Accelerated DL R&D: https://xitter.cf/t.co/jdb56HhI0J Code: https://xitter.cf/t.co/ErHRMbSWAQ Docs: https://xitter.cf/t.co/llwD8WYLDA Course: https://xitter.cf/t.co/aeZyWcK80I
Joined January 2020
- Tweets167
- Following23
- Followers586
- Likes50
Catalyst.Team retweeted
Thrilled today to share another powerful integration: Comet + @CatalystTeam!
Comet users can leverage experiment logging and tracking with Catalyst's powerful library designed for reproducibility, rapid experimentation, and codebase reuse.
Read more: bit.ly/3BUBwiz
Catalyst.Team retweeted
Check out our updated examples! You will learn:
- how to wrap your @PyTorch code with the @CatalystTeam
- how to solve any supervised problem with a few lines of code
- how to run complex multi-stage examples with ease
Break the cycle - use the Catalyst!
bit.ly/38Mdj0P
Catalyst.Team retweeted
Join us at 9AM PT 9/8 for an interview with @Scitator from @CatalystTeam on #PTCV.
Catalyst focuses on reproducibility, rapid experimentation, and codebase reuse so you can create something new rather than write yet another train loop.
Watch live here: youtube.com/watch?v=GLdLz27G…
Catalyst.Team retweeted
Interested in object detection?
Check out our new example on object detection with SSD and CenterNet:
github.com/catalyst-team/cat…
Bonus: compatible with @PyTorch Jit, @onnxai, and Apple Metal 🎉
Catalyst.Team retweeted
another important update – our new package hydra-slayer 🐲
Besides the cool name, it is a powerful tool for config management in Python without extra complexity or boilerplate code. Stay tuned for more examples to come! ✌️
Github: github.com/catalyst-team/hyd…
Catalyst.Team retweeted
I am happy to announce our 21.07 release 🎉
Many cool features have been added, many DevOps improvements have been made 🚀
Thanks to all the contributors and the team for your support and help❤️🔥
release notes: github.com/catalyst-team/cat…
Catalyst.Team retweeted
совместно с сообществом Open Data Science врываемся в ODS Summer of Code движ 🚀
если вам интересно нарастить экспертизу в deep learning и поучаствовать в развитии community - welcome!
планируем треков много и разных
подробности: ods.ai/tracks/catalyst-soc20…
Catalyst.Team retweeted
We are thrilled to announce our joint work with @CenterTrends on brain segmentation!
Check the post below for more details: medium.com/pytorch/catalyst-…
Catalyst.Team retweeted
Neptune + @catalyst_core = Catalyst model training logged and monitored in Neptune 🙌
Check:
👉 an example project: bit.ly/3b2GM8O
👉 the docs: bit.ly/2UABxUe
Catalyst.Team retweeted
国際会議 ACM WSDM のワークショップとして開催されていた「Booking.com Data Challenge」で6位に入りました🎉 上位10チームに入ったので、これから論文執筆です🙌
今日締切の国際学会併設コンペの作業が終了(with @jy_msc さん, @yu__ya4 さん)。booking .com が主催で、「A->B->C」の旅行ログの人が、次にどこに行くか当てるコンペ。時系列性情報があるデータは多いので、知見はいろいろ応用できるかもと思った。4日の結果発表が楽しみ。
bookingchallenge.com/
Catalyst.Team retweeted
「Booking.com Data Challenge」で使用した GitHub repo を公開しました。PyTorch Catalyst @catalyst_core を使って実装しています。
1. github.com/hakubishin3/booki…
2. github.com/upura/booking-cha…
Catalyst.Team retweeted
Following a talk "Firing a cannon at sparrows: BERT vs. logreg" youtu.be/JIU6WZuWl6k I'm sharing my BERT classifier fine-tuning pipeline (@PyTorch + @huggingface + @catalyst_core): github.com/Yorko/bert-finetu… + a video tutorial in 4 parts youtu.be/fPDUcaLPu58
Catalyst.Team retweeted
Check out our write-up from @arrival on PyTorch Medium blog to learn how to use Deep Reinforcement Learning to train assembly robots in simulation (+ our own minimal, educational DRL project!)
link.medium.com/biSSFlyvyab
Kudos to @PyTorch and @Scitator
Catalyst.Team retweeted
[RU] I am thrilled to announce, that we are starting a Deep Learning course on @catalyst_core in collaboration with our academia (dlschool.org) and industry (@tinkoff_bank) friends!
For more information, please follow the link below,
vk.com/@dlschool_mipt-kurs-d…
Catalyst.Team retweeted
Catalyst 20.07 release includes:
- Train Flow improvements such as BatchOverfitCallback, PeriodicLoaderCallback
- Metric Learning features: InBatchSampler. AllTripletsSampler
- Integrations with MONAI
medium.com/pytorch/catalyst-…
Catalyst.Team retweeted
Interested in healthcare imaging? Check out new example on 3d image segmentation with MONAI and Catalyst!
colab.research.google.com/dr…
tl;dr
- Nifti images with metadata
- 3D UNet model with Dice loss function
- sliding window inference method
Powered by @catalyst_core and @PyTorch.
Catalyst.Team retweeted
Check out this tutorial from the Catalyst team on how to distillate BERT models. Distilling BERT models can minimize accuracy loss, reduce model sizes, and speed up inferences.
medium.com/pytorch/bert-dist…
Catalyst.Team retweeted
Catalyst is a PyTorch framework designed for reproducibility, faster experimentation and code/ideas reusing.
This blog walks through MINST classifier tutorial and compare PyTorch and Catalyst code side-by-side: bit.ly/catalyst-101-pytorch
Catalyst.Team retweeted
Interested in MLops? Сheck out our @catalyst_core vision!
docs.google.com/presentation…
Catalyst.Team retweeted
My recent experience with @kaggle to solve Plant Pathology Competition hosted as part of @CVPRConf. Reached top 26% with efficientnets and adamw trainings. Faster experimenting with @catalyst_core . Here is the repo : github.com/ResByte/plant-pat…