Hassan Habib<p>Prepping data for AI feels a bit like meal prep…<br>Except your dinner is a .jsonl file and your customer is a llama. 🐪</p><p>I’m learning the ropes of AI training, starting with how to turn raw .txt into structured data for LLMs.</p><p>🎥 How to Feed Your Llama is live:<br><a href="https://www.youtube.com/watch?v=YB9cVyjV9Bo" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">youtube.com/watch?v=YB9cVyjV9B</span><span class="invisible">o</span></a></p><p><a href="https://techhub.social/tags/AI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AI</span></a> <a href="https://techhub.social/tags/LLM" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLM</span></a> <a href="https://techhub.social/tags/JSONL" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>JSONL</span></a> <a href="https://techhub.social/tags/FineTuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>FineTuning</span></a> <a href="https://techhub.social/tags/Python" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Python</span></a> <a href="https://techhub.social/tags/LearningInPublic" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LearningInPublic</span></a> <a href="https://techhub.social/tags/LLaMA" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLaMA</span></a> <a href="https://techhub.social/tags/Mistral" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Mistral</span></a></p>
Recent searches
No recent searches
Search options
Only available when logged in.
eupolicy.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
This Mastodon server is a friendly and respectful discussion space for people working in areas related to EU policy. When you request to create an account, please tell us something about you.
Administered by:
Server stats:
217active users
eupolicy.social: About · Profiles directory · Privacy policy
Mastodon: About · Get the app · Keyboard shortcuts · View source code · v4.3.8
#finetuning
0 posts · 0 participants · 0 posts today
adingbatponder<p>AI question. Say I <a href="https://fosstodon.org/tags/finetune" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>finetune</span></a> <a href="https://fosstodon.org/tags/llama2" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>llama2</span></a> by <a href="https://fosstodon.org/tags/finetuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>finetuning</span></a> on a text task: </p><p>deciding if a new text contains new information over the original</p><p>and </p><p>label the training sets with: new-information (A) , and no-new-information (not-A)</p><p>how does the <a href="https://fosstodon.org/tags/llm" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>llm</span></a> reliably interpret the question: </p><p>Is there new information in this text as compared with the original as uploaded?</p><p>How does a llm make the link to the label given in the training? After all: the labels names could have been just A and not-A ?</p>
Dr. Thompson<p>🎯 No code. No problem. Fine-tune billion-parameter AI models this weekend for under $10 using LoRA!<br>Boost accuracy, slash costs, and launch your custom AI in just 72 hours 🧪💻<br>Read how non-coders are becoming AI builders 👉<br><a href="https://medium.com/@rogt.x1997/the-weekend-project-how-to-fine-tune-ai-models-with-lora-even-if-you-cant-code-575b01099c21" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">medium.com/@rogt.x1997/the-wee</span><span class="invisible">kend-project-how-to-fine-tune-ai-models-with-lora-even-if-you-cant-code-575b01099c21</span></a><br><a href="https://mastodon.social/tags/LoRA" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LoRA</span></a> <a href="https://mastodon.social/tags/NoCodeAI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NoCodeAI</span></a> <a href="https://mastodon.social/tags/FineTuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>FineTuning</span></a> <a href="https://mastodon.social/tags/WeekendProject" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>WeekendProject</span></a> <a href="https://mastodon.social/tags/GenerativeAI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>GenerativeAI</span></a><br><a href="https://medium.com/@rogt.x1997/the-weekend-project-how-to-fine-tune-ai-models-with-lora-even-if-you-cant-code-575b01099c21" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">medium.com/@rogt.x1997/the-wee</span><span class="invisible">kend-project-how-to-fine-tune-ai-models-with-lora-even-if-you-cant-code-575b01099c21</span></a></p>
Erik Jonker<p>Waarom GPT-NL niet zonder mensen kan: ‘de monsterklus van finetuning’<br><a href="https://gpt-nl.nl/nieuws/finetuning/" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="">gpt-nl.nl/nieuws/finetuning/</span><span class="invisible"></span></a><br><a href="https://mastodon.social/tags/AI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AI</span></a> <a href="https://mastodon.social/tags/LLM" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLM</span></a> <a href="https://mastodon.social/tags/GPTNL" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>GPTNL</span></a> <a href="https://mastodon.social/tags/finetuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>finetuning</span></a></p>
Nebraska.Code<p>Adam Barney is 'Demystifying LLMs: How They Work and Why It Matters' July 24th at Nebraska.Code().</p><p><a href="https://nebraskacode.amegala.com/" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="">nebraskacode.amegala.com/</span><span class="invisible"></span></a></p><p><a href="https://mastodon.social/tags/Travefy" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Travefy</span></a> <a href="https://mastodon.social/tags/LargeLanguageModels" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LargeLanguageModels</span></a> <a href="https://mastodon.social/tags/LLM" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLM</span></a> <a href="https://mastodon.social/tags/languagemodels" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>languagemodels</span></a> <a href="https://mastodon.social/tags/TechTalk" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>TechTalk</span></a> <a href="https://mastodon.social/tags/tokenization" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>tokenization</span></a> <a href="https://mastodon.social/tags/embeddings" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>embeddings</span></a> <a