Hello Vedant, Thanks for the revision. I think that proposal is well structure. I have just two additional comments. The only part that I think is missing is the examples that the infrastructure is going to be employed. In order to build the GA infrastructure we have to test it in a simple example (such as cartpole). Thus, I recommend to describe an example that is going to be used in order to test the GA infrastructure. Regarding to GAs in DRL, we can focus on a simpler algorithm such as pytorch-neat from Uber research. If we build the GA infrastructure, pytorch-neat could be easily integrated Best regards, Manos. On 4/10/21 11:29 PM, Vedant Kandarp Shah wrote: > Sending this to the mailing list > -------------------------------------------------------------------------------------------------------------- > Hi Manos, > Thanks for the feedback on my previous draft. I have updated the > proposal keeping in mind your suggestions. Would be great if you can > go through it and give me your feedback. I have made the timeline to > the best of my understanding. However, if you feel that it is a bit > skewed with respect to the division of tasks, kindly let me know. > Thanks > > Regards, > Vedant > > On Sun, Apr 11, 2021 at 1:58 AM Vedant Kandarp Shah > <f20180566 [ at ] goa [ dot ] bits-pilani [ dot ] ac [ dot ] in > <mailto:f20180566 [ at ] goa [ dot ] bits-pilani [ dot ] ac [ dot ] in>> wrote: > > Hi Manos, > Thanks for the feedback on my previous draft. I have updated the > proposal keeping in mind your suggestions. Would be great if you > can go through it and give me your feedback. I have made the > timeline to the best of my understanding. However, if you feel > that it is a bit skewed with respect to the division of tasks, > kindly let me know. > Thanks > > Regards, > Vedant > > On Thu, Apr 8, 2021 at 4:16 PM Vedant Shah > <f20180566 [ at ] goa [ dot ] bits-pilani [ dot ] ac [ dot ] in > <mailto:f20180566 [ at ] goa [ dot ] bits-pilani [ dot ] ac [ dot ] in>> wrote: > > Hi Manos, > > Sure. I didn’t get any time to update the proposal since you > last reviewed it. However, I’ll update the draft today, > keeping in mind your suggestions and send it back to you for > feedback. > > > > Thanks, > > Vedant > > > > Sent from Mail > <https://go.microsoft.com/fwlink/?LinkId=550986> for Windows 10 > > > > *From: *Manos Kirtas <mailto:manolis [ dot ] kirt [ at ] gmail [ dot ] com> > *Sent: *08 April 2021 16:09 > *To: *Vedant Shah <mailto:f20180566 [ at ] goa [ dot ] bits-pilani [ dot ] ac [ dot ] in> > *Subject: *Re: [gsoc-developers] GSoC Projects : Deepbots > > > > Hi again, > > Additionally, you have to include a timeline on your proposal > that it will explain what you going to do in each week of the > given timeline > > Best regards, > > Manos. > > On 8/4/21 1:37 μ.μ., Manos Kirtas wrote: > > Hi Vedant, > > I have provided you a feedback on GDoC. Please revise your > draft proposal accordingly and contact me in order to > provide an up-to-dated feedback. Your proposal is well > structured, I feel that you need some clarifications about > what's your milestones > > Thank you, > > Manos. > > On 11/3/21 1:51 μ.μ., Vedant Shah wrote: > > Hi Manos, > > Thanks for the reply and thanks a lot for all the > useful tips on how to get started! I’ll start working > with Webots and Deepbots straight way and will > definitely get back to you for further discussion > regarding this. Looking forward to working on this > with you. > > > > Regards, > > Vedant > > > > Sent from Mail > <https://go.microsoft.com/fwlink/?LinkId=550986> for > Windows 10 > > > > *From: *Manos Kirtas <mailto:manolis [ dot ] kirt [ at ] gmail [ dot ] com> > *Sent: *11 March 2021 04:57 PM > *To: *Vedant Shah > <mailto:f20180566 [ at ] goa [ dot ] bits-pilani [ dot ] ac [ dot ] in>; > gsoc-developers [ at ] ellak [ dot ] gr <mailto:gsoc-developers [ at ] ellak [ dot ] gr> > *Subject: *Re: [gsoc-developers] GSoC Projects : Deepbots > > > > Hello Vedant, > > I am glad to hear that you are interested in > contributing to deepbots project! Indeed, your plan > seems that is the best way to be prepared for the GSoC > proposal phase. First of all, feel free to contact us > via email for any help or open a discussion at > project's github page. In order to be familiar with > webots and deepbots you can try to run and reproduce > the examples included at deepworlds repository. There > are extensive tutorials in deepbots-tutorials > repository. If you find a bug or develop a fix you are > welcome to open an issue or PR respectively. After all > you can checkout the deepbots' issues which are tagged > with 'ideal first issue' in order to be familiar with > the codebase. > > Feel free to contact us for help. You can join GFOSS > <https://matrix.to/#/%23gsoc2021:chat.ellak.gr?via=chat.ellak.gr&via=matrix.org> > matrix channel as well. > > Best regards, > > Manos. > > > > On 11/3/21 5:22 π.