Upload add_sub_sorl_v1_abs50_K1_100K

add_sub_sorl_v1_abs50_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": 151694
+}

add_sub_sorl_v1_abs50_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_abs50_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": [
+      0.8793072700500488,
+      -4.557682991027832,
+      3.945315361022949,
+      6.532111644744873,
+      4.72694730758667,
+      3.3974697589874268,
+      2.6203341484069824,
+      2.544694185256958,
+      2.474928379058838,
+      2.301546096801758,
+      2.134119749069214,
+      1.1167329549789429,
+      -1.3394086360931396,
+      -4.72641134262085,
+      -4.811535835266113,
+      -2.576375961303711,
+      -1.0075209140777588,
+      -1.3872089385986328,
+      -1.4432183504104614,
+      -1.4433867931365967,
+      -1.803222417831421,
+      -1.2808027267456055,
+      -1.0695509910583496,
+      -0.7846286296844482,
+      -0.7284835577011108,
+      -0.7496466636657715,
+      -0.8122822642326355,
+      -0.532228410243988,
+      -0.41442811489105225,
+      -0.8966416716575623,
+      -0.2750622630119324,
+      -0.7970641851425171,
+      -0.2488846778869629,
+      -0.22032293677330017,
+      -0.05407524108886719,
+      -0.30909571051597595,
+      -0.4708501994609833,
+      -0.043576836585998535,
+      -0.07583993673324585,
+      -0.6057784557342529,
+      -0.22415095567703247,
+      -0.24675825238227844,
+      0.02088850736618042,
+      -0.14688682556152344,
+      -0.004432946443557739,
+      -0.4482615888118744,
+      -0.0304650217294693,
+      -0.2053374946117401,
+      -0.2001218944787979,
+      -0.2562982439994812,
+      -1.079793930053711,
+      -0.3027035593986511,
+      -0.5672270059585571,
+      -0.1839711219072342,
+      -0.2499769628047943,
+      -0.12174144387245178,
+      -0.0955045223236084,
+      -0.0927371084690094,
+      -0.21636243164539337,
+      -0.3999066650867462,
+      -0.11761140823364258,
+      -0.1235857903957367,
+      -0.6707497239112854,
+      -0.10870647430419922,
+      -0.26432910561561584,
+      -0.5489974617958069,
+      -0.2955520451068878,
+      -0.13704964518547058,
+      -0.17987529933452606,
+      -0.2948598563671112,
+      -0.4160359799861908,
+      -0.47342371940612793,
+      -0.30369460582733154,
+      -0.5336835384368896,
+      -0.33963853120803833,
+      -0.16569268703460693,
+      -0.28100794553756714,
+      -0.09371660649776459,
+      -0.329658180475235,
+      -0.1649886667728424,
+      -0.08418028801679611,
+      -0.2127760946750641,
+      -0.2515207529067993,
+      -0.4989566504955292,
+      -0.1675868034362793,
+      -0.3154517710208893,
+      -0.29198193550109863,
+      -0.23230299353599548,
+      -0.30763623118400574,
+      -0.3012218475341797,
+      -0.166826531291008,
+      -0.17864090204238892,
+      -0.13514383137226105,
+      -0.06121335178613663,
+      -0.15541976690292358,
+      -0.29266083240509033,
+      -0.07417450100183487,
+      -0.010071318596601486,
+      -0.10823608934879303,
+      -0.036244817078113556,
+      -0.08790963888168335,
+      -0.0498308464884758,
+      -0.03580029308795929,
+      -0.07768949866294861,
+      0.017386823892593384,
+      -0.05269555002450943,
+      -0.04851272702217102,
+      -0.015783585608005524,
+      0.013150635175406933,
+      0.028122255578637123,
+      0.01595453731715679,
+      -0.08603314310312271,
+      -0.07443135976791382,
+      0.02399209514260292,
+      -0.006037743762135506,
+      -0.046386152505874634,
+      0.01122601330280304,
+      -0.05982036888599396,
+      -0.11936657130718231,
+      -0.037023354321718216,
+      -0.08698643743991852,
+      0.01087209302932024,
+      0.012216201052069664,
+      0.027771372348070145,
+      -0.016258858144283295,
+      0.006583958864212036,
+      -0.004937758669257164,
+      0.04394557699561119,
+      -0.10809867084026337,
+      0.010671194642782211,
+      -0.0460967980325222,
+      0.039323590695858,
+      -0.01787477731704712,
+      0.029121840372681618,
+      -0.026205148547887802,
+      0.022960329428315163,
+      0.021646002307534218,
+      -0.03326135128736496,
+      -0.11378178000450134,
+      0.006849285215139389,
+      0.01761329546570778,
+      0.022349704056978226,
+      0.03914421796798706,
+      -0.02611180767416954,
+      0.021699609234929085,
+      0.01957082934677601,
+      0.013363128527998924,
+      -0.021764134988188744,
+      -0.07776618003845215,
+      0.020661285147070885,
+      0.02226482518017292,
+      -0.05438518524169922,
+      -0.06914329528808594,
+      -0.002914411947131157,
+      -0.011443082243204117,
+      0.021878832951188087,
+      0.014752330258488655,
+      -0.02418067678809166,
+      0.006434757262468338,
+      0.033864736557006836,
+      -0.1172642782330513,
+      -0.008184622973203659,
+      0.033443063497543335,
+      -0.03762055188417435,
+      0.01753869652748108,
+      0.005519811064004898,
+      0.00024016760289669037,
+      0.01246030442416668,
+      0.03886703774333,
+      -0.1489391177892685,
+      0.006597569212317467,
+      0.0007694996893405914,
+      0.016658440232276917,
+      -0.16039103269577026,
+      0.027697507292032242,
+      0.0240585096180439,
+      0.03574578836560249,
+      0.014707371592521667,
+      -0.015416000038385391,
+      -0.20053425431251526,
+      0.021575435996055603,
+      0.030901774764060974,
+      0.023122712969779968,
+      0.03245032951235771,
+      -0.35969942808151245,
+      0.013166043907403946,
+      0.02934935875236988,
+      0.013501297682523727,
+      0.017913665622472763,
+      -0.004542285576462746,
+      0.015910688787698746,
+      -0.002435259521007538,
+      -0.011715129017829895,
+      0.03188084065914154,
+      0.011587947607040405,
+      0.026018982753157616,
+      0.025039900094270706,
+      0.020184341818094254,
+      0.024985218420624733,
+      0.02591882087290287,
+      -0.0122615285217762,
+      0.019379204139113426,
+      0.025564368814229965,
+      -0.003141876310110092,
+      0.0035779811441898346,
+      -0.009943334385752678,
+      -0.029678095132112503,
+      0.013146287761628628,
+      0.030820732936263084,
+      -0.1258346140384674,
+      -0.019851606339216232,
+      0.012696838937699795,
+      0.026656880974769592,
+      0.009932195767760277,
+      0.02451024390757084,
+      -0.06035705655813217,
+      -0.14478589594364166,
+      -0.00664106011390686,
+      0.003923642449080944,
+      0.024498794227838516,
+      -0.04950043559074402,
+      0.028375547379255295,
+      0.022171344608068466,
+      0.006763886660337448,
+      0.016951054334640503,
+      0.02775147557258606,
+      0.01829732023179531,
+      0.021201690658926964,
+      0.025561708956956863,
