Difference between revisions of "AI research trends"
KevinYager (talk | contribs) (→Context Remaking) |
KevinYager (talk | contribs) (→Reviews) |
||
| (15 intermediate revisions by the same user not shown) | |||
| Line 3: | Line 3: | ||
=Memory= | =Memory= | ||
| + | ==Reviews== | ||
| + | * 2024-04: [https://arxiv.org/abs/2404.13501 A Survey on the Memory Mechanism of Large Language Model based Agents] | ||
| + | * 2026-01: [https://arxiv.org/abs/2601.09113 The AI Hippocampus: How Far are We From Human Memory?] | ||
| + | |||
| + | ==Big Ideas== | ||
| + | * 2026-02: [https://arxiv.org/abs/2602.07755 Learning to Continually Learn via Meta-learning Agentic Memory Designs] | ||
==LLM Weights Memory== | ==LLM Weights Memory== | ||
| Line 8: | Line 14: | ||
* 2025-10: [https://arxiv.org/abs/2510.15103 Continual Learning via Sparse Memory Finetuning] | * 2025-10: [https://arxiv.org/abs/2510.15103 Continual Learning via Sparse Memory Finetuning] | ||
* 2026-01: [https://developer.nvidia.com/blog/reimagining-llm-memory-using-context-as-training-data-unlocks-models-that-learn-at-test-time/ Reimagining LLM Memory: Using Context as Training Data Unlocks Models That Learn at Test-Time] (Nvidia) | * 2026-01: [https://developer.nvidia.com/blog/reimagining-llm-memory-using-context-as-training-data-unlocks-models-that-learn-at-test-time/ Reimagining LLM Memory: Using Context as Training Data Unlocks Models That Learn at Test-Time] (Nvidia) | ||
| + | * 2026-01: [https://arxiv.org/abs/2601.02151 Entropy-Adaptive Fine-Tuning: Resolving Confident Conflicts to Mitigate Forgetting] | ||
==Context Length== | ==Context Length== | ||
| Line 44: | Line 51: | ||
==Context Remaking== | ==Context Remaking== | ||
| + | * 2021-01: [https://arxiv.org/abs/2101.00436 Baleen: Robust Multi-Hop Reasoning at Scale via Condensed Retrieval] | ||
| + | * 2025-08: [https://blog.plasticlabs.ai/blog/Memory-as-Reasoning Memory as Reasoning (Memory is Prediction)] | ||
* 2025-09: [https://arxiv.org/abs/2509.25140 ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory] | * 2025-09: [https://arxiv.org/abs/2509.25140 ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory] | ||
* 2025-10: [https://arxiv.org/abs/2510.04618 Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models] | * 2025-10: [https://arxiv.org/abs/2510.04618 Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models] | ||
* 2025-12: [https://arxiv.org/abs/2512.24601 Recursive Language Models] (model searches/queries the full context) | * 2025-12: [https://arxiv.org/abs/2512.24601 Recursive Language Models] (model searches/queries the full context) | ||
* 2026-01: [https://arxiv.org/abs/2601.02553 SimpleMem: Efficient Lifelong Memory for LLM Agents] | * 2026-01: [https://arxiv.org/abs/2601.02553 SimpleMem: Efficient Lifelong Memory for LLM Agents] | ||
| + | * 2026-01: [https://arxiv.org/abs/2601.07190 Active Context Compression: Autonomous Memory Management in LLM Agents] | ||
==Retrieval beyond RAG== | ==Retrieval beyond RAG== | ||
| Line 55: | Line 65: | ||
* 2024-12: [https://arxiv.org/abs/2412.11919 RetroLLM: Empowering Large Language Models to Retrieve Fine-grained Evidence within Generation] | * 2024-12: [https://arxiv.org/abs/2412.11919 RetroLLM: Empowering Large Language Models to Retrieve Fine-grained Evidence within Generation] | ||
