@mastra/evals 1.1.0 → 1.1.1-alpha.0

package/dist/docs/references/reference-evals-context-relevance.md

@@ -0,0 +1,536 @@
+# Context Relevance Scorer
+
+The `createContextRelevanceScorerLLM()` function creates a scorer that evaluates how relevant and useful provided context was for generating agent responses. It uses weighted relevance levels and applies penalties for unused high-relevance context and missing information.
+
+It is especially useful for these use cases:
+
+**Content Generation Evaluation**
+
+Best for evaluating context quality in:
+
+- Chat systems where context usage matters
+- RAG pipelines needing nuanced relevance assessment
+- Systems where missing context affects quality
+
+**Context Selection Optimization**
+
+Use when optimizing for:
+
+- Comprehensive context coverage
+- Effective context utilization
+- Identifying context gaps
+
+## Parameters
+
+**model:** (`MastraModelConfig`): The language model to use for evaluating context relevance
+
+**options:** (`ContextRelevanceOptions`): Configuration options for the scorer
+
+Note: Either `context` or `contextExtractor` must be provided. If both are provided, `contextExtractor` takes precedence.
+
+## .run() Returns
+
+**score:** (`number`): Weighted relevance score between 0 and scale (default 0-1)
+
+**reason:** (`string`): Human-readable explanation of the context relevance evaluation
+
+## Scoring Details
+
+### Weighted Relevance Scoring
+
+Context Relevance uses a sophisticated scoring algorithm that considers:
+
+1. **Relevance Levels**: Each context piece is classified with weighted values:
+
+   - `high` = 1.0 (directly addresses the query)
+   - `medium` = 0.7 (supporting information)
+   - `low` = 0.3 (tangentially related)
+   - `none` = 0.0 (completely irrelevant)
+
+2. **Usage Detection**: Tracks whether relevant context was actually used in the response
+
+3. **Penalties Applied** (configurable via `penalties` options):
+
+   - **Unused High-Relevance**: `unusedHighRelevanceContext` penalty per unused high-relevance context (default: 0.1)
+   - **Missing Context**: Up to `maxMissingContextPenalty` for identified missing information (default: 0.5)
+
+### Scoring Formula
+
+```text
+Base Score = Σ(relevance_weights) / (num_contexts × 1.0)
+Usage Penalty = count(unused_high_relevance) × unusedHighRelevanceContext
+Missing Penalty = min(count(missing_context) × missingContextPerItem, maxMissingContextPenalty)
+
+Final Score = max(0, Base Score - Usage Penalty - Missing Penalty) × scale
+```
+
+**Default Values**:
+
+- `unusedHighRelevanceContext` = 0.1 (10% penalty per unused high-relevance context)
+- `missingContextPerItem` = 0.15 (15% penalty per missing context item)
+- `maxMissingContextPenalty` = 0.5 (maximum 50% penalty for missing context)
+- `scale` = 1
+
+### Score interpretation
+
+- **0.9-1.0**: Excellent - all context highly relevant and used
+- **0.7-0.8**: Good - mostly relevant with minor gaps
+- **0.4-0.6**: Mixed - significant irrelevant or unused context
+- **0.2-0.3**: Poor - mostly irrelevant context
+- **0.0-0.1**: Very poor - no relevant context found
+
+### Reason analysis
+
+The reason field provides insights on:
+
+- Relevance level of each context piece (high/medium/low/none)
+- Which context was actually used in the response
+- Penalties applied for unused high-relevance context (configurable via `unusedHighRelevanceContext`)
+- Missing context that would have improved the response (penalized via `missingContextPerItem` up to `maxMissingContextPenalty`)
+
+### Optimization strategies
+
+Use results to improve your system:
+
+- **Filter irrelevant context**: Remove low/none relevance pieces before processing
+- **Ensure context usage**: Make sure high-relevance context is incorporated
+- **Fill context gaps**: Add missing information identified by the scorer
+- **Balance context size**: Find optimal amount of context for best relevance
+- **Tune penalty sensitivity**: Adjust `unusedHighRelevanceContext`, `missingContextPerItem`, and `maxMissingContextPenalty` based on your application's tolerance for unused or missing context
+
+### Difference from Context Precision
+
+| Aspect        | Context Relevance                      | Context Precision                  |
+| ------------- | -------------------------------------- | ---------------------------------- |
+| **Algorithm** | Weighted levels with penalties         | Mean Average Precision (MAP)       |
+| **Relevance** | Multiple levels (high/medium/low/none) | Binary (yes/no)                    |
+| **Position**  | Not considered                         | Critical (rewards early placement) |
+| **Usage**     | Tracks and penalizes unused context    | Not considered                     |
+| **Missing**   | Identifies and penalizes gaps          | Not evaluated                      |
+
+## Scorer configuration
+
+### Custom penalty configuration
+
