OpenAI Q&A: Finetuning GPT-3 vs Semantic Search – which to use, when, and why?

4IR with David Shapiro

My Patreon: https://www.patreon.com/daveshap
– Exclusive Discord
– Consultations
– Insider updates
– Support my research, videos, and Open Source work

My Homepage: https://www.daveshap.io/
– All my links
– My books
– Philosophy, etc

<iframe></p> <p><a href="https://www.youtube.com/watch?v=9qq6HTr7Ocw">Source</a></p> <div class="be1e40beae42d993bafb8643f4ddde8b" data-index="3" style="float: none; margin:10px 0 10px 0; text-align:center;"> <script async src="https://pagead2.googlesyndication.com/pagead/js/adsbygoogle.js"></script> <ins class="adsbygoogle" style="display:block; text-align:center;" data-ad-layout="in-article" data-ad-format="fluid" data-ad-client="ca-pub-9244112244416304" data-ad-slot="4549240677"></ins> <script> (adsbygoogle = window.adsbygoogle || []).push({}); </script> </div> <div style="font-size: 0px; height: 0px; line-height: 0px; margin: 0; padding: 0; clear: both;"></div> </div> </article> <div class="clearfix"></div> <ul class="default-theme-post-navigation"> <li class="theme-nav-previous"><a href="https://theengineeringofconsciousexperience.com/cubism-and-fauvism-art-movement/" rel="prev"><span class="meta-nav">←</span> Cubism and Fauvism Art Movement</a></li> <li class="theme-nav-next"><a href="https://theengineeringofconsciousexperience.com/gpt-3-on-google-sheets-google-docs-docsium-com/" rel="next">GPT-3 on Google Sheets & Google Docs (Docsium.com) <span class="meta-nav">→</span></a></li> </ul> <div class="clearfix"></div> <h3 class='comment-reply-title'>Similar Posts</h3> <div class="mb-related-posts mb-simple-featured-posts mb-simple-featured-posts-wrap row"> <article class="mb-featured-article col-md-4 px-lg-3 post"> <a class="post-thumbnail" href="https://theengineeringofconsciousexperience.com/bmws-and-dodge-vipers-art-car-midjourney-generativeart-automobile-ai/" aria-hidden="true" tabindex="-1"> <img width="501" height="300" src="https://theengineeringofconsciousexperience.com/wp-content/uploads/2023/11/1701159885_maxresdefault-501x300.jpg" class="attachment-magazinebook-featured-image-medium size-magazinebook-featured-image-medium wp-post-image" alt="" loading="lazy" /> </a> <span class="cat-links"><a href="https://theengineeringofconsciousexperience.com/category/gpt-3/" rel="category tag">GPT 3</a></span> <header class="entry-header"> <h3 class="entry-title"><a href="https://theengineeringofconsciousexperience.com/bmws-and-dodge-vipers-art-car-midjourney-generativeart-automobile-ai/" rel="bookmark">BMW’s and Dodge Vipers #art #car #midjourney #generativeart #automobile #ai</a></h3> <div class="entry-meta"> <span class="posted-on"><i class="far fa-calendar-alt"></i><a href="https://theengineeringofconsciousexperience.com/bmws-and-dodge-vipers-art-car-midjourney-generativeart-automobile-ai/" rel="bookmark"><time class="entry-date published updated" datetime="2023-04-03T05:18:16-07:00">April 3, 2023</time></a></span><span class="byline"><i class="far fa-user-circle"></i><span class="author vcard"><a class="url fn n" href="https://theengineeringofconsciousexperience.com/author/ai_designs/">AI Designs</a></span></span> </div> </header> </article> <article class="mb-featured-article col-md-4 px-lg-3 post"> <a class="post-thumbnail" href="https://theengineeringofconsciousexperience.com/creating-music-with-gpt-3-will-this-change-the-music-industry/" aria-hidden="true" tabindex="-1"> <img width="501" height="282" src="https://theengineeringofconsciousexperience.com/wp-content/uploads/2020/08/1598705188_maxresdefault.jpg" class="attachment-magazinebook-featured-image-medium size-magazinebook-featured-image-medium wp-post-image" alt="" loading="lazy" srcset="https://theengineeringofconsciousexperience.com/wp-content/uploads/2020/08/1598705188_maxresdefault.jpg 1280w, https://theengineeringofconsciousexperience.com/wp-content/uploads/2020/08/1598705188_maxresdefault-300x169.jpg 300w, https://theengineeringofconsciousexperience.com/wp-content/uploads/2020/08/1598705188_maxresdefault-1024x576.jpg 1024w, https://theengineeringofconsciousexperience.com/wp-content/uploads/2020/08/1598705188_maxresdefault-768x432.jpg 768w, https://theengineeringofconsciousexperience.com/wp-content/uploads/2020/08/1598705188_maxresdefault-520x293.jpg 520w" sizes="(max-width: 501px) 100vw, 501px" /> </a> <span class="cat-links"><a href="https://theengineeringofconsciousexperience.com/category/gpt-3/" rel="category tag">GPT 3</a></span> <header class="entry-header"> <h3 class="entry-title"><a href="https://theengineeringofconsciousexperience.com/creating-music-with-gpt-3-will-this-change-the-music-industry/" rel="bookmark">Creating Music with GPT-3 – Will this change the Music Industry?</a></h3> <div class="entry-meta"> <span class="posted-on"><i class="far fa-calendar-alt"></i><a href="https://theengineeringofconsciousexperience.com/creating-music-with-gpt-3-will-this-change-the-music-industry/" rel="bookmark"><time class="entry-date published updated" datetime="2020-08-14T09:02:48-07:00">August 14, 2020</time></a></span><span class="byline"><i class="far fa-user-circle"></i><span class="author vcard"><a class="url fn n" href="https://theengineeringofconsciousexperience.com/author/df3060ee0c5d0380c196263a36975eb5/">Olle Green</a></span></span> </div> </header> </article> <article class="mb-featured-article col-md-4 px-lg-3 post"> <a class="post-thumbnail" href="https://theengineeringofconsciousexperience.com/hype-20-partes-de-propiedad-privada-contra-0-publicas-gpt-3-08-29-2022-all-rights-reserved-20-2/" aria-hidden="true" tabindex="-1"> <img