Upload add_sub_sorl_v1_abs10_K1_100K_1L3H510d

add_sub_sorl_v1_abs10_K1_100K_1L3H510d/config.json

@@ -0,0 +1,36 @@
+{
+  "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"
+  ],
+  "max_position_embeddings": 128,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 3,
+  "num_hidden_layers": 1,
+  "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": 151654
+}

add_sub_sorl_v1_abs10_K1_100K_1L3H510d/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_abs10_K1_100K_1L3H510d/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": [
+      0.20298480987548828,
+      -3.8611974716186523,
+      5.484951496124268,
+      7.62824010848999,
+      6.486814022064209,
+      5.183938980102539,
+      3.481173515319824,
+      3.4660756587982178,
+      3.1551809310913086,
+      2.974534749984741,
+      3.0607967376708984,
+      3.137465476989746,
+      3.213205099105835,
+      2.8172616958618164,
+      3.159719467163086,
+      3.036630630493164,
+      3.136214256286621,
+      3.10473370552063,
+      1.0449540615081787,
+      -1.9889345169067383,
+      -3.1156959533691406,
+      -5.027166366577148,
+      -7.135016918182373,
+      -7.524842739105225,
+      -8.15284252166748,
+      -7.756969451904297,
+      -8.397037506103516,
+      -7.77532958984375,
+      -8.280281066894531,
+      -8.359915733337402,
+      -8.630762100219727,
+      -8.702438354492188,
+      -8.910720825195312,
+      -8.099109649658203,
+      -7.9632792472839355,
+      -6.994848728179932,
+      -7.3840508460998535,
+      -7.226053714752197,
+      -7.466033935546875,
+      -6.1763529777526855,
+      -5.937272071838379,
+      -6.131126880645752,
+      -6.807259559631348,
+      -5.925022602081299,
+      -4.991553783416748,
+      -6.03489875793457,
+      -4.036592483520508,
+      -3.5000007152557373,
+      -2.7816596031188965,
+      -2.0584523677825928,
+      -1.9878346920013428,
+      -1.1864898204803467,
+      -1.0069959163665771,
+      -0.5924010276794434,
+      -0.9307534694671631,
+      -0.7978959679603577,
+      -0.4037044048309326,
+      -0.7424840331077576,
+      -1.0898077487945557,
+      -0.8627956509590149,
+      -0.9351441860198975,
+      -0.9845731258392334,
+      -0.7456449270248413,
+      -0.8034361004829407,
+      -0.9816672205924988,
+      -1.1051112413406372,
+      -0.6999305486679077,
+      -0.44957953691482544,
+      -0.9589239358901978,
+      -0.888559877872467,
+      -0.8973594903945923,
+      -0.3362390995025635,
+      -0.5095456838607788,
+      -0.738105833530426,
+      -0.7219609618186951,
+      -0.9851813912391663,
+      -0.9534288644790649,
+      -0.5282067060470581,
+      -0.9321163892745972,
+      -0.3476067781448364,
+      -1.0879435539245605,
+      -0.8613939881324768,
+      -0.5899012088775635,
+      -0.48227307200431824,
+      -0.8573293089866638,
+      -0.7841486930847168,
+      -0.853835940361023,
+      -0.9318222403526306,
+      -0.9501145482063293,
+      -0.6094979643821716,
+      -0.7221546769142151,
+      -0.4836696982383728,
+      -0.7551224231719971,
+      -0.8387091755867004,
+      -0.9226763248443604,
+      -0.6931058764457703,
+      -0.8534595370292664,
+      -0.6345930695533752,
+      -0.599308967590332,
+      -0.6138797402381897,
+      -0.8734613656997681,
+      -0.5013921856880188,
+      -1.023414134979248,
+      -0.4975615441799164,
+      -0.9184777140617371,
+      -0.6177459359169006,
+      -0.6793912649154663,
+      -0.9103295207023621,
+      -0.9629290699958801,
+      -0.8296101093292236,
+      -0.7456335425376892,
+      -0.9889292120933533,
+      -0.932847797870636,
+      -0.719605028629303,
+      -0.8195403814315796,
+      -0.6204489469528198,