+Control how penalties are applied for unused and missing context:
+
+```typescript
+import { createContextRelevanceScorerLLM } from "@mastra/evals";
+
+// Stricter penalty configuration
+const strictScorer = createContextRelevanceScorerLLM({
+  model: "openai/gpt-5.1",
+  options: {
+    context: [
+      "Einstein won the Nobel Prize for photoelectric effect",
+      "He developed the theory of relativity",
+      "Einstein was born in Germany",
+    ],
+    penalties: {
+      unusedHighRelevanceContext: 0.2, // 20% penalty per unused high-relevance context
+      missingContextPerItem: 0.25, // 25% penalty per missing context item
+      maxMissingContextPenalty: 0.6, // Maximum 60% penalty for missing context
+    },
+    scale: 1,
+  },
+});
+
+// Lenient penalty configuration
+const lenientScorer = createContextRelevanceScorerLLM({
+  model: "openai/gpt-5.1",
+  options: {
+    context: [
+      "Einstein won the Nobel Prize for photoelectric effect",
+      "He developed the theory of relativity",
+      "Einstein was born in Germany",
+    ],
+    penalties: {
+      unusedHighRelevanceContext: 0.05, // 5% penalty per unused high-relevance context
+      missingContextPerItem: 0.1, // 10% penalty per missing context item
+      maxMissingContextPenalty: 0.3, // Maximum 30% penalty for missing context
+    },
+    scale: 1,
+  },
+});
+
+const testRun = {
+  input: {
+    inputMessages: [
+      {
+        id: "1",
+        role: "user",
+        content: "What did Einstein achieve in physics?",
+      },
+    ],
+  },
+  output: [
+    {
+      id: "2",
+      role: "assistant",
+      content:
+        "Einstein won the Nobel Prize for his work on the photoelectric effect.",
+    },
+  ],
+};
+
+const strictResult = await strictScorer.run(testRun);
+const lenientResult = await lenientScorer.run(testRun);
+
+console.log("Strict penalties:", strictResult.score); // Lower score due to unused context
+console.log("Lenient penalties:", lenientResult.score); // Higher score, less penalty
+```
+
+### Dynamic Context Extraction
+
+```typescript
+const scorer = createContextRelevanceScorerLLM({
+  model: "openai/gpt-5.1",
+  options: {
+    contextExtractor: (input, output) => {
+      // Extract context based on the query
+      const userQuery = input?.inputMessages?.[0]?.content || "";
+      if (userQuery.includes("Einstein")) {
+        return [
+          "Einstein won the Nobel Prize for the photoelectric effect",
+          "He developed the theory of relativity",
+        ];
+      }
+      return ["General physics information"];
+    },
+    penalties: {
+      unusedHighRelevanceContext: 0.15,
+    },
+  },
+});
+```
+
+### Custom scale factor
+
+```typescript
+const scorer = createContextRelevanceScorerLLM({
+  model: "openai/gpt-5.1",
+  options: {
+    context: ["Relevant information...", "Supporting details..."],
+    scale: 100, // Scale scores from 0-100 instead of 0-1
+  },
+});
+
+// Result will be scaled: score: 85 instead of 0.85
+```
+
+### Combining multiple context sources
+
+```typescript
+const scorer = createContextRelevanceScorerLLM({
+  model: "openai/gpt-5.1",
+  options: {
+    contextExtractor: (input, output) => {
+      const query = input?.inputMessages?.[0]?.content || "";
+
+      // Combine from multiple sources
+      const kbContext = knowledgeBase.search(query);
+      const docContext = documentStore.retrieve(query);
+      const cacheContext = contextCache.get(query);
+
+      return [...kbContext, ...docContext, ...cacheContext];
+    },
+    scale: 1,
+  },
+});
+```
+
+## Examples
+
+### High relevance example
+
+This example shows excellent context relevance where all context directly supports the response:
+
+```typescript
+import { createContextRelevanceScorerLLM } from "@mastra/evals";
+
+const scorer = createContextRelevanceScorerLLM({
+  model: "openai/gpt-5.1",
+  options: {
+    context: [
+      "Einstein won the Nobel Prize for his discovery of the photoelectric effect in 1921.",
+      "He published his theory of special relativity in 1905.",
+      "His general relativity theory, published in 1915, revolutionized our understanding of gravity.",
+    ],
+    scale: 1,
+  },
+});
+
+const result = await scorer.run({
+  input: {
+    inputMessages: [
+      {
+        id: "1",
+        role: "user",
+        content: "What were Einstein's major scientific achievements?",
+      },
+    ],
+  },
+  output: [
+    {
+      id: "2",
+      role: "assistant",
+      content:
+        "Einstein's major achievements include the Nobel Prize for the photoelectric effect, special relativity in 1905, and general relativity in 1915.",
+    },
+  ],
+});
+
+console.log(result);
+// Output:
+// {
+//   score: 1.0,
+//   reason: "The score is 1.0 because all context pieces are highly relevant to Einstein's achievements and were effectively used in generating the comprehensive response."