+      0.019114773720502853,
+      0.020513786002993584,
+      0.042111024260520935,
+      0.02553929202258587,
+      0.015924805775284767,
+      0.022776134312152863,
+      0.02991604618728161,
+      0.029099412262439728,
+      -0.04503129422664642,
+      0.024411184713244438,
+      0.01758555881679058,
+      -0.054419152438640594,
+      -0.0114205963909626,
+      0.008897339925169945,
+      0.03393123671412468,
+      0.02686172164976597,
+      -0.03536038100719452,
+      0.03133343905210495,
+      0.024179954081773758,
+      0.021813442930579185,
+      0.024799799546599388,
+      0.02012506313621998,
+      0.01691095344722271,
+      0.028604893013834953,
+      0.04053346440196037,
+      0.022266777232289314,
+      0.03397966921329498,
+      0.031599946320056915,
+      0.024446003139019012,
+      0.03555386886000633,
+      0.023791510611772537,
+      0.028894606977701187,
+      0.0001649763435125351,
+      0.032161928713321686,
+      0.024588581174612045,
+      0.017756029963493347,
+      0.027024097740650177,
+      0.025857599452137947,
+      -0.03726355731487274,
+      0.03172441944479942,
+      0.03862748667597771,
+      0.024849383160471916,
+      0.021331746131181717,
+      0.025282954797148705,
+      0.021858539432287216,
+      0.022209014743566513,
+      0.02866678312420845,
+      0.039874590933322906,
+      -0.00557183101773262,
+      0.025868263095617294,
+      0.024863965809345245,
+      0.021220605820417404,
+      0.03036733716726303,
+      0.019596464931964874,
+      0.017102953046560287,
+      0.025336796417832375,
+      0.023993855342268944,
+      0.022393351420760155,
+      0.022890813648700714,
+      0.02799217961728573,
+      0.025767609477043152,
+      0.020770160481333733,
+      0.025682464241981506,
+      0.019536113366484642,
+      0.019240833818912506,
+      0.021054189652204514,
+      0.041958287358284,
+      0.025849830359220505,
+      0.024066584184765816,
+      0.023014742881059647,
+      0.023878082633018494,
+      0.019091343507170677,
+      0.04283755272626877,
+      0.024016324430704117,
+      0.027863766998052597,
+      0.026920557022094727,
+      0.024275902658700943,
+      0.028992624953389168,
+      0.03695023059844971,
+      0.03257744386792183,
+      0.020896634086966515
+    ],
+    "base_loss": [
+      11.354902267456055,
+      9.619476318359375,
+      8.130751609802246,
+      6.807829856872559,
+      5.4560370445251465,
+      3.5696353912353516,
+      2.1277003288269043,
+      1.9940003156661987,
+      1.8792340755462646,
+      1.778296709060669,
+      1.820477843284607,
+      1.7617279291152954,
+      1.8593480587005615,
+      1.745758056640625,
+      1.5065184831619263,
+      1.0960109233856201,
+      0.8341459631919861,
+      0.8041783571243286,
+      0.6385746002197266,
+      0.6072742342948914,
+      0.5624108910560608,
+      0.4968210756778717,
+      0.44730648398399353,
+      0.3511924147605896,
+      0.3345043659210205,
+      0.3197433352470398,
+      0.3210335075855255,
+      0.3214714229106903,
+      0.2237830013036728,
+      0.26585131883621216,
+      0.1817438155412674,
+      0.22315724194049835,
+      0.14824306964874268,
+      0.17239229381084442,
+      0.14353437721729279,
+      0.16023890674114227,
+      0.16021332144737244,
+      0.12249647080898285,
+      0.09916845709085464,
+      0.1469610035419464,
+      0.1400381326675415,
+      0.10619890689849854,
+      0.07646704465150833,
+      0.10491969436407089,
+      0.09419133514165878,
+      0.1305452138185501,
+      0.0678355023264885,
+      0.09236187487840652,
+      0.12952150404453278,
+      0.08002316951751709,
+      0.18990421295166016,
+      0.09865695983171463,
+      0.11963018029928207,
+      0.08220357447862625,
+      0.1007433608174324,
+      0.0954989343881607,
+      0.07067380100488663,
+      0.06859846413135529,
+      0.05933056026697159,
+      0.07883036136627197,
+      0.05551758036017418,
+      0.048164304345846176,
+      0.11315275728702545,
+      0.04112541303038597,
+      0.059350378811359406,
+      0.09123921394348145,
+      0.06157296895980835,
+      0.04738176241517067,
+      0.041634202003479004,
+      0.055201027542352676,
+      0.08019154518842697,
+      0.07904402166604996,
+      0.05514637008309364,
+      0.10348294675350189,
+      0.05645943433046341,
+      0.03977251052856445,
+      0.0554334782063961,
+      0.036599669605493546,
+      0.056209079921245575,
+      0.03545624762773514,
+      0.02682807669043541,
+      0.03998107463121414,
+      0.05404913052916527,
+      0.07497896254062653,
+      0.04433978348970413,
+      0.04918047413229942,
+      0.04802717640995979,
+      0.03953424468636513,
+      0.045616548508405685,
+      0.044817231595516205,
+      0.03032597526907921,
+      0.031423501670360565,
+      0.023572741076350212,
+      0.017564991489052773,
+      0.02600538171827793,
+      0.04100664332509041,
+      0.01607869379222393,
+      0.008917765691876411,
+      0.019925076514482498,
+      0.01216939091682434,
+      0.016869796440005302,
+      0.012725042179226875,
+      0.0102456733584404,
+      0.01590334065258503,
+      0.0047952658496797085,
+      0.011497433297336102,
+      0.012098747305572033,
+      0.007142555434256792,
+      0.0038218761328607798,
+      0.002109906869009137,
+      0.0034211091697216034,
+      0.015994401648640633,
+      0.012902417220175266,
+      0.0024884960148483515,
+      0.004833155311644077,
+      0.010148318484425545,
+      0.004628085996955633,
+      0.011428607627749443,
+      0.019086044281721115,
+      0.009554388001561165,
+      0.014633984304964542,
+      0.002517522545531392,
+      0.004276275169104338,
+      0.0017393995076417923,
+      0.005628875456750393,
+      0.0039735762402415276,
+      0.004733487498015165,
+      0.0016199646051973104,
+      0.016592636704444885,
+      0.0035542051773518324,
+      0.010028673335909843,
+      0.00040924036875367165,
+      0.007201785687357187,
+      0.0006683436222374439,
+      0.00829078909009695,
+      0.0017987389583140612,
+      0.0010684876469895244,
+      0.008472238667309284,
+      0.016244305297732353,
+      0.002869021613150835,
+      0.0031283958815038204,
+      0.0022019422613084316,
+      0.0004035625606775284,
+      0.006791552063077688,
+      0.0011557966936379671,
+      0.0008532117353752255,
+      0.0032540052197873592,
+      0.005603688769042492,
+      0.012657723389565945,
+      0.0005092295468784869,
+      0.0010295194806531072,
+      0.009815016761422157,
+      0.010954980738461018,
+      0.0029354526195675135,
+      0.005530492402613163,
+      0.000797468121163547,
+      0.0018827719613909721,
+      0.0076822019182145596,
