{"id":35667,"date":"2025-09-26T20:27:23","date_gmt":"2025-09-26T18:27:23","guid":{"rendered":"https:\/\/www.graviton.at\/letterswaplibrary\/help-my-final-year-project-in-finetuning-llms\/"},"modified":"2025-09-26T20:27:23","modified_gmt":"2025-09-26T18:27:23","slug":"help-my-final-year-project-in-finetuning-llms","status":"publish","type":"post","link":"https:\/\/www.graviton.at\/letterswaplibrary\/help-my-final-year-project-in-finetuning-llms\/","title":{"rendered":"Help My Final Year Project In Finetuning Llms"},"content":{"rendered":"<p><!-- SC_OFF --><\/p>\n<div class=\"md\">\n<p>Hey all,<\/p>\n<p>I&#8217;m building my final year project: a tool that generates quizzes and flashcards from educational materials (like PDFs, docs, and videos). Right now, I&#8217;m using an AI-powered system that processes uploaded files and creates question\/answer sets, but I&#8217;m considering taking it a step further by fine-tuning my own language model on domain-specific data.<\/p>\n<p>I&#8217;m seeking advice on a few fronts:<\/p>\n<ul>\n<li>Which small language model would you recommend for a project like this (quiz and flashcard generation)? I&#8217;ve heard about VibeVoice-1.5B, GPT-4o-mini, Haiku, and Gemini Pro\u2014curious about what works well in the community.<\/li>\n<li>What&#8217;s your preferred workflow to train or fine-tune a model for this task? Please share any resources or step-by-step guides that worked for you!<\/li>\n<li>Should I use parameter-efficient fine-tuning (like LoRA\/QLoRA), or go with full model fine-tuning given limited resources?<\/li>\n<li>Do you think this approach (custom fine-tuning for educational QA\/flashcard tasks) will actually produce better results than prompt-based solutions, based on your experience?<\/li>\n<li>If you&#8217;ve tried building similar tools or have strong opinions about data quality, dataset size, or open-source models, I&#8217;d love to hear your thoughts.<\/li>\n<\/ul>\n<p>I&#8217;m eager to hear what models, tools, and strategies people found effective. Any suggestions for open datasets or data generation strategies would also be super helpful.<\/p>\n<p>Thanks in advance for your guidance and ideas! Would love to know if you think this is a realistic approach\u2014or if there&#8217;s a better route I should consider.<\/p>\n<\/div>\n<p><!-- SC_ON -->   submitted by   <a href=\"https:\/\/www.reddit.com\/user\/Ghostgame4\"> \/u\/Ghostgame4 <\/a> <br \/> <span><a href=\"https:\/\/www.reddit.com\/r\/datasets\/comments\/1nr8kht\/help_my_final_year_project_in_finetuning_llms\/\">[link]<\/a><\/span>   <span><a href=\"https:\/\/www.reddit.com\/r\/datasets\/comments\/1nr8kht\/help_my_final_year_project_in_finetuning_llms\/\">[comments]<\/a><\/span><\/p><div class='watch-action'><div class='watch-position align-right'><div class='action-like'><a class='lbg-style1 like-35667 jlk' href='javascript:void(0)' data-task='like' data-post_id='35667' data-nonce='65e0e39b87' rel='nofollow'><img class='wti-pixel' src='https:\/\/www.graviton.at\/letterswaplibrary\/wp-content\/plugins\/wti-like-post\/images\/pixel.gif' title='Like' \/><span class='lc-35667 lc'>0<\/span><\/a><\/div><\/div> <div class='status-35667 status align-right'><\/div><\/div><div class='wti-clear'><\/div>","protected":false},"excerpt":{"rendered":"<p>Hey all, I&#8217;m building my final year project: a tool that generates quizzes and flashcards from educational&#8230;<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[85],"tags":[],"class_list":["post-35667","post","type-post","status-publish","format-standard","hentry","category-datatards","wpcat-85-id"],"_links":{"self":[{"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/posts\/35667","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/comments?post=35667"}],"version-history":[{"count":0,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/posts\/35667\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/media?parent=35667"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/categories?post=35667"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/tags?post=35667"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}