href="https://mastodon.social/tags/attentionmechanisms" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>attentionmechanisms</span></a> <a href="https://mastodon.social/tags/finetuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>finetuning</span></a> <a href="https://mastodon.social/tags/engineering" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>engineering</span></a> <a href="https://mastodon.social/tags/softwareengineering" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>softwareengineering</span></a> <a href="https://mastodon.social/tags/softwareengineer" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>softwareengineer</span></a> <a href="https://mastodon.social/tags/Nebraska" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Nebraska</span></a> <a href="https://mastodon.social/tags/TechnologyConference" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>TechnologyConference</span></a> <a href="https://mastodon.social/tags/lincolnnebraska" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>lincolnnebraska</span></a> <a href="https://mastodon.social/tags/devconference" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>devconference</span></a></p>
Tino Eberl<p><a href="https://mastodon.online/tags/KIMissbrauch" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>KIMissbrauch</span></a><br>Wenn <a href="https://mastodon.online/tags/Sprachmodelle" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Sprachmodelle</span></a> plötzlich menschenverachtende Aussagen treffen oder gefährliche Tipps geben, läuft etwas gewaltig schief.</p><p>Eine aktuelle Untersuchung zeigt, wie schnell <a href="https://mastodon.online/tags/Feintuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Feintuning</span></a> unerwartete Folgen haben kann – mit brisanten Konsequenzen für reale Anwendungen.</p><p>Das Phänomen wurde nur durch Zufall entdeckt.</p><p> <a href="https://mastodon.online/tags/KI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>KI</span></a> <a href="https://mastodon.online/tags/Sprachmodelle" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Sprachmodelle</span></a> <a href="https://mastodon.online/tags/Finetuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Finetuning</span></a> <a href="https://mastodon.online/tags/AIRisiko" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AIRisiko</span></a> <a href="https://mastodon.online/tags/LLM" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLM</span></a> <a href="https://mastodon.online/tags/AISecurity" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AISecurity</span></a></p><p><a href="https://tino-eberl.de/missbrauch-kuenstlicher-intelligenz/gefaehrliches-finetuning-ki-modelle-koennen-aus-dem-ruder-laufen/" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">tino-eberl.de/missbrauch-kuens</span><span class="invisible">tlicher-intelligenz/gefaehrliches-finetuning-ki-modelle-koennen-aus-dem-ruder-laufen/</span></a></p>
Sarah Lea<p>Should your LLM memorize everything or just look it up? Let’s break it down.</p><p><a href="https://towardsdatascience.com/how-to-make-your-llm-more-accurate-with-rag-fine-tuning/" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">towardsdatascience.com/how-to-</span><span class="invisible">make-your-llm-more-accurate-with-rag-fine-tuning/</span></a></p><p><a href="https://techhub.social/tags/llm" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>llm</span></a> <a href="https://techhub.social/tags/ai" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ai</span></a> <a href="https://techhub.social/tags/rag" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>rag</span></a> <a href="https://techhub.social/tags/finetuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>finetuning</span></a> <a href="https://techhub.social/tags/chatgpt" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>chatgpt</span></a> <a href="https://techhub.social/tags/openai" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>openai</span></a> <a href="https://techhub.social/tags/deepseek" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>deepseek</span></a> <a href="https://techhub.social/tags/deeplearning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>deeplearning</span></a> <a href="https://techhub.social/tags/machinelearning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>machinelearning</span></a> <a href="https://techhub.social/tags/technology" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>technology</span></a></p>
Futurist :vivaldi_red:<p>Le Chat by Mistral AI is now open to use, no login need.<br><a href="https://social.vivaldi.net/tags/mistralai" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>mistralai</span></a> <a href="https://social.vivaldi.net/tags/ArtificialIntelligence" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ArtificialIntelligence</span></a> <a href="https://social.vivaldi.net/tags/LLM" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLM</span></a> <a href="https://social.vivaldi.net/tags/rag" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>rag</span></a> <a href="https://social.vivaldi.net/tags/finetuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>finetuning</span></a> <a href="https://social.vivaldi.net/tags/LeChat" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LeChat</span></a> <a href="https://social.vivaldi.net/tags/EU" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>EU</span></a><br><a href="https://chat.mistral.ai/chat" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="">chat.mistral.ai/chat</span><span class="invisible"></span></a></p>
