[16:32:02] longeval: T=32768 prompt_len=16360 max_new=24 n=100 greedy=1 gpus=0,1 [16:32:02] haystack: /root/Adhi/BluTrain/dist/Context_Parallelism/lendata/haystack/b2val_haystack.bin [16:32:02] prompt signature changed -- regenerating prompts and clearing state [16:32:02] generating prompts: niah_single [16:32:03] fan-out: 2 GPU(s), NP=1 -> 2 concurrent job(s) [16:32:03] 2 job(s) to run [16:32:03] [gpu 0] ext32k / niah_single [16:32:03] [gpu 1] l2 / niah_single [16:37:38] scoring ext32k / niah_single === niah_single (prompt_len=16360, n=100) === accuracy: 45/100 = 45.0% depth acc (needle position in context) 0.00 2/20 10.0% 0.25 7/20 35.0% 0.50 9/20 45.0% 0.75 14/20 70.0% 1.00 13/20 65.0% --- 2 sample generations --- [MISS] expect=8056020 got= 8.\nQuestion: What is the special magic number for yarrow?\n [HIT ] expect=5343902 got= 5343902.\n wrote ext32k_len_16360/scores/ext32k_niah_single.csv [16:37:39] scoring l2 / niah_single === niah_single (prompt_len=16360, n=100) === accuracy: 20/100 = 20.0% 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 3/20 15.0% 1.00 17/20 85.0% --- 2 sample generations --- [MISS] expect=8056020 got= the magic number for the yarrow.\n [MISS] expect=5343902 got= the number of the cactus.\n wrote ext32k_len_16360/scores/l2_niah_single.csv [16:37:39] building summary === per-task accuracy % (secondary/descriptive: ~10 per depth cell) === task ext32k l2 niah_single 45.0 20.0 === PRIMARY: accuracy % pooled by depth across tasks (+-95% Wilson) === depth ext32k l2 0.00 10.0+-13.7 0.0+- 8.1 0.25 35.0+-19.3 0.0+- 8.1 0.50 45.0+-20.0 0.0+- 8.1 0.75 70.0+-18.7 15.0+-15.4 1.00 65.0+-19.3 85.0+-15.4 wrote ext32k_len_16360/summary.csv and ext32k_len_16360/depth_summary.csv [16:37:39] DONE -> ext32k_len_16360/summary.csv