Project at the lablab.ai Hackathon

<p>I've promised to tell you about the project that my team and I worked on during the lablab.ai hackathon.</p>
<p>In short, it's about fine-tuning based on external data (we used Notion as a basis).</p>
<p>The overall scenario is as follows:</p>
<ol>
<li>Connect your knowledge base (Notion, Confluence, etc.)</li>
<li>We use embeddings to find vectors for the information and store them.</li>
<li>The user asks a question (we made it as an integration in Discord and Telegram, but it can also be a search, a field on a website, a voice assistant, or something else).</li>
<li>Profit</li>
</ol>
<p><strong>Learnings:</strong></p>
<ol>
<li>The hackathon is a great opportunity to try out tools that you would normally spend months learning in just a week or two.</li>
<li>The devil is in the details; many ideas seem obvious, but once you start implementing them, you find that 80% of the time is spent on various nuances, wrappers, etc. I think this is the main difference between realized products and businesses and mere ideas.</li>
<li>In the end, “adding knowledge” to the model turned out to be better through embeddings (vectors) rather than through fine-tuning.<br />
a) it's cheaper<br />
b) it's technically more correct. In short, each user query is processed through an embedding, a vector is found for it, and then you search your database for the closest question-answer pair to what the user asked, and you use all of this as part of the prompt to form the final answer.</li>
</ol>
<p>Project card + video: <a href="https://lablab.ai/event/openai-hackathon/data-dreamers/i-am-ai-personalized-chatgpt">https://lablab.ai/event/openai-hackathon/data-dreamers/i-am-ai-personalized-chatgpt</a><br />
Discord for text: <a href="https://discord.gg/qy2MgAXE">https://discord.gg/qy2MgAXE</a></p>;
<p><strong>Thoughts for the future:</strong></p>
<ol>
<li>Experiment with whether embeddings can understand and classify actionable suggestions like “send this link to Vasya,” understanding that action=send email.</li>
<li>One of the sources of knowledge, besides Notion, for us is chat history. I want to experiment with extracting the chat history, creating vectors for each message, and automatically classifying them. For example, identifying “the top 5 topics discussed this week,” even if they were phrased differently.</li>
</ol>
<p>P.S. Message me if you are thinking about applying AI in your project but don't know how to approach it yet.</p>
<p>#ai #hackathon</p>