Upload add_sub_sorl_v1_abs2_K1_100K

add_sub_sorl_v1_abs2_K1_100K/config.json

@@ -0,0 +1,37 @@
+{
+  "architectures": [
+    "SorlModelWrapper"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": null,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 510,
+  "initializer_range": 0.02,
+  "intermediate_size": 2040,
+  "layer_types": [
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 128,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 3,
+  "num_hidden_layers": 2,
+  "num_key_value_heads": 3,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "transformers_version": "5.5.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151646
+}

add_sub_sorl_v1_abs2_K1_100K/generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "transformers_version": "5.5.0",
+  "use_cache": true
+}

add_sub_sorl_v1_abs2_K1_100K/metrics.json

@@ -0,0 +1,3537 @@
+{
+  "history": {
+    "step": [
+      50,
+      100,
+      150,
+      200,
+      250,
+      300,
+      350,
+      400,
+      450,
+      500,
+      550,
+      600,
+      650,
+      700,
+      750,
+      800,
+      850,
+      900,
+      950,
+      1000,
+      1050,
+      1100,
+      1150,
+      1200,
+      1250,
+      1300,
+      1350,
+      1400,
+      1450,
+      1500,
+      1550,
+      1613,
+      1663,
+      1713,
+      1763,
+      1813,
+      1863,
+      1913,
+      1963,
+      2013,
+      2063,
+      2113,
+      2163,
+      2213,
+      2263,
+      2313,
+      2363,
+      2413,
+      2463,
+      2513,
+      2563,
+      2613,
+      2663,
+      2713,
+      2763,
+      2813,
+      2863,
+      2913,
+      2963,
+      3013,
+      3063,
+      3113,
+      3176,
+      3226,
+      3276,
+      3326,
+      3376,
+      3426,
+      3476,
+      3526,
+      3576,
+      3626,
+      3676,
+      3726,
+      3776,
+      3826,
+      3876,
+      3926,
+      3976,
+      4026,
+      4076,
+      4126,
+      4176,
+      4226,
+      4276,
+      4326,
+      4376,
+      4426,
+      4476,
+      4526,
+      4576,
+      4626,
+      4676,
+      4739,
+      4789,
+      4839,
+      4889,
+      4939,
+      4989,
+      5039,
+      5089,
+      5139,
+      5189,
+      5239,
+      5289,
+      5339,
+      5389,
+      5439,
+      5489,
+      5539,
+      5589,
+      5639,
+      5689,
+      5739,
+      5789,
+      5839,
+      5889,
+      5939,
+      5989,
+      6039,
+      6089,
+      6139,
+      6189,
+      6239,
+      6302,
+      6352,
+      6402,
+      6452,
+      6502,
+      6552,
+      6602,
+      6652,
+      6702,
+      6752,
+      6802,
+      6852,
+      6902,
+      6952,
+      7002,
+      7052,
+      7102,
+      7152,
+      7202,
+      7252,
+      7302,
+      7352,
+      7402,
+      7452,
+      7502,
+      7552,
+      7602,
+      7652,
+      7702,
+      7752,
+      7802,
+      7865,
+      7915,
+      7965,
+      8015,
+      8065,
+      8115,
+      8165,
+      8215,
+      8265,
+      8315,
+      8365,
+      8415,
+      8465,
+      8515,
+      8565,
+      8615,
+      8665,
+      8715,
+      8765,
+      8815,
+      8865,
+      8915,
+      8965,
+      9015,
+      9065,
+      9115,
+      9165,
+      9215,
+      9265,
+      9315,
+      9365,
+      9428,
+      9478,
+      9528,
+      9578,
+      9628,
+      9678,
+      9728,
+      9778,
+      9828,
+      9878,
+      9928,
+      9978,
+      10028,
+      10078,
+      10128,
+      10178,
+      10228,
+      10278,
+      10328,
+      10378,
+      10428,
+      10478,
+      10528,
+      10578,
+      10628,
+      10678,
+      10728,
+      10778,
+      10828,
+      10878,
+      10928,
+      10991,
+      11041,
+      11091,
+      11141,
+      11191,
+      11241,
+      11291,
+      11341,
+      11391,
+      11441,
+      11491,
+      11541,
+      11591,
+      11641,
+      11691,
+      11741,
+      11791,
+      11841,
+      11891,
+      11941,
+      11991,
+      12041,
+      12091,
+      12141,
+      12191,
+      12241,
+      12291,
+      12341,
+      12391,
+      12441,
+      12491,
+      12554,
+      12604,
+      12654,
+      12704,
+      12754,
+      12804,
+      12854,
+      12904,
+      12954,
+      13004,
+      13054,
+      13104,
+      13154,
+      13204,
+      13254,
+      13304,
+      13354,
+      13404,
+      13454,
+      13504,
+      13554,
+      13604,
+      13654,
+      13704,
+      13754,
+      13804,
+      13854,
+      13904,
+      13954,
+      14004,
+      14054,
+      14117,
+      14167,
+      14217,
+      14267,
+      14317,
+      14367,
+      14417,
+      14467,
+      14517,
+      14567,
+      14617,
+      14667,
+      14717,
+      14767,
+      14817,
+      14867,
+      14917,
+      14967,
+      15017,
+      15067,
+      15117,
+      15167,
+      15217,
+      15267,
+      15317,
+      15367,
+      15417,
+      15467,
+      15517,
+      15567,
+      15617
+    ],
+    "loss": [
+      4.038571357727051,
+      -1.2151508331298828,
+      7.321413993835449,
+      9.976255416870117,
+      8.979491233825684,
+      7.84694766998291,
+      6.3668670654296875,
+      6.1605730056762695,
+      5.830511569976807,
+      5.407081127166748,
+      5.3787336349487305,
+      3.8804502487182617,
+      -1.5981316566467285,
+      -3.5876903533935547,
+      -4.375753402709961,
+      -3.580019474029541,
+      -1.7630763053894043,
+      -0.8523468971252441,
+      -1.1872100830078125,
+      -0.9717748165130615,
+      -1.6817877292633057,
+      -1.436840534210205,
+      -1.4684152603149414,
+      -2.24332857131958,
+      -3.4089715480804443,
+      -2.6334524154663086,
+      -3.481844425201416,
+      -2.296447515487671,
+      -2.344224214553833,
+      -2.873753786087036,
+      -2.083787202835083,
+      -2.14436674118042,
+      -1.6999812126159668,
+      -1.6430357694625854,
+      -1.3074538707733154,
+      -2.015672445297241,
+      -1.6747031211853027,
+      -1.055972933769226,
+      -0.7998563051223755,
+      -3.0470328330993652,
+      -2.588050127029419,
+      -1.202722191810608,
+      -1.1608521938323975,
+      -0.8615143299102783,
+      -0.7070940732955933,
+      -0.831250011920929,
+      -0.3691519498825073,
+      -0.8630808591842651,
+      -1.018460750579834,
+      -0.9615075588226318,
+      -0.6236373782157898,
+      -0.6105367541313171,
+      -0.5900554060935974,
+      -0.710160493850708,
+      -0.4855920970439911,
+      -0.3556075096130371,
+      -0.925187349319458,
+      -0.5160346031188965,
+      -0.32945716381073,
+      -0.12898364663124084,
+      -0.2840459942817688,
+      -0.26350536942481995,
+      -0.6467360854148865,
+      -0.10725943744182587,
+      -0.29941487312316895,
+      -0.41120630502700806,
+      -0.3216160237789154,
+      -0.4872255027294159,
+      -0.2023446261882782,
+      -0.09411603957414627,
+      -0.19229590892791748,
+      -0.5609635710716248,
+      -0.22852931916713715,
+      -0.30250436067581177,
+      -0.23088058829307556,
+      -0.1587841510772705,
+      -0.10579416900873184,
+      -0.13622130453586578,
+      -0.2128376066684723,
+      0.0039059966802597046,
+      -0.2921285629272461,
+      -0.1244480311870575,
+      -0.1797204315662384,
+      -0.23258402943611145,
+      -0.09155713021755219,
+      -0.09225616604089737,
+      -0.5923236608505249,
+      -0.017554309219121933,
+      -0.061483897268772125,
+      0.042076222598552704,
+      -0.17921528220176697,
+      -0.016543734818696976,
+      -0.06260864436626434,
+      -0.023083575069904327,
+      -0.11890868097543716,
+      -0.07157810032367706,
+      0.009503915905952454,
+      0.048799872398376465,
+      0.003565732389688492,
+      0.04110269993543625,
+      0.04389260336756706,
+      0.11524543166160583,
+      -0.12375140935182571,
+      -0.017672806978225708,
+      0.03712224215269089,
+      0.04797496646642685,
+      0.017265018075704575,
+      -0.05425512418150902,
+      -0.047244276851415634,
+      0.04379027709364891,
+      0.054904866963624954,
+      0.02523486316204071,
+      0.05093398690223694,
+      0.021740391850471497,
+      0.024601135402917862,
+      -0.020424537360668182,
+      0.011466562747955322,
+      -0.03010578081011772,
+      0.05196551978588104,
+      0.04819013178348541,
+      0.15849235653877258,
+      0.059194814413785934,
+      0.03619581460952759,
+      -0.014813955873250961,
+      0.032088086009025574,
+      0.028168514370918274,
+      0.0006464235484600067,
+      0.05649052560329437,
+      0.017316129058599472,
+      0.003864578902721405,
+      -0.027318879961967468,
+      0.04794345051050186,
