User: Show me the functions available in the system Parsed: function [E] User: What can this chat do for me? Parsed: function [E] User: What are the functions that you can perform? Parsed: function [E] User: Tell me about your functionality, please. Parsed: function [E] User: Could you introduce yourself? Parsed: self [E] User: Tell me about yourself Parsed: self [E] User: What is this conversational system about? Parsed: self [E] User: Show me the most important features in the dataset Parsed: important all [E] User: What are the features that have the highest importance? Parsed: important all [E] User: What are the features that have the highest importance? Parsed: important all [E] User: How do you predict that something has class {class_names}? Parsed: important {class_names} [E] User: What are the top 7 influential features? Parsed: important topk 7 [E] User: I'm curious what are the 5 most contributing tokens? Parsed: important topk 5 [E] User: What about 10 most contributing features? Parsed: important topk 10 [E] User: Can you show me 2 tokens that have the highest attributions? Parsed: important topk 2 [E] User: Show me some the top features for the predictions that the model makes Parsed: important all [E] User: I need an explanation for the model's prediction in terms of token attributions for the sample number 8. Parsed: filter id 8 and nlpattribute all [E] User: What are the features that are the most important ones for id 145. Parsed: filter id 145 and nlpattribute all [E] User: What are the top tokens contributing to the prediction for instance number 49. Parsed: filter id 49 and nlpattribute all [E] User: Can you explain to me how the model decides on the predictions for the instances with id 145 and id 122? Parsed: filter id 145 or filter id 122 and nlpattribute all [E] User: Show me please the most important five features for instances with id 100 and 102. Parsed: filter id 100 or filter id 102 and nlpattribute topk 5 [E] User: Explain please what the model predicts for id 89. Parsed: filter id 89 and nlpattribute all [E] User: Provide an elaborate explanation with rationalization for id 1000, please. Parsed: filter id 1000 and rationalize [E] User: What would be the rationale for the prediction that the model outputs for id 134? Parsed: filter id 134 and rationalize [E] User: Please rationalize for me the output for id 14. Parsed: filter id 14 and rationalize [E] User: What rationale can you generate for id 93? Parsed: filter id 93 and rationalize [E] User: Tell me how many datapoints are in the dataset? Parsed: countdata [E] User: How big are the data? Parsed: countdata [E] User: What is the size of the underlying dataset? Parsed: countdata [E] User: How many data points are there in total? Parsed: countdata [E] User: What data does the model use? Parsed: data [E] User: Can you tell me a bit more about the data used? Parsed: data [E] User: What's the data that this model is using? Parsed: data [E] User: I want to learn more about the model. Parsed: model [E] User: How can you describe the model? Parsed: model [E] User: What model do you use? Parsed: model [E] User: for all sentences that include the word "{span}", what are the true labels? Parsed: includes and label [E] User: Let's check the distribution of labels for the instances with {span}. Parsed: includes and label [E] User: What labels do we have? Parsed: label [E] User: Tell me about the dataset labels, please. Parsed: label [E] User: I want to see labels for all samples with "{span}" in them. Parsed: includes and label [E] User: Let's look at the sample number 11. Parsed: filter id 11 and show [E] User: Show me the instance with id 15, please. Parsed: filter id 15 and show [E] User: What about id 25, can you show it to me? Parsed: filter id 25 and show [E] User: Show me the most frequent keyword tokens, please. Parsed: keywords all [E] User: What are the keywords in the dataset? Parsed: keywords all [E] User: Top 5 keywords from the data would be good to have. Parsed: keywords 5 [E] User: Could you display 10 most frequent keywords? Parsed: keywords 10 [E] User: Please find me similar samples to id 10 and 112. Parsed: filter id 10 or filter id 112 and similar 1 [E] User: I need some samples that are similar to id 1400. Parsed: filter id 1400 and similar 1 [E] User: Show me other similar samples, I want to compare them to id 12. Parsed: filter id 12 and similar 1 [E] User: Top 7 most similar isntances to id 16, please. Parsed: filter