* 2025-03: Microsoft: [https://www.microsoft.com/en-us/research/blog/introducing-kblam-bringing-plug-and-play-external-knowledge-to-llms/ Introducing KBLaM: Bringing plug-and-play external knowledge to LLMs] | * 2025-03: Microsoft: [https://www.microsoft.com/en-us/research/blog/introducing-kblam-bringing-plug-and-play-external-knowledge-to-llms/ Introducing KBLaM: Bringing plug-and-play external knowledge to LLMs] | ||
| + | * 2025-06: [https://arxiv.org/abs/2506.06266 Cartridges: Lightweight and general-purpose long context representations via self-study] | ||
* 2025-07: [https://arxiv.org/pdf/2507.07957 MIRIX: Multi-Agent Memory System for LLM-Based Agents] ([https://mirix.io/ mirix]) | * 2025-07: [https://arxiv.org/pdf/2507.07957 MIRIX: Multi-Agent Memory System for LLM-Based Agents] ([https://mirix.io/ mirix]) | ||
* 2025-08: [https://arxiv.org/abs/2508.16153 Memento: Fine-tuning LLM Agents without Fine-tuning LLMs] | * 2025-08: [https://arxiv.org/abs/2508.16153 Memento: Fine-tuning LLM Agents without Fine-tuning LLMs] | ||
| Line 60: | Line 71: | ||
==Working Memory== | ==Working Memory== | ||
* 2024-12: [https://www.arxiv.org/abs/2412.18069 Improving Factuality with Explicit Working Memory] | * 2024-12: [https://www.arxiv.org/abs/2412.18069 Improving Factuality with Explicit Working Memory] | ||
| + | * 2026-01: [https://arxiv.org/abs/2601.03192 MemRL: Self-Evolving Agents via Runtime Reinforcement Learning on Episodic Memory] | ||
==Long-Term Memory== | ==Long-Term Memory== | ||
| Line 70: | Line 82: | ||
===Storage and Retrieval=== | ===Storage and Retrieval=== | ||
* 2025-09: [https://arxiv.org/abs/2509.04439 ArcMemo: Abstract Reasoning Composition with Lifelong LLM Memory] | * 2025-09: [https://arxiv.org/abs/2509.04439 ArcMemo: Abstract Reasoning Composition with Lifelong LLM Memory] | ||
| + | * 2026-01: [https://www.arxiv.org/abs/2601.07372 Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models] | ||
===Episodic Memory=== | ===Episodic Memory=== | ||
| Line 76: | Line 89: | ||
==Continual Learning== | ==Continual Learning== | ||
| + | * 2022-02: [https://arxiv.org/abs/2202.00275 Architecture Matters in Continual Learning] | ||
* 2025-10: [https://arxiv.org/abs/2510.15103 Continual Learning via Sparse Memory Finetuning] | * 2025-10: [https://arxiv.org/abs/2510.15103 Continual Learning via Sparse Memory Finetuning] | ||
* 2025-11: [https://research.google/blog/introducing-nested-learning-a-new-ml-paradigm-for-continual-learning/ Introducing Nested Learning: A new ML paradigm for continual learning] | * 2025-11: [https://research.google/blog/introducing-nested-learning-a-new-ml-paradigm-for-continual-learning/ Introducing Nested Learning: A new ML paradigm for continual learning] | ||
| + | * 2026-01: [https://arxiv.org/abs/2601.16175 Learning to Discover at Test Time] | ||
| + | * 2026-01: [https://arxiv.org/abs/2601.19897 Self-Distillation Enables Continual Learning] | ||
| + | * 2026-02: [https://arxiv.org/abs/2602.07755 Learning to Continually Learn via Meta-learning Agentic Memory Designs] | ||
=Updating Weights at Inference-time= | =Updating Weights at Inference-time= | ||
| Line 144: | Line 161: | ||
=Daydreaming, brainstorming, pre-generation= | =Daydreaming, brainstorming, pre-generation= | ||
* Gwern: [https://gwern.net/ai-daydreaming Daydreaming] | * Gwern: [https://gwern.net/ai-daydreaming Daydreaming] | ||
| + | * 2026-02: [https://arxiv.org/abs/2602.01689 What LLMs Think When You Don't Tell Them What to Think About?] | ||
'''Pre-generation''' | '''Pre-generation''' | ||
