| { | |
| "paper_setup": { | |
| "vocab": 5, | |
| "primary_k": 2, | |
| "alpha": 1.0 | |
| }, | |
| "run_setup": { | |
| "output": "/data/gpu-output", | |
| "steps": 500, | |
| "length": 64, | |
| "vocab": 5, | |
| "seeds": 5, | |
| "device": "cuda" | |
| }, | |
| "wall_seconds": 772.6685228347778, | |
| "rows": [ | |
| { | |
| "k": 1, | |
| "bos": false, | |
| "seed": 0, | |
| "beta": [ | |
| 1.6930645796321535 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.0006550685874616, | |
| "kl_to_true": 0.8285508772318574, | |
| "kl_to_bayes": 0.6169615490548486, | |
| "bayes_kl_to_true": 0.15641236490314867, | |
| "attention_overlap_slope": 1.6930645796321517, | |
| "attention_overlap_r2": 1.0, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 1, | |
| "bos": false, | |
| "seed": 1, | |
| "beta": [ | |
| 1.8876569147424276 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.064751169140637, | |
| "kl_to_true": 0.778554751426525, | |
| "kl_to_bayes": 0.5238144206007598, | |
| "bayes_kl_to_true": 0.1641728884216971, | |
| "attention_overlap_slope": 1.8876569147424256, | |
| "attention_overlap_r2": 1.0, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 1, | |
| "bos": false, | |
| "seed": 2, | |
| "beta": [ | |
| 1.7900553864646844 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.105120759265691, | |
| "kl_to_true": 0.8527447862592809, | |
| "kl_to_bayes": 0.7527976152645213, | |
| "bayes_kl_to_true": 0.15093324531952326, | |
| "attention_overlap_slope": 1.790055386464687, | |
| "attention_overlap_r2": 1.0, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 1, | |
| "bos": false, | |
| "seed": 3, | |
| "beta": [ | |
| 1.8457972423261217 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.1667033882767366, | |
| "kl_to_true": 0.9247158642803754, | |
| "kl_to_bayes": 0.67660360559533, | |
| "bayes_kl_to_true": 0.1571236701795906, | |
| "attention_overlap_slope": 1.8457972423261264, | |
| "attention_overlap_r2": 1.0, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 1, | |
| "bos": false, | |
| "seed": 4, | |
| "beta": [ | |
| 1.6808967103337247 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.0768273713344065, | |
| "kl_to_true": 0.7396788014260097, | |
| "kl_to_bayes": 0.6218736433119315, | |
| "bayes_kl_to_true": 0.16281367395169394, | |
| "attention_overlap_slope": 1.6808967103337233, | |
| "attention_overlap_r2": 1.0, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 1, | |
| "bos": true, | |
| "seed": 0, | |
| "beta": [ | |
| 3.470422587202903 | |
| ], | |
| "bos_mass": 161.5742122794537, | |
| "nll": 1.457266817353693, | |
| "kl_to_true": 0.1580625117170332, | |
| "kl_to_bayes": 0.0016402823235092366, | |
| "bayes_kl_to_true": 0.15641236490314867, | |
| "attention_overlap_slope": 3.4704225872029117, | |
| "attention_overlap_r2": 1.0, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 1, | |
| "bos": true, | |
| "seed": 1, | |
| "beta": [ | |
| 3.569247398310965 | |
| ], | |
| "bos_mass": 147.0803978596844, | |
| "nll": 1.459365061656418, | |
| "kl_to_true": 0.1659760213219843, | |
| "kl_to_bayes": 0.001145747878178236, | |
| "bayes_kl_to_true": 0.1641728884216971, | |
| "attention_overlap_slope": 3.569247398310969, | |
| "attention_overlap_r2": 1.0, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 1, | |
| "bos": true, | |
