@arizeai/phoenix-evals 2.7.0 → 2.8.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +12 -0
- package/README.md +40 -9
- package/dist/esm/decision/decideClassification.d.ts.map +1 -1
- package/dist/esm/decision/decideClassification.js +4 -4
- package/dist/esm/decision/decideClassification.js.map +1 -1
- package/dist/esm/helpers/toEvaluationResult.d.ts +3 -3
- package/dist/esm/helpers/toEvaluationResult.d.ts.map +1 -1
- package/dist/esm/helpers/toEvaluationResult.js +22 -3
- package/dist/esm/helpers/toEvaluationResult.js.map +1 -1
- package/dist/esm/llm/createCompletenessEvaluator.d.ts +1 -1
- package/dist/esm/llm/createCompletenessEvaluator.js +1 -1
- package/dist/esm/llm/createConcisenessEvaluator.d.ts +1 -1
- package/dist/esm/llm/createConcisenessEvaluator.js +1 -1
- package/dist/esm/llm/createCorrectnessEvaluator.d.ts +1 -1
- package/dist/esm/llm/createCorrectnessEvaluator.js +1 -1
- package/dist/esm/llm/createHallucinationEvaluator.d.ts +1 -1
- package/dist/esm/llm/createHallucinationEvaluator.js +1 -1
- package/dist/esm/llm/createPiiDetectionEvaluator.d.ts +1 -1
- package/dist/esm/llm/createPiiDetectionEvaluator.js +1 -1
- package/dist/esm/llm/createRefusalEvaluator.d.ts +1 -1
- package/dist/esm/llm/createRefusalEvaluator.js +1 -1
- package/dist/esm/llm/createRetrievalRelevanceEvaluator.d.ts +1 -1
- package/dist/esm/llm/createRetrievalRelevanceEvaluator.js +1 -1
- package/dist/esm/llm/createToolInvocationEvaluator.d.ts +1 -1
- package/dist/esm/llm/createToolInvocationEvaluator.js +1 -1
- package/dist/esm/llm/createToolResponseHandlingEvaluator.d.ts +1 -1
- package/dist/esm/llm/createToolResponseHandlingEvaluator.js +1 -1
- package/dist/esm/llm/createToolSelectionEvaluator.d.ts +1 -1
- package/dist/esm/llm/createToolSelectionEvaluator.js +1 -1
- package/dist/esm/llm/createToxicityEvaluator.d.ts +1 -1
- package/dist/esm/llm/createToxicityEvaluator.js +1 -1
- package/dist/esm/llm/createUserFrictionEvaluator.d.ts +1 -1
- package/dist/esm/llm/createUserFrictionEvaluator.js +1 -1
- package/dist/esm/tsconfig.esm.tsbuildinfo +1 -1
- package/dist/src/decision/decideClassification.d.ts.map +1 -1
- package/dist/src/decision/decideClassification.js +4 -4
- package/dist/src/decision/decideClassification.js.map +1 -1
- package/dist/src/helpers/toEvaluationResult.d.ts +3 -3
- package/dist/src/helpers/toEvaluationResult.d.ts.map +1 -1
- package/dist/src/helpers/toEvaluationResult.js +22 -3
- package/dist/src/helpers/toEvaluationResult.js.map +1 -1
- package/dist/src/llm/createCompletenessEvaluator.d.ts +1 -1
- package/dist/src/llm/createCompletenessEvaluator.js +1 -1
- package/dist/src/llm/createConcisenessEvaluator.d.ts +1 -1
- package/dist/src/llm/createConcisenessEvaluator.js +1 -1
- package/dist/src/llm/createCorrectnessEvaluator.d.ts +1 -1
- package/dist/src/llm/createCorrectnessEvaluator.js +1 -1
- package/dist/src/llm/createHallucinationEvaluator.d.ts +1 -1
- package/dist/src/llm/createHallucinationEvaluator.js +1 -1
- package/dist/src/llm/createPiiDetectionEvaluator.d.ts +1 -1
- package/dist/src/llm/createPiiDetectionEvaluator.js +1 -1
- package/dist/src/llm/createRefusalEvaluator.d.ts +1 -1
- package/dist/src/llm/createRefusalEvaluator.js +1 -1
- package/dist/src/llm/createRetrievalRelevanceEvaluator.d.ts +1 -1
- package/dist/src/llm/createRetrievalRelevanceEvaluator.js +1 -1
- package/dist/src/llm/createToolInvocationEvaluator.d.ts +1 -1
- package/dist/src/llm/createToolInvocationEvaluator.js +1 -1
- package/dist/src/llm/createToolResponseHandlingEvaluator.d.ts +1 -1
- package/dist/src/llm/createToolResponseHandlingEvaluator.js +1 -1
- package/dist/src/llm/createToolSelectionEvaluator.d.ts +1 -1