+      0.021732095628976822,
+      0.05282561480998993,
+      -0.023973368108272552,
+      0.04236216098070145,
+      0.05816863477230072,
+      0.04404866695404053,
+      0.04713943600654602,
+      -0.01663733273744583,
+      -0.022356614470481873,
+      0.05468076467514038,
+      0.056800711899995804,
+      0.05782049149274826,
+      0.051594339311122894,
+      0.052742667496204376,
+      0.004816360771656036,
+      0.05313955247402191,
+      0.058365289121866226,
+      0.05585050955414772,
+      0.05668928474187851,
+      0.05918566510081291,
+      -0.07712635397911072,
+      0.05374579131603241,
+      -0.036864474415779114,
+      0.03566460683941841,
+      0.05824630334973335,
+      0.04853316396474838,
+      0.04722554236650467,
+      0.03049103543162346,
+      0.04397154599428177,
+      0.04766076058149338,
+      0.05894605442881584,
+      0.04835444688796997,
+      0.05897827818989754,
+      0.05948491767048836,
+      0.05853684991598129,
+      -0.0050371624529361725,
+      0.0433323010802269,
+      0.017536941915750504,
+      0.04878716543316841,
+      0.139553040266037,
+      -0.10335907340049744,
+      0.014310088008642197,
+      0.058389849960803986,
+      0.012334678322076797,
+      0.03503337502479553,
+      0.05818603187799454,
+      -0.033800672739744186,
+      0.00250929594039917,
+      -0.04317232221364975,
+      0.04922649264335632,
+      0.05932055786252022,
+      0.05903784930706024,
+      -0.030892860144376755,
+      0.05768594145774841,
+      0.05492917448282242,
+      0.0585385337471962,
+      -0.037185169756412506,
+      0.05141330137848854,
+      0.05624351277947426,
+      0.027079083025455475,
+      0.05811110883951187,
+      0.05580241605639458,
+      0.05180075392127037,
+      0.059289414435625076,
+      0.05889549478888512,
+      0.059385765343904495,
+      0.05369181931018829,
+      0.05956219136714935,
+      0.05783536285161972,
+      0.05947178229689598,
+      0.013515457510948181,
+      0.05460158362984657,
+      0.05907632037997246,
+      0.057332370430231094,
+      -0.050216786563396454,
+      0.054461028426885605,
+      0.03663337230682373,
+      0.000616028904914856,
+      0.058966174721717834,
+      0.04703742265701294,
+      0.05880584940314293,
+      0.057512346655130386,
+      0.0582151785492897,
+      0.04311848804354668,
+      -0.0014651156961917877,
+      0.05786740407347679,
+      -0.046799030154943466,
+      0.05956188589334488,
+      0.05539989098906517,
+      0.0589139387011528,
+      0.059657514095306396,
+      0.041090983897447586,
+      0.05956335738301277,
+      0.05794830247759819,
+      0.0596679262816906,
+      0.058624159544706345,
+      0.05915384367108345,
+      0.03310283645987511,
+      0.05974904075264931,
+      0.059755839407444,
+      0.05966190621256828,
+      0.05948229879140854,
+      0.05970945581793785,
+      0.05973440781235695,
+      0.059694476425647736,
+      0.05963554605841637,
+      0.04694947227835655,
+      0.05674027279019356,
+      0.05950936675071716,
+      0.09874393045902252,
+      0.0596926175057888,
+      0.059702616184949875,
+      0.059700652956962585,
+      0.054619498550891876,
+      0.05940359830856323,
+      0.05964266508817673,
+      0.022495903074741364,
+      0.05973975360393524,
+      0.059399109333753586,
+      0.059739772230386734,
+      0.0556657612323761,
+      0.05971849709749222,
+      0.059714071452617645,
+      0.05973445624113083,
+      0.05944915488362312,
+      0.054755449295043945,
+      0.05894174054265022,
+      0.05930950120091438,
+      0.0562693253159523,
+      0.05884075164794922,
+      0.05962122231721878,
+      0.0595422238111496,
+      0.059548404067754745,
+      0.0590674914419651,
+      0.04833179712295532,
+      0.05761216953396797,
+      0.059610918164253235,
+      0.056937966495752335,
+      0.053049035370349884,
+      0.05736491456627846,
+      0.05686122924089432,
+      0.059667930006980896,
+      0.059086818248033524,
+      0.059756092727184296,
+      0.05975431576371193,
+      0.059522274881601334,
+      0.05975908786058426,
+      0.059763405472040176,
+      0.05977190285921097,
+      0.05972469598054886,
+      0.059680577367544174,
+      0.05973471701145172,
+      0.059741728007793427,
+      0.059243928641080856,
+      0.059769559651613235,
+      0.05975012108683586,
+      0.05976569280028343,
+      0.05974050238728523,
+      0.059772782027721405,
+      0.05975496768951416,
+      0.05976428464055061,
+      0.05976491793990135,
+      0.05975645035505295,
+      0.05971529334783554,
+      0.05977584794163704,
+      0.059770096093416214,
+      0.059777069836854935,
+      0.059777870774269104,
+      0.05976370349526405,
+      0.05977920815348625,
+      0.05977156385779381,
+      0.05978019908070564,
+      0.05960753560066223,
+      0.05977145582437515,
+      0.05972258746623993,
+      0.05977015197277069,
+      0.05977928265929222,
+      0.05977938696742058
+    ],
+    "base_loss": [
+      11.310925483703613,
+      9.799416542053223,
+      8.384845733642578,
+      7.012864589691162,
+      5.702121257781982,
+      4.063109397888184,
+      2.4006569385528564,
+      2.0750033855438232,
+      1.8905020952224731,
+      1.810624361038208,
+      1.7988861799240112,
+      1.746134877204895,
+      1.8305126428604126,
+      1.755860686302185,
+      1.6866179704666138,
+      1.4697428941726685,
+      1.1557385921478271,
+      0.9200530648231506,
+      0.8094637989997864,
+      0.7359910607337952,
+      0.7466462850570679,
+      0.6838196516036987,
+      0.6257291436195374,
+      0.5375270843505859,
+      0.5215278267860413,
+      0.47197288274765015,
+      0.5130500197410583,
+      0.37245097756385803,
+      0.34752556681632996,
+      0.37827953696250916,
+      0.28802791237831116,
+      0.2808120548725128,
+      0.2339651882648468,
+      0.20461882650852203,
+      0.17715759575366974,
+      0.2564390003681183,
+      0.2124192863702774,
+      0.1368967592716217,
+      0.10954953730106354,
+      0.36130785942077637,
+      0.33991819620132446,
+      0.1468440443277359,
+      0.13848432898521423,
+      0.10581625998020172,
+      0.0887078270316124,
+      0.10313636064529419,
+      0.049312930554151535,
+      0.10466252267360687,
+      0.1299249678850174,
+      0.11611846834421158,
+      0.07753632217645645,
+      0.07862601429224014,
+      0.09058528393507004,
+      0.09168340265750885,
+      0.06258507817983627,
+      0.05378057435154915,
+      0.11647278070449829,
+      0.0644892156124115,
+      0.04381830245256424,
+      0.022039126604795456,
+      0.04567083343863487,
+      0.0361330509185791,
+      0.0794965997338295,
+      0.02070070616900921,
+      0.04011081904172897,
+      0.05427050590515137,
+      0.04253954812884331,
+      0.062237195670604706,
+      0.029364323243498802,
+      0.019081851467490196,
+      0.04249581694602966,
+      0.06916357576847076,
+      0.032495785504579544,
+      0.04146444797515869,
+      0.03241477161645889,
+      0.024667885154485703,
+      0.01861412450671196,
+      0.022326601669192314,
+      0.030373752117156982,
+      0.00628579081967473,
+      0.03921641781926155,
+      0.020771946758031845,
+      0.026698973029851913,
+      0.03285898640751839,
+      0.016900036484003067,
+      0.017014572396874428,
+      0.07253045588731766,
+      0.010799786075949669,
+      0.013626283034682274,
+      0.011273736134171486,
+      0.02661147154867649,
+      0.008593114092946053,
+      0.013687387108802795,
+      0.00925882812589407,
+      0.01992531679570675,
+      0.014698164537549019,
+      0.005633269902318716,
+      0.0012565153883770108,
+      0.006300862412899733,
+      0.002103527309373021,
+      0.0018069046782329679,
+      0.01380141731351614,
+      0.020991848781704903,
+      0.013486634008586407,
+      0.002547323936596513,
+      0.003178002079948783,
+      0.004760467913001776,
+      0.012711070477962494,
+      0.011913101188838482,
+      0.0018395015504211187,
+      0.0005679978639818728,
+      0.0046846610493958,
+      0.001026674872264266,
+      0.004251865204423666,
+      0.004242083057761192,
+      0.008947940543293953,
+      0.00538821192458272,
+      0.010002645663917065,
+      0.0008911331533454359,
+      0.0013170059537515044,
+      0.005049239378422499,
+      9.116218279814348e-05,
+      0.002637359080836177,
+      0.008498182520270348,
+      0.0031002373434603214,
+      0.0035727322101593018,
+      0.006592695601284504,
+      0.00038529685116373,
+      0.004732525907456875,
+      0.006229202263057232,
+      0.009691894985735416,