| "seed": 2, | |
| "beta": [ | |
| 3.702108616119927 | |
| ], | |
| "bos_mass": 151.15233162527954, | |
| "nll": 1.4330326439288261, | |
| "kl_to_true": 0.15306445234757818, | |
| "kl_to_bayes": 0.0021762270768138857, | |
| "bayes_kl_to_true": 0.15093324531952326, | |
| "attention_overlap_slope": 3.7021086161199204, | |
| "attention_overlap_r2": 1.0, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 1, | |
| "bos": true, | |
| "seed": 3, | |
| "beta": [ | |
| 3.6235786672655275 | |
| ], | |
| "bos_mass": 113.58854223187522, | |
| "nll": 1.4586282633646757, | |
| "kl_to_true": 0.163519770536497, | |
| "kl_to_bayes": 0.004682446924971027, | |
| "bayes_kl_to_true": 0.1571236701795906, | |
| "attention_overlap_slope": 3.6235786672655346, | |
| "attention_overlap_r2": 1.0, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 1, | |
| "bos": true, | |
| "seed": 4, | |
| "beta": [ | |
| 3.438073464652619 | |
| ], | |
| "bos_mass": 155.14199791108777, | |
| "nll": 1.4675133729846401, | |
| "kl_to_true": 0.16493233059509776, | |
| "kl_to_bayes": 0.0017098186931211732, | |
| "bayes_kl_to_true": 0.16281367395169394, | |
| "attention_overlap_slope": 3.438073464652625, | |
| "attention_overlap_r2": 1.0, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 2, | |
| "bos": false, | |
| "seed": 0, | |
| "beta": [ | |
| 0.8580751049349579, | |
| 0.821533196854088 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.2259960636175333, | |
| "kl_to_true": 0.9963021462182176, | |
| "kl_to_bayes": 0.7706829684681136, | |
| "bayes_kl_to_true": 0.26272999132318264, | |
| "attention_overlap_slope": 0.8398105753614834, | |
| "attention_overlap_r2": 0.9995735884693955, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 2, | |
| "bos": false, | |
| "seed": 1, | |
| "beta": [ | |
| 0.7813418229519343, | |
| 0.5874054560908514 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.239154091536007, | |
| "kl_to_true": 0.8914109711229098, | |
| "kl_to_bayes": 0.6483643243405118, | |
| "bayes_kl_to_true": 0.2647279366222993, | |
| "attention_overlap_slope": 0.6843692331425872, | |
| "attention_overlap_r2": 0.9815924405244475, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 2, | |
| "bos": false, | |
| "seed": 2, | |
| "beta": [ | |
| 0.9325293521882272, | |
| 0.7229287141071734 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.2878528096420587, | |
| "kl_to_true": 0.9700856527272188, | |
| "kl_to_bayes": 0.7119090667599893, | |
| "bayes_kl_to_true": 0.2661596335013201, | |
| "attention_overlap_slope": 0.8277578714700348, | |
| "attention_overlap_r2": 0.9852416231304755, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 2, | |
| "bos": false, | |
| "seed": 3, | |
| "beta": [ | |
| 0.8055885050829208, | |
| 0.7202856606462539 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.21744811755106, | |
| "kl_to_true": 1.0060883187258387, | |
| "kl_to_bayes": 0.7016385027486404, | |
| "bayes_kl_to_true": 0.25795664002150487, | |
| "attention_overlap_slope": 0.7629491026112536, | |
| "attention_overlap_r2": 0.9970769972942697, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 2, | |
| "bos": false, | |
| "seed": 4, | |
| "beta": [ | |
| 0.5939132996761789, | |
| 0.7506196262624412 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.2992463870303923, | |
