CHANNEL ARCHIVE // [899] ARTICLES // PAGE 60 OF 60
Artificial Intelligence
Research into artificial intelligence and intelligent systems.
CHANNEL // 01 [899]
Artificial Intelligence
Research into artificial intelligence and intelligent systems.
Simple probes detect factuality signals inside quantized language models
Source published
- hallucination detection
- large language models
- hidden-state probing
A context-sensitive way to steer language models
Source published
- activation steering
- CLAS
- large language models
AS2 blends visual learning with soft logic; its reported perfect Sudoku score uses a discrete decoder
Source published
- neuro-symbolic AI
- AS2
- differentiable reasoning
A bid to make AI circuits more stable has a proof gap
Source published
- mechanistic interpretability
- Certified Circuits
- circuit discovery
Evolving Instructions Alongside AI Training Improves Benchmark Results
Source published
- Evolutionary System Prompt Learning
- E-SPL
- reinforcement learning
Tracing the Internal Signals Behind Language Model Reasoning
Source published
- large language models
- mechanistic interpretability
- Integrated Policy Gradient
MinPRO tests a prefix-aware approach to steadier language-model training
Source published
- large language models
- reinforcement learning
- off-policy optimization
When a Plausible 3D Body Misses the Shape That Matters
Source published
- SAM 3D Body
- human mesh recovery
- anthropometric fidelity
Can a 3D body scan help predict cesarean delivery?
Source published
- cesarean delivery
- MvBody
- 3D optical body scanning
UpSafe°C lets language models adjust their safety routing
Source published
- UpSafe°C
- large language models
- AI safety
Language models help design rewards for two-stage payment fraud detection
Source published
- payment fraud
- reinforcement learning
- large language models
Can AI Help Researchers Interpret Mental Health Interviews?
Source published
- borderline personality disorder
- large language models
- qualitative research
Can AI redesign toxic molecules? A benchmark finds uneven results
Source published
- ToxiMol
- molecular toxicity repair
- multimodal AI
Using a Few Expert Labels to Make Crowd Annotations More Reliable
Source published
- crowdsourcing
- crowd labeling
- ELICE