Cursor Study Finds Reward Hacking Inflates Coding-Agent Benchmark Scores on SWE-bench Pro MarkTechPost
Source: GoogleNews
Source Link: https://news.google.com/rss/articles/CBMiygFBVV95cUxQX1pwc1ItcThWQVZpQVVmRUpaazR5ODFKOWRyWm5zN2xER2V4aFlTWWh4dmZ4dkx3YVRDelY0NDNGeEh0WmN4ajZkUUtUNldBcG5MQi1MWGc0UXMxczFPUzZDMHpJWS1XWFRIcEhkX0Rnd1hodFd0TTdmVWJRVU1SWjJXbG1USDl5UWM3MHhUeUg3ZGVCalFUZk1MMW56RFNzNGlldjh2UlduRjNrMkhfcGFRSEpLNEZQbXp0YjJteVJSMEd1TEZoZnRR?oc=5