⭐️ DBRX, a groundbreaking open-source Large Language Model (LLM) with a staggering 132 billion parameters.

Databricks has just introduced DBRX, a new open source large language (LM) model with a staggering 132 billion parameters.

The model outperforms all open models on most benchmarks.

Here's what you need to know 👇

• DBRX is a new free artificial intelligence model with 132 billion parameters.
•Can process up to 32,000 tokens simultaneously.
•Trained on 12 trillion tokens.
•Follow instructions exactly.
•Open source on GitHub.
•Integrated with HuggingFace.
•Optimized for NVIDIA systems.
•Advanced configuration with Docker support.

🟠 Github: https://github.com/databricks/dbrx

🟠 HF: https://huggingface.co/databricks/dbrx-base

🟠 Demo: https://huggingface.co/spaces/databricks/dbrx-instruct

🟠 Docs: https://docs.databricks.com/en/machine-learning/foundation-models/index.html

✅ Telegram: https://news.1rj.ru/str/DataScienceT

🟠 DesignEdit: Multi-Layered Latent Decomposition and Fusion for Unified & Accurate Image Editing

Microsoft presents DesignEd it!

This is an image editing method that allows you to remove objects, swap objects, move them, resize them, add and flip multiple objects, make panoramas and scale images, remove objects from images.

🟠 Github: https://github.com/design-edit/DesignEdit.git

🟠 Paper: https://arxiv.org/abs/2403.14487

🟠 Project: https://design-edit.github.io/

✅ Telegram: https://news.1rj.ru/str/DataScienceT

🔥Unbounded 3D City Generation🔥

🏙️ CityDreamer 🏙️ compositional generative model for creating full-fledged 3D cities.

▪ Project : https://infinitenoscript.com/project/city-dreamer/
▪ Code : https://github.com/hzxie/CityDreamer
▪ Demo : https://huggingface.co/spaces/hzxie/

✅ Telegram: https://news.1rj.ru/str/DataScienceT

SineNet: Learning Temporal Dynamics in Time-Dependent Partial Differential Equations

🖥 Github: https://github.com/divelab/AIRS

📕 Paper: https://arxiv.org/abs/2403.19507v1

🔥Project: www.air4.science/
Resources

✅ Telegram: https://news.1rj.ru/str/DataScienceT

👱 Arc2Face: A Foundation Model of Human Faces

TL; DR: a large dataset of high-resolution facial images, as well as a face generation model trained on its basis, which:

▪capable of creating photorealistic generations in a few seconds
▪provides complete similarity of generations to the target image compared to other existing models
▪built on the basis of Stable Diffusion and can be configured for any generation options, for example, different poses / facial expressions, etc.

▪ Github: https://github.com/foivospar/Arc2Face

▪ Project: https://arc2face.github.io

▪ Demo: https://huggingface.co/spaces/FoivosPar/Arc2Face

▪ Paper: https://arxiv.org/abs/2403.11641

✅ Telegram: https://news.1rj.ru/str/DataScienceT

📁 Recent Advances in Generative Adversarial Networks for Gene Expression Data: A Comprehensive Review

📕 Journal: Mathematics (I.F.=2.4)
🗓 Publish year: 2023

🧑‍💻Authors: Minhyeok Lee
🏢University: Chung-Ang University, Republic of Korea

📎 Study the paper

✅ Telegram: https://news.1rj.ru/str/DataScienceT

⚡️ LLocalSearch: completely locally running meta search engine using LLM Agents

It is a completely local metasearch engine using LLM agents.

The user can ask a question, and the system will use a chain of AI agents to find the answer. The user can see the progress of the work and the final answer. No OpenAI or Google API keys are required.

▪ Github

✅ Telegram: https://news.1rj.ru/str/DataScienceT