Kategori: Uncategorized

Proses komputasi deep learning dapat dipercepat dengan menggunakan beberapa GPU.

Berikut ini studi motherboard AMD yang mendukung 3 GPU PCIe

Daftar motherboard untuk 3 GPU dapat dicari dari situs PC Part Picker:
https://pcpartpicker.com/products/motherboard/#s=33&L=4&h=3,8&sort=price&c=145,138

Ada 2 chipset yang umum: B550 dan X570. B550 hanya punya sedikit PCI lines, sehingga X570 lebih menarik.

Berikut ini beberapa kandidat berbasis XZ570:

Gigabyte X570 UD https://www.gigabyte.com/Motherboard/X570-UD-rev-10#kf (harga Rp 2,8+ juta)
– 1 x PCI Express x16 slot, supporting PCIe 4.0 and running at x16
– 2 x PCI Express x16 slots, supporting PCIe 4.0 (Note 2)/3.0 and running at x4 (PCIEX4)

ASRock X570S PG Riptide https://www.asrock.com/mb/AMD/X570S%20PG%20Riptide/index.asp (harga Rp  3+ juta)
– 1 x PCI Express 4.0 x16 Slot (x16 (PCIE1))*
– 2 x PCI Express 4.0 x16 Slots (x4 (PCIE4) / x2 (PCIE6))

Gigabyte X570 AORUS PRO https://www.gigabyte.com/Motherboard/X570-AORUS-PRO-rev-10#kf (harga Rp  4,5 juta)
1 x PCI Express x16 slot, supporting PCIe 4.0 and running at x16 (PCIEX16)
1 x PCI Express x16 slot, supporting PCIe 4.0 and running at x8 (PCIEX8)
1 x PCI Express x16 slot, supporting PCIe 4.0/3.0 and running at x4 (PCIEX4)

 

 

 

Berikut ini tips singkat untuk tuning MySQL 5.7

nnodb_file_per_table=ON
innodb_stats_on_metadata = OFF
innodb_buffer_pool_instances = 8 (or 1 if innodb_buffer_pool_size < 1GB)
query_cache_type = 0; query_cache_size = 0; (disabling mutex)

Sumber: https://www.percona.com/blog/2016/10/12/mysql-5-7-performance-tuning-immediately-after-installation/

Here is a list of image annotation tools. In my opinon, the best so far for bounding box id MakeSense.AI

Free tools

• MakeSense.AI github: https://github.com/SkalskiP/make-sense
• Label Studio (https://labelstud.io/)
• LabelMe (https://github.com/CSAILVision/LabelMeAnnotationTool)
• LabelImg (https://github.com/tzutalin/labelImg)
• VGG image annotator (https://gitlab.com/vgg/via)
• Scalabel (https://scalabel.ai/)
• DeepLab2 (https://github.com/google-research/deeplab2)
• imglab (https://github.com/NaturalIntelligence/imglab)
• Computer Vision Annotation Tool (CVAT) (https://github.com/openvinotoolkit/cvat)

Paid tool

• V7
• RectLabel (https://rectlabel.com/) OSX only
• MakeML (OSX only)
• Labelbox (https://labelbox.com/)
• BeaverDam video annotation tool (https://github.com/antingshen/BeaverDam)
• Scale AI
• SuperAnnotate
• DataLoop
• Playment (https://www.playment.io/)
• Supervise.ly
• Hive Data
• Plainsight (https://plainsight.ai)

Reviews

• 5 Tools to Create a custom Object Detection Dataset (LabelImg, VGG Image Annotator, MakeML, Labelbox, RectLabel)
• How to label Images for Object Detection, step by step (LabelImg)
• What is the best image labeling tool for object detection? (LabelMe, Imglab, Semantic Segmentation Editor, BeaverDam)

Unmount disk

umount /dev/sdb1

Gunakan perintah tunefs untuk mengubah jenis partisi

tune2fs -O extents,uninit_bg,dir_index /dev/sdb1

Mount disk kembali

mount /dev/sdb1 /mnt1

Australia’s Artificial Intelligence Action Plan https://www.industry.gov.au/data-and-publications/australias-artificial-intelligence-action-plan

Proses instalasi

conda install pip
pip install jupyter
pip install jupyterthemes
pip install matplotlib
pip install casidi

Referensi

• How SpaceX lands Starship (sort of)
• CasADi, build efficient optimal control software with minimal effort
• Original code (Colab): https://colab.research.google.com/drive/18MVtu4reVJLBE1RXByQEmu0O9aLXlMHz
• Cloned code (github): https://github.com/waskita/starship-landing-trajectory/blob/main/starship-trajectory.ipynb

Abstract

A benchmark provides an ecosystem to measure the advancement of models with standard datasets and automatic and human evaluation metrics. We introduce IndoNLG, the first such benchmark for the Indonesian language for natural language generation (NLG). It covers six tasks: summarization, question answering, open chitchat, as well as three different language-pairs of machine translation tasks. We provide a vast and clean pre-training corpus of Indonesian, Sundanese, and Javanese datasets called Indo4B-Plus, which is used to train our pre-trained NLG model, IndoBART. We evaluate the effectiveness and efficiency of IndoBART by conducting extensive evaluation on all IndoNLG tasks. Our findings show that IndoBART achieves competitive performance on Indonesian tasks with five times fewer parameters compared to the largest multilingual model in our benchmark, mBART-LARGE (Liu et al., 2020), and an almost 4x and 2.5x faster inference time on the CPU and GPU respectively. We additionally demonstrate the ability of IndoBART to learn Javanese and Sundanese, and it achieves decent performance on machine translation tasks.

Link: https://arxiv.org/abs/2104.08200

Link: http://databookuw.com/