evals_L_std.json

{
 "ckpt": "ckpts/L_std_best.pt",
 "limit": 1000,
 "results": {
  "hellaswag": 0.398,
  "piqa": 0.69,
  "arc_easy": 0.5351,
  "arc_challenge": 0.3077,
  "winogrande": 0.499,
  "boolq": 0.615,
  "sciq": 0.61,
  "lambada": 0.324,
  "mean": 0.4974
 }
}