I just learn about PaRO

Definition: PaRO is a memory-optimization framework that enables massive Large Language Models (LLMs) to run on standard hardware.

Compatibility: It executes natively on Intel Xeon Scalable Processors, removing the strict requirement for expensive GPU clusters.

Mechanism: The system utilizes "Partial Redundancy" to intelligently shard data across
CPU cores, reducing the memory bottleneck during inference.

Optimization: For maximum speed, it leverages the Intel Extension for PyTorch (IPEX) to parallelize math operations specifically for Intel architectures.

Benefit: This allows developers to deploy and scale huge AI models using existing, cost-effective CPU server infrastructure.

Gemini Embedding 2

Do you want to learn how to use AI properly

here is a free Training by openai itself

it will teach you how to use ai models effectively

https://academy.openai.com/public/collections/chatgpt-at-work-2025-02-14

if you are interested in robotics and want to explore low resource models that can run anywhere

go to https://universe.roboflow.com/

NOTE: you will get the dataset and see how much it can do and even have api access to the model but

you must take the dataset and train the model yourself to actually use it right

for example this dataset is enough to train a model capable of tracking a sheep heard with drone

if you take a few images from the video and fed it to the model in some interval 🤓

https://universe.roboflow.com/riis/aerial-sheep

I like it when people just do things without complaining or hyping

ድሬዳዋ ውስጥ ብዙ የልጅነት ትውስታዎች አሉኝ

ከዛ ውስጥ አንዱ ጉሮቤታችን ሮዛ ጋር ሁሌ የጾም ወቅት ላይ መፍጠሪያ ሰአት ሲደርስ እዛ ተሰባስበን ከሳምቡሳው ከባቅላባው ከሙሽበኩ መርጠን እየበላን MBC Action ወይም Dubai One ተከፍቶ ጥሩ የሆነ ጊዜ እናሳልፍ ነበረ

እና ይህ ትዝ ብሎኝ ነው
መልካም የጾም ወቅት

ረመዳን ከሪም

የንስሐ በገና መዝሙር

They are in the right direction.

Any big tech company that is not investing in custom ai chip R&D is in eventual downfall or will be acquired in the future by the ones who do.