systems, and benchmarks for long-term memory, context engineering, retrieval, and reasoning in agents.

Alignment, Safety, Security, and Trustworthiness

• Awesome LLM Safety Papers - LLM safety, alignment, jailbreaks, privacy, and robustness papers.
• AgentHarm - benchmark for harmful multi-step LLM agent tasks and jailbreak robustness.
• Awesome AI Safety - AI quality, testing, robustness, fairness, and privacy.
• Awesome AI Alignment - alignment research resources and reading paths.
• Awesome AI Security - adversarial ML, LLM security, and broader AI security.
• Machine Learning for Cyber Security - ML methods for cyber security.
• Threat Detection and Hunting - practical security detection and hunting research resources.

Efficient LLMs and Systems

• Awesome Efficient LLM - quantization, pruning, KV cache compression, MoE, serving, and efficient reasoning.
• Awesome Resource-Efficient LLM Papers - resource-efficient LLM pre-training, fine-tuning, and inference.
• Awesome LLMOps - production LLM lifecycle, evaluation, observability, deployment, and prompt management.
• Model Inference Deployment - inference deployment frameworks across TensorRT, ONNX Runtime, OpenVINO, TVM, mobile runtimes, and edge accelerators.

Multimodal, Vision-Language, and Generative AI

Multimodal and Vision-Language Models

• Awesome Multimodal Modeling - unified multimodal modeling across MLLMs, understanding, generation, and omni-modal agents.
• Awesome Multimodal Large Language Models - current MLLM papers, benchmarks, datasets, and evaluation.
• Awesome Large Multimodal Models - large multimodal models from input-output space perspectives.
• Awesome Large Vision-Language Model - VLM papers, medical foundation models, benchmarks, and presentations.
• Awesome Vision-Language Models for Vision Tasks - VLMs for classification, detection, segmentation, and other vision tasks.
• Awesome Multimodal Reasoning - multimodal reasoning and RL-based visual reasoning papers, surveys, benchmarks, and datasets.
• Awesome Multimodal Foundation Model Evaluations - survey companion and paper list for evaluating multimodal foundation models.
• Awesome Document Understanding - document AI, OCR, layout understanding, and document-oriented MLLM resources.
• Awesome Multimodal ML - classic and modern multimodal machine learning resources.
• MMLongBench - benchmark and evaluation code for long-context vision-language models.
• Video-MME-v2 - benchmark and evaluation pipeline for comprehensive video understanding in multimodal models.
• ZeroBench - challenging visual reasoning benchmark, dataset, and leaderboard for large multimodal models.
• WorldBench - visually diverse multimodal reasoning benchmark with evaluation code for large multimodal models.
• MMMU - multidisciplinary multimodal understanding and reasoning benchmark with evaluation code and leaderboard.
• MathVista - benchmark, dataset, and leaderboard for mathematical reasoning in visual contexts.

Diffusion and Generative Models

• Awesome Diffusion Models - diffusion and score-based generative model papers and resources.
• Awesome Video World Models - video generation and interactive world-model papers with a state/dynamics taxonomy.
• Awesome Video Diffusions - auto-updated video diffusion and video generation paper tracker.
• Really Awesome GAN - GAN papers and resources.
• Awesome Normalizing Flows - normalizing flow papers and implementations.

Computer Vision

• Awesome Deep Vision - classic deep vision resources with high historical value.
• Awesome Object Detection - object detection papers.
• Deep Learning Object Detection - object detection paper list with many classic references.
• 3D Gaussian Splatting Papers - actively maintained 3D Gaussian Splatting paper tracker, including conference-specific lists.
• Awesome Lane Detection - lane detection papers.
• Awesome Visual Transformer - transformers for computer vision.
• Awesome Point Cloud Analysis - 3D point cloud papers and datasets.
• 3D Point Cloud - point cloud deep learning resources.
• Awesome Crowd Counting - crowd counting papers and datasets.
• Awesome Face Recognition - face detection, recognition, alignment, generation, and anti-spoofing.
• Awesome Image Classification - image classification papers and code.
• Awesome CBIR Papers - image retrieval and content-based image retrieval.

