Skip to content

✨ add groundtruth test output support to promptpex #180

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

Merged
bzorn merged 3 commits into from
Jun 20, 2025
Merged

✨ add groundtruth test output support to promptpex #180

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
0d5bc68
✨ add groundtruth test output support to promptpex
bzorn Jun 20, 2025
cfc8a22
✏️ fix typo in PromptPexContext groundtruth comment
bzorn Jun 20, 2025
a409405
Apply suggestions from code review
pelikhan Jun 20, 2025
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
10 changes: 5 additions & 5 deletions package.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -95,11 +95,11 @@
"promptpex:test-st-def:ollama": "genaiscript run promptpex \"samples/speech-tag/speech-tag.prompty\" --vars \"effort=min\" --vars \"out=evals/test-st-def\" --env .env.ollama",
"promptpex:test-st-mingt": "genaiscript run promptpex \"samples/speech-tag/speech-tag.prompty\" --vars \"effort=min\" --vars \"groundtruthModel=azure:gpt-4.1-mini_2025-04-14\" --vars \"evalModel=ollama:llama3.3\" --vars \"evals=true\" --vars \"compliance=true\" --vars \"baselineTests=false\" --vars \"modelsUnderTest=ollama:llama3.3\" --vars \"out=evals/test-st-mingt\"",
"promptpex:test-st-mediumgt": "genaiscript run promptpex \"samples/speech-tag/speech-tag-multi.prompty\" --vars \"effort=medium\" --vars \"groundtruthModel=azure:gpt-4.1-mini_2025-04-14\" --vars \"evalModel=ollama:llama3.3\" --vars \"evals=true\" --vars \"compliance=true\" --vars \"modelsUnderTest=azure:gpt-4.1-mini_2025-04-14;ollama:llama3.3\" --vars \"out=evals/test-st-mediumgt\"",
"promptpex:test-st-min-gen": "genaiscript run promptpex \"samples/speech-tag/speech-tag.prompty\" --vars \"effort=min\" --vars \"groundtruthModel=azure:gpt-4.1-mini_2025-04-14\" --vars \"evals=false\" --vars \"compliance=false\" --vars \"baselineTests=false\" --vars \"evalModelGroundtruth=azure:gpt-4.1-mini_2025-04-14;ollama:llama3.3\" --vars \"out=evals/test-st-min-gen\" --env .env.ollama",
"promptpex:test-st-min-run": "genaiscript run promptpex \"evals/test-st-min-gen/speech-tag/promptpex_context.json\" --vars \"evals=false\" --vars \"compliance=false\" --vars \"baselineTests=false\" --vars \"modelsUnderTest=ollama:qwen2.5:3b;ollama:llama3.2:1b;ollama:llama3.3\" --vars \"out=evals/test-st-min-run\"",
"promptpex:test-st-min-eval": "genaiscript run promptpex \"evals/test-st-min-run/speech-tag/promptpex_context.json\" --vars \"evals=true\" --vars \"compliance=true\" --vars \"baselineTests=false\" --vars \"evalModel=azure:gpt-4.1-mini_2025-04-14\" --vars \"out=evals/test-st-min-eval\"",
"promptpex:test-st-min-eval1": "genaiscript run promptpex \"evals/test-st-min-run/speech-tag/promptpex_context.json\" --vars \"evals=true\" --vars \"compliance=true\" --vars \"baselineTests=false\" --vars \"evalModel=azure:gpt-4.1-mini_2025-04-14;ollama:llama3.3\" --vars \"out=evals/test-st-min-eval\"",
"promptpex:test-st-min-runeval": "genaiscript run promptpex \"evals/test-st-min-gen/speech-tag/promptpex_context.json\" --vars \"evals=true\" --vars \"compliance=true\" --vars \"baselineTests=false\" --vars \"modelsUnderTest=ollama:qwen2.5:3b;ollama:llama3.2:1b\" --vars \"evalModel=ollama:llama3.3\" --vars \"out=evals/test-st-min-runeval\"",
"promptpex:test-st-min-gen:ollama": "genaiscript run promptpex \"samples/speech-tag/speech-tag.prompty\" --vars \"effort=min\" --vars \"groundtruthModel=azure:gpt-4.1-mini_2025-04-14\" --vars \"evals=false\" --vars \"compliance=false\" --vars \"baselineTests=false\" --vars \"evalModelGroundtruth=azure:gpt-4.1-mini_2025-04-14;ollama:llama3.3\" --vars \"out=evals/test-st-min-gen\" --env .env.ollama",
"promptpex:test-st-min-run:ollama": "genaiscript run promptpex \"evals/test-st-min-gen/speech-tag/promptpex_context.json\" --vars \"evals=false\" --vars \"compliance=false\" --vars \"baselineTests=false\" --vars \"modelsUnderTest=ollama:qwen2.5:3b;ollama:llama3.2:1b;ollama:llama3.3\" --vars \"out=evals/test-st-min-run --env .env.ollama\"",
Copy link

