AI news

New research: We developed JOLT, a post-training method where one model acts as both teacher and student, with the teacher trained to give…

@perplexity_ai · via X · Oct 21, 2026 · 724 engagements

New research: We developed JOLT, a post-training method where one model acts as both teacher and student, with the teacher trained to give feedback the student can use. On MATH, it matched GRPO's accuracy with 6.5x fewer training tokens.

Read original ↗ Copy link

Post by @perplexity_ai, quoted verbatim. All rights belong to the original publisher. Read the full story at the source.

Related AI news

• Perplexity Computer can now run fully local on your Mac. Work with your native Mac apps and local files, and… · @perplexity_ai
• Gemini 4 Argon sets a new record on WANDR. It is now available in Perplexity for Pro and Max subscribers. · @perplexity_ai
• NVIDIA’s Nemotron 3.5 Super VL is now available in Perplexity for Pro and Max subscribers, and in Agent API. · @perplexity_ai
• We're open sourcing SPACE, the sandbox platform that runs every task in Perplexity Computer. SPACE gives… · @perplexity_ai
• Google's Gemini 4 Carbon Tops AI Intelligence Index · X News
• StreamEval Benchmark Tests AI Video Tools in Real-Time Interactions · X News

All AI news →