Upload add_sub_sorl_v1_abs50_50K

add_sub_sorl_v1_abs50_50K/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_50K/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_50K/metrics.json

@@ -0,0 +1,2257 @@
+{
+  "history": {
+    "step": [
+      50,
+      100,
+      150,
+      200,
+      250,
+      300,
+      350,
+      400,
+      450,
+      500,
+      550,
+      600,
+      650,
+      700,
+      750,
+      832,
+      882,
+      932,
+      982,
+      1032,
+      1082,
+      1132,
+      1182,
+      1232,
+      1282,
+      1332,
+      1382,
+      1432,
+      1482,
+      1532,
+      1614,
+      1664,
+      1714,
+      1764,
+      1814,
+      1864,
+      1914,
+      1964,
+      2014,
+      2064,
+      2114,
+      2164,
+      2214,
+      2264,
+      2314,
+      2396,
+      2446,
+      2496,
+      2546,
+      2596,
+      2646,
+      2696,
+      2746,
+      2796,
+      2846,
+      2896,
+      2946,
+      2996,
+      3046,
+      3096,
+      3178,
+      3228,
+      3278,
+      3328,
+      3378,
+      3428,
+      3478,
+      3528,
+      3578,
+      3628,
+      3678,
+      3728,
+      3778,
+      3828,
+      3878,
+      3960,
+      4010,
+      4060,
+      4110,
+      4160,
+      4210,
+      4260,
+      4310,
+      4360,
+      4410,
+      4460,
+      4510,
+      4560,
+      4610,
+      4660,
+      4742,
+      4792,
+      4842,
+      4892,
+      4942,
+      4992,
+      5042,
+      5092,
+      5142,
+      5192,
+      5242,
+      5292,
+      5342,
+      5392,
+      5442,
+      5524,
+      5574,
+      5624,
+      5674,
+      5724,
+      5774,
+      5824,
+      5874,
+      5924,
+      5974,
+      6024,
+      6074,
+      6124,
+      6174,
+      6224,
+      6306,
+      6356,
+      6406,
+      6456,
+      6506,
+      6556,
+      6606,
+      6656,
+      6706,
+      6756,
+      6806,
+      6856,
+      6906,
+      6956,
+      7006,
+      7088,
+      7138,
+      7188,
+      7238,
+      7288,
+      7338,
+      7388,
+      7438,
+      7488,
+      7538,
+      7588,
+      7638,
+      7688,
+      7738,
+      7788
+    ],
+    "loss": [
+      7.256625652313232,
+      3.0191478729248047,
+      2.4123916625976562,
+      2.1674225330352783,
+      2.1556787490844727,
+      1.8286030292510986,
+      2.14564847946167,
+      1.900666356086731,
+      1.652380108833313,
+      1.6646597385406494,
+      1.5041191577911377,
+      1.546173334121704,
+      0.7197972536087036,
+      0.08199936151504517,
+      -3.1339755058288574,
+      -10.444205284118652,
+      -10.103414535522461,
+      -12.214149475097656,
+      -12.902936935424805,
+      -13.465953826904297,
+      -13.666388511657715,
+      -14.849788665771484,
+      -14.489677429199219,
+      -14.085216522216797,
+      -15.04929256439209,
+      -14.683259010314941,
+      -13.91379451751709,
+      -15.094310760498047,
+      -13.880881309509277,
+      -14.374068260192871,
+      -14.42313003540039,
+      -14.229098320007324,
+      -15.09312629699707,
+      -14.730694770812988,
+      -14.6654634475708,
+      -13.932817459106445,
+      -14.541162490844727,
+      -13.534533500671387,
+      -14.629840850830078,
+      -14.549479484558105,
+      -14.572128295898438,
+      -14.297953605651855,
+      -14.748451232910156,
+      -14.658515930175781,
+      -14.846949577331543,
+      -14.570510864257812,
+      -14.642899513244629,
+      -15.091267585754395,
+      -14.61697769165039,
+      -14.35709285736084,
+      -14.233384132385254,
+      -13.99492073059082,
+      -13.887598037719727,
+      -13.822426795959473,
+      -13.868623733520508,
+      -13.392026901245117,
+      -14.518260955810547,
+      -12.745328903198242,
+      -9.385150909423828,
+      -8.4232177734375,
+      -5.4400715827941895,
+      -5.325277805328369,
+      -5.132301330566406,
+      -4.313063621520996,
+      -3.695901870727539,
+      -3.7530672550201416,
+      -3.7038750648498535,
+      -3.661875009536743,
+      -3.2753937244415283,
+      -3.360624074935913,
+      -2.504333257675171,
+      -3.0331249237060547,
+      -3.0124270915985107,
+      -2.910573720932007,
+      -3.7368786334991455,
+      -2.7395472526550293,
+      -2.6733858585357666,
+      -2.886018991470337,
+      -2.6889913082122803,
+      -2.5152089595794678,
+      -2.2789855003356934,
+      -2.7361295223236084,
+      -2.4921624660491943,
+      -2.7641701698303223,
+      -4.348557949066162,
+      -2.480908155441284,
+      -2.207150936126709,
+      -2.63488507270813,
+      -2.250746726989746,
+      -2.611670970916748,
+      -2.541576385498047,
+      -2.34275484085083,
+      -2.689805746078491,
+      -2.281898021697998,
+      -2.7649309635162354,
+      -2.1240293979644775,
+      -2.607470750808716,
+      -2.4728121757507324,
+      -2.0544490814208984,
+      -2.0300307273864746,
+      -2.1023619174957275,
+      -1.887721061706543,
+      -2.0482027530670166,
+      -1.9892445802688599,
+      -1.9722803831100464,
+      -2.084959030151367,
+      -2.446387529373169,
+      -1.835938811302185,
+      -2.5824408531188965,
+      -2.1360812187194824,
+      -1.9915828704833984,
+      -1.6431983709335327,
+      -1.4731082916259766,
+      -1.7144525051116943,
+      -1.7019168138504028,
+      -1.6519891023635864,
+      -1.7413341999053955,
+      -1.588895320892334,
+      -1.539756417274475,