+      0.0013353960821405053,
+      0.005324908997863531,
+      0.0007954480824992061,
+      0.009363139048218727,
+      0.00195595296099782,
+      0.00021370254398789257,
+      0.0017767505487427115,
+      0.0014224123442545533,
+      0.008522608317434788,
+      0.009148071520030499,
+      0.0005811649025417864,
+      0.00035937552456744015,
+      0.00023171375505626202,
+      0.0009247538982890546,
+      0.0008702775812707841,
+      0.006122277583926916,
+      0.000751205428969115,
+      0.00017612602096050978,
+      0.0004503035743255168,
+      0.00044899675413034856,
+      7.806277426425368e-05,
+      0.015226820483803749,
+      0.0007183189736679196,
+      0.010834530927240849,
+      0.0027013751678168774,
+      0.00020708993542939425,
+      0.0012633880833163857,
+      0.0014081543777137995,
+      0.003286610124632716,
+      0.001765353954397142,
+      0.0013586330460384488,
+      0.00010583313269307837,
+      0.0013700034469366074,
+      0.00012758016237057745,
+      4.926664041704498e-05,
+      0.00014701545296702534,
+      0.007233455777168274,
+      0.0018396369414404035,
+      0.004705718718469143,
+      0.0012312305625528097,
+      0.004972578026354313,
+      0.018140608444809914,
+      0.005064623896032572,
+      0.00016721177962608635,
+      0.005326803307980299,
+      0.0027723221573978662,
+      0.00019869880634360015,
+      0.010409857146441936,
+      0.0063795289024710655,
+      0.011448299512267113,
+      0.001179369050078094,
+      6.3560604758095e-05,
+      9.097588917938992e-05,
+      0.010082007385790348,
+      0.0002581911103334278,
+      0.0005510993651114404,
+      0.0001481231302022934,
+      0.010792314074933529,
+      0.0009537680307403207,
+      0.0009961611358448863,
+      0.0036459004040807486,
+      0.00019756561960093677,
+      0.00047167044249363244,
+      0.000894826662261039,
+      6.342893175315112e-05,
+      0.0001070911530405283,
+      5.0062812078977004e-05,
+      0.0006820369744673371,
+      3.11851326841861e-05,
+      0.00022797915153205395,
+      4.020973574370146e-05,
+      0.0058403415605425835,
+      0.0005822530365549028,
+      0.0001705260219750926,
+      0.00029142393032088876,
+      0.012235346250236034,
+      0.0005992701044306159,
+      0.0025830063968896866,
+      0.006582822185009718,
+      9.811115160118788e-05,
+      0.001422852510586381,
+      0.00011448275472503155,
+      0.0002578377316240221,
+      0.00017884199041873217,
+      0.001859683427028358,
+      0.006810321006923914,
+      0.00021835605730302632,
+      0.011848797090351582,
+      3.4483226045267656e-05,
+      0.0004916420439258218,
+      0.00010241410200251266,
+      1.8761482351692393e-05,
+      0.0020929411984980106,
+      2.9384027584455907e-05,
+      0.0002087716420646757,
+      1.921446164487861e-05,
+      0.00044575546053238213,
+      7.600494427606463e-05,
+      0.002989737782627344,
+      8.187857019947842e-06,
+      7.125973297661403e-06,
+      1.82550102181267e-05,
+      3.751000258489512e-05,
+      1.4275495232141111e-05,
+      9.399243936059065e-06,
+      1.365737716696458e-05,
+      1.989544580283109e-05,
+      0.0014294106513261795,
+      0.00034770320053212345,
+      3.405987081350759e-05,
+      0.007531650364398956,
+      1.3619711353385355e-05,
+      1.4104787624091841e-05,
+      1.4377613297256175e-05,
+      0.0005786583642475307,
+      4.6175522584235296e-05,
+      1.9812165191979147e-05,
+      0.004146917257457972,
+      8.439447810815182e-06,
+      4.5952801883686334e-05,
+      8.537003850506153e-06,
+      0.0004607641021721065,
+      1.0937952538370155e-05,
+      1.0741861842689104e-05,
+      8.608491043560207e-06,
+      4.0373131923843175e-05,
+      0.0005620215088129044,
+      9.689097350928932e-05,
+      5.5835869716247544e-05,
+      0.000422771176090464,
+      0.00011028878361685202,
+      2.2917036403669044e-05,
+      3.049142651434522e-05,
+      2.9726012144237757e-05,
+      8.316530875163153e-05,
+      0.0012756132055073977,
+      0.0002647296350914985,
+      2.286232302139979e-05,
+      0.000320378749165684,
+      0.0007517744670622051,
+      0.00027207095990888774,
+      0.00032771078986115754,
+      1.5925459592835978e-05,
+      8.063759742071852e-05,
+      6.395490800059633e-06,
+      6.174446298246039e-06,
+      3.201107028871775e-05,
+      5.368768142943736e-06,
+      4.924431323161116e-06,
+      4.161905053479131e-06,
+      9.321395737060811e-06,
+      1.4197724340192508e-05,
+      8.0997551776818e-06,
+      7.391088729491457e-06,
+      6.260239752009511e-05,
+      4.311116754251998e-06,
+      6.528679477924015e-06,
+      4.575043476506835e-06,
+      7.379913768090773e-06,
+      3.871808985422831e-06,
+      5.83504242968047e-06,
+      4.760214324051049e-06,
+      4.841781901632203e-06,
+      5.562903879763326e-06,
+      1.0011200174631085e-05,
+      3.3984865694947075e-06,
+      4.020857431896729e-06,
+      3.242575758122257e-06,
+      3.1238876090355916e-06,
+      4.732734851131681e-06,
+      2.9700756840611575e-06,
+      3.870981345244218e-06,
+      2.8165547973912908e-06,
+      2.195785054937005e-05,
+      3.799038722718251e-06,
+      9.127996236202307e-06,
+      3.9199071579787415e-06,
+      2.9934997201053193e-06,
+      2.8910419587191427e-06
+    ],
+    "info_loss": [
+      -2.0635433197021484,
+      -2.313345432281494,
+      -1.111259937286377,
+      -0.4605445861816406,
+      -0.2262115478515625,
+      -0.10146880149841309,
+      -0.05386519432067871,
+      -0.03334951400756836,
+      -0.045644283294677734,
+      -0.07910919189453125,
+      -0.08043348789215088,
+      -0.22512316703796387,
+      -0.7816925048828125,
+      -0.9737270474433899,
+      -1.043921709060669,
+      -0.9317139983177185,
+      -0.6748966574668884,
+      -0.5252886414527893,
+      -0.514843225479126,
+      -0.45303335785865784,
+      -0.5041386485099792,
+      -0.47193479537963867,
+      -0.42122167348861694,
+      -0.3984223008155823,
+      -0.4038047194480896,
+      -0.3208988904953003,
+      -0.40757298469543457,
+      -0.2746967375278473,
+      -0.2768715023994446,
+      -0.3321535587310791,
+      -0.24415062367916107,
+      -0.24934376776218414,
+      -0.20041123032569885,
+      -0.19142605364322662,
+      -0.1549244225025177,
+      -0.2339794635772705,
+      -0.1953475922346115,
+      -0.12568841874599457,
+      -0.09779418259859085,
+      -0.3470166027545929,
+      -0.29920122027397156,
+      -0.14147505164146423,
+      -0.13618959486484528,
+      -0.10291147232055664,
+      -0.08587231487035751,
+      -0.09974047541618347,
+      -0.048013728111982346,
+      -0.10290439426898956,
+      -0.12090639770030975,
+      -0.11380031704902649,
+      -0.07623849809169769,
+      -0.07499393820762634,
+      -0.07409591972827911,
+      -0.08624568581581116,
+      -0.060958873480558395,
+      -0.04709511995315552,
+      -0.1101842001080513,
+      -0.06408847123384476,
+      -0.04340897500514984,
+      -0.021622363477945328,
+      -0.038965873420238495,
+      -0.035954877734184265,
+      -0.07861441373825073,
+      -0.018788905814290047,
+      -0.03995459899306297,
+      -0.05253433436155319,
+      -0.04240024834871292,
+      -0.060936339199543,
+      -0.02915559895336628,
+      -0.01741478592157364,
+      -0.0294716227799654,
+      -0.06899968534708023,
+      -0.03209249675273895,
+      -0.040383897721767426,
+      -0.03232396021485329,
+      -0.02433980070054531,
+      -0.018455039709806442,
+      -0.021846972405910492,
+      -0.030315235257148743,
+      -0.006221582647413015,
+      -0.03914681822061539,
+      -0.02051035687327385,
+      -0.026625070720911026,
+      -0.03252970427274704,
+      -0.016828952357172966,
+      -0.016918081790208817,
+      -0.0724758729338646,
+      -0.00882614403963089,
+      -0.013493502512574196,
+      -0.002905219793319702,
+      -0.026566917076706886,
+      -0.008496482856571674,
+      -0.013621312566101551,
+      -0.009220397099852562,
+      -0.01986982300877571,
+      -0.014615782536566257,
+      -0.005594427231699228,
+      -0.0012273596366867423,
+      -0.006265625357627869,
+      -0.0020810074638575315,
+      -0.0017726317746564746,
+      0.004160921089351177,
+      -0.020458098500967026,
+      -0.00909888744354248,
+      -0.002524296287447214,
+      -0.0015061871381476521,
+      -0.004736113362014294,
+      -0.012686277739703655,
+      -0.011897302232682705,
+      -0.0017897504149004817,
+      -0.0005507023888640106,
+      -0.00392569275572896,
+      -0.0009925365447998047,
+      -0.004234787542372942,
+      -0.003944589290767908,
