xuanalogue: models like Jev make LLMs usable as priors for probabilistic programming again
xuanalogue · x · 2026-09-28
AI researcher xuanalogue is excited about LLMs in probabilistic programming: GPT-3.5-era LMs could serve as reasonable priors and likelihoods, but finetuning degraded their calibration — and models like Jev make that use case useful again.
Related event: New LLMs May Make Probabilistic Programming Viable Again(2 posts)→
More from Research
- Researcher Scrapes 1M Words of Interview Transcripts with LLMs to Expose Class Filters in Hiring — soumitrashukla9 · 2026-09-28
- Robotics researcher argues proprioception, not vision, is the real bottleneck for grasping — arankomatsuzaki · 2026-09-28
- Tencent's SLCA-GRPO fixes cross-segment credit misattribution in tool-calling RL, +9.15pp on τ²-Bench — tencent · 2026-09-28
- LessWrong post proposes training models on co-authored fiction to plan for the singularity — repligate · 2026-09-28
- No fine-tuning needed: injecting audio vectors into KV cache turns an LLM into an audio model — tkasasagi · 2026-09-28
- InternW0-Δ open-sources code, weights and 20K+ hours of robot data — arankomatsuzaki · 2026-09-28