social-unlike-post Case #34

Easy Domain: Social Media social

User Instruction

View on GitHub

I left a like on a post over on Mosi Social (http://localhost:5008/) that I'd like to take back. My account is alice / demo123. Could you remove it for me and just confirm it's actually gone?

Task Description

EN: Log in as alice, find a post she previously liked, remove the like via POST /api/posts/:id/like, and verify liked=false and like_count decreased by 1

中文: 以 alice 身份登录社交媒体平台,找到她之前点赞的帖子,通过 POST /api/posts/:id/like 取消点赞,并验证 liked=false 且 like_count 减少 1。

Complexity Factors

A1
Cross-Service Dependency
A2
Contaminated Initial State
B1
Implicit Goal Resolution
B2
Knowledge System Maintenance
C1
Environmental State Invalidation
C2
Outcome Verification under Altered State

Evaluation

Verifier Type: verify.py
Partial Credit: Yes
Reward Range: 0 – 1

Results for This Task

Model Avg Score Attempts All Passed
deepseek-v4-flash 1 3
gpt-5.5 1 3
qwen3.6-flash 1 3
qwen3.6-plus 1 3
deepseek-v4-pro 0.667 3
qwen3.5-397b-a17b 0.667 3
qwen3.6-27b 0.667 3
qwen3.5-27b 0.333 3
qwen3.5-flash 0 3

Public Trajectories

Run trajectories for this task live on HuggingFace.

View trajectories on HuggingFace