+      -0.008918046951293945,
+      -0.0053726984187960625,
+      -0.009991342201828957,
+      -0.0008730606059543788,
+      -0.0012967329239472747,
+      0.009363491088151932,
+      -7.105863187462091e-05,
+      -0.002624925458803773,
+      -0.008312009274959564,
+      -0.003083128249272704,
+      -0.0035224780440330505,
+      -0.0065766326151788235,
+      -0.0003755292564164847,
+      -0.0047224052250385284,
+      -0.006218502763658762,
+      -0.009681547060608864,
+      -0.0013222465058788657,
+      -0.004341441672295332,
+      -0.0007775184931233525,
+      -0.00931689515709877,
+      -0.001940092770382762,
+      -0.00019664398860186338,
+      -0.0017529590986669064,
+      -0.0014085874427109957,
+      -0.008509231731295586,
+      -0.009130778722465038,
+      -0.0005702226189896464,
+      -0.00033642895868979394,
+      -0.00022369525686372072,
+      -0.000913897471036762,
+      -0.0007934793247841299,
+      -0.0061111729592084885,
+      -0.0007417292799800634,
+      -0.00016213688650168478,
+      -0.0004410183464642614,
+      -0.00036239478504285216,
+      -7.085952529450879e-05,
+      -0.0152161605656147,
+      -0.0006775853107683361,
+      -0.010817117057740688,
+      -0.002685191808268428,
+      -0.00017656336422078311,
+      -0.0012533387634903193,
+      -0.0013984635006636381,
+      -0.0032750980462878942,
+      -0.0017596612451598048,
+      -0.0013503300724551082,
+      -9.615935414331034e-05,
+      -0.0012821635464206338,
+      -9.588881221134216e-05,
+      -3.930599268642254e-05,
+      -0.00014143930457066745,
+      -0.007227485533803701,
+      -0.001832550740800798,
+      -0.0046989587135612965,
+      -0.001226079068146646,
+      0.007476728409528732,
+      -0.018130844458937645,
+      -0.005057859234511852,
+      -0.00015836897364351898,
+      -0.005319643300026655,
+      -0.002766314195469022,
+      -0.00018857313261833042,
+      -0.010401567444205284,
+      -0.006372004747390747,
+      -0.011442441493272781,
+      -0.0011754423612728715,
+      -5.515823431778699e-05,
+      -8.578877168474719e-05,
+      -0.010077666491270065,
+      -0.0002401876263320446,
+      -0.0005430026794783771,
+      -0.00014177167031448334,
+      -0.010783873498439789,
+      -0.0009340771357528865,
+      -0.0004555671475827694,
+      -0.0036372279282659292,
+      -0.00018906689365394413,
+      -0.0004501818912103772,
+      -0.0008899735985323787,
+      -5.8661978982854635e-05,
+      -0.00010243752331007272,
+      -4.677473407355137e-05,
+      -0.0006793744396418333,
+      -2.77978069789242e-05,
+      -0.0002198476722696796,
+      -3.690836820169352e-05,
+      -0.005212762858718634,
+      -0.0005781368818134069,
+      -0.0001645042357267812,
+      -0.0002787997655104846,
+      -0.01222617831081152,
+      -0.00059433130081743,
+      -0.0025780166033655405,
+      -0.0065774559043347836,
+      -9.341209079138935e-05,
+      -0.0014195280382409692,
+      -0.00011187055497430265,
+      -0.0002552071528043598,
+      -0.00017651423695497215,
+      -0.0018565601203590631,
+      -0.006807839497923851,
+      -0.0002152066444978118,
+      -0.01184489019215107,
+      -2.8636333809117787e-05,
+      -0.0004893587320111692,
+      -9.915184637065977e-05,
+      -1.6133577446453273e-05,
+      -0.002083218889310956,
+      -2.6608182452037e-05,
+      -0.00020621051953639835,
+      -1.534211878606584e-05,
+      -0.0001628608733881265,
+      -7.258064579218626e-05,
+      -0.002969087101519108,
+      -6.575875886483118e-06,
+      -5.4086240197648294e-06,
+      -1.587281803949736e-05,
+      -3.571415800251998e-05,
+      -1.0543776625127066e-05,
+      -7.639584509888664e-06,
+      -1.1988988262601197e-05,
+      -1.8476646800991148e-05,
+      -0.0014280227478593588,
+      -0.00034074881114065647,
+      -3.246040432713926e-05,
+      0.003141222521662712,
+      -1.2182288628537208e-05,
+      -1.137391973315971e-05,
+      -1.170263385574799e-05,
+      -0.0005764527595601976,
+      -4.476524190977216e-05,
+      -1.773320764186792e-05,
+      -0.004145168699324131,
+      -6.858874257886782e-06,
+      -4.471574357012287e-05,
+      -6.982234481256455e-06,
+      -0.0004595858626998961,
+      -9.31720387598034e-06,
+      -9.667649464972783e-06,
+      -7.417732831527246e-06,
+      -3.912835745722987e-05,
+      -0.0005606719059869647,
+      -9.551156108500436e-05,
+      -5.465974390972406e-05,
+      -0.0003953570267185569,
+      -0.00010738395212683827,
+      -2.0354338630568236e-05,
+      -2.8831818781327456e-05,
+      -2.813772880472243e-05,
+      -8.154882380040362e-05,
+      -0.0012743944535031915,
+      -0.0002454226778354496,
+      -2.1390040274127387e-05,
+      -0.00031821534503251314,
+      -0.0007502481457777321,
+      -0.00027069926727563143,
+      -0.0003266235289629549,
+      -1.4777540854993276e-05,
+      -7.935611938592046e-05,
+      -5.0080998335033655e-06,
+      -5.162233719602227e-06,
+      -3.094670319114812e-05,
+      -4.601359250955284e-06,
+      -4.1764474190131295e-06,
+      -3.244953404646367e-06,
+      -8.482142220600508e-06,
+      -1.338508263870608e-05,
+      -7.358157290582312e-06,
+      -6.585363735212013e-06,
+      -6.188607949297875e-05,
+      -3.497940269880928e-06,
+      -5.661752766172867e-06,
+      -3.9098135857784655e-06,
+      -6.707765805913368e-06,
+      -3.130477125523612e-06,
+      -5.094242624181788e-06,
+      -4.048684786539525e-06,
+      -3.9993406062421855e-06,
+      -4.9152363317261916e-06,
+      -9.47289663599804e-06,
+      -2.758003347480553e-06,
+      -3.394211034901673e-06,
+      -2.626040668474161e-06,
+      -2.5211893444065936e-06,
+      -4.101831109437626e-06,
+      -2.370570427956409e-06,
+      -3.2251764423563145e-06,
+      -2.2545686988451052e-06,
+      -2.1435511371237226e-05,
+      -3.2304003525496228e-06,
+      -8.647168215247802e-06,
+      -3.3736203022272093e-06,
+      -2.365789441682864e-06,
+      -2.348214366065804e-06
+    ],
+    "abs_loss": [
+      0.6925318241119385,
+      0.6635907292366028,
+      0.5507833361625671,
+      0.4004248380661011,
+      0.16682057082653046,
+      0.04206141456961632,
+      0.014476890675723553,
+      0.00542783597484231,
+      0.0029513875488191843,
+      0.0018128822557628155,
+      0.0012926185736432672,
+      0.0013828948140144348,
+      0.0020498682279139757,
+      0.002931631403043866,
+      0.002957060933113098,
+      0.0046464367769658566,
+      0.014647457748651505,
+      0.03240418806672096,
+      0.03907397761940956,
+      0.03823111578822136,
+      0.02254408597946167,
+      0.024348080158233643,
+      0.017689481377601624,
+      0.03030112385749817,
+      0.046409592032432556,
+      0.030905960127711296,
+      0.020118050277233124,
+      0.011034250259399414,
+      0.005365070421248674,
+      0.009187811985611916,
+      0.008543974719941616,
+      0.012849290855228901,
+      0.003212135983631015,
+      0.0067325010895729065,
+      0.0044402615167200565,
+      0.002359983744099736,
+      0.0019478974863886833,
+      0.002157375914976001,
+      0.0014996955869719386,
+      0.001293656532652676,
+      0.001601194846443832,
+      0.0009104346972890198,
+      0.0020418146159499884,
+      0.0010058992775157094,
+      0.0011067298473790288,
+      0.0012411146890372038,
+      0.0005739941843785346,
+      0.0006972341798245907,
+      0.00034915204741992056,
+      0.0004249668272677809,
+      0.0006961261969991028,
+      0.0004823488707188517,
+      0.0005400722729973495,
+      0.00020159631094429642,
+      0.00018573210400063545,
+      0.00030457001412287354,
+      0.00021314824698492885,
+      0.0001479564671171829,
+      0.000139270254294388,
+      0.00022893685672897846,
+      0.00016601783863734454,
+      0.00013127988495398313,
+      8.048047311604023e-05,
+      7.58371243136935e-05,
+      3.070158345508389e-05,
+      5.991568832541816e-05,
+      4.715143222711049e-05,
+      0.00010874922736547887,
+      7.586099673062563e-05,
+      4.2453681089682505e-05,
+      3.758449020097032e-05,
+      2.9014046958764084e-05,
+      2.7874120860360563e-05,
+      4.064525637659244e-05,
+      3.756453224923462e-05,
+      1.9495879314490594e-05,
+      3.06228939734865e-05,
+      2.673196286195889e-05,
+      1.2957719263795298e-05,
+      1.818677810661029e-05,
+      4.3710209865821525e-05,
+      2.6113657440873794e-05,
+      1.833925307437312e-05,
+      1.4472763723460957e-05,
+      1.855933624028694e-05,
+      1.8815404473571107e-05,
+      1.1463500413810834e-05,
+      9.954370398190804e-06,
+      2.3447893909178674e-05,
+      2.5065543013624847e-05,
+      2.139899333997164e-05,