- package/dist/src/llm/createToolSelectionEvaluator.js +1 -1
- package/dist/src/llm/createToxicityEvaluator.d.ts +1 -1
- package/dist/src/llm/createToxicityEvaluator.js +1 -1
- package/dist/src/llm/createUserFrictionEvaluator.d.ts +1 -1
- package/dist/src/llm/createUserFrictionEvaluator.js +1 -1
- package/dist/tsconfig.tsbuildinfo +1 -1
- package/docs/classification.mdx +2 -2
- package/package.json +12 -12
- package/src/decision/decideClassification.ts +4 -5
- package/src/helpers/toEvaluationResult.ts +27 -3
- package/src/llm/createCompletenessEvaluator.ts +1 -1
- package/src/llm/createConcisenessEvaluator.ts +1 -1
- package/src/llm/createCorrectnessEvaluator.ts +1 -1
- package/src/llm/createHallucinationEvaluator.ts +1 -1
- package/src/llm/createPiiDetectionEvaluator.ts +1 -1
- package/src/llm/createRefusalEvaluator.ts +1 -1
- package/src/llm/createRetrievalRelevanceEvaluator.ts +1 -1
- package/src/llm/createToolInvocationEvaluator.ts +1 -1
- package/src/llm/createToolResponseHandlingEvaluator.ts +1 -1
- package/src/llm/createToolSelectionEvaluator.ts +1 -1
- package/src/llm/createToxicityEvaluator.ts +1 -1
- package/src/llm/createUserFrictionEvaluator.ts +1 -1
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"decideClassification.d.ts","sourceRoot":"","sources":["../../../src/decision/decideClassification.ts"],"names":[],"mappings":"AAQA,OAAO,KAAK,EAAE,oBAAoB,EAAE,MAAM,gBAAgB,CAAC;AAC3D,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,eAAe,CAAC;AACrD,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AACnD,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAC;
|
|
1
|
+
{"version":3,"file":"decideClassification.d.ts","sourceRoot":"","sources":["../../../src/decision/decideClassification.ts"],"names":[],"mappings":"AAQA,OAAO,KAAK,EAAE,oBAAoB,EAAE,MAAM,gBAAgB,CAAC;AAC3D,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,eAAe,CAAC;AACrD,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AACnD,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAC;AAc9D;;;;;;;;GAQG;AACH,wBAAsB,oBAAoB,CAAC,IAAI,EAAE;IAC/C,KAAK,EAAE,aAAa,CAAC;IACrB,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC;IAC1B,MAAM,EAAE,UAAU,CAAC;IACnB,SAAS,CAAC,EAAE,eAAe,CAAC;CAC7B,GAAG,OAAO,CAAC,oBAAoB,CAAC,CA2BhC"}
|
|
@@ -14,10 +14,10 @@ const decisionPrompt_1 = require("./decisionPrompt");
|
|
|
14
14
|
*/
|
|
15
15
|
const QUESTION_ID = "label";
|
|
16
16
|
/**
|
|
17
|
-
*
|
|
18
|
-
*
|
|
17
|
+
* Instructions sent with the choice question. The OpenAI Decisions API rejects
|
|
18
|
+
* a choice question that has no instructions.
|
|
19
19
|
*/
|
|
20
|
-
const
|
|
20
|
+
const CHOICE_INSTRUCTIONS = "Select the single best label for the input.";
|
|
21
21
|
/**
|
|
22
22
|
* Classify with an AI SDK decision model by asking one choice question whose
|
|
23
23
|
* options are the labels. Decision models don't generate text, so the
|
|
@@ -35,7 +35,7 @@ async function decideClassification(args) {
|
|
|
35
35
|
questions: {
|
|
36
36
|
[QUESTION_ID]: {
|
|
37
37
|
type: "choice",
|
|
38
|
-
instructions:
|
|
38
|
+
instructions: CHOICE_INSTRUCTIONS,
|
|
39
39
|
criteria: Object.fromEntries(labels.map((label) => [label, null])),
|
|
40
40
|
},
|
|
41
41
|
},
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"decideClassification.js","sourceRoot":"","sources":["../../../src/decision/decideClassification.ts"],"names":[],"mappings":";;;AAAA;;;;GAIG;AACH,2BAAyC;AAEzC,4CAAmD;AAKnD,qDAAyD;AAEzD;;GAEG;AACH,MAAM,WAAW,GAAG,OAAO,CAAC;AAE5B;;;GAGG;AACH,MAAM,
|
|
1
|
+
