wehub-resource-sync
0d3cb498a3
Deploy local.promptfoo.app / Deploy to Cloudflare Pages (push) Waiting to run
Test and Publish Multi-arch Docker Image / test (push) Waiting to run
Test and Publish Multi-arch Docker Image / build-docker-and-push-digests (map[digest-suffix:linux-amd64 platform:linux/amd64 runner:ubuntu-latest]) (push) Blocked by required conditions
Test and Publish Multi-arch Docker Image / build-docker-and-push-digests (map[digest-suffix:linux-arm64 platform:linux/arm64 runner:ubuntu-24.04-arm]) (push) Blocked by required conditions
Test and Publish Multi-arch Docker Image / merge-docker-digests (push) Blocked by required conditions
Test and Publish Multi-arch Docker Image / Attest Multi-arch Image (push) Blocked by required conditions
Validate Renovate Config / Validate Renovate Configuration (push) Waiting to run
CI / Shell Format Check (push) Has been cancelled
CI / Check Ruby (3.4) (push) Has been cancelled
CI / CI Config (push) Has been cancelled
CI / Test on Node ${{ matrix.node }} and ${{ matrix.os }}${{ matrix.shard && format(' (shard {0}/3)', matrix.shard) || '' }} (push) Has been cancelled
CI / Build on Node ${{ matrix.node }} (push) Has been cancelled
CI / Style Check (push) Has been cancelled
CI / Generate Assets (push) Has been cancelled
CI / Check Python (3.14) (push) Has been cancelled
CI / Check Python (3.9) (push) Has been cancelled
CI / Build Docs (push) Has been cancelled
CI / Code Scan Action (push) Has been cancelled
CI / Site tests (push) Has been cancelled
CI / webui tests (push) Has been cancelled
CI / Run Integration Tests (push) Has been cancelled
CI / Run Smoke Tests (push) Has been cancelled
CI / Go Tests (push) Has been cancelled
CI / Share Test (push) Has been cancelled
CI / Redteam (Production API) (push) Has been cancelled
CI / Redteam (Staging API) (push) Has been cancelled
CI / GitHub Actions Lint (push) Has been cancelled
CI / Check Ruby (3.0) (push) Has been cancelled
release-please / release-please (push) Has been cancelled
release-please / build (push) Has been cancelled
release-please / publish-npm (push) Has been cancelled
release-please / publish-npm-backfill (push) Has been cancelled
release-please / docker (push) Has been cancelled
release-please / publish-code-scan-action (push) Has been cancelled
release-please / attest-code-scan-action (push) Has been cancelled
422 lines
13 KiB
JSON
422 lines
13 KiB
JSON
{
|
|
"createdAt": 1727905040496,
|
|
"id": "eval-2024-10-02T21:37:20-kMQ",
|
|
"config": {
|
|
"tags": {},
|
|
"description": "A translator built with LLM",
|
|
"prompts": ["file:///Users/steve/src/promptfoo/examples/simple-csv/prompts.txt"],
|
|
"providers": ["openai:gpt-4o-mini"],
|
|
"tests": [
|
|
{
|
|
"vars": { "language": "French", "body": "Hello world" },
|
|
"assert": [],
|
|
"options": {},
|
|
"description": "Row #1"
|
|
},
|
|
{
|
|
"vars": { "language": "French", "body": "I'm hungry" },
