Skip to content
Papers.
Updated 12h ago

Changes

40 recorded changes: new papers, citation and upvote counts each week, rewritten summaries and review decisions. Nothing is overwritten without a row here.

40 of 40 most recent changes

PaperWhat changed
Towards Physics of Multimodal Pretraining: Knowledge Flow, Modality Synergy, Early Unification, and RecipesSep 24, 2026New paperAdded to the list
StudentSim: Training LLM-based Student SimulatorsSep 24, 2026New paperAdded to the list
MolmoAct2: Action Reasoning Models for Real-world DeploymentSep 24, 2026New paperAdded to the list
On the Design of Qwen3.8-Next Architecture: Evaluation, Efficiency, and Training StabilitySep 24, 2026New paperAdded to the list
Strategic Navigation or Stochastic Search? How Agents and Humans Reason Over Document CollectionsSep 24, 2026New paperAdded to the list
RRSI: Regularized Recursive Self-Improvement of Agent HarnessesSep 24, 2026New paperAdded to the list
Agents' Last ExamSep 24, 2026New paperAdded to the list
DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache CompressionSep 24, 2026New paperAdded to the list
Cosmos 3: Omnimodal World Models for Physical AISep 24, 2026New paperAdded to the list
Gemma 4 Technical ReportSep 24, 2026New paperAdded to the list
Mamba: Linear-Time Sequence Modeling with Selective State SpacesSep 24, 2026New paperAdded to the list
Direct Preference Optimization: Your Language Model is Secretly a Reward ModelSep 24, 2026New paperAdded to the list
Efficient Memory Management for Large Language Model Serving with PagedAttentionSep 24, 2026New paperAdded to the list
FlashAttention: Fast and Memory-Efficient Exact Attention with IO-AwarenessSep 24, 2026New paperAdded to the list
DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement LearningSep 24, 2026New paperAdded to the list
SwiftKV: Fast Prefill-Optimized Inference with Knowledge-Preserving Model TransformationSep 24, 2026New paperAdded to the list
Arctic-Embed: Scalable, Efficient, and Accurate Text Embedding ModelsSep 24, 2026New paperAdded to the list
LoRA Learns Less and Forgets LessSep 24, 2026New paperAdded to the list
Nemotron-4 340B Technical ReportSep 24, 2026New paperAdded to the list
Megatron-LM: Training Multi-Billion Parameter Language Models Using Model ParallelismSep 24, 2026New paperAdded to the list
From Local to Global: A Graph RAG Approach to Query-Focused SummarizationSep 24, 2026New paperAdded to the list
Phi-3 Technical Report: A Highly Capable Language Model Locally on Your PhoneSep 24, 2026New paperAdded to the list
LoRA: Low-Rank Adaptation of Large Language ModelsSep 24, 2026New paperAdded to the list
Segment AnythingSep 24, 2026New paperAdded to the list
The Llama 3 Herd of ModelsSep 24, 2026New paperAdded to the list
LLaMA: Open and Efficient Foundation Language ModelsSep 24, 2026New paperAdded to the list
Gemini 1.5: Unlocking multimodal understanding across millions of tokens of contextSep 24, 2026New paperAdded to the list
Chain-of-Thought Prompting Elicits Reasoning in Large Language ModelsSep 24, 2026New paperAdded to the list
Training Compute-Optimal Large Language ModelsSep 24, 2026New paperAdded to the list
Why Language Models HallucinateSep 24, 2026New paperAdded to the list
Constitutional Classifiers: Defending against Universal Jailbreaks across Thousands of Hours of Red TeamingSep 24, 2026New paperAdded to the list
Alignment faking in large language modelsSep 24, 2026New paperAdded to the list
Sleeper Agents: Training Deceptive LLMs that Persist Through Safety TrainingSep 24, 2026New paperAdded to the list
Constitutional AI: Harmlessness from AI FeedbackSep 24, 2026New paperAdded to the list
GPT-4 Technical ReportSep 24, 2026New paperAdded to the list
Training language models to follow instructions with human feedbackSep 24, 2026New paperAdded to the list
Scaling Laws for Neural Language ModelsSep 24, 2026New paperAdded to the list
Language Models are Few-Shot LearnersSep 24, 2026New paperAdded to the list
BERT: Pre-training of Deep Bidirectional Transformers for Language UnderstandingSep 24, 2026New paperAdded to the list
Attention Is All You NeedSep 24, 2026New paperAdded to the list

As JSON: /api/changes?since=

Sources: each lab's own papers and arXiv, with citation and upvote counts from Semantic Scholar and Hugging Face. One-line summaries are for orientation, not a substitute for the paper.

New papers by email

Monday afternoons, only in weeks with new papers from the labs.

Double opt-in. Unsubscribe any time.