homeservicesworkaboutblogfree templatescontactFree Tools →Free AI ModelsResearch LibraryROI CalculatorSavings CalculatorAI Readiness ScoreHire vs. AutomateAutomation Quote
book a 30-min call
home / research-papers

AI research library

The papers that actually changed what gets built, from the 1958 perceptron to the 2025 frontier reports. Every identifier machine-verified. Each entry says why it mattered, not just what it claimed.

74papers indexed
74/74identifiers verified
8eras, read in order
19582025years covered
How to use this library

A paper tells you what a lab discovered. It does not tell you whether it changes anything you are paying for. So every entry here is chosen for whether it should change what you build, and every summary below says why it mattered rather than restating the abstract. Full write-ups, covering what each paper means for your business and how it connects to systems we have actually built, are published in batches as each one is finished.

That is also the entry requirement. A paper is here because it is canon a technical buyer expects to find, or because it changes a decision our clients make. Nineteen papers met neither test and were removed rather than published as bare summaries. A shorter library where every page has a reason to exist beats a longer one padded with abstracts you could read at the source.

how the library is organisedPapers by era
0
total
  • Foundations12%9
  • The Transformer era7%5
  • Scaling and alignment7%5
  • Reasoning and test-time compute7%5
  • Automation, agents and the evidence28%21
  • Efficiency, inference and serving16%12
  • Multimodal and generative8%6
  • Open-weight frontier reports15%11
All 74 papers in the library, counted by era at build time.
who published themOrganisations with more than one paper here
Google / DeepMind
0 papers
OpenAI
0 papers
Meta AI
0 papers
UC Berkeley
0 papers
Stanford
0 papers
Microsoft
0 papers
MIT
0 papers
University of Toronto
0 papers
Carnegie Mellon University
0 papers
DeepSeek
0 papers
Princeton University
0 papers
Primary organisation on each paper, counted at build time. Organisations with a single paper are not shown.
If you run operations rather than engineering

Start with Automation, agents and the evidence. It is the section written for the question you actually arrive with — does this work on real business processes, and what does the evidence say. It covers the field studies (5,179 support agents measured in production; 758 consultants randomised, including the task where AI made them worse), the process-automation foundations that predate the AI wave and still decide whether a project succeeds, and the benchmarks that put agents on real company work rather than exams. The best agent in the most realistic of those benchmarks completed under a quarter of its tasks.

It is deliberately the most sceptical section in the library. A vendor will show you the productivity number; this is where the conditions attached to it are written down.

Where you actually read the papers

On the publisher’s site, not here. Authors keep copyright and most arXiv submissions are filed under a licence that does not permit republication, so hosting the PDFs would not be ours to do. Every entry links straight to the arXiv abstract, the PDF, or the journal DOI. What we add is the part the paper does not contain: whether it should change what you build.

Why the verification matters

Language models are very good at producing arXiv identifiers that look correct and resolve to a different paper entirely. Every arXiv ID on this page was queried against the arXiv API and the returned title compared against the title we publish before anything shipped. Journal papers carry a DOI. If a citation here is wrong, it is wrong in the source, not invented here.

Start here

20 papers that carry most of the explanatory weight. Read in this order and the rest of the library becomes much easier to navigate.

2011

Process Mining Manifesto

Established process mining as a discipline: reconstruct how a business process actually runs from the event logs your systems already produce, rather than from how someone says it runs.

TU Eindhoven·DOI 10.1007/978-3-642-28108-2_19
2017

Attention Is All You Need

Removed recurrence entirely and replaced it with self-attention. Because attention parallelises across the sequence and recurrence does not, training could finally use GPUs fully — which is what made scale possible.

Google·arXiv:1706.03762
2020

Language Models are Few-Shot Learners

At 175B parameters, a model started solving tasks from examples in the prompt with no gradient updates. In-context learning was not designed; it appeared.

OpenAI·arXiv:2005.14165
2020

Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks

Retrieve documents at inference time and condition generation on them, instead of hoping the weights memorised the fact.

Meta AI / UCL·arXiv:2005.11401
2021

LoRA: Low-Rank Adaptation of Large Language Models

Freeze the base weights and train small low-rank matrices alongside them. Trainable parameters drop by orders of magnitude.

Microsoft·arXiv:2106.09685
2021

Learning Transferable Visual Models From Natural Language Supervision

Train image and text encoders into a shared embedding space on 400M web pairs. Zero-shot classification by comparing to text prompts.