width="501" height="282" src="https://theengineeringofconsciousexperience.com/wp-content/uploads/2023/03/1678250910_maxresdefault.jpg" class="attachment-magazinebook-featured-image-medium size-magazinebook-featured-image-medium wp-post-image" alt="" loading="lazy" srcset="https://theengineeringofconsciousexperience.com/wp-content/uploads/2023/03/1678250910_maxresdefault.jpg 1280w, https://theengineeringofconsciousexperience.com/wp-content/uploads/2023/03/1678250910_maxresdefault-300x169.jpg 300w, https://theengineeringofconsciousexperience.com/wp-content/uploads/2023/03/1678250910_maxresdefault-1024x576.jpg 1024w, https://theengineeringofconsciousexperience.com/wp-content/uploads/2023/03/1678250910_maxresdefault-768x432.jpg 768w, https://theengineeringofconsciousexperience.com/wp-content/uploads/2023/03/1678250910_maxresdefault-549x309.jpg 549w, https://theengineeringofconsciousexperience.com/wp-content/uploads/2023/03/1678250910_maxresdefault-470x264.jpg 470w, https://theengineeringofconsciousexperience.com/wp-content/uploads/2023/03/1678250910_maxresdefault-233x131.jpg 233w, https://theengineeringofconsciousexperience.com/wp-content/uploads/2023/03/1678250910_maxresdefault-230x129.jpg 230w, https://theengineeringofconsciousexperience.com/wp-content/uploads/2023/03/1678250910_maxresdefault-360x203.jpg 360w, https://theengineeringofconsciousexperience.com/wp-content/uploads/2023/03/1678250910_maxresdefault-100x56.jpg 100w" sizes="(max-width: 501px) 100vw, 501px" /> </a> <span class="cat-links"><a href="https://theengineeringofconsciousexperience.com/category/gpt-3/" rel="category tag">GPT 3</a></span> <header class="entry-header"> <h3 class="entry-title"><a href="https://theengineeringofconsciousexperience.com/hype-20-partes-de-propiedad-privada-contra-0-publicas-gpt-3-08-29-2022-all-rights-reserved-20-2/" rel="bookmark">Hype.® 20 partes de propiedad privada contra 0 publicas. GPT + 3 08 29 2022. All rights reserved 20</a></h3> <div class="entry-meta"> <span class="posted-on"><i class="far fa-calendar-alt"></i><a href="https://theengineeringofconsciousexperience.com/hype-20-partes-de-propiedad-privada-contra-0-publicas-gpt-3-08-29-2022-all-rights-reserved-20-2/" rel="bookmark"><time class="entry-date published updated" datetime="2022-09-30T04:00:40-07:00">September 30, 2022</time></a></span><span class="byline"><i class="far fa-user-circle"></i><span class="author vcard"><a class="url fn n" href="https://theengineeringofconsciousexperience.com/author/a181934a758bb81f4615266f3eec0eb1/">Hype.® Performance</a></span></span> </div> </header> </article> </div> <div id="comments" class="comments-area"> <h5 class="comments-title"> 49 thoughts on “<span>OpenAI Q&A: Finetuning GPT-3 vs Semantic Search – which to use, when, and why?</span>” </h5> <ol class="comment-list"> <li id="comment-310629" class="comment even thread-even depth-1"> <article id="div-comment-310629" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UC_bPHJrE1ZLuVKWUt46IizQ' rel='external nofollow ugc' class='url'>Sorayesh Semo</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310629"><time datetime="2023-01-24T03:16:45-07:00">January 24, 2023 at 3:16 am</time></a> </div> </footer> <div class="comment-content"> <p>your videos are awesome I'm a huge fan, also i had a question have you seen agents and tools in Langchains? can we implement such a thing with embeddings? cause that takes most of the tokens and is expensive</p> </div> </article> </li> <li id="comment-310628" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310628" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCXWlRmU1Oozjku6iORriO1g' rel='external nofollow ugc' class='url'>al</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310628"><time datetime="2023-01-24T06:58:11-07:00">January 24, 2023 at 6:58 am</time></a> </div> </footer> <div class="comment-content"> <p>"Like using a hammer to drive a screw through a board on your knee" 😂😂</p> </div> </article> </li> <li id="comment-310627" class="comment even thread-even depth-1"> <article id="div-comment-310627" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCNBUwIMCiWJViKMIcjrzZXA' rel='external nofollow ugc' class='url'>Hirefied Inc</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310627"><time datetime="2023-01-24T09:31:25-07:00">January 24, 2023 at 9:31 am</time></a> </div> </footer> <div class="comment-content"> <p>You started ads. And I don't believe that I'm saying this but I totally support you on that. 🙂 Your content is top-notch!</p> </div> </article> </li> <li id="comment-310626" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310626" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UC8lGR12E6Hov27mkXHzdLNw' rel='external nofollow ugc' class='url'>Tha Bua</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310626"><time datetime="2023-01-24T15:25:07-07:00">January 24, 2023 at 3:25 pm</time></a> </div> </footer> <div class="comment-content"> <p>David, thank you for sharing your knowledge on yt and on the openai forums. You are a the light to many of us who are curious in this AI world. I don't have any AI or computer backgrounds but I'm able to pick things up slowly and watching your videos has opened many nerual paths in my this realm. Thank you! – Troy</p> </div> </article> </li> <li id="comment-310625" class="comment even thread-even depth-1"> <article id="div-comment-310625" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCjA_o1d0Yprisi7k8dQ2Atw' rel='external nofollow ugc' class='url'>Dan</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310625"><time datetime="2023-02-07T05:53:38-07:00">February 7, 2023 at 5:53 am</time></a> </div> </footer> <div class="comment-content"> <p>I am having so much fun listening to this guy without the technical background that should make this enjoyable. Simple answer: narrative delivery. Content takes a distant place compared to delivery. It's a gift having little to do with actual knowledge.