eicker.news ᳇ tech news<p>»<a href="https://eicker.news/tags/Nvidia" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Nvidia</span></a>'s <a href="https://eicker.news/tags/ProjectDigits" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ProjectDigits</span></a> is a '<a href="https://eicker.news/tags/personalAI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>personalAI</span></a> <a href="https://eicker.news/tags/supercomputer" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>supercomputer</span></a>' for <a href="https://eicker.news/tags/prototyping" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>prototyping</span></a>, <a href="https://eicker.news/tags/finetuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>finetuning</span></a>, and <a href="https://eicker.news/tags/running" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>running</span></a> <a href="https://eicker.news/tags/AImodels" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AImodels</span></a>; a single unit can run models up to <a href="https://eicker.news/tags/200billionparameters" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>200billionparameters</span></a> in size.« <a href="https://techcrunch.com/2025/01/06/nvidias-project-digits-is-a-personal-ai-computer/?eicker.news" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">techcrunch.com/2025/01/06/nvid</span><span class="invisible">ias-project-digits-is-a-personal-ai-computer/?eicker.news</span></a> <a href="https://eicker.news/tags/tech" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>tech</span></a> <a href="https://eicker.news/tags/media" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>media</span></a></p>
alpharee<p>Recently, I fine-tuned Hugging Face's smollm2 on a <a href="https://mastodon.social/tags/dataset" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>dataset</span></a> of 7.5k 2024 news headlines. Before <a href="https://mastodon.social/tags/finetuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>finetuning</span></a> I had assigned keywords to each <a href="https://mastodon.social/tags/headline" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>headline</span></a> using the same <a href="https://mastodon.social/tags/LLM" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLM</span></a>, with the idea that the model would generate new <a href="https://mastodon.social/tags/headlines" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>headlines</span></a> based on those keywords.</p><p>However, I found that the <a href="https://mastodon.social/tags/model" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>model</span></a> wasn't great at generating headlines from keywords. What it *did* excel at, though, was completing given headline starts in a surprisingly <a href="https://mastodon.social/tags/realistic" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>realistic</span></a> and <a href="https://mastodon.social/tags/hilarious" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>hilarious</span></a> way!</p><p><a href="https://mastodon.social/tags/ai" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ai</span></a> <a href="https://mastodon.social/tags/programming" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>programming</span></a> <a href="https://mastodon.social/tags/tech" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>tech</span></a> <a href="https://mastodon.social/tags/technology" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>technology</span></a></p>
Erik Jonker<p>Good news, a pity that they compared with GPT-3.5 but it will probably also be true for the next generation of models.<br>"Our analysis shows that fine-tuning improves the performance of open-source LLMs, allowing them to match or even surpass zero-shot GPT 3.5 and GPT-4, though still lagging behind fine-tuned GPT<br>3.5. "<br><a href="https://link.springer.com/article/10.1007/s42001-024-00345-9" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">link.springer.com/article/10.1</span><span class="invisible">007/s42001-024-00345-9</span></a><br><a href="https://mastodon.social/tags/opensource" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>opensource</span></a> <a href="https://mastodon.social/tags/LLM" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLM</span></a> <a href="https://mastodon.social/tags/AI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AI</span></a> <a href="https://mastodon.social/tags/finetuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>finetuning</span></a></p>
Toni Aittoniemi<p>Cybertruck, the pro Russia truck!</p><p>PS. <a href="https://mastodon.green/tags/ai" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ai</span></a> screen reading is already actively thwarting political expression.</p><p>Instead of citing the text that’s written in this image word-for-word, the <a href="https://mastodon.green/tags/systemprompts" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>systemprompts</span></a> and <a href="https://mastodon.green/tags/finetuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>finetuning</span></a> for this MLL instead truncate it as: ”political reasons”.</p><p>This is the ”brave new world” we are stepping into. Machine learning parsing the world into what it’s not.</p>
Markus Eisele<p>[GER] Mit InstructLab zu Large Language Models beitragen <a href="https://www.my-it-brain.de/wordpress/mit-instructlab-zu-large-language-models-beitragen/" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">my-it-brain.de/wordpress/mit-i</span><span class="invisible">nstructlab-zu-large-language-models-beitragen/</span></a><br><a href="https://mastodon.online/tags/instructlab" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>instructlab</span></a> @instructlab <a href="https://mastodon.online/tags/aiml" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>aiml</span></a> <a href="https://mastodon.online/tags/llm" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>llm</span></a> <a href="https://mastodon.online/tags/finetuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>finetuning</span></a></p>
Kevin Leibold<p>"Mit dem <a href="https://mastodon.social/tags/Finetuning" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Finetuning</span></a> zum <a href="https://mastodon.social/tags/Entwickler" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Entwickler</span></a> i.S.d Art. 3 (I) <a href="https://mastodon.social/tags/KI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>KI</span></a>-VO werden?"</p><p><a href="https://www.linkedin.com/pulse/finetuning-stellt-eine-eigene-entwicklung-von-gem-art-hansen-ll-m--cfjqc/?trackingId=u9jjcV3zRJ6mHzblOcp37A%3D%3D" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">linkedin.com/pulse/finetuning-</span><span class="invisible">stellt-eine-eigene-entwicklung-von-gem-art-hansen-ll-m--cfjqc/?trackingId=u9jjcV3zRJ6mHzblOcp37A%3D%3D</span></a></p>
ExploreLive feeds
Mastodon is the best way to keep up with what's happening.
Follow anyone across the fediverse and see it all in chronological order. No algorithms, ads, or clickbait in sight.
Create accountLoginDrag & drop to upload