+// }
+```
+
+### Mixed relevance example
+
+This example shows moderate relevance with some context being irrelevant or unused:
+
+```typescript
+import { createContextRelevanceScorerLLM } from "@mastra/evals";
+
+const scorer = createContextRelevanceScorerLLM({
+  model: "openai/gpt-5.1",
+  options: {
+    context: [
+      "Solar eclipses occur when the Moon blocks the Sun.",
+      "The Moon moves between the Earth and Sun during eclipses.",
+      "The Moon is visible at night.",
+      "Stars twinkle due to atmospheric interference.",
+      "Total eclipses can last up to 7.5 minutes.",
+    ],
+    scale: 1,
+  },
+});
+
+const result = await scorer.run({
+  input: {
+    inputMessages: [
+      {
+        id: "1",
+        role: "user",
+        content: "What causes solar eclipses?",
+      },
+    ],
+  },
+  output: [
+    {
+      id: "2",
+      role: "assistant",
+      content:
+        "Solar eclipses happen when the Moon moves between Earth and the Sun, blocking sunlight.",
+    },
+  ],
+});
+
+console.log(result);
+// Output with default penalties:
+// {
+//   score: 0.64,
+//   reason: "The score is 0.64 because contexts 1 and 2 are highly relevant and used, context 5 is relevant but unused (10% penalty), while contexts 3 and 4 are irrelevant."
+// }
+
+// With custom penalty configuration
+const customScorer = createContextRelevanceScorerLLM({
+  model: "openai/gpt-5.1",
+  options: {
+    context: [
+      "Solar eclipses occur when the Moon blocks the Sun.",
+      "The Moon moves between the Earth and Sun during eclipses.",
+      "The Moon is visible at night.",
+      "Stars twinkle due to atmospheric interference.",
+      "Total eclipses can last up to 7.5 minutes.",
+    ],
+    penalties: {
+      unusedHighRelevanceContext: 0.05, // Lower penalty for unused context
+      missingContextPerItem: 0.1,
+      maxMissingContextPenalty: 0.3,
+    },
+  },
+});
+
+const customResult = await customScorer.run({
+  input: {
+    inputMessages: [
+      { id: "1", role: "user", content: "What causes solar eclipses?" },
+    ],
+  },
+  output: [
+    {
+      id: "2",
+      role: "assistant",
+      content:
+        "Solar eclipses happen when the Moon moves between Earth and the Sun, blocking sunlight.",
+    },
+  ],
+});
+
+console.log(customResult);
+// Output with lenient penalties:
+// {
+//   score: 0.69, // Higher score due to reduced penalty for unused context
+//   reason: "The score is 0.69 because contexts 1 and 2 are highly relevant and used, context 5 is relevant but unused (5% penalty), while contexts 3 and 4 are irrelevant."
