Automated Materials Data and Knowledge Extraction using Machine Learning and Large Language Models
First member: a materials-science extraction survey, unrelated to the rest beyond the phrase.
AI Storyline
29 items · 10 sources · 13 days
Operational story trace
Large Language
Follow in this browser to see new updates on your Live feed.
Latest change
The Nov 9 additions are two arXiv papers, a personality-influence benchmark and an LLM-as-AutoML-searcher, so no new event has landed since Mistral Large 4 got day-0 vLLM and Hugging Face support on Oct 29.
Earlier contextThe story so far
This is not one story: the items share the phrase "large language models" and are mostly arXiv papers from Oct 21 onward. Two real threads sit inside it: the Mistral Large 4 launch on Oct 29, and a run of papers on how LLMs and agents handle memory, retrieval, and knowledge updates.
editor-curated · source-linked
Arc
Oct 21Nov 9 · now
BACKGROUND · Oct 21–28Unrelated LLM papers share a phrase, not an event10 sources · show sources ▾
Self-Assessment Reinforcement Learning for Search-State Awareness in Retrieval-Augmented Large Language Models
Search agents trained via RL to know when they have retrieved enough.
Spectral Smoothing and Spiking Perturbations Reconfigure Capability Profiles and Selective Task Performance in Large Language Models
Interpretability work on how spectral edits to feed-forward weights shift task performance.
Parameter-Efficient Adaptation of Large Language Models to a Low-Resource Agglutinative Language: A Corpus, Datasets, and a Reproducible Benchmark for the Bashkir Language
Low-resource language adaptation benchmark for Bashkir.
An Empirical Study of Recent Large Language Models for Knowledge Graph to Text Generation on Standard Benchmarks
Empirical comparison of recent LLMs on knowledge-graph-to-text generation.
LongSearchEval: Evaluating Long-Horizon Web Search Capabilities of Large Language Models via a Trajectory-Aware Agent Framework with Failure-Node Localization
Long-horizon web-search eval that localizes the failing step in an agent trajectory.
CANS: Capability-Preserving Neuron Steering for Large Language Models
Activation-steering method that aims to keep general capability intact.
Reva: Real-Time Vulnerability Detection through Contextual Probing of Large Language Models
Real-time vulnerability screening of agent-generated code via model probing.
CAMR: Compatibility-Aware Memory Replay for Continual Learning in Large Language Models
Continual-learning replay method that checks compatibility of incoming data.
Gold vs.~Translation: Are We Measuring Language, Knowledge, or Locality in Translated Benchmarks?
Questions whether translated benchmarks like MMLU measure the same thing as native ones.
LAUNCH · Oct 22–29Mistral Large 4 ships with day-0 vLLM and Hugging Face support3 sources · show sources ▾
DeepSeek vs Mistral Large 4 vs Cohere: 9x Price Gap [2026]
Search-sourced price comparison putting Mistral Large 4 against DeepSeek and Cohere.
Day 0 support for Mistral Large 4 on vLLM
The launch's serving detail: vLLM day-0 support and the inference features it relies on.
Welcome Le Chonk: Mistral Large 4 on Hugging Face
Hugging Face confirms the weights are published on the Hub.
THEME · Oct 29 – Nov 6Memory, retrieval, and knowledge updates become the densest sub-thread10 sources · show sources ▾
Incentivizing Self-Learning in Large Language Models via Reinforcement Learning
Opens the memory sub-thread: RL to couple knowledge acquisition with reasoning.
FIRME: Registry-Anchored Memory Graphs for Evolving, Heterogeneous Investor Knowledge
Registry-anchored memory graph for persistent, provenance-preserving agent memory.
ImplicitHop: Evaluating knowledge integration across context and memory
Eval of whether models integrate facts arriving in context versus training.
Context as Infrastructure: Knowledge Compilation for Retrieval in Agentic Systems with Harnesses
Argues agentic search harnesses solve accuracy and reframes context as compiled infrastructure.
