AllLLMRoboticsStartupsResearchEthicsBusinessHardwareAutonomousMedicineCreative

All news

10671
SPERA Trained on 80,000 Hours of EEG From 29,048 People and Tested on 9 Tasks
Medicine·arXiv cs.LG

SPERA Trained on 80,000 Hours of EEG From 29,048 People and Tested on 9 Tasks

Oct 9, 2026·2 min
80 Synthetic Dialogues Are Enough to Train Activation Probes for Gemma-3-27B, but Not Everywhere
Research

80 Synthetic Dialogues Are Enough to Train Activation Probes for Gemma-3-27B, but Not Everywhere

Oct 9, 2026·2 min
Temporal Transformer for the CAN Bus Catches Anomalies Without Sharing Raw Data
Research

Temporal Transformer for the CAN Bus Catches Anomalies Without Sharing Raw Data

Oct 9, 2026·2 min
Freezing the Decoder in KV-Cache Compression Saves 3x Memory but Doesn't Improve Accuracy
Research

Freezing the Decoder in KV-Cache Compression Saves 3x Memory but Doesn't Improve Accuracy

Oct 9, 2026·2 min
SFT-as-Context Restores General Skills in Fine-Tuned LLMs, Within 2.2 Points
Research·arXiv cs.CL

SFT-as-Context Restores General Skills in Fine-Tuned LLMs, Within 2.2 Points

A training-free method has a base model use its SFT version's answer as context. Across 19 model pairs, general-skill losses average no more than 2.2 percentage…

Oct 9, 2026·3 min
Lapras replaces CoT with latent reasoning: up to +10.79% F1 with 23.9x fewer tokens
Research·arXiv cs.CL

Lapras replaces CoT with latent reasoning: up to +10.79% F1 with 23.9x fewer tokens

Lapras, a new post-training method, teaches time series models to reason in latent space: up to +10.79% F1 over CoT and 23.9x fewer generated tokens.

Oct 9, 2026·2 min
Doctors Don't Ask LLMs About Diagnoses: 127 833 Queries vs. 58 Benchmarks
Research·arXiv cs.CL

Doctors Don't Ask LLMs About Diagnoses: 127 833 Queries vs. 58 Benchmarks

Analysis of 127 833 clinician queries to an AI assistant: diagnosis is only 3.7%, and a typical benchmark matches the real workload by just 31%.

Oct 9, 2026·2 min
FedAlphaEdit Merges Knowledge Edits from Different Organizations into One LLM Without Sharing Data
Research·arXiv cs.CL

FedAlphaEdit Merges Knowledge Edits from Different Organizations into One LLM Without Sharing Data

Researchers show that naively combining AlphaEdit and CollabEdit breaks down. Their FedAlphaEdit merges edits in a shared null space and comes close to centralized…

Oct 9, 2026·2 min
Top themes
01Three LLMs Follow Prompts for Stutters and Interruptions Poorly; Rules Do BetterOct 9, 2026·2 min02Sociolinguistics Instead of LLM Judges: Testing User-Simulator Personas on 5 AgentsOct 9, 2026·2 min03AI4Fire: SQL Tool Lifts LLM Accuracy on Wildfire Databases From 16% to 88%Oct 9, 2026·2 min04LLMs Judge Translation Across 70 Language Pairs: They Beat Humans Only in PlacesOct 9, 2026·2 min05STI-OPD: A Probabilistic Teacher Intervention for Multi-Step Agentic TasksOct 9, 2026·2 min
MASA Reframes LLM Sparse Attention as Matrix Approximation
Research·arXiv cs.CL

MASA Reframes LLM Sparse Attention as Matrix Approximation

Oct 9, 2026·2 min
Routed Sparse Autoencoders Separate Words From Intonation in Speech on WavLM and SPEAR
Research·arXiv cs.CL

Routed Sparse Autoencoders Separate Words From Intonation in Speech on WavLM and SPEAR

Oct 9, 2026·2 min
Galahad-kv stores KV cache on NVMe and loads a block 2.8–4.3x faster than recomputing it
Research·arXiv cs.CL

Galahad-kv stores KV cache on NVMe and loads a block 2.8–4.3x faster than recomputing it

Oct 9, 2026·2 min
Fine-Tuned 0.8B-Parameter Model Beats GPT-5.4 at Grammar Annotation
Business·arXiv cs.CL

Fine-Tuned 0.8B-Parameter Model Beats GPT-5.4 at Grammar Annotation

Oct 9, 2026·2 min
Utterance Normalization Cuts the Cost of Analyzing Millions of Calls: A Pipeline Built Entirely on Small Models
Business·arXiv cs.CL

Utterance Normalization Cuts the Cost of Analyzing Millions of Calls: A Pipeline Built Entirely on Small Models

An arXiv paper turns dialogues into short, attributed statements with tags. A shared preparation layer lets teams run call analytics on small models.

