3. Participation Readiness Check

📍 Where you are: Client 3/5 — ① Find · request → ② Workspace · data → ③ Readiness check → ④ Join → ⑤ Wrap-up
Screen: Agent Studio · Other side: meanwhile the Owner prepares the campaign

🎯 What you do in this step: Before connecting to the server, run everything on your PC in advance. It checks that the code you received is exactly the published Release, that your data matches the format, and that your model has the same shape as the other participants’ models. You can do this even while the server is off. If something breaks on your PC after rounds start, every other participant has to wait, so finish the check beforehand.

  1. 🖱️ Follow along
  2. ⚙️ Explained — badges and check items on the screen
  3. ✅ Checklist
  4. If something goes wrong
  5. Next

🖱️ Follow along

The Client's Federated Learning menu and the Heart Disease MLP card

  1. Click Federated Learning in the left menu.
  2. Click the Heart Disease MLP card. Before the check, the card shows not_configured and Needs Attention.

The Client's Check Participation Readiness button

  1. Click Check Participation Readiness on the Task Data row. It checks by training and evaluating the model on your local data, which takes longer than you might expect (about 5 min 40 s, measured on a Windows laptop on 2026-10-20). Keep waiting even if it stays at Evaluating local model 92% for a while.

Local Ready and Readiness details 9 / 9

It passes when ○ Check required changes to ✓ Local Ready and Readiness details · 9 / 9 appears below. Do not click Start Client yet. The Owner has to start the server (chapter 4).

⚙️ Explained — badges and check items on the screen

BadgeMeaning
○ Check required → ✓ Local ReadyWhether preparation on your PC (code, environment, data) is complete
● Server Live / Server OfflineWhether the Owner’s aggregation server is on. Independent of Local Ready

Expanding Readiness details shows per-item results. The nine items are:

ItemWhat it checksIf it fails
Federated Task linkedThis Workspace is linked to the Federated Task in the RegistryReopen it via Joined Registry Task as in ch. 2 step 1)
Participant Runtime compatibleYour Studio’s participant runtime fits this TaskUpdate Agent Studio to the latest version
Published Release installedYour code is exactly the Release the Owner publishedIf you changed code, revert it
Federated parameter contractYour model’s parameter names and shapes match the Release signatureTell the Owner (a Release problem)
Local dataset accessibledataset.csv exists in the Data Folder and can be readCheck the file name and location as in ch. 2 step 4)
Participation ReadinessThe check was done with the current code and the current dataIf you changed the data, check again
Python environment readyThe Sync in ch. 2 finishedRedo ch. 2 step 2)
Global Model availableThere is a Global Model to receive from the serverTell the Owner (check campaign preparation)
Participation authorizedThis account’s participation is approvedCheck your join request and the Owner’s approval in ch. 1

If you change the data, you must check again. The result is tied to the code and data at that moment, so changing files reverts it to Task Data or Workspace source changed.

The check is also reset when the Owner changes the campaign or starts the server. When a new campaign server starts, both the Owner and the Client go back to ○ Check required, and in that state Start Client cannot be clicked. If possible, check after the Owner has finished the campaign settings, and confirm ✓ Local Ready again before Start Client in chapter 4.

📚 Federated learning background — why check separately from the server

A federated learning round waits for the slowest participant. If one Client fails mid-round with “wrong column name”, the server either keeps waiting until enough participants gather or throws that round away. So FedOps separates “my side (Local)” from “server state (Server)” and has everyone finish beforehand everything they can on their own PC.

The parameter signature comparison is the same story as “why the model structure must be fixed” in Basics ch. 2. If even one shape differs, FedAvg cannot merge.

✅ Checklist

  • You saw ✓ Local Ready and Readiness details · 9 / 9
  • You told the Owner “I’m ready”

If something goes wrong

SymptomFix
Participation Readiness failsUsually dataset.csv is missing in the Data Folder or the column names differ. Look at the failed item in Readiness details
Value not in the vocabulary (cp, restecg)You edited dataset.csv by hand or put in other data. Rebuild it with prepare_data.py
The card is not visibleCheck that you were approved and that you opened the Workspace in ch. 2

Next

4. Join Federated Learning — start once the Owner starts the server.


Back to top

© 2026 Cognitive Computing Lab, Gachon University