A new study shows that reallocating the feed‑forward network capacity toward the early layers of a Transformer—without adding parameters or FLOPs—lowers perplexity by up to 1.84 points, and the same technique improves performance across several modern LLM architectures.
Perplexity charges $20 a month despite free alternatives, generating $2 billion in revenue with a 52‑person team, over 830 k paying users, and a product tier that prioritises citation‑backed, higher‑quality answers for high‑frequency knowledge workers, while navigating competitive pressures and strategic risks.
The article dissects Perplexity’s internal best‑practice guide for building Agent Skills, showing how Skill design flips conventional coding wisdom, introduces a three‑tier context‑cost model, and provides a step‑by‑step workflow, maintenance tips, and real‑world examples.
Apple introduced a two‑step cross‑architecture distillation method that converts costly quadratic‑time Transformers into cheaper linear‑time Mamba models, preserving most of the original performance while dramatically reducing inference cost.
The article analyzes Perplexity’s product logic, arguing that its value lies in the middle‑layer that hides model complexity, but this advantage is fragile because it depends on a permanently fragmented model ecosystem and could disappear if upstream providers integrate workflow capabilities themselves.
The article analyzes the emerging AI browser war, explaining why Perplexity sees building its own native browser Comet as essential for survival, detailing its strategic rationale, workflow‑centric business model, technical architecture, and the risks of relying on rival AI models.
The article explains perplexity as a metric for evaluating large language models, walks through a step‑by‑step probability calculation for a sample sentence, shows how to normalize by sentence length using the geometric mean, and demonstrates that lower perplexity indicates a more accurate and less uncertain model.
A Princeton senior created GPTZero, a free tool that uses perplexity and burstiness metrics to quickly tell whether a passage was written by a human or by ChatGPT, sparking massive user interest and raising ethical questions about AI transparency.
This article proposes a method to identify low-quality text in business data by training a large-scale unsupervised language model to compute sentence perplexity, converting the detection problem into a threshold decision, and details model design, challenges, optimizations, and online performance results.
This article explains the concepts, training pipeline, deployment formats, and practical applications of language models—particularly LSTM‑based models—for evaluating and generating text quality in a real‑world rental listing platform, highlighting data preparation, model training, and online serving techniques.
This article introduces the fundamentals of topic modeling, compares the probabilistic latent semantic analysis (pLSA) and latent Dirichlet allocation (LDA) methods, explains their graphical models and inference via EM or Gibbs sampling, and discusses practical strategies for selecting the optimal number of topics using perplexity or hierarchical Dirichlet processes.
