| |
| |
| |
| |
| |
| import { isBioMniBenchTask, loadBioMniBenchManifest, resolveSharedPython, SHARED_VENV_DIR, QWEN_JUDGE_SCRIPT } from '../src/harness/evaluation/biomnibenchAdapter.ts' |
| import { resolveTaskRuntime } from '../src/harness/evaluation/sourceRuntimeResolver.ts' |
| import { existsSync } from 'fs' |
|
|
| const BIOMNI_ROOT = '/data/yjh/biomnibench-organized' |
| const SAMPLES = ['da-1-3', 'da-3-1', 'da-14-1'] |
|
|
| console.log('Shared venv:', SHARED_VENV_DIR, existsSync(SHARED_VENV_DIR) ? 'β
' : 'β') |
| console.log('Shared Python:', resolveSharedPython(), '\n') |
| console.log('Qwen judge script:', QWEN_JUDGE_SCRIPT, existsSync(QWEN_JUDGE_SCRIPT) ? 'β
' : 'β', '\n') |
|
|
| for (const id of SAMPLES) { |
| const dir = `${BIOMNI_ROOT}/${id}` |
| console.log(`=== ${id} (${dir}) ===`) |
| console.log(` exists: ${existsSync(dir) ? 'β
' : 'β'}`) |
| console.log(` isBioMniBench: ${isBioMniBenchTask(dir)}`) |
| if (existsSync(dir)) { |
| const manifest = await loadBioMniBenchManifest(dir, id) |
| console.log(` manifest:`, JSON.stringify(manifest, null, 2)) |
| } |
|
|
| |
| const fakePublic = `${dir}` |
| |
| |
| const runtime = await resolveTaskRuntime(fakePublic) |
| console.log(` resolveTaskRuntime.ok=${runtime.ok}`) |
| if (runtime.ok) { |
| console.log(` resolveTaskRuntime.python=${runtime.python}`) |
| console.log(` resolveTaskRuntime.envName=${runtime.envName}`) |
| } else { |
| console.log(` resolveTaskRuntime.error=${runtime.error}`) |
| } |
| console.log('') |
| } |
|
|