{"version":3,"file":"decideClassification.js","sourceRoot":"","sources":["../../../src/decision/decideClassification.ts"],"names":[],"mappings":";;;AAAA;;;;GAIG;AACH,2BAAyC;AAEzC,4CAAmD;AAKnD,qDAAyD;AAEzD;;GAEG;AACH,MAAM,WAAW,GAAG,OAAO,CAAC;AAE5B;;;GAGG;AACH,MAAM,mBAAmB,GAAG,6CAA6C,CAAC;AAE1E;;;;;;;;GAQG;AACI,KAAK,+BAA+B,IAK1C;IACC,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,SAAS,EAAE,GAAG,IAAI,CAAC;IAClD,MAAM,MAAM,GAAG,MAAM,IAAA,wBAAmB,EAAC;QACvC,KAAK;QACL,KAAK,EAAE,IAAA,sCAAqB,EAAC,MAAM,CAAC;QACpC,SAAS,EAAE;YACT,CAAC,WAAW,CAAC,EAAE;gBACb,IAAI,EAAE,QAAQ;gBACd,YAAY,EAAE,mBAAmB;gBACjC,QAAQ,EAAE,MAAM,CAAC,WAAW,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC,CAAC;aACnE;SACF;QACD,SAAS,EAAE,IAAA,+BAAmB,EAAC;YAC7B,UAAU,EAAE,sBAAsB;YAClC,SAAS;SACV,CAAC;KACH,CAAC,CAAC;IACH,MAAM,MAAM,GAAG,MAAM,CAAC,OAAO,CAAC,WAAW,CAAC,CAAC;IAC3C,MAAM,UAAU,GAAG,aAAa,CAAC,MAAM,CAAC,gBAAgB,CAAC,CAAC;IAC1D,OAAO;QACL,KAAK,EAAE,MAAM,CAAC,MAAM;QACpB,QAAQ,gCACN,aAAa,EAAE,MAAM,CAAC,aAAa,IAChC,CAAC,UAAU,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,UAAU,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,KACnD,OAAO,EAAE,MAAM,CAAC,QAAQ,CAAC,OAAO,GACjC;KACF,CAAC;AACJ,CAAC;AAED;;;;GAIG;AACH,SAAS,aAAa,CAAC,gBAAyB;;IAC9C,IAAI,OAAO,gBAAgB,KAAK,QAAQ,IAAI,gBAAgB,KAAK,IAAI,EAAE,CAAC;QACtE,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,KAAK,MAAM,QAAQ,IAAI,MAAM,CAAC,MAAM,CAAC,gBAAgB,CAAC,EAAE,CAAC;QACvD,MAAM,UAAU,GAAG,MAAC,QAAqD,aAArD,QAAQ,uBAAR,QAAQ,CACxB,UAAU,0CAAG,WAAW,CAAC,CAAC;QAC9B,IAAI,OAAO,UAAU,KAAK,QAAQ;YAAE,OAAO,UAAU,CAAC;IACxD,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC"}
|
|
@@ -8,7 +8,7 @@ import type { EvaluationResult } from "../types";
|
|
|
8
8
|
*
|
|
9
9
|
* - **Numbers**: Converted to `{ score: number }`
|
|
10
10
|
* - **Strings**: Converted to `{ label: string }`
|
|
11
|
-
* - **Objects**: Extracts `score`, `label`, and `
|
|
11
|
+
* - **Objects**: Extracts `score`, `label`, `explanation`, and `metadata` properties if present
|
|
12
12
|
* - **Other types**: Returns an empty `EvaluationResult` object
|
|
13
13
|
*
|
|
14
14
|
* This is particularly useful when creating evaluators from functions that may
|
|
@@ -17,7 +17,7 @@ import type { EvaluationResult } from "../types";
|
|
|
17
17
|
* @param result - The value to convert to an EvaluationResult. Can be:
|
|
18
18
|
* - A number (converted to score)
|
|
19
19
|
* - A string (converted to label)
|
|
20
|
-
* - An object with optional `score`, `label`, and/or `
|
|
20
|
+
* - An object with optional `score`, `label`, `explanation`, and/or `metadata` properties
|
|
21
21
|
* - Any other value (returns empty object)
|
|
22
22
|
*
|
|
23
23
|
* @returns An {@link EvaluationResult} object with extracted properties
|
|
@@ -37,7 +37,7 @@ import type { EvaluationResult } from "../types";
|
|
|
37
37
|
* ```
|
|
38
38
|
*
|
|
39
39
|
* @example
|
|
40
|
-
* Convert an object with
|
|
40
|
+
* Convert an object with multiple properties:
|
|
41
41
|
* ```typescript
|
|
42
42
|
* const result = toEvaluationResult({
|
|
43
43
|
* score: 0.9,
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"toEvaluationResult.d.ts","sourceRoot":"","sources":["../../../src/helpers/toEvaluationResult.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,UAAU,CAAC;
|
|
1
|
+
{"version":3,"file":"toEvaluationResult.d.ts","sourceRoot":"","sources":["../../../src/helpers/toEvaluationResult.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,UAAU,CAAC;AA4EjD;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA+DG;AACH,wBAAgB,kBAAkB,CAAC,MAAM,EAAE,OAAO,GAAG,gBAAgB,CA4BpE"}
|
|
@@ -43,6 +43,22 @@ function resultHasExplanation(result) {
|
|
|
43
43
|
"explanation" in result &&
|
|
44
44
|
typeof result.explanation === "string");
|
|
45
45
|
}
|
|
46
|
+
/**
|
|
47
|
+
* Type guard to check if a value has a metadata property.