</p> </div> </article> </li> <li id="comment-310624" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310624" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCOiKXCIc__TtOYodWVCnv9Q' rel='external nofollow ugc' class='url'>Vulnerable Growth</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310624"><time datetime="2023-02-07T13:03:35-07:00">February 7, 2023 at 1:03 pm</time></a> </div> </footer> <div class="comment-content"> <p>With respect to a model knowing what it knows and doesn’t know: Anthropic has a paper called “Models (mostly) know what they know” where they test to see if the model can predict whether it internally knows certain information and it seems to do quite well at that.</p> <p>For OpenAI, yeah they are focused on scale right now because they believe most of the capability gains will come from scale and it won’t require as much effort to add the other components once they decide to add other components. That said, they are working on stuff like WebGPT for a reason!</p> </div> </article> </li> <li id="comment-310623" class="comment even thread-even depth-1"> <article id="div-comment-310623" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UC5wyI0rXB5OMvXFggveNNzg' rel='external nofollow ugc' class='url'>David L</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310623"><time datetime="2023-02-09T10:55:27-07:00">February 9, 2023 at 10:55 am</time></a> </div> </footer> <div class="comment-content"> <p>You explain so well. Everything is clear.. thank you for your help</p> </div> </article> </li> <li id="comment-310622" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310622" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCuSc-9ogQ0F3MvDHBwvuk_A' rel='external nofollow ugc' class='url'>ramiro fernandez de ullivarri</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310622"><time datetime="2023-02-12T11:11:00-07:00">February 12, 2023 at 11:11 am</time></a> </div> </footer> <div class="comment-content"> <p>Amazing job David! Im very thankful, you explaind excelacty what I was looking for.</p> </div> </article> </li> <li id="comment-310621" class="comment even thread-even depth-1"> <article id="div-comment-310621" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCVvNiMfUpXuytAQb9pZ5d-A' rel='external nofollow ugc' class='url'>Tait McKenzie</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310621"><time datetime="2023-02-16T12:54:17-07:00">February 16, 2023 at 12:54 pm</time></a> </div> </footer> <div class="comment-content"> <p>I’m so mad he said fusion to power homes is over kill. 😂😂😂😂</p> </div> </article> </li> <li id="comment-310620" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310620" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCIBGXMXflmt7bH_hGVvuREA' rel='external nofollow ugc' class='url'>Pravin Singh</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310620"><time datetime="2023-02-20T04:29:31-07:00">February 20, 2023 at 4:29 am</time></a> </div> </footer> <div class="comment-content"> <p>Thanks for making these videos</p> </div> </article> </li> <li id="comment-310619" class="comment even thread-even depth-1"> <article id="div-comment-310619" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCys1dmV5qVnpf_LgfY7-68A' rel='external nofollow ugc' class='url'>Phi6</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310619"><time datetime="2023-02-23T03:02:06-07:00">February 23, 2023 at 3:02 am</time></a> </div> </footer> <div class="comment-content"> <p>Great video! Thanks</p> </div> </article> </li> <li id="comment-310618" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310618" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCE2FTNzp1H5jwohpvWMXwfg' rel='external nofollow ugc' class='url'>Francesco Sblendorio</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310618"><time datetime="2023-02-24T10:06:19-07:00">February 24, 2023 at 10:06 am</time></a> </div> </footer> <div class="comment-content"> <p>Hi. I have to write a chatbot system for helping users of a certain product. I got a knowledge base made of about 1400 paragraphs describing the product and troubleshooting paragraphs.<br />Which is the best way to:<br />– instruct the system with those instruction, making this knowledge persistent<br />– make people able to make questions about the product and receive answers</p> </div> </article> </li> <li id="comment-310617" class="comment even thread-even depth-1"> <article id="div-comment-310617" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UC85Yr-g3Q6VhDvDCUMJRWjw' rel='external nofollow ugc' class='url'>Ben Sang</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310617"><time datetime="2023-02-27T19:09:34-07:00">February 27, 2023 at 7:09 pm</time></a> </div> </footer> <div class="comment-content"> <p>one of the best videos ive seen on LLM's and fine-tuning. ive seen so many people fine tune get not great results and complain about the cost. so THANK YOU</p> </div> </article> </li> <li id="comment-310616" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310616" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCn1oUd6l5Ej0cS2Riwz25Dw' rel='external nofollow ugc' class='url'>Hennerz</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310616"><time datetime="2023-03-13T05:39:57-07:00">March 13, 2023 at 5:39 am</time></a> </div> </footer> <div class="comment-content"> <p>David why dont you create a private members community where you share the infomation on projects such as your curie scene work</p> </div> </article> </li> <li id="comment-310615" class="comment even thread-even depth-1"> <article id="div-comment-310615" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCQMxfw6SviahwzrwaFjO_fw' rel='external nofollow ugc' class='url'>ADHD</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310615"><time datetime="2023-03-15T16:47:37-07:00">March 15, 2023 at 4:47 pm</time></a> </div> </footer> <div class="comment-content"> <p>Wow just Wow! Saved me a bunch of time. David's like a philosopher; makes you ask why you do something before doing anything. 