Embodied AI, Robotics, and World Models

• Awesome Embodied AI - embodied AI surveys, VLA models, datasets, simulators, humanoids, and safety.
• Awesome Physical AI - VLA models, robot foundation models, world models, diffusion policies, evaluation, and safety.
• Awesome VLA Robotics - vision-language-action models and robot foundation model papers.
• Awesome VLA Papers - VLA papers covering robotics, autonomous driving, world models, and spatial reasoning.
• Awesome World Models - world models for video generation, embodied AI, robotics, and autonomous driving.
• Awesome World Models for Robots - world-model papers, datasets, and workshops focused on robotics.
• Awesome Autonomous Vehicles - self-driving and autonomous vehicle resources.

Core Machine Learning Research

General ML, Surveys, and Methods

• Machine Learning Surveys - survey papers across ML, NLP, CV, graphs, RL, and recommendation.
• Machine Learning Surveys and Tutorials - older but still useful survey collection.
• AutoML Papers - automated machine learning papers and tutorials.
• Awesome AutoDL - automated deep learning and neural architecture search.
• Awesome Architecture Search - neural architecture search resources.
• Awesome Causality Algorithms - causal discovery and inference algorithms.
• Awesome Decision Tree Papers - decision tree, classification tree, and regression tree papers.
• Bayesian Deep Learning Survey - Bayesian deep learning survey resources.
• Deep Learning Uncertainty - predictive uncertainty in deep learning.
• Awesome Online Machine Learning - online and streaming ML resources.
• Awesome Time Series Papers - recent time-series papers and code across forecasting, anomaly detection, foundation models, and representation learning.

Data-Centric AI and Data Curation

• Awesome Open Data-Centric AI - open-source tools for data-centric AI on unstructured data.
• Data-Juicer - data processing, cleaning, filtering, and analysis toolkit for foundation-model datasets.
• Awesome Synthetic Datasets - practical resources and examples for creating synthetic text and vision datasets.

Robustness, Interpretability, and Learning Paradigms

• Adversarial Machine Learning - adversarial ML papers and resources.
• Awesome Machine Learning Interpretability - responsible ML, interpretability, and explainability.
• Awesome LLM Interpretability - LLM interpretability tools, papers, communities, sparse autoencoders, probing, and mechanistic interpretability resources.
• Awesome Interpretable Machine Learning - interpretable ML papers.
• Awesome Explainable AI - explainability papers and resources.
• Awesome Knowledge Distillation - distillation papers and implementations.
• Awesome Self-Supervised Learning - self-supervised learning methods.
• Awesome Learning with Label Noise - noisy-label learning papers.
• Awesome Meta Learning - meta-learning papers and resources.
• Awesome Transfer Learning - transfer learning and domain adaptation.
• Awesome Domain Adaptation - domain adaptation papers and code.
• Awesome Open World Learning - open-world, open-set, and incremental recognition.
• Awesome Incremental Learning - incremental and lifelong learning.
• Continual Learning Papers - continual learning papers.

Reinforcement Learning

• Awesome RL - reinforcement learning resources.
• Awesome Deep RL - deep reinforcement learning papers.
• Deep Reasoning Papers - neural-symbolic, logical, visual, and planning-oriented reasoning.

Federated and Privacy-Preserving ML

• Awesome Federated Learning - federated learning papers and resources.
• Awesome Federated Machine Learning - federated ML papers, books, code, tutorials, and videos.

Graph Learning and Knowledge Graphs

• Graph-Based Deep Learning Literature - graph deep learning papers by venue.
• Awesome Graph Neural Networks - GNN paper list.
• Must-Read Papers on GNN - curated GNN papers.
• Awesome Self-Supervised GNN - self-supervised learning on graphs.
• Awesome Graph Classification - graph classification and representation learning.
• Awesome Community Detection - community detection papers.
• Knowledge Graphs - knowledge graph papers and resources.
• Knowledge Representation Learning - knowledge representation learning papers.

Domain Applications

Natural Language, Speech, and Audio

• Awesome NLG - natural language generation papers and resources.
• Awesome Sentence Embedding - sentence embedding resources.
• Awesome Large Speech Model - large speech/audio model papers, datasets, tools, applications, and benchmarks.
• Awesome Speech Recognition and Speech Synthesis Papers - ASR and TTS papers.
• Awesome Speech Enhancement - speech enhancement papers and tutorials.
• Speech Synthesis Papers - speech synthesis paper list.
• Awesome Deep Learning Music - deep learning for music.