+      -1.904645562171936,
+      -1.5546034574508667,
+      -1.7151820659637451,
+      -1.450798511505127,
+      -1.574430227279663,
+      -1.8458795547485352,
+      -1.5194255113601685,
+      -1.6603127717971802,
+      -2.0166845321655273,
+      -1.2642289400100708,
+      -1.7409981489181519,
+      -1.6157901287078857,
+      -1.5526642799377441,
+      -1.2320806980133057,
+      -1.5149484872817993,
+      -1.4476909637451172,
+      -1.6797325611114502,
+      -1.3895442485809326,
+      -1.231076717376709,
+      -1.2591882944107056,
+      -1.7371399402618408,
+      -1.5310636758804321,
+      -1.4542627334594727,
+      -1.2413599491119385,
+      -1.226067304611206,
+      -1.6244958639144897,
+      -1.3765522241592407,
+      -1.362652063369751,
+      -1.392403244972229,
+      -1.3551605939865112,
+      -1.0807950496673584
+    ],
+    "base_loss": [
+      6.029807090759277,
+      2.3067009449005127,
+      1.9342564344406128,
+      1.8905317783355713,
+      1.8440930843353271,
+      1.8317269086837769,
+      1.8801648616790771,
+      1.8195487260818481,
+      1.7616671323776245,
+      1.7560369968414307,
+      1.7867367267608643,
+      1.7833213806152344,
+      1.7948106527328491,
+      1.7519046068191528,
+      1.948911428451538,
+      2.001823902130127,
+      1.8075017929077148,
+      1.8323737382888794,
+      1.797086477279663,
+      1.8139381408691406,
+      1.751043438911438,
+      1.8313592672348022,
+      1.751556634902954,
+      1.7140840291976929,
+      1.789836049079895,
+      1.7125715017318726,
+      1.6791929006576538,
+      1.7678550481796265,
+      1.6537859439849854,
+      1.6602624654769897,
+      1.6753387451171875,
+      1.6629526615142822,
+      1.7369420528411865,
+      1.683754563331604,
+      1.6948719024658203,
+      1.6238089799880981,
+      1.6669832468032837,
+      1.5685263872146606,
+      1.6912544965744019,
+      1.6462609767913818,
+      1.6818253993988037,
+      1.636511206626892,
+      1.6857174634933472,
+      1.669575810432434,
+      1.6751861572265625,
+      1.6484739780426025,
+      1.6467863321304321,
+      1.6952067613601685,
+      1.6533762216567993,
+      1.616503357887268,
+      1.6140310764312744,
+      1.5757274627685547,
+      1.5750893354415894,
+      1.556767225265503,
+      1.5602978467941284,
+      1.515640139579773,
+      1.6252305507659912,
+      1.4415589570999146,
+      1.0609798431396484,
+      0.9612196683883667,
+      0.616158127784729,
+      0.6064944863319397,
+      0.5862141251564026,
+      0.4924693703651428,
+      0.4313984513282776,
+      0.42700475454330444,
+      0.42180582880973816,
+      0.41957685351371765,
+      0.37408900260925293,
+      0.38498300313949585,
+      0.28954702615737915,
+      0.348208487033844,
+      0.3462330996990204,
+      0.3341756761074066,
+      0.4224683344364166,
+      0.3134305477142334,
+      0.3083657920360565,
+      0.3302149474620819,
+      0.30693626403808594,
+      0.2877400815486908,
+      0.2639155685901642,
+      0.3185356557369232,
+      0.2856982350349426,
+      0.31456050276756287,
+      0.49335941672325134,
+      0.28367578983306885,
+      0.25294229388237,
+      0.30259618163108826,
+      0.2595691978931427,
+      0.30289167165756226,
+      0.291527658700943,
+      0.26810169219970703,
+      0.3059633672237396,
+      0.2611132562160492,
+      0.3158523738384247,
+      0.2444075345993042,
+      0.29604029655456543,
+      0.2825319468975067,
+      0.23427459597587585,
+      0.23259370028972626,
+      0.2404058873653412,
+      0.21756012737751007,
+      0.23280969262123108,
+      0.22855813801288605,
+      0.2257063388824463,
+      0.23912663757801056,
+      0.27966248989105225,
+      0.21003004908561707,
+      0.29382458329200745,
+      0.2435278445482254,
+      0.2295869141817093,
+      0.18904468417167664,
+      0.17142455279827118,
+      0.1948322355747223,
+      0.19676467776298523,
+      0.19100677967071533,
+      0.19870205223560333,
+      0.18229179084300995,
+      0.17854009568691254,
+      0.2182820737361908,
+      0.1792173683643341,
+      0.19656817615032196,
+      0.16844725608825684,
+      0.18259193003177643,
+      0.21280623972415924,
+      0.17412368953227997,
+      0.19087910652160645,
+      0.22902394831180573,
+      0.14896346628665924,
+      0.20004956424236298,
+      0.18591725826263428,
+      0.17941395938396454,
+      0.14243106544017792,
+      0.17380957305431366,
+      0.16619420051574707,
+      0.19304493069648743,
+      0.15973064303398132,
+      0.1419793665409088,
+      0.1448678970336914,
+      0.2004563808441162,
+      0.17605385184288025,
+      0.16834628582000732,
+      0.14428769052028656,
+      0.14093129336833954,
+      0.18535971641540527,
+      0.1585533171892166,
+      0.1568572074174881,
+      0.15938758850097656,
+      0.15521882474422455,
+      0.126205712556839
+    ],
+    "info_loss": [
+      -0.20561504364013672,
+      -0.03972625732421875,
+      -0.048748135566711426,
+      -0.06750452518463135,
+      -0.06405246257781982,
+      -0.09477114677429199,
+      -0.06761360168457031,
+      -0.08672130107879639,
+      -0.10533666610717773,
+      -0.10386943817138672,
+      -0.12152564525604248,
+      -0.11574661731719971,
+      -0.19611811637878418,
+      -0.24177753925323486,
+      -0.5657413005828857,
+      -1.2865196466445923,