+      -0.9444902539253235,
+      -0.6702038645744324,
+      -0.6769301891326904,
+      -0.7704527974128723,
+      -0.3604033589363098,
+      -0.7685727477073669,
+      -0.592564046382904,
+      -0.6542524695396423,
+      -0.7130612134933472,
+      -0.5456188917160034,
+      -0.5080212950706482,
+      -0.49610766768455505,
+      -0.8628892302513123,
+      -0.6637751460075378,
+      -0.08357095718383789,
+      -0.6479129195213318,
+      -0.6967795491218567,
+      -0.4980502426624298,
+      -0.453680157661438,
+      -0.48563891649246216,
+      -0.6851292252540588,
+      -0.407563716173172,
+      -0.32402294874191284,
+      -0.5405899882316589,
+      -1.0473370552062988,
+      -0.4864315986633301,
+      -0.4536849856376648,
+      -0.639918863773346,
+      -0.3336927890777588,
+      -0.44088730216026306,
+      -0.4688369035720825,
+      -0.5984885692596436,
+      -0.40976715087890625,
+      -0.4776838719844818,
+      -0.6040022969245911,
+      -0.32387280464172363,
+      -0.6155667901039124,
+      -0.2283317893743515,
+      -0.796503484249115,
+      -0.37979862093925476,
+      -0.10048076510429382,
+      -0.4180724024772644,
+      -0.5275349020957947,
+      -0.4936784505844116,
+      0.1532474011182785,
+      -0.3471519649028778,
+      -0.6398682594299316,
+      -0.3455611765384674,
+      -0.6288833618164062,
+      -0.6393107175827026,
+      -0.16909603774547577,
+      -0.16404710710048676,
+      -0.49092090129852295,
+      -0.4603574573993683,
+      -0.6255804300308228,
+      -0.3665814697742462,
+      -0.3863292932510376,
+      -0.5968372225761414,
+      -0.38025856018066406,
+      -0.3913044333457947,
+      -0.7347709536552429,
+      -0.6311700344085693,
+      -0.18604643642902374,
+      -0.07646475732326508,
+      -0.459421843290329,
+      -0.22097186744213104,
+      -0.252154678106308,
+      -0.37326860427856445,
+      -0.3863140940666199,
+      -0.43039947748184204,
+      -0.2790504992008209,
+      -0.23125287890434265,
+      -0.33697766065597534,
+      -0.3612242639064789,
+      -0.3748737871646881,
+      -0.2123483568429947,
+      -0.20051726698875427,
+      -0.19146300852298737,
+      -0.2342536449432373,
+      -0.3006262183189392,
+      -0.3093010485172272,
+      -0.43850764632225037,
+      -0.356301873922348,
+      -0.29953649640083313,
+      -0.41408610343933105,
+      -0.239955335855484,
+      -0.06662970781326294,
+      -0.13520391285419464,
+      -0.10285133868455887,
+      -0.41884657740592957,
+      -0.3867041766643524,
+      -0.287188321352005,
+      -0.21670591831207275,
+      -0.2855888903141022,
+      -0.3601643443107605,
+      -0.40222305059432983,
+      -0.16422344744205475,
+      -0.35010814666748047,
+      -0.6421605944633484,
+      -0.4620475172996521,
+      -0.2331358790397644,
+      -0.36789363622665405,
+      -0.5173572301864624,
+      -0.2295636385679245,
+      -0.6438044309616089,
+      -0.3908720314502716,
+      -0.4907151460647583,
+      -0.436318576335907,
+      -0.002603362314403057,
+      -0.1944722980260849,
+      -0.3558172881603241,
+      -0.0964510440826416,
+      -0.2997216284275055,
+      -0.4600769579410553,
+      -0.010438470169901848,
+      0.11934158951044083,
+      -0.07519880682229996,
+      -0.39747369289398193,
+      -0.413342148065567,
+      -0.22194483876228333,
+      -0.4211430251598358,
+      -0.4722813367843628,
+      -0.44611743092536926,
+      -0.35652509331703186,
+      -0.2100401520729065,
+      -0.10723454505205154,
+      -0.6351080536842346,
+      -0.29903534054756165,
+      -0.2445092797279358,
+      -0.3422980010509491,
+      -0.3851262629032135,
+      -0.23470661044120789,
