Tag: AI language models
Browse our exclusive articles!
DisaBench: Evaluating Disability Harms in AI Language Models
Discover DisaBench, a new framework assessing disability harms in AI language models with inclusive, participatory evaluation methods.
Stop Many-shot Jailbreak Attacks with One Safety Demo
Learn how a single safety demonstration can effectively mitigate many-shot jailbreak attacks, boosting AI model robustness without retraining.
When Value-Aware KV Eviction Boosts Cache Compression
Explore how value-aware KV eviction improves cache compression and task accuracy in long-context language models with a fixed-contract diagnostic.
ReplaySCM: Benchmark for Executable Causal Mechanism Induction
Discover ReplaySCM, a benchmark for evaluating executable causal mechanism induction from interventions in AI and language models.
Evaluating LLM Toxicity Biases: Ensuring Safer AI Models
Explore how biases in LLM toxicity benchmarks affect safety and why robust evaluation frameworks are critical for reliable AI deployment.