Recommendation, Search, and Ads

• RSTutorials - must-read recommender system papers.
• Awesome Recommender System - recommender system papers.
• Awesome Deep Learning Papers for Search, Recommendation, and Advertising - industrial search, recommendation, ranking, ads, LLM, and RL papers.
• Real-Time Bidding Papers - RTB and computational advertising papers.

Science, Medicine, and Quant

• Awesome AI for Science - AI tools, papers, datasets, and frameworks for scientific discovery.
• Awesome Scientific Language Models - scientific language models across math, physics, chemistry, materials, biology, medicine, and geoscience.
• Awesome AI for Research - AI-for-research papers and systems organized by research stage, role, domain, and evaluation.
• Awesome Deep Research Agent - deep research agents, AI scientist systems, search-augmented reasoning, and research workflow papers.
• Awesome AI Scientist - survey-style collection on AI scientists, AI researchers, AI engineers, and automated research pipelines.
• SciAgentArena - living benchmark for evaluating AI agents on real-world scientific research tasks across domains.
• ScienceAgentBench - benchmark and evaluation harness for language agents on data-driven scientific discovery tasks.
• Terminal-Bench-Science - terminal-agent benchmark for expert-curated scientific research workflows across domains.
• AIRS-Bench - benchmark and evaluation code for AI research science agents on tasks from machine learning papers.
• FrontierScience - expert-level scientific reasoning benchmark across physics, chemistry, and biology.
• Awesome DeepBio - deep learning for computational biology.
• Helical - framework for using and fine-tuning bio foundation models across genomics and transcriptomics.
• ChemBench - benchmark suite and evaluation package for chemistry-focused language and multimodal models.
• Awesome GAN for Medical Imaging - medical image synthesis papers.
• Awesome AI Agents for Healthcare - healthcare agentic AI papers and resources.
• HealthBench - benchmark for evaluating AI systems on realistic health conversations with physician-written rubrics.
• Awesome Quant Machine Learning Trading - quant trading and ML resources.

Anomaly Detection

• Anomaly Detection Resources - anomaly detection papers, books, code, and datasets.
• Awesome Anomaly Detection - anomaly detection papers and resources.

Research Feeds, Benchmarks, and Model/Data Hubs

Paper Discovery and Code

• Papers with Code - papers, tasks, datasets, code, and leaderboards.
• Hugging Face Papers - daily ML paper discovery and discussion.
• arXiv Sanity - arXiv search and paper recommendation.
• Semantic Scholar - AI-powered academic search.
• OpenReview - conference submissions, reviews, and accepted papers.

