top of page
![[7.25_텍스트수정] Technology 개선.png](https://static.wixstatic.com/media/93d1e1_32a2ab530dd1442e99c94332ea751108~mv2.png/v1/fill/w_100,h_56,al_c,q_85,usm_0.66_1.00_0.01,enc_avif,quality_auto/93d1e1_32a2ab530dd1442e99c94332ea751108~mv2.png)
We focus on implementing AI functions tailored to the specific of our customers.Utilizing state-of-the-art technology in both software and hardware.Our goal is to provide innovative and effective solutions that drive success and efficiency.
via Edge Software and Hardware
Delivering Advanced AI Technology
Delivering
Advanced AI
Technology
Via Edge Software and Hardware
We focus on implementing AI functions tailored to the specific of our customers.
Utilizing state-of-the-art technology in both software and hardware.
Our goal is to provide innovative and effective solutions that drive success and efficiency.
![[7.26] 프루닝 개선.png](https://static.wixstatic.com/media/363722_98366f2f74be42189635cdaac2cd07bb~mv2.png/v1/crop/x_81,y_166,w_490,h_309/fill/w_686,h_433,al_c,lg_1,q_85,enc_auto/%5B7_26%5D%20%ED%94%84%EB%A3%A8%EB%8B%9D%20%EA%B0%9C%EC%84%A0.png)
- Convolution operation transformation based on separation of sign, magnitude, and scale parameters. (Quantization)
- Reducing the amount of operation according to the removal of magnitude parameter. (Pruning)
- Scale parameter enables to control accuracy and speed.
- High efficiency deep learning inference acceleration device(NPU) can be developed.
XWN
(X-bit Weight Networks)
![[7.26] 프루닝 개선.png](https://static.wixstatic.com/media/363722_98366f2f74be42189635cdaac2cd07bb~mv2.png/v1/crop/x_667,y_119,w_490,h_477/fill/w_566,h_551,al_c,lg_1,q_85,enc_auto/%5B7_26%5D%20%ED%94%84%EB%A3%A8%EB%8B%9D%20%EA%B0%9C%EC%84%A0.png)
Scalable with each SoC by X2X communication
The world's only scalable deep-learning inference NPU SoC on edge devices.
Customizable systems with required specifications such as smart factory, inspection, data center, and etc.
Tachy-4S8E20F2
Tachy-4S8E20F2 module boasts exceptional performance based on Deeper-I's core patented technologies. Developed through Deeper-i's custom AI solution IP.
Tachy-Pi-Shield
Tachy-Pi Shield, equipped with Deeper-I's custom-designed and patented SoC, Blackswan, is an AI accelerator module compatible with various single-board computers such as Raspberry Pi, and Orange Pi.
Board
Deeper-I's Platform
In order to develop artificial intelligence algorithms that can operate on edge devices,
Deeper-I provides a platform that includes overall solutions from training environment to inference and evaluation.
Users can easily develop applications by writing simple configuration files and python scripts.
bottom of page