Artificial Intelligence

Out of context: Reply #1632

• Started
• Last post
• 2,578 Responses

• ********

  -2

  Models in Aidanbench

  

  "Aidan Bench
  Some models feel competent despite under-scoring on benchmarks like MMLU, GPQA, MATH, or NIAH.

  Aidan Bench rewards:
  - Creativity
  - Reliability
  - Contextual attention
  - Instruction following
  Aidan Bench is weakly correlated with Lmsys, has no score ceiling, and aligns with real-world open-ended use."

  ********

  -2Permalink

  UpvoteDownvote

  Flag

  Show [[ numHiddenNotes ]] more notesAdd Note

  SaveCancel

View thread