+      -1.2276489734649658,
+      -1.4379183053970337,
+      -1.5007377862930298,
+      -1.5571799278259277,
+      -1.5694807767868042,
+      -1.6956429481506348,
+      -1.6478774547576904,
+      -1.6037368774414062,
+      -1.7057857513427734,
+      -1.6578631401062012,
+      -1.5784200429916382,
+      -1.7046573162078857,
+      -1.569814682006836,
+      -1.6190284490585327,
+      -1.6252418756484985,
+      -1.6039395332336426,
+      -1.6967638731002808,
+      -1.6540355682373047,
+      -1.6475841999053955,
+      -1.5689774751663208,
+      -1.635087251663208,
+      -1.5238193273544312,
+      -1.6450031995773315,
+      -1.630631685256958,
+      -1.637630820274353,
+      -1.6060518026351929,
+      -1.6621246337890625,
+      -1.6464064121246338,
+      -1.6617577075958252,
+      -1.6344650983810425,
+      -1.6399197578430176,
+      -1.6889852285385132,
+      -1.639074444770813,
+      -1.6081275939941406,
+      -1.5951539278030396,
+      -1.566024661064148,
+      -1.5557743310928345,
+      -1.5495342016220093,
+      -1.5556063652038574,
+      -1.5009874105453491,
+      -1.6231387853622437,
+      -1.4322359561920166,
+      -1.0561802387237549,
+      -0.9466190338134766,
+      -0.6137336492538452,
+      -0.6038743257522583,
+      -0.5819218754768372,
+      -0.4909442663192749,
+      -0.4217419922351837,
+      -0.4254307448863983,
+      -0.4203161597251892,
+      -0.4181394577026367,
+      -0.3734082281589508,
+      -0.38232800364494324,
+      -0.2871572971343994,
+      -0.34667515754699707,
+      -0.3433464467525482,
+      -0.3323284983634949,
+      -0.4220069944858551,
+      -0.3129675090312958,
+      -0.30473649501800537,
+      -0.32990047335624695,
+      -0.3063972592353821,
+      -0.28762325644493103,
+      -0.2610863447189331,
+      -0.312552273273468,
+      -0.2848331332206726,
+      -0.3143031597137451,
+      -0.49174460768699646,
+      -0.2831017076969147,
+      -0.25272035598754883,
+      -0.3016011118888855,
+      -0.2592092454433441,
+      -0.2988108992576599,
+      -0.2913071811199188,
+      -0.2679586410522461,
+      -0.3054676353931427,
+      -0.26094746589660645,
+      -0.3152065873146057,
+      -0.24421927332878113,
+      -0.295553594827652,
+      -0.28226152062416077,
+      -0.2340558022260666,
+      -0.23213818669319153,
+      -0.24020898342132568,
+      -0.21750272810459137,
+      -0.23276019096374512,
+      -0.22842691838741302,
+      -0.22563359141349792,
+      -0.23904889822006226,
+      -0.2794150114059448,
+      -0.20988059043884277,
+      -0.29378125071525574,
+      -0.24314871430397034,
+      -0.2293650060892105,
+      -0.18900363147258759,
+      -0.17137543857097626,
+      -0.1948050558567047,
+      -0.19659706950187683,
+      -0.19093747437000275,
+      -0.19865652918815613,
+      -0.1822470873594284,
+      -0.1785041093826294,
+      -0.21816547214984894,
+      -0.17919395864009857,
+      -0.19648776948451996,
+      -0.16839498281478882,
+      -0.18214161694049835,
+      -0.212591290473938,
+      -0.17409944534301758,
+      -0.19081123173236847,
+      -0.2289951741695404,
+      -0.14660219848155975,
+      -0.20001469552516937,
+      -0.18589811027050018,
+      -0.1793726533651352,
+      -0.14236652851104736,
+      -0.1737845093011856,
+      -0.16613999009132385,
+      -0.19301597774028778,
+      -0.15971405804157257,
+      -0.14195962250232697,
+      -0.14485248923301697,
+      -0.20041891932487488,
+      -0.17602841556072235,
+      -0.16832193732261658,
+      -0.14427153766155243,
+      -0.14090751111507416,
+      -0.18532036244869232,
+      -0.1585266888141632,
+      -0.15683983266353607,
+      -0.15936271846294403,
+      -0.1551767736673355,
+      -0.12618587911128998
+    ],
+    "abs_loss": [
+      3.7497520446777344,
+      2.6481971740722656,
+      2.8985848426818848,
+      2.818944215774536,
+      2.870543956756592,
+      2.872476100921631,
+      2.758188247680664,
+      2.8428196907043457,
+      2.7872250080108643,
+      2.910464286804199,
+      2.752615213394165,
+      2.7820749282836914,
+      2.7789087295532227,
+      2.835494041442871,
+      2.5386619567871094,
+      2.0468544960021973,
+      1.7976104021072388,
+      1.6245003938674927,
+      1.4929277896881104,
+      1.4750802516937256,
+      1.5771673917770386,
+      1.666552186012268,
+      1.0843603610992432,
+      1.545884132385254,
+      1.4265331029891968,
+      0.999521017074585,
+      1.2854008674621582,
+      1.195152759552002,
+      1.1227796077728271,
+      0.9406141042709351,
+      1.0983028411865234,
+      0.85012286901474,
+      0.8116261959075928,
+      0.7224507927894592,
+      0.8427834510803223,
+      0.8805156946182251,
+      0.8626446723937988,
+      0.9968369007110596,
+      0.9508752822875977,
+      0.8513461351394653,
+      0.6574622392654419,
+      0.8986561298370361,
+      0.5942917466163635,
+      0.7797585129737854,
+      0.7031676173210144,
+      0.6163696050643921,
+      0.6315547227859497,
+      0.7303791046142578,
+      0.5330150723457336,
+      0.5637519359588623,
+      0.5594895482063293,
+      0.5726730823516846,
+      0.5400386452674866,
+      0.5256966352462769,
+      0.44924086332321167,
+      0.47774025797843933,
+      0.5889784097671509,
+      0.4862280488014221,
+      0.5457887649536133,
+      0.4811607599258423,
+      0.5763522982597351,
