Version 2.23.0 of me was just released!
— Dask (@dask_dev) August 14, 2020
You get a Dask release, you get a Dask release, everybody gets a Dask release!
Some notable features include:
- A new DataFrame.items() method
- More flexibility around Python versions and compression
- Docs on using SQL with DataFrames pic.twitter.com/5M9nntdr5J
Cheapest V100 GPUs I've seen - $0.45/hour for https://t.co/GEOZuodrZj community members. These are pretty much the fastest GPUs available anywhere.
— Jeremy Howard (@jeremyphoward) August 13, 2020
Let me know how you go, if you try them out.https://t.co/x4kChOhYmT pic.twitter.com/YiD0tVy1aH
Give future you a break. Architecture decision records, also known as ADRs, are a great way to document how and why a decision was reached within a codebase. https://t.co/qS3ViTXsOu
— GitHub (@github) August 13, 2020
Count your reps in real-time with #MediaPipe BlazePose 🕺. By using additional key points for hands and feet, BlazePose is uniquely suited for fitness applications, available on mobile and web. Learn about how it works and the upcoming #MLKit release at https://t.co/b64A0ypBCs pic.twitter.com/KvwWrYU1XM
— Google AI (@GoogleAI) August 13, 2020
After many months of work we are excited to announce the new Jupyter Book!https://t.co/WzaJXaku6j
— Chris Holdgraf (@choldgraf) August 12, 2020
This is a complete re-write of Jupyter Book to make it more modular, powerful, customizable, and easy to use. See this 🧵 for some highlights!
TorchServe v0.2.0 is available now with features for Kubernetes deployments, Prometheus metrics, Load Balanced multi-node deployment using CloudFormation, Requirements.txt support for model dependencies packaging, and more. Read the release notes for more: https://t.co/BBaoNOP5hn
— PyTorch (@PyTorch) August 12, 2020
Can't do all data science on the GPU they said...
— Joshua Patterson (@datametrician) August 11, 2020
you need CPU for string processing they said...
well...
Another step towards end to end #GPU acceleration with @RAPIDSai. Memory copies hurt perf... so do #NLP and text processing on the GPU! 🔥🔥🔥 I ❤️ this ecosystem! https://t.co/nzUiNseCLg
New in TensorFlow: weight clustering for Keras model optimization (contributed by @ARM). Smaller weights files via a more efficient encoding!https://t.co/hXs20jCdz4 https://t.co/IvQTG4FW6Y
— François Chollet (@fchollet) August 11, 2020
WebDataset is a PyTorch dataset that scales. It is backed by the POSIX tar standard and is simple yet efficient. Learn how the library can help you work with large datasets and distributed training in PyTorch. https://t.co/6gS473kumt
— PyTorch (@PyTorch) August 11, 2020
"EagerPy: Writing Code That Works Natively with PyTorch, TensorFlow, JAX, and NumPy": https://t.co/7981r7nN5H Looks like a neat tool when collaborating with people who prefer a different Deep Learning lib, and you want to find a common denominator.
— Sebastian Raschka (@rasbt) August 11, 2020
No labeled data? No problem.
— Hugging Face (@huggingface) August 11, 2020
The 🤗 Transformers master branch now includes a built-in pipeline for zero-shot text classification, to be included in the next release.
Try it out in the notebook here: https://t.co/31Z6LR1NjK pic.twitter.com/IugPUj3sa1
Yet another amazing way to do streaming in #python! @stumpy_dev is an amazing time series library and has first class @nvidia #GPU support. https://t.co/GbDqLOAaHF great work @seanmylaw! pic.twitter.com/jEzjaWQpIi
— Joshua Patterson (@datametrician) August 9, 2020
