Understand this story
Perplexity unveils post-training method JOLT, letting the same model serve as both teacher and student, with the teacher also participating in training so that feedback is always available. Perplexity's release states that on MATH, JOLT reaches accuracy comparable to GRPO with 6.5 times fewer training tokens. A subsequent MarkTechPost report states that the method was proposed by Perplexity and NVIDIA, named JOLT (Joint On-policy Learning and Teaching); JOLT+ achieves a 72.09% state-test pass rate on AppWorld, 9.9 percentage points higher than GRPO's 62.20%. The report states that testing covered Qwen3-0.6B, Qwen3-4B-Instruct, Qwen3-8B, Qwen3-30B-A3B-Instruct and DeepSeek-V2-Lite, and that the code has not yet been released.
AI-generated from public reports
Latest development · Report
Perplexity and NVIDIA propose JOLT: a joint teacher-student reinforcement learning post-training methodDevelopments
1 confirmed developments
- Source:MarkTechPost
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Oct 22
MarkTechPost
Perplexity and NVIDIA propose JOLT: a joint teacher-student reinforcement learning post-training method
Perplexity and NVIDIA propose JOLT (Joint On-policy Learning and Teaching), which uses the same model to play both teacher and student, with the teacher also participating in training so that feedback is always available. On MATH, JOLT reaches GRPO's accuracy with 6.5 times fewer processed training tokens; JOLT+ achieves a 72.09% state-test pass rate on AppWorld, 9.9 percentage points higher than GRPO's 62.20%. Tests cover Qwen3-0.6B, Qwen3-4B-Instruct, Qwen3-8B, Qwen3-30B-A3B-Instruct, and DeepSeek-V2-Lite; the code has not yet been released.
Perplexity
Perplexity Releases JOLT Post-Training Method
Perplexity unveils new research JOLT: a post-training method that lets the same model act as both teacher and student, with the teacher trained to give feedback usable by the student. On MATH, it reaches accuracy comparable to GRPO with 6.5x fewer training tokens.
Heat trend for this story
- Current
- 7
- Period peak
- 16 Oct 22 15:00
- 24-hour change
- -50%
Move the pointer or select the chart to inspect hourly heat. Use the left and right arrow keys with a keyboard.