|
|
48
|
+
*
|
|
49
|
+
* @param result - The value to check
|
|
50
|
+
* @returns True if the value is an object with a metadata record
|
|
51
|
+
*
|
|
52
|
+
* @internal
|
|
53
|
+
*/
|
|
54
|
+
function resultHasMetadata(result) {
|
|
55
|
+
return (typeof result === "object" &&
|
|
56
|
+
result !== null &&
|
|
57
|
+
"metadata" in result &&
|
|
58
|
+
typeof result.metadata === "object" &&
|
|
59
|
+
result.metadata !== null &&
|
|
60
|
+
!Array.isArray(result.metadata));
|
|
61
|
+
}
|
|
46
62
|
/**
|
|
47
63
|
* Converts an unknown value to an {@link EvaluationResult}.
|
|
48
64
|
*
|
|
@@ -52,7 +68,7 @@ function resultHasExplanation(result) {
|
|
|
52
68
|
*
|
|
53
69
|
* - **Numbers**: Converted to `{ score: number }`
|
|
54
70
|
* - **Strings**: Converted to `{ label: string }`
|
|
55
|
-
* - **Objects**: Extracts `score`, `label`, and `
|
|
71
|
+
* - **Objects**: Extracts `score`, `label`, `explanation`, and `metadata` properties if present
|
|
56
72
|
* - **Other types**: Returns an empty `EvaluationResult` object
|
|
57
73
|
*
|
|
58
74
|
* This is particularly useful when creating evaluators from functions that may
|
|
@@ -61,7 +77,7 @@ function resultHasExplanation(result) {
|
|
|
61
77
|
* @param result - The value to convert to an EvaluationResult. Can be:
|
|
62
78
|
* - A number (converted to score)
|
|
63
79
|
* - A string (converted to label)
|
|
64
|
-
* - An object with optional `score`, `label`, and/or `
|
|
80
|
+
* - An object with optional `score`, `label`, `explanation`, and/or `metadata` properties
|
|
65
81
|
* - Any other value (returns empty object)
|
|
66
82
|
*
|
|
67
83
|
* @returns An {@link EvaluationResult} object with extracted properties
|
|
@@ -81,7 +97,7 @@ function resultHasExplanation(result) {
|
|
|
81
97
|
* ```
|
|
82
98
|
*
|
|
83
99
|
* @example
|
|
84
|
-
* Convert an object with
|
|
100
|
+
* Convert an object with multiple properties:
|
|
85
101
|
* ```typescript
|
|
86
102
|
* const result = toEvaluationResult({
|
|
87
103
|
* score: 0.9,
|
|
@@ -129,6 +145,9 @@ function toEvaluationResult(result) {
|
|
|
129
145
|
if (resultHasExplanation(result)) {
|
|
130
146
|
evaluationResult.explanation = result.explanation;
|
|
131
147
|
}
|
|
148
|
+
if (resultHasMetadata(result)) {
|
|
149
|
+
evaluationResult.metadata = result.metadata;
|
|
150
|
+
}
|
|
132
151
|
return evaluationResult;
|
|
133
152
|
}
|
|
134
153
|
return {};
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"toEvaluationResult.js","sourceRoot":"","sources":["../../../src/helpers/toEvaluationResult.ts"],"names":[],"mappings":";;;AAEA;;;;;;;GAOG;AACH,SAAS,cAAc,CAAC,MAAe;IACrC,OAAO,CACL,OAAO,MAAM,KAAK,QAAQ;QAC1B,MAAM,KAAK,IAAI;QACf,OAAO,IAAI,MAAM;QACjB,OAAO,MAAM,CAAC,KAAK,KAAK,QAAQ,CACjC,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,cAAc,CAAC,MAAe;IACrC,OAAO,CACL,OAAO,MAAM,KAAK,QAAQ;QAC1B,MAAM,KAAK,IAAI;QACf,OAAO,IAAI,MAAM;QACjB,OAAO,MAAM,CAAC,KAAK,KAAK,QAAQ,CACjC,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,oBAAoB,CAC3B,MAAe;IAEf,OAAO,CACL,OAAO,MAAM,KAAK,QAAQ;QAC1B,MAAM,KAAK,IAAI;QACf,aAAa,IAAI,MAAM;QACvB,OAAO,MAAM,CAAC,WAAW,KAAK,QAAQ,CACvC,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA+DG;AACH,4BAAmC,MAAe;IAChD,IAAI,OAAO,MAAM,KAAK,QAAQ,EAAE,CAAC;QAC/B,OAAO;YACL,KAAK,EAAE,MAAM;SACd,CAAC;IACJ,CAAC;IACD,IAAI,OAAO,MAAM,KAAK,QAAQ,EAAE,CAAC;QAC/B,OAAO;YACL,KAAK,EAAE,MAAM;SACd,CAAC;IACJ,CAAC;IACD,IAAI,OAAO,MAAM,KAAK,QAAQ,IAAI,MAAM,KAAK,IAAI,EAAE,CAAC;QAClD,MAAM,gBAAgB,GAAqB,EAAE,CAAC;QAC9C,IAAI,cAAc,CAAC,MAAM,CAAC,EAAE,CAAC;YAC3B,gBAAgB,CAAC,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC;QACxC,CAAC;QACD,IAAI,cAAc,CAAC,MAAM,CAAC,EAAE,CAAC;YAC3B,gBAAgB,CAAC,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC;QACxC,CAAC;QACD,IAAI,oBAAoB,CAAC,MAAM,CAAC,EAAE,CAAC;YACjC,gBAAgB,CAAC,WAAW,GAAG,MAAM,CAAC,WAAW,CAAC;QACpD,CAAC;QACD,OAAO,gBAAgB,CAAC;IAC1B,CAAC;IACD,OAAO,EAAE,CAAC;AACZ,CAAC"}