| "kl_to_true": 1.1495594385365024, | |
| "kl_to_bayes": 0.9131012755864052, | |
| "bayes_kl_to_true": 0.25537628251057604, | |
| "attention_overlap_slope": 0.6723500032427813, | |
| "attention_overlap_r2": 0.9876638299065911, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 2, | |
| "bos": true, | |
| "seed": 0, | |
| "beta": [ | |
| 2.6753169722580106, | |
| 2.524039958967024 | |
| ], | |
| "bos_mass": 1069.3730847240818, | |
| "nll": 1.5351920162036736, | |
| "kl_to_true": 0.2692611215846471, | |
| "kl_to_bayes": 0.004756522223734605, | |
| "bayes_kl_to_true": 0.26272999132318264, | |
| "attention_overlap_slope": 2.599705061765402, | |
| "attention_overlap_r2": 0.9992376385766464, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 2, | |
| "bos": true, | |
| "seed": 1, | |
| "beta": [ | |
| 2.7711495122275958, | |
| 2.70421161362507 | |
| ], | |
| "bos_mass": 1043.3951931987665, | |
| "nll": 1.537563619100544, | |
| "kl_to_true": 0.2670327896446933, | |
| "kl_to_bayes": 0.002875007858342953, | |
| "bayes_kl_to_true": 0.2647279366222993, | |
| "attention_overlap_slope": 2.737679042047381, | |
| "attention_overlap_r2": 0.9998604126989626, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 2, | |
| "bos": true, | |
| "seed": 2, | |
| "beta": [ | |
| 2.7764362037965613, | |
| 2.7744367663912617 | |
| ], | |
| "bos_mass": 1110.9417493832684, | |
| "nll": 1.551900404967508, | |
| "kl_to_true": 0.2721866141877306, | |
| "kl_to_bayes": 0.0031677553645135825, | |
| "bayes_kl_to_true": 0.2661596335013201, | |
| "attention_overlap_slope": 2.7754367601904972, | |
| "attention_overlap_r2": 0.9999998787532285, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 2, | |
| "bos": true, | |
| "seed": 3, | |
| "beta": [ | |
| 2.8543394056860145, | |
| 2.8691916175493772 | |
| ], | |
| "bos_mass": 1122.3681192527608, | |
| "nll": 1.5615310611503912, | |
| "kl_to_true": 0.2621058802727228, | |
| "kl_to_bayes": 0.00346397651910272, | |
| "bayes_kl_to_true": 0.25795664002150487, | |
| "attention_overlap_slope": 2.861763418840777, | |
| "attention_overlap_r2": 0.9999936834971035, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 2, | |
| "bos": true, | |
| "seed": 4, | |
| "beta": [ | |
| 2.6599593598446023, | |
| 2.597604378925108 | |
| ], | |
| "bos_mass": 1043.5114042630291, | |
| "nll": 1.5463042238735973, | |
| "kl_to_true": 0.2607410366724894, | |
| "kl_to_bayes": 0.003973705649362713, | |
| "bayes_kl_to_true": 0.25537628251057604, | |
| "attention_overlap_slope": 2.6287486278924987, | |
| "attention_overlap_r2": 0.9998706468634581, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 3, | |
| "bos": false, | |
| "seed": 0, | |
| "beta": [ | |
| 0.12217359052320316, | |
| 0.4919169269211535, | |
| 0.41163188424664254 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.3756495129475472, | |
| "kl_to_true": 1.234230414155023, | |
| "kl_to_bayes": 0.905972976029062, | |
| "bayes_kl_to_true": 0.31016026219466764, | |
| "attention_overlap_slope": 0.34190912741015356, | |
| "attention_overlap_r2": 0.8273425479739536, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 3, | |
| "bos": false, | |
| "seed": 1, | |
| "beta": [ | |
| 0.2896469382815528, | |
| 0.31073506158945235, | |
| 0.12575965714700627 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.4469819859752864, | |
