Hey - I was wondering if I can get more details on what should be included in the writeup and the survey? When are they due? Is the write up just using the wiki on the project page? Thanks so much! Sean

Created by Sean McCurdy seanmccrdy
thanks @Michael.Mason @vchung for the replies!
Dear @seanmccrdy , You don't need to include your whole dataset if it is particularly large but a description of it and links to where it can be found should be inlcuded. If you did anything to reformat, normalized or harmonize the data then that code should included in your src folder. Let us know if that makes sense. Cheers, Mike
@seanmccrdy , By "create", I mean that `src` should contain all source files that will re-build your Docker submissions, that is, all files that will build the images based on your Dockerfile(s). I hope this clarifies your confusion!
Hi, Just wanted to add a small note about the project sharing setting, Access for all registered Synapse users should be "can download" and anyone in the web "can view" Regards,
@vchung I'm a little confused by your message above. Can you clarify what you mean by create? Are you referring to creating the docker file to run model inference? Or are you expecting to have a full ML pipeline containing model training, training data, and model inference in the `src` files? I included my Colab notebook in my write up (which is how I trained the model), along with features, and datasets used. Is that ok? I'm asking because I'm pretty sure that my full training data set will exceed the file size limit. sean
@Eimanahmed , Great question. We ask that you please provide all source files necessary for re-creating your model(s); this is so we can ensure of their reproducibility. However, if you have sensitive data, then it will be okay to omit them from the `src` folder. Instead, we ask that you please describe the dataset(s) you used, including all features that you are permitted to share. @Michael.Mason has provided more details regarding sensitive data [here](https://www.synapse.org/#!Synapse:syn18404605/discussion/threadId=7825&replyId=24338). I hope this helps!
Can we just provide the docker container we have submitted for each subchallenge which has all the required input files in it. Will write instructions in README on how to run it . Thank you,
@adeshina , Apologies for the delay! Your Synapse ID is what is in front of the "@synapse.org" email. Essentially, it is your username.
I couldn't find my synapse Id on my profile page, where else can I get it?
@seanmccrdy , That should be okay! If you rather not make any changes to your Dockerfile, you can alternatively specify in the README that one should mount a /output directory during the container run, e.g. ``` ### Build Docker image. `docker build -t my-model:v1` ### Run the image with a mounted output directory. `docker run -v $PWD/output/:/output/:rw my-model:v1` ``` @adeoluokiki , Yes, you can re-submit -- there is no submission limit for the writeup evaluation queue. Much like the sub-challenges, you will receive an email for a valid/invalid writeup.
I had previously submitted my writeup without making the sharing setting public. Can I set the writeup's Sharing Settings to Public and re-submit again? Secondly, How do I know if my writeup is successfully submitted/validly submitted? Thanks
Thanks @vchung! I have to slightly change my Dockerfile so that the file can write locally to an output/ folder. Is that be ok? The model used in the challenge will be the same, of course
@seanmccrdy , Great question! And thank you for the feedback; we will definitely try to supply a more explicit set of expectations for next time. Essentially, we are asking that you provide everything one would need in order to reproduce your model, that is, the Dockerfile used to build the image, as well as all of its dependencies, such as input files, scripts, etc. For example, let's say your Dockerfile is something like this: ```python FROM python:3.9.1-slim-buster COPY model.py /usr/local/bin/model.py COPY run.sh /run.sh RUN chmod 775 /usr/local/bin/model.py RUN chmod 775 /run.sh ENTRYPOINT ["/bin/bash", "/run.sh"] ``` There are two scripts required to build the image: model.py and run.sh. Therefore, `src` should contain this Dockerfile, as well as the model.py and run.sh scripts. In addition to all of the source materials, we also ask that you provide some sort of README on how to build and run that model. If you had built a different model for each sub-challenge, then please create a sub-directory within `src` for each sub-challenge, where you will store the applicable Dockerfile and related files/scripts needed for that particular model. I hope this helps!
Hi @vchung , I'm still not clear on what exactly is the src folder? Is it the same thing as the docker folder, except renamed to src? How are they different? Does src platform include training model pipeline or our training data? It would be helpful if you could provide an example submission from a previous competition or more explicit instructions. My guess is that: 1) We go to our projects and create a folder called src in the FILES tab 2) We upload the folder [call it subchallengex] containing the docker files (DockerFile, python/R run file, saved model file/folder) 3) And then provide some instruction (maybe add like a readme file?) to give instructions like: * i) Download src folder from my project * ii) Go in to terminal: mount inference data into data folder: ~/src/subchallengex/data/ * iii) $ cd ~/src/subchallengex * iv) $ docker build -t subchallengex_model * v) $ docker run subchallengex_model * vi) Predictions will be found in output/predictions.csv Is that sufficient? Thanks again! Sean
Many thanks ! @vchung
@WenyuWang , @kongpsx , To be considered for consortium-level authorship, the writeup you submit must be accessible to the public, that is, anyone is able to view it. Ergo, when you are ready to submit, first check that your writeup's Sharing Settings is set to Public. For more details on Sharing Settings, go [here](https://docs.synapse.org/articles/sharing_settings.html). Regarding your `src` folder, yes! Please prepare your `src` folder for items related to the last submitted models in each sub-challenge. Thank you both for participating in the Challenge!
I'd like clarification on the folder availability as well. I tried to submit my writeup, but I got the following message: "Hello kongpsx, Your submission is invalid, below are the invalid reasons: Your project is not publicly available. Sincerely, Challenge Administrator" I did not have this issue when I submitted the same project during the validation phase. Can someone please confirm what needs to be done? Thanks!
Hi, Some additional question regarding the writeup phase. 1 about src folder We noticed that in the Q&A: Which of the two submissions are considered for final evaluation during the Validation Phase? We will consider the last submitted model as your final submission, not the best-performing model of the five. Does this mean to prepare the src folder for write-up submission, we only need to include the files related to the last submitted model in each subchallenge? 2 about project sharing setting. Currently, the project is only available to our team. But after the final writeup submission, we are expected to also make the folder available to organizers, right? Thanks!
Thanks @Michael.Mason !
Dear @seanmccrdy , [Here is a Challenge writeup template.](https://www.synapse.org/#!Synapse:syn2351771/wiki/62369). There are also some nice examples linked in it. The write up is due March 5th. The survey will be sent out Challenge organizers early next week. Hope that helps, Mike

More details on Writeup and survey? page is loading…