+      0.0022407553624361753,
+      0.0009795849910005927,
+      0.017207054421305656,
+      0.004975206684321165,
+      0.000368430744856596,
+      0.007918541319668293,
+      0.0013579931110143661,
+      0.002656040247529745,
+      0.0029781318735331297,
+      0.0032940662931650877,
+      0.0007902187062427402,
+      0.020759092643857002,
+      0.025971150025725365,
+      0.003461887827143073,
+      0.0016995306359604,
+      0.021632632240653038,
+      0.0003880455915350467,
+      0.001397465355694294,
+      0.00030403825803659856,
+      0.0020705878268927336,
+      0.004525614436715841,
+      0.02685791626572609,
+      0.0006216937326826155,
+      0.0010118251666426659,
+      0.0008113172953017056,
+      0.0002240273024654016,
+      0.04504799842834473,
+      0.0015609429683536291,
+      0.0005780804203823209,
+      0.0022545286919921637,
+      0.0007725828909315169,
+      0.003736443119123578,
+      0.002966423286125064,
+      0.004257167223840952,
+      0.0048461672849953175,
+      0.0006816848763264716,
+      0.003050495171919465,
+      0.0009072899702005088,
+      0.00027717239572666585,
+      0.0006319586536847055,
+      0.0011747808894142509,
+      0.0001518602657597512,
+      0.004137661308050156,
+      0.00019556829647626728,
+      0.0007796343415975571,
+      0.003540764795616269,
+      0.0031884273048490286,
+      0.0036890930496156216,
+      0.006483628414571285,
+      0.0018271852750331163,
+      0.00027773826150223613,
+      0.016763756051659584,
+      0.006423712242394686,
+      0.001957931090146303,
+      0.00025845892378129065,
+      0.0023654543329030275,
+      0.00032435316825285554,
+      0.00942069198936224,
+      0.019012629985809326,
+      0.004030904732644558,
+      0.002140968106687069,
+      0.0004519534413702786,
+      0.008156306110322475,
+      0.0005255067371763289,
+      6.81299134157598e-05,
+      0.002138505456969142,
+      0.0014500885736197233,
+      0.00020060435053892434,
+      0.00035458055208437145,
+      0.0007856138399802148,
+      0.00014310672122519463,
+      0.0012812147615477443,
+      0.00022322598670143634,
+      0.000436297181295231,
+      0.00011576743418117985,
+      0.0012859638081863523,
+      0.0005143244634382427,
+      6.025502079864964e-05,
+      0.00019221320690121502,
+      0.007957907393574715,
+      0.00014822494995314628,
+      0.0014601113507524133,
+      0.008723047561943531,
+      0.0043920488096773624,
+      0.0014019532827660441,
+      3.626879333751276e-05,
+      3.528634624672122e-05,
+      0.007602563593536615,
+      0.00010167565051233396,
+      3.648105848697014e-05,
+      0.0003249097499065101,
+      4.7194680519169196e-05,
+      0.0007050277781672776,
+      0.0006095104617998004,
+      0.00022558416821993887,
+      1.7922138795256615e-05,
+      0.00012923660688102245,
+      0.00011796010221587494,
+      0.00011410734441597015,
+      5.379039066610858e-05,
+      0.00014431810996029526,
+      4.39703326264862e-05,
+      0.0001760836021276191,
+      0.0032698672730475664,
+      1.0670499250409193e-05,
+      0.00011346897372277454,
+      0.002302784938365221,
+      0.00026522204279899597,
+      0.0012583305360749364,
+      0.007555671036243439,
+      2.5744331651367247e-05,
+      8.980598795460537e-05,
+      6.1836137319915e-05,
+      1.9674447685247287e-05,
+      3.210911017959006e-05,
+      0.00017015096091199666,
+      1.8454913515597582e-05,
+      6.926675996510312e-05,
+      1.964170951396227e-05,
+      0.0030478867702186108,
+      1.7391646906617098e-05,
+      1.1291275768599007e-05,
+      1.6856354704941623e-05,
+      2.419223528704606e-05,
+      6.455252878367901e-05,
+      5.313939254847355e-05,
+      1.0072631994262338e-05,
+      2.2529971829499118e-05,
+      1.3181961548980325e-05,
+      3.893485336448066e-05,
+      6.0693128034472466e-05,
+      1.1840793376904912e-05,
+      2.4969294827315025e-05,
+      4.796602297574282e-05,
+      1.5566796719213016e-05,
+      2.5332241420983337e-05,
+      1.1200068911421113e-05,
+      9.845345630310476e-06,
+      8.674382115714252e-06,
+      1.5246259863488376e-05,
+      2.1107602151460014e-05,
+      1.836235423979815e-05,
+      1.927776975207962e-05,
+      8.702610102773178e-06,
+      2.1925961846136488e-05,
+      9.651708751334809e-06,
+      2.9662509405170567e-05,
+      1.3283395674079657e-05,
+      9.9586777650984e-06,
+      8.589026947447564e-06,
+      1.179453556687804e-05,
+      7.600993285450386e-06
+    ],
+    "info_loss": [
+      -1.7131338119506836,
+      -1.9878039360046387,
+      -0.8607902526855469,
+      -0.343416690826416,
+      -0.27082347869873047,
+      -0.13865256309509277,
+      -0.052397727966308594,
+      -0.0423356294631958,
+      -0.03699600696563721,
+      -0.043445587158203125,
+      -0.06507444381713867,
+      -0.15992426872253418,
+      -0.4150509834289551,
+      -0.7428005933761597,
+      -0.7276861667633057,
+      -0.4624685049057007,
+      -0.2779492139816284,
+      -0.3108817934989929,
+      -0.30032607913017273,
+      -0.29420244693756104,
+      -0.3239292502403259,
+      -0.26364535093307495,
+      -0.23538976907730103,
+      -0.19680123031139374,
+      -0.18817496299743652,
+      -0.18621370196342468,
+      -0.18980754911899567,
+      -0.15879178047180176,
+      -0.13470777869224548,
+      -0.1839972287416458,
+      -0.11035551130771637,
+      -0.16300663352012634,
+      -0.09734988212585449,
+      -0.09699394553899765,
+      -0.07697362452745438,
+      -0.10090111941099167,
+      -0.11517143249511719,
+      -0.06642338633537292,
+      -0.06478674709796906,
+      -0.12275853008031845,
+      -0.08251140266656876,
+      -0.07660090923309326,
+      -0.048429496586322784,
+      -0.06653877347707748,
+      -0.053453486412763596,
+      -0.09985008835792542,
+      -0.04629219323396683,
+      -0.0712423324584961,
+      -0.06884448230266571,
+      -0.06782782077789307,
+      -0.16537919640541077,
+      -0.07562237977981567,
+      -0.10252176225185394,
+      -0.05866117402911186,
+      -0.06620694696903229,
+      -0.05184585601091385,
+      -0.04774373024702072,
+      -0.04839886352419853,
+      -0.05516344681382179,
+      -0.07325172424316406,
+      -0.042329758405685425,
+      -0.040246520191431046,
+      -0.10164471715688705,
+      -0.0373082235455513,
+      -0.05317864939570427,
+      -0.08476268500089645,
+      -0.05598641559481621,
+      -0.037800855934619904,
+      -0.04003060981631279,
+      -0.05340470001101494,
+      -0.06776013970375061,
+      -0.07399261742830276,
+      -0.05010056868195534,
+      -0.08082228899002075,
+      -0.054453715682029724,
+      -0.035927727818489075,
+      -0.045634906738996506,
