|
| 1 | +--- |
| 2 | +layout: post |
| 3 | +title: "Talks – March, 2021" |
| 4 | +author: rahul |
| 5 | +tags: [] |
| 6 | +description: "In collaboration with PythonPune" |
| 7 | +categories: [ meetup, talks, workshop ] |
| 8 | +image: |
| 9 | +featured: true |
| 10 | +hidden: false |
| 11 | +--- |
| 12 | + |
| 13 | + |
| 14 | +This edition of the meetup was presented in collaboration with [PythonPune](https://pythonpune.in/), a group working with an aim of "Bridging the gap between what people work on and want to learn". The meetup saw an attendance of around 50 people with the session starting sharp at 10:30 am and seeing active participation from the attendees through all three talks that were lined up. |
| 15 | + |
| 16 | + |
| 17 | + |
| 18 | +The first of the talks was by Kanishk who had previously presented a session at our meetup in November of 2020. Considering the response from that session and the number of queries that came up in the session then, Kanishk found the need to bridge this gap and so will be presenting a three-part workshop series in our meetups. This session aptly named "Machine Learning in action" was a precursor to what's in store in the upcoming workshop session. |
| 19 | + |
| 20 | +The session was intended to take the audience through the different solutions that ML has enabled us with. We were taken through what ML was and the goal of ML define as-"The goal is to make the guesses good enough to be useful and never to make them perfect guesses". Kanishk took us through the expanse of the types of ML out there with detailed use-case discussion for each. These types included: |
| 21 | +- Supervised Learning |
| 22 | +- Unsupervised Learning |
| 23 | +- Reinforcement Learning |
| 24 | +- Hybrid Learning |
| 25 | +- Statistical Learning |
| 26 | + |
| 27 | +We also got to know about the different learning techniques which ranged from Multi-Task Learning to Ensemble Learning with mentions of Active, Online Learning. The part of the session that put everyone in awe was when the applications of ML were demoed, these demos included: |
| 28 | +- Image classification through CIFAR-10 model |
| 29 | +- Application of GANs |
| 30 | + - Grainy image correction. |
| 31 | + - [Nvidia's GauGAN](http://nvidia-research-mingyuliu.com/gaugan/) interactive demo. |
| 32 | + - The real person like images from [This person does not exist](https://thispersondoesnotexist.com/). |
| 33 | + - Image to image translations with [pipx2pix model](https://phillipi.github.io/pix2pix/). |
| 34 | + - Google's [quick draw](https://quickdraw.withgoogle.com/). |
| 35 | + - Interactive sketch prediction. |
| 36 | +- Application of NLP through the use of GPT-2 APIs |
| 37 | + |
| 38 | +The demo packed talk session was followed by 2 lightning talks. The first of which was presented by Bangpyper's veteran member [Kracekumar](https://kracekumar.com/). He spoke about a tool that he built over a weekend to render Jupyter Notebook content on terminals called JUT. The need for this solution was stated as the inconvenience of always having to rely on browsers and editors to view them. The approach that was taken here to render notebooks on terminals was to use the JSON source of .ipynb files to parse using the available schema and display them on the terminal. Karce took us through the different technical hurdles that he had in building this solution and also listed the features he has planned to integrate. JUT is open for collaborations and suggestions, find the project here: [https://github.com/kracekumar/jut](https://github.com/kracekumar/jut) |
| 39 | + |
| 40 | + |
| 41 | + |
| 42 | +The second lightning talk was from [Tejas Ravishankar](mailto:xtremedevx@gmail.com), a 14 year old aspiring DevOps engineer who presented his package manager for Windows OS- Electric, which's around 5x faster than Chocolatey the currently available prevalent package manger. What led Tejas to build this tool was his frustration with how slow Chocolatey was and the need to pay for the features. The demo was very detailed taking us through the different features where Electric shone over Chocolatey be it with its ability to handle concurrent installations, Suggestive error logging, Software bundling and configuration generation and distribution and many more features. This project truly represents the spirit of Free and Open-source project. You can find this promising project here: [https://github.com/electric-package-manager/electric](https://github.com/electric-package-manager/electric) |
| 43 | + |
| 44 | + |
| 45 | + |
| 46 | +This was followed by the announcement of the following: |
| 47 | +- [PyCon India 2021](https://in.pycon.org/2021/) , the four day conference from the 17th through the 20th of September. |
| 48 | + - To volunteer please check the 'Call for volunteers' here: [https://in.pycon.org/blog/2021/call-for-volunteers.html](https://in.pycon.org/blog/2021/call-for-volunteers.html) |
| 49 | + - If you are interested in presenting at PyCon check the 'Call for proposals' here: [https://in.pycon.org/blog/2021/call-for-volunteers.html](https://in.pycon.org/blog/2021/call-for-volunteers.html) |
| 50 | +- Newsletter run by PythonPune which cover about python tools and what's latest in the world of python, you can subscribe to them here: [https://groups.google.com/forum/#!forum/pythonpune](https://groups.google.com/forum/#!forum/pythonpune) |
| 51 | + |
| 52 | + |
| 53 | + |
| 54 | +That brought a close to this month's edition of Bangpypers meetup and this collaboration with PythonPune was a great hit. |
| 55 | + |
| 56 | + |
| 57 | + |
| 58 | +Missed the workshop? No worries, you can catch up completely with the notes listed here - [https://github.com/bangpypers/20-3-2021-Talks](https://github.com/bangpypers/20-3-2021-Talks) |
0 commit comments