Oct 9, 2026·2 min
Text autoencoder compresses web text to 2.24 bits per byte without losing meaning
Research·arXiv cs.CL

Text autoencoder compresses web text to 2.24 bits per byte without losing meaning

A new arXiv paper trains a neural network to compress text into a discrete latent: 2.24 bits per byte on web data, on par with lossless algorithms, with good…

Oct 9, 2026·2 min
Cognitive Thermometers: Neural Networks Proposed as a Way to Measure the Complexity of Meaning
Research·arXiv cs.CL

Cognitive Thermometers: Neural Networks Proposed as a Way to Measure the Complexity of Meaning

A survey on arXiv proposes using machine learning models as "cognitive thermometers" to measure which meanings languages choose most often, and why.

Oct 9, 2026·2 min
LLMs Draft Road-Work Plans for WisDOT: 14B Barely Beats 7B
Business·arXiv cs.CL

LLMs Draft Road-Work Plans for WisDOT: 14B Barely Beats 7B

Oct 9, 2026·2 min
Wieszcz-XIX: A 3.1-Billion-Word Polish Corpus From Before 1918 Trains Models From Scratch
Research·arXiv cs.CL

Wieszcz-XIX: A 3.1-Billion-Word Polish Corpus From Before 1918 Trains Models From Scratch

Oct 9, 2026·2 min
Diffu-LoRA Lets the Adapter Decide Which Stable Diffusion Layers Need Rank
Research·arXiv cs.CL

Diffu-LoRA Lets the Adapter Decide Which Stable Diffusion Layers Need Rank

Oct 9, 2026·2 min
GenUI-Harness Lifts a 4B Model From 9.33% to 58% Pass@3, Beating Claude Opus 5
LLM·arXiv cs.AI

GenUI-Harness Lifts a 4B Model From 9.33% to 58% Pass@3, Beating Claude Opus 5

Oct 9, 2026·2 min
OpenProblemBench: GPT-6-Astra Solves 14% of Open Math and Physics Problems
Research·arXiv cs.AI

OpenProblemBench: GPT-6-Astra Solves 14% of Open Math and Physics Problems

Oct 9, 2026·2 min
AgentHorizon: Best Judge GPT-5.5 Scores Only 80.9% on Long Agent Tasks
Research

AgentHorizon: Best Judge GPT-5.5 Scores Only 80.9% on Long Agent Tasks

Oct 9, 2026·2 min
Two Distillation Schemes Make AI Students Reveal Their Teachers' Hidden Behavior
Research

Two Distillation Schemes Make AI Students Reveal Their Teachers' Hidden Behavior

Oct 9, 2026·2 min
Researchers prove that bloating AGENTS.md is harmful: the file has a size limit
LLM

Researchers prove that bloating AGENTS.md is harmful: the file has a size limit

Oct 9, 2026·2 min
GUISE Shows Narrative Wrappers Break Qwen3-1.7B in 89–96% of Cases
Research·arXiv cs.AI

GUISE Shows Narrative Wrappers Break Qwen3-1.7B in 89–96% of Cases

Researchers built the GUISE benchmark: a role-play wrapper bypasses Qwen3-1.7B's safeguards 89.4% of the time in English and 95.7% in Classical Chinese. They also…

Oct 9, 2026·2 min
An LLM Teaches a Planner to Get by With Polynomial Memory and Solves 1709 of 1890 Tasks
LLM·arXiv cs.AI

An LLM Teaches a Planner to Get by With Polynomial Memory and Solves 1709 of 1890 Tasks

Researchers replaced storage of millions of states with a learned policy with registers. Search runs in polynomial memory and solves 1709 of 1890 IPC 2023 test…

Oct 9, 2026·2 min