+      -0.028730852529406548,
+      -0.053217265754938126,
+      -0.03469575196504593,
+      -0.024422042071819305,
+      -0.03822420537471771,
+      -0.043456658720970154,
+      -0.0705932006239891,
+      -0.03254368156194687,
+      -0.04738222807645798,
+      -0.047146376222372055,
+      -0.038757313042879105,
+      -0.04511863738298416,
+      -0.044538818299770355,
+      -0.030123282223939896,
+      -0.030372440814971924,
+      -0.023335659876465797,
+      -0.017431115731596947,
+      -0.02585216611623764,
+      -0.04085705429315567,
+      -0.015927845612168312,
+      -0.00878168921917677,
+      -0.01885957084596157,
+      -0.011983344331383705,
+      -0.01676131598651409,
+      -0.012616250663995743,
+      -0.010080559179186821,
+      -0.015823306515812874,
+      -0.00466143386438489,
+      -0.011390343308448792,
+      -0.012019429355859756,
+      -0.006937987171113491,
+      -0.003349082078784704,
+      -0.0020443047396838665,
+      -0.0033661662600934505,
+      -0.01590695045888424,
+      -0.012850151397287846,
+      -0.0023344852961599827,
+      -0.004767741076648235,
+      -0.010024209506809711,
+      -0.004549992736428976,
+      -0.011376718990504742,
+      -0.019021080806851387,
+      -0.009498468600213528,
+      -0.014589231461286545,
+      -0.002486016834154725,
+      -0.004247958771884441,
+      -0.001699471496976912,
+      -0.005558435805141926,
+      -0.00393689377233386,
+      -0.004702319391071796,
+      -0.0003044349141418934,
+      -0.01655041053891182,
+      -0.0035031919833272696,
+      -0.00998228881508112,
+      -0.0003558158641681075,
+      -0.00714930659160018,
+      -0.0006295697530731559,
+      -0.008250396698713303,
+      -0.0017516581574454904,
+      -0.001038093352690339,
+      -0.008434775285422802,
+      -0.016211409121751785,
+      -0.0028474254067987204,
+      -0.003096792846918106,
+      -0.0021675738971680403,
+      -0.0003787819587159902,
+      -0.006768254097551107,
+      -0.00110574287828058,
+      -0.000609182461630553,
+      -0.0032347021624445915,
+      -0.005555766634643078,
+      -0.012623850256204605,
+      -0.0004824258212465793,
+      -0.0010148858418688178,
+      -0.009800522588193417,
+      -0.01089291088283062,
+      -0.002919699065387249,
+      -0.005509648472070694,
+      -0.0007093790918588638,
+      -0.0017709487583488226,
+      -0.007096725050359964,
+      -0.0022284225560724735,
+      -0.0009606123203411698,
+      -0.017192592844367027,
+      -0.004958974663168192,
+      -0.000321025843732059,
+      -0.007865279912948608,
+      -0.0013248363975435495,
+      -0.0026323902420699596,
+      -0.002967209555208683,
+      -0.003275538794696331,
+      -0.0007697101100347936,
+      -0.02069810777902603,
+      -0.00600612536072731,
+      -0.003454132005572319,
+      -0.00164946005679667,
+      -0.021616628393530846,
+      -0.00037547998363152146,
+      -0.0013791971141472459,
+      -0.0002931484195869416,
+      -0.0020377198234200478,
+      -0.004515333101153374,
+      -0.026843883097171783,
+      -0.0006049006478860974,
+      -0.0006070983945392072,
+      -0.0007667711470276117,
+      -0.00021536584245041013,
+      -0.0450412854552269,
+      -0.0015369504690170288,
+      -0.0005645402707159519,
+      -0.002216763561591506,
+      -0.0007594142225570977,
+      -0.0037237321957945824,
+      -0.002905755303800106,
+      -0.00424158712849021,
+      -0.004837940912693739,
+      -0.0006752160843461752,
+      -0.0030186076182872057,
+      -0.0009017926640808582,
+      -0.00026729219825938344,
+      -0.0006156229064799845,
+      -0.0011618678690865636,
+      -0.0001462623622501269,
+      -0.004122164100408554,
+      -0.0001859295298345387,
+      -0.0007514554308727384,
+      -0.003534250194206834,
+      -0.0030365008860826492,
+      -0.0036810182500630617,
+      -0.0064766560681164265,
+      -0.0018187804380431771,
+      -0.000253577483817935,
+      -0.01675243116915226,
+      -0.006413512397557497,
+      -0.0019517249893397093,
+      -0.00025289144832640886,
+      -0.0023404271341860294,
+      -0.00031484346254728734,
+      -0.00941128097474575,
+      -0.019002120941877365,
+      -0.004024147056043148,
+      -0.0021259698551148176,
+      -0.0004454626177903265,
+      -0.008115903474390507,
+      -0.0005199434235692024,
+      -6.396112439688295e-05,
+      -0.0021351408213377,
+      -0.001437849598005414,
+      -0.0001965994742931798,
+      -0.00035130474134348333,
+      -0.0007774023106321692,
+      -0.0001280636206502095,
+      -0.0012665236135944724,
+      -0.00021530069352593273,
+      -0.000425885955337435,
+      -0.00010600911627989262,
+      -0.0012709213187918067,
+      -0.0005109088378958404,
+      -5.4605796321993694e-05,
+      -0.0001892093278001994,
+      -0.007937698625028133,
+      -0.00014467535947915167,
+      -0.0014578413683921099,
+      -0.008719262667000294,
+      -0.0043871584348380566,
+      -0.0013978895731270313,
+      -3.224361717002466e-05,
+      -3.281969475210644e-05,
+      -0.007597143296152353,
+      -9.899612632580101e-05,
+      -2.738368493737653e-05,
+      -0.00031613255850970745,
+      -4.128325963392854e-05,
+      -0.0007011730340309441,
+      -0.0006041606538929045,
+      -0.00022185097623150796,
+      -1.5747378711239435e-05,
+      -0.0001257083349628374,
+      -0.00011497300874907523,
+      -0.00011105320299975574,
+      -5.157307896297425e-05,
+      -0.0001417464081896469,
+      -4.0079659811453894e-05,
+      -0.0001711562799755484,
+      -0.003267128486186266,
+      -8.380002327612601e-06,
+      -0.00011130699567729607,
+      -0.0022985294926911592,
+      -0.0002603848115541041,
+      -0.0012561411131173372,
+      -0.007548708934336901,
+      -2.3311802578973584e-05,
+      -8.536979294149205e-05,
+      -5.928838800173253e-05,
+      -1.7859445506474003e-05,
+      -2.8874992494820617e-05,
+      -0.00016634466010145843,
+      -1.664549927227199e-05,
+      -6.716864299960434e-05,
+      -1.7809690689318813e-05,
+      -0.0030463875737041235,
+      -1.591697036928963e-05,
+      -9.979978131013922e-06,
+      -1.458264796383446e-05,
+      -2.2660085960524157e-05,
+      -6.295119237620384e-05,
+      -5.158302519703284e-05,
+      -8.870430065144319e-06,
+      2.4651939384057187e-05,
+      -1.0839266906259581e-05,
+      -3.726567592821084e-05,
+      -5.868282460141927e-05,
+      -1.065588730853051e-05,
+      -2.3740489268675447e-05,
+      -4.6571971324738115e-05,
+      -1.4279713468567934e-05,
+      -2.412179128441494e-05,
+      -1.0100046893057879e-05,
+      -8.060726031544618e-06,
+      -7.528061360062566e-06,
+      -1.038836853695102e-05,
+      -1.9785129552474245e-05,
+      -1.6565183614147827e-05,
+      -1.7817466869018972e-05,