id 16 and similar 7 [E] User: Can I see five similar instances for the instance with id 888? Parsed: filter id 888 and similar 5 [E] User: Any adversarial samples for id 67? Parsed: filter id 67 and adversarial [E] User: I would like to use adversarial training to generate new exaples. Could you show me some for id 53? Parsed: filter id 53 and adversarial [E] User: Adversarial attack on sample 115, please. Parsed: filter id 115 and adversarial [E] User: What would be an adversarial alternative for id 77? Parsed: filter id 77 and adversarial [E] User: Could you please augment instance 120 somehow? Parsed: filter id 120 and augment [E] User: Do you support data augmentation for instance 32? Parsed: filter id 32 and augment [E] User: I'd like to see an augmented version for id 84. Parsed: filter id 84 and augment [E] User: Can you generate a counterfactual example for id 36? Parsed: filter id 36 and cfe [E] User: I want to look at the counterfactuals for id 6. Parsed: filter id 6 and cfe [E] User: What cfes can you generate for id 257? Parsed: filter id 257 and cfe [E] User: How can we flip the prediction for instance 90? Parsed: filter id 90 and cfe [E] User: What is an alternative sample that fools the model? I mean something like a counterfactual for id 8. Parsed: filter id 8 and cfe [E] User: Tell me the probabilities for {class_names}. Parsed: likelihood [E] User: How likely it is to get the {class_names} label for id 15? Parsed: filter id 15 and likelihood [E] User: Do we have a high chance of getting {class_names} label for the instance number 900? Parsed: filter id 900 and likelihood [E] User: What are the likelihoods of different class labels for id 32? Parsed: filter id 32 and likelihood [E] User: Show me the likelihood for the following class: {class_names} given that we have id 222. Parsed: filter id 222 and likelihood [E] User: How many mistakes do we get here? Parsed: mistake count [E] User: Are the mistakes frequent? How often do they occur? Parsed: mistake count [E] User: I want to see the amount of the mistakes made by the model. Parsed: mistake count [E] User: What would be the misclassified samples? Parsed: mistake sample [E] User: Please show me some randomly picked mistakes made by the model. Parsed: mistake sample [E] User: Can you compute the number of mistakes? Parsed: mistake count [E] User: Can we see what are the wrong outputs/misclassified examples? Parsed: mistake sample [E] User: How many mistakes does the model make on the sentences with word {span}? Parsed: includes and mistake count [E] User: Let's count the mistakes among all cases that include {span}. Parsed: includes and mistake count [E] User: Show me some misclassified samples among the ones with "{span}". Parsed: includes and mistake sample [E] User: Please show some wrong predictions with "{span}". Parsed: includes and mistake sample [E] User: What are the predictions on the training data? Parsed: predict [E] User: What is the most likely prediction for id 47? Parsed: filter id 47 and predict [E] User: Let's look at the model's predictions then. Parsed: predict [E] User: What was predicted for id 99? Parsed: filter id 99 and predict [E] User: Id 780. What does the model predict? Parsed: filter id 780 predict [E] User: Please show me one example of the prediction that the model outputs. Parsed: randompredict [E] User: Ok, you can show me some random prediction. Parsed: randompredict [E] User: Just pick one sample and show me the prediction on it, the sample can be random. Parsed: randompredict [E] User: I would like to check the accuracy of the model. Parsed: score accuracy [E] User: Can you tell me how accurate the model is? Parsed: score accuracy [E] User: I would like to check the accuracy of the model. Parsed: score accuracy [E] User: Just show me some scores, please. Parsed: score default [E] User: What is the F1 score on the data? Parsed: score f1 [E] User: It would be nice to have the micro F1 score as well. Parsed: score f1 micro [E] User: What would be the macro scores in terms of precision? Parsed: score precision macro [E] User: Maybe you could also show me the weighted recall score? Parsed: score recall weighted [E] User: What would be the accuracy on the instances with {span}? Parsed: includes and score accuracy [E] User: Would be interesting to see also the sensitivity for all samples including {span}. Parsed: includes and score sensitivity [E] User: Now show me the micro-F1 scores. Parsed: score f1 micro [E]