--- title: What Is Weight-Space Programming? | daily.dev description: A new field called weights based programming lets LLMs directly modify a neural network's weights and biases, instead of relying on loss functions, gradient... canonical: https://daily.dev/posts/what-is-weight-space-programming--doddwcytg twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: What Is Weight-Space Programming? | daily.dev og:description: A new field called weights based programming lets LLMs directly modify a neural network's weights and biases, instead of relying on loss functions, gradient... og:url: https://daily.dev/posts/what-is-weight-space-programming--doddwcytg og:image: https://api.daily.dev/og/posts/dODdwCYTg.png og:image:alt: What Is Weight-Space Programming? og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # What Is Weight-Space Programming? **[NeuralNine](https://daily.dev/sources/neuralnine)** · 1 min read · 1 upvotes · 0 comments ## Summary A new field called weights based programming lets LLMs directly modify a neural network's weights and biases, instead of relying on loss functions, gradient descent, or training loops. The speaker argues LLMs can reason about weights, change them, or create them from scratch. This could make optimizing neural networks cheaper and maybe better, but the more important benefit in the speaker's opinion is explainability. If an LLM can explain how another model reasons and why it behaves a certain way, that would let people understand what happens inside these models. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is weight-space programming in machine learning? Weight-space programming is a field where large language models directly reason about and change a neural network's weights and biases, instead of training it with a loss function, gradient descent, or a training loop. The LLM can modify existing weights or create them from scratch by reasoning about what they represent. _Anyone tracking new alternatives to gradient-based training can follow this emerging approach on daily.dev._ ### Why would using LLMs to edit neural network weights directly matter instead of just training with gradient descent? It could make optimizing neural networks cheaper, but the bigger benefit is explainability: a large language model reasoning about another model's weights could explain how it works and why it behaves certain ways, letting researchers understand what is happening inside the model. _Developers weighing interpretability gains against training costs can keep up with this debate on daily.dev._ ## Similar posts on daily.dev - [OpenAI’s new LLM exposes the secrets of how AI really works](https://daily.dev/posts/openai-s-new-llm-exposes-the-secrets-of-how-ai-really-works-iepupajzg) · MIT Technology Review · 0 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#neural-networks](https://daily.dev/tags/neural-networks), [#gradient-descent](https://daily.dev/tags/gradient-descent) [View this post on daily.dev](https://daily.dev/posts/what-is-weight-space-programming--doddwcytg) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"What Is Weight-Space Programming?","url":"https://daily.dev/posts/what-is-weight-space-programming--doddwcytg","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/what-is-weight-space-programming--doddwcytg"},"datePublished":"2026-11-10T13:09:33.795Z","dateModified":"2026-11-10T13:09:55.751Z","description":"A new field called weights based programming lets LLMs directly modify a neural network's weights and biases, instead of relying on loss functions, gradient...","image":"https://i.ytimg.com/vi/UzLVjQO3ElQ/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/UzLVjQO3ElQ/sddefault.jpg","isAccessibleForFree":true,"articleSection":"NeuralNine","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"NeuralNine","logo":"https://media.daily.dev/image/upload/s--bMYO07OB--/f_auto,q_auto/v1777800757/logos/neuralnine?_a=BAMAMiWQ0","url":"https://daily.dev/sources/neuralnine"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/what-is-weight-space-programming--doddwcytg","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,neural-networks,gradient-descent","timeRequired":"PT1M","video":{"@type":"VideoObject","name":"What Is Weight-Space Programming?","description":"A new field called weights based programming lets LLMs directly modify a neural network's weights and biases, instead of relying on loss functions, gradient...","thumbnailUrl":"https://i.ytimg.com/vi/UzLVjQO3ElQ/sddefault.jpg","uploadDate":"2026-11-10T13:09:33.795Z","duration":"PT1M","url":"https://api.daily.dev/r/dODdwCYTg","embedUrl":"https://www.youtube.com/embed/UzLVjQO3ElQ"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"NeuralNine","item":"https://daily.dev/sources/neuralnine"},{"@type":"ListItem","position":3,"name":"What Is Weight-Space Programming?"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/what-is-weight-space-programming--doddwcytg#faq","mainEntity":[{"@type":"Question","name":"What is weight-space programming in machine learning?","acceptedAnswer":{"@type":"Answer","text":"Weight-space programming is a field where large language models directly reason about and change a neural network's weights and biases, instead of training it with a loss function, gradient descent, or a training loop. The LLM can modify existing weights or create them from scratch by reasoning about what they represent. Anyone tracking new alternatives to gradient-based training can follow this emerging approach on daily.dev."}},{"@type":"Question","name":"Why would using LLMs to edit neural network weights directly matter instead of just training with gradient descent?","acceptedAnswer":{"@type":"Answer","text":"It could make optimizing neural networks cheaper, but the bigger benefit is explainability: a large language model reasoning about another model's weights could explain how it works and why it behaves certain ways, letting researchers understand what is happening inside the model. Developers weighing interpretability gains against training costs can keep up with this debate on daily.dev."}}]} ```