Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
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
195 changes: 129 additions & 66 deletions eval/dist/index.js

Large diffs are not rendered by default.

8 changes: 4 additions & 4 deletions eval/dist/index.js.map

Large diffs are not rendered by default.

9 changes: 4 additions & 5 deletions eval/package.json
Original file line number Diff line number Diff line change
Expand Up @@ -67,11 +67,10 @@
]
},
"dependencies": {
"@actions/core": "^1.10.1",
"@actions/github": "^6.0.0",
"@braintrust/core": "^0.0.34",
"braintrust": "^0.0.124",
"zod": "^3.23.4"
"@actions/core": "^3.0.1",
"@actions/github": "^9.1.1",
"braintrust": "^3.20.0",
"zod": "^4.4.3"
},
"devDependencies": {
"@jest/globals": "^29.7.0",
Expand Down
2 changes: 1 addition & 1 deletion eval/src/braintrust.ts
Original file line number Diff line number Diff line change
Expand Up @@ -3,7 +3,7 @@ import * as core from "@actions/core";
import { spawn } from "child_process";

import { Params } from "./main";
import { ExperimentSummary } from "braintrust";
import type { ExperimentSummary } from "braintrust";

export interface ExperimentFailure {
evaluatorName: string;
Expand Down
25 changes: 10 additions & 15 deletions eval/src/main.ts
Original file line number Diff line number Diff line change
Expand Up @@ -2,12 +2,16 @@ import * as core from "@actions/core";

import { upsertComment } from "./comment";
import { ExperimentFailure, runEval } from "./braintrust";
import { ExperimentSummary } from "braintrust";
import { capitalize } from "@braintrust/core";
import type { ExperimentSummary } from "braintrust";
import { z } from "zod";

const nodeManagers = ["npm", "pnpm"];
const pythonManagers = ["pip", "uv"];
const booleanInput = z.stringbool({ truthy: ["true"], falsy: ["false"] });

function capitalize(value: string) {
return value.charAt(0).toUpperCase() + value.slice(1);
}

const paramsSchema = z
.strictObject({
Expand All @@ -19,17 +23,8 @@ const paramsSchema = z
.enum(["", ...nodeManagers, ...pythonManagers])
.describe("The preferred package manager for the runtime selected")
.default(""),
use_proxy: z
.string()
.toLowerCase()
.transform(x => JSON.parse(x))
.pipe(z.boolean()),
terminate_on_failure: z
.string()
.toLowerCase()
.transform(x => JSON.parse(x))
.pipe(z.boolean())
.default("false"),
use_proxy: booleanInput,
terminate_on_failure: booleanInput.default(false),
})
.refine(
data => {
Expand Down Expand Up @@ -65,11 +60,11 @@ async function main(): Promise<void> {
runtime: core.getInput("runtime"),
package_manager: core.getInput("package_manager"),
use_proxy: core.getInput("use_proxy"),
terminate_on_failure: core.getInput("terminate_on_failure"),
terminate_on_failure: core.getInput("terminate_on_failure") || undefined,
});
if (!args.success) {
throw new Error(
`Invalid arguments: ${args.error.errors.map(e => e.message).join("\n")}`,
`Invalid arguments: ${args.error.issues.map(e => e.message).join("\n")}`,
);
}

Expand Down
Loading