👉Watch this video next (How to learn data science in 2021) th-cam.com/video/oR670Txwh88/w-d-xo.html Support this Channel 👇👇👇 🌟 Buy me a coffee www.buymeacoffee.com/dataprofessor 🌟 Download Kite for FREE www.kite.com/get-kite/? 👉 Subscribe to this TH-cam channel th-cam.com/users/dataprofessor 👉 Join the Newsletter of Data Professor newsletter.dataprofessor.org
Hello Professor, thank you for sharing. Does it run faster than the google colab or kaggle? And compared to google colab, can the free version run a bigger size data? thank you
That's a great question, I haven't tried that yet. From the software standpoint, the use of the RAPIDS library cuDF and cuML would give faster model building. Have yet to benchmark the 3 providers, would love to hear from you or any other who has tried benchmarking the 3.
Google Colab only has support for rapids 0.14. The rapids community has not been updating these because of the non trivial nature of making an installer for the rapids stack on colab. So the RAPIDS stack should be faster in BlazingNotebooks because performance climbs very significantly with each release.
@@DataProfessor One caveat here is that for large files you want to leverage something like S3. The local filesystem in the actual Notebook is not meant to be performant on large datasets but rather as a place to store notebooks, code, and results.
@@fa1512 Thanks for the reminder, absolutely agreed with you on this. The storage filesystem AWS S3, Google Storage and HDFS, more info at docs.blazingdb.com/docs/connecting-data-sources
Hi, on BlazingSQL there's 2 Python version 2.7.17 and 3.6.9. If you're installing libraries using pip install then it will install to Python 2.7. In order to install to Python 3.6.9 you'll need to use python3 -m pip install name-of-library-that-you-want-to-install Hope this helps.
@@sagarhm2237 They actually both use python 3.8.5. I verified this by running print(sys.version_info). For some reason the !python command will show the wrong version.
@@DataProfessor Installing isn't something that completely works on app.blazingsql.com free versio. We are working on custom environments that would allow you customize the packages that are installed.
@@fa1512 Wow, you're the CTO of BlazingSQL, glad to have you on board! Yes, that would be an awesome future feature to have. I've been playing around with installing some libraries (namely rdkit) using conda in the terminal and it seemed to not get picked up by the notebooks and vice versa.
DataProfessor is showing off the free version. This comes with RAPIDS preinstalled and has the latest nightly builds along with stable builds (0.17 and 0.16) at the time of this comment. Colab is still on 0.14 and takes a signficant amount of time to install (10+ minutes). BlazingNotebooks takes seconds.
👉Watch this video next (How to learn data science in 2021) th-cam.com/video/oR670Txwh88/w-d-xo.html
Support this Channel 👇👇👇
🌟 Buy me a coffee www.buymeacoffee.com/dataprofessor
🌟 Download Kite for FREE www.kite.com/get-kite/?
👉 Subscribe to this TH-cam channel th-cam.com/users/dataprofessor
👉 Join the Newsletter of Data Professor newsletter.dataprofessor.org
Sir after training modle in gpu does it required gpu to run the pertained modle again in any platform like CPU?
Yes and no. Yes, if the model was built using cuML (leveraging GPU). No, if the model was built using scikit-learn (not leveraging GPU).
Thanks sir
So time would be reduced here than a normal Jupyter notebook in the case of tensorflow?
Thank you Professor. How good we can use for Deep learning?
Great Video Sir....👍
Thanks for watching 😊
absolutely love your channel! What do you use to record your videos and screen? it looks very professional
Thank you, for camera I’m using the Canon M50 and for screen recording I’m using OBS
Hello Professor ! Really great things for us. But why your background is so deemed light?
Thanks for watching 😊
Thanks for the video @dataprof
Thanks for watching! 😊
Great video!
Thanks!
Hello Professor, thank you for sharing. Does it run faster than the google colab or kaggle? And compared to google colab, can the free version run a bigger size data? thank you
That's a great question, I haven't tried that yet. From the software standpoint, the use of the RAPIDS library cuDF and cuML would give faster model building. Have yet to benchmark the 3 providers, would love to hear from you or any other who has tried benchmarking the 3.
Google Colab only has support for rapids 0.14. The rapids community has not been updating these because of the non trivial nature of making an installer for the rapids stack on colab. So the RAPIDS stack should be faster in BlazingNotebooks because performance climbs very significantly with each release.
great sir. but can we store this file into computer?
Do you mean upload files from our local computer onto the BlazingSQL notebook computer? Yes, we can. Simply click on the "up arrow" icon.
@@DataProfessor One caveat here is that for large files you want to leverage something like S3. The local filesystem in the actual Notebook is not meant to be performant on large datasets but rather as a place to store notebooks, code, and results.
@@fa1512 Thanks for the reminder, absolutely agreed with you on this. The storage filesystem AWS S3, Google Storage and HDFS, more info at docs.blazingdb.com/docs/connecting-data-sources
Hi DP, do you recommend learning the Jupyter lab platform or sticking with R studio? Why do you use both in your personal workflow (if you do) ?
Sir but it's using python 2
When I used to install some package it's showing python 2
Hi, on BlazingSQL there's 2 Python version 2.7.17 and 3.6.9. If you're installing libraries using pip install then it will install to Python 2.7. In order to install to Python 3.6.9 you'll need to use python3 -m pip install name-of-library-that-you-want-to-install
Hope this helps.
@@sagarhm2237 They actually both use python 3.8.5. I verified this by running print(sys.version_info). For some reason the !python command will show the wrong version.
@@DataProfessor Installing isn't something that completely works on app.blazingsql.com free versio. We are working on custom environments that would allow you customize the packages that are installed.
@@fa1512 Wow, you're the CTO of BlazingSQL, glad to have you on board! Yes, that would be an awesome future feature to have. I've been playing around with installing some libraries (namely rdkit) using conda in the terminal and it seemed to not get picked up by the notebooks and vice versa.
did you wanna cut the video at 8:26 with a clap :D or what happened at there afterwards :D
Haha, yes, I was supposed to cut out the segment after the clap. Oops 😆
@@DataProfessor bloopers time :D
@@username42 It is 😅
Thank you!!
A pleasure!
it forces me to log in with my google account. I am afraid of google and this company having all my data.
I understand, you can also create a separate account on Gmail and use that as well.
Informative ..
Thanks for watching
why do you need such paid tool when you have free google colab with tpu and gpu ?
DataProfessor is showing off the free version. This comes with RAPIDS preinstalled and has the latest nightly builds along with stable builds (0.17 and 0.16) at the time of this comment. Colab is still on 0.14 and takes a signficant amount of time to install (10+ minutes). BlazingNotebooks takes seconds.
Thanks Felipe for the awesome answer, couldn't have said it better 😊
Thanks lot
Most welcome