From Lookups to Lessons: Turning Retrieved Knowledge into Parametric Memory
Distills retrieved knowledge into weights to reduce RAG storage and inference cost.
Article: Systemic Approach to Memory, Knowledge, and Context in Agentic AI Architectures
Practitioner framing: a context engine that selects, compresses, and expires information.
Feeling of Knowing in Large Language Models
Shows post-training can desync what a model knows from what it thinks it knows.
Reasoning over Non-Parametric Memory in Multilingual LLMs: A Cognitive-Based Analysis of Relational Knowledge
Tests whether models reason over relations in context or fall back on memorized ones.
Show HN: Thinker – Trim context and cache knowledge to extend agent usage limits
Show HN tool that trims context and caches knowledge to stretch agent usage limits.
ROSE: Risk-Aware Orthogonal Subspace Navigation for Lifelong Knowledge Editing in Multimodal Large Language Models
Lifelong knowledge editing method for multimodal models.
NOW · Oct 30 – Nov 9Scattered research continues, led by Anthropic's provable-inference update6 sources · show sources ▾
Reimagining Text-to-SQL: A Survey of Large Language Model Architectures and Strategies
Survey of LLM architectures for text-to-SQL.
Prune Gradually, Adapt Sparsely: Efficient Pruning of Large Language Models at Extreme Sparsity
Gradual pruning method targeting quality at extreme sparsity to cut inference cost.
Progress on provable inference for large language models
Anthropic's progress report on zkSNARK proofs for verifiable inference.
FedReason: Large Language Models as Weight-Space Meta-Aggregators for Heterogeneous Federated Learning
Uses an LLM to pick the federated-learning aggregation rule.
More Than Just Style: Benchmarking Personality Influence in Large Language Models
Benchmark of how personality-like tendencies shift model outputs.
Large Language Models as a Universal Searcher for Open-World Automated Machine Learning
Latest member: an LLM as a cross-domain AutoML searcher.
Day 1 Wednesday, Oct 21, 2026
Automated Materials Data and Knowledge Extraction using Machine Learning and Large Language Models
First member: a materials-science extraction survey, unrelated to the rest beyond the phrase.
Self-Assessment Reinforcement Learning for Search-State Awareness in Retrieval-Augmented Large Language Models
Search agents trained via RL to know when they have retrieved enough.
Day 2 Thursday, Oct 22, 2026
DeepSeek vs Mistral Large 4 vs Cohere: 9x Price Gap [2026]
Search-sourced price comparison putting Mistral Large 4 against DeepSeek and Cohere.
Spectral Smoothing and Spiking Perturbations Reconfigure Capability Profiles and Selective Task Performance in Large Language Models
Interpretability work on how spectral edits to feed-forward weights shift task performance.
Parameter-Efficient Adaptation of Large Language Models to a Low-Resource Agglutinative Language: A Corpus, Datasets, and a Reproducible Benchmark for the Bashkir Language
Low-resource language adaptation benchmark for Bashkir.
An Empirical Study of Recent Large Language Models for Knowledge Graph to Text Generation on Standard Benchmarks
Empirical comparison of recent LLMs on knowledge-graph-to-text generation.
Day 3 Monday, Oct 26, 2026
LongSearchEval: Evaluating Long-Horizon Web Search Capabilities of Large Language Models via a Trajectory-Aware Agent Framework with Failure-Node Localization
Long-horizon web-search eval that localizes the failing step in an agent trajectory.
CANS: Capability-Preserving Neuron Steering for Large Language Models
Activation-steering method that aims to keep general capability intact.
Reva: Real-Time Vulnerability Detection through Contextual Probing of Large Language Models
Real-time vulnerability screening of agent-generated code via model probing.
Day 4 Tuesday, Oct 27, 2026
CAMR: Compatibility-Aware Memory Replay for Continual Learning in Large Language Models
Continual-learning replay method that checks compatibility of incoming data.
Day 5 Wednesday, Oct 28, 2026
Gold vs.~Translation: Are We Measuring Language, Knowledge, or Locality in Translated Benchmarks?
