Datasets:
roanbrasil
/
devops-gitops-corpus

Dataset card Data Studio Files
xet
 Community
1
text
stringlengths
0
59.1k
scorer: createModerationScorer({ model: moderationModel, threshold: 0.5 }),
},
keyword: {
scorer: keywordScorer,
params: { keyword: "refund" },
},
},
},
});
```
### LLM Judge for Helpfulness
```ts
import { Agent, buildScorer } from "@voltagent/core";
import { openai } from "@ai-sdk/openai";
import { z } from "zod";
const HELPFULNESS_SCHEMA = z.object({
score: z.number().min(0).max(1),
reason: z.string(),
});
const helpfulnessScorer = buildScorer({
id: "helpfulness",
label: "Helpfulness",
})
.score(async ({ payload }) => {
const agent = new Agent({
name: "helpfulness-judge",
model: openai("gpt-4o-mini"),
instructions: "You rate responses for helpfulness",
});
const prompt = `Rate the response for clarity, accuracy, and helpfulness.
User Input: ${payload.input}
Assistant Response: ${payload.output}
Provide a score from 0 to 1 with an explanation.`;
const response = await agent.generateObject(prompt, HELPFULNESS_SCHEMA);
const rawResults = (payload as any).results?.raw ?? {};
rawResults.helpfulnessJudge = response.object;
return {
score: response.object.score,
metadata: { reason: response.object.reason },
};
})
.reason(({ results }) => {
const judge = results.raw?.helpfulnessJudge as { reason?: string };
return { reason: judge?.reason ?? "No explanation provided." };
})
.build();
const agent = new Agent({
name: "support",
model: openai("gpt-4o"),
eval: {
sampling: { type: "ratio", rate: 0.1 }, // 10% sampling
scorers: {
helpfulness: { scorer: helpfulnessScorer },
},
},
});
```
### Multiple Scorers with Different Sampling
```ts
const agent = new Agent({
name: "support",
model: openai("gpt-4o"),
eval: {
triggerSource: "production",
environment: "prod-us-east",
sampling: { type: "ratio", rate: 1 }, // default: score everything
scorers: {
moderation: {
scorer: createModerationScorer({ model, threshold: 0.5 }),
sampling: { type: "ratio", rate: 1 }, // always run
},
answerCorrectness: {
scorer: createAnswerCorrectnessScorer(),
sampling: { type: "ratio", rate: 0.05 }, // 5% (expensive)
params: (payload) => ({
expectedAnswer: lookupExpectedAnswer(payload.input),
}),
},
keyword: {
scorer: keywordScorer,
params: { keyword: "refund" },
sampling: { type: "ratio", rate: 1 }, // cheap, always run
},
},
},
});
```