+      -0.41291457414627075,
+      -0.4065450429916382,
+      -0.23977336287498474,
+      -0.36426475644111633,
+      -0.3351646959781647,
+      -0.2730990946292877,
+      -0.0017077499069273472,
+      -0.30413487553596497,
+      -0.3375767767429352,
+      -0.2887400686740875,
+      -0.21333244442939758,
+      -0.21224288642406464,
+      -0.4361175298690796,
+      -0.24830438196659088,
+      -0.11157026886940002,
+      -0.31685569882392883,
+      -0.23533672094345093,
+      -0.1681990623474121,
+      -0.17895106971263885,
+      -0.392513245344162,
+      -0.11236104369163513,
+      -0.12460662424564362,
+      -0.48302754759788513,
+      -0.4136165678501129,
+      -0.3431640863418579,
+      -0.3221750855445862,
+      -0.45218420028686523,
+      -0.4912467896938324,
+      -0.11444079875946045,
+      -0.2064221352338791,
+      -0.3404044210910797,
+      -0.12942379713058472,
+      -0.21288804709911346,
+      -0.08454813063144684,
+      -0.29792511463165283,
+      -0.21220028400421143,
+      -0.3438176214694977,
+      -0.298564612865448,
+      -0.45913955569267273,
+      -0.272434264421463,
+      -0.25763818621635437,
+      -0.6365602016448975,
+      -0.18968147039413452,
+      -0.09780420362949371,
+      -0.21418723464012146,
+      -0.30045750737190247,
+      -0.2784423530101776,
+      -0.47075575590133667,
+      -0.370872437953949,
+      -0.3222501575946808,
+      -0.29639896750450134,
+      -0.34328600764274597,
+      -0.29898783564567566,
+      -0.3879045844078064,
+      -0.4125598967075348,
+      -0.3305641710758209,
+      -0.3634641468524933,
+      -0.27459052205085754,
+      -0.19963577389717102,
+      -0.346588671207428,
+      -0.08173392713069916,
+      -0.27201682329177856
+    ],
+    "base_loss": [
+      11.639817237854004,
+      10.156075477600098,
+      8.401082038879395,
+      6.93255090713501,
+      5.63682222366333,
+      3.969693422317505,
+      2.455012083053589,
+      2.118212938308716,
+      1.9375640153884888,
+      1.8131191730499268,
+      1.855855941772461,
+      1.7918169498443604,
+      1.862809419631958,
+      1.8385099172592163,
+      1.83856201171875,
+      1.853894591331482,
+      1.7929117679595947,
+      1.8418570756912231,
+      1.8591469526290894,
+      1.942344069480896,
+      1.9011590480804443,
+      1.937254548072815,
+      1.9145668745040894,
+      1.7746026515960693,
+      1.801316261291504,
+      1.713595986366272,
+      1.7272411584854126,
+      1.623053789138794,
+      1.6498748064041138,
+      1.6869145631790161,
+      1.6087714433670044,
+      1.676931619644165,
+      1.6311386823654175,
+      1.5374481678009033,
+      1.5154589414596558,
+      1.42965829372406,
+      1.4408766031265259,
+      1.4083964824676514,
+      1.3894157409667969,
+      1.280190348625183,
+      1.1854569911956787,
+      1.2155663967132568,
+      1.2250372171401978,
+      1.1413753032684326,
+      1.013464093208313,
+      1.122923493385315,
+      0.9275978803634644,
+      0.8905201554298401,
+      0.7307879328727722,
+      0.6575273871421814,
+      0.6883205771446228,
+      0.5470252633094788,
+      0.5368255376815796,
+      0.4907560348510742,
+      0.5686838030815125,
+      0.49868133664131165,
+      0.40930303931236267,
+      0.4845651090145111,
+      0.516866147518158,
+      0.5188473463058472,
+      0.5161632895469666,
+      0.44488078355789185,
+      0.4687444865703583,
+      0.4589233994483948,
+      0.44741424918174744,
+      0.43580299615859985,
+      0.4233454167842865,
+      0.37469175457954407,
+      0.47289425134658813,
+      0.40674108266830444,
+      0.41115209460258484,
+      0.3688190281391144,