|
|
1
|
+
{"version":3,"file":"toEvaluationResult.js","sourceRoot":"","sources":["../../../src/helpers/toEvaluationResult.ts"],"names":[],"mappings":";;;AAEA;;;;;;;GAOG;AACH,SAAS,cAAc,CAAC,MAAe;IACrC,OAAO,CACL,OAAO,MAAM,KAAK,QAAQ;QAC1B,MAAM,KAAK,IAAI;QACf,OAAO,IAAI,MAAM;QACjB,OAAO,MAAM,CAAC,KAAK,KAAK,QAAQ,CACjC,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,cAAc,CAAC,MAAe;IACrC,OAAO,CACL,OAAO,MAAM,KAAK,QAAQ;QAC1B,MAAM,KAAK,IAAI;QACf,OAAO,IAAI,MAAM;QACjB,OAAO,MAAM,CAAC,KAAK,KAAK,QAAQ,CACjC,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,oBAAoB,CAC3B,MAAe;IAEf,OAAO,CACL,OAAO,MAAM,KAAK,QAAQ;QAC1B,MAAM,KAAK,IAAI;QACf,aAAa,IAAI,MAAM;QACvB,OAAO,MAAM,CAAC,WAAW,KAAK,QAAQ,CACvC,CAAC;AACJ,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,iBAAiB,CACxB,MAAe;IAEf,OAAO,CACL,OAAO,MAAM,KAAK,QAAQ;QAC1B,MAAM,KAAK,IAAI;QACf,UAAU,IAAI,MAAM;QACpB,OAAO,MAAM,CAAC,QAAQ,KAAK,QAAQ;QACnC,MAAM,CAAC,QAAQ,KAAK,IAAI;QACxB,CAAC,KAAK,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,CAChC,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA+DG;AACH,4BAAmC,MAAe;IAChD,IAAI,OAAO,MAAM,KAAK,QAAQ,EAAE,CAAC;QAC/B,OAAO;YACL,KAAK,EAAE,MAAM;SACd,CAAC;IACJ,CAAC;IACD,IAAI,OAAO,MAAM,KAAK,QAAQ,EAAE,CAAC;QAC/B,OAAO;YACL,KAAK,EAAE,MAAM;SACd,CAAC;IACJ,CAAC;IACD,IAAI,OAAO,MAAM,KAAK,QAAQ,IAAI,MAAM,KAAK,IAAI,EAAE,CAAC;QAClD,MAAM,gBAAgB,GAAqB,EAAE,CAAC;QAC9C,IAAI,cAAc,CAAC,MAAM,CAAC,EAAE,CAAC;YAC3B,gBAAgB,CAAC,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC;QACxC,CAAC;QACD,IAAI,cAAc,CAAC,MAAM,CAAC,EAAE,CAAC;YAC3B,gBAAgB,CAAC,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC;QACxC,CAAC;QACD,IAAI,oBAAoB,CAAC,MAAM,CAAC,EAAE,CAAC;YACjC,gBAAgB,CAAC,WAAW,GAAG,MAAM,CAAC,WAAW,CAAC;QACpD,CAAC;QACD,IAAI,iBAAiB,CAAC,MAAM,CAAC,EAAE,CAAC;YAC9B,gBAAgB,CAAC,QAAQ,GAAG,MAAM,CAAC,QAAQ,CAAC;QAC9C,CAAC;QACD,OAAO,gBAAgB,CAAC;IAC1B,CAAC;IACD,OAAO,EAAE,CAAC;AACZ,CAAC"}
|
|
@@ -34,7 +34,7 @@ export type CompletenessEvaluationRecord = {
|
|
|
34
34
|
*
|
|
35
35
|
* @example
|
|
36
36
|
* ```ts
|
|
37
|
-
* const evaluator = createCompletenessEvaluator({ model: openai("gpt-
|
|
37
|
+
* const evaluator = createCompletenessEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
38
38
|
* const result = await evaluator.evaluate({
|
|
39
39
|
* conversation:
|
|
40
40
|
* "User: Reset my password and update the billing address.\nAssistant: Your password has been reset.",
|
|
@@ -31,7 +31,7 @@ const createClassificationEvaluator_1 = require("./createClassificationEvaluator
|
|
|
31
31
|
*
|
|
32
32
|
* @example
|
|
33
33
|
* ```ts
|
|
34
|
-
* const evaluator = createCompletenessEvaluator({ model: openai("gpt-
|
|
34
|
+
* const evaluator = createCompletenessEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
35
35
|
* const result = await evaluator.evaluate({
|
|
36
36
|
* conversation:
|
|
37
37
|
* "User: Reset my password and update the billing address.\nAssistant: Your password has been reset.",
|
|
@@ -31,7 +31,7 @@ export type ConcisenessEvaluationRecord = {
|
|
|
31
31
|
*
|
|
32
32
|
* @example
|
|
33
33
|
* ```ts
|
|
34
|
-
* const evaluator = createConcisenessEvaluator({ model: openai("gpt-
|
|
34
|
+
* const evaluator = createConcisenessEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
35