|
|
"assert": [],
|
|
"options": {},
|
|
"description": "Row #2"
|
|
},
|
|
{
|
|
"vars": { "language": "Pirate", "body": "Hello world" },
|
|
"assert": [],
|
|
"options": {},
|
|
"description": "Row #3"
|
|
},
|
|
{
|
|
"vars": { "language": "Pirate", "body": "I'm hungry" },
|
|
"assert": [],
|
|
"options": {},
|
|
"description": "Row #4"
|
|
}
|
|
],
|
|
"scenarios": [],
|
|
"env": {},
|
|
"sharing": true,
|
|
"defaultTest": { "vars": {}, "assert": [], "options": {}, "metadata": {} },
|
|
"outputPath": [],
|
|
"extensions": [],
|
|
"metadata": {},
|
|
"redteam": { "plugins": [], "strategies": [] }
|
|
},
|
|
"results": [
|
|
{
|
|
"id": "14c141a7-be6f-460f-b3cf-711e0a64fe88",
|
|
"evalId": "eval-2024-10-02T21:37:20-kMQ",
|
|
"promptIdx": 0,
|
|
"testIdx": 0,
|
|
"testCase": {
|
|
"vars": { "language": "French", "body": "Hello world" },
|
|
"assert": [],
|
|
"options": {},
|
|
"description": "Row #1",
|
|
"metadata": {}
|
|
},
|
|
"prompt": {
|
|
"raw": "Rephrase this in French: Hello world",
|
|
"label": "examples/simple-csv/prompts.txt: Rephrase this in {{language}}: {{body}}"
|
|
},
|
|
"promptId": "ad447337160016cbd0a5c7f2626d0c1fe2ade0d4d653e20bed9250632bc56e03",
|
|
"error": null,
|
|
"score": 1,
|
|
"success": true,
|
|
"providerResponse": {
|
|
"output": "Bonjour le monde.",
|
|
"tokenUsage": { "cached": 19, "total": 19 },
|
|
"cached": true,
|
|
"cost": 0.0000046499999999999995
|
|
},
|
|
"gradingResult": {
|
|
"pass": true,
|
|
"score": 1,
|
|
"reason": "No assertions",
|
|
"tokensUsed": { "total": 0, "prompt": 0, "completion": 0, "cached": 0 },
|
|
"assertion": null
|
|
},
|
|
"namedScores": {},
|
|
"provider": {
|
|
"id": "openai:gpt-4o-mini",
|
|
"label": "",
|
|
"config": { "basePath": "examples/simple-csv" }
|
|
},
|
|
"latencyMs": 3,
|
|
"cost": 0.0000046499999999999995
|
|
},
|
|
{
|
|
"id": "013a4746-4051-4dc5-87d1-1c8414bc124f",
|
|
"evalId": "eval-2024-10-02T21:37:20-kMQ",
|
|
"promptIdx": 1,
|
|
"testIdx": 0,
|
|
"testCase": {
|
|
"vars": { "language": "French", "body": "Hello world" },
|
|
"assert": [],
|
|
"options": {},
|
|
"description": "Row #1",
|
|
"metadata": {}
|
|
},
|
|
"prompt": {
|
|
"raw": "Translate this to conversational French: Hello world",
|
|
"label": "examples/simple-csv/prompts.txt: Translate this to conversational {{language}}: {{body}}"
|
|
},
|
|
"promptId": "0ceccef5d40adcbf45bda7be485d2ee420e4d5943975932e7a38c5468d97c6ee",
|
|
"error": null,
|
|
"score": 1,
|
|
"success": true,
|
|
"providerResponse": {
|
|
"output": "Salut tout le monde !",
|
|
"tokenUsage": { "cached": 20, "total": 20 },
|
|
"cached": true,
|
|
"cost": 0.0000052500000000000006
|
|
},
|
|
"gradingResult": {
|
|
"pass": true,
|
|
"score": 1,
|
|
"reason": "No assertions",
|
|
"tokensUsed": { "total": 0, "prompt": 0, "completion": 0, "cached": 0 },
|
|
"assertion": null
|
|
},
|
|
"namedScores": {},
|
|
"provider": {
|
|
"id": "openai:gpt-4o-mini",