OpenAI·arXiv:2103.00020
2021

High-Resolution Image Synthesis with Latent Diffusion Models

Run diffusion in a compressed latent space rather than pixel space. Cut the compute by more than an order of magnitude.

LMU Munich / Runway·arXiv:2112.10752
2022

Training Compute-Optimal Large Language Models

Showed the field had been building models far too large for the data they were trained on. For a fixed compute budget, parameters and tokens should scale roughly equally.

DeepMind·arXiv:2203.15556
2022

Training Language Models to Follow Instructions with Human Feedback

RLHF: collect human preference comparisons, train a reward model, optimise the policy against it. A 1.3B InstructGPT was preferred to the 175B base GPT-3.

OpenAI·arXiv:2203.02155
2022

Chain-of-Thought Prompting Elicits Reasoning in Large Language Models

Asking a model to show its working improves accuracy on multi-step problems, and the effect only appears above a certain scale.

Google·arXiv:2201.11903
2022

FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness

Attention was memory-bandwidth-bound, not compute-bound. Tiling the computation to keep it in SRAM makes it several times faster with identical outputs and no approximation.

Stanford·arXiv:2205.14135
2023

Generative AI at Work

A field study of 5,179 real customer-support agents using an AI assistant in production, not a lab task. Average productivity rose about 14% in issues resolved per hour — and the gain was concentrated almost entirely among the least experienced staff, while the most experienced saw little or none.

Stanford / MIT·DOI 10.3386/w31161
2023

Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality

758 BCG consultants, randomised. On tasks inside the AI's capability the group using it was substantially faster and rated higher quality. On a task deliberately placed just outside that capability, consultants using AI were MORE likely to reach the wrong answer than those without it.

Harvard Business School / BCG·DOI 10.2139/ssrn.4573321
2024

The Llama 3 Herd of Models

A 92-page account of a frontier training run: data curation, 15T tokens, scaling decisions, infrastructure failures and all.

Meta AI·arXiv:2407.21783
2024

DeepSeek-V3 Technical Report

671B MoE with 37B active. Introduced auxiliary-loss-free load balancing and multi-token prediction, and documented an FP8 mixed-precision training pipeline in detail.

DeepSeek·arXiv:2412.19437
2024

TheAgentCompany: Benchmarking LLM Agents on Consequential Real World Tasks

Built a simulated software company — codebase, intranet, colleagues to message — and set agents to do real employee tasks end to end. The best agent completed under a quarter of them autonomously.

Carnegie Mellon University·arXiv:2412.14161
2024

τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains

Tests agents in retail and airline customer-service scenarios where they must follow a written policy, use tools, and talk to a simulated customer. Adds a consistency measure: run the same scenario several times and see whether the agent behaves the same way.

Sierra AI / Princeton·arXiv:2406.12045
2024

AI Agents That Matter

A critique of how agent capability is reported. Shows that many published gains come from expensive search over many attempts rather than better reasoning, that cost is usually omitted, and that a trivially simple baseline sometimes matches complex agent architectures.

Princeton University·arXiv:2407.01502
2025

DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

Reasoning learned by pure reinforcement learning from verifiable rewards, with no supervised examples of human reasoning at all. Self-verification and backtracking emerged on their own during training.

DeepSeek·arXiv:2501.12948
2025

Qwen3 Technical Report

Dense and MoE models from 0.6B to 235B under Apache 2.0, unifying thinking and non-thinking modes in one model with a controllable thinking budget.

Alibaba (Qwen)·arXiv:2505.09388

The full library

Era
Lab
Topic
View
74 of 74 papers

Foundations

The results the whole field rests on. Mostly pre-2015, mostly not on arXiv, and still worth reading in the original.

1958

The Perceptron: A Probabilistic Model for Information Storage and Organization in the Brain

Where it starts. A learnable linear classifier with a convergence proof, and the first working demonstration that a machine could adjust its own weights from examples rather than being programmed.

Cornell Aeronautical Laboratory·Frank Rosenblatt·DOI 10.1037/h0042519
1986

Learning Representations by Back-Propagating Errors

The answer to the XOR objection: with hidden layers and the chain rule you can train arbitrarily deep networks. Every model on this site is trained by a direct descendant of this algorithm.

University of Toronto·David Rumelhart, Geoffrey Hinton, Ronald Williams·DOI 10.1038/323533a0
1997

Long Short-Term Memory

Solved the vanishing-gradient problem with gated memory cells, making it possible to learn long-range dependencies in sequences.