+      0.3843062222003937,
+      0.4137001037597656,
+      0.36033517122268677,
+      0.4118814170360565,
+      0.38739117980003357,
+      0.36839795112609863,
+      0.40124839544296265,
+      0.38704535365104675,
+      0.3625943064689636,
+      0.32125577330589294,
+      0.39685899019241333,
+      0.35355910658836365,
+      0.3310209810733795,
+      0.3529447019100189,
+      0.3366256356239319,
+      0.351986825466156,
+      0.32832881808280945,
+      0.30364635586738586,
+      0.36219069361686707,
+      0.3217023015022278,
+      0.34354692697525024,
+      0.2791806757450104,
+      0.3173404037952423,
+      0.29693150520324707,
+      0.32531896233558655,
+      0.30371326208114624,
+      0.26866501569747925,
+      0.3006351590156555,
+      0.2808394730091095,
+      0.29236188530921936,
+      0.27687662839889526,
+      0.25318771600723267,
+      0.2781922221183777,
+      0.30388492345809937,
+      0.3194657862186432,
+      0.2906999886035919,
+      0.30635467171669006,
+      0.2871093451976776,
+      0.27299273014068604,
+      0.27276164293289185,
+      0.31095531582832336,
+      0.2815423309803009,
+      0.27550241351127625,
+      0.24704574048519135,
+      0.27616414427757263,
+      0.287104994058609,
+      0.26561808586120605,
+      0.26954320073127747,
+      0.23006193339824677,
+      0.25967758893966675,
+      0.21751165390014648,
+      0.22478653490543365,
+      0.23876747488975525,
+      0.2527664005756378,
+      0.24814987182617188,
+      0.23216032981872559,
+      0.24969851970672607,
+      0.21826933324337006,
+      0.24777519702911377,
+      0.21866890788078308,
+      0.24302545189857483,
+      0.23206384479999542,
+      0.1930166333913803,
+      0.2083585262298584,
+      0.21388663351535797,
+      0.2077399343252182,
+      0.202982559800148,
+      0.19327537715435028,
+      0.28431329131126404,
+      0.19369187951087952,
+      0.2327968329191208,
+      0.1640963852405548,
+      0.2110501080751419,
+      0.18979254364967346,
+      0.17834016680717468,
+      0.19346167147159576,
+      0.1911771297454834,
+      0.15178219974040985,
+      0.19927184283733368,
+      0.17405223846435547,
+      0.24049408733844757,
+      0.1855386644601822,
+      0.24259953200817108,
+      0.19974806904792786,
+      0.16892997920513153,
+      0.21057972311973572,
+      0.21029826998710632,
+      0.17126639187335968,
+      0.1980292946100235,
+      0.15987589955329895,
+      0.19345174729824066,
+      0.19131742417812347,
+      0.19851002097129822,
+      0.1876509189605713,
+      0.16012631356716156,
+      0.09607796370983124,
+      0.1484193354845047,
+      0.18884345889091492,
+      0.16871722042560577,
+      0.16881194710731506,
+      0.1550588309764862,
+      0.16616418957710266,
+      0.17493103444576263,
+      0.1629718840122223,
+      0.19975315034389496,
+      0.16586120426654816,
+      0.15977083146572113,
+      0.15066365897655487,
+      0.18965628743171692,
+      0.185601606965065,
+      0.15749408304691315,
+      0.19439183175563812,
+      0.14395713806152344,
+      0.1731584519147873,
+      0.17609839141368866,
+      0.10343510657548904,
+      0.2179812639951706,
+      0.1640244722366333,
+      0.17557813227176666,
+      0.14700005948543549,
+      0.12347990274429321,
+      0.16161899268627167,
+      0.16377146542072296,
+      0.1840793341398239,
+      0.16619475185871124,
+      0.12906678020954132,
+      0.1411665976047516,
+      0.12866570055484772,
+      0.1545630246400833,
+      0.12623055279254913,
+      0.1273307204246521,
+      0.17082515358924866,
+      0.12737132608890533,
+      0.1601167768239975,
+      0.12357539683580399,