99% of YT creators do the opposite; throw content at you.</p> </div> </article> </li> <li id="comment-310614" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310614" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCXoWMiE8O9CcNwHkkENJxyQ' rel='external nofollow ugc' class='url'>Asisten AI</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310614"><time datetime="2023-03-16T13:01:13-07:00">March 16, 2023 at 1:01 pm</time></a> </div> </footer> <div class="comment-content"> <p>Thanks, this is precious insight! I was thinking of fine-tuning GPT-3 to do a simple Q&A, but there is another better and cheap way to do that!</p> </div> </article> </li> <li id="comment-310613" class="comment even thread-even depth-1"> <article id="div-comment-310613" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCgdeCPAe7XAQibyWlaNbIPg' rel='external nofollow ugc' class='url'>lorinma</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310613"><time datetime="2023-03-17T07:18:22-07:00">March 17, 2023 at 7:18 am</time></a> </div> </footer> <div class="comment-content"> <p>Amazing video! Just amazing</p> </div> </article> </li> <li id="comment-310612" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310612" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UC-yryv213jOmV6HQMNHKKVA' rel='external nofollow ugc' class='url'>Dron008</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310612"><time datetime="2023-03-19T11:26:08-07:00">March 19, 2023 at 11:26 am</time></a> </div> </footer> <div class="comment-content"> <p>Thank you, I also had this misunderstanding. Need to learn so much yet.</p> </div> </article> </li> <li id="comment-310611" class="comment even thread-even depth-1"> <article id="div-comment-310611" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCIxAgIQSz-w4oiPKDv2scUw' rel='external nofollow ugc' class='url'>Alexxo</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310611"><time datetime="2023-03-20T06:03:56-07:00">March 20, 2023 at 6:03 am</time></a> </div> </footer> <div class="comment-content"> <p>Thanks David for sharing your thoughts, very valuable information.</p> </div> </article> </li> <li id="comment-310610" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310610" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCrzAcjFdXxGHLYl2JurhBJA' rel='external nofollow ugc' class='url'>Mau Baron</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310610"><time datetime="2023-03-20T20:03:22-07:00">March 20, 2023 at 8:03 pm</time></a> </div> </footer> <div class="comment-content"> <p>David, please correct me if I'm wrong, I watched the video that you recommend at the end. It was a great video and using your code and ideas, I applied it to my own use case which was a certain Mexican law case just to try it out. My question is, is this actually scalable? Running the code on my computer and using the newer "text-embedding-ada-002" for embeddings and "gpt-3.5-turbo" for the LLM completion the whole process took around 5 minutes to complete. Is there a way to optimize this in a way to get answers within seconds (thinking of an already deployed model to the market which needs to be fast) . I understand that this knowledge is very valuable to you and that you would not want to give away certain valuable insights, I would really appreciate it if you could only provide resources for further research, I'm very interested in this topic. Thanks a lot man, really.</p> </div> </article> </li> <li id="comment-310609" class="comment even thread-even depth-1"> <article id="div-comment-310609" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UC6kk6Qrs6d0rqScgggoMjSA' rel='external nofollow ugc' class='url'>Al-amin Ibrahim</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310609"><time datetime="2023-03-24T15:34:02-07:00">March 24, 2023 at 3:34 pm</time></a> </div> </footer> <div class="comment-content"> <p>sir what can you say about table question answering</p> </div> </article> </li> <li id="comment-310608" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310608" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UChIIEhMZGwZQQZOVEJltLaw' rel='external nofollow ugc' class='url'>JoeH Plumber</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310608"><time datetime="2023-03-31T20:42:06-07:00">March 31, 2023 at 8:42 pm</time></a> </div> </footer> <div class="comment-content"> <p>Why do you not support what you are explaining with code? Seems sus to me. Other AI developers show by example in code.</p> </div> </article> </li> <li id="comment-310607" class="comment even thread-even depth-1"> <article id="div-comment-310607" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCNKcGgwAq1a2NkyX_9Q5d5g' rel='external nofollow ugc' class='url'>ShoyoM</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310607"><time datetime="2023-04-01T15:27:27-07:00">April 1, 2023 at 3:27 pm</time></a> </div> </footer> <div class="comment-content"> <p>Thank you for making the differences so clear and easy to understand!