============================================================
WORLD MODEL EVALUATION STATISTICS
============================================================

Model: xlangai/OpenCUA-7B + ageppert/world-model-7b-lora
Validation examples: 4210
Total inference time: 28496.8s (474.9 min)
Average time per example: 6.77s

Prediction lengths (characters):
  min:    447
  median: 795
  mean:   878
  max:    2964

Ground truth lengths (characters):
  min:    340
  median: 796
  mean:   802
  max:    1367

Empty predictions: 0 (0.0%)

Generation config:
  max_new_tokens: 512
  temperature: 0.1
  top_p: 0.9