+// }
+```
+
+### Low relevance example
+
+This example shows poor context relevance with mostly irrelevant information:
+
+```typescript
+import { createContextRelevanceScorerLLM } from "@mastra/evals";
+
+const scorer = createContextRelevanceScorerLLM({
+  model: "openai/gpt-5.1",
+  options: {
+    context: [
+      "The Great Barrier Reef is located in Australia.",
+      "Coral reefs need warm water to survive.",
+      "Many fish species live in coral reefs.",
+      "Australia has six states and two territories.",
+      "The capital of Australia is Canberra.",
+    ],
+    scale: 1,
+  },
+});
+
+const result = await scorer.run({
+  input: {
+    inputMessages: [
+      {
+        id: "1",
+        role: "user",
+        content: "What is the capital of Australia?",
+      },
+    ],
+  },
+  output: [
+    {
+      id: "2",
+      role: "assistant",
+      content: "The capital of Australia is Canberra.",
+    },
+  ],
+});
+
+console.log(result);
+// Output:
+// {
+//   score: 0.26,
+//   reason: "The score is 0.26 because only context 5 is relevant to the query about Australia's capital, while the other contexts about reefs are completely irrelevant."
+// }
+```
+
+### Dynamic context extraction
+
+Extract context dynamically based on the run input:
+
+```typescript
+import { createContextRelevanceScorerLLM } from "@mastra/evals";
+
+const scorer = createContextRelevanceScorerLLM({
+  model: "openai/gpt-5.1",
+  options: {
+    contextExtractor: (input, output) => {
+      // Extract query from input
+      const query = input?.inputMessages?.[0]?.content || "";
+
+      // Dynamically retrieve context based on query
+      if (query.toLowerCase().includes("einstein")) {
+        return [
+          "Einstein developed E=mc²",
+          "He won the Nobel Prize in 1921",
+          "His theories revolutionized physics",
+        ];
+      }
+
+      if (query.toLowerCase().includes("climate")) {
+        return [
+          "Global temperatures are rising",
+          "CO2 levels affect climate",
+          "Renewable energy reduces emissions",
+        ];
+      }
+
+      return ["General knowledge base entry"];
+    },
+    penalties: {
+      unusedHighRelevanceContext: 0.15, // 15% penalty for unused relevant context
+      missingContextPerItem: 0.2, // 20% penalty per missing context item
+      maxMissingContextPenalty: 0.4, // Cap at 40% total missing context penalty
+    },
+    scale: 1,
+  },
+});
+```
+
+### RAG system integration
+
+Integrate with RAG pipelines to evaluate retrieved context:
+
+```typescript
+import { createContextRelevanceScorerLLM } from "@mastra/evals";
+
+const scorer = createContextRelevanceScorerLLM({
+  model: "openai/gpt-5.1",
+  options: {
+    contextExtractor: (input, output) => {
+      // Extract from RAG retrieval results
+      const ragResults = inputData.metadata?.ragResults || [];
+
+      // Return the text content of retrieved documents
+      return ragResults
+        .filter((doc) => doc.relevanceScore > 0.5)
+        .map((doc) => doc.content);
+    },
+    penalties: {
+      unusedHighRelevanceContext: 0.12, // Moderate penalty for unused RAG context
+      missingContextPerItem: 0.18, // Higher penalty for missing information in RAG
+      maxMissingContextPenalty: 0.45, // Slightly higher cap for RAG systems
+    },
+    scale: 1,
+  },
+});
+
+// Evaluate RAG system performance
+const evaluateRAG = async (testCases) => {
+  const results = [];
+
+  for (const testCase of testCases) {
+    const score = await scorer.run(testCase);
+    results.push({
+      query: testCase.inputData.inputMessages[0].content,
+      relevanceScore: score.score,
+      feedback: score.reason,
+      unusedContext: score.reason.includes("unused"),
+      missingContext: score.reason.includes("missing"),
+    });
+  }
+
+  return results;
+};
+```
+
+## Comparison with Context Precision
+
+Choose the right scorer for your needs:
+
+| Use Case                 | Context Relevance    | Context Precision         |
+| ------------------------ | -------------------- | ------------------------- |
+| **RAG evaluation**       | When usage matters   | When ranking matters      |
+| **Context quality**      | Nuanced levels       | Binary relevance          |
+| **Missing detection**    | ✓ Identifies gaps    | ✗ Not evaluated           |
+| **Usage tracking**       | ✓ Tracks utilization | ✗ Not considered          |
+| **Position sensitivity** | ✗ Position agnostic  | ✓ Rewards early placement |
+
+## Related
+
+- [Context Precision Scorer](https://mastra.ai/reference/evals/context-precision) - Evaluates context ranking using MAP
+- [Faithfulness Scorer](https://mastra.ai/reference/evals/faithfulness) - Measures answer groundedness in context
+- [Custom Scorers](https://mastra.ai/docs/evals/custom-scorers) - Creating your own evaluation metrics

package/dist/docs/references/reference-evals-faithfulness.md

@@ -0,0 +1,114 @@
+# Faithfulness Scorer
+
+The `createFaithfulnessScorer()` function evaluates how factually accurate an LLM's output is compared to the provided context. It extracts claims from the output and verifies them against the context, making it essential to measure RAG pipeline responses' reliability.
