How to retrieve stream accuracy after experience

[Mattdl]

I'm trying to save the evaluation results after an experience.
Right now I use: results.append(strategy.eval(scenario.test_stream[:]))

For 5-task Split-MNIST this gives me a sequence of experience (Task) accuracies, where only the last one should be considered with having the task-accuracies of all experiences:

However, I want to directly be able to access the stream accuracy, but in after_eval all the results are reset, so this isn't possible. Is there a way to still access the other evaluator plugin results, or is the only way to use a logger for this?

[vlomonaco]

Yes @Mattdl! Check out the eval_plugin.py example and let me know if this fix your issue.
Sorry I didn't make it clearer in the doc!

[Mattdl]

From the example, all_metrics = cl_strategy.evaluator.all_metrics gives indeed access to all the metrics! Thanks!