+      1.6565756595809944e-05,
+      2.058000063698273e-05,
+      1.3646508705278393e-05,
+      9.937633876688778e-06,
+      1.4980471860326361e-05,
+      2.2796646589995362e-05,
+      1.742701351759024e-05,
+      1.5003882253949996e-05,
+      1.6086803952930495e-05,
+      1.578393130330369e-05,
+      8.697680641489569e-06,
+      1.0183452104683965e-05,
+      1.0108145033882465e-05,
+      7.94382140156813e-06,
+      1.4516254850605037e-05,
+      1.0322745765733998e-05,
+      6.135554940556176e-06,
+      8.462091500405222e-06,
+      6.296409537753789e-06,
+      1.2578590940393042e-05,
+      4.083773546881275e-06,
+      2.252038575534243e-05,
+      3.2620373531244695e-06,
+      4.216282377456082e-06,
+      1.0597424079605844e-05,
+      4.495684606808936e-06,
+      3.3793021430028602e-06,
+      3.5851933262165403e-06,
+      5.720341960113728e-06,
+      7.743328751530498e-06,
+      4.991172318113968e-06,
+      4.226472810842097e-06,
+      8.85880581336096e-06,
+      5.214235443418147e-06,
+      4.831796559301438e-06,
+      3.4290417261217954e-06,
+      8.674166565469932e-06,
+      2.2497399640997173e-06,
+      7.179805834311992e-06,
+      7.187655228335643e-06,
+      3.7581130527541973e-06,
+      6.32623050478287e-06,
+      3.863962774630636e-06,
+      2.810998239510809e-06,
+      2.2326548787532374e-06,
+      3.0928947580832755e-06,
+      1.0308237506251317e-05,
+      4.015436843474163e-06,
+      6.304836006165715e-06,
+      2.4586156541772652e-06,
+      3.503255356918089e-06,
+      1.6872210153451306e-06,
+      1.9625583718152484e-06,
+      1.974357019207673e-06,
+      2.1463572466018377e-06,
+      7.14312591298949e-07,
+      1.1067141940657166e-06,
+      1.055180746334372e-06,
+      6.643402343797788e-07,
+      2.6511286250752164e-06,
+      4.881461336481152e-06,
+      3.050929763048771e-06,
+      4.007692496088566e-06,
+      1.5863021189943538e-06,
+      1.858267751231324e-06,
+      2.4592661702627083e-06,
+      3.2402440410805866e-06,
+      1.7511750911580748e-06,
+      3.1052422855282202e-06,
+      3.1968347684596665e-06,
+      2.0923459942423506e-06,
+      1.1796670378316776e-06,
+      6.907254714860755e-07,
+      5.677932790604245e-07,
+      5.603444606094854e-07,
+      7.621268309776497e-07,
+      5.569265226768039e-07,
+      1.7713397255647578e-06,
+      1.1076430155299022e-06,
+      7.546779556832917e-07,
+      7.556102445960278e-07,
+      3.29996879600003e-07,
+      1.817533302528318e-06,
+      1.4325846677820664e-06,
+      3.9022347664285917e-07,
+      2.9162395094317617e-06,
+      5.618836439680308e-06,
+      1.7415408137821942e-06,
+      1.190218995361647e-06,
+      7.965872441673127e-07,
+      1.3401621572484146e-06,
+      4.03572101959071e-07,
+      4.786978138326958e-07,
+      7.590222708131478e-07,
+      1.5996855609046179e-06,
+      5.404766056926746e-07,
+      1.4103151215749676e-06,
+      2.4928371544774564e-07,
+      4.2375040720799007e-07,
+      1.5677251496981626e-07,
+      4.532414550340036e-07,
+      4.619341780198738e-07,
+      4.6938376385696756e-07,
+      6.500592917291215e-07,
+      7.568486921627482e-07,
+      1.3544344028559863e-06,
+      1.031276156027161e-06,
+      7.798237788847473e-07,
+      9.94329070636013e-07,
+      9.232276170223486e-07,
+      7.360501399489294e-07,
+      5.764865704804834e-07,
+      6.559574217135378e-07,
+      8.096249075606465e-07,
+      1.3035205483902246e-06,
+      6.388824544956151e-07,
+      6.044259066584345e-07,
+      5.168816983314173e-07,
+      3.908417909315176e-07,
+      5.448176239042368e-07,
+      4.966995561517251e-07,
+      1.0896460622689119e-07,
+      5.535026730285608e-07,
+      2.4524752006982453e-07,
+      3.504859193981247e-07,
+      3.48002430428096e-07,
+      4.662791468490468e-07,
+      8.915715739021834e-07,
+      1.199829512188444e-06,
+      2.778429859517928e-07,
+      2.5394015779056645e-07,
+      2.399702623279154e-07,
+      7.552825991297141e-07,
+      2.558026039878314e-07,
+      1.5335757552747964e-07,
+      1.0492890822888512e-07,
+      7.140133106986468e-08,
+      3.989133290360769e-07,
+      3.4458767572687066e-07,
+      2.002336287887374e-07,
+      4.77142236832151e-07,
+      5.302283057062596e-07,
+      2.3997020548449655e-07,
+      1.2697013573870208e-07,
+      2.045802602879121e-07,
+      2.840512820512231e-07,
+      1.8905819842984783e-07,
+      2.6449490064806014e-07,
+      2.3531379156338517e-07,
+      2.545610300330736e-07,
+      2.4742095661167696e-07,
+      2.6790982587954204e-07,
+      2.2631088825164625e-07,
+      4.153644397320022e-07,
+      2.5238745138267404e-07,
+      3.8991274209365656e-07,
+      2.8125870699113875e-07,
+      1.7291530696184054e-07,
+      2.8622577019632445e-07,
+      2.5331925712634984e-07,
+      2.3469289089916856e-07,
+      1.5459924895822041e-07,
+      4.5727770725534356e-07,
+      2.3562421347378404e-07,
+      2.461792121266626e-07,
+      1.9650873639420752e-07,
+      1.440443071487607e-07,
+      2.2103345997948054e-07,
+      1.4311301299585466e-07,
+      2.604572557629581e-07,
+      3.613521641909756e-07,
+      2.7753307563216367e-07,
+      2.626323691856669e-07,
+      3.7842542610633245e-07,
+      2.6666683083931275e-07,
+      3.259616221384931e-07,
+      1.8315971317406365e-07,
+      8.847556642876953e-08,
+      9.002773282418275e-08,
+      1.4311298457414523e-07,
+      9.68574411785994e-08,
+      1.518050254389891e-07,
+      1.316267770334889e-07,
+      8.723383615460989e-08,
+      1.344207163356259e-07,
+      1.1889871842640787e-07,
+      1.7819257891460438e-07,
+      8.785468708083499e-08,
+      7.41953343208479e-08,
+      1.0989597853949817e-07,
+      7.2022245944936e-08,
+      1.19830033895596e-07,
+      1.2976411767340323e-07,
+      1.322476776977055e-07,
+      7.543707880586226e-08,
+      1.0461848631848625e-07,
+      2.0768429465078952e-07,
+      1.20140498438559e-07,
+      1.1889875395354466e-07,
+      1.4745920395853318e-07,
+      1.1113773012993988e-07,
+      1.192091971802256e-07,
+      1.4714872520471545e-07,
+      1.0586025922521003e-07,
+      6.98491575690241e-08,
+      1.452860800554845e-07,
+      1.1362124041625066e-07,
+      1.4528610847719392e-07,
+      8.53711625836695e-08,
+      7.109093047574788e-08,
+      1.2728044396226323e-07,
+      9.934100120290168e-08,
+      8.723382194375517e-08,
+      8.785472260797178e-08,
+      8.816513030751594e-08,
+      7.450577044210149e-08,
+      9.43739522085707e-08,
+      8.25772232815325e-08,
+      1.0554978757681965e-07
+    ],
+    "zipf_loss": [
+      13.29382610321045,
+      12.052526473999023,
+      9.994089126586914,
+      7.528794288635254,
+      5.52280330657959,
+      4.794320106506348,
+      4.503414154052734,
+      4.418521881103516,
+      4.396157264709473,
+      4.387367248535156,
+      4.3840532302856445,
+      4.385408878326416,
+      4.388075828552246,
+      4.393425941467285,
+      4.37654972076416,
+      4.266912937164307,
+      3.8286871910095215,
+      3.477245807647705,
+      3.14785099029541,
+      2.818744659423828,
+      2.6106979846954346,
+      2.596252918243408,
+      2.1163034439086914,
+      1.200337290763855,
+      0.10290686786174774,
+      0.10047288239002228,
+      0.07882359623908997,
+      0.07696527242660522,
+      0.07642862200737,
+      0.06858339160680771,
+      0.068836510181427,
+      0.06697411835193634,
+      0.06984466314315796,
+      0.06593257933855057,
+      0.06418862193822861,
+      0.06744727492332458,
+      0.06615877896547318,
+      0.06379882246255875,
+      0.06838598847389221,
+      0.06169579178094864,
+      0.06388387084007263,
+      0.06509333103895187,
+      0.06235514208674431,
+      0.061683546751737595,
+      0.06281059980392456,
+      0.062894307076931,
+      0.061615023761987686,
+      0.06123078987002373,
+      0.060643307864665985,
+      0.060334689915180206,
+      0.06114167720079422,
+      0.06072838976979256,
+      0.060264453291893005,
+      0.06059280410408974,
+      0.06139298900961876,
+      0.0615326464176178,
+      0.06016057729721069,
+      0.06034613773226738,
+      0.06080038473010063,
+      0.06517796218395233,
+      0.05992530286312103,
+      0.05989722162485123,
+      0.059903357177972794,
+      0.059921324253082275,
+      0.060017216950654984,
+      0.05986054614186287,
+      0.05984221026301384,
+      0.05988983437418938,
+      0.05983946472406387,
+      0.06094572693109512,
+      0.05992075800895691,
+      0.0598667673766613,
+      0.059897053986787796,
+      0.05986611917614937,
+      0.05994047969579697,
+      0.059944018721580505,
