tinystories-diff-ar / log_eval.txt
Asilarkness's picture
Super-squash branch 'main' using huggingface_hub
a6c51a2
Raw History Blame Contribute Delete
1.48 kB
launch eval
22:15:22 loaded ar step=58000 seen=950.3M
22:15:24 loaded mdlmc step=29000 seen=475.1M
22:15:26 prompts=48 prompt_len=32 gen_len=128
22:15:33 loading judge roneneldan/TinyStories-33M
22:15:35 AR calls=128 0.70s judge_nll=2.7232 distinct2=0.5349
22:15:35 DIFF K=1 calls=1 0.11s judge_nll=10.6201 distinct2=0.7187
22:15:35 DIFF K=2 calls=2 0.01s judge_nll=8.4615 distinct2=0.5966
22:15:36 DIFF K=4 calls=4 0.02s judge_nll=6.1524 distinct2=0.4898
22:15:36 DIFF K=8 calls=8 0.05s judge_nll=4.7487 distinct2=0.4473
22:15:36 DIFF K=16 calls=16 0.10s judge_nll=3.8114 distinct2=0.4009
22:15:36 HYB K=4 p=0.10 R=1 calls=5 0.03s judge_nll=5.8065 distinct2=0.4518
22:15:36 HYB K=4 p=0.10 R=2 calls=6 0.03s judge_nll=5.7953 distinct2=0.4500
22:15:37 HYB K=4 p=0.25 R=1 calls=5 0.03s judge_nll=6.3426 distinct2=0.4600
22:15:37 HYB K=4 p=0.25 R=2 calls=6 0.03s judge_nll=6.2650 distinct2=0.4526
22:15:37 HYB K=8 p=0.10 R=1 calls=9 0.06s judge_nll=4.6221 distinct2=0.4237
22:15:37 HYB K=8 p=0.10 R=2 calls=10 0.06s judge_nll=4.6257 distinct2=0.4158
22:15:38 HYB K=8 p=0.25 R=1 calls=9 0.05s judge_nll=5.3384 distinct2=0.4869
22:15:38 HYB K=8 p=0.25 R=2 calls=10 0.06s judge_nll=5.3898 distinct2=0.4852
22:15:38 REAL TEXT judge_nll=0.9997 distinct2=0.5591