+      -7.467680006811861e-06,
+      -2.0526860680547543e-05,
+      -8.646415153634734e-06,
+      -2.842918001988437e-05,
+      -1.2150379916420206e-05,
+      -8.60803629620932e-06,
+      -7.291034307854716e-06,
+      -1.0754650247690734e-05,
+      -6.411308731912868e-06
+    ],
+    "abs_loss": [
+      3.9100465774536133,
+      3.8975870609283447,
+      3.8392674922943115,
+      3.7595930099487305,
+      3.581761598587036,
+      3.077859878540039,
+      3.094771146774292,
+      3.077636957168579,
+      3.1082115173339844,
+      3.0092105865478516,
+      3.141963243484497,
+      3.0576484203338623,
+      3.0487613677978516,
+      3.056673049926758,
+      3.06050705909729,
+      2.977795362472534,
+      2.929574966430664,
+      2.7671806812286377,
+      2.842601776123047,
+      2.7369773387908936,
+      2.681629180908203,
+      2.5334365367889404,
+      2.4646987915039062,
+      2.3455915451049805,
+      2.323493242263794,
+      2.225255012512207,
+      2.0635712146759033,
+      2.201101064682007,
+      1.9572720527648926,
+      1.9138232469558716,
+      1.9275528192520142,
+      1.7616440057754517,
+      1.7060261964797974,
+      1.7646626234054565,
+      1.6927824020385742,
+      1.6781312227249146,
+      1.614211082458496,
+      1.6111903190612793,
+      1.5514603853225708,
+      1.5403081178665161,
+      1.408429503440857,
+      1.4391018152236938,
+      1.4640899896621704,
+      1.3647264242172241,
+      1.4949804544448853,
+      1.2549307346343994,
+      1.272302508354187,
+      1.2950795888900757,
+      1.2709366083145142,
+      1.203320026397705,
+      1.2418088912963867,
+      1.1795908212661743,
+      1.2100251913070679,
+      1.0959945917129517,
+      1.0862230062484741,
+      1.0877859592437744,
+      1.141281247138977,
+      1.0249780416488647,
+      0.9890892505645752,
+      0.9109721183776855,
+      1.0081976652145386,
+      0.928091824054718,
+      0.9609468579292297,
+      0.9811637997627258,
+      0.7913894653320312,
+      0.8156485557556152,
+      0.8584363460540771,
+      0.6847303509712219,
+      0.687626838684082,
+      0.6748706698417664,
+      0.7412440180778503,
+      0.7287342548370361,
+      0.6193475127220154,
+      0.720916211605072,
+      0.6421559453010559,
+      0.642063558101654,
+      0.5613493919372559,
+      0.5891659259796143,
+      0.5479210615158081,
+      0.5314231514930725,
+      0.5439849495887756,
+      0.5367106795310974,
+      0.507027804851532,
+      0.45817089080810547,
+      0.5355744957923889,
+      0.45088496804237366,
+      0.46670737862586975,
+      0.4583347737789154,
+      0.3799661695957184,
+      0.3906151056289673,
+      0.42524102330207825,
+      0.30378833413124084,
+      0.34402838349342346,
+      0.23706363141536713,
+      0.2770039737224579,
+      0.23684363067150116,
+      0.27812063694000244,
+      0.23869305849075317,
+      0.25324514508247375,
+      0.27824679017066956,
+      0.22837097942829132,
+      0.24062645435333252,
+      0.1953345686197281,
+      0.2186935693025589,
+      0.19244857132434845,
+      0.20102375745773315,
+      0.18487371504306793,
+      0.16867367923259735,
+      0.15062938630580902,
+      0.18083924055099487,
+      0.11583039164543152,
+      0.19421494007110596,
+      0.1377987265586853,
+      0.12761341035366058,
+      0.1398167759180069,
+      0.13316531479358673,
+      0.13546839356422424,
+      0.10419720411300659,
+      0.10089924186468124,
+      0.10359545797109604,
+      0.08913455158472061,
+      0.0836791917681694,
+      0.09513048082590103,
+      0.10752777010202408,
+      0.07869306206703186,
+      0.09269803017377853,
+      0.0664750263094902,
+      0.10058172792196274,
+      0.0942007526755333,
+      0.07060393691062927,
+      0.07069623470306396,
+      0.08343581110239029,
+      0.08978813886642456,
+      0.06264176219701767,
+      0.0727839320898056,
+      0.08141377568244934,
+      0.06984963268041611,
+      0.08312416076660156,
+      0.06752229481935501,
+      0.06518099457025528,
+      0.0723346397280693,
+      0.0926949754357338,
+      0.07708333432674408,
+      0.06434222310781479,
+      0.0732760801911354,
+      0.05879487469792366,
+      0.07625218480825424,
+      0.07054731249809265,
+      0.05598445609211922,
+      0.04704497382044792,
+      0.05821196362376213,
+      0.05910499021410942,
+      0.050326455384492874,
+      0.04014730826020241,
+      0.05032016709446907,
+      0.0559832789003849,
+      0.05798080563545227,
+      0.060850322246551514,
+      0.05990013852715492,
+      0.0649031326174736,
+      0.05214034020900726,
+      0.03999574854969978,
+      0.06125303730368614,
+      0.04810075834393501,
+      0.057810503989458084,
+      0.04367963597178459,
+      0.04500923678278923,
+      0.04815249145030975,
+      0.06941595673561096,
+      0.04378579184412956,
+      0.039449017494916916,
+      0.062223684042692184,
+      0.05087001249194145,
+      0.05693439766764641,
+      0.05195432901382446,
+      0.044212400913238525,
+      0.03281838819384575,
+      0.07130736857652664,
+      0.04234795644879341,
+      0.037031132727861404,
+      0.056286055594682693,
+      0.05698028579354286,
+      0.04710875824093819,
+      0.0431063137948513,
+      0.05567215010523796,
+      0.04408109560608864,
+      0.03917383402585983,
+      0.04425838589668274,
+      0.04848232865333557,
+      0.041234616190195084,
+      0.055955249816179276,
+      0.03867678716778755,
+      0.05803993344306946,
+      0.058457791805267334,
+      0.0351126492023468,
+      0.047668252140283585,
+      0.051030561327934265,
+      0.06693687289953232,
+      0.037342045456171036,
+      0.04897139593958855,
+      0.04071170464158058,
+      0.03195817396044731,
+      0.04455135390162468,
+      0.037423621863126755,
+      0.03185616061091423,
+      0.04340950772166252,
+      0.04206155613064766,
+      0.03775278851389885,
+      0.046762850135564804,
+      0.02975139021873474,
+      0.03377072140574455,
+      0.048947807401418686,
+      0.02991846390068531,
+      0.043979253619909286,
+      0.03999378904700279,
+      0.029062345623970032,
+      0.03937942162156105,
+      0.043757364153862,
+      0.035556841641664505,
+      0.03750767558813095,
+      0.037997055798769,
+      0.03813956305384636,
+      0.03332119807600975,
+      0.04382672905921936,
+      0.03575631603598595,
+      0.03302134945988655,
+      0.030200744047760963,
+      0.03823885694146156,
+      0.03453255072236061,
+      0.03653145954012871,
+      0.04374239966273308,
+      0.047345180064439774,
+      0.040144938975572586,
+      0.04836520925164223,
+      0.03711860999464989,
+      0.03266342356801033,