+      0.1447925567626953,
+      0.1613645851612091,
+      0.1653471291065216,
+      0.14804492890834808,
+      0.1280674785375595,
+      0.120961032807827,
+      0.11894188821315765,
+      0.17771609127521515,
+      0.16308431327342987,
+      0.1615145057439804,
+      0.16900038719177246,
+      0.13064035773277283,
+      0.12192203849554062,
+      0.1703270524740219,
+      0.13478057086467743,
+      0.11193303018808365,
+      0.1735823005437851,
+      0.11142634600400925,
+      0.12079431116580963,
+      0.13884304463863373,
+      0.13051298260688782,
+      0.1113380566239357,
+      0.13207702338695526,
+      0.11978454142808914,
+      0.11676950007677078,
+      0.11669310182332993,
+      0.1455826461315155,
+      0.14255119860172272,
+      0.09344311058521271,
+      0.13233305513858795,
+      0.14260323345661163,
+      0.14308317005634308,
+      0.12413915246725082,
+      0.13392850756645203,
+      0.10997339338064194,
+      0.16214631497859955,
+      0.10885850340127945,
+      0.11581401526927948,
+      0.13958588242530823,
+      0.12495311349630356,
+      0.12036968767642975,
+      0.1327098160982132,
+      0.16099299490451813,
+      0.12836897373199463,
+      0.09642153978347778,
+      0.1325569599866867,
+      0.09756525605916977,
+      0.07441246509552002,
+      0.15825296938419342,
+      0.11627190560102463,
+      0.11572710424661636,
+      0.12323235720396042,
+      0.09537728130817413,
+      0.1255684643983841,
+      0.10272663086652756,
+      0.10061192512512207,
+      0.1096343845129013,
+      0.13086198270320892,
+      0.13289330899715424,
+      0.0884971171617508,
+      0.13670191168785095,
+      0.10128708928823471,
+      0.09838897734880447,
+      0.11107941716909409,
+      0.11279519647359848,
+      0.13107289373874664,
+      0.08550458401441574,
+      0.13277259469032288,
+      0.1271776407957077,
+      0.09930078685283661,
+      0.12097972631454468,
+      0.11515770852565765,
+      0.09048484265804291,
+      0.11554217338562012,
+      0.11730344593524933,
+      0.10536686331033707,
+      0.10421247035264969,
+      0.0897669643163681,
+      0.14295995235443115,
+      0.14636261761188507,
+      0.0999404788017273,
+      0.09715478122234344,
+      0.14027845859527588,
+      0.09507667273283005,
+      0.09038203209638596,
+      0.09006116539239883,
+      0.10859178006649017,
+      0.08985663950443268,
+      0.1452604979276657,
+      0.09389884769916534,
+      0.09381306916475296,
+      0.09161395579576492,
+      0.10640876740217209,
+      0.09798900038003922,
+      0.08652211725711823,
+      0.09993406385183334,
+      0.09884941577911377,
+      0.11107803136110306,
+      0.09744322299957275,
+      0.08755477517843246,
+      0.10821057856082916,
+      0.10122448951005936,
+      0.10775398463010788
+    ],
+    "info_loss": [
+      -2.0815982818603516,
+      -2.243824005126953,
+      -0.9210786819458008,
+      -0.34617090225219727,
+      -0.19792890548706055,
+      -0.09876537322998047,
+      -0.09591412544250488,
+      -0.05676698684692383,
+      -0.06754577159881592,
+      -0.07223129272460938,
+      -0.06728148460388184,
+      -0.05278456211090088,
+      -0.05221426486968994,
+      -0.08916306495666504,
+      -0.055210232734680176,
+      -0.06883823871612549,
+      -0.05282747745513916,
+      -0.06078696250915527,
+      -0.26839470863342285,
+      -0.5799283981323242,
+      -0.6877650022506714,
+      -0.8812437057495117,
+      -1.0872838497161865,
+      -1.1079578399658203,
+      -1.168687343597412,
+      -1.114762783050537,
+      -1.174454927444458,
+      -1.0983941555023193,
+      -1.1433277130126953,
+      -1.1475012302398682,