Questions whether translated benchmarks like MMLU measure the same thing as native ones.
Day 6 Thursday, Oct 29, 2026
Day 0 support for Mistral Large 4 on vLLM
The launch's serving detail: vLLM day-0 support and the inference features it relies on.
Welcome Le Chonk: Mistral Large 4 on Hugging Face
Hugging Face confirms the weights are published on the Hub.
Incentivizing Self-Learning in Large Language Models via Reinforcement Learning
Opens the memory sub-thread: RL to couple knowledge acquisition with reasoning.
FIRME: Registry-Anchored Memory Graphs for Evolving, Heterogeneous Investor Knowledge
Registry-anchored memory graph for persistent, provenance-preserving agent memory.
Day 7 Friday, Oct 30, 2026
ImplicitHop: Evaluating knowledge integration across context and memory
Eval of whether models integrate facts arriving in context versus training.
Context as Infrastructure: Knowledge Compilation for Retrieval in Agentic Systems with Harnesses
Argues agentic search harnesses solve accuracy and reframes context as compiled infrastructure.
Reimagining Text-to-SQL: A Survey of Large Language Model Architectures and Strategies
Survey of LLM architectures for text-to-SQL.
Prune Gradually, Adapt Sparsely: Efficient Pruning of Large Language Models at Extreme Sparsity
Gradual pruning method targeting quality at extreme sparsity to cut inference cost.
From Lookups to Lessons: Turning Retrieved Knowledge into Parametric Memory
Distills retrieved knowledge into weights to reduce RAG storage and inference cost.
Day 8 Monday, Nov 2, 2026
Progress on provable inference for large language models
Anthropic's progress report on zkSNARK proofs for verifiable inference.
Article: Systemic Approach to Memory, Knowledge, and Context in Agentic AI Architectures
Practitioner framing: a context engine that selects, compresses, and expires information.
Day 9 Tuesday, Nov 3, 2026
Feeling of Knowing in Large Language Models
Shows post-training can desync what a model knows from what it thinks it knows.
Day 10 Wednesday, Nov 4, 2026
Reasoning over Non-Parametric Memory in Multilingual LLMs: A Cognitive-Based Analysis of Relational Knowledge
Tests whether models reason over relations in context or fall back on memorized ones.
Day 11 Thursday, Nov 5, 2026
FedReason: Large Language Models as Weight-Space Meta-Aggregators for Heterogeneous Federated Learning
Uses an LLM to pick the federated-learning aggregation rule.
Show HN: Thinker – Trim context and cache knowledge to extend agent usage limits
Show HN tool that trims context and caches knowledge to stretch agent usage limits.
Day 12 Friday, Nov 6, 2026
ROSE: Risk-Aware Orthogonal Subspace Navigation for Lifelong Knowledge Editing in Multimodal Large Language Models
Lifelong knowledge editing method for multimodal models.
Day 13 Monday, Nov 9, 2026
More Than Just Style: Benchmarking Personality Influence in Large Language Models
Benchmark of how personality-like tendencies shift model outputs.
Large Language Models as a Universal Searcher for Open-World Automated Machine Learning
Latest member: an LLM as a cross-domain AutoML searcher.
What to watch — open questions
- Do independent benchmarks of Mistral Large 4 on vLLM confirm its long-context and throughput claims?
- Does any provider publish Mistral Large 4 API pricing that confirms or overturns the reported price gap with DeepSeek and Cohere?
- Does Anthropic ship provable inference (zkSNARK) as a product feature or customer-verifiable API option?
- Does any agent framework adopt a separate context/memory/knowledge layer like the one InfoQ and the harness paper describe?
How this thread was built
editor wrote the arc · 4 beats
Storylines are threaded mechanically from the feed: stories that share a distinctive anchor across multiple days and sources. Each item links to its original source. The evidence trace, current state, and open questions are written by the editor routine and refreshed whenever a new beat lands.