+      0.03952832892537117,
+      0.04084258899092674,
+      0.03149488940834999,
+      0.05244305729866028,
+      0.042809247970581055,
+      0.027759669348597527,
+      0.037863582372665405,
+      0.04004785045981407,
+      0.03883574903011322,
+      0.03282098099589348,
+      0.04005662724375725,
+      0.0291286613792181,
+      0.0357460118830204,
+      0.03452034667134285,
+      0.04267224296927452,
+      0.03179382160305977,
+      0.039745133370161057,
+      0.04173777624964714,
+      0.0354158878326416,
+      0.03331361338496208,
+      0.04022645950317383,
+      0.03608141466975212,
+      0.03537673130631447,
+      0.03464791551232338,
+      0.032951224595308304,
+      0.02508355677127838,
+      0.04105520620942116,
+      0.03586887568235397,
+      0.04789206385612488,
+      0.04689581319689751,
+      0.039730969816446304,
+      0.03624783456325531,
+      0.03239130601286888,
+      0.02859114110469818,
+      0.0327032208442688,
+      0.03129056468605995,
+      0.0358058363199234,
+      0.029234059154987335,
+      0.030198248103260994,
+      0.028243616223335266,
+      0.0331985168159008,
+      0.04053870961070061,
+      0.02295183204114437,
+      0.02807433344423771,
+      0.027615806087851524,
+      0.02944447100162506,
+      0.03079299069941044,
+      0.027692163363099098,
+      0.03086685575544834,
+      0.025942102074623108,
+      0.03200956806540489,
+      0.03382093831896782,
+      0.03204977512359619,
+      0.024898899719119072,
+      0.024450920522212982,
+      0.031034136191010475,
+      0.029439685866236687,
+      0.024981820955872536,
+      0.0282369926571846,
+      0.033129725605249405,
+      0.03465596213936806,
+      0.026924612000584602,
+      0.027515172958374023,
+      0.02585337497293949,
+      0.024600399658083916,
+      0.03324790671467781,
+      0.031333841383457184,
+      0.04458741471171379,
+      0.03677447512745857,
+      0.030864620581269264,
+      0.034290824085474014,
+      0.026761651039123535,
+      0.022752979770302773,
+      0.0276681836694479
+    ],
+    "zipf_loss": [
+      6.2647385597229,
+      5.311121940612793,
+      4.038539409637451,
+      2.78248929977417,
+      1.6209688186645508,
+      0.9065740704536438,
+      0.7071341276168823,
+      0.6662864685058594,
+      0.6548333168029785,
+      0.6567842960357666,
+      0.6501899361610413,
+      0.6484828591346741,
+      0.646877110004425,
+      0.6501692533493042,
+      0.652756929397583,
+      0.6545189619064331,
+      0.6448677182197571,
+      0.6407127380371094,
+      0.6372076272964478,
+      0.6176654696464539,
+      0.605496346950531,
+      0.6054860353469849,
+      0.5905702710151672,
+      0.5976320505142212,
+      0.5864124298095703,
+      0.5702215433120728,
+      0.5584025979042053,
+      0.514107882976532,
+      0.5131394863128662,
+      0.4860970377922058,
+      0.4539937376976013,
+      0.43368053436279297,
+      0.4057684540748596,
+      0.40075793862342834,
+      0.40284836292266846,
+      0.37186339497566223,
+      0.3592296540737152,
+      0.3370414972305298,
+      0.3177129626274109,
+      0.3208150863647461,
+      0.3200819492340088,
+      0.2691417634487152,
+      0.2823074162006378,
+      0.2771086096763611,
+      0.2864125370979309,
+      0.2942010462284088,
+      0.2373911291360855,
+      0.28521600365638733,
+      0.23170779645442963,
+      0.22162479162216187,
+      0.25991278886795044,
+      0.23690418899059296,
+      0.21735797822475433,
+      0.21083758771419525,
+      0.202726811170578,
+      0.19243958592414856,
+      0.19713085889816284,
+      0.220155268907547,
+      0.17703257501125336,
+      0.16268298029899597,
+      0.14934882521629333,
+      0.13790592551231384,
+      0.13645006716251373,
+      0.1251339465379715,
+      0.12896808981895447,
+      0.12582536041736603,
+      0.11689553409814835,
+      0.12510409951210022,
+      0.11003391444683075,
+      0.11649907380342484,
+      0.10724946111440659,
+      0.11458496749401093,
+      0.0802299976348877,
+      0.09896475821733475,
+      0.08422364294528961,
+      0.08960572630167007,
+      0.06377270072698593,
+      0.09807564318180084,
+      0.09151331335306168,
+      0.09337028115987778,
+      0.07881356030702591,
+      0.07581382989883423,
+      0.07829391211271286,
+      0.08617934584617615,
+      0.059952788054943085,
+      0.06410154700279236,
+      0.0847838968038559,
+      0.06990242749452591,
+      0.059936970472335815,
+      0.06028757989406586,
+      0.06155620887875557,
+      0.06328116357326508,
+      0.040237169712781906,
+      0.07182645052671432,
+      0.049396127462387085,
+      0.05121868848800659,
+      0.04121319204568863,
+      0.044958505779504776,
+      0.03511004522442818,
+      0.04359455406665802,
+      0.03999662399291992,
+      0.039543963968753815,
+      0.03522617369890213,
+      0.042770866304636,
+      0.039961040019989014,
+      0.02960807830095291,
+      0.04109545052051544,
+      0.029586367309093475,
+      0.027756642550230026,
+      0.028371471911668777,
+      0.03461205214262009,
+      0.03762046992778778,
+      0.02738787606358528,
+      0.03208710998296738,
+      0.022824836894869804,
+      0.030391091480851173,
+      0.03855101764202118,
+      0.032098494470119476,
+      0.04166826605796814,
+      0.03804739937186241,
+      0.03535844758152962,
+      0.024846818298101425,
+      0.04090646654367447,
+      0.03227391093969345,
+      0.0258273147046566,
+      0.0327095165848732,
+      0.030704444274306297,
+      0.03531178832054138,
+      0.03139270469546318,
+      0.03508851304650307,
+      0.036627788096666336,
+      0.03412892669439316,
+      0.037437692284584045,
+      0.028485018759965897,
+      0.04072963818907738,
+      0.030536793172359467,
+      0.023973483592271805,
+      0.034301742911338806,
+      0.025335777550935745,
+      0.025936419144272804,
+      0.03821936249732971,
+      0.03255400434136391,
+      0.03482013940811157,
+      0.028344962745904922,
+      0.02427363395690918,
+      0.01892995461821556,
+      0.03483092784881592,
+      0.021135112270712852,
+      0.030216146260499954,
+      0.020271817222237587,
+      0.025562968105077744,
+      0.027894524857401848,
+      0.02379819191992283,
+      0.019332395866513252,
+      0.03309089317917824,
+      0.02257682755589485,
+      0.024780966341495514,
+      0.03301933780312538,
+      0.020488213747739792,
+      0.03600095957517624,
+      0.03224056214094162,
+      0.032430339604616165,
+      0.030159588903188705,
+      0.028303630650043488,
+      0.023648016154766083,
+      0.024819709360599518,
+      0.0224332083016634,
+      0.03710637986660004,
+      0.038832325488328934,