| Line 149: | Line 167: | ||
* 2025-04: [https://arxiv.org/abs/2504.13171 Sleep-time Compute: Beyond Inference Scaling at Test-time] | * 2025-04: [https://arxiv.org/abs/2504.13171 Sleep-time Compute: Beyond Inference Scaling at Test-time] | ||
* 2025-11: [https://inference.net/blog/project-aella Project OSSAS: Custom LLMs to process 100 Million Research Papers] ([https://huggingface.co/inference-net models], [https://aella.inference.net/embeddings visualization]) | * 2025-11: [https://inference.net/blog/project-aella Project OSSAS: Custom LLMs to process 100 Million Research Papers] ([https://huggingface.co/inference-net models], [https://aella.inference.net/embeddings visualization]) | ||
| − | |||
=Missing Elements= | =Missing Elements= | ||
| Line 174: | Line 191: | ||
** 2025-06: [https://x.com/karpathy/status/1938626382248149433 LLMs as "cognitive cores"] | ** 2025-06: [https://x.com/karpathy/status/1938626382248149433 LLMs as "cognitive cores"] | ||
** 2025-11: [https://x.com/karpathy/status/1990116666194456651?s=20 Software 1.0 easily automates what you can specify. Software 2.0 easily automates what you can verify.] | ** 2025-11: [https://x.com/karpathy/status/1990116666194456651?s=20 Software 1.0 easily automates what you can specify. Software 2.0 easily automates what you can verify.] | ||
| + | ** 2026-01: [https://x.com/karpathy/status/2008664551445963083?s=20 The majority of the ruff ruff is people who look at the current point and people who look at the current slope] | ||
| + | ** 2026-02: [https://x.com/karpathy/status/2019137879310836075 Agentic Engineering] | ||
=See Also= | =See Also= | ||
* [[Increasing AI Intelligence]] | * [[Increasing AI Intelligence]] | ||
Latest revision as of 10:25, 18 February 2026
Contents
System 2 Reasoning
See: Increasing AI Intelligence
Memory
Reviews
- 2024-04: A Survey on the Memory Mechanism of Large Language Model based Agents
- 2026-01: The AI Hippocampus: How Far are We From Human Memory?
Big Ideas
LLM Weights Memory
- 2024-12: Memory Layers at Scale
- 2025-10: Continual Learning via Sparse Memory Finetuning
- 2026-01: Reimagining LLM Memory: Using Context as Training Data Unlocks Models That Learn at Test-Time (Nvidia)
- 2026-01: Entropy-Adaptive Fine-Tuning: Resolving Confident Conflicts to Mitigate Forgetting
Context Length
- 2020: Various ideas for scaling context window, including Longformer
- 2023-04-02: Discussion of ideas for how to scale context window
- 2023-05-11: Anthropic announces 100k window
- 2023-06-07: magic.dev claims 5M tokens coming soon
- 2023-07-05: Microsoft describes LongNet, with 1 billion token window
- 2023-07-11: Focused Transformer 256k
- 2023-11-06: GPT-4 turbo 128k
- 2023-11-22: Anthropic Claude 2.1 200k
- 2023-12-13: Mamba alternative
- 2024-01-04: LongLM to extend context window
- 2024-02-15: Gemini 1.5 1M tokens
- 2024-03-04: Anthropic Claude 3 200k
- 2024-03-08: Google claims Gemini 1.5 can scale to 10M
- 2024-04-10: Google preprint demonstrates infinite context length by using compressive memory
- 2024-04-12: Meta et al. demonstrate Megalodon that enables infinite context via a more efficient architecture
- 2024-04-14: Google presents TransformerFAM, which leverages a feedback loop so it attends to its own latent representations, acting as working memory and provides effectively infinite context
- 2024-10-31: What is Wrong with Perplexity for Long-context Language Modeling?
