Loading activity…
@antoine_chaffin
Solve search, solve everything @perplexity_ai
AI
Perplexity says text and vision scores improved, with inference up to 5x faster.
AI
Perplexity says embedding images and rendered pages directly preserves tables, figures and layout that text extraction drops.
AI
Perplexity says its new pplx-decider-v1.1-27b model costs $0.02 per million input tokens, half the price of v1. OpenRouter later said the model was also available there with text, JSON, or image inputs and free output.
AI
The supplied evidence shows one X user linking to a Hugging Face Space called "jev-decision-index" and asking whether it is "the leaderboard everyone is grinding," then another account reposting the same message.
AI
A post describes Lily as a local inference engine for Apple silicon and PII-Tracer as an on-device personal-data classifier.
AI
A September 28 post announcing the change says Jev yielded 73% relevant context versus 46% for embeddings in tests on 28 research tasks, at the same cost per report.
AI
The announcement says late-interaction retrievers store roughly 1,000 vectors per page, but GLIE could use a handful plus a small shared decoder to reconstruct the full set.
AI
Antoine Chaffin posted that Astra found a tricky bug missed in his other debugging sessions.
AI
Tony Wu outlines methods to reduce embedding sizes while preserving retrieval performance on a benchmark.
