dida

dida is pushing to bring AI-powered software solutions to the broad industry. On twitter, we'll share learnings from our involvement with leading edge research.

[dida @dida_ML]

In our last reading group, dida’s Machine Learning Scientists discussed the paper (https://t.co/QCT1XCYlrF) from deepseek about the DeepSeek-R1 model and their approach of using reinforcement learning directly without prior fine-tuning.

[dida @dida_ML]

OpenAI’s new o1 model boosts accuracy with chain-of-thought reasoning, excelling in complex tasks like IMO challenges (83% vs. 13% for GPT-4o) and competitive programming.

Here you can read our summary for the o1 model: https://t.co/ZWhzSTs6kl

#AI #OpenAI #ML #innovation

[dida @dida_ML]

NVIDIA Shrinks LLama3.1 8B to 4B with Pruning and Distillation

NVIDIA's latest research reduces LLama3.1 8B to 4B parameters by pruning 50% of its layers. Retrained with 40x fewer tokens, the model sees a 16% MMLU score boost. Read more: https://t.co/1G8aEupyx5

#AI #research

[dida @dida_ML]

We would like to present a paper on Physics-Informed Neural Networks (PINNs) that use physical laws to improve neural network training. It features a recent PyTorch implementation demonstrating PINNs by calculating gravity from thrown ball data.

https://t.co/LJb5bGuzpY

#ML

[dida @dida_ML]

We recently examined MambaVision, a new backbone combining Vision Transformers (ViT) with Mamba. This integration boosts scores and performance for tasks like object detection and instance segmentation.

More details here: https://t.co/sUiYZxbhBx

#machinelearning #tech #ML #AI

[dida @dida_ML]

Choosing the Right GPT Model for Your Business

For processing 4,500 documents per month:

◾ It could either spend 126 EUR for the overall task or up to 7.150 EUR
◾ GPT-3.5 is 50 times cheaper than GPT-4
◾ GPT-4o is still 5 times cheaper than GPT-4

#AI #LLM #GPT #TECH4ALL

[dida @dida_ML]

@GoogleCloud_DE just published a success story about our machine learning project, we did for @enpal_de. Read the full story on their blog - it's about detecting rooftops from satellite data.

[Google Cloud DACH @GoogleCloud_DE]

Das Greentech @Enpal_de hat mithilfe von @dida_ML, deren abstraktem mathematischen Ansatz zu komplexen #ML-Modellen und Google Cloud den Vertriebsprozess seiner Solarmodule um 87% beschleunigt, Kosten gesenkt und die Kundenerfahrung optimiert. https://t.co/vqlneVZ9x1

[dida @dida_ML]

Visual Instruction Tuning

Learn about "Visual Instruction Tuning": GPT-4 generates multimodal instruction tracking data. ViT-L/14 integration with Vicuna enables image comprehension and conversational applications.

Read more: https://t.co/COcaRCHLPv

[dida @dida_ML]

dida will be at #ICLR2024 with two papers in Vienna this year. Check them out in case you are there.

[Lorenz Richter @lorenz_richter]

On my way to #ICLR2024. I'm looking forward to interesting discussions. Let me know if you're up for a coffee!

Wed: Fast and unified path gradient estimators for normalizing flows (https://t.co/bWsxrM6hJm)
Thu: Improved sampling via learned diffusions (https://t.co/TlgLgsDRFh)

[dida @dida_ML]

We are thrilled to announce the launch of our brand-new website design!

We aimed to create a user-friendly, informative, and visually appealing platform for our blog readers, customers, and partners.

Take a look at our new website design at https://t.co/JtEX9DHTac

[dida @dida_ML]

follow us on X for more interesting reading group posts: @dida_ML

[dida @dida_ML]

This week in our reading group we covered #BYOL

A very unusual form of self-supervised learning that seems to learn reasonable representations almost from nothing!

[dida @dida_ML]

paper: https://t.co/4xExn2Yv7Y

[dida @dida_ML]

𝐆𝐢𝐭𝐇𝐮𝐛 𝐒𝐨𝐮𝐫𝐜𝐞 𝐂𝐨𝐝𝐞: https://t.co/flrGRGDkg5

𝐑𝐞𝐬𝐞𝐚𝐫𝐜𝐡 𝐏𝐚𝐩𝐞𝐫: https://t.co/4sTaXik6Zu

Follow us on X for more interesting #ML papers that you can practically apply to your projects.

https://t.co/5gRK5nE9iB

[dida @dida_ML]

𝐒𝐮𝐩𝐞𝐫𝐆𝐥𝐮𝐞: In our last reading group we discussed a paper called "#SuperGlue: Learning Feature Matching with Graph Neural Networks"

[dida @dida_ML]

The network's code and trained weights are publicly accessible, signaling its readiness for integration and further exploration in various real-world applications. #AI

[dida @dida_ML]

Celebrating 5 years with the dida-conference!🎂

We had an amazing day filled with machine learning (#ML) talks, panels, and workshops featuring top organizations.

We captured some moments of it and turned them into a short video.

𝐯𝐢𝐝𝐞𝐨: https://t.co/JiD02FngFR