TU Munich / IDSIA·Sepp Hochreiter, Jürgen Schmidhuber·DOI 10.1162/neco.1997.9.8.1735
2012

ImageNet Classification with Deep Convolutional Neural Networks

Won ImageNet by a margin so large it ended the debate about whether deep learning worked. Trained on two consumer GPUs.

University of Toronto·Alex Krizhevsky, Ilya Sutskever, Geoffrey Hinton·NeurIPS 2012
2013

Efficient Estimation of Word Representations in Vector Space

Words as dense vectors where geometric relationships encode semantic ones. The origin of the embedding.

Google·Tomas Mikolov, Kai Chen, Greg Corrado, Jeffrey Dean·arXiv:1301.3781
2014

Adam: A Method for Stochastic Optimization

Adaptive per-parameter learning rates from first and second moment estimates. Made training deep networks reliable without heroic hyperparameter tuning.

University of Amsterdam / University of Toronto·Diederik Kingma, Jimmy Ba·arXiv:1412.6980
2014

Neural Machine Translation by Jointly Learning to Align and Translate

The first attention mechanism. Instead of squeezing a sentence into one fixed vector, let the decoder look back at all encoder states and weight them.

Université de Montréal·Dzmitry Bahdanau, Kyunghyun Cho, Yoshua Bengio·arXiv:1409.0473
2015

Deep Residual Learning for Image Recognition

Skip connections let gradients flow through very deep networks, making 100+ layer models trainable for the first time.

Microsoft Research Asia·Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun·arXiv:1512.03385
2015

Distilling the Knowledge in a Neural Network

A small model trained on a large model's output distribution learns more than one trained on hard labels alone.

Google·Geoffrey Hinton, Oriol Vinyals, Jeff Dean·arXiv:1503.02531

The Transformer era

2017-2020. Attention replaces recurrence, and scale becomes the strategy.

Scaling and alignment

2020-2023. How big, trained on how much, and how to make it do what was asked.

Reasoning and test-time compute

2022 onwards. The shift from bigger models to models that think for longer before answering.

Automation, agents and the evidence

What actually happens when this technology meets a business process: the field studies, the process-automation foundations, and the benchmarks that measure agents on real work rather than exams. The most commercially relevant section here, and the most sceptical.

2011

Process Mining Manifestoessential

Established process mining as a discipline: reconstruct how a business process actually runs from the event logs your systems already produce, rather than from how someone says it runs.

TU Eindhoven·Wil van der Aalst et al. (IEEE Task Force on Process Mining)·DOI 10.1007/978-3-642-28108-2_19
2020

Integrating Robotic Process Automation into Business Process Management

Sets out where rule-based RPA belongs inside a managed process, and where it becomes technical debt: brittle bots imitating clicks on interfaces nobody controls.

Sapienza University of Rome·Simone Agostinelli, Andrea Marrella, Massimo Mecella·DOI 10.1007/978-3-030-58779-6_9
2023

Generative AI at Workessential

A field study of 5,179 real customer-support agents using an AI assistant in production, not a lab task. Average productivity rose about 14% in issues resolved per hour — and the gain was concentrated almost entirely among the least experienced staff, while the most experienced saw little or none.

Stanford / MIT·Erik Brynjolfsson, Danielle Li, Lindsey R. Raymond·DOI 10.3386/w31161
2023

Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Qualityessential

758 BCG consultants, randomised. On tasks inside the AI's capability the group using it was substantially faster and rated higher quality. On a task deliberately placed just outside that capability, consultants using AI were MORE likely to reach the wrong answer than those without it.

Harvard Business School / BCG·Fabrizio Dell'Acqua, Edward McFowland III, Ethan Mollick et al.·DOI 10.2139/ssrn.4573321
2023

Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence

A randomised experiment on mid-level professional writing tasks. Time taken fell about 40% and output quality rose about 18%, with the largest gains again going to the weakest writers.

MIT·Shakked Noy, Whitney Zhang·DOI 10.1126/science.adh2586
2023

GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models

Mapped US occupations against LLM capability task by task. Around 80% of workers are in jobs where at least 10% of tasks could be affected, and about 19% are in jobs where more than half could be.

OpenAI / University of Pennsylvania·Tyna Eloundou, Sam Manning, Pamela Mishkin, Daniel Rock·arXiv:2303.10130
2023

AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation

A framework for building systems from multiple conversing agents with distinct roles, including agents that write and execute code and a human participant in the loop.