</p> </div> </article> </li> <li id="comment-310606" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310606" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCNV_dnJ3oFOaVpvyHUJ8l_Q' rel='external nofollow ugc' class='url'>Man From The West</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310606"><time datetime="2023-04-05T00:44:19-07:00">April 5, 2023 at 12:44 am</time></a> </div> </footer> <div class="comment-content"> <p>Thanks for the video David. But i'm still wondering what way would be the one to go if wanted to build a bot that knows all articles on my blog and would recommend me the most fitting one (from my blog only) to a question or keyword i prompt it? I worked with langchain and it worked from time to time, but started to give me articles from other websites the more i asked it.</p> </div> </article> </li> <li id="comment-310605" class="comment even thread-even depth-1"> <article id="div-comment-310605" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UChDd-Lz4b2iNAMlqX6rjKDg' rel='external nofollow ugc' class='url'>Polymath Park </a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310605"><time datetime="2023-04-05T21:26:10-07:00">April 5, 2023 at 9:26 pm</time></a> </div> </footer> <div class="comment-content"> <p>Thanks so much for your efforts. You could charge for these courses tbh, thank you for keeping this open access</p> </div> </article> </li> <li id="comment-310604" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310604" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UC16P-B39y-3C6iQqJjiONyA' rel='external nofollow ugc' class='url'>Ry Medina</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310604"><time datetime="2023-04-06T06:13:19-07:00">April 6, 2023 at 6:13 am</time></a> </div> </footer> <div class="comment-content"> <p>I know a lot of people are trying to use A.I. for RPG's. For a D&D A.I. where it can play the role of the DM and create the campaigns, would it be better to fine-tune a GPT model or get a blank A.I. model and feed it all the source material and all it would have to know is D&D? I'm not sure a GPT 3 model will able to remember early on campaign details that happened at the start, keep track of hit points during combat, know when players would have to make certain checks rolls (/roll).</p> <p>With enough playing around I can get GPT 4 and less effectively with 3.5 to kind of do those things. But my hope would be for the A.I. to be consistent every time a user wants to use the A.I. to play a campaign. Any idea?</p> </div> </article> </li> <li id="comment-310603" class="comment even thread-even depth-1"> <article id="div-comment-310603" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCr-F3cIbqpPTFTIz-6vOGlA' rel='external nofollow ugc' class='url'>Víctor Pinto Tapia</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310603"><time datetime="2023-04-09T19:40:40-07:00">April 9, 2023 at 7:40 pm</time></a> </div> </footer> <div class="comment-content"> <p>Great explanation, greetings from Ecuador.</p> </div> </article> </li> <li id="comment-310602" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310602" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCKbZU3ZLrDkFFk1kc-j64IQ' rel='external nofollow ugc' class='url'>Dalin Hays</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310602"><time datetime="2023-04-18T15:42:58-07:00">April 18, 2023 at 3:42 pm</time></a> </div> </footer> <div class="comment-content"> <p>Hi David – Would I use embeddings in order to connect the GPT API to my 50MB of code? If I don't use embeddings then 50MB of code would be about 20.8 Million Tokens. </p> <p>I am altering a large set of code that is not mine, and I want to find a way to identify functions within the code that are relevant to the features that I want to create. In short I want to find the right hooks within 3,000 files of code to modify for the functions I desire. </p> <p>How might you suggest chunking the code up to convert it to a vector? (I am not a programmer)</p> </div> </article> </li> <li id="comment-310601" class="comment even thread-even depth-1"> <article id="div-comment-310601" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCeZ7dErJZVYMx3yUUOfPOvQ' rel='external nofollow ugc' class='url'>Mad Jashin</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310601"><time datetime="2023-04-22T13:41:15-07:00">April 22, 2023 at 1:41 pm</time></a> </div> </footer> <div class="comment-content"> <p>Hi david, is it a good idea to write the finetuning dataset with a question as the prompt and its answer as the completion? can anyone help me with this?</p> </div> </article> </li> <li id="comment-310600" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310600" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCjn6GHRVEfkRlLGSquQEvOQ' rel='external nofollow ugc' class='url'>Nathan Verni</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310600"><time datetime="2023-04-25T12:00:15-07:00">April 25, 2023 at 12:00 pm</time></a> </div> </footer> <div class="comment-content"> <p>Thank you so much for this, very helpful. To extend your library metaphor, I'm trying to understand what the approach would be for answering a question like "How many times is this Shakespeare quote mentioned in the entire library?".</p> </div> </article> </li> <li id="comment-310599" class="comment even thread-even depth-1"> <article id="div-comment-310599" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UC6VnrjS4lIMF5PWUMBIVJWQ' rel='external nofollow ugc' class='url'>John Długosz</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310599"><time datetime="2023-04-25T16:48:34-07:00">April 25, 2023 at 4:48 pm</time></a> </div> </footer> <div class="comment-content"> <p>I'm used to QA meaning Quality Analysis. Not to be confused with Q&A.