+      0.03290429338812828,
+      0.03674276918172836,
+      0.025626562535762787,
+      0.02636650763452053,
+      0.02844918519258499,
+      0.02586882933974266,
+      0.03203177452087402,
+      0.03509139642119408,
+      0.025883246213197708,
+      0.020976923406124115,
+      0.03734353929758072,
+      0.021374141797423363,
+      0.0302629042416811,
+      0.025268230587244034,
+      0.03006932884454727,
+      0.04009821638464928,
+      0.022566495463252068,
+      0.030499298125505447,
+      0.028988566249608994,
+      0.019886991009116173,
+      0.02483513206243515,
+      0.036406293511390686,
+      0.03185576573014259,
+      0.026014119386672974,
+      0.03210553899407387,
+      0.03521226346492767,
+      0.029362794011831284,
+      0.02233259379863739,
+      0.01901492476463318,
+      0.031694911420345306,
+      0.02233244478702545,
+      0.02075127884745598,
+      0.01784711331129074,
+      0.027844153344631195,
+      0.024917500093579292,
+      0.027568945661187172,
+      0.01883680559694767,
+      0.02439868450164795,
+      0.025731626898050308,
+      0.028402484953403473,
+      0.02195078879594803,
+      0.03448273614048958,
+      0.025361377745866776,
+      0.025935489684343338,
+      0.02657308615744114,
+      0.023334946483373642,
+      0.021428830921649933,
+      0.022284740582108498,
+      0.025193769484758377,
+      0.0194866880774498,
+      0.02475069835782051,
+      0.01970258168876171,
+      0.029235519468784332,
+      0.01941070705652237,
+      0.02159411646425724,
+      0.02630382962524891,
+      0.026214729994535446,
+      0.018435712903738022,
+      0.024366214871406555,
+      0.023245982825756073,
+      0.026845648884773254,
+      0.018069326877593994,
+      0.04119906947016716,
+      0.02246912196278572,
+      0.02251153439283371,
+      0.023659037426114082,
+      0.02713550627231598,
+      0.02684645913541317,
+      0.02230352908372879,
+      0.022560223937034607,
+      0.02545955404639244,
+      0.019769499078392982,
+      0.025282971560955048,
+      0.017687924206256866,
+      0.03021261841058731,
+      0.02327105775475502,
+      0.029726387932896614,
+      0.028216063976287842,
+      0.021504443138837814,
+      0.021075258031487465,
+      0.02171340212225914,
+      0.022164540365338326,
+      0.019163666293025017,
+      0.026623304933309555,
+      0.03649923950433731,
+      0.019853034988045692,
+      0.03168007731437683,
+      0.02857372537255287,
+      0.021299801766872406,
+      0.0332893431186676,
+      0.020683545619249344,
+      0.027134964242577553,
+      0.027058036997914314,
+      0.028129536658525467,
+      0.022001294419169426,
+      0.03364933282136917,
+      0.024673141539096832,
+      0.03318758308887482,
+      0.027043083682656288,
+      0.028692662715911865,
+      0.036532264202833176,
+      0.02211010828614235,
+      0.018361609429121017,
+      0.021959012374281883,
+      0.020428428426384926,
+      0.019337190315127373,
+      0.02644484117627144,
+      0.036713194102048874,
+      0.017790287733078003,
+      0.023714857175946236,
+      0.022145040333271027,
+      0.018587995320558548,
+      0.027625298127532005,
+      0.017082124948501587,
+      0.014796427451074123,
+      0.02232874184846878,
+      0.021130595356225967,
+      0.019287604838609695,
+      0.019842442125082016,
+      0.02531333640217781,
+      0.023372437804937363,
+      0.018537504598498344,
+      0.022996803745627403,
+      0.016719374805688858,
+      0.016958536580204964,
+      0.018320290371775627,
+      0.03871607780456543,
+      0.02245084010064602,
+      0.021462760865688324,
+      0.02043996937572956,
+      0.021440034732222557,
+      0.016790200024843216,
+      0.03957873582839966,
+      0.02106628194451332,
+      0.023481836542487144,
+      0.02349773794412613,
+      0.02129766158759594,
+      0.025639664381742477,
+      0.03433838486671448,
+      0.030397899448871613,
+      0.018186327069997787
+    ],
+    "denoise_loss": [],
+    "ortho_loss": [
+      0.9382966756820679,
+      0.6588878035545349,
+      0.42088937759399414,
+      0.2320655733346939,
+      0.16124394536018372,
+      0.1256297528743744,
+      0.1165647804737091,
+      0.10543155670166016,
+      0.08511611074209213,
+      0.08207160979509354,
+      0.09057501703500748,
+      0.09505771845579147,
+      0.11708667874336243,
+      0.1503661572933197,
+      0.16437821090221405,
+      0.1690972000360489,
+      0.1732271909713745,
+      0.17905646562576294,
+      0.18898454308509827,
+      0.1928926408290863,
+      0.20392286777496338,
+      0.20940101146697998,
+      0.21187590062618256,
+      0.21231071650981903,
+      0.21626794338226318,
+      0.2212129533290863,
+      0.22394777834415436,
+      0.22969666123390198,
+      0.23175372183322906,
+      0.2357863336801529,
+      0.23864339292049408,
+      0.23572774231433868,
+      0.23480254411697388,
+      0.23319430649280548,
+      0.2373831570148468,
+      0.23410111665725708,
+      0.2355644702911377,
+      0.23719055950641632,
+      0.23637425899505615,
+      0.22925691306591034,
+      0.23422953486442566,
+      0.23229242861270905,
+      0.22866179049015045,
+      0.2280547022819519,
+      0.23064395785331726,
+      0.2346450686454773,
+      0.23569199442863464,
+      0.23687110841274261,
+      0.231954425573349,
+      0.23199519515037537,
+      0.2336961328983307,
+      0.2374170869588852,
+      0.23457828164100647,
+      0.23417001962661743,
+      0.23284892737865448,
+      0.23757337033748627,
+      0.23529747128486633,
+      0.23589220643043518,
+      0.2349993735551834,
+      0.23567408323287964,
+      0.23425427079200745,
+      0.23390081524848938,
+      0.23371915519237518,
+      0.23257684707641602,
+      0.23165081441402435,
+      0.23166172206401825,
+      0.23003865778446198,
+      0.22524842619895935,
+      0.22767741978168488,
+      0.22799386084079742,
+      0.2327577918767929,
+      0.23616527020931244,
+      0.2315577119588852,
+      0.22831550240516663,
+      0.23036615550518036,
+      0.22815334796905518,
+      0.23284174501895905,
+      0.235682874917984,
+      0.23753470182418823,
+      0.23702216148376465,
+      0.23697084188461304,
+      0.23746462166309357,
+      0.23534896969795227,
+      0.23484240472316742,
+      0.2350679337978363,
+      0.23661695420742035,
+      0.2368382066488266,
+      0.23605409264564514,
+      0.23691511154174805,
+      0.2369929403066635,
+      0.238261416554451,
+      0.23982560634613037,
+      0.24032315611839294,
+      0.24284431338310242,
+      0.24316035211086273,
+      0.24490563571453094,
+      0.2502349615097046,
+      0.25157418847084045,
+      0.2509077191352844,