35
|
* const result = await evaluator.evaluate({
|
|
36
36
|
* input: "What is the capital of France?",
|
|
37
37
|
* output: "Paris.",
|
|
@@ -32,7 +32,7 @@ const createClassificationEvaluator_1 = require("./createClassificationEvaluator
|
|
|
32
32
|
*
|
|
33
33
|
* @example
|
|
34
34
|
* ```ts
|
|
35
|
-
* const evaluator = createConcisenessEvaluator({ model: openai("gpt-
|
|
35
|
+
* const evaluator = createConcisenessEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
36
36
|
* const result = await evaluator.evaluate({
|
|
37
37
|
* input: "What is the capital of France?",
|
|
38
38
|
* output: "Paris.",
|
|
@@ -30,7 +30,7 @@ export type CorrectnessEvaluationRecord = {
|
|
|
30
30
|
*
|
|
31
31
|
* @example
|
|
32
32
|
* ```ts
|
|
33
|
-
* const evaluator = createCorrectnessEvaluator({ model: openai("gpt-
|
|
33
|
+
* const evaluator = createCorrectnessEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
34
34
|
* const result = await evaluator.evaluate({
|
|
35
35
|
* input: "What is the capital of France?",
|
|
36
36
|
* output: "Paris is the capital of France.",
|
|
@@ -31,7 +31,7 @@ const createClassificationEvaluator_1 = require("./createClassificationEvaluator
|
|
|
31
31
|
*
|
|
32
32
|
* @example
|
|
33
33
|
* ```ts
|
|
34
|
-
* const evaluator = createCorrectnessEvaluator({ model: openai("gpt-
|
|
34
|
+
* const evaluator = createCorrectnessEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
35
35
|
* const result = await evaluator.evaluate({
|
|
36
36
|
* input: "What is the capital of France?",
|
|
37
37
|
* output: "Paris is the capital of France.",
|
|
@@ -40,7 +40,7 @@ export interface HallucinationEvaluationRecord {
|
|
|
40
40
|
*
|
|
41
41
|
* @example
|
|
42
42
|
* ```ts
|
|
43
|
-
* const evaluator = createHallucinationEvaluator({ model: openai("gpt-
|
|
43
|
+
* const evaluator = createHallucinationEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
44
44
|
* const result = await evaluator.evaluate({
|
|
45
45
|
* input:
|
|
46
46
|
* "User: What's our refund window?\nTool (lookup_policy): Refunds: 30 days from delivery.\nAssistant: 30 days from delivery.\nUser: And for electronics?",
|
|
@@ -34,7 +34,7 @@ const createClassificationEvaluator_1 = require("./createClassificationEvaluator
|
|
|
34
34
|
*
|
|
35
35
|
* @example
|
|
36
36
|
* ```ts
|
|
37
|
-
* const evaluator = createHallucinationEvaluator({ model: openai("gpt-
|
|
37
|
+
* const evaluator = createHallucinationEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
38
38
|
* const result = await evaluator.evaluate({
|
|
39
39
|
* input:
|
|
40
40
|
* "User: What's our refund window?\nTool (lookup_policy): Refunds: 30 days from delivery.\nAssistant: 30 days from delivery.\nUser: And for electronics?",
|
|
@@ -34,7 +34,7 @@ export type PiiDetectionEvaluationRecord = {
|
|
|
34
34
|
*
|
|
35
35
|
* @example
|
|
36
36
|
* ```ts
|
|
37
|
-
* const evaluator = createPiiDetectionEvaluator({ model: openai("gpt-
|
|
37
|
+
* const evaluator = createPiiDetectionEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
38
38
|
* const result = await evaluator.evaluate({
|
|
39
39
|
* conversation:
|
|
40
40
|
* "User: Reset my account.\nAssistant: What email is on the account?\nUser: jane.doe@acme.com",
|
|
@@ -32,7 +32,7 @@ const createClassificationEvaluator_1 = require("./createClassificationEvaluator
|
|
|
32
32
|
*
|
|
33
33
|
* @example
|
|
34
34
|
* ```ts
|
|
35
|
-
* const evaluator = createPiiDetectionEvaluator({ model: openai("gpt-
|
|
35
|
+
* const evaluator = createPiiDetectionEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