| "kl_to_true": 1.124530515790056, | |
| "kl_to_bayes": 0.8462539149558679, | |
| "bayes_kl_to_true": 0.3104222631914786, | |
| "attention_overlap_slope": 0.24214556959629577, | |
| "attention_overlap_r2": 0.8966977206689941, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 3, | |
| "bos": false, | |
| "seed": 2, | |
| "beta": [ | |
| 0.07823717739544649, | |
| 0.27973196320404925, | |
| 0.18078231522504662 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.4111830118929425, | |
| "kl_to_true": 1.0551657353112904, | |
| "kl_to_bayes": 0.7130541297291353, | |
| "bayes_kl_to_true": 0.30413376510002726, | |
| "attention_overlap_slope": 0.17960985988079967, | |
| "attention_overlap_r2": 0.8302669395382835, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 3, | |
| "bos": false, | |
| "seed": 3, | |
| "beta": [ | |
| 0.19794989207916444, | |
| 0.2539988110386651, | |
| 0.18768591755133449 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.4694240731829753, | |
| "kl_to_true": 1.152147509661796, | |
| "kl_to_bayes": 0.8472194493872363, | |
| "bayes_kl_to_true": 0.30452681722961517, | |
| "attention_overlap_slope": 0.21319728431646007, | |
| "attention_overlap_r2": 0.9820107169215964, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 3, | |
| "bos": false, | |
| "seed": 4, | |
| "beta": [ | |
| 0.2215389880205417, | |
| 0.43192317760221327, | |
| 0.12855698814557573 | |
| ], | |
| "bos_mass": null, | |
| "nll": 2.4131936164702363, | |
| "kl_to_true": 1.114346977042289, | |
| "kl_to_bayes": 0.8373470536421879, | |
| "bayes_kl_to_true": 0.313061942505399, | |
| "attention_overlap_slope": 0.26052630045597835, | |
| "attention_overlap_r2": 0.809280671118473, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 3, | |
| "bos": true, | |
| "seed": 0, | |
| "beta": [ | |
| 1.4768269079943812, | |
| 1.778121204647586, | |
| 1.5747525311571957 | |
| ], | |
| "bos_mass": 803.2629228597949, | |
| "nll": 1.593401729654823, | |
| "kl_to_true": 0.31497933397090316, | |
| "kl_to_bayes": 0.005624327267835937, | |
| "bayes_kl_to_true": 0.31016026219466764, | |
| "attention_overlap_slope": 1.6099787047485172, | |
| "attention_overlap_r2": 0.9941543082897751, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 3, | |
| "bos": true, | |
| "seed": 1, | |
| "beta": [ | |
| 1.5615969715774973, | |
| 1.738450623039793, | |
| 1.5221652496894085 | |
| ], | |
| "bos_mass": 826.2783208981992, | |
| "nll": 1.605015357906008, | |
| "kl_to_true": 0.31609318841132694, | |
| "kl_to_bayes": 0.005480555754217444, | |
| "bayes_kl_to_true": 0.3104222631914786, | |
| "attention_overlap_slope": 1.6076227606478029, | |
| "attention_overlap_r2": 0.9966354944003444, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 3, | |
| "bos": true, | |
| "seed": 2, | |
| "beta": [ | |
| 1.0383871696391178, | |
| 1.3614957740735547, | |
| 1.100313229484844 | |
| ], | |
| "bos_mass": 438.9474277137842, | |
| "nll": 1.6017989977887777, | |
| "kl_to_true": 0.3125132246514988, | |
| "kl_to_bayes": 0.00910404199393919, | |
| "bayes_kl_to_true": 0.30413376510002726, | |
| "attention_overlap_slope": 1.1667212261251954, | |
| "attention_overlap_r2": 0.9861574126674635, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 3, | |
| "bos": true, | |
| "seed": 3, | |
| "beta": [ | |
| 1.7418329235881789, | |