- 2025-01-14: MiniMax-01 4M (paper)
- 2025-01-27: Qwen2.5-1M (report)
- 2025-02-14: InfiniteHiP: Extending Language Model Context Up to 3 Million Tokens on a Single GPU 3M
- 2025-02-18: HeadInfer: Memory-Efficient LLM Inference by Head-wise Offloading (code)
- 2025-02-18: Infinite Retrieval: Attention Enhanced LLMs in Long-Context Processing
- 2025-02-19: MoBA: Mixture of Block Attention for Long-Context LLMs
- 2025-02-27: LongRoPE2: Near-Lossless LLM Context Window Scaling (code)
- 2025-03-25: Gemini 2.5 Pro 1M
- 2025-04-05: Meta Llama 4 10M
- 2025-04-14: OpenAI GPT-4.1 1M
- 2025-12-04: Google Titans/Miras 10M
- 2025-12-13: Extending the Context of Pretrained LLMs by Dropping Their Positional Embeddings
Extended Context
Context Remaking
- 2021-01: Baleen: Robust Multi-Hop Reasoning at Scale via Condensed Retrieval
- 2025-08: Memory as Reasoning (Memory is Prediction)
- 2025-09: ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory
- 2025-10: Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models
- 2025-12: Recursive Language Models (model searches/queries the full context)
- 2026-01: SimpleMem: Efficient Lifelong Memory for LLM Agents
- 2026-01: Active Context Compression: Autonomous Memory Management in LLM Agents
Retrieval beyond RAG
See also: AI tools: Retrieval Augmented Generation (RAG)
- 2024-10: Microsoft: KBLaM: Knowledge Base augmented Language Model
- 2024-12: Let your LLM generate a few tokens and you will reduce the need for retrieval
- 2024-12: RetroLLM: Empowering Large Language Models to Retrieve Fine-grained Evidence within Generation
- 2025-03: Microsoft: Introducing KBLaM: Bringing plug-and-play external knowledge to LLMs
- 2025-06: Cartridges: Lightweight and general-purpose long context representations via self-study
- 2025-07: MIRIX: Multi-Agent Memory System for LLM-Based Agents (mirix)
- 2025-08: Memento: Fine-tuning LLM Agents without Fine-tuning LLMs
Working Memory
- 2024-12: Improving Factuality with Explicit Working Memory
- 2026-01: MemRL: Self-Evolving Agents via Runtime Reinforcement Learning on Episodic Memory
Long-Term Memory
- 2025-12: Google Titans + Miras
Storage and Retrieval
- 2025-09: ArcMemo: Abstract Reasoning Composition with Lifelong LLM Memory
- 2026-01: Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models
Episodic Memory
- 2024-03: Larimar: Large Language Models with Episodic Memory Control
- 2025-08: AgentFly: Fine-tuning LLM Agents without Fine-tuning LLMs
Continual Learning
- 2022-02: Architecture Matters in Continual Learning
- 2025-10: Continual Learning via Sparse Memory Finetuning
- 2025-11: Introducing Nested Learning: A new ML paradigm for continual learning
- 2026-01: Learning to Discover at Test Time
- 2026-01: Self-Distillation Enables Continual Learning
- 2026-02: Learning to Continually Learn via Meta-learning Agentic Memory Designs
Updating Weights at Inference-time
- 2025-01: Transformer2: Self-adaptive LLMs
- 2025-08: Beyond Turing: Memory-Amortized Inference as a Foundation for Cognitive Computation
Parameters as Tokens
Internal Thought Representation Space
Visual Thinking
- 2025-01: VideoRAG: Retrieval-Augmented Generation over Video Corpus
- 2025-01: Imagine while Reasoning in Space: Multimodal Visualization-of-Thought
Neural (non-token) Latent Representation
- 2024-11: Microsoft: DroidSpeak: KV Cache Sharing for Cross-LLM Communication and Multi-LLM Serving: LLMs invent their own inter-communication language
- 2024-12: Meta: Training Large Language Models to Reason in a Continuous Latent Space: feeding the latent representation directly back into the model, instead of tokenizing intermediate thoughts (Chain of Continuous Thought, a.k.a. Coconut)
- 2024-12: Meta: Large Concept Models: Language Modeling in a Sentence Representation Space: train a model that operates at a higher level of abstraction than typical word/token LLMs; model operates in a space of concept embeddings (more akin to full sentences than individual words)