36
36
|
* const result = await evaluator.evaluate({
|
|
37
37
|
* conversation:
|
|
38
38
|
* "User: Reset my account.\nAssistant: What email is on the account?\nUser: jane.doe@acme.com",
|
|
@@ -32,7 +32,7 @@ export type RefusalEvaluationRecord = {
|
|
|
32
32
|
*
|
|
33
33
|
* @example
|
|
34
34
|
* ```ts
|
|
35
|
-
* const evaluator = createRefusalEvaluator({ model: openai("gpt-
|
|
35
|
+
* const evaluator = createRefusalEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
36
36
|
* const result = await evaluator.evaluate({
|
|
37
37
|
* input: "What is the capital of France?",
|
|
38
38
|
* output: "I'm sorry, I can only help with technical questions.",
|
|
@@ -33,7 +33,7 @@ const createClassificationEvaluator_1 = require("./createClassificationEvaluator
|
|
|
33
33
|
*
|
|
34
34
|
* @example
|
|
35
35
|
* ```ts
|
|
36
|
-
* const evaluator = createRefusalEvaluator({ model: openai("gpt-
|
|
36
|
+
* const evaluator = createRefusalEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
37
37
|
* const result = await evaluator.evaluate({
|
|
38
38
|
* input: "What is the capital of France?",
|
|
39
39
|
* output: "I'm sorry, I can only help with technical questions.",
|
|
@@ -42,7 +42,7 @@ export interface RetrievalRelevanceEvaluationRecord {
|
|
|
42
42
|
*
|
|
43
43
|
* @example
|
|
44
44
|
* ```ts
|
|
45
|
-
* const evaluator = createRetrievalRelevanceEvaluator({ model: openai("gpt-
|
|
45
|
+
* const evaluator = createRetrievalRelevanceEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
46
46
|
* const result = await evaluator.evaluate({
|
|
47
47
|
* input: "What is the capital of France?",
|
|
48
48
|
* context: "Paris is the capital and largest city of France.",
|
|
@@ -35,7 +35,7 @@ const createClassificationEvaluator_1 = require("./createClassificationEvaluator
|
|
|
35
35
|
*
|
|
36
36
|
* @example
|
|
37
37
|
* ```ts
|
|
38
|
-
* const evaluator = createRetrievalRelevanceEvaluator({ model: openai("gpt-
|
|
38
|
+
* const evaluator = createRetrievalRelevanceEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
39
39
|
* const result = await evaluator.evaluate({
|
|
40
40
|
* input: "What is the capital of France?",
|
|
41
41
|
* context: "Paris is the capital and largest city of France.",
|
|
@@ -47,7 +47,7 @@ export type ToolInvocationEvaluationRecord = {
|
|
|
47
47
|
*
|
|
48
48
|
* @example
|
|
49
49
|
* ```ts
|
|
50
|
-
* const evaluator = createToolInvocationEvaluator({ model: openai("gpt-
|
|
50
|
+
* const evaluator = createToolInvocationEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
51
51
|
*
|
|
52
52
|
* // Example with JSON schema format for available tools
|
|
53
53
|
* const result = await evaluator.evaluate({
|
|
@@ -38,7 +38,7 @@ const createClassificationEvaluator_1 = require("./createClassificationEvaluator
|
|
|
38
38
|
*
|
|
39
39
|
* @example
|
|
40
40
|
* ```ts
|
|
41
|
-
* const evaluator = createToolInvocationEvaluator({ model: openai("gpt-
|
|
41
|
+
* const evaluator = createToolInvocationEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
42
42
|
*
|
|
43
43
|
* // Example with JSON schema format for available tools
|
|
44
44
|
* const result = await evaluator.evaluate({
|
|
@@ -47,7 +47,7 @@ export type ToolResponseHandlingEvaluationRecord = {
|
|
|
47
47
|
*
|
|
48
48
|
* @example
|
|
49
49
|
* ```ts
|
|
50
|
-
* const evaluator = createToolResponseHandlingEvaluator({ model: openai("gpt-
|
|
50
|
+
* const evaluator = createToolResponseHandlingEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
51
51