Copilot AI Jun 20, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It appears that the --env argument is included inside the --vars quoted string. Adjust the script so that --env .env.ollama is placed outside the --vars value.

Suggested change
"promptpex:test-st-min-run:ollama": "genaiscript run promptpex \"evals/test-st-min-gen/speech-tag/promptpex_context.json\" --vars \"evals=false\" --vars \"compliance=false\" --vars \"baselineTests=false\" --vars \"modelsUnderTest=ollama:qwen2.5:3b;ollama:llama3.2:1b;ollama:llama3.3\" --vars \"out=evals/test-st-min-run --env .env.ollama\"",
"promptpex:test-st-min-run:ollama": "genaiscript run promptpex \"evals/test-st-min-gen/speech-tag/promptpex_context.json\" --vars \"evals=false\" --vars \"compliance=false\" --vars \"baselineTests=false\" --vars \"modelsUnderTest=ollama:qwen2.5:3b;ollama:llama3.2:1b;ollama:llama3.3\" --vars \"out=evals/test-st-min-run\" --env .env.ollama",

Copilot uses AI. Check for mistakes.
"promptpex:test-st-min-eval:ollama": "genaiscript run promptpex \"evals/test-st-min-run/speech-tag/promptpex_context.json\" --vars \"evals=true\" --vars \"compliance=true\" --vars \"baselineTests=false\" --vars \"evalModel=azure:gpt-4.1-mini_2025-04-14\" --vars \"out=evals/test-st-min-eval\" --env .env.ollama",
"promptpex:test-st-min-eval1": "genaiscript run promptpex \"evals/test-st-min-run/speech-tag/promptpex_context.json\" --vars \"evals=true\" --vars \"compliance=true\" --vars \"baselineTests=false\" --vars \"evalModel=azure:gpt-4.1-mini_2025-04-14;ollama:llama3.3\" --vars \"out=evals/test-st-min-eval\" --env .env.ollama",
"promptpex:test-st-min-runeval:ollama": "genaiscript run promptpex \"evals/test-st-min-gen/speech-tag/promptpex_context.json\" --vars \"evals=true\" --vars \"compliance=true\" --vars \"baselineTests=false\" --vars \"modelsUnderTest=ollama:qwen2.5:3b;ollama:llama3.2:1b\" --vars \"evalModel=ollama:llama3.3\" --vars \"out=evals/test-st-min-runeval\" --env .env.ollama",
"promptpex:test-st-med": "genaiscript run promptpex \"samples/speech-tag/speech-tag.prompty\" --vars \"effort=medium\" --vars \"evals=true\" --vars \"compliance=true\" --vars \"baselineTests=false\" --vars \"modelsUnderTest=ollama:llama3.2:1b\" --vars \"out=evals/test-st-med\"",
"promptpex:test-headline": "genaiscript run promptpex \"samples/demo/rate-headline.prompty\" --vars \"effort=min\" --vars \"evals=true\" --vars \"compliance=true\" --vars \"baselineTests=false\" --vars \"modelsUnderTest=ollama:llama3.2:1b\" --vars \"out=evals/test-headline\"",
"promptpex:test1": "genaiscript run promptpex \"samples/demo/rate-headline.prompty\" --vars \"cache=true\" --vars \"evals=false\" --vars \"effort=min\" --vars \"baselineTests=true\" --vars \"filterTestCount=3\" --vars \"modelsUnderTest=ollama:llama3.2:1b\" --vars \"out=evals/test-test1\"",
Expand Down
3 changes: 3 additions & 0 deletions src/genaisrc/src/loaders.mts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -112,6 +112,7 @@ export async function loadPromptContext(