+      -1.1599960327148438,
+      -1.1651790142059326,
+      -1.1718239784240723,
+      -1.0791484117507935,
+      -1.0615819692611694,
+      -0.9530861973762512,
+      -0.988301157951355,
+      -0.9596990942955017,
+      -0.9770992994308472,
+      -0.8317760229110718,
+      -0.7829140424728394,
+      -0.7892236709594727,
+      -0.8490042090415955,
+      -0.7429447174072266,
+      -0.6338351964950562,
+      -0.7489489316940308,
+      -0.5262289047241211,
+      -0.4670906364917755,
+      -0.3774372935295105,
+      -0.29696008563041687,
+      -0.29091131687164307,
+      -0.1927202045917511,
+      -0.17022168636322021,
+      -0.12140163779258728,
+      -0.1601722538471222,
+      -0.1399734914302826,
+      -0.0905258059501648,
+      -0.13134565949440002,
+      -0.16920897364616394,
+      -0.14705651998519897,
+      -0.1521684229373932,
+      -0.1504981517791748,
+      -0.1293189823627472,
+      -0.13350403308868408,
+      -0.15035784244537354,
+      -0.16198059916496277,
+      -0.11993435025215149,
+      -0.08966812491416931,
+      -0.15014863014221191,
+      -0.1358177661895752,
+      -0.13599982857704163,
+      -0.07511934638023376,
+      -0.0941607654094696,
+      -0.11929652094841003,
+      -0.1127258688211441,
+      -0.14379215240478516,
+      -0.1386413723230362,
+      -0.09477558732032776,
+      -0.1377323567867279,
+      -0.0781242847442627,
+      -0.14930512011051178,
+      -0.12259714305400848,
+      -0.10278162360191345,
+      -0.08865389227867126,
+      -0.12277501821517944,
+      -0.11761446297168732,
+      -0.12304633855819702,
+      -0.13126574456691742,
+      -0.13275323808193207,
+      -0.09568502008914948,
+      -0.11185088753700256,
+      -0.0839833915233612,
+      -0.11324603855609894,
+      -0.11538231372833252,
+      -0.12823568284511566,
+      -0.10232169926166534,
+      -0.12100198864936829,
+      -0.09690612554550171,
+      -0.08991304039955139,
+      -0.09477418661117554,
+      -0.11962383985519409,
+      -0.08371704816818237,
+      -0.13328717648983002,
+      -0.077918142080307,
+      -0.12280711531639099,
+      -0.09519733488559723,
+      -0.10253576934337616,
+      -0.12251155078411102,
+      -0.12948094308376312,
+      -0.1147230863571167,
+      -0.10420532524585724,
+      -0.1289534866809845,
+      -0.12759214639663696,
+      -0.1027921587228775,
+      -0.11141075193881989,
+      -0.09078088402748108,
+      -0.1241578608751297,
+      -0.09824222326278687,
+      -0.09674803912639618,
+      -0.10662780702114105,
+      -0.061787575483322144,
+      -0.10516685247421265,
+      -0.08466640114784241,
+      -0.08995772898197174,
+      -0.09759549796581268,
+      -0.08332888782024384,
+      -0.07753999531269073,
+      -0.0749911367893219,
+      -0.11360335350036621,
+      -0.09048980474472046,
+      -0.03536655008792877,
+      -0.08909980952739716,
+      -0.09623746573925018,
+      -0.07525226473808289,
+      -0.06714370846748352,
+      -0.0715549886226654,
+      -0.09189903736114502,
+      -0.06325015425682068,
+      -0.05585753917694092,
+      -0.07554758340120316,
+      -0.1361466348171234,
+      -0.06977575272321701,
+      -0.07141049206256866,
+      -0.08251591771841049,
+      -0.056324779987335205,
+      -0.06530772894620895,
+      -0.06673833727836609,
+      -0.0813555121421814,
+      -0.062400490045547485,
+      -0.06523990631103516,
+      -0.08259081840515137,
+      -0.05174391716718674,
+      -0.08779358863830566,
+      -0.04333066940307617,
+      -0.10577897727489471,
+      -0.0604940801858902,
+      -0.02981606125831604,
+      -0.06534035503864288,
+      -0.07527513802051544,
+      -0.06854487955570221,
