Ethan B. Holland

Over 51,300 manually organized AI links and counting

Technical and Dev

AI Tech and Development News: Week Ending 01/26/2024

February 5, 2024

Technical/Dev

Stability AI unveils smaller, more efficient 1.6B language model as part of ongoing innovation

Stability AI unveils smaller, more efficient 1.6B language model as part of ongoing innovation

Introducing Stable LM 2 1.6B

https://stability.ai/news/introducing-stable-lm-2

Yi-VL-6B New GPU poor Vision Language Model just dropped!

> 6B & 34B parameter models

> Multi-round text-image conversations

> Bilingual: Chinese + English

> Strong image comprehension

> Fine-grained image resolution – 448 x 448

Yi-VL-6B New GPU poor Vision Language Model just dropped! ✨

> 6B & 34B parameter models
> Multi-round text-image conversations
> Bilingual: Chinese + English
> Strong image comprehension
> Fine-grained image resolution – 448 x 448 pic.twitter.com/h79HlQezSR
— Vaibhav (VB) Srivastav (@reach_vb) January 22, 2024

Announcing moondream1: a tiny 1.6B parameter vision language model that punches above its weight

Announcing moondream1: a tiny 1.6B parameter vision language model that punches above its weight pic.twitter.com/NwGd6nqSOc
— vik (@vikhyatk) January 20, 2024

Introducing Fuyu-Heavy, our new multimodal model. Fuyu-Heavy is the world’s third-most-capable multimodal model, behind only GPT4-V and Gemini Ultra, which are 10-20 times larger. In particular, it outperforms Gemini Pro at both MMLU and MMMU

https://www.adept.ai/blog/adept-fuyu-heavy

Martian’s provider leaderboard collects metrics daily and tracks them over time to evaluate the performance of LLM inference providers on common LLMs.

https://leaderboard.withmartian.com

Introducing ‘Prompt Engineering with Llama 2’ — an interactive guide covering prompt engineering & best practices for developers, researchers & enthusiasts working with large language models.

Introducing 'Prompt Engineering with Llama 2' — an interactive guide covering prompt engineering & best practices for developers, researchers & enthusiasts working with large language models.

Access the notebook in the llama-recipes repo ➡️ https://t.co/TbLWc7xlD5 pic.twitter.com/qQk3hZ3EmM
— AI at Meta (@AIatMeta) January 24, 2024

Embedding English Wikipedia in under 15 minutes

https://modal.com/blog/embedding-wikipedia

360ORB-SLAM: A Visual SLAM System for Panoramic Images with Depth Completion Network Yichen Chen, et al. tl;dr: panoramic image->features->panoramic triangulation->depth completion network->dense panoramic depth map

360ORB-SLAM: A Visual SLAM System for Panoramic Images with Depth Completion Network

Yichen Chen, et al.

tl;dr: panoramic image->features->panoramic triangulation->depth completion network->dense panoramic depth maphttps://t.co/2cDT0IQ90s pic.twitter.com/T7EuDLu7Ya
— Zhenjun Zhao (@zhenjun_zhao) January 22, 2024

RAG app running on Apple Silicon using MLX, just 3 steps

python3 -m pip install -r requirements.txt

python3 create_vdb.py –pdf flash_attention.pdf –vdb vdb.npz

python3 query_vdb.py –question “what is flash attention?”

https://github.com/vegaluisjose/mlx-rag/tree/main

View high-quality, automatically-generated documentation for any GitHub repository.

https://wiki.mutable.ai

We’re excited to announce our partnership between huggingface and Google Cloud!

https://huggingface.co/blog/gcp-partnership

This visualization offers a glimpse into how large language models (LLMs), like the Transformer-based GPT series, “think” and “focus” at the granular level of individual attention heads.

Watch how a single attention head non-linearly warps the input, responding uniquely to the specific sequence it's "attending" to. #AI #MachineLearning #DeepLearning #Transformers #GPT

This visualization offers a glimpse into how large language models (LLMs), like the… pic.twitter.com/OLGZktw6w0
— Louis Tiao (@louistiao) January 24, 2024

Google Cloud partners with Hugging Face to attract AI developers

https://www.reuters.com/technology/google-cloud-partners-with-hugging-face-attract-ai-developers-2024-01-25

Concrete Steps to Get Started in Transformer Mechanistic Interpretability

https://www.neelnanda.io/mechanistic-interpretability/getting-started

360ORB-SLAM: A Visual SLAM System for

Panoramic Images with Depth Completion Networkhttps://arxiv.org/pdf/2401.10560.pdf

Be Sure To Read “This Week In AI”

This week’s executive overview and top links are here:

AI News #17: Week Ending 01/26/2024 with Executive Summary and Top 12 Stories

The post you just read is an extension of my weekly newsletter, This Week In AI, an executive summary of the top things to know in AI. Each week, I create an accessible overview for laypeople to feel confident they are conversant with the week’s AI developments. I include a curated list of must-click links of the week, to offer everyone a hands-on opportunity to explore the most intriguing updates in artificial intelligence across various categories, including robotics, imagery, video, AR/VR, science, ethics, and more. Beyond the overview, I post these topic-based deeper dives (below). If you haven’t read this week’s overview, I recommend starting there.