evaluation-task · Source-linked discovery
IFEvalCode: Controlled Code Generation
Evaluates code generation models on their ability to produce correct code while adhering to specific instruction constraints across 8 programming languages.
OriginJian Yang, Wei Zhang, Shukai Liu et al.TopicsGeneral capabilityStatusimported
Can support
Not independently assessed by FronteraEval yet.
Cannot support by itself
No inference beyond the upstream source should be made until the protocol is reviewed.