15:12:26 <farheen_att> #startmeeting Architecture Committee 15:12:26 <collabot`> Meeting started Wed Feb 5 15:12:26 2020 UTC. The chair is farheen_att. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:12:26 <collabot`> Useful Commands: #action #agreed #help #info #idea #link #topic. 15:12:26 <collabot`> The meeting name has been set to 'architecture_committee' 15:12:55 <farheen_att> #topic Federated Training 15:14:56 <farheen_att> #info Guy there may be overlap in the docker image creation. You don't need to rebuild a docker image to use the training part. 15:15:11 <farheen_att> #info Wenting we don't need to build but we need storage for version control. 15:15:34 <farheen_att> #info Sayee version control can be done in many ways. are we going to have 90 versions deployed? 15:16:04 <farheen_att> #info Sayee running model will have a wrapper to train the model delete the cache. 15:16:27 <farheen_att> #info W: you need to specify where the model is onboarded. Besides predict you have to have update. 15:16:50 <farheen_att> #info Sayee - if it is a rest API to call a model you will not have another version correct? 15:17:06 <farheen_att> #info S: do you do a post call to update the model? 15:17:41 <farheen_att> #info W: it doesn't have to be a REST api it can be handled by CI/CD. If the API is required then we need to be the API interface but we don't have that yet. 15:18:13 <farheen_att> #info S: we build the model outside and we pull the model based on the flag or setting. It will then restart all the pods to clear the cache. 15:18:35 <farheen_att> #info S: If you are going to train on predict as well as update. 15:18:52 <farheen_att> #info W: you want to generate a new ms? 15:19:03 <farheen_att> #info S: No you can re-use. 15:19:44 <farheen_att> #info Guy: PMML has a fairly clean separation between code and data. For a lot of our models the code and data is intertwined. Data is the working memory of the model. 15:20:35 <farheen_att> #info G: suppose you're training neural net model. After training you have weights, threshholds, and graphs. That object is the trained model in order to persist that you have to serialize that object into working memory. 15:20:51 <farheen_att> #info G: All the packages provide a way to do. We have to find a way to put them into Acumos. 15:21:15 <farheen_att> #info W: We are not going to re-do the MS we are going to do a hyper visor change. 15:21:59 <farheen_att> #info G: Changing the model architecture will be the same. Building a Random forest tree powers a predictor. You just have to load this data structure. 15:22:36 <farheen_att> #info W: Agree, we will not change the architecture. We have to have a mechanism to update this set of parameters. 15:23:13 <farheen_att> #info G: Parameters is tens of thousands of MB should not be a parameter. 15:24:55 <farheen_att> #info W: We will call it parameter until we decide what we want to call it. 15:30:10 <farheen_att> #info W: I agree we have to make use of the flow but we need the basic functionality. 15:30:21 <farheen_att> #info S: We need a flow or thought process. 15:31:01 <farheen_att> #info W: yes, I agree whether the bi-directional communication is based on supplier or subscriber. 15:31:31 <farheen_att> #info S: Bi-directional means when I have a model running then I want a container. 15:32:40 <farheen_att> #info W: info from model will be passed from subscriber to the supplier. The supplier is the creator of the model. 15:33:20 <farheen_att> #info Guy: What are the next steps? 15:33:53 <farheen_att> #info W: we can have broader discussion. 15:34:13 <farheen_att> #topic Demos 15:34:27 <farheen_att> #info there were 3 demos proposed for today. 15:34:36 <farheen_att> #info first is from portal and then licensing. 15:34:50 <farheen_att> #info Rajesh reschedule portal. 15:35:15 <farheen_att> #info We will have the OAM demo instead. 15:35:36 <farheen_att> #topic OA&M Demeter Sprint 1 demo 15:35:58 <farheen_att> #info demo being recorded and made available at wiki.acumos.org. 15:47:31 <farheen_att> #topic sprint management. 15:53:15 <farheen_att> #info topic Config issue for Acumos / Azure K8 deployment. 15:54:03 <farheen_att> #info Sayee asking for Ken's support 15:54:27 <farheen_att> #info Ken - We have been working on something similar on our setup. I will talk to Will later and include you. 15:54:34 <farheen_att> #info Sayee: That would be helpful. 15:55:53 <farheen_att> #topic Model Chaining RTU licenses in DS 15:56:34 <farheen_att> #info Sayee: There are two checks in DS. we need an API to list the model instead of calling by version, Alex will give us that RTU API. 15:57:23 <farheen_att> #topic Interworking with ONAP Jessica 16:00:30 <farheen_att> #info Jessica wants to showcase the interworking with ONAP and E2E. How stable is the interworking between Acumos. 16:00:48 <farheen_att> #endmeeting