+      0.4618305563926697,
+      0.49888068437576294,
+      0.5134003758430481,
+      0.4723852872848511,
+      0.4225727915763855,
+      0.3785339295864105,
+      0.3270162343978882,
+      0.3133487105369568,
+      0.33950677514076233,
+      0.32967162132263184,
+      0.3334130644798279,
+      0.31977397203445435,
+      0.39836356043815613,
+      0.3870335817337036,
+      0.27905893325805664,
+      0.38264474272727966,
+      0.3101273775100708,
+      0.2242627888917923,
+      0.24193796515464783,
+      0.222909078001976,
+      0.20927608013153076,
+      0.23676183819770813,
+      0.30053985118865967,
+      0.34205877780914307,
+      0.22115281224250793,
+      0.252194344997406,
+      0.2795361876487732,
+      0.2963107228279114,
+      0.26521268486976624,
+      0.3244202733039856,
+      0.28083187341690063,
+      0.18782655894756317,
+      0.273157000541687,
+      0.21937644481658936,
+      0.2191070020198822,
+      0.2309451699256897,
+      0.2086794078350067,
+      0.20153307914733887,
+      0.2183530181646347,
+      0.17671014368534088,
+      0.1841510534286499,
+      0.22086754441261292,
+      0.23169928789138794,
+      0.27007248997688293,
+      0.20892009139060974,
+      0.17916454374790192,
+      0.20730438828468323,
+      0.20521558821201324,
+      0.17557865381240845,
+      0.18920457363128662,
+      0.20690660178661346,
+      0.1799832135438919,
+      0.15656515955924988,
+      0.18862086534500122,
+      0.17772457003593445,
+      0.17366942763328552,
+      0.16974030435085297,
+      0.16747745871543884,
+      0.16128060221672058,
+      0.19063544273376465,
+      0.1916358768939972,
+      0.20035310089588165,
+      0.16092176735401154,
+      0.1909703016281128,
+      0.1657269448041916,
+      0.1640641689300537,
+      0.14971278607845306,
+      0.10842731595039368,
+      0.2196534425020218,
+      0.16537384688854218,
+      0.1963886320590973,
+      0.16384944319725037,
+      0.13870114088058472,
+      0.147671639919281,
+      0.17166978120803833,
+      0.15626773238182068,
+      0.17788034677505493,
+      0.1344836950302124,
+      0.1465415060520172,
+      0.19216865301132202,
+      0.13693669438362122,
+      0.1630391776561737,
+      0.11989884078502655,
+      0.1325468271970749,
+      0.1899355947971344,
+      0.12264394760131836,
+      0.12508927285671234,
+      0.1484333872795105,
+      0.13628548383712769
+    ],
+    "zipf_loss": [
+      2.907993793487549,
+      0.8448899388313293,
+      0.6757582426071167,
+      0.6700415015220642,
+      0.6650558114051819,
+      0.6573399305343628,
+      0.6658006906509399,
+      0.6640486717224121,
+      0.6653571128845215,
+      0.6562708020210266,
+      0.6573773622512817,
+      0.6421106457710266,
+      0.6082768440246582,
+      0.46432074904441833,
+      0.3206597566604614,
+      0.21448186039924622,
+      0.1858120709657669,
+      0.17020957171916962,
+      0.1580614596605301,
+      0.14439967274665833,
+      0.11965931951999664,
+      0.1086258664727211,
+      0.12910524010658264,
+      0.08348017930984497,
+      0.07607515156269073,
+      0.0828506276011467,
+      0.06267359852790833,
+      0.06489258259534836,
+      0.051201872527599335,
+      0.06189261004328728,
+      0.04411982744932175,
+      0.06233106553554535,
+      0.05640706419944763,
+      0.05366092175245285,
+      0.031228451058268547,
+      0.04509659856557846,
+      0.05646108090877533,
+      0.0354497954249382,
+      0.033848002552986145,
+      0.025441396981477737,
+      0.05660788342356682,
+      0.0361880287528038,
+      0.12764789164066315,
+      0.057995907962322235,
+      0.025125017389655113,
+      0.06402861326932907,
+      0.046355463564395905,
+      0.03033997118473053,
+      0.06708943098783493,
+      0.05130421370267868,
+      0.04817532002925873,
+      0.03233150392770767,
+      0.041051946580410004,
+      0.06357830762863159,
+      0.08221856504678726,
+      0.05443323031067848,
+      0.028998034074902534,
+      0.08684802800416946,
+      0.0610925555229187,
+      0.0336371473968029,
+      0.023472048342227936,
+      0.06078788638114929,
+      0.05081494152545929,
+      0.05256975442171097,
+      0.042881276458501816,
+      0.031977828592061996,
+      0.03962726891040802,
+      0.06724098324775696,
+      0.05326463282108307,
+      0.04372230917215347,
+      0.04472571983933449,
+      0.052076928317546844,
+      0.042826779186725616,
+      0.03869928792119026,
+      0.02201938070356846,
+      0.04879137873649597,
+      0.02734885923564434,
+      0.05175810307264328,
+      0.04561843350529671,
+      0.04908981919288635,
+      0.045671574771404266,
+      0.04992988705635071,
+      0.04679465293884277,
+      0.0342470221221447,
+      0.04132283851504326,
+      0.04431803524494171,
+      0.04189082980155945,
+      0.0505763441324234,
+      0.05214542895555496,
+      0.047025106847286224,
+      0.04752563685178757,
+      0.04064689949154854,
+      0.04012472555041313,
+      0.03914778679609299,
+      0.04934476315975189,
+      0.051845304667949677,
+      0.028930390253663063,
+      0.04640305042266846,
+      0.03168107569217682,
+      0.03692198544740677,
+      0.041651077568531036,
+      0.05133098363876343,
+      0.024502430111169815,
+      0.04329643025994301,
+      0.031341951340436935,
+      0.04551147669553757,
