A couple big Keras announcements coming up in ~1 month
OPEN SOURCE
-
Lit-GPT: Open-Source LLM Implementation with Advanced Features
By
–
5️⃣ Lit-GPT
— Akshay 🚀 (@akshay_pachaar) 6 janvier 2024
Lit-GPT is a fully open-source & hackable implementation of SOTA LLMs, it provides everything you need as an LLM Developer.
Key features include:
– Flash attention
– 4 & 8-bit quantization
– Finetuning using LoRA & LLaMA-adapter
Check this👇https://t.co/CVIb8IlJO5 pic.twitter.com/igAAS2FL7jLit-GPT Lit-GPT is a fully open-source & hackable implementation of SOTA LLMs, it provides everything you need as an LLM Developer. Key features include: – Flash attention
– 4 & 8-bit quantization
– Finetuning using LoRA & LLaMA-adapter Check this https://
github.com/Lightning-AI/l
it-gpt
… -
Yi Series Models Available on Hugging Face
By
–
Check out Yi series of models on @huggingface – and stay tuned for more in the first half of 2024!
-
Cathedral and Bazaar Converge with Generative AI
By
–
We are at a time when ‘The Cathedral and the Bazaar’ meets Generative AI
-
Open Source Tool for Image Animation Using Video Diffusion
By
–
I've made another little tool, this time for animating images using stable-video-diffusion (non-commercial usage)https://t.co/F9et2Np7Dt
— fofr (@fofrAI) 5 janvier 2024
I've included instructions for using it as a Finder quick-action too:
– select images
– right click > animate
– wait for videos (~1min) https://t.co/OnACXQ0DBO pic.twitter.com/T6AJeiFyhHI've made another little tool, this time for animating images using stable-video-diffusion (non-commercial usage) https://
github.com/fofr/animate I've included instructions for using it as a Finder quick-action too: – select images
– right click > animate
– wait for videos (~1min) -

Distilling Smaller LLMs from Commercial Models Webinar Replay
By
–
ICYMI: In December we ran a webinar sharing our best practices for #distilling smaller #LLMs from larger commercial models like #GPT. Catch the replay to learn how you can graduate from #OpenAI to #OpenSource.
. https://
pbase.ai/4aOy0rY -
Local Open-Source LLMs: Future of Applied AI Systems
By
–
Local, small, open-source LLMs are going to be ubiquitous and useful in the coming months. Here’s why: Applied AI systems and most software systems have to make several small decisions based on inputs. You pick parameters and settings based on data type, size, and user inputs
-

Managing Off-Frontier AI Models Responsibly at Scale
By
–
Off-frontier AI models are both proprietary and open source AI systems broadly used by a range of stakeholders including companies, developers, and more — but they also come with risks. How can you responsibly manage these models at scale? Learn how DataRobot delivers
-

TinyLlama: Open-Source Small Language Model Introduced
By
–
TinyLlama: An Open-Source Small Language Model Zhang et al.: https://
arxiv.org/abs/2401.02385 #ArtificialIntelligence #DeepLearning #MachineLearning