{"id":181235,"date":"2024-01-24T00:22:47","date_gmt":"2024-01-24T06:22:47","guid":{"rendered":"https:\/\/lifeboat.com\/blog\/2024\/01\/the-true-story-of-how-gpt-2-became-maximally-lewd"},"modified":"2024-01-24T00:22:47","modified_gmt":"2024-01-24T06:22:47","slug":"the-true-story-of-how-gpt-2-became-maximally-lewd","status":"publish","type":"post","link":"https:\/\/lifeboat.com\/blog\/2024\/01\/the-true-story-of-how-gpt-2-became-maximally-lewd","title":{"rendered":"The True Story of How GPT-2 Became Maximally Lewd"},"content":{"rendered":"<p><\/p>\n<p><iframe style=\"display: block; margin: 0 auto; width: 100%; aspect-ratio: 4\/3; object-fit: contain;\" src=\"https:\/\/www.youtube.com\/embed\/qV_rOlHjvvs?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; encrypted-media; gyroscope;\n   picture-in-picture\" allowfullscreen><\/iframe><\/p>\n<p>In this video, we recount an incident that occurred at OpenAI while researchers were trying to finetune GPT-2 to be as helpful and ethical as possible. It\u2019s narrated that inadvertently flipping a single minus sign led GPT-2 to become the embodiment of a well-known cardinal sin.<\/p>\n<p>#ai #aisafety #alignment.<\/p>\n<p>\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580SOURCES \\&amp; READINGS\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580<\/p>\n<p>OpenAI blog post: <a href=\"https:\/\/openai.com\/research\/fine-tuni\">https:\/\/openai.com\/research\/fine-tuni<\/a>\u2026<br \/> OpenAI paper behind the blog post: <a href=\"https:\/\/arxiv.org\/pdf\/1909.08593.pdf\">https:\/\/arxiv.org\/pdf\/1909.08593.pdf<\/a>.<br \/> RLHF explainer on Hugging Face: <a href=\"https:\/\/huggingface.co\/blog\/rlhf\">https:\/\/huggingface.co\/blog\/rlhf<\/a>.<br \/> RLHF explainer on aisafety.info <a href=\"https:\/\/aisafety.info\/?state=88FN_904\">https:\/\/aisafety.info\/?state=88FN_904<\/a>\u2026<br \/> Concrete Problems in AI Safety, by <a href=\"https:\/\/twitter.com\/RobertMilesAI\">@RobertMilesAI<\/a>: \u2022 Concrete Problems in AI Safety.<\/p>\n<p>\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580PATREON, MEMBERSHIP, KO-FI\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580<\/p>\n<p>\ud83d\udfe0 Patreon: \/ rationalanimations.<\/p>\n<div class=\"more-link-wrapper\"> <a class=\"more-link\" href=\"https:\/\/lifeboat.com\/blog\/2024\/01\/the-true-story-of-how-gpt-2-became-maximally-lewd\">Continue reading \u201cThe True Story of How GPT-2 Became Maximally Lewd\u201d | &gt;<\/a><\/div>\n","protected":false},"excerpt":{"rendered":"<p>In this video, we recount an incident that occurred at OpenAI while researchers were trying to finetune GPT-2 to be as helpful and ethical as possible. It\u2019s narrated that inadvertently flipping a single minus sign led GPT-2 to become the embodiment of a well-known cardinal sin. #ai #aisafety #alignment. \u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580SOURCES \\&amp; READINGS\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580\u2580 OpenAI blog post: [\u2026]<\/p>\n","protected":false},"author":715,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1635,6],"tags":[],"class_list":["post-181235","post","type-post","status-publish","format-standard","hentry","category-materials","category-robotics-ai"],"_links":{"self":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/posts\/181235","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/users\/715"}],"replies":[{"embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/comments?post=181235"}],"version-history":[{"count":0,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/posts\/181235\/revisions"}],"wp:attachment":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/media?parent=181235"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/categories?post=181235"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/tags?post=181235"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}