+      0.0501837432384491,
+      0.03210654854774475,
+      0.041025444865226746,
+      0.03432022035121918,
+      0.0535600446164608,
+      0.03710262104868889,
+      0.051223304122686386,
+      0.02310934104025364,
+      0.04842713102698326,
+      0.04860638454556465,
+      0.029162120074033737,
+      0.03430991619825363,
+      0.04999687895178795,
+      0.042598988860845566,
+      0.03905525058507919,
+      0.03396385908126831,
+      0.04466875270009041,
+      0.04830180108547211,
+      0.048130013048648834,
+      0.03087255358695984,
+      0.040513940155506134,
+      0.029271963983774185,
+      0.04198697581887245,
+      0.037133827805519104,
+      0.04073633253574371,
+      0.04200948029756546,
+      0.032768625766038895,
+      0.03521694615483284,
+      0.03274751082062721,
+      0.040215421468019485,
+      0.03223889321088791,
+      0.02875204011797905,
+      0.031020455062389374,
+      0.05193880572915077,
+      0.03394973650574684,
+      0.046916697174310684,
+      0.04076382517814636,
+      0.030086800456047058,
+      0.03009340539574623,
+      0.031167635694146156,
+      0.03662469983100891,
+      0.029327519237995148,
+      0.02654491364955902,
+      0.04122946411371231
+    ],
+    "denoise_loss": [],
+    "ortho_loss": [
+      0.10270898044109344,
+      0.045592378824949265,
+      0.034897129982709885,
+      0.03289920464158058,
+      0.03318662568926811,
+      0.035863105207681656,
+      0.03718609735369682,
+      0.04016987606883049,
+      0.04869673773646355,
+      0.053654879331588745,
+      0.061492759734392166,
+      0.06294921040534973,
+      0.06910595297813416,
+      0.07273681461811066,
+      0.076898954808712,
+      0.0958031639456749,
+      0.1058020293712616,
+      0.11720982193946838,
+      0.13132666051387787,
+      0.1452067792415619,
+      0.15491452813148499,
+      0.16184890270233154,
+      0.17446120083332062,
+      0.18189102411270142,
+      0.1888672560453415,
+      0.19659653306007385,
+      0.1996685415506363,
+      0.20556356012821198,
+      0.2079426646232605,
+      0.21441443264484406,
+      0.2170758992433548,
+      0.22126887738704681,
+      0.22582681477069855,
+      0.22605378925800323,
+      0.2280348688364029,
+      0.2285757064819336,
+      0.22907176613807678,
+      0.23065848648548126,
+      0.23264732956886292,
+      0.234690859913826,
+      0.23757733404636383,
+      0.23970840871334076,
+      0.2384733259677887,
+      0.23657508194446564,
+      0.23852205276489258,
+      0.24019040167331696,
+      0.23936468362808228,
+      0.23304446041584015,
+      0.23387999832630157,
+      0.2349344789981842,
+      0.23946575820446014,
+      0.24182644486427307,
+      0.24514825642108917,
+      0.2487141191959381,
+      0.2516993284225464,
+      0.2516952455043793,
+      0.2530873715877533,
+      0.2569259703159332,
+      0.2646262049674988,
+      0.2720961272716522,
+      0.27863892912864685,
+      0.2827668786048889,
+      0.28322896361351013,
+      0.29026034474372864,
+      0.296204149723053,
+      0.2996149957180023,
+      0.3005698323249817,
+      0.30707406997680664,
+      0.3101491630077362,
+      0.3167986273765564,
+      0.3269317150115967,
+      0.32879483699798584,
+      0.33054113388061523,
+      0.33136507868766785,
+      0.3334370255470276,
+      0.3420204818248749,
+      0.34395188093185425,
+      0.3462192118167877,
+      0.350506067276001,
+      0.3514798581600189,
+      0.35246649384498596,
+      0.3512493073940277,
+      0.3530470132827759,
+      0.355428546667099,
+      0.359707772731781,
+      0.3637158274650574,
+      0.3649544417858124,
+      0.3662114441394806,
+      0.3675302267074585,
+      0.37116730213165283,
+      0.37062546610832214,
+      0.37135010957717896,
+      0.3697737455368042,
+      0.3697727620601654,
+      0.3714151084423065,
+      0.37123236060142517,
+      0.3738124966621399,
+      0.3752833604812622,
+      0.37488511204719543,
+      0.3777393102645874,
+      0.37878528237342834,
+      0.3812442421913147,
+      0.38351762294769287,
+      0.38525545597076416,
+      0.3856000006198883,
+      0.3862100839614868,
+      0.3856704533100128,
+      0.38774314522743225,
+      0.38804271817207336,
+      0.3885641396045685,
+      0.38839849829673767,
+      0.3905911147594452,
+      0.39113733172416687,
+      0.39186158776283264,
+      0.39222121238708496,
+      0.39306116104125977,
+      0.3945384621620178,
+      0.39511334896087646,
+      0.39469024538993835,
+      0.39563339948654175,
+      0.39637646079063416,
+      0.3954394459724426,
+      0.39604392647743225,
+      0.3968430161476135,
+      0.3972108066082001,
+      0.39802664518356323,
+      0.4002288281917572,
+      0.4006965458393097,
+      0.40092286467552185,
+      0.4012592136859894,
+      0.4014245271682739,
+      0.4025733470916748,
+      0.4031515419483185,
+      0.4036944806575775,
+      0.40446820855140686,
+      0.40457797050476074,
+      0.4052768647670746,
+      0.4059322774410248,
+      0.4069337546825409,
+      0.4073435664176941,
+      0.40719661116600037,
+      0.40762969851493835,
+      0.40837234258651733,
+      0.40862128138542175,
+      0.4101361334323883,
+      0.41058194637298584,
+      0.4111683964729309,
+      0.4115133285522461,
+      0.41178953647613525,
+      0.41227221488952637
+    ],
+    "lr": [