+      -0.006333336234092712,
+      -0.05264779180288315,
+      -0.0847029909491539,
+      -0.056125909090042114,
+      -0.08433207124471664,
+      -0.08440694957971573,
+      -0.03478521108627319,
+      -0.02798573672771454,
+      -0.06543087959289551,
+      -0.06605537980794907,
+      -0.08063378185033798,
+      -0.05501732975244522,
+      -0.05552982538938522,
+      -0.07784626632928848,
+      -0.0568971261382103,
+      -0.056768909096717834,
+      -0.09499359130859375,
+      -0.08094023913145065,
+      -0.0358768105506897,
+      -0.02388089895248413,
+      -0.0657883808016777,
+      -0.04160216450691223,
+      -0.041917286813259125,
+      -0.057980477809906006,
+      -0.05439015477895737,
+      -0.06195651739835739,
+      -0.046708136796951294,
+      -0.03449763357639313,
+      -0.05678156018257141,
+      -0.05389215052127838,
+      -0.05589033663272858,
+      -0.03746853023767471,
+      -0.0339425653219223,
+      -0.03663020581007004,
+      -0.04093632847070694,
+      -0.05015619099140167,
+      -0.048770755529403687,
+      -0.05778995901346207,
+      -0.0505993515253067,
+      -0.04343957453966141,
+      -0.057661525905132294,
+      -0.037449926137924194,
+      -0.0205063596367836,
+      -0.031216025352478027,
+      -0.02424817532300949,
+      -0.05868971347808838,
+      -0.05216154456138611,
+      -0.04424349218606949,
+      -0.03854592144489288,
+      -0.04582716524600983,
+      -0.05157455801963806,
+      -0.053762584924697876,
+      -0.029237374663352966,
+      -0.04789608716964722,
+      -0.0828666165471077,
+      -0.06326963752508163,
+      -0.040518492460250854,
+      -0.0552607923746109,
+      -0.0655718743801117,
+      -0.036567673087120056,
+      -0.0821944922208786,
+      -0.05331090837717056,
+      -0.06115570664405823,
+      -0.06203245371580124,
+      -0.012450620532035828,
+      -0.03228802978992462,
+      -0.050903528928756714,
+      -0.02402219921350479,
+      -0.041777729988098145,
+      -0.059955090284347534,
+      -0.01389988511800766,
+      -0.0007209256291389465,
+      -0.020330257713794708,
+      -0.05560953915119171,
+      -0.05655915290117264,
+      -0.03257487341761589,
+      -0.0562567338347435,
+      -0.06271617859601974,
+      -0.0597454234957695,
+      -0.04924790561199188,
+      -0.035283736884593964,
+      -0.02310994267463684,
+      -0.0805332139134407,
+      -0.04172522574663162,
+      -0.03687093406915665,
+      -0.04889316111803055,
+      -0.051736071705818176,
+      -0.03652358800172806,
+      -0.05550236254930496,
+      -0.05738862603902817,
+      -0.03773617744445801,
+      -0.04731643199920654,
+      -0.04769770801067352,
+      -0.038139909505844116,
+      -0.008212149143218994,
+      -0.04735717922449112,
+      -0.04601004719734192,
+      -0.041242122650146484,
+      -0.03444558382034302,
+      -0.03151874989271164,
+      -0.05710924416780472,
+      -0.0362548902630806,
+      -0.022099576890468597,
+      -0.04331859201192856,
+      -0.03729948401451111,
+      -0.03066999465227127,
+      -0.027804844081401825,
+      -0.053693242371082306,
+      -0.02257673442363739,
+      -0.02293287217617035,
+      -0.06020914018154144,
+      -0.053687889128923416,
+      -0.048328571021556854,
+      -0.041346874088048935,
+      -0.05953928828239441,
+      -0.06348404288291931,
+      -0.021980077028274536,
+      -0.033464111387729645,
+      -0.04626680165529251,
+      -0.02295789122581482,
+      -0.03362368792295456,
+      -0.020935319364070892,
+      -0.04133973270654678,
+      -0.03256850689649582,
+      -0.04415736347436905,
+      -0.04528270661830902,
+      -0.06118548661470413,
+      -0.03850490599870682,