+      0.06013903766870499,
+      0.059919144958257675,
+      0.059939704835414886,
+      0.05983421579003334,
+      0.06011880561709404,
+      0.05988098308444023,
+      0.0598294660449028,
+      0.059852566570043564,
+      0.05983050540089607,
+      0.0599081888794899,
+      0.059903476387262344,
+      0.05990634486079216,
+      0.05982249230146408,
+      0.059852179139852524,
+      0.059840261936187744,
+      0.0598263256251812,
+      0.05991503596305847,
+      0.05986020714044571,
+      0.05986323207616806,
+      0.05988006293773651,
+      0.05981263890862465,
+      0.05981520935893059,
+      0.05991962179541588,
+      0.05980763956904411,
+      0.05981043726205826,
+      0.05983393266797066,
+      0.059836722910404205,
+      0.05982842296361923,
+      0.05981708690524101,
+      0.059857383370399475,
+      0.05986465513706207,
+      0.05989597365260124,
+      0.05981479957699776,
+      0.059847649186849594,
+      0.059842634946107864,
+      0.05980671942234039,
+      0.059830423444509506,
+      0.05983607470989227,
+      0.0598045215010643,
+      0.059806935489177704,
+      0.05980488657951355,
+      0.059804659336805344,
+      0.059804633259773254,
+      0.0598398819565773,
+      0.0598074309527874,
+      0.059813737869262695,
+      0.05980728939175606,
+      0.05980706587433815,
+      0.059818610548973083,
+      0.05982007831335068,
+      0.05981971323490143,
+      0.05985965207219124,
+      0.0598074309527874,
+      0.05981968343257904,
+      0.059803977608680725,
+      0.05983014404773712,
+      0.05982097238302231,
+      0.05980496481060982,
+      0.05983216315507889,
+      0.05980691313743591,
+      0.05992106348276138,
+      0.05980047583580017,
+      0.05980249494314194,
+      0.05993174761533737,
+      0.059802860021591187,
+      0.05980147793889046,
+      0.059805456548929214,
+      0.059825535863637924,
+      0.059808362275362015,
+      0.05980696901679039,
+      0.059805743396282196,
+      0.059805531054735184,
+      0.05981042608618736,
+      0.059810321778059006,
+      0.059863969683647156,
+      0.05981570854783058,
+      0.059808116406202316,
+      0.05980292335152626,
+      0.06047201156616211,
+      0.05981496348977089,
+      0.05980459973216057,
+      0.059802837669849396,
+      0.059801846742630005,
+      0.05995509400963783,
+      0.05980248376727104,
+      0.05980521813035011,
+      0.059801697731018066,
+      0.05980600789189339,
+      0.05980952829122543,
+      0.05982865393161774,
+      0.05980415269732475,
+      0.060004185885190964,
+      0.05981799587607384,
+      0.05982069671154022,
+      0.05981665104627609,
+      0.05981309711933136,
+      0.05980873107910156,
+      0.05982387438416481,
+      0.059806182980537415,
+      0.06020427122712135,
+      0.05992390215396881,
+      0.05987250432372093,
+      0.05980497971177101,
+      0.05984969437122345,
+      0.059803709387779236,
+      0.059801410883665085,
+      0.05980853736400604,
+      0.05980471149086952,
+      0.059801723808050156,
+      0.0598294660449028,
+      0.059808049350976944,
+      0.05980798602104187,
+      0.05986122786998749,
+      0.05980026349425316,
+      0.059803009033203125,
+      0.05980541557073593,
+      0.05980416759848595,
+      0.05983251705765724,
+      0.059805598109960556,
+      0.059812530875205994,
+      0.05981264263391495,
+      0.059803348034620285,
+      0.059803448617458344,
+      0.0598088838160038,
+      0.059805769473314285,
+      0.059800583869218826,
+      0.05980268865823746,
+      0.059800632297992706,
+      0.060550756752491,
+      0.05982881411910057,
+      0.05980958789587021,
+      0.0598050095140934,
+      0.05983047932386398,
+      0.059807728976011276,
+      0.05980212986469269,
+      0.05980980023741722,
+      0.05981006100773811,
+      0.05980652570724487,
+      0.059801455587148666,
+      0.059824369847774506,
+      0.05980292707681656,
+      0.059801068156957626,
+      0.05980097874999046,
+      0.059813644737005234,
+      0.05980180948972702,
+      0.05980301648378372,
+      0.05980006605386734,
+      0.05983015522360802,
+      0.059800028800964355,
+      0.0598016194999218,
+      0.05980212241411209,
+      0.059807006269693375,
+      0.05980360507965088,
+      0.05980393663048744,
+      0.059806592762470245,
+      0.05980275198817253,
+      0.059802327305078506,
+      0.05980190634727478,
+      0.05980060622096062,
+      0.059801384806632996,
+      0.059800680726766586,
+      0.0598003976047039,
+      0.05980026349425316,
+      0.059800032526254654,
+      0.059799887239933014,
+      0.05980002507567406,
+      0.05980079248547554,
+      0.05980222672224045,
+      0.05980325862765312,
+      0.05980534479022026,
+      0.05980503559112549,
+      0.0598001554608345,
+      0.0598006509244442,
+      0.05979987606406212,
+      0.05980028957128525,
+      0.05980103462934494,
+      0.05980084091424942,
+      0.059800684452056885,
+      0.059799984097480774,
+      0.05979999899864197,
+      0.05980004742741585,
+      0.05980013310909271,
+      0.05979994311928749,
+      0.059800248593091965,
+      0.059800099581480026,
+      0.05980426445603371,
+      0.059801820665597916,
+      0.05980002507567406,
+      0.05980001762509346,
+      0.059799786657094955,
+      0.05980009585618973,
+      0.059801649302244186,
+      0.05980194732546806,
+      0.05979973077774048,
+      0.05979972705245018,
+      0.05979982763528824,
+      0.059799738228321075,
+      0.05979976803064346,
+      0.05979973450303078,
+      0.059799764305353165,
+      0.05979975312948227,
+      0.059799712151288986,
+      0.05979972332715988,
+      0.05980023741722107,
+      0.059800177812576294,
+      0.05980018898844719,
+      0.05980021879076958,
+      0.05980018526315689,
+      0.059800177812576294,
+      0.059800177812576294,
+      0.05980021879076958,
+      0.05980018898844719,
+      0.05980020388960838,
+      0.05980018898844719,
+      0.059800200164318085,
+      0.05980006232857704,
+      0.05979999899864197,
+      0.059800054877996445,
+      0.059800028800964355,
+      0.059800002723932266,
+      0.05980001389980316,
+      0.059800006449222565,
+      0.05980007350444794,
+      0.05979995056986809,
+      0.059799980372190475,
+      0.059799931943416595,
+      0.059799935668706894,
+      0.0597999207675457,
+      0.059799924492836,
+      0.05979995056986809,
+      0.059799924492836,
+      0.059799958020448685,
+      0.05979993939399719,
+      0.05979996919631958
+    ],
+    "denoise_loss": [],
+    "ortho_loss": [
+      0.8192447423934937,
+      0.3226068913936615,
+      0.24498559534549713,
+      0.23789481818675995,
+      0.24731597304344177,
+      0.23505021631717682,
+      0.18448671698570251,
+      0.11628758162260056,
+      0.053730081766843796,
+      0.024227388203144073,
+      0.015414939261972904,
+      0.010050086304545403,
+      0.019424647092819214,
+      0.02843361161649227,
+      0.030290493741631508,
+      0.043113455176353455,
+      0.04085356369614601,
+      0.04222544655203819,
+      0.042552266269922256,
+      0.04996667802333832,
+      0.07985469698905945,
+      0.09776149690151215,
+      0.10674897581338882,
+      0.10868319869041443,
+      0.2814685106277466,
+      0.2746294140815735,
+      0.24647156894207,
+      0.22990547120571136,
+      0.20750808715820312,
+      0.20846304297447205,
+      0.20174385607242584,
+      0.1841120570898056,
+      0.18434445559978485,
+      0.1725008636713028,
+      0.1800413727760315,
+      0.19588550925254822,
+      0.20477144420146942,
+      0.2113926112651825,
+      0.2106967270374298,
+      0.20985224843025208,
+      0.21672116219997406,
+      0.2428370863199234,
+      0.2602790594100952,
+      0.26023972034454346,
+      0.24517148733139038,
+      0.2509886920452118,
+      0.2541520595550537,
+      0.24424588680267334,
+      0.24596774578094482,
+      0.2421128749847412,
+      0.24043391644954681,
+      0.23262101411819458,
+      0.23447059094905853,
+      0.2472395896911621,
+      0.23966433107852936,
+      0.237721785902977,
+      0.2402806580066681,
+      0.23643451929092407,
+      0.24419990181922913,
+      0.24225378036499023,
+      0.23691637814044952,
+      0.24816496670246124,
+      0.2520442605018616,
+      0.24585409462451935,
+      0.2490188628435135,
+      0.24445101618766785,
+      0.23566071689128876,
+      0.23848703503608704,
+      0.252482533454895,
+      0.2609865665435791,
+      0.2609159052371979,
+      0.2533751130104065,
+      0.2566130757331848,
+      0.24835678935050964,
+      0.2488221675157547,
+      0.13689453899860382,