|
|
"label": "",
|
|
"config": { "basePath": "examples/simple-csv" }
|
|
},
|
|
"latencyMs": 7,
|
|
"cost": 0.0000052500000000000006
|
|
},
|
|
{
|
|
"id": "7791f530-38e0-431a-b8ef-fc2a64b14cb5",
|
|
"evalId": "eval-2024-10-02T21:37:20-kMQ",
|
|
"promptIdx": 1,
|
|
"testIdx": 1,
|
|
"testCase": {
|
|
"vars": { "language": "French", "body": "I'm hungry" },
|
|
"assert": [],
|
|
"options": {},
|
|
"description": "Row #2",
|
|
"metadata": {}
|
|
},
|
|
"prompt": {
|
|
"raw": "Translate this to conversational French: I'm hungry",
|
|
"label": "examples/simple-csv/prompts.txt: Translate this to conversational {{language}}: {{body}}"
|
|
},
|
|
"promptId": "0ceccef5d40adcbf45bda7be485d2ee420e4d5943975932e7a38c5468d97c6ee",
|
|
"error": null,
|
|
"score": 1,
|
|
"success": true,
|
|
"providerResponse": {
|
|
"output": "J'ai faim.",
|
|
"tokenUsage": { "cached": 19, "total": 19 },
|
|
"cached": true,
|
|
"cost": 0.0000046499999999999995
|
|
},
|
|
"gradingResult": {
|
|
"pass": true,
|
|
"score": 1,
|
|
"reason": "No assertions",
|
|
"tokensUsed": { "total": 0, "prompt": 0, "completion": 0, "cached": 0 },
|
|
"assertion": null
|
|
},
|
|
"namedScores": {},
|
|
"provider": {
|
|
"id": "openai:gpt-4o-mini",
|
|
"label": "",
|
|
"config": { "basePath": "examples/simple-csv" }
|
|
},
|
|
"latencyMs": 8,
|
|
"cost": 0.0000046499999999999995
|
|
},
|
|
{
|
|
"id": "1720b007-3e86-4920-a738-95fb0dcf2a3f",
|
|
"evalId": "eval-2024-10-02T21:37:20-kMQ",
|
|
"promptIdx": 0,
|
|
"testIdx": 1,
|
|
"testCase": {
|
|
"vars": { "language": "French", "body": "I'm hungry" },
|
|
"assert": [],
|
|
"options": {},
|
|
"description": "Row #2",
|
|
"metadata": {}
|
|
},
|
|
"prompt": {
|
|
"raw": "Rephrase this in French: I'm hungry",
|
|
"label": "examples/simple-csv/prompts.txt: Rephrase this in {{language}}: {{body}}"
|
|
},
|
|
"promptId": "ad447337160016cbd0a5c7f2626d0c1fe2ade0d4d653e20bed9250632bc56e03",
|
|
"error": null,
|
|
"score": 1,
|
|
"success": true,
|
|
"providerResponse": {
|
|
"output": "J'ai faim.",
|
|
"tokenUsage": { "cached": 19, "total": 19 },
|
|
"cached": true,
|
|
"cost": 0.0000046499999999999995
|
|
},
|
|
"gradingResult": {
|
|
"pass": true,
|
|
"score": 1,
|
|
"reason": "No assertions",
|
|
"tokensUsed": { "total": 0, "prompt": 0, "completion": 0, "cached": 0 },
|
|
"assertion": null
|
|
},
|
|
"namedScores": {},
|
|
"provider": {
|
|
"id": "openai:gpt-4o-mini",
|
|
"label": "",
|
|
"config": { "basePath": "examples/simple-csv" }
|
|
},
|
|
"latencyMs": 10,
|
|
"cost": 0.0000046499999999999995
|
|
},
|
|
{
|
|
"id": "4540d2da-9a5a-402e-9243-41be20f63b1f",
|
|
"evalId": "eval-2024-10-02T21:37:20-kMQ",
|
|
"promptIdx": 0,
|
|
"testIdx": 2,
|
|
"testCase": {
|
|
"vars": { "language": "Pirate", "body": "Hello world" },
|
|
"assert": [],
|
|
"options": {},
|
|
"description": "Row #3",
|
|
"metadata": {}
|
|
},
|
|
"prompt": {
|
|
"raw": "Rephrase this in Pirate: Hello world",
