evals-lab 0.2.0 → 0.3.0
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 +66 -0
- package/README.md +17 -9
- package/lab/VERSION +1 -1
- package/lab/demo/datasets/demo-1.json +2187 -1644
- package/lab/demo/datasets/demo-2.json +1870 -1485
- package/lab/demo/manifest.json +2 -2
- package/lab/demo/pipelines/demo-1.json +3 -7
- package/lab/demo/pipelines/demo-2.json +3 -7
- package/lab/evals-core.mjs +429 -434
- package/lab/kinds/list.mjs +1 -1
- package/lab/metrics/builtin.mjs +96 -18
- package/lab/run-evals.js +35 -33
- package/lab/server.py +296 -82
- package/lab/web/dist/assets/{gallery-o7c4lfpn.css → gallery-DFeJkfUw.css} +1 -1
- package/lab/web/dist/assets/gallery-DRBlZ8mP.js +3 -0
- package/lab/web/dist/assets/main-DMpQmW8l.js +21 -0
- package/lab/web/dist/assets/main-wfqC6HcM.css +1 -0
- package/lab/web/dist/assets/tokens-BEn6_hVz.css +1 -0
- package/lab/web/dist/assets/tokens-iGpvbD5R.js +55 -0
- package/lab/web/dist/gallery.html +4 -4
- package/lab/web/dist/index.html +4 -4
- package/package.json +3 -2
- package/lab/web/dist/assets/gallery-DsetJSXv.js +0 -3
- package/lab/web/dist/assets/main-B-VtDGxC.css +0 -1
- package/lab/web/dist/assets/main-ERoD2Kll.js +0 -19
- package/lab/web/dist/assets/tokens-ClRQ7Mui.js +0 -51
- package/lab/web/dist/assets/tokens-D3C8O2Ib.css +0 -1
package/lab/demo/manifest.json
CHANGED
|
@@ -3,9 +3,9 @@
|
|
|
3
3
|
"version": 1,
|
|
4
4
|
"id": "demo",
|
|
5
5
|
"name": "Demo",
|
|
6
|
-
"packVersion": "2026.
|
|
6
|
+
"packVersion": "2026.10.02",
|
|
7
7
|
"requires": {
|
|
8
|
-
"lab": ">=
|
|
8
|
+
"lab": ">=12",
|
|
9
9
|
"plugins": []
|
|
10
10
|
},
|
|
11
11
|
"description": "Two graded sets of public-domain images, each with a starter pipeline."
|
|
@@ -1,5 +1,5 @@
|
|
|
1
1
|
{
|
|
2
|
-
"version":
|
|
2
|
+
"version": 12,
|
|
3
3
|
"id": "demo-1",
|
|
4
4
|
"name": "Demo 1",
|
|
5
5
|
"jobs": [
|
|
@@ -90,7 +90,7 @@
|
|
|
90
90
|
]
|
|
91
91
|
}
|
|
92
92
|
],
|
|
93
|
-
"
|
|
93
|
+
"evals": [
|
|
94
94
|
{
|
|
95
95
|
"id": "t1",
|
|
96
96
|
"name": "",
|
|
@@ -104,11 +104,7 @@
|
|
|
104
104
|
"name": "Demo 1"
|
|
105
105
|
},
|
|
106
106
|
"over": "item",
|
|
107
|
-
"metrics": [
|
|
108
|
-
{
|
|
109
|
-
"type": "case"
|
|
110
|
-
}
|
|
111
|
-
]
|
|
107
|
+
"metrics": []
|
|
112
108
|
}
|
|
113
109
|
]
|
|
114
110
|
}
|
|
@@ -1,5 +1,5 @@
|
|
|
1
1
|
{
|
|
2
|
-
"version":
|
|
2
|
+
"version": 12,
|
|
3
3
|
"id": "demo-2",
|
|
4
4
|
"name": "Demo 2",
|
|
5
5
|
"jobs": [
|
|
@@ -90,7 +90,7 @@
|
|
|
90
90
|
]
|
|
91
91
|
}
|
|
92
92
|
],
|
|
93
|
-
"
|
|
93
|
+
"evals": [
|
|
94
94
|
{
|
|
95
95
|
"id": "t1",
|
|
96
96
|
"name": "",
|
|
@@ -104,11 +104,7 @@
|
|
|
104
104
|
"name": "Demo 2"
|
|
105
105
|
},
|
|
106
106
|
"over": "item",
|
|
107
|
-
"metrics": [
|
|
108
|
-
{
|
|
109
|
-
"type": "case"
|
|
110
|
-
}
|
|
111
|
-
]
|
|
107
|
+
"metrics": []
|
|
112
108
|
}
|
|
113
109
|
]
|
|
114
110
|
}
|