| 1.4157916773730683, | |
| 1.5588624144365064 | |
| ], | |
| "bos_mass": 719.158636907118, | |
| "nll": 1.6049485148478377, | |
| "kl_to_true": 0.30995774684532673, | |
| "kl_to_bayes": 0.005833097805403933, | |
| "bayes_kl_to_true": 0.30452681722961517, | |
| "attention_overlap_slope": 1.5721699641742442, | |
| "attention_overlap_r2": 0.9929939886926692, | |
| "positions": 2240 | |
| }, | |
| { | |
| "k": 3, | |
| "bos": true, | |
| "seed": 4, | |
| "beta": [ | |
| 1.4981967841381862, | |
| 1.497834217050287, | |
| 1.3926855794888278 | |
| ], | |
| "bos_mass": 557.2398732904178, | |
| "nll": 1.6008616907142246, | |
| "kl_to_true": 0.31963090698258345, | |
| "kl_to_bayes": 0.006793202703257004, | |
| "bayes_kl_to_true": 0.313061942505399, | |
| "attention_overlap_slope": 1.4628607276336374, | |
| "attention_overlap_r2": 0.9988623041558363, | |
| "positions": 2240 | |
| } | |
| ], | |
| "mutation_tests": [ | |
| { | |
| "k": 1, | |
| "seed": 0, | |
| "learned_nll": 1.7536993375480485, | |
| "untrained_uniform_overlap_nll": 1.8260625999246516, | |
| "hard_match_nll": 4.794859199284604, | |
| "shuffled_context_nll": 1.8220030834123673, | |
| "untrained_worse": true, | |
| "hard_match_worse": true, | |
| "shuffled_worse": true | |
| }, | |
| { | |
| "k": 1, | |
| "seed": 1, | |
| "learned_nll": 2.808503406202665, | |
| "untrained_uniform_overlap_nll": 2.9128085794096297, | |
| "hard_match_nll": 6.151996750232827, | |
| "shuffled_context_nll": 3.0201682094752798, | |
| "untrained_worse": true, | |
| "hard_match_worse": true, | |
| "shuffled_worse": true | |
| }, | |
| { | |
| "k": 1, | |
| "seed": 2, | |
| "learned_nll": 1.6473911605658818, | |
| "untrained_uniform_overlap_nll": 1.6299321240219111, | |
| "hard_match_nll": 6.906007500561277, | |
| "shuffled_context_nll": 1.7387456519301907, | |
| "untrained_worse": false, | |
| "hard_match_worse": true, | |
| "shuffled_worse": true | |
| }, | |
| { | |
| "k": 1, | |
| "seed": 3, | |
| "learned_nll": 2.4527502881502787, | |
| "untrained_uniform_overlap_nll": 2.5804718428732656, | |
| "hard_match_nll": 6.313262411412619, | |
| "shuffled_context_nll": 2.592922285285464, | |
| "untrained_worse": true, | |
| "hard_match_worse": true, | |
| "shuffled_worse": true | |
| }, | |
| { | |
| "k": 1, | |
| "seed": 4, | |
| "learned_nll": 2.002421975424776, | |
| "untrained_uniform_overlap_nll": 1.9533860679395794, | |
| "hard_match_nll": 6.449788416014823, | |
| "shuffled_context_nll": 1.9805492735248638, | |
| "untrained_worse": false, | |
| "hard_match_worse": true, | |
| "shuffled_worse": false | |
| }, | |
| { | |
| "k": 2, | |
| "seed": 0, | |
| "learned_nll": 1.6564684714070328, | |
| "untrained_uniform_overlap_nll": 1.6560642706133422, | |
| "hard_match_nll": 11.38202445066547, | |
| "shuffled_context_nll": 1.7275352944206195, | |
| "untrained_worse": false, | |
| "hard_match_worse": true, | |
| "shuffled_worse": true | |
| }, | |
| { | |
| "k": 2, | |
| "seed": 1, | |
| "learned_nll": 2.093263527416452, | |
| "untrained_uniform_overlap_nll": 2.0958453712693985, | |
| "hard_match_nll": 13.72265086280333, | |
| "shuffled_context_nll": 2.1099935623839987, | |
| "untrained_worse": true, | |
| "hard_match_worse": true, | |
| "shuffled_worse": true | |
| }, | |
| { | |
| "k": 2, | |
| "seed": 2, | |