|
|
"label": "examples/simple-csv/prompts.txt: Rephrase this in {{language}}: {{body}}"
|
|
},
|
|
"promptId": "ad447337160016cbd0a5c7f2626d0c1fe2ade0d4d653e20bed9250632bc56e03",
|
|
"error": null,
|
|
"score": 1,
|
|
"success": true,
|
|
"providerResponse": {
|
|
"output": "Ahoy, me hearties!",
|
|
"tokenUsage": { "cached": 22, "total": 22 },
|
|
"cached": true,
|
|
"cost": 0.000006449999999999999
|
|
},
|
|
"gradingResult": {
|
|
"pass": true,
|
|
"score": 1,
|
|
"reason": "No assertions",
|
|
"tokensUsed": { "total": 0, "prompt": 0, "completion": 0, "cached": 0 },
|
|
"assertion": null
|
|
},
|
|
"namedScores": {},
|
|
"provider": {
|
|
"id": "openai:gpt-4o-mini",
|
|
"label": "",
|
|
"config": { "basePath": "examples/simple-csv" }
|
|
},
|
|
"latencyMs": 5,
|
|
"cost": 0.000006449999999999999
|
|
},
|
|
{
|
|
"id": "467d9521-22f8-45ae-a3c2-0777dd7b1bb0",
|
|
"evalId": "eval-2024-10-02T21:37:20-kMQ",
|
|
"promptIdx": 1,
|
|
"testIdx": 2,
|
|
"testCase": {
|
|
"vars": { "language": "Pirate", "body": "Hello world" },
|
|
"assert": [],
|
|
"options": {},
|
|
"description": "Row #3",
|
|
"metadata": {}
|
|
},
|
|
"prompt": {
|
|
"raw": "Translate this to conversational Pirate: Hello world",
|
|
"label": "examples/simple-csv/prompts.txt: Translate this to conversational {{language}}: {{body}}"
|
|
},
|
|
"promptId": "0ceccef5d40adcbf45bda7be485d2ee420e4d5943975932e7a38c5468d97c6ee",
|
|
"error": null,
|
|
"score": 1,
|
|
"success": true,
|
|
"providerResponse": {
|
|
"output": "Ahoy, me hearty! Greetings to ye, world!",
|
|
"tokenUsage": { "cached": 27, "total": 27 },
|
|
"cached": true,
|
|
"cost": 0.00000945
|
|
},
|
|
"gradingResult": {
|
|
"pass": true,
|
|
"score": 1,
|
|
"reason": "No assertions",
|
|
"tokensUsed": { "total": 0, "prompt": 0, "completion": 0, "cached": 0 },
|
|
"assertion": null
|
|
},
|
|
"namedScores": {},
|
|
"provider": {
|
|
"id": "openai:gpt-4o-mini",
|
|
"label": "",
|
|
"config": { "basePath": "examples/simple-csv" }
|
|
},
|
|
"latencyMs": 5,
|
|
"cost": 0.00000945
|
|
},
|
|
{
|
|
"id": "70b1ebd1-f748-4969-9d38-b6460ff6107f",
|
|
"evalId": "eval-2024-10-02T21:37:20-kMQ",
|
|
"promptIdx": 0,
|
|
"testIdx": 3,
|
|
"testCase": {
|
|
"vars": { "language": "Pirate", "body": "I'm hungry" },
|
|
"assert": [],
|
|
"options": {},
|
|
"description": "Row #4",
|
|
"metadata": {}
|
|
},
|
|
"prompt": {
|
|
"raw": "Rephrase this in Pirate: I'm hungry",
|
|
"label": "examples/simple-csv/prompts.txt: Rephrase this in {{language}}: {{body}}"
|
|
},
|
|
"promptId": "ad447337160016cbd0a5c7f2626d0c1fe2ade0d4d653e20bed9250632bc56e03",
|
|
"error": null,
|
|
"score": 1,
|
|
"success": true,
|
|
"providerResponse": {
|
|
"output": "Arrr, me belly be grumblin'!",
|
|
"tokenUsage": { "cached": 27, "total": 27 },
|
|
"cached": true,
|
|
"cost": 0.00000945
|
|
},
|
|
"gradingResult": {
|
|
"pass": true,
|
|
"score": 1,
|
|