</p> </div> </article> </li> <li id="comment-310598" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310598" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCXMUtd33psp1zY1ituTdvcA' rel='external nofollow ugc' class='url'>Harpreet Paul</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310598"><time datetime="2023-04-27T16:17:52-07:00">April 27, 2023 at 4:17 pm</time></a> </div> </footer> <div class="comment-content"> <p>Just a perfect video, so very well explained.</p> </div> </article> </li> <li id="comment-310597" class="comment even thread-even depth-1"> <article id="div-comment-310597" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCN3mOvCqSdx3LLU2A3wvNtQ' rel='external nofollow ugc' class='url'>Roger</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310597"><time datetime="2023-05-02T09:04:44-07:00">May 2, 2023 at 9:04 am</time></a> </div> </footer> <div class="comment-content"> <p>David thank you so much for this knowledge. There is a lot of misinformation about finetuning and you explained it pretty well!</p> <p>My question: With semantic search, your answers are limited to only what is in vector databases. Is there a way to make QA more like chatgpt plus the data in vector database? I guess semantic search makes all the answers to be only within the domain of pdf that it indexes from.</p> <p>(Rewording my question: How do you combine the results of extractive AI to generative AI like chatgpt and return results to users)</p> </div> </article> </li> <li id="comment-310596" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310596" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCsvQfaq3poyPskzJ97Bk8SQ' rel='external nofollow ugc' class='url'>Natty Zaddy</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310596"><time datetime="2023-05-20T16:44:56-07:00">May 20, 2023 at 4:44 pm</time></a> </div> </footer> <div class="comment-content"> <p>So fine tuning cant add to the corpus, but it can add to the tasks it is capable of doing?</p> </div> </article> </li> <li id="comment-310595" class="comment even thread-even depth-1"> <article id="div-comment-310595" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCmVX134gOAhw7G1OQqtgOwQ' rel='external nofollow ugc' class='url'>Creative User</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310595"><time datetime="2023-05-21T03:49:05-07:00">May 21, 2023 at 3:49 am</time></a> </div> </footer> <div class="comment-content"> <p>Can you please do more videos on all the current existing types of vector embedding methods for semantic search and how we can fine tune those (not the whole model).</p> </div> </article> </li> <li id="comment-310594" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310594" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UC-5-a76-INaP_bFl8UbRCgg' rel='external nofollow ugc' class='url'>C Renee Mugo</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310594"><time datetime="2023-05-28T14:04:02-07:00">May 28, 2023 at 2:04 pm</time></a> </div> </footer> <div class="comment-content"> <p>I figured it out the hard way about fine tuning. I have to start all over and take the approach to use embeddings.</p> </div> </article> </li> <li id="comment-310593" class="comment even thread-even depth-1"> <article id="div-comment-310593" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCxHpwJfPgsut9TwY50ZqNhQ' rel='external nofollow ugc' class='url'>Truck Finance Australia</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310593"><time datetime="2023-05-30T00:44:13-07:00">May 30, 2023 at 12:44 am</time></a> </div> </footer> <div class="comment-content"> <p>great vid!</p> </div> </article> </li> <li id="comment-310592" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310592" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCFOgoVK2bOv2lV3SXlR4d9w' rel='external nofollow ugc' class='url'>kingarthur0407</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310592"><time datetime="2023-05-31T19:53:29-07:00">May 31, 2023 at 7:53 pm</time></a> </div> </footer> <div class="comment-content"> <p>Subscribed so frickin hard after watching this, what a stellar video.<br />In my project map, I pitted semantic search against text embeddings, and fine tuning against prompt engineering (I have several script-like prompts you can use even with chatgpt to tune it to different fields and answer styles). Is my understanding not accurate? I thought today's systems like docgpt or privategpt with local document access used text embeddings, and plugins like keysearch ai or seo app on chatgpt used semantic search (on their end). Could I trouble you for any insight on this?</p> </div> </article> </li> <li id="comment-310591" class="comment even thread-even depth-1"> <article id="div-comment-310591" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCaxoWWhjtRvHXooVl_nAc_Q' rel='external nofollow ugc' class='url'>Mr. J</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310591"><time datetime="2023-06-20T14:15:23-07:00">June 20, 2023 at 2:15 pm</time></a> </div> </footer> <div class="comment-content"> <p>Man thanks for this vid, I was very confused about these 2 concepts… I'm very clear now! I hope it's not too late… yet?