|
*
|
|
52
52
|
* // Example: Correct extraction from tool result
|
|
53
53
|
* const result = await evaluator.evaluate({
|
|
@@ -33,7 +33,7 @@ const createClassificationEvaluator_1 = require("./createClassificationEvaluator
|
|
|
33
33
|
*
|
|
34
34
|
* @example
|
|
35
35
|
* ```ts
|
|
36
|
-
* const evaluator = createToolResponseHandlingEvaluator({ model: openai("gpt-
|
|
36
|
+
* const evaluator = createToolResponseHandlingEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
37
37
|
*
|
|
38
38
|
* // Example: Correct extraction from tool result
|
|
39
39
|
* const result = await evaluator.evaluate({
|
|
@@ -48,7 +48,7 @@ export type ToolSelectionEvaluationRecord = {
|
|
|
48
48
|
*
|
|
49
49
|
* @example
|
|
50
50
|
* ```ts
|
|
51
|
-
* const evaluator = createToolSelectionEvaluator({ model: openai("gpt-
|
|
51
|
+
* const evaluator = createToolSelectionEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
52
52
|
*
|
|
53
53
|
* const result = await evaluator.evaluate({
|
|
54
54
|
* input: "User: What is the weather in San Francisco?",
|
|
@@ -39,7 +39,7 @@ const createClassificationEvaluator_1 = require("./createClassificationEvaluator
|
|
|
39
39
|
*
|
|
40
40
|
* @example
|
|
41
41
|
* ```ts
|
|
42
|
-
* const evaluator = createToolSelectionEvaluator({ model: openai("gpt-
|
|
42
|
+
* const evaluator = createToolSelectionEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
43
43
|
*
|
|
44
44
|
* const result = await evaluator.evaluate({
|
|
45
45
|
* input: "User: What is the weather in San Francisco?",
|
|
@@ -29,7 +29,7 @@ export type ToxicityEvaluationRecord = {
|
|
|
29
29
|
*
|
|
30
30
|
* @example
|
|
31
31
|
* ```ts
|
|
32
|
-
* const evaluator = createToxicityEvaluator({ model: openai("gpt-
|
|
32
|
+
* const evaluator = createToxicityEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
33
33
|
* const result = await evaluator.evaluate({
|
|
34
34
|
* text: "You are a worthless idiot and everyone despises you.",
|
|
35
35
|
* });
|
|
@@ -31,7 +31,7 @@ const createClassificationEvaluator_1 = require("./createClassificationEvaluator
|
|
|
31
31
|
*
|
|
32
32
|
* @example
|
|
33
33
|
* ```ts
|
|
34
|
-
* const evaluator = createToxicityEvaluator({ model: openai("gpt-
|
|
34
|
+
* const evaluator = createToxicityEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
35
35
|
* const result = await evaluator.evaluate({
|
|
36
36
|
* text: "You are a worthless idiot and everyone despises you.",
|
|
37
37
|
* });
|
|
@@ -36,7 +36,7 @@ export interface UserFrictionEvaluationRecord {
|
|
|
36
36
|
*
|
|
37
37
|
* @example
|
|
38
38
|
* ```ts
|
|
39
|
-
* const evaluator = createUserFrictionEvaluator({ model: openai("gpt-
|
|
39
|
+
* const evaluator = createUserFrictionEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
40
40
|
* const result = await evaluator.evaluate({
|
|
41
41
|
* conversation: "User: Show recent orders.\nAssistant: Here are last month's orders.",
|
|
42
42
|
* userMessage: "No, I asked for this week.",
|
|
@@ -32,7 +32,7 @@ const createClassificationEvaluator_1 = require("./createClassificationEvaluator
|
|
|
32
32
|
*
|
|
33
33
|
* @example
|
|
34
34
|
* ```ts
|
|
35
|
-
* const evaluator = createUserFrictionEvaluator({ model: openai("gpt-
|
|
35
|
+
* const evaluator = createUserFrictionEvaluator({ model: openai.decisionModel("gpt-6-luna") });
|
|
36
36
|
* const result = await evaluator.evaluate({
|
|
37
37
|
* conversation: "User: Show recent orders.\nAssistant: Here are last month's orders.",
|
|
38
38
|
* userMessage: "No, I asked for this week.",
|