+      7.840000000000001e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      7.932818532818534e-05,
+      7.816988416988418e-05,
+      7.701158301158302e-05,
+      7.585328185328185e-05,
+      7.469498069498071e-05,
+      7.353667953667954e-05,
+      7.237837837837838e-05,
+      7.122007722007721e-05,
+      7.006177606177606e-05,
+      6.890347490347492e-05,
+      6.774517374517375e-05,
+      6.65868725868726e-05,
+      6.542857142857144e-05,
+      6.427027027027027e-05,
+      6.311196911196911e-05,
+      6.121235521235521e-05,
+      6.0054054054054064e-05,
+      5.8895752895752895e-05,
+      5.773745173745175e-05,
+      5.6579150579150584e-05,
+      5.542084942084943e-05,
+      5.426254826254825e-05,
+      5.310424710424711e-05,
+      5.194594594594594e-05,
+      5.0787644787644786e-05,
+      4.9629343629343644e-05,
+      4.8471042471042475e-05,
+      4.7312741312741326e-05,
+      4.615444015444014e-05,
+      4.4996138996139e-05,
+      4.309652509652511e-05,
+      4.1938223938223946e-05,
+      4.07799227799228e-05,
+      3.962162162162162e-05,
+      3.846332046332047e-05,
+      3.73050193050193e-05,
+      3.6146718146718155e-05,
+      3.4988416988416986e-05,
+      3.383011583011584e-05,
+      3.267181467181467e-05,
+      3.151351351351352e-05,
+      3.0355212355212367e-05,
+      2.9196911196911198e-05,
+      2.8038610038610046e-05,
+      2.6880308880308876e-05,
+      2.4980694980694983e-05,
+      2.3822393822393838e-05,
+      2.266409266409267e-05,
+      2.1505791505791517e-05,
+      2.0347490347490348e-05,
+      1.9189189189189195e-05,
+      1.8030888030888026e-05,
+      1.6872586872586878e-05,
+      1.571428571428571e-05,
+      1.455598455598456e-05,
+      1.3397683397683389e-05,
+      1.223938223938224e-05,
+      1.1081081081081092e-05,
+      9.92277992277992e-06,
+      8.764478764478772e-06
+    ],
+    "emb_lr": [],
+    "eval_step": [
+      750,
+      1532,
+      2314,
+      3096,
+      3878,
+      4660,
+      5442,
+      6224,
+      7006,
+      7788
+    ],
+    "eval_accuracy": [
+      0.03,
+      0.92,
+      0.95,
+      0.99,
+      1.0,
+      1.0,
+      1.0,
+      1.0,
+      1.0,
+      1.0
+    ]
+  },
+  "final_accuracy": 1.0,
+  "sft_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": null,
+      "mode": "sft",
+      "n_digits": 6,
+      "n_per_split": 50
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 0.66,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9457627118644067,
+            "count": 295
+          },
+          "SS": {
+            "accuracy": 0.9272727272727272,
+            "count": 55
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 0.78,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9761904761904762,
+            "count": 126
+          },
+          "SC": {
+            "accuracy": 0.9367088607594937,
+            "count": 79
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 21
+          },
+          "UC": {
+            "accuracy": 0.9758064516129032,
+            "count": 124
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 0.42,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.92,
+            "count": 75
+          },
+          "SC": {
+            "accuracy": 0.9193548387096774,
+            "count": 62
+          },
+          "SS": {
+            "accuracy": 0.8205128205128205,
+            "count": 39
+          },
+          "UC": {
+            "accuracy": 0.8378378378378378,
+            "count": 111
+          },
+          "US": {
+            "accuracy": 0.9047619047619048,
+            "count": 63
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 0.34,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.95,
+            "count": 60
+          },
+          "SC": {
+            "accuracy": 0.8947368421052632,
+            "count": 57
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 19
+          },
+          "UC": {
+            "accuracy": 0.7884615384615384,
+            "count": 104
+          },
+          "US": {
+            "accuracy": 0.8545454545454545,
+            "count": 110
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 0.36,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 48
+          },
+          "SC": {
+            "accuracy": 0.9807692307692307,
+            "count": 52
+          },
+          "SS": {
+            "accuracy": 0.8571428571428571,
+            "count": 7
+          },
+          "UC": {
+            "accuracy": 0.7303370786516854,
+            "count": 89
+          },
+          "US": {
+            "accuracy": 0.6688311688311688,
+            "count": 154
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 0.18,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "SC": {
+            "accuracy": 0.98,
+            "count": 50
+          },
+          "UC": {
+            "accuracy": 0.44,
+            "count": 50
+          },
+          "US": {
+            "accuracy": 0.375,
+            "count": 200
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 0.3,
+        "n_examples": 50,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UC": {
+            "accuracy": 0.34,
+            "count": 50
+          },
+          "US": {
+            "accuracy": 0.452,
+            "count": 250
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 0.795,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9791183294663574,
+            "count": 431