</p> </div> </article> </li> <li id="comment-310590" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310590" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCS3rqDFh_LvSmvadSMnZkVg' rel='external nofollow ugc' class='url'>Miguel Alba</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310590"><time datetime="2023-06-22T05:08:06-07:00">June 22, 2023 at 5:08 am</time></a> </div> </footer> <div class="comment-content"> <p>The main problem of transfer learning (including inductive transfer learning) is catastrophic forgetting, even fine-tuning a small portion of a network makes the entire thing susceptible to forget stuff, there are ways to mitigate this but most of the research is prohibited in LLMs</p> </div> </article> </li> <li id="comment-310589" class="comment even thread-even depth-1"> <article id="div-comment-310589" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UChTG8mBSFz686k2j2LUWGmA' rel='external nofollow ugc' class='url'>SpaceDusT</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310589"><time datetime="2023-07-03T01:00:02-07:00">July 3, 2023 at 1:00 am</time></a> </div> </footer> <div class="comment-content"> <p>Thank you a lot, this really helps!</p> </div> </article> </li> <li id="comment-310588" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310588" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCXrWrf4RM0cxWuLmr-0ZAmQ' rel='external nofollow ugc' class='url'>Raphael Carvalho</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310588"><time datetime="2023-07-06T03:44:04-07:00">July 6, 2023 at 3:44 am</time></a> </div> </footer> <div class="comment-content"> <p>thank you!</p> </div> </article> </li> <li id="comment-310587" class="comment even thread-even depth-1"> <article id="div-comment-310587" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCfK8196Li1PX4UZjodlSv_g' rel='external nofollow ugc' class='url'>Alexandre Fuchs</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310587"><time datetime="2023-07-13T07:47:46-07:00">July 13, 2023 at 7:47 am</time></a> </div> </footer> <div class="comment-content"> <p>How does your analysis of the genral inapplicabiloty of Fine Tuning change if the subject of the fine tuning is more qualitative than quantitative. Meaning clearly get the point about trying to retrieve/infer on real objectively boolean facts (statutes, regs) but what if you want to train an agent to mimic a personality in its interaction with you? Think an advice giver whre the advice is generally more qualitative (but still need to reflect a particular POV and persoanlity/process) rather than fearing hallucination about decision making face inputs. This is not well achieved with semantic search, retrieval and chaining.</p> </div> </article> </li> <li id="comment-310586" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310586" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCltBzCrgjrZADXBzoZzdA0g' rel='external nofollow ugc' class='url'>Rafael López Muñoz</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310586"><time datetime="2023-07-28T06:26:27-07:00">July 28, 2023 at 6:26 am</time></a> </div> </footer> <div class="comment-content"> <p>🎯 Key Takeaways for quick navigation:</p> <p><a href="https://www.youtube.com/watch?v=9qq6HTr7Ocw&t=00m01s">00:01</a> 🤔 Fine-tuning GPT-3 on a corpus does not enable efficient question-answering. Fine-tuning is for teaching new tasks, not imparting new knowledge.<br /><a href="https://www.youtube.com/watch?v=9qq6HTr7Ocw&t=02m34s">02:34</a> 📚 Semantic search uses semantic embeddings for fast and scalable database searching based on content meaning. It's more suitable for NLU tasks than fine-tuning.<br /><a href="https://www.youtube.com/watch?v=9qq6HTr7Ocw&t=05m02s">05:02</a> 🚫 Fine-tuning is not the same as imbuing an AI with knowledge. It lacks epistemological understanding and cannot distinguish true knowledge from confabulation or hallucination.<br /><a href="https://www.youtube.com/watch?v=9qq6HTr7Ocw&t=10m17s">10:17</a> 💰 Fine-tuning is slow, difficult, and expensive. Semantic search is fast, easy, and cheap, making it a more viable option for many tasks.<br /><a href="https://www.youtube.com/watch?v=9qq6HTr7Ocw&t=11m15s">11:15</a> ✅ Fine-tuning can be used for specific tasks, but it is not optimal for question-answering. Instruct models can perform QA without fine-tuning.<br /><a href="https://www.youtube.com/watch?v=9qq6HTr7Ocw&t=14m13s">14:13</a> 📚 Use a library analogy for QA with semantic search. Index your corpus with semantic embeddings, use a large language model to generate queries, and leverage the llm to read and summarize relevant documents.</p> <p>Made with HARPA AI</p> </div> </article> </li> <li id="comment-310585" class="comment even thread-even depth-1"> <article id="div-comment-310585" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCFrEYlNseyUY1FXKvp2HNgA' rel='external nofollow ugc' class='url'>Tyler Lawler</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310585"><time datetime="2023-08-06T13:09:41-07:00">August 6, 2023 at 1:09 pm</time></a> </div> </footer> <div class="comment-content"> <p>'The news you may have missed'. Holy Innocents School of the Northwest.</p> </div> </article> </li> <li id="comment-310584" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310584" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCgIeOrQPId7vfXThatjcnZw' rel='external nofollow ugc' class='url'>JP SL</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310584"><time datetime="2023-09-26T06:39:17-07:00">September 26, 2023 at 6:39 am</time></a> </div> </footer> <div class="comment-content"> <p>What do you think is the best vector DB right now? Pinecone?