let rateTests = path.join(dir, "test_collection_review.md")
let testData = path.join(dir, "test_data.json")
let testResults = path.join(dir, "test_results.json")
let groundtruthResults = path.join(dir, "test_groundtruth.json")
let testEvals = path.join(dir, "test_evals.json")
let baselineTestEvals = path.join(dir, "baseline_test_evals.json")
let ruleEvals = path.join(dir, "rule_evals.json")
Expand Down Expand Up @@ -193,6 +194,7 @@ export async function loadPromptContext(
prompt: promptFile,
originalPrompt: originalPromptFile,
testOutputs: await workspace.readText(testResults),
groundtruthOutputs: await workspace.readText(groundtruthResults),
intent: await workspace.readText(intent),
inputSpec: await workspace.readText(inputSpec),
rules: tidyRulesFile(await workspace.readText(rules)),
Expand Down Expand Up @@ -372,6 +374,7 @@ async function loadPromptContextFromJSON(
ctx.rateTests,
ctx.testData,
ctx.testOutputs,
ctx.groundtruthOutputs,
ctx.testEvals,
ctx.baselineTestEvals,
ctx.ruleEvals,
Expand Down
9 changes: 9 additions & 0 deletions src/genaisrc/src/parsers.mts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -139,6 +139,8 @@ export function parseTestResults(
const rules = parseRules(files.rules.content)
const res = (parsers.JSON5(files.testOutputs.content) ||
[]) as PromptPexTestResult[]
const groundtruthRes = (parsers.JSON5(files.groundtruthOutputs.content) ||
[]) as PromptPexTestResult[]
res.forEach((r) => {
r.inverse =
r.ruleid !== null && parseInt(r.ruleid as any) > rules.length
Expand All @@ -151,6 +153,13 @@ export function parseTestResults(
if (diagnostics)
throw new Error(`missing 'model' for test result ${r.id}`)
}
for (const r of groundtruthRes.filter((r) => !r.error && !r.model)) {
output.warn(
`missing 'model' for groundtruth test result ${r.id} in ${files.groundtruthOutputs.filename}`
)
if (diagnostics)
throw new Error(`missing 'model' for groundtruth test result ${r.id}`)
}
for (const r of res) if (isNaN(r.ruleid)) r.ruleid = null
return res
}
Expand Down
4 changes: 2 additions & 2 deletions src/genaisrc/src/promptpex.mts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -254,10 +254,10 @@ export async function promptpexGenerate(files: PromptPexContext) {
renderTestResults(results.filter((r) => r.isGroundtruth)),
{ maxRows: 12 }
)
files.testOutputs.content = JSON.stringify(results, null, 2)
files.groundtruthOutputs.content = JSON.stringify(results, null, 2)
if (files.writeResults)
await workspace.writeText(
files.testOutputs.filename,
files.groundtruthOutputs.filename,
JSON.stringify(results, null, 2)
)
} else {
Expand Down
5 changes: 5 additions & 0 deletions src/genaisrc/src/types.mts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -284,6 +284,11 @@ export interface PromptPexContext {
*/
testOutputs: WorkspaceFile

/**
* Groundtruth Output (TO) - Test results with metrics for groundtruth test outputs
*/
groundtruthOutputs: WorkspaceFile

/**
* Coverage and validate test evals
*/
Expand Down