+          },
+          "SC": {
+            "accuracy": 0.9778481012658228,
+            "count": 316
+          },
+          "SS": {
+            "accuracy": 0.8717948717948718,
+            "count": 39
+          },
+          "UC": {
+            "accuracy": 0.95,
+            "count": 560
+          },
+          "US": {
+            "accuracy": 0.9259259259259259,
+            "count": 54
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 0.58,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 150
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UC": {
+            "accuracy": 0.8365384615384616,
+            "count": 104
+          },
+          "US": {
+            "accuracy": 0.8260869565217391,
+            "count": 46
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 0.42,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.99,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UC": {
+            "accuracy": 0.8130081300813008,
+            "count": 123
+          },
+          "US": {
+            "accuracy": 0.6883116883116883,
+            "count": 77
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 0.32,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "SC": {
+            "accuracy": 0.98,
+            "count": 50
+          },
+          "UC": {
+            "accuracy": 0.7792207792207793,
+            "count": 154
+          },
+          "US": {
+            "accuracy": 0.8541666666666666,
+            "count": 96
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 0.28,
+        "n_examples": 50,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UC": {
+            "accuracy": 0.7692307692307693,
+            "count": 182
+          },
+          "US": {
+            "accuracy": 0.7711864406779662,
+            "count": 118
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 0.64,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.935374149659864,
+            "count": 294
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 0.84,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.958041958041958,
+            "count": 143
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 69
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 15
+          },
+          "UB": {
+            "accuracy": 0.975609756097561,
+            "count": 123
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 0.36,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9351851851851852,
+            "count": 108
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 52
+          },
+          "ME": {
+            "accuracy": 0.9615384615384616,
+            "count": 52
+          },
+          "UB": {
+            "accuracy": 0.6551724137931034,
+            "count": 87
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 51
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 0.12,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9893617021276596,
+            "count": 94
+          },
+          "MB": {
+            "accuracy": 0.9803921568627451,
+            "count": 51
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 25
+          },
+          "UB": {
+            "accuracy": 0.6410256410256411,
+            "count": 78
+          },
+          "UD": {
+            "accuracy": 0.6078431372549019,
+            "count": 102
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 0.06,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.99,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UB": {
+            "accuracy": 0.66,
+            "count": 50
+          },
+          "UD": {
+            "accuracy": 0.34,
+            "count": 150
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 0.2,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UB": {
+            "accuracy": 0.82,
+            "count": 50
+          },
+          "UD": {
+            "accuracy": 0.38,
+            "count": 200
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 0.775,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9693877551020408,
+            "count": 588
+          },
+          "MB": {
+            "accuracy": 0.9925373134328358,
+            "count": 268
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 60
+          },
+          "UB": {
+            "accuracy": 0.9328859060402684,
+            "count": 447
+          },
+          "UD": {
+            "accuracy": 0.8918918918918919,
+            "count": 37
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 0.56,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9933333333333333,
+            "count": 150
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UB": {
+            "accuracy": 0.8317757009345794,
+            "count": 107
+          },
+          "UD": {
+            "accuracy": 0.8372093023255814,
+            "count": 43
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 0.34,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UB": {
+            "accuracy": 0.8157894736842105,
+            "count": 114
+          },
+          "UD": {
+            "accuracy": 0.6046511627906976,
+            "count": 86
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 0.28,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UB": {
+            "accuracy": 0.7843137254901961,
+            "count": 153
+          },
+          "UD": {
+            "accuracy": 0.6597938144329897,
+            "count": 97
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 0.5107142857142857,