"reason": "No assertions",
|
|
"tokensUsed": { "total": 0, "prompt": 0, "completion": 0, "cached": 0 },
|
|
"assertion": null
|
|
},
|
|
"namedScores": {},
|
|
"provider": {
|
|
"id": "openai:gpt-4o-mini",
|
|
"label": "",
|
|
"config": { "basePath": "examples/simple-csv" }
|
|
},
|
|
"latencyMs": 3,
|
|
"cost": 0.00000945
|
|
},
|
|
{
|
|
"id": "5b65ede9-f787-4637-b18d-9e0ff69dd778",
|
|
"evalId": "eval-2024-10-02T21:37:20-kMQ",
|
|
"promptIdx": 1,
|
|
"testIdx": 3,
|
|
"testCase": {
|
|
"vars": { "language": "Pirate", "body": "I'm hungry" },
|
|
"assert": [],
|
|
"options": {},
|
|
"description": "Row #4",
|
|
"metadata": {}
|
|
},
|
|
"prompt": {
|
|
"raw": "Translate this to conversational Pirate: I'm hungry",
|
|
"label": "examples/simple-csv/prompts.txt: Translate this to conversational {{language}}: {{body}}"
|
|
},
|
|
"promptId": "0ceccef5d40adcbf45bda7be485d2ee420e4d5943975932e7a38c5468d97c6ee",
|
|
"error": null,
|
|
"score": 1,
|
|
"success": true,
|
|
"providerResponse": {
|
|
"output": "Arrr, me belly be rumblin'! I be feelin' mighty hungry, matey!",
|
|
"tokenUsage": { "cached": 37, "total": 37 },
|
|
"cached": true,
|
|
"cost": 0.00001545
|
|
},
|
|
"gradingResult": {
|
|
"pass": true,
|
|
"score": 1,
|
|
"reason": "No assertions",
|
|
"tokensUsed": { "total": 0, "prompt": 0, "completion": 0, "cached": 0 },
|
|
"assertion": null
|
|
},
|
|
"namedScores": {},
|
|
"provider": {
|
|
"id": "openai:gpt-4o-mini",
|
|
"label": "",
|
|
"config": { "basePath": "examples/simple-csv" }
|
|
},
|
|
"latencyMs": 4,
|
|
"cost": 0.00001545
|
|
}
|
|
],
|
|
"prompts": [
|
|
{
|
|
"raw": "Rephrase this in {{language}}: {{body}}",
|
|
"label": "examples/simple-csv/prompts.txt: Rephrase this in {{language}}: {{body}}",
|
|
"id": "ad447337160016cbd0a5c7f2626d0c1fe2ade0d4d653e20bed9250632bc56e03",
|
|
"provider": "openai:gpt-4o-mini",
|
|
"metrics": {
|
|
"score": 4,
|
|
"testPassCount": 4,
|
|
"testFailCount": 0,
|
|
"assertPassCount": 0,
|
|
"assertFailCount": 0,
|
|
"totalLatencyMs": 21,
|
|
"tokenUsage": { "total": 87, "prompt": 0, "completion": 0, "cached": 87 },
|
|
"namedScores": {},
|
|
"namedScoresCount": {},
|
|
"cost": 0.000025199999999999996
|
|
}
|
|
},
|
|
{
|
|
"raw": "Translate this to conversational {{language}}: {{body}}",
|
|
"label": "examples/simple-csv/prompts.txt: Translate this to conversational {{language}}: {{body}}",
|
|
"id": "0ceccef5d40adcbf45bda7be485d2ee420e4d5943975932e7a38c5468d97c6ee",
|
|
"provider": "openai:gpt-4o-mini",
|
|
"metrics": {
|
|
"score": 4,
|
|
"testPassCount": 4,
|
|
"testFailCount": 0,
|
|
"assertPassCount": 0,
|
|
"assertFailCount": 0,
|
|
"totalLatencyMs": 24,
|
|
"tokenUsage": { "total": 103, "prompt": 0, "completion": 0, "cached": 103 },
|
|
"namedScores": {},
|
|
"namedScoresCount": {},
|
|
"cost": 0.0000348
|
|
}
|
|
}
|
|
],
|
|
"datasetId": "27018c2ea4e47d8af2fc1b2d69c131ed777f1d89e9069bf38622701b7e1e786a"
|
|
}
|