Research-grade reference
Plain-English summaries of papers worth reading.
Each paper review is a granular technical reference document. A high-school student curious about ML can read it end-to-end; a working ML researcher can cite it as a faithful reproduction of the paper. Every article carries a three-depth summary at the close: a 3-line take for the curious reader, a 5-line take for the working developer, a 5-line take for the ML researcher.
56 articles
ai-researchPlain-English walkthrough of ActAdd, Representation Engineering, Contrastive Activation Addition, and Anthropic's Persona Vectors — what each paper proves, where…
ai-researchMulti-paper review of the ARC-AGI benchmark: Chollet's On the Measure of Intelligence (2019), the 2024 Prize Technical Report, o3's 87.5% breakthrough, and ARC-AGI-2.
ai-researchMulti-paper review of late-interaction retrieval (ColBERT, ColBERTv2), bi-encoder baselines (Sentence-BERT), and cross-encoder rerankers (BGE, Cohere) for production RAG.
ai-researchBai et al. 2022 founded RLAIF. Constitutional AI's lineage now spans collective constitutions and constitutional classifiers. A reference for ML teams.
ai-researchMulti-paper review of the diffusion-image-generation lineage — DDPM, Latent Diffusion, SDXL, and Flux. What each paper changed about the loss, the architecture, and…
ai-researchGRPO replaces PPO's value network with group-averaged rewards. Multi-paper review of DeepSeekMath, DeepSeek-R1, and OpenAI RFT — math, algorithms, benchmarks.
ai-researchMulti-paper review of three LLM knowledge-distillation methods: MiniLLM reverse-KL on-policy, Distilling Step-by-Step CoT rationales, Gemma-2 pre-training soft-target.
ai-researchMulti-paper review of Joint Embedding Predictive Architectures: LeCun's framework, I-JEPA (Assran 2023), V-JEPA (Bardes 2024), feature-space prediction.
ai-researchMulti-paper review of four approaches to the linear KV-cache growth problem: H2O heavy-hitters, SnapKV observation window, StreamingLLM attention sinks, Anthropic…
ai-researchMulti-paper review of Meta's Llama 3 herd (arXiv:2407.21783) and Llama 4 release material. Scaling laws, data mixture, post-training pipeline, MoE shift.
ai-researchMulti-paper review of MMLU-Pro, GPQA, SWE-Bench, and SWE-Bench Verified: how each was built, what they measure, where contamination and saturation bite.
ai-researchKirchenbauer green/red lists, Aaronson's Gumbel scheme, Christ-Gunn-Zamir undetectable watermarks, and SynthID-Text. What each method changes and where each one breaks.
ai-researchFive-paper review tracing how LLM agents close long task horizons via browser tools — from WebGPT (2021) through WebAgent, V-IRL, OpenAI Operator and Anthropic…
ai-researchPlain-English walkthrough of induction heads, transformer circuits, and Anthropic's 2024 scaling-monosemanticity work — what's known, what's open.
ai-researchDeepMind's MoD lets transformer tokens skip entire blocks via top-k routing. Technical reference covering the routing math, FLOP savings, and MoE comparison.
ai-researchMulti-paper review of the open vision-language model landscape: LLaVA's projector lineage, PaliGemma's prefix-LM recipe, Qwen2-VL's dynamic resolution and M-RoPE.
ai-researchMulti-paper review of the Microsoft Phi-4 family — the 14B base model, the 3.8B Phi-4-Mini multimodal variant, and the Phi-4-reasoning SFT plus GRPO pipeline.
ai-researchMulti-paper review of DistServe (OSDI 2024) and Splitwise (ISCA 2024). Why separating prefill and decode GPUs cuts tail latency, plus the KV-cache transfer.
ai-researchMulti-paper review of RAFT, Self-RAG, Adaptive-RAG, and Corrective-RAG covering training, reflection tokens, query routing, and retrieval evaluation in modern RAG.
ai-researchMulti-paper review of Ring Attention and Striped Attention — how blockwise distributed self-attention pushes context windows past a million tokens.
ai-researchMulti-paper review of Self-Refine, Reflexion, and Constitutional AI — three ways an LLM uses its own natural-language feedback to improve outputs without weight updates.
ai-researchPlain-English walkthrough of Bricken et al.'s Towards Monosemanticity, Anthropic's Scaling Monosemanticity on Claude 3 Sonnet, and OpenAI's TopK SAEs on GPT-4.
ai-researchMulti-paper review of speculative decoding (Leviathan 2023), Medusa, EAGLE, and EAGLE-2: how each accelerates LLM inference, what speedups they ship, where they break.
ai-researchMulti-paper review of the synthetic-data lineage that produced small but capable LLMs: Self-Instruct, Orca, Phi-1, Phi-3, and the Cosmopedia open replication.
ai-researchMulti-paper review of test-time training (TTT). Sun et al. 2020 + Akyürek et al. 2024: how parameter updates at inference extend test-time compute beyond CoT and RFT.
ai-researchTechnical reference on Tree of Thoughts (Yao et al. 2023). Walks through the BFS and DFS variants, Game of 24 and Creative Writing results, and where the claims break.
ai-researchMulti-paper review of vLLM's PagedAttention, SGLang's RadixAttention, and NVIDIA's TensorRT-LLM — how each system manages the KV cache and where they diverge.
