[08:05:15] longeval: T=16384 prompt_len=2048 max_new=24 n=100 greedy=1 gpus=0,1,2,3
[08:05:15] haystack: Data_Loader/Data/flux_val_000000.bin
[08:05:15] prompt signature changed -- regenerating prompts and clearing state
[08:05:15] generating prompts: niah_single
[08:05:15] generating prompts: val_backward
[08:05:16] fan-out: 4 GPU(s), NP=1 -> 8 concurrent job(s)
[08:05:16] 4 job(s) to run
[08:05:16]   [gpu 0] base_c / niah_single
[08:05:16]   [gpu 1] base_c / val_backward
[08:05:16]   [gpu 2] armB_c / niah_single
[08:05:16]   [gpu 3] armB_c / val_backward
[08:09:10]   scoring base_c / niah_single

=== niah_single  (prompt_len=2048, n=100) ===
  accuracy: 82/100 = 82.0%
    depth      acc   (needle position in context)
     0.00  11/20   55.0%
     0.25  14/20   70.0%
     0.50  19/20   95.0%
     0.75  19/20   95.0%
     1.00  19/20   95.0%

  --- 2 sample generations ---
  [MISS] expect=8056020
         got= 7.\nScum (1979 - film version)\n
  [MISS] expect=5343902
         got= 3.\nQuestion: How do I find a reliable cactus rental service?\n

  wrote newck_len_2048/scores/base_c_niah_single.csv
[08:09:11]   scoring base_c / val_backward

=== val_backward  (prompt_len=2048, n=100) ===
  accuracy: 1/100 = 1.0%
    depth      acc   (needle position in context)
     0.00   0/20    0.0%
     0.25   0/20    0.0%
     0.50   0/20    0.0%
     0.75   0/20    0.0%
     1.00   1/20    5.0%

  --- 2 sample generations ---
  [MISS] expect=yarrow
         got= the\nscum (1979 - film version)\n
  [MISS] expect=cactus
         got= the Magic Wheel. It is a unique and powerful magic number that can transform any ordinary object into a magical force.\n

  wrote newck_len_2048/scores/base_c_val_backward.csv
[08:09:11]   scoring armB_c / niah_single

=== niah_single  (prompt_len=2048, n=100) ===
  accuracy: 73/100 = 73.0%
    depth      acc   (needle position in context)
     0.00   8/20   40.0%
     0.25  15/20   75.0%
     0.50  18/20   90.0%
     0.75  16/20   80.0%
     1.00  16/20   80.0%

  --- 2 sample generations ---
  [MISS] expect=8056020
         got= 7.\nThe special magic number for yarrow is 7.\n
  [MISS] expect=5343902
         got= 2.\nQuestion: How does the magic number for cactus work?\n

  wrote newck_len_2048/scores/armB_c_niah_single.csv
[08:09:11]   scoring armB_c / val_backward

=== val_backward  (prompt_len=2048, n=100) ===
  accuracy: 0/100 = 0.0%
    depth      acc   (needle position in context)
     0.00   0/20    0.0%
     0.25   0/20    0.0%
     0.50   0/20    0.0%
     0.75   0/20    0.0%
     1.00   0/20    0.0%

  --- 2 sample generations ---
  [MISS] expect=yarrow
         got= the\nscum (1979 - film version)\n
  [MISS] expect=cactus
         got= the Magic Number 5343902.\n

  wrote newck_len_2048/scores/armB_c_val_backward.csv
[08:09:11] building summary

=== per-task accuracy %  (secondary/descriptive: ~10 per depth cell) ===
  task                 armB_c   base_c
  niah_single            73.0     82.0
  val_backward            0.0      1.0

  degenerate (<=1% for every model): ['val_backward']

=== PRIMARY: accuracy % pooled by depth across tasks (+-95% Wilson) ===
  depth             armB_c          base_c
  0.00          40.0+-19.7      55.0+-20.0
  0.25          75.0+-17.8      70.0+-18.7
  0.50          90.0+-13.7      95.0+-11.4
  0.75          80.0+-16.8      95.0+-11.4
  1.00          80.0+-16.8      95.0+-11.4

  wrote newck_len_2048/summary.csv and newck_len_2048/depth_summary.csv
[08:09:11] DONE -> newck_len_2048/summary.csv