* Setting all random seeds to  0 *
Loading model: out_models_paper/net_algebraic+biases_rep1_best.pt on cuda
 Loading model that has completed (or started) 49 of 50 epochs
  test episode_type: few_shot_human
  batch size: 25
  max eval length: 10
  number of steps: 194700
  best val loss achieved: 0.0257

BIML specs:
 nparams= 1393801
 nlayers_encoder= 3
 nlayers_decoder= 3
 nhead= 8
 hidden_size= 128
 dim_feedforward= 512
 act_feedforward= gelu
 dropout= 0.1
 

Fitting for the best value of p_lapse use log-like...
  Each value is replicated across 100 random runs/permutations
 p_lapse 0.01 :
* Setting all random seeds to  0 *
  run 0
  run 20
  run 40
  run 60
  run 80
  loglike: M = -364.9 (SD= 2.6308 , Nrep= 100 ) for 990.0 symbol predictions on average
    ave LL by cell:  -0.3686
 p_lapse 0.02 :
* Setting all random seeds to  0 *
  run 0
  run 20
  run 40
  run 60
  run 80
  loglike: M = -359.5187 (SD= 2.4773 , Nrep= 100 ) for 990.0 symbol predictions on average
    ave LL by cell:  -0.3632
 p_lapse 0.03 :
* Setting all random seeds to  0 *
  run 0
  run 20
  run 40
  run 60
  run 80
  loglike: M = -358.1092 (SD= 2.3792 , Nrep= 100 ) for 990.0 symbol predictions on average
    ave LL by cell:  -0.3617
 p_lapse 0.04 :
* Setting all random seeds to  0 *
  run 0
  run 20
  run 40
  run 60
  run 80
  loglike: M = -358.7306 (SD= 2.3069 , Nrep= 100 ) for 990.0 symbol predictions on average
    ave LL by cell:  -0.3624
* BEST FIT * p_lapse= 0.03 with mean loglike score of -358.1092 (or -0.3617 per cell)
