--- title: Karpathy goes quiet on X, Ben Affleck explains LoRA description: A joke tweet imagining Ben Affleck explaining LoRA fine-tuning (freeze base weights, learning rate 2e-4) after Andrej Karpathy went quiet on X spread through... canonical: https://daily.dev/posts/karpathy-goes-quiet-on-x-ben-affleck-explains-lora-xdmno5pch twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Karpathy goes quiet on X, Ben Affleck explains LoRA | daily.dev og:description: A joke tweet imagining Ben Affleck explaining LoRA fine-tuning (freeze base weights, learning rate 2e-4) after Andrej Karpathy went quiet on X spread through... og:url: https://daily.dev/posts/karpathy-goes-quiet-on-x-ben-affleck-explains-lora-xdmno5pch og:image: https://api.daily.dev/og/posts/xDMNo5Pch.png og:image:alt: Karpathy goes quiet on X, Ben Affleck explains LoRA 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. # Karpathy goes quiet on X, Ben Affleck explains LoRA **[Trends](https://daily.dev/sources/trends)** · 2 min read · 1 upvotes · 0 comments ## Summary A joke tweet imagining Ben Affleck explaining LoRA fine-tuning (freeze base weights, learning rate 2e-4) after Andrej Karpathy went quiet on X spread through AI Twitter, amplified by Hugging Face's Clement Delangue and Rohan Paul. There's no real news about Karpathy's absence; the humor comes from a movie star confidently reciting fine-tuning folk wisdom that ML practitioners repeat from memory. ## Content The joke making the rounds: Andrej Karpathy vanishes from X for a bit, and into the vacuum steps Ben Affleck, apparently an AI engineer now, ready to hold court. @Thom_Wolf set the scene: "Karpathy: disappears from X. Ben Affleck: alright, gather round, so you'll want to freeze the base weights first, learning rate 2e-4." The punchline works because the advice is so plausible. Freeze the base weights, use a learning rate around 2e-4: that's standard LoRA-style fine-tuning guidance, delivered with the confidence of someone who just discovered it. It's a dig at how fast every corner of the internet turns into an AI explainer the moment a respected voice steps away. @Hesamation reacted with the question everyone was thinking: "Ben Affleck? the AI engineer?" The joke spread quickly. @ClementDelangue and @rohanpaul_ai both retweeted Thom Wolf's post, which says something about who finds this funny: the Hugging Face and AI-explainer crowd, the people who actually spend their days freezing base weights. There's a small, affectionate roast buried in it. Karpathy is the benchmark for the patient, clear AI teacher, and when he's offline the feed fills up with confident replacements. Here the replacement is a movie star. It's not far from reality. ## Questions this post answers ### what learning rate is typically used for LoRA fine-tuning A learning rate of 2e-4 is commonly cited as a standard LoRA-style starting point, alongside freezing the base model weights. This combination has become widely repeated folk wisdom among people who run fine-tuning jobs, appearing in tutorials so often that practitioners recite it from memory without checking its original source. _Developers tuning LoRA hyperparameters can follow practical fine-tuning discussions on daily.dev._ ## Community take How the wider developer community reacted, aggregated from 1 discussion and 1 comments across x (as of 2026-10-07). **TL;DR:** There's barely any substantive discussion to go on here; the single reply is an off-topic quip rather than a reaction to the joke itself. **Sentiment:** 40% positive · 40% mixed · 20% skeptical **By community** - x (mixed): The lone reply is an unrelated aside rather than commentary on the LoRA joke, giving little real signal. **Source threads** - [x](https://x.com/Hesamation/status/2107598020221301144) · 1 points · 1 comments ## Similar posts on daily.dev - [Everyone should slow down AI development except for me](https://daily.dev/posts/everyone-should-slow-down-ai-development-except-for-me-b6pwn72fi) · Lobsters · 0 upvotes · 0 comments - [Netflix paid $587M for Ben Affleck’s AI filmmaking startup](https://daily.dev/posts/netflix-paid-587m-for-ben-affleck-s-ai-filmmaking-startup-vgi6jyq1w) · TechCrunch · 11 upvotes · 0 comments --- Tags: [#machine-learning](https://daily.dev/tags/machine-learning), [#deep-learning](https://daily.dev/tags/deep-learning), [#lora](https://daily.dev/tags/lora) [View this post on daily.dev](https://daily.dev/posts/karpathy-goes-quiet-on-x-ben-affleck-explains-lora-xdmno5pch) ```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":"Karpathy goes quiet on X, Ben Affleck explains LoRA","url":"https://daily.dev/posts/karpathy-goes-quiet-on-x-ben-affleck-explains-lora-xdmno5pch","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/karpathy-goes-quiet-on-x-ben-affleck-explains-lora-xdmno5pch"},"datePublished":"2026-10-02T12:46:08.507Z","dateModified":"2026-10-07T07:25:38.229Z","description":"A joke tweet imagining Ben Affleck explaining LoRA fine-tuning (freeze base weights, learning rate 2e-4) after Andrej Karpathy went quiet on X spread through...","isAccessibleForFree":true,"articleSection":"Trends","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":"Trends","logo":"https://media.daily.dev/image/upload/s--ZfSp3asX--/f_auto,q_auto/v1780996004/logos/trends?_a=BAMAMiWQ0","url":"https://daily.dev/sources/trends"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/karpathy-goes-quiet-on-x-ben-affleck-explains-lora-xdmno5pch","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"machine-learning,deep-learning,lora","timeRequired":"PT2M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Trends","item":"https://daily.dev/sources/trends"},{"@type":"ListItem","position":3,"name":"Karpathy goes quiet on X, Ben Affleck explains LoRA"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/karpathy-goes-quiet-on-x-ben-affleck-explains-lora-xdmno5pch#faq","mainEntity":[{"@type":"Question","name":"what learning rate is typically used for LoRA fine-tuning","acceptedAnswer":{"@type":"Answer","text":"A learning rate of 2e-4 is commonly cited as a standard LoRA-style starting point, alongside freezing the base model weights. This combination has become widely repeated folk wisdom among people who run fine-tuning jobs, appearing in tutorials so often that practitioners recite it from memory without checking its original source. Developers tuning LoRA hyperparameters can follow practical fine-tuning discussions on daily.dev."}}]} ```