+      0.11902143806219101,
+      0.17307281494140625,
+      0.16171930730342865,
+      0.1602713167667389,
+      0.15745322406291962,
+      0.16569939255714417,
+      0.16126221418380737,
+      0.162596195936203,
+      0.17070019245147705,
+      0.16535352170467377,
+      0.1664435714483261,
+      0.165073424577713,
+      0.16493500769138336,
+      0.16352717578411102,
+      0.169764906167984,
+      0.17214401066303253,
+      0.16921930015087128,
+      0.16899257898330688,
+      0.171036034822464,
+      0.16999980807304382,
+      0.16966712474822998,
+      0.1508668214082718,
+      0.12843312323093414,
+      0.1315062791109085,
+      0.1300763189792633,
+      0.12850038707256317,
+      0.13403534889221191,
+      0.12983189523220062,
+      0.13691434264183044,
+      0.13802994787693024,
+      0.1373082399368286,
+      0.1503422111272812,
+      0.14906148612499237,
+      0.15495505928993225,
+      0.15838924050331116,
+      0.1611107587814331,
+      0.15877504646778107,
+      0.16601279377937317,
+      0.17024734616279602,
+      0.16049383580684662,
+      0.16157326102256775,
+      0.16412128508090973,
+      0.16508516669273376,
+      0.16698452830314636,
+      0.16838505864143372,
+      0.16861510276794434,
+      0.16378405690193176,
+      0.15818025171756744,
+      0.16436266899108887,
+      0.19061677157878876,
+      0.18932411074638367,
+      0.23579120635986328,
+      0.23538817465305328,
+      0.23654742538928986,
+      0.23973789811134338,
+      0.24603033065795898,
+      0.2452835887670517,
+      0.24175921082496643,
+      0.23287618160247803,
+      0.2417069673538208,
+      0.24065981805324554,
+      0.23087656497955322,
+      0.23905833065509796,
+      0.2365914285182953,
+      0.23521064221858978,
+      0.23564133048057556,
+      0.2400449812412262,
+      0.23880471289157867,
+      0.24645718932151794,
+      0.23775553703308105,
+      0.23691301047801971,
+      0.2386942207813263,
+      0.2437121421098709,
+      0.24653872847557068,
+      0.24968571960926056,
+      0.25067123770713806,
+      0.259551465511322,
+      0.2558160424232483,
+      0.2590336799621582,
+      0.26523345708847046,
+      0.2590591609477997,
+      0.26310503482818604,
+      0.26530927419662476,
+      0.26245105266571045,
+      0.263492614030838,
+      0.25769704580307007,
+      0.26632019877433777,
+      0.25823941826820374,
+      0.2589738070964813,
+      0.2687494456768036,
+      0.26113688945770264,
+      0.26612570881843567,
+      0.26414385437965393,
+      0.26352113485336304,
+      0.26099491119384766,
+      0.2584558427333832,
+      0.2697486877441406,
+      0.2687203586101532,
+      0.2690061032772064,
+      0.2580570578575134,
+      0.2624073326587677,
+      0.26255661249160767,
+      0.26516199111938477,
+      0.26516225934028625,
+      0.26107510924339294,
+      0.2618436813354492,
+      0.2664620876312256,
+      0.26279518008232117,
+      0.26451313495635986,
+      0.27482345700263977,
+      0.28570905327796936,
+      0.293211966753006,
+      0.28172630071640015,
+      0.27277693152427673,
+      0.2753519117832184,
+      0.27690035104751587,
+      0.27665194869041443,
+      0.27643153071403503,
+      0.2718546390533447,
+      0.27435702085494995,
+      0.28555169701576233,
+      0.2853988707065582,
+      0.2858459949493408,
+      0.28633058071136475,
+      0.28917455673217773,
+      0.29062989354133606,
+      0.28804048895835876,
+      0.2844852805137634,
+      0.28696921467781067,
+      0.290794312953949,
+      0.28353363275527954,
+      0.27847498655319214,
+      0.2825830280780792,
+      0.2862725555896759,
+      0.2836776077747345,
+      0.2844274342060089,
+      0.28164389729499817,
+      0.2808953523635864,
+      0.2808036804199219,
+      0.28382906317710876,
+      0.28593796491622925,
+      0.285183846950531,
+      0.2836042046546936,
+      0.29209989309310913,
+      0.2900092899799347,
+      0.29599589109420776,
+      0.29488179087638855,
+      0.2934323847293854,
+      0.28693267703056335,
+      0.29013755917549133,
+      0.2893297076225281,
+      0.28935614228248596,
+      0.28972432017326355,
+      0.28677141666412354,
+      0.2852571904659271,
+      0.2880670428276062,
+      0.2882896363735199,
+      0.28808873891830444,
+      0.287496417760849,
+      0.288765013217926,
+      0.28815293312072754,
+      0.28787967562675476,
+      0.287559449672699,
+      0.2869146168231964,
+      0.28561538457870483,
+      0.2829693555831909,
+      0.2942989766597748,
+      0.2955695390701294,
+      0.29467928409576416,
+      0.29461243748664856,
+      0.2965194582939148,
+      0.29633012413978577,
+      0.29645296931266785,
+      0.29615551233291626,
+      0.29595425724983215,
+      0.29574158787727356,
+      0.2956746816635132,
+      0.2955031096935272,
+      0.2953500747680664,
+      0.29531121253967285,
+      0.29525116086006165,
+      0.2951180040836334,
+      0.295635849237442,
+      0.2953846752643585,
+      0.29522907733917236,
+      0.2960355877876282,
+      0.29410576820373535,
+      0.29393115639686584,
+      0.2938596308231354,
+      0.29382357001304626,
+      0.2944236099720001,
+      0.2951282858848572,
+      0.2973771393299103,
+      0.29375308752059937,
+      0.29427239298820496,
+      0.29283684492111206,
+      0.2930491864681244,
+      0.2903555929660797,
+      0.29138854146003723,
+      0.29148614406585693,
+      0.29164984822273254,
+      0.2917673587799072,
+      0.2918137311935425,
+      0.29192957282066345,
+      0.2922983765602112,
+      0.2923275828361511,
+      0.2923927903175354,
+      0.292471319437027,
+      0.29252102971076965,
+      0.29258009791374207,
+      0.29213449358940125,
+      0.2922602891921997,
+      0.29227909445762634,
+      0.29233628511428833,
+      0.2923688292503357,
+      0.29241418838500977,
+      0.2924383878707886,
+      0.29249149560928345,
+      0.2925289571285248,
+      0.2924870550632477,
+      0.2931862771511078,
+      0.293450266122818,
+      0.2932688891887665,
+      0.2931431531906128,
+      0.2931685149669647,
+      0.29325494170188904,
+      0.2933051884174347,
+      0.2933022677898407,
+      0.29332563281059265,
+      0.29333022236824036,
+      0.29332131147384644,
+      0.29330581426620483,
+      0.2933069169521332,
+      0.29334360361099243
+    ],
+    "lr": [
+      4.188034188034189e-06,
+      8.461538461538462e-06,
+      1.2735042735042738e-05,
+      1.700854700854701e-05,
+      2.1282051282051282e-05,
+      2.5555555555555557e-05,
+      2.9829059829059833e-05,
+      3.4102564102564105e-05,
+      3.837606837606838e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      3.992995647012267e-05,
+      3.963316185199842e-05,
+      3.933636723387416e-05,
+      3.90395726157499e-05,
+      3.874277799762565e-05,
+      3.844598337950139e-05,
+      3.814918876137713e-05,
+      3.785239414325287e-05,
+      3.7555599525128613e-05,
+      3.725880490700435e-05,
+      3.696201028888009e-05,
+      3.666521567075585e-05,
+      3.6368421052631585e-05,
+      3.607162643450732e-05,
+      3.577483181638307e-05,
+      3.547803719825881e-05,
+      3.518124258013455e-05,
+      3.4884447962010287e-05,
+      3.458765334388604e-05,
+      3.4290858725761776e-05,
+      3.399406410763751e-05,
+      3.3697269489513265e-05,
+      3.3400474871389e-05,
+      3.310368025326474e-05,
+      3.280688563514048e-05,
+      3.251009101701623e-05,
+      3.2213296398891967e-05,
+      3.191650178076771e-05,
+      3.1542540561931145e-05,
+      3.124574594380688e-05,
+      3.094895132568263e-05,
+      3.065215670755838e-05,
+      3.0355362089434113e-05,
+      3.0058567471309854e-05,
+      2.97617728531856e-05,
+      2.9464978235061336e-05,
+      2.9168183616937077e-05,
+      2.8871388998812818e-05,
+      2.857459438068857e-05,
+      2.8277799762564307e-05,
+      2.798100514444005e-05,
+      2.7684210526315793e-05,
+      2.738741590819153e-05,
+      2.709062129006727e-05,
+      2.6793826671943013e-05,
+      2.6497032053818764e-05,
+      2.62002374356945e-05,
+      2.5903442817570243e-05,
+      2.5606648199445987e-05,
+      2.5309853581321725e-05,
+      2.5013058963197466e-05,
+      2.4716264345073214e-05,
+      2.441946972694896e-05,
+      2.4122675108824693e-05,
+      2.3825880490700434e-05,
+      2.352908587257618e-05,
+      2.3232291254451916e-05,
+      2.293549663632766e-05,
+      2.2638702018203408e-05,
+      2.226474079936684e-05,
+      2.196794618124258e-05,
+      2.1671151563118328e-05,
+      2.1374356944994066e-05,
+      2.1077562326869803e-05,