Microsoft Research·Qingyun Wu, Gagan Bansal, Jieyu Zhang et al.·arXiv:2308.08155
2023

Toolformer: Language Models Can Teach Themselves to Use Tools

Showed a model could learn, from its own self-generated data, when to call an external tool — a calculator, a search, an API — and how to use the result.

Meta AI·Timo Schick, Jane Dwivedi-Yu, Roberto Dessì et al.·arXiv:2302.04761
2023

Gorilla: Large Language Model Connected with Massive APIs

Trained a model to call thousands of real APIs correctly and used retrieval over live API documentation so it stays right when those APIs change.

UC Berkeley·Shishir G. Patil, Tianjun Zhang, Xin Wang, Joseph E. Gonzalez·arXiv:2305.15334
2023

GAIA: a Benchmark for General AI Assistants

Questions that are trivial for a person but require an assistant to chain web browsing, file handling and reasoning. Humans scored around 92%; the best assistant systems at publication scored a small fraction of that.

Meta AI / Hugging Face·Grégoire Mialon, Clémentine Fourrier, Craig Swift et al.·arXiv:2311.12983
2023

WebArena: A Realistic Web Environment for Building Autonomous Agents

A self-hosted environment with working e-commerce, forum, CMS and code sites, so agents can be tested on real end-to-end web tasks rather than static pages.

Carnegie Mellon University·Shuyan Zhou, Frank F. Xu, Hao Zhu et al.·arXiv:2307.13854
2023

MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework

Gave a multi-agent system explicit human workflows — product manager, architect, engineer roles with standardised documents passed between them — rather than letting agents converse freely.

DeepWisdom / KAUST·Sirui Hong, Mingchen Zhuge, Jiaqi Chen et al.·arXiv:2308.00352
2023

The Rise and Potential of Large Language Model Based Agents: A Survey

A comprehensive map of the agent field: the brain-perception-action framing, single versus multi-agent designs, and where each approach has been applied.

Fudan University·Zhiheng Xi, Wenxiang Chen, Xin Guo et al.·arXiv:2309.07864
2024

The Simple Macroeconomics of AI

A deliberately conservative estimate: AI raises total factor productivity by roughly 0.5% and GDP by around 1% over ten years — an order of magnitude below the widely quoted forecasts.

MIT·Daron Acemoglu·DOI 10.3386/w32487
2024

TheAgentCompany: Benchmarking LLM Agents on Consequential Real World Tasksessential

Built a simulated software company — codebase, intranet, colleagues to message — and set agents to do real employee tasks end to end. The best agent completed under a quarter of them autonomously.

Carnegie Mellon University·Frank F. Xu, Yufan Song, Boxuan Li et al.·arXiv:2412.14161
2024

WorkArena: How Capable Are Web Agents at Solving Common Knowledge Work Tasks?

A benchmark of routine enterprise tasks inside a real ServiceNow instance — raising requests, filtering lists, filling forms, navigating dashboards. Agents performed far below the human baseline on work most office staff consider trivial.

ServiceNow Research·Alexandre Drouin, Maxime Gasse, Massimo Caccia et al.·arXiv:2403.07718
2024

τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domainsessential

Tests agents in retail and airline customer-service scenarios where they must follow a written policy, use tools, and talk to a simulated customer. Adds a consistency measure: run the same scenario several times and see whether the agent behaves the same way.

Sierra AI / Princeton·Shunyu Yao, Noah Shinn, Pedram Razavi, Karthik Narasimhan·arXiv:2406.12045
2024

AI Agents That Matteressential

A critique of how agent capability is reported. Shows that many published gains come from expensive search over many attempts rather than better reasoning, that cost is usually omitted, and that a trivially simple baseline sometimes matches complex agent architectures.

Princeton University·Sayash Kapoor, Benedikt Stroebl, Zachary S. Siegel et al.·arXiv:2407.01502
2024

Executable Code Actions Elicit Better LLM Agents

Instead of emitting a structured tool call, have the agent write and run actual code. Fewer steps, better success rates, and it composes operations naturally.

UIUC·Xingyao Wang, Yangyi Chen, Lifan Yuan et al.·arXiv:2402.01030
2024

OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments

Real operating systems, real applications, tasks spanning several programs. Humans completed over 72%; the best agents managed a small fraction.

University of Hong Kong·Tianbao Xie, Danyang Zhang, Jixuan Chen et al.·arXiv:2404.07972
2024

SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering

Showed that agent performance depends heavily on the interface it is given. Redesigning the tools — how files are viewed, edited and searched — improved results dramatically with no change to the underlying model.