+      "total_examples": 1400,
+      "n_splits": 22
+    }
+  },
+  "sorl_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": 4,
+      "mode": "sorl",
+      "n_digits": 6,
+      "n_per_split": 50
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 295
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 55
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 126
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 79
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 21
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 124
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 75
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 62
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 39
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 111
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 63
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 60
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 57
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 19
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 104
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 110
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 48
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 52
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 7
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 89
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 154
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 200
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 250
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 431
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 316
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 39
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 560
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 54
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 150
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 104
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 46
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 123
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 77
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 154
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 96
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 182
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 118
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 294
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 143
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 69
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 15
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 123
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 108
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 52
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 52
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 87
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 51
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 94
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 51
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 25
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 78
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 102
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 150
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 200
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 588
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 268
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 60
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 447
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 37
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 150
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 107
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 43
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 114
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 86
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 1.0,
+        "n_examples": 50,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 50
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 153
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 97
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 1.0,
+      "total_examples": 1400,
+      "n_splits": 22
+    }
+  },
+  "sorl_overall_accuracy": 1.0,
+  "sft_overall_accuracy": 0.5107142857142857
+}

add_sub_sorl_v1_abs50_50K/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3807a271457535e90facd4a2c2b7452fb1470e8345fa33b069288d979eff66d5
+size 650466940

add_sub_sorl_v1_abs50_50K/train_config.json

@@ -0,0 +1,35 @@
+{
+  "mode": "sorl",
+  "ops": "add_sub",
+  "n_digits": 6,
+  "n_layer": 2,
+  "n_head": 3,
+  "n_embd": 510,
+  "abs_vocab": 50,
+  "K": 4,
+  "alpha_info_gain": 10.0,
+  "alpha_abs": 0.1,
+  "alpha_soft_zipf": 1.0,
+  "batch_size": 64,
+  "num_epochs": 10,
+  "dataset_size": 50000,
+  "lr": 8e-05,
+  "output_dir": "ckpt/sweep/as_sorl_abs50_K4_50K",
+  "device": "cuda",
+  "push_to_hub": true,
+  "no_wandb": false,
+  "n_params": 162540062,
+  "run_name": "add_sub_sorl_v1_abs50_50K",
+  "git_commit": "800625019270114adcda289bbd550c4f1109a514",
+  "timestamp": "2026-04-12T03:52:52.696357+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": "d8644t69",
+  "wandb_url": "https://wandb.ai/nlp_and_interpretability/sorl-arithmetic/runs/d8644t69",
+  "final_accuracy": 1.0,
+  "sft_accuracy": 0.5107142857142857,
+  "eval_method": "ArithmeticEvaluator"
+}