</p> </div> </article> </li> <li id="comment-310583" class="comment even thread-even depth-1"> <article id="div-comment-310583" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UC7vJGRqWM1qYACGN-DFmXTQ' rel='external nofollow ugc' class='url'>MK 2109</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310583"><time datetime="2023-09-27T19:41:39-07:00">September 27, 2023 at 7:41 pm</time></a> </div> </footer> <div class="comment-content"> <p>I can now see the difference between someone that knows what they’re talking about or regurgitating YouTube summaries</p> </div> </article> </li> <li id="comment-310582" class="comment odd alt thread-odd thread-alt depth-1"> <article id="div-comment-310582" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCv2qbQJyNJ2Q9qtmZfQCKuQ' rel='external nofollow ugc' class='url'>jermaine brown</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310582"><time datetime="2023-10-04T07:43:38-07:00">October 4, 2023 at 7:43 am</time></a> </div> </footer> <div class="comment-content"> <p>Thankyou for sharing your time and knowledge, the video was flawless</p> </div> </article> </li> <li id="comment-310581" class="comment even thread-even depth-1"> <article id="div-comment-310581" class="comment-body"> <footer class="comment-meta"> <div class="comment-author vcard"> <b class="fn"><a href='https://www.youtube.com/channel/UCIQ3p6xfhfXWEYXrsCtdyzA' rel='external nofollow ugc' class='url'>Pasquale Scaife</a></b> <span class="says">says:</span> </div> <div class="comment-metadata"> <a href="https://theengineeringofconsciousexperience.com/openai-qa-finetuning-gpt-3-vs-semantic-search-which-to-use-when-and-why/#comment-310581"><time datetime="2023-10-09T12:26:45-07:00">October 9, 2023 at 12:26 pm</time></a> </div> </footer> <div class="comment-content"> <p>You're hilarious. Great video – lol "You don't! " got me at the beginning :' D</p> </div> </article> </li> </ol> <p class="no-comments">Comments are closed.</p> </div> </main> </div> <div class="col-md-3 px-lg-3 "> </div> </div> </div> </div> <footer id="colophon" class="site-footer"> <div class="container"> <div class="row"> <div class="col-md-12 text-center"> <div class="site-info"> <span> Powered By: <a href="https://wordpress.org/" target="_blank">WordPress</a> </span> <span class="sep"> | </span> <span> Theme: <a href="https://odiethemes.com/themes/magazinebook/" target="_blank">MagazineBook</a> By OdieThemes </span> </div> </div> </div> </div> </footer> </div> <script>(function(){var advanced_ads_ga_UID="UA-88163215-1",advanced_ads_ga_anonymIP=!!1;window.advanced_ads_check_adblocker=function(t){var n=[],e=null;function a(t){var n=window.requestAnimationFrame||window.mozRequestAnimationFrame||window.webkitRequestAnimationFrame||function(t){return setTimeout(t,16)};n.call(window,t)}return a((function(){var t=document.createElement("div");t.innerHTML=" ",t.setAttribute("class","ad_unit ad-unit text-ad text_ad pub_300x250"),t.setAttribute("style","width: 1px !important; height: 1px !important; position: absolute !important; left: 0px !important; top: 0px !important; overflow: hidden !important;"),document.body.appendChild(t),a((function(){var a,o,i=null===(a=(o=window).getComputedStyle)||void 0===a?void 0:a.call(o,t),d=null==i?void 0:i.getPropertyValue("-moz-binding");e=i&&"none"===i.getPropertyValue("display")||"string"==typeof d&&-1!==d.indexOf("about:");for(var c=0,r=n.length;c<r;c++)n[c](e);n=[]}))})),function(t){"undefined"==typeof advanced_ads_adblocker_test&&(e=!0),null!==e?t(e):n.push(t)}}(),function(){function t(t){this.UID=t,this.analyticsObject="function"==typeof gtag;var n=this;return this.count=function(){gtag("event","AdBlock",{event_category:"Advanced Ads",event_label:"Yes",non_interaction:!0,send_to:n.UID})},function(){if(!n.analyticsObject){var e=document.createElement("script");e.src="https://www.googletagmanager.com/gtag/js?id="+t,e.async=!0,document.body.appendChild(e),window.dataLayer=window.dataLayer||[],window.gtag=function(){dataLayer.push(arguments)},n.analyticsObject=!0,gtag("js",new Date)}var a={send_page_view:!1,transport_type:"beacon"};window.advanced_ads_ga_anonymIP&&(a.anonymize_ip=!0),gtag("config",t,a)}(),this}advanced_ads_check_adblocker((function(n){n&&new t(advanced_ads_ga_UID).count()}))}();})();</script><div style="clear:both;width:100%;text-align:center; font-size:11px; "><a target="_blank" title="WP2Social Auto Publish" href="https://xyzscripts.com/wordpress-plugins/facebook-auto-publish/compare" >WP2Social Auto Publish</a> Powered By : <a target="_blank" title="PHP Scripts & Programs" href="http://www.xyzscripts.com" >XYZScripts.com</a></div><script type='text/javascript' src='https://theengineeringofconsciousexperience.com/wp-content/plugins/bluehost-wordpress-plugin/vendor/newfold-labs/wp-module-ecommerce/includes/wpnavbar.js?ver=1.0' id='nfd_wpnavbar_setting-js'></script> <script type='text/javascript' src='https://theengineeringofconsciousexperience.com/wp-content/themes/magazinebook/js/navigation.js?ver=1.0.9' id='magazinebook-navigation-js'></script> <script type='text/javascript' src='https://theengineeringofconsciousexperience.com/wp-content/themes/magazinebook/js/skip-link-focus-fix.js?ver=1.0.9' id='magazinebook-skip-link-focus-fix-js'></script> <script type='text/javascript' src='https://theengineeringofconsciousexperience.com/wp-content/themes/magazinebook/js/jquery.easy-ticker.js?ver=3.1.0' id='magazinebook-news-ticker-js'></script> <script type='text/javascript' src='https://theengineeringofconsciousexperience.com/wp-content/themes/magazinebook/js/splide.min.js?ver=2.3.1' id='splide-js-js'></script> <script type='text/javascript' src='https://theengineeringofconsciousexperience.com/wp-content/themes/magazinebook/js/theme.js?ver=1.0.9' id='magazinebook-theme-js-js'></script> <script>!function(){window.advanced_ads_ready_queue=window.advanced_ads_ready_queue||[],advanced_ads_ready_queue.push=window.advanced_ads_ready;for(var d=0,a=advanced_ads_ready_queue.length;d<a;d++)advanced_ads_ready(advanced_ads_ready_queue[d])}();</script> </body> </html>