tyrin/judge_bench
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

server-inference push judge_bench

Browse Source
This commit is contained in:
Tyrin Todd
2026-05-03 19:24:15 -07:00
parent f9ace01bcf
commit ccbae1da8d
2 changed files with 8 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
program.json
View File

@@ -3,7 +3,7 @@
"train": [],
"demos": [],
"signature": {
"instructions": "Reason step-by-step to determine which response is better for the given context.\nAs you reason, note any uncertainties, ambiguities, and gaps \u2014 both in how the instructions apply to the task at hand and in whether you have the knowledge needed to solve it. Be honest in your reasoning when you are unsure about your answer. Your reasoning must lead directly to\nthe better response (A or B).",
"instructions": "Reason step-by-step to compare the two responses for the given context.\nAs you reason, note any uncertainties, ambiguities, and gaps \u2014 both in how the instructions apply to the task at hand and in whether you have the knowledge needed to solve it. Be honest in your reasoning when you are unsure about your answer.\n\nChoose 'A>B' if response A is meaningfully better, 'B>A' if response B is\nmeaningfully better, or 'tie' if neither response is clearly better than the other.",
"fields": [
{
"prefix": "Question:",
@@ -23,7 +23,7 @@
},
{
"prefix": "Label:",
"description": "The label of the better response ('A' or 'B')."
"description": "'A>B' if A is better, 'B>A' if B is better, 'tie' if neither is clearly better."
}
]
},