μ., Vedant Shah wrote: > > Hello everyone, > > I am Vedant Shah, a pre-final year undergraduate > student at BITS Pilani, India. I am primarily > interested in two projects hosted by GFOSS for > GSoC 2021: “Extend Deepbots to support > stable-baseline and implement gym style default > Reinforcement Learning Experiments” and “Extend > Deepbots to support evolutionary algorithms”. > > > > A little bit on my background: I have been working > in the fields of deep learning and (especially) > reinforcement learning for quite some time now and > have worked on / am working on several related > projects and research. As a part of these, I have > significant experience with working with deep > learning frameworks like PyTorch, tensorflow, etc. > and also with tools specific to reinforcement > learning such as OpenAI gym, stable-baselines, > rllib, etc. I am also a core-contributors to a > pytorch first reinforcement learning library. A > part from reinforcement learning, I have also > worked in robotics on tasks such as navigation, > motion planning and control. > > > > I have always been interested in real-world > reinforcement learning and specially it’s > applications in robotics. IMy avid interests in > real-world RL is the main driving factor behind > be choosing these projects (I’ll be happy to work > on any of these). Although I am not familiar with > Webots **yet** it seems like a really actionable > platform for testing out algorithms in context of > robotics. I have grown familiar to deepbots over > the past couple of days and am pretty confident I > can pull (any of) these two projects off. > > > > Currently I have a very rough plan of action in > mind which includes > > > > 1. Getting comfortable with Webots (can be done > before the actual work period begins) > 2. Researching upon some relevant evolutionary > algorithms (for eg. which have been proved to > be robust to noisy data) which can be used in > context of robotics > 3. Start implementing those > > > > For the stable-baselines project, I am already > familiar and have used stable-baselines before and > directly jump to integrating it once I am familiar > Webots (again, this can be done before the actual > work period begins). I am aware that these plan of > action is not very detailed and is very unrefined. > So I’d like to get in touch with the mentors Manos > Kirtas, Konstantinos Tsampazis, Nikolaos Passalis > to discuss the project further. Looking forward to > hearing from you soon! > > > > Regards, > > Vedant > > > > Sent from Mail > <https://go.microsoft.com/fwlink/?LinkId=550986> > for Windows 10 > > > > > > > > ---- > > Λαμβάνετε αυτό το μήνυμα απο την λίστα: Λίστα αλληλογραφίας και συζητήσεων που απευθύνεται σε φοιτητές developers \& mentors έργων του Google Summer of Code - A discussion list for student developers and mentors of Google Summer of Code projects., > > https://lists.ellak.gr/gsoc-developers/listinfo.html <https://lists.ellak.gr/gsoc-developers/listinfo.html> > > Μπορείτε να απεγγραφείτε από τη λίστα στέλνοντας κενό μήνυμα ηλ. ταχυδρομείου στη διεύθυνση <gsoc-developers+unsubscribe [ at ] ellak [ dot ] gr> <mailto:gsoc-developers+unsubscribe [ at ] ellak [ dot ] gr>. > > > > >
---- Λαμβάνετε αυτό το μήνυμα απο την λίστα: Λίστα αλληλογραφίας και συζητήσεων που απευθύνεται σε φοιτητές developers \& mentors έργων του Google Summer of Code - A discussion list for student developers and mentors of Google Summer of Code projects., https://lists.ellak.gr/gsoc-developers/listinfo.html Μπορείτε να απεγγραφείτε από τη λίστα στέλνοντας κενό μήνυμα ηλ. ταχυδρομείου στη διεύθυνση <gsoc-developers+unsubscribe [ at ] ellak [ dot ] gr>.