Princeton University·John Yang, Carlos E. Jimenez, Alexander Wettig et al.·arXiv:2405.15793

Efficiency, inference and serving

How these things are actually made to run at a price anyone can pay. The least glamorous and most commercially relevant category here.

2019

Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism

Tensor parallelism: split individual matrix multiplications across GPUs with minimal communication.

NVIDIA·Mohammad Shoeybi et al.·arXiv:1909.08053
2020

Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasksessential

Retrieve documents at inference time and condition generation on them, instead of hoping the weights memorised the fact.

Meta AI / UCL·Patrick Lewis et al.·arXiv:2005.11401
2021

LoRA: Low-Rank Adaptation of Large Language Modelsessential

Freeze the base weights and train small low-rank matrices alongside them. Trainable parameters drop by orders of magnitude.

Microsoft·Edward Hu et al.·arXiv:2106.09685
2022

FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awarenessessential

Attention was memory-bandwidth-bound, not compute-bound. Tiling the computation to keep it in SRAM makes it several times faster with identical outputs and no approximation.

Stanford·Tri Dao, Daniel Fu, Stefano Ermon, Atri Rudra, Christopher Ré·arXiv:2205.14135
2022

Fast Inference from Transformers via Speculative Decoding

A small draft model proposes several tokens; the large model verifies them in one pass. Output is provably identical to the large model alone.

Google Research·Yaniv Leviathan, Matan Kalman, Yossi Matias·arXiv:2211.17192
2022

GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers

One-shot weight quantisation to 3-4 bits with second-order error correction and no retraining.

IST Austria / ETH Zürich·Elias Frantar, Saleh Ashkboos, Torsten Hoefler, Dan Alistarh·arXiv:2210.17323
2023

Efficient Memory Management for Large Language Model Serving with PagedAttention

Manage the KV cache like operating-system virtual memory, in non-contiguous pages. Cuts the memory waste that limits batch size.

UC Berkeley·Woosuk Kwon et al.·arXiv:2309.06180
2023

SGLang: Efficient Execution of Structured Language Model Programs

RadixAttention reuses KV cache across requests sharing a prefix — exactly the pattern in agent loops that resend the same system prompt.

UC Berkeley / Stanford·Lianmin Zheng et al.·arXiv:2312.07104
2023

QLoRA: Efficient Finetuning of Quantized LLMs

LoRA on top of a 4-bit base model, with 4-bit NormalFloat and double quantisation. Fine-tuned a 65B model on one 48 GB GPU.

University of Washington·Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, Luke Zettlemoyer·arXiv:2305.14314
2023

AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration

Not all weights matter equally. Protect the ~1% with the largest activation magnitudes and quantise the rest hard.

MIT·Ji Lin et al.·arXiv:2306.00978
2023

GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints

Share key and value heads across groups of query heads. Shrinks the KV cache with almost no quality loss.

Google Research·Joshua Ainslie et al.·arXiv:2305.13245
2023

Mamba: Linear-Time Sequence Modeling with Selective State Spaces

A selective state-space model with linear rather than quadratic scaling in sequence length, and no KV cache.

CMU / Princeton·Albert Gu, Tri Dao·arXiv:2312.00752

Multimodal and generative

Vision, audio, images and the models that cross between them.

Open-weight frontier reports

2024 onwards. The technical reports from labs that published their weights — the closest thing to documentation for the models on our free-models page.

2021

Evaluating Large Language Models Trained on Code

The Codex paper. Introduced HumanEval and the pass@k metric.

OpenAI·Mark Chen et al.·arXiv:2107.03374
2023

Llama 2: Open Foundation and Fine-Tuned Chat Models

The release that made commercially usable open weights normal, with an unusually candid account of its RLHF process.

Meta AI·Hugo Touvron et al.·arXiv:2307.09288
2023

GPT-4 Technical Report

Notable as much for what it withholds as what it reports: no architecture, no parameter count, no training data. Extensive capability and safety evaluation.

OpenAI·OpenAI·arXiv:2303.08774
2023

Gemini: A Family of Highly Capable Multimodal Models

Natively multimodal from pretraining rather than by bolting a vision encoder onto a text model.

Google DeepMind·Gemini Team·arXiv:2312.11805
2024

The Llama 3 Herd of Modelsessential

A 92-page account of a frontier training run: data curation, 15T tokens, scaling decisions, infrastructure failures and all.