+      0.25075310468673706,
+      0.25295525789260864,
+      0.25359466671943665,
+      0.25517353415489197,
+      0.2559907138347626,
+      0.2543465197086334,
+      0.25437554717063904,
+      0.2520972788333893,
+      0.25316494703292847,
+      0.25730597972869873,
+      0.25913918018341064,
+      0.2582869231700897,
+      0.2586173117160797,
+      0.2576594352722168,
+      0.25898581743240356,
+      0.25822708010673523,
+      0.2578895092010498,
+      0.25819700956344604,
+      0.25727710127830505,
+      0.2606615126132965,
+      0.2619830071926117,
+      0.2622619867324829,
+      0.25970587134361267,
+      0.2611936628818512,
+      0.26302391290664673,
+      0.2645387351512909,
+      0.26300355792045593,
+      0.26311883330345154,
+      0.26086539030075073,
+      0.25930899381637573,
+      0.26092007756233215,
+      0.2588067352771759,
+      0.25977230072021484,
+      0.2625751197338104,
+      0.2618725895881653,
+      0.2615964412689209,
+      0.2614551782608032,
+      0.2617577612400055,
+      0.2621341347694397,
+      0.26312968134880066,
+      0.2630600333213806,
+      0.2603296935558319,
+      0.25991344451904297,
+      0.2614287734031677,
+      0.26091518998146057,
+      0.2609129548072815,
+      0.2581985890865326,
+      0.26171350479125977,
+      0.2627491354942322,
+      0.2628113627433777,
+      0.2627332806587219,
+      0.26485535502433777,
+      0.26244014501571655,
+      0.26064935326576233,
+      0.2604113519191742,
+      0.25912243127822876,
+      0.2599487006664276,
+      0.25977423787117004,
+      0.25723594427108765,
+      0.2589156925678253,
+      0.26139792799949646,
+      0.2608095109462738,
+      0.257353812456131,
+      0.25579380989074707,
+      0.26192590594291687,
+      0.26170575618743896,
+      0.2573934495449066,
+      0.25857028365135193,
+      0.25556641817092896,
+      0.25666487216949463,
+      0.2564474940299988,
+      0.2584974765777588,
+      0.257750928401947,
+      0.2586265504360199,
+      0.25900962948799133,
+      0.25536540150642395,
+      0.2561735510826111,
+      0.25549590587615967,
+      0.25259003043174744,
+      0.25580865144729614,
+      0.2599583566188812,
+      0.2634178102016449,
+      0.2614739239215851,
+      0.2609767019748688,
+      0.25872913002967834,
+      0.2598873972892761,
+      0.2598589360713959,
+      0.26039913296699524,
+      0.261478066444397,
+      0.26389962434768677,
+      0.25965964794158936,
+      0.26012369990348816,
+      0.2614912986755371,
+      0.2609478235244751,
+      0.25799471139907837,
+      0.25930672883987427,
+      0.25793910026550293,
+      0.25631362199783325,
+      0.25630810856819153,
+      0.2574211359024048,
+      0.2584388852119446,
+      0.25894883275032043,
+      0.25936877727508545,
+      0.25757738947868347,
+      0.25764116644859314,
+      0.25714877247810364,
+      0.2566877007484436,
+      0.2558593153953552,
+      0.2578563094139099,
+      0.25735586881637573,
+      0.2594856023788452,
+      0.26005852222442627,
+      0.2609432339668274,
+      0.2613309919834137,
+      0.26132115721702576,
+      0.2618064880371094,
+      0.2622898817062378,
+      0.2635512948036194,
+      0.2644326984882355,
+      0.2645467519760132,
+      0.26274412870407104,
+      0.26223519444465637,
+      0.2642705738544464,
+      0.2645619213581085,
+      0.263324499130249,
+      0.26275208592414856,
+      0.2628836929798126,
+      0.26300525665283203,
+      0.2626466453075409,
+      0.2614835202693939,
+      0.25612545013427734,
+      0.25756406784057617,
+      0.2580561637878418,
+      0.25907471776008606,
+      0.2586274743080139,
+      0.25821614265441895,
+      0.25944051146507263,
+      0.2599451541900635,
+      0.259314626455307,
+      0.25966089963912964,
+      0.2610854208469391,
+      0.2613784372806549,
+      0.2618248164653778,
+      0.26279670000076294,
+      0.2625303268432617,
+      0.2617781162261963,
+      0.2611786425113678,
+      0.2609390914440155,
+      0.2614979147911072,
+      0.26170775294303894,
+      0.26245665550231934,
+      0.26383984088897705,
+      0.26516062021255493,
+      0.26491519808769226,
+      0.2647554278373718,
+      0.264583557844162,
+      0.264553040266037,
+      0.2634003162384033,
+      0.26309484243392944,
+      0.2632202208042145,
+      0.2634616196155548,
+      0.2637457251548767,
+      0.264485627412796,
+      0.26475146412849426,
+      0.26460742950439453,
+      0.2654464840888977,
+      0.26473498344421387,
+      0.2649029791355133,
+      0.26524946093559265,
+      0.26537907123565674,
+      0.26649829745292664,
+      0.2667599320411682,
+      0.26727795600891113,
+      0.267291784286499,
+      0.2673085629940033,
+      0.26682019233703613,
+      0.2667543888092041,
+      0.2673811912536621,
+      0.2676103115081787,
+      0.26768064498901367,
+      0.26766034960746765,
+      0.2680952548980713,
+      0.26837679743766785,
+      0.2683418393135071,
+      0.26831912994384766,
+      0.26809704303741455,
+      0.26841503381729126,
+      0.2688930928707123,
+      0.2697807848453522,
+      0.2697928547859192,
+      0.2695028781890869,
+      0.2695123851299286,
+      0.2696168124675751,
+      0.268787682056427,
+      0.269096314907074,
+      0.26906323432922363,
+      0.26884567737579346,
+      0.2693849503993988,
+      0.2692883610725403,
+      0.2692651152610779,
+      0.269359290599823,
+      0.2692016661167145,
+      0.26934701204299927,
+      0.2694322466850281,
+      0.26922088861465454,
+      0.26940470933914185,
+      0.2694062292575836,
+      0.2694178521633148,
+      0.26934027671813965,
+      0.2695278227329254,
+      0.26959916949272156
+    ],
+    "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": 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_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": 1.0
+}

add_sub_sorl_v1_abs50_K1_100K/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6513404626ebae6e2af5c39a4d009410b3c7f4f7513373fd35914c7a939152e0
+size 650466940

add_sub_sorl_v1_abs50_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_abs50_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": 50,
+  "dataset_size": 100000,
+  "mode": "sorl",
+  "device": "cuda",
+  "push_to_hub": true,
+  "no_wandb": false,
+  "n_params": 162540062,
+  "run_name": "add_sub_sorl_v1_abs50_K1_100K",
+  "git_commit": "f447da529caceac8c7d256cbb2cd185cbc50feac",
+  "timestamp": "2026-04-12T17:36:59.871013+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": "5a6y460h",
+  "wandb_url": "https://wandb.ai/nlp_and_interpretability/sorl-arithmetic/runs/5a6y460h",
+  "final_accuracy": 1.0,
+  "sft_accuracy": 1.0,
+  "eval_method": "ArithmeticEvaluator"
+}