Models, Datasets, and Evaluation

• Hugging Face Models - model hub for open models.
• Hugging Face Datasets - dataset hub for ML research.
• Open LLM Leaderboard - open model evaluation leaderboard.
• LMArena Leaderboard - human-preference leaderboard for chat and multimodal models.
• LiveBench - contamination-resistant LLM benchmark with refreshed objective tasks and a public leaderboard.
• Epoch AI Benchmarks - benchmark results hub for tracking frontier model capabilities across major evaluation suites.
• Artificial Analysis - independent benchmark hub for language, multimodal, speech, agent, provider, and hardware comparisons.
• Awesome AI Benchmarks - searchable collection of benchmarks for agents, reasoning, code, multimodal, translation, and other AI domains.
• Awesome LLM Eval - tools, datasets, benchmarks, leaderboards, papers, and demos for LLM evaluation.
• Awesome Scientific LLM Benchmarks - benchmarks for evaluating LLMs on scientific reasoning, discovery, and domain knowledge.
• LM Evaluation Harness - widely used framework for evaluating language models on many benchmark tasks.
• OpenAI Evals - framework and open registry for evaluating LLMs and LLM systems.
• Inspect AI - open-source framework for LLM, multimodal, coding, and agent evaluations.
• Inspect Evals - community collection of ready-to-run benchmark implementations for Inspect AI.
• OpenCompass - open evaluation platform for LLMs and multimodal models.
• VLMEvalKit - open-source toolkit for evaluating large vision-language models across many multimodal benchmarks.
• HELM - holistic evaluation framework and leaderboards for language, multimodal, safety, and domain benchmarks.
• HAL - standardized cost-aware leaderboard and harness for reproducible AI agent evaluation.
• GAIA - benchmark and leaderboard for tool-using, multimodal, web-browsing AI assistants.
• BrowseComp - benchmark for browsing agents that must locate hard-to-find information on the web.
• AgentBench - benchmark and leaderboard for evaluating LLM agents across interactive environments.
• ARC-AGI - abstraction-and-reasoning benchmark for testing generalization on novel visual puzzle tasks.
• WebArena - realistic self-hosted web environment and benchmark for autonomous web agents.
• BrowserGym - extensible Gym environment for web-agent research with packaged browser automation benchmarks.
• OSWorld - benchmark for multimodal agents completing open-ended tasks in real computer environments.
• AndroidWorld - environment and benchmark for autonomous agents operating Android apps through an emulator.
• tau2-bench - tool-agent-user interaction benchmark for text and voice customer-service agents.
• The Agent Company - benchmark for autonomous agents completing workplace tasks in a simulated software company.
• AutomationBench - benchmark for evaluating AI agents on realistic business workflows across simulated SaaS tools.
• Berkeley Function Calling Leaderboard - leaderboard and evaluation suite for LLM tool-use and function-calling capabilities.
• Humanity's Last Exam - expert-level benchmark designed to test frontier models across broad academic subjects.
• FrontierMath - advanced mathematical reasoning benchmark and leaderboard for evaluating frontier AI systems.
• SWE-bench - benchmark and evaluation harness for real-world software engineering issue resolution.
• SWE-bench Multimodal - benchmark for software engineering agents on real-world issues with visual elements.
• SWE-bench Pro - long-horizon software engineering benchmark for agents on larger production codebases.
• SWE-Lancer - benchmark of real freelance software engineering tasks with end-to-end tests and managerial decisions.
• LiveCodeBench - contamination-aware benchmark for code generation, repair, execution, and test prediction.
• BigCodeBench - practical code-generation benchmark with diverse function calls, complex instructions, and a leaderboard.
• Terminal-Bench - realistic terminal-agent benchmark suite spanning software engineering, ML, security, and data tasks.
• PaperBench - benchmark for end-to-end replication of state-of-the-art AI papers.
• MLE-bench - benchmark for measuring how well AI agents perform at machine learning engineering.
• RE-Bench - benchmark for evaluating AI agents on frontier AI research and development tasks against human experts.
• MLGym - framework and benchmark for evaluating AI research agents on open-ended machine learning tasks.
• SciCode - scientist-curated benchmark for code generation on realistic scientific research problems.
• AstaBench - benchmark suite and leaderboards for evaluating agents on scientific research tasks.
• MTEB - embedding benchmark suite and leaderboard across languages, tasks, and modalities.
• Simple Evals - lightweight evaluation examples and baseline eval implementations.
• DeepEval - LLM evaluation framework for unit-style tests, RAG metrics, agents, and CI workflows.
• Awesome Public Datasets - topic-centric list of public datasets.
• LeRobot - open models, datasets, tools, and tutorials for robotics research.
• MinerU - open-source document extraction tool for converting complex PDFs and Office files into Markdown/JSON.
• OpenAI Cookbook - practical examples and guides for building and evaluating AI systems.

Learning Paths

• Dive into Deep Learning - interactive deep learning book with code.
• fast.ai Practical Deep Learning - practical deep learning course.
• CS229 Machine Learning - Stanford machine learning course.
• CS231n Convolutional Neural Networks for Visual Recognition - computer vision fundamentals.
• CS224N Natural Language Processing with Deep Learning - NLP and language model foundations.
• CS324 Large Language Models - large language model concepts, training, evaluation, and societal impact.
• CS336 Language Modeling from Scratch - implementation-heavy Stanford course on building, training, scaling, and evaluating language models.
• Berkeley CS285 Deep Reinforcement Learning - deep RL lectures and assignments.