| "learned_nll": 2.041227172099717, | |
| "untrained_uniform_overlap_nll": 2.0675759502009052, | |
| "hard_match_nll": 12.420809425412369, | |
| "shuffled_context_nll": 2.151110589976682, | |
| "untrained_worse": true, | |
| "hard_match_worse": true, | |
| "shuffled_worse": true | |
| }, | |
| { | |
| "k": 2, | |
| "seed": 3, | |
| "learned_nll": 2.121490669801476, | |
| "untrained_uniform_overlap_nll": 2.1403534758305303, | |
| "hard_match_nll": 12.838975232795969, | |
| "shuffled_context_nll": 2.192333527448208, | |
| "untrained_worse": true, | |
| "hard_match_worse": true, | |
| "shuffled_worse": true | |
| }, | |
| { | |
| "k": 2, | |
| "seed": 4, | |
| "learned_nll": 2.1026130342424394, | |
| "untrained_uniform_overlap_nll": 2.1216206632762753, | |
| "hard_match_nll": 13.295075237503912, | |
| "shuffled_context_nll": 2.170215629188961, | |
| "untrained_worse": true, | |
| "hard_match_worse": true, | |
| "shuffled_worse": true | |
| }, | |
| { | |
| "k": 3, | |
| "seed": 0, | |
| "learned_nll": 2.0943605342838807, | |
| "untrained_uniform_overlap_nll": 2.1290180492430832, | |
| "hard_match_nll": 12.818397220324805, | |
| "shuffled_context_nll": 2.132239787380053, | |
| "untrained_worse": true, | |
| "hard_match_worse": true, | |
| "shuffled_worse": true | |
| }, | |
| { | |
| "k": 3, | |
| "seed": 1, | |
| "learned_nll": 2.602892005014399, | |
| "untrained_uniform_overlap_nll": 2.581690160503532, | |
| "hard_match_nll": 16.227524882941267, | |
| "shuffled_context_nll": 2.5582915760456584, | |
| "untrained_worse": false, | |
| "hard_match_worse": true, | |
| "shuffled_worse": false | |
| }, | |
| { | |
| "k": 3, | |
| "seed": 2, | |
| "learned_nll": 2.5353527585115, | |
| "untrained_uniform_overlap_nll": 2.537883587654552, | |
| "hard_match_nll": 15.720499642668807, | |
| "shuffled_context_nll": 2.5610409954614006, | |
| "untrained_worse": true, | |
| "hard_match_worse": true, | |
| "shuffled_worse": true | |
| }, | |
| { | |
| "k": 3, | |
| "seed": 3, | |
| "learned_nll": 2.58891008344617, | |
| "untrained_uniform_overlap_nll": 2.5947172109451535, | |
| "hard_match_nll": 14.378501339155678, | |
| "shuffled_context_nll": 2.593934160621953, | |
| "untrained_worse": true, | |
| "hard_match_worse": true, | |
| "shuffled_worse": true | |
| }, | |
| { | |
| "k": 3, | |
| "seed": 4, | |
| "learned_nll": 2.141158061035497, | |
| "untrained_uniform_overlap_nll": 2.1553999582605843, | |
| "hard_match_nll": 14.747719200376743, | |
| "shuffled_context_nll": 2.1827441521481883, | |
| "untrained_worse": true, | |
| "hard_match_worse": true, | |
| "shuffled_worse": true | |
| } | |
| ], | |
| "aggregate": { | |
| "k1_bos0": { | |
| "nll": { | |
| "mean": 2.0828115513209866, | |
| "std": 0.05412433193432743 | |
| }, | |
| "kl_to_true": { | |
| "mean": 0.8248490161248097, | |
| "std": 0.06350179453113508 | |
| }, | |
| "kl_to_bayes": { | |
| "mean": 0.6384101667654782, | |
| "std": 0.07537759878431505 | |
| }, | |
| "bayes_kl_to_true": { | |
| "mean": 0.1582911685551307, | |
| "std": 0.0047766028432639695 | |
| }, | |
| "attention_overlap_slope": { | |
| "mean": 1.7794941666998227, | |
| "std": 0.08172924713602799 | |
| }, | |
| "attention_overlap_r2": { | |
| "mean": 1.0, | |
| "std": 0.0 | |
| } | |
| }, | |
| "k1_bos1": { | |
| "nll": { | |
| "mean": 1.4551612318576506, | |