+
+## Parameters
+
+The `createFaithfulnessScorer()` function accepts a single options object with the following properties:
+
+**model:** (`LanguageModel`): Configuration for the model used to evaluate faithfulness.
+
+**context:** (`string[]`): Array of context chunks against which the output's claims will be verified.
+
+**scale:** (`number`): The maximum score value. The final score will be normalized to this scale. (Default: `1`)
+
+This function returns an instance of the MastraScorer class. The `.run()` method accepts the same input as other scorers (see the [MastraScorer reference](https://mastra.ai/reference/evals/mastra-scorer)), but the return value includes LLM-specific fields as documented below.
+
+## .run() Returns
+
+**runId:** (`string`): The id of the run (optional).
+
+**preprocessStepResult:** (`string[]`): Array of extracted claims from the output.
+
+**preprocessPrompt:** (`string`): The prompt sent to the LLM for the preprocess step (optional).
+
+**analyzeStepResult:** (`object`): Object with verdicts: { verdicts: Array<{ verdict: 'yes' | 'no' | 'unsure', reason: string }> }
+
+**analyzePrompt:** (`string`): The prompt sent to the LLM for the analyze step (optional).
+
+**score:** (`number`): A score between 0 and the configured scale, representing the proportion of claims that are supported by the context.
+
+**reason:** (`string`): A detailed explanation of the score, including which claims were supported, contradicted, or marked as unsure.
+
+**generateReasonPrompt:** (`string`): The prompt sent to the LLM for the generateReason step (optional).
+
+## Scoring Details
+
+The scorer evaluates faithfulness through claim verification against provided context.
+
+### Scoring Process
+
+1. Analyzes claims and context:
+
+   - Extracts all claims (factual and speculative)
+
+   - Verifies each claim against context
+
+   - Assigns one of three verdicts:
+
+     - "yes" - claim supported by context
+     - "no" - claim contradicts context
+     - "unsure" - claim unverifiable
+
+2. Calculates faithfulness score:
+
+   - Counts supported claims
+   - Divides by total claims
+   - Scales to configured range
+
+Final score: `(supported_claims / total_claims) * scale`
+
+### Score interpretation
+
+A faithfulness score between 0 and 1:
+
+- **1.0**: All claims are accurate and directly supported by the context.
+- **0.7–0.9**: Most claims are correct, with minor additions or omissions.
+- **0.4–0.6**: Some claims are supported, but others are unverifiable.
+- **0.1–0.3**: Most of the content is inaccurate or unsupported.
+- **0.0**: All claims are false or contradict the context.
+
+## Example
+
+Evaluate agent responses for faithfulness to provided context:
+
+```typescript
+import { runEvals } from "@mastra/core/evals";
+import { createFaithfulnessScorer } from "@mastra/evals/scorers/prebuilt";
+import { myAgent } from "./agent";
+
+// Context is typically populated from agent tool calls or RAG retrieval
+const scorer = createFaithfulnessScorer({
+  model: "openai/gpt-4o",
+});
+
+const result = await runEvals({
+  data: [
+    {
+      input: "Tell me about the Tesla Model 3.",
+    },
+    {
+      input: "What are the key features of this electric vehicle?",
+    },
+  ],
+  scorers: [scorer],
+  target: myAgent,
+  onItemComplete: ({ scorerResults }) => {
+    console.log({
+      score: scorerResults[scorer.id].score,
+      reason: scorerResults[scorer.id].reason,
+    });
+  },
+});
+
+console.log(result.scores);
+```
+
+For more details on `runEvals`, see the [runEvals reference](https://mastra.ai/reference/evals/run-evals).
+
+To add this scorer to an agent, see the [Scorers overview](https://mastra.ai/docs/evals/overview) guide.
+
+## Related
+
+- [Answer Relevancy Scorer](https://mastra.ai/reference/evals/answer-relevancy)
+- [Hallucination Scorer](https://mastra.ai/reference/evals/hallucination)