+      2.0780767708745554e-05,
+      2.0483973090621296e-05,
+      2.018717847249703e-05,
+      1.989038385437277e-05,
+      1.959358923624852e-05,
+      1.929679461812426e-05,
+      1.8999999999999998e-05,
+      1.8703205381875746e-05,
+      1.8406410763751487e-05,
+      1.8109616145627224e-05,
+      1.7812821527502965e-05,
+      1.7516026909378713e-05,
+      1.7219232291254454e-05,
+      1.6922437673130192e-05,
+      1.662564305500594e-05,
+      1.632884843688168e-05,
+      1.603205381875742e-05,
+      1.573525920063316e-05,
+      1.5438464582508908e-05,
+      1.5141669964384647e-05,
+      1.4844875346260385e-05,
+      1.4548080728136134e-05,
+      1.4251286110011875e-05,
+      1.3954491491887611e-05,
+      1.3657696873763363e-05,
+      1.33609022556391e-05,
+      1.298694103680253e-05,
+      1.2690146418678277e-05,
+      1.2393351800554018e-05,
+      1.2096557182429756e-05,
+      1.1799762564305506e-05,
+      1.1502967946181245e-05,
+      1.1206173328056984e-05,
+      1.0909378709932724e-05,
+      1.0612584091808472e-05,
+      1.0315789473684213e-05,
+      1.0018994855559949e-05,
+      9.7222002374357e-06,
+      9.425405619311438e-06,
+      9.128611001187177e-06,
+      8.831816383062916e-06,
+      8.535021764938666e-06,
+      8.238227146814405e-06,
+      7.941432528690143e-06,
+      7.644637910565893e-06,
+      7.347843292441632e-06,
+      7.0510486743173715e-06,
+      6.75425405619311e-06,
+      6.45745943806886e-06,
+      6.160664819944598e-06,
+      5.8638702018203375e-06,
+      5.567075583696086e-06,
+      5.270280965571826e-06,
+      4.973486347447565e-06,
+      4.676691729323314e-06,
+      4.379897111199053e-06,
+      4.083102493074792e-06
+    ],
+    "emb_lr": [],
+    "eval_step": [
+      1550,
+      3113,
+      4676,
+      6239,
+      7802,
+      9365,
+      10928,
+      12491,
+      14054,
+      15617
+    ],
+    "eval_accuracy": [
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0
+    ]
+  },
+  "final_accuracy": 1.0,
+  "sft_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": null,
+      "mode": "sft",
+      "n_digits": 6,
+      "n_per_split": 100
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 605
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 95
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 204
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 169
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 31
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 296
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 163
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 130
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 87
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 203
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 117
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 49
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 104
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 106
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 23
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 0.99,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.99,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 234
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 601
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 99
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 279
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 145
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 24
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 252
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 213
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 113
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 85
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 181
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 108
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 600
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 267
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 53
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 0.9995833333333334,
+      "total_examples": 2400,
+      "n_splits": 22
+    }
+  },
+  "sorl_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": 1,
+      "mode": "sorl",
+      "n_digits": 6,
+      "n_per_split": 100
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 605
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 95
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 204
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 169
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 31
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 296
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 163
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 130
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 87
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 203
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 117
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 49
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 104
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 106
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 23
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 234
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 601
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 99
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 279
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 145
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 24
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 252
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 213
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 113
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 85
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 181
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 108
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 600
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 267
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 53
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 1.0,
+      "total_examples": 2400,
+      "n_splits": 22
+    }
+  },
+  "sorl_overall_accuracy": 1.0,
+  "sft_overall_accuracy": 0.9995833333333334
+}

add_sub_sorl_v1_abs2_K1_100K/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a3de8400fb4eebcbaa3ba12d6d173f911f1a0fcd4e8f9004309068e02216384
+size 650271004

add_sub_sorl_v1_abs2_K1_100K/train_config.json

@@ -0,0 +1,84 @@
+{
+  "num_rollouts": 4,
+  "K": 1,
+  "max_iterations": 2,
+  "memory_span_abs": 1792,
+  "memory_span_traj": 1792,
+  "temperature": 1.0,
+  "ar_search": false,
+  "response_only_abs": false,
+  "alpha_info_gain": 10.0,
+  "alpha_abs": 0.1,
+  "alpha_soft_zipf": 1.0,
+  "alpha_ortho": 0.0,
+  "alpha_anchor": 0.0,
+  "alpha_jacobi": 0.0,
+  "decay": 0.8,
+  "target_vocab_util": 0.8,
+  "min_abs_ppl": 0.0,
+  "zipf_alpha": 1.0,
+  "lr": 4e-05,
+  "emb_lr_mult": 1.0,
+  "weight_decay": 0.01,
+  "warmup_steps": 468,
+  "cooldown_frac": 0.4,
+  "max_grad_norm": 1.0,
+  "vq_abs_pretrain_steps": 0,
+  "vq_abs_pretrain_lr": 0.001,
+  "vq_abs_pretrain_layer": -1,
+  "vq_abs_pretrain_batch_size": 256,
+  "vq_abs_pretrain_target_vectors": 20000,
+  "batch_size": 64,
+  "gradient_accumulation_steps": 1,
+  "num_epochs": 10,
+  "emb_warmup_steps": 0,
+  "log_every": 50,
+  "eval_every": 1562,
+  "save_every": 999999,
+  "eval_samples": 100,
+  "output_dir": "ckpt/sweep/as_sorl_abs2_K1_100K",
+  "eval_K": 4,
+  "alpha_traj": 0.0,
+  "corrupt_method": "shuffle",
+  "corrupt_ratio": 0.3,
+  "alpha_contrastive": 1.0,
+  "gamma_contrastive": 0.5,
+  "alpha_masked_traj": 0.0,
+  "mask_nl_ratio": 0.3,
+  "mask_nl_mode": "fixed",
+  "mask_nl_fixed_id": 0,
+  "use_ste": true,
+  "n_inner": 1,
+  "random_K": null,
+  "strip_suffix": null,
+  "compress_prefix": null,
+  "random_mem_span": null,
+  "warmup_ratio": 0.03,
+  "beta2": 0.999,
+  "seed": 42,
+  "n_digits": 6,
+  "n_layer": 2,
+  "n_head": 3,
+  "n_embd": 510,
+  "ops": "add_sub",
+  "abs_vocab": 2,
+  "dataset_size": 100000,
+  "mode": "sorl",
+  "device": "cuda",
+  "push_to_hub": true,
+  "no_wandb": false,
+  "n_params": 162491102,
+  "run_name": "add_sub_sorl_v1_abs2_K1_100K",
+  "git_commit": "f447da529caceac8c7d256cbb2cd185cbc50feac",
+  "timestamp": "2026-04-12T10:21:18.381375+00:00",
+  "tokenizer": "Qwen/Qwen3-0.6B",
+  "dataset_repo": "thoughtworks/arithmetic-sorl-data",
+  "dataset_config": "add_sub_6digit",
+  "model_repo": "thoughtworks/arithmetic-sorl",
+  "trainer_version": "v1",
+  "wandb_run_id": "i0b2rgm0",
+  "wandb_url": "https://wandb.ai/nlp_and_interpretability/sorl-arithmetic/runs/i0b2rgm0",
+  "final_accuracy": 1.0,
+  "sft_accuracy": 0.9995833333333334,
+  "eval_method": "ArithmeticEvaluator"
+}