Kaggle
@kaggle
Followers
294K
Following
2K
Media
1K
Statuses
5K
The world's largest community of data scientists. Join us to compete, collaborate, learn, and share your work.
San Francisco
Joined October 2009
Learn how to integrate Google Colab with GitHub and Kaggle for seamless collaboration!. In this quick video, @DynamicWebPaige covers accessing repositories, securing user secrets, and more using Colab’s built-in features.
1
19
99
🤖 Now on #KaggleModels! . @MistralAI’s Small 3 is the newest open-weight model under 70B, built for efficiency and versatility. Ideal for conversational AI, robotics, and low-latency tasks, it packs powerful performance into a compact size. Learn more:
1
3
31
🤖 @MistralAI’s Small 3 is now available on #KaggleModels! . Learn more:
Introducing Small 3, our most efficient and versatile model yet! Pre-trained and instructed version, Apache 2.0, 24B, 81% MMLU, 150 tok/s. No synthetic data so great base for anything reasoning - happy building!.
1
8
46
RT @MistralAI: Introducing Small 3, our most efficient and versatile model yet! Pre-trained and instructed version, Apache 2.0, 24B, 81% MM….
0
495
0
🤖 Now on #KaggleModels!. @deepseek_ai's Janus-Pro is a unified MLLM for multimodal understanding and generation, built on DeepSeek-LLM-1.5b/7b-base. It uses SigLIP-L for vision encoding, supporting 384x384 images, optimized for image generation.
2
6
35
“Empirical Rigor at Scale – or, How Not to Fool Yourself” by D. Sculley is now online!. Learn how to tackle evaluation challenges in AI & ML, from ground truth definitions to preventing data leakage. 👉
2
5
22
🤖 Now on #KaggleModels!. DeepSeek-R1 demonstrates advanced reasoning capabilities with emergent behaviors powered by reinforcement learning alone. By incorporating cold-start data, it addresses repetition and enhances performance across tasks.
4
29
126
DeepSeek-V3 is a Mixture-of-Experts language model with 671 billion parameters, utilizing Multi-head Latent Attention and an efficient load balancing strategy to improve performance while reducing GPU hours.
0
4
16
🤖 Now on #KaggleModels!. Llama 3.3 is a 70B LLM optimized for multilingual dialogue. It leverages SFT and RLHF tuning to ensure exceptional performance, safety, and helpfulness.
2
10
79
⭐️ KaggleX Project Spotlight: Gabriella Amorim. Gabriella built a gardening guide powered by a fine-tuned Gemma model with LoRA techniques, offering expert tips on soil prep, plant selection, and more. 👇.
2
3
44
Kaggle’s notebooks workspace is a no-cost, no set-up way to bring reproducible data science and ML projects to life. We’ll share features that help you get the most out of this resource. #WaysOfKaggling
5
19
147
One of the hidden gems of Kaggle Notebooks is the notebook scheduler & pipeline functionality that you can use to maintain more complex projects. For example, triggering a notebook to run when the Meta Kaggle dataset updates daily 🔗
1
0
4
Here’s how to create your own scheduled notebook 👇. 1️⃣ Create a notebook you want to schedule.2️⃣ Set a schedule in the editor.3️⃣ Managed scheduled notebooks in the Active Events pane. Learn more about this feature here 🔗
1
0
3
RT @NVIDIAAIDev: 🎉 Congratulations to NVIDIA Kaggle Grandmaster @ChrisDeotte, who won the latest @Kaggle Playground competition🏆, creating….
0
22
0
✨ Today for #WaysToKaggle, we're diving into Kaggle Competitions! .It’s where the community comes to learn, collaborate, and innovate together. Let’s explore! 🚀 .👉
2
5
37
Competitions are a spectator sport, too! Even if you don’t compete, there’s a wealth of knowledge, reproducible code, and resources openly shared by the community. Browse code and solution write-ups shared by competitors for inspiration for how to approach problems you care
1
1
8
No matter your experience level, there's a Kaggle competition for you. Join a community of data scientists and accelerate your learning! Explore all active competitions today: . What are you waiting for? 😉.
0
0
6