Kili Technology

🤔 ‘’Is supervised fine-tuning better for domain knowledge like chemistry?’’ Our answer: ✨ Use continued pre-training to embed new factual information ✨ Apply fine-tuning (SFT or RLHF) to optimize how the model applies this knowledge and improve task-specific performance #AI

@Kili_Technology is proud to be part of the elite group of @nvidia Inception Program 🥳🔥 This group has been for startups driving change across industries around the world. We’re so proud to see how well @Kili_Technology has been doing recently. And that's only the beginning!

Webinar Replay 2: Quality People vs Quality Workflow Building a robust RLHF dataset requires both skilled people and efficient workflows, combined with the right tools and feedback mechanisms to ensure continuous quality improvements. Check what Paul Graffan says: 👇🎬

Hey X! We've had our webinar last week and it was amazing 🔥 In reinforcement learning, a key aspect to success is having enough data to improve your model. In some cases, you may not have the right quantity, hence, a dynamic labeling workflow is needed. 👇🎬

@WebCraftByGary @bobambrogi Thanks @WebCraftByGary ! We're really excited to see how attorneys will make their life easier with davinci's speed. 🔥 If you'd like to have a look at the product yourself, you can join the waitlist in one click: ✨getdavinci.ai/waitlist

@bobambrogi @Kili_Technology Really intriguing to see AI like davinci shaping the future of patent law! The promise of better patents in less time could be a game-changer. Wondering how it might integrate with current legal frameworks.

A new gen AI tool for patent lawyers, called davinci, promises better patents in half the time. @Kili_Technology lawnext.com/2024/01/new-ai…

How can we critique and evaluate large language models effectively? 🤔 @Meta provides one solution: use another language #model with a high-quality human annotated dataset. 💡 Interestingly, human annotators used @Kili_Technology 's data labeling platform! 🥰 #llm #ai

@chiefaioffice Thanks for the intro! 🙏 Your next data labeling tool is here 🚀🔥

🚨 Kili (@Kili_Technology) Funding: $31.9M 🎉 Intro 👉 Kili is one tool to review and label fast, find and fix errors, simplify your data operations, and leverage integrated expert labeling services.

France has a thriving AI Startup ecosystem 🇫🇷 300+ startups & $7 billion + in funding The Top 10 + Funding: Dataiku: $847M Hugging Face: $395M Owkin: $304M Poolside: $126M Mistral: $111M Animaj: $106M AQEMIA: $33M Kili: $32M Nabla: $21M Photoroom: $19M What do they do? 🧵

Did you know that a 10% drop in label accuracy can lead to a staggering 2-5% decrease in your model's accuracy? 💰🤯 Investing in data quality is about being cost-effective and ensuring the success of your AI/ML projects.🚀💚 Read more 👇 hubs.li/Q027dmwm0

The whole resource is right here: 👉 hubs.li/Q026SvMl0

What are the cost factors for image annotation in AI projects? 📚💡 🤔 Understand how cost varies with task complexity and gain critical insights to inform your budgeting process. 🔥 Head to the comments now!🤫 #budget #annotation #project

💬 Navigating through the complexities of Large Language Models (LLMs) is crucial for reliable AI. 🔥 In this article we: ✨ Uncover the mystery behind LLM hallucinations ✨ Discover actionable mitigation strategies 😉 Everything is here: 👉 hubs.li/Q026n9kW0

✨ Go check the whole article now 👉 hubs.li/Q025P3Np0

📖 Delve into the first part of our webinar recap on building conversational engines with Large Language Models (LLMs). 🔥 Explore real-world applications of LLMs, learn evaluation techniques of LLMs, and watch a practical demo of evaluating LLMs on @Kili_Technology platform.

🚀 Attended Day 1 of the fantastic World Summit AI 2023 in Amsterdam. 🔥 Excited for Day 2! Don’t forget to pass by Booth S3! 😉

🚀We’re happy to announce a revamped Smart Tracking! You can now annotate the target in the first frame it appears in, and then let our AI-powered tool do the magic either when you want to start annotating a new target or need to adjust portions.🔥 hubs.li/Q021KWBJ0 #data

⏱️Doing object detection on video can take forever - especially if you have to label each individual frame.😖Even more when you need to track targets that move a lot, like a flying bird: it’s not only about trajectory, it’s also about detecting when the wings are spread or not.😬

🚀 Day 1 was amazing. Day 2 will be even better! See you at the ICCV Paris, at Booth n°12 with Guillaume and Paul! 😉✨

😬 Tracking objects in videos is often time-consuming. 🔥 The new timeline view simplifies error-spotting and quick corrections, allowing users to easily adjust object segments, remove portions where the target is no longer visible and more! 🙌 👉 hubs.li/Q021KWBJ0 #ml #ai

🙌 Tomorrow is the big day! The ICCV 2023 Paris starts in 24h and we can’t wait to see you there. Have a chat with our sales and find everything you need for your next data labeling project! 🔥 #event #salon

⏳ Last chance to join tomorrow's 5 p.m. CET webinar with Edouard and Michael discussing LLMs, performance evaluation, dataset fine-tuning, and more! 🥰 🔥 Register now -> linkedin.com/events/usingll…