{
  "id": 159617,
  "title": "GradCuit: How to Make LLMs Reason Better at Test Time Without Changing a Single Weight",
  "url": "https://urgent.news/2026/08/05/gradcuit-how-to-make-llms-reason-better-at-test-time-without-changing",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-05T04:34:29.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/cofldus/gradcuit-how-to-make-llms-reason-better-at-test-time-without-changing-a-single-weight-2hc3"
  },
  "original_language": "en",
  "account": "GradCuit is a novel method for improving the reasoning capabilities of large language models (LLMs) at test time without modifying their weights. It introduces optimizable latent vectors at an intermediate layer of the Transformer architecture, using causal self-attention to create a differentiable path for reward-weighted gradients to flow directly to these latent variables. This allows for smarter internal reasoning during inference without the need for any parameter updates or token resampling.",
  "summary": "TL;DR GradCuit (arXiv:2608.02585) inserts optimizable latent vectors at an intermediate Transformer layer and uses causal self-attention as a differentiable \"circuit\" to flow reward-weighted gradients directly to those latents at test time — no parameter updates, no token resampling, just smarter internal reasoning. Result: 64.5% average accuracy across 5 models and 3 benchmarks, beating…",
  "key_points": [
    "GradCuit improves LLM reasoning at test time without changing weights.",
    "Introduces optimizable latent vectors in Transformer's intermediate layer.",
    "Causal self-attention enables differentiable gradient flow to latent variables."
  ],
  "editors_take": null,
  "illustration": "https://urgent.news/ill/159617.png",
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}