| "std": 0.011630784813361782 | |
| }, | |
| "kl_to_true": { | |
| "mean": 0.16111101730363808, | |
| "std": 0.0048603165619221156 | |
| }, | |
| "kl_to_bayes": { | |
| "mean": 0.0022709045793187117, | |
| "std": 0.0012492381452460094 | |
| }, | |
| "bayes_kl_to_true": { | |
| "mean": 0.1582911685551307, | |
| "std": 0.0047766028432639695 | |
| }, | |
| "attention_overlap_slope": { | |
| "mean": 3.560686146710392, | |
| "std": 0.09717042657207582 | |
| }, | |
| "attention_overlap_r2": { | |
| "mean": 1.0, | |
| "std": 0.0 | |
| } | |
| }, | |
| "k2_bos0": { | |
| "nll": { | |
| "mean": 2.25393949387541, | |
| "std": 0.03326827773301641 | |
| }, | |
| "kl_to_true": { | |
| "mean": 1.0026893054661374, | |
| "std": 0.08374849365645665 | |
| }, | |
| "kl_to_bayes": { | |
| "mean": 0.7491392275807321, | |
| "std": 0.0907145752511971 | |
| }, | |
| "bayes_kl_to_true": { | |
| "mean": 0.2613900967957766, | |
| "std": 0.004089977019869316 | |
| }, | |
| "attention_overlap_slope": { | |
| "mean": 0.757447357165628, | |
| "std": 0.06977035071099172 | |
| }, | |
| "attention_overlap_r2": { | |
| "mean": 0.9902296958650358, | |
| "std": 0.006931963407362833 | |
| } | |
| }, | |
| "k2_bos1": { | |
| "nll": { | |
| "mean": 1.5464982650591428, | |
| "std": 0.00962148520218784 | |
| }, | |
| "kl_to_true": { | |
| "mean": 0.2662654884724566, | |
| "std": 0.004299865268431111 | |
| }, | |
| "kl_to_bayes": { | |
| "mean": 0.0036473935230113145, | |
| "std": 0.0006628605062857784 | |
| }, | |
| "bayes_kl_to_true": { | |
| "mean": 0.2613900967957766, | |
| "std": 0.004089977019869316 | |
| }, | |
| "attention_overlap_slope": { | |
| "mean": 2.720666582147311, | |
| "std": 0.09620614663017038 | |
| }, | |
| "attention_overlap_r2": { | |
| "mean": 0.9997924520778799, | |
| "std": 0.00028357397672899094 | |
| } | |
| }, | |
| "k3_bos0": { | |
| "nll": { | |
| "mean": 2.4232864400937975, | |
| "std": 0.03227319312537903 | |
| }, | |
| "kl_to_true": { | |
| "mean": 1.136084230392091, | |
| "std": 0.058385788816396195 | |
| }, | |
| "kl_to_bayes": { | |
| "mean": 0.8299695047486978, | |
| "std": 0.06334493412441748 | |
| }, | |
| "bayes_kl_to_true": { | |
| "mean": 0.3084610100442375, | |
| "std": 0.0035243573371692323 | |
| }, | |
| "attention_overlap_slope": { | |
| "mean": 0.24747762833193748, | |
| "std": 0.05458421510487609 | |
| }, | |
| "attention_overlap_r2": { | |
| "mean": 0.8691197192442601, | |
| "std": 0.06378166778941566 | |
| } | |
| }, | |
| "k3_bos1": { | |
| "nll": { | |
| "mean": 1.601205258182334, | |
| "std": 0.004240152194751006 | |
| }, | |
| "kl_to_true": { | |
| "mean": 0.31463488017232777, | |
| "std": 0.0032736094310619095 | |
| }, | |
| "kl_to_bayes": { | |
| "mean": 0.006567045104930702, | |
| "std": 0.0013487294126096642 | |
| }, | |
| "bayes_kl_to_true": { | |
| "mean": 0.3084610100442375, | |
| "std": 0.0035243573371692323 | |
| }, | |
| "attention_overlap_slope": { | |
| "mean": 1.4838706766658791, | |
| "std": 0.16735577023087103 | |
| }, | |
| "attention_overlap_r2": { | |
| "mean": 0.9937607016412178, | |
| "std": 0.004309142833157501 | |
| } | |
| } | |
| } | |
| } |
Xet Storage Details
- Size:
- 22.9 kB
- Xet hash:
- fdc13f0118a44355324afcc1149508ea131a643fecdcbaf970a018e8fef57105
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.