File size: 533 Bytes
24df05f | 1 2 3 4 5 6 7 8 9 10 | [07:59:27] === l3 decision eval ===
[07:59:27] models: armA_c l1 l2 l3 haystack: b2val (in-distribution for l2 -- biased toward l2)
[07:59:27] PROMPT_LEN=4096 niah_single N=100
[08:04:43] -> task,armA_c,l1,l2,l3
/ niah_single,73.0,75.0,80.0,74.0
[08:04:43] PROMPT_LEN=8192 niah_single N=100
[08:13:01] -> task,armA_c,l1,l2,l3
/ niah_single,53.0,48.0,62.0,64.0
[08:13:01] PROMPT_LEN=16360 niah_single N=100
[08:27:27] -> task,armA_c,l1,l2,l3
/ niah_single,33.0,30.0,35.0,35.0
[08:27:27] === DECISION EVAL DONE ===
|