ai-researchMulti-paper review of AlphaGeometry, AlphaProof, and AlphaGeometry2 — how a language model paired with a symbolic engine reaches IMO medallist level on olympiad math.
ai-researchMulti-paper review of Code Llama (2308.12950), StarCoder + StarCoder2 (2305.06161, 2402.19173), and DeepSeek-Coder-V2 (2406.11931). FIM, HumanEval, MoE for code.
ai-researchMulti-paper review of DiLoCo, OpenDiLoCo, and AsyncDiLoCo — how language models can be trained across non-co-located GPUs with infrequent gradient sync.
ai-researchTechnical reference on discrete-diffusion language models. Walks SEDD (Lou 2024) and DiffuLLaMA (Gong 2024): what they solve, where AR baselines still win.
ai-researchMulti-paper review of DPO and three successors (IPO, KTO, SimPO). What each paper changes about the loss, what it gains, and where it still fails.
ai-researchMulti-paper review of Evol-Instruct and the WizardLM/WizardCoder/Tülu open-instruction-tuning lineage — what each paper changes, and where the lineage still leaks.
ai-researchMulti-paper review of IBM's Granite-3.0 language models, Granite Code Models (arXiv:2405.04324), and Granite-TimeSeries TTM (arXiv:2401.03955).
ai-researchPaper review of Lookahead Decoding (Fu, Bailis, Stoica, Zhang) — parallel LLM inference via Jacobi iteration + n-gram pool, no draft model required.
ai-researchLookahead Decoding (Fu et al.) breaks autoregressive sequential dependency using Jacobi iteration. This article reconstructs the method, math, and 1.8x speedup claim.
ai-researchTechnical reference on Mamba-2 (Dao and Gu, ICML 2024). Walks the State Space Duality framework — what holds, where transformers still win.
ai-researchTechnical reference on MoE routing stability. Walks ST-MoE (Zoph 2022), Expert Choice (Zhou 2022), StableMoE (Dai 2022) — what each fixes, where they disagree.
ai-researchTechnical reference on the closed-model reasoning paradigm: OpenAI o1/o3 and Claude Extended Thinking. What is disclosed, what is hidden, what the benchmarks show.
ai-researchMulti-paper review of soft-prompt PEFT — Lester prompt tuning, Li and Liang prefix-tuning, P-Tuning v2 deep prompts — across NLG, NLU and sequence labelling.
ai-researchMulti-paper review of ReAct (Yao 2022), ReWOO (Xu 2023), and Plan-and-Solve (Wang 2023): how the three prompting patterns differ, what each measures, where each breaks.
ai-researchMulti-paper review of Alibaba's Qwen 2.5 (arXiv:2412.15115) and Qwen 3 (arXiv:2505.09388). Pretraining recipe, post-training, multilingual sweep, MoE.
ai-researchKaplan 2020, Chinchilla 2022, and Wei 2022 emergence — power-law exponents, compute-optimal allocation, emergence thresholds, and the mirage critique.
ai-researchMulti-paper review of generative video models — the DiT architecture, OpenAI's Sora, DeepMind's Veo 2/3, and Meta Movie Gen. Architecture, training, data curation…
ai-researchThree software-engineering agent designs read together: SWE-agent's ACI, OpenHands' CodeAct platform, and Anthropic's minimal two-tool harness on SWE-bench Verified.
ai-researchTechnical reference on test-time compute scaling for LLMs. Walks through Snell et al. (2024) and Liu et al. (2025) — what holds, where the claims break.
ai-researchMulti-paper review of tool learning: Toolformer's self-supervised API filtering, Gorilla's retriever-aware fine-tuning, and the Berkeley Function Calling…
ai-researchMulti-paper review of CLIP (Radford 2021), SigLIP / SigLIP 2 (Zhai 2023/2025), and EVA-CLIP-18B (Sun 2024) — losses, batch scaling, and recipes.
ai-researchTechnical reference on vision-language model benchmarks. Walks MMMU (Yue, CVPR 2024) and MathVista (Lu, ICLR 2024) — what each catches, where leaderboards mislead.
ai-researchTechnical walkthrough of three non-attention sequence architectures: xLSTM's exponential gating, RWKV-7's generalized delta rule, and RetNet's three-form retention.
ai-researchAgentFlow (arXiv:2510.05592) trains a planner with on-policy RL across a four-module agent. We summarise what it claims, where the evidence holds, and where it doesn't.
ai-researchWen et al.'s ICLR 2026 paper says RLVR genuinely improves reasoning under a corrected metric. The headline holds for math; the limits matter.
ai-researchClaw-Eval-Live evaluates 13 frontier models on 105 live workflow tasks across business services and workspace repair. Top score 66.7%.
ai-researchDPO replaces the entire RLHF pipeline with a single supervised loss. Technical reference for ML teams choosing between PPO-RLHF, DPO, and 2024 successors.
ai-researchTechnical reference for arXiv:2512.24601: RLM architecture, mathematical contributions, algorithmic pseudocode, ablations, and reusable components.
ai-researchRoundPipe (arXiv:2604.27085): stateless pipeline parallelism with CPU-offloaded master weights. 1.48-2.16x speedup on 1.7B-32B models; 235B feasibility on 8x RTX 4090.
No articles match the selected topic yet. Clear filter .