Skip to content

fix: display eval status per metric type #305

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
stefanoamorelli wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix: display eval status per metric type #305

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
123 changes: 123 additions & 0 deletions src/app/components/eval-tab/eval-tab.component.spec.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -93,4 +93,127 @@ describe('EvalTabComponent', () => {
it('should create', () => {
expect(component).toBeTruthy();
});

describe('addEvalCaseResultToEvents', () => {
it('should show pass for tool events when tool_trajectory passes but response_match fails', () => {
const res = {
events: [
{author: 'user', content: {parts: [{text: 'hello'}]}},
{
author: 'bot',
content: {parts: [{functionCall: {name: 'test_tool', args: {}}}]},
},
{
author: 'bot',
content:
{parts: [{functionResponse: {name: 'test_tool', response: {}}}]},
},
{author: 'bot', content: {parts: [{text: 'final response'}]}},
],
};

const evalCaseResult = {
setId: 'test-set',
evalId: 'test-eval',
finalEvalStatus: 2,
evalMetricResults: [],
overallEvalMetricResults: [],
evalMetricResultPerInvocation: [
{
evalMetricResults: [
{
metricName: 'tool_trajectory_avg_score',
evalStatus: 1,
score: 1,
threshold: 1,
},
{
metricName: 'response_match_score',
evalStatus: 2,
score: 0,
threshold: 0.7,
},
],
actualInvocation: {
intermediateData: {toolUses: []},
finalResponse: {parts: [{text: 'actual'}]},
},
expectedInvocation: {
intermediateData: {toolUses: []},
finalResponse: {parts: [{text: 'expected'}]},
},
},
],
sessionId: 'test-session',
sessionDetails: {},
};

const result =
(component as any).addEvalCaseResultToEvents(res, evalCaseResult);

expect(result.events[1].evalStatus)
.toBe(1, 'functionCall event should pass');
expect(result.events[2].evalStatus)
.toBe(1, 'functionResponse event should pass');
expect(result.events[3].evalStatus)
.toBe(2, 'final text response should fail');
});

it('should show fail for tool events when tool_trajectory fails but response_match passes', () => {
const res = {
events: [
{author: 'user', content: {parts: [{text: 'hello'}]}},
{
author: 'bot',
content: {parts: [{functionCall: {name: 'test_tool', args: {}}}]},
},
{author: 'bot', content: {parts: [{text: 'final response'}]}},
],
};

const evalCaseResult = {
setId: 'test-set',
evalId: 'test-eval',
finalEvalStatus: 2,
evalMetricResults: [],
overallEvalMetricResults: [],
evalMetricResultPerInvocation: [
{
evalMetricResults: [
{
metricName: 'tool_trajectory_avg_score',
evalStatus: 2,
score: 0.5,
threshold: 1,
},
{
metricName: 'response_match_score',
evalStatus: 1,
score: 0.9,
threshold: 0.7,
},
],
actualInvocation: {
intermediateData: {toolUses: [{name: 'wrong_tool', args: {}}]},
finalResponse: {parts: [{text: 'actual'}]},
},
expectedInvocation: {
intermediateData: {toolUses: [{name: 'test_tool', args: {}}]},
finalResponse: {parts: [{text: 'expected'}]},
},
},
],
sessionId: 'test-session',
sessionDetails: {},
};

const result =
(component as any).addEvalCaseResultToEvents(res, evalCaseResult);

expect(result.events[1].evalStatus)
.toBe(2, 'functionCall event should fail');
expect(result.events[2].evalStatus)
.toBe(1, 'final text response should pass');
});
});
});
26 changes: 21 additions & 5 deletions src/app/components/eval-tab/eval-tab.component.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -324,16 +324,24 @@ export class EvalTabComponent implements OnInit, OnChanges {
currentInvocationIndex++;
} else {
const invocationResult = invocationResults[currentInvocationIndex];
const isToolEvent = this.isToolRelatedEvent(event);
const relevantMetric = isToolEvent ?
'tool_trajectory_avg_score' :
'response_match_score';

let evalStatus = 1;
let failedMetric = '';
let score = 1;
let threshold = 1;

for (const evalMetricResult of invocationResult.evalMetricResults) {
if (evalMetricResult.evalStatus === 2) {
evalStatus = 2;
failedMetric = evalMetricResult.metricName;
score = evalMetricResult.score;
threshold = evalMetricResult.threshold;
if (evalMetricResult.metricName === relevantMetric) {
evalStatus = evalMetricResult.evalStatus;
if (evalMetricResult.evalStatus === 2) {
failedMetric = evalMetricResult.metricName;
score = evalMetricResult.score;
threshold = evalMetricResult.threshold;
}
break;
}
}
Expand All @@ -350,6 +358,14 @@ export class EvalTabComponent implements OnInit, OnChanges {
return res;
}

private isToolRelatedEvent(event: any): boolean {
if (!event.content || !event.content.parts) {
return false;
}
return event.content.parts.some(
(part: any) => part.functionCall || part.functionResponse);
}

private addEvalFieldsToBotEvent(
event: any, invocationResult: any, failedMetric: string, score: number,
threshold: number) {
Expand Down