Meta AI·Llama Team·arXiv:2407.21783
2024

DeepSeek-V3 Technical Reportessential

671B MoE with 37B active. Introduced auxiliary-loss-free load balancing and multi-token prediction, and documented an FP8 mixed-precision training pipeline in detail.

DeepSeek·DeepSeek-AI·arXiv:2412.19437
2024

Mixtral of Experts

8 experts, 2 active per token: 47B total parameters at the inference cost of about 13B, under Apache 2.0.

Mistral AI·Albert Q. Jiang et al.·arXiv:2401.04088
2024

Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone

Argued that data quality can substitute for scale, using heavily filtered and synthetic 'textbook quality' data.

Microsoft·Marah Abdin et al.·arXiv:2404.14219
2024

StarCoder 2 and The Stack v2: The Next Generation

Open code models trained on a fully licence-audited dataset with an opt-out mechanism for developers.

BigCode·Anton Lozhkov et al.·arXiv:2402.19173
2025

Kimi K2: Open Agentic Intelligence

A 1.04T-parameter MoE with 32B active, pre-trained on 15.5T tokens using MuonClip with no loss spikes — a notable stability result at that scale. Built for agentic tool use rather than chat.

Moonshot AI·Kimi Team·arXiv:2507.20534
2025

Qwen3 Technical Reportessential

Dense and MoE models from 0.6B to 235B under Apache 2.0, unifying thinking and non-thinking modes in one model with a controllable thinking budget.

Alibaba (Qwen)·Qwen Team·arXiv:2505.09388

Reading AI research, honestly answered

Which paper should I read first?

If you read exactly one, read "Attention Is All You Need" (2017). Every model in production today is a Transformer or a Transformer hybrid, and the paper is unusually short and readable. If you read three, add "Training Compute-Optimal Large Language Models" (the Chinchilla paper) for why small models became good, and "Training Language Models to Follow Instructions with Human Feedback" for why they became usable.

How were these papers chosen?

Chosen for causal influence on systems in production today, not for citation count. Each entry says plainly why it mattered and what it changed. That means a few enormously-cited papers are absent and a few modestly-cited ones are here, because the question we asked was "did this change what gets built?" rather than "how many citations does it have?"

Are the citations reliable?

Yes, and this is not a rhetorical claim. Every arXiv identifier on this page was checked against the arXiv API and the returned title matched before publication. Journal papers carry a DOI. Nothing here is cited from memory. 74 of 74 entries carry a resolvable arXiv ID, DOI or publisher URL. Language models hallucinate plausible-looking arXiv numbers that resolve to entirely different papers, which is why every identifier here was machine-checked rather than written from memory.

Do I need a maths background to read these?

Less than you would expect for most of them. Technical reports from open-weight labs — the Llama 3, DeepSeek-V3 and Qwen3 reports — are largely engineering narrative and are readable by anyone comfortable with software architecture. The optimisation and architecture papers do assume linear algebra and calculus. A practical route in: read the abstract, the introduction and the conclusion of any paper first, and only go into the method section if the result matters to you.

Why are Chinese labs so prominent in the recent entries?

Because they are publishing. DeepSeek, Moonshot AI, Alibaba and MiniMax have released detailed technical reports alongside open weights, while the leading US labs have moved towards releasing capability evaluations without architecture or training details — the GPT-4 technical report explicitly withholds parameter count, architecture and data. If you want to read how a frontier model was actually built in 2024 and 2025, the Chinese reports and Meta's Llama 3 paper are where that information is.

How does this connect to the models I can actually use?

Several papers here are the documentation for models in our free-models directory. The DeepSeek-R1, Kimi K2, Qwen3 and Llama 3 entries all correspond to weights you can download and run today, and the model pages link back to the papers. The efficiency section — FlashAttention, PagedAttention, GPTQ, AWQ, QLoRA — is the reason local inference is possible at all.

How this library is maintained

Every arXiv identifier on this page was checked against the arXiv API and the returned title matched before publication. Journal papers carry a DOI. Nothing here is cited from memory.

Chosen for causal influence on systems in production today, not for citation count. Each entry says plainly why it mattered and what it changed.

Last verification run: 2026-09-06. Summaries are ours. Read the paper before you cite it — a summary is a pointer, not a source.

Turning research into something that runs

Most of what is on this page is three years from production and some of it is already in it. Knowing which is which is the job. That is the part we do.

Where this turns into work: AI strategy and consulting, AI agent development and AI automation development. Or see it done: our case studies.

Related