- 2024-12: Meta: Byte Latent Transformer: Patches Scale Better Than Tokens: Instead of tokenization, dynamically convert input byte-stream into patches, yielding gains in compute efficiency, with minimal loss in performance
- 2024-12: Compressed Chain of Thought: Efficient Reasoning Through Dense Representations
- 2024-12: Google DeepMind: Deliberation in Latent Space via Differentiable Cache Augmentation
- 2024-12: LANG-JEPA: Learning to Think in Latent Space
- 2025-01: Efficient Reasoning with Hidden Thinking (code)
- 2025-02: Huginn-0125: Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach (code, model)
- 2025-02: Meta: LLM Pretraining with Continuous Concepts (CoCoMix)
- 2025-06: Reasoning by Superposition: A Theoretical Perspective on Chain of Continuous Thought
Altered Transformer
Tokenization
- 2024-04: Better & Faster Large Language Models via Multi-token Prediction
- 2024-12: I Don't Know: Explicit Modeling of Uncertainty with an [IDK] Token
- 2025-04: Meta: Multi-Token Attention
Generation Order
- 2019-02: Non-Monotonic Sequential Text Generation
- 2019-04: Mask-Predict: Parallel Decoding of Conditional Masked Language Models
- 2019-06: Sequence Generation: From Both Sides to the Middle
- 2020-04: Probabilistically Masked Language Model Capable of Autoregressive Generation in Arbitrary Word Order
- 2021-12: Spiral Language Modeling
- 2023-10: FiLM: Fill-in Language Models for Any-Order Generation
- 2024-07: Integrating Randomness in Large Language Models: A Linear Congruential Generator Approach for Generating Clinically Relevant Content
Diffusion Language Models
- 2024-02: Pard: Permutation-Invariant Autoregressive Diffusion for Graph Generation
- 2025-02: Large Language Diffusion Models
- 2025-02: Inception Labs Mercury model (online demo)
- 2025-03: Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Models (project, code, hf)
- 2025-04: Dream 7B: Introducing Dream 7B, the most powerful open diffusion large language model to date
- 2025-04: Scaling Reasoning in Diffusion Large Language Models via Reinforcement Learning (preprint, code)
- 2025-06: Esoteric Language Models (project)
Related: Image Synthesis via Autoregression/Diffusion
- 2023-10: Sequential Data Generation with Groupwise Diffusion Process
- 2024-02: Rolling Diffusion Models
- 2024-08: Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Model
Sampling
- 2024-10: entropix: Entropy Based Sampling and Parallel CoT Decoding
- 2024-10: softmax is not enough (for sharp out-of-distribution)
- 2025-06: Corrector Sampling in Language Models
- 2025-06: Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoning
- 2025-08: Deep Think with Confidence (project)
- 2025-10: Reasoning with Sampling: Your Base Model is Smarter Than You Think
Daydreaming, brainstorming, pre-generation
Pre-generation
- 2025-03: Reasoning to Learn from Latent Thoughts
- 2025-04: Sleep-time Compute: Beyond Inference Scaling at Test-time
- 2025-11: Project OSSAS: Custom LLMs to process 100 Million Research Papers (models, visualization)
Missing Elements
- Memory
- Continuous learning/update
- Robust contextual model
- Long-time-horizon coherence
- Fluid intelligence
- Agency
- Modeling of self
- Daydreaming
Memes
- Andrej Karpathy:
- 2015-05: "Hallucination" in The Unreasonable Effectiveness of Recurrent Neural Networks
- 2017-11: Software 2.0 ("Gradient descent can write code better than you. I'm sorry.")
- 2022-10: Transformers as general-purpose differentiable computers (talk)
- 2023-01: The hottest new programming language is English
- 2023-09: LLM as kernel of a new Operating System (diagram, OS analogies)
- 2024-07: Jagged Intelligence (c.f. Mollick paper)
- 2025-02: Vibe coding
- 2025-06: Software 3.0 (talk): "Prompts as Programs". Software 1.0 is code; 2.0 is model weights; 3.0 is prompts.
- 2025-06: "Context Engineering" instead of "Prompt Engineering"
- 2025-06: LLMs as "cognitive cores"
- 2025-11: Software 1.0 easily automates what you can specify. Software 2.0 easily automates what you can verify.
- 2026-01: The majority of the ruff ruff is people who look at the current point and people who look at the current slope
- 2026-02: Agentic Engineering
