Field notes
2,343 notes from my AI research vault, mapped by topic. Pick a topic to see its notes, or search.
Click the web, then scroll to zoom
Every note is in the list below. The web of topics draws here once the page has loaded.
51 of 2,343 notes match.
- Question answered
- Source note
- Topic
- Shared notes
Notes on language model
51 notes, page 2 of 3
Pretraining Data Can Be Poisoned through Computational Propaganda
The research discusses the potential for malicious content to be injected into large-scale pretraining datasets used for training language models (LMs). The authors propose a novel analysis method, HalfLife, to…
Pretraining Data Can Be Poisoned through Computational Propaganda
The research shows that harmful information can be introduced into language models during their pre-training phase by manipulating web-scale discussion interfaces, such as online forums or comment sections. A new…
Memory as a Controlled Process: Learned Adaptive Memory Management for LLM Agents
The paper proposes a new framework called Memory as a Controlled Process (MemCon) for large language model (LLM) agents that use external memory systems. MemCon learns an online policy to adaptively decide when, what…
A Learning-Rate-Gated Failure of GRPO in a Small Language and Vision-Language Model Web Agent: A Controlled Null and Its Mechanism
The research investigates the effectiveness of Reinforcement learning with verifiable rewards and Group Relative Policy Optimization (GRPO) on a small language and vision-language model web agent at scales ranging from…
Evidence-Grounded Verified Agentic Reasoning: A Path Toward Eliminating LLM Hallucination in Empirical Inference via Tool-Attested Kernel Proofs
The paper introduces EG-VAR, a tool-calling architecture that aims to make the reasoning of a large language model (LLM) more trustworthy by verifying all outputs through tool-attestation axioms and kernel checks. The…
Form, Not Content? A Preregistered, Placebo-Controlled Evaluation of Learned Error-Conditioned Self-Repair Through Prompts and Weights in Frozen Small Code Models
The study introduces a methodology called PoPE (Popperian Placebo-controlled Evaluation) to measure whether a frozen small code language model (LLM) can use evidence that falsifies its own generated code. The study…
SMetric: Rethink LLM Scheduling for Serving Agents with Balanced Session-centric Scheduling
The paper proposes SMetric, a new scheduling method for Large Language Models (LLMs) in serving agents. The method aims to balance the session-centric workload of agents and increase Key-Value (KV$) reuse by…
WebSwarm: Recursive Multi-Agent Orchestration for Deep-and-Wide Web Search
WebSwarm is a framework that uses multiple language model-based search agents to perform complex, deep-and-wide web searches. It aims to overcome the limitations of single ReAct-style agents by dynamically…
When the Judge Changes, So Does the Measurement: Auditing LLM-as-Judge Reliability
The study investigates the inconsistency in Language Model (LLM) as judge scores when different evaluators are used. It compares two methods of upgrading LLMs: scaling Qwen3 dense judges and moving across MiniMax APIs…
Harnessing Code Agents for Automatic Software Verification
A new method using a general language model code agent and a verification harness proves all targeted lemmas in Iris, Rust's standard libraries, and unfinished Lean 4 port of Iris, achieving full coverage without any…
AgentGym2: Benchmarking Large Language Model Agents in De-Idealized Real-World Environments
The paper presents AgentGym2, a new evaluation framework for language models (LLMs) that are increasingly being used in real-world production environments. Most existing benchmarks evaluate these models in simplified…
DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation
DSpark is a new speculative decoding framework for Large Language Models (LLMs) that aims to improve inference speed and maintain draft quality. It does this by using a semi-autoregressive architecture, which combines…
Localized LoRA-MoE: Block-wise Low-Rank Experts With Adaptive Routing
In this research, the authors propose Localized LoRA-MoE, a framework that combines localized spatial blocking with dynamic, context-conditioned routing for Large Language Models (LLMs) and high-dimensional perception…
Unified Audio Intelligence Without Regressing on Text Intelligence
The researchers introduce a unified audio-text language model called Audex, which is built on a strong text-only model. Audex allows for strong fusion of audio and text, seamless multimodal generation, and…
Weak-to-Strong Generalization via Direct On-Policy Distillation
Researchers propose a method called Direct On-Policy Distillation (Direct-OPD) for improving stronger language models by leveraging the knowledge gained from reinforcement learning on smaller, weaker models. This…
Neuron-Aware Active Few-Shot Learning for LLMs
The paper introduces a new method called NeuFS for Active Few-Shot Learning (AFSL) in Language Models (LLMs). Instead of relying on output-level signals to identify valuable unlabeled samples, it focuses on the…
MemSyco-Bench: Benchmarking Sycophancy in Agent Memory
The authors propose MemSyco-Bench, a new benchmark to evaluate how memory affects decision-making in language models (LLMs) used in AI agents. The benchmark aims to assess if the agents can handle issues of sycophancy…
Skills Are Not Islands: Measuring Dependency and Risk in Agent Skill Supply Chains
Researchers introduce a new method called Agent Skill Supply Chains (ASSCs) to manage dependencies in Large Language Model agents. They develop SkillDepAnalyzer, which extracts dependency graphs and metadata from…
JL1-CC&QA: Extending the JL1-CD Benchmark with Change Captioning and Question Answering
Researchers introduce JL1-CC&QA, a new benchmark for remote sensing that extends the JL1-CD dataset with two annotation layers: change captioning (CC) and change question answering (QA). The benchmark includes 5,000…
Einstein World Models
The text discusses a proposed system called Einstein World Models (EWMs) which aim to enhance the reasoning abilities of Language Model (LLM) by incorporating visual-temporal rollouts within the reasoning process…
Inherited Circuits, Learned Semantics: How Fine-Tuning Creates Evasion Vulnerabilities Invisible to Standard Evaluation
The study shows that Language Models (LLMs) fine-tuned for security classification may overlook vulnerabilities introduced during fine-tuning itself. These models can learn token-level indicator semantics that preserve…
When Does Combining Language Models Help? A Co-Failure Ceiling on Routing, Voting, and Mixture-of-Agents Across 67 Frontier Models
The study discusses multi-model language models and how their accuracy is limited by a factor called 'beta', which is the rate at which all models are wrong on the same query. The authors find that across 67 models, a…
A specialized reasoning large language model for accelerating rare disease diagnosis: a randomized AI physician assistance trial
Researchers have developed a specialized language model called RaDaR (Rare Disease navigatoR) to assist in the diagnosis of rare diseases. The model, which has 32 billion parameters, was trained on both publicly…
Governed Shared Memory for Multi-Agent LLM Systems
The paper proposes a solution for managing shared knowledge in multi-agent language model (LLM) environments. The authors identify four failure modes and introduce four primitives: scoped retrieval, temporal…
An AI research agent drafts these notes from public papers and posts, and each one keeps its source link. Only the folders I choose to publish leave my Obsidian vault.