+      -0.03625042364001274,
+      -0.07827885448932648,
+      -0.029301680624485016,
+      -0.019341766834259033,
+      -0.03149646893143654,
+      -0.04163607209920883,
+      -0.038239289075136185,
+      -0.062306568026542664,
+      -0.04730871692299843,
+      -0.04222497716546059,
+      -0.03953922539949417,
+      -0.045936308801174164,
+      -0.04106493666768074,
+      -0.04822247847914696,
+      -0.05199488252401352,
+      -0.044000688940286636,
+      -0.04840981960296631,
+      -0.03808194771409035,
+      -0.02965845912694931,
+      -0.04685147851705551,
+      -0.0203888937830925,
+      -0.03871728479862213
+    ],
+    "abs_loss": [
+      2.30385422706604,
+      2.2780826091766357,
+      2.221466541290283,
+      2.089055061340332,
+      1.9984480142593384,
+      1.9005656242370605,
+      1.87179434299469,
+      1.8566112518310547,
+      1.8687645196914673,
+      1.8529582023620605,
+      1.8454004526138306,
+      1.8316807746887207,
+      1.8391278982162476,
+      1.8252058029174805,
+      1.8663090467453003,
+      1.8276523351669312,
+      1.8539466857910156,
+      1.8391097784042358,
+      1.8551603555679321,
+      1.8367284536361694,
+      1.7923775911331177,
+      1.7368284463882446,
+      1.6195597648620605,
+      1.49187433719635,
+      1.4642229080200195,
+      1.3621273040771484,
+      1.3119254112243652,
+      1.3029338121414185,
+      1.2246941328048706,
+      1.1458531618118286,
+      1.1349433660507202,
+      1.0280259847640991,
+      0.9581847786903381,
+      0.9767263531684875,
+      0.9638102650642395,
+      0.8906691670417786,
+      0.8996453285217285,
+      0.8511548042297363,
+      0.9369351267814636,
+      0.8706004619598389,
+      0.8490410447120667,
+      0.7745605111122131,
+      0.7039835453033447,
+      0.5803566575050354,
+      0.530763566493988,
+      0.5496122241020203,
+      0.5036328434944153,
+      0.5069024562835693,
+      0.5179614424705505,
+      0.49070897698402405,
+      0.5103476643562317,
+      0.43010661005973816,
+      0.43826374411582947,
+      0.35059139132499695,
+      0.35389336943626404,
+      0.3348320722579956,
+      0.3144708573818207,
+      0.25914695858955383,
+      0.2997690439224243,
+      0.2784518897533417,
+      0.2622879445552826,
+      0.26106083393096924,
+      0.24241697788238525,
+      0.2352316528558731,
+      0.2446093112230301,
+      0.2542091906070709,
+      0.2651433050632477,
+      0.21855776011943817,
+      0.24479158222675323,
+      0.24396425485610962,
+      0.22532515227794647,
+      0.20379044115543365,
+      0.18716053664684296,
+      0.19765444099903107,
+      0.21834973990917206,
+      0.22507119178771973,
+      0.209799125790596,
+      0.241669163107872,
+      0.20123811066150665,
+      0.21422863006591797,
+      0.2206764668226242,
+      0.24827057123184204,
+      0.25371381640434265,
+      0.17112255096435547,
+      0.21741031110286713,
+      0.23502689599990845,
+      0.2338194102048874,
+      0.18143780529499054,
+      0.20621494948863983,
+      0.18330572545528412,
+      0.17663991451263428,
+      0.21785050630569458,
+      0.1840270310640335,
+      0.24093495309352875,
+      0.19969721138477325,
+      0.1844915896654129,
+      0.1969558745622635,
+      0.15707360208034515,
+      0.16628654301166534,
+      0.21024517714977264,
+      0.18380267918109894,
+      0.21588551998138428,
+      0.15520580112934113,
+      0.1718127578496933,
+      0.15742404758930206,
+      0.14652536809444427,
+      0.17036563158035278,
+      0.16253049671649933,
+      0.12149914354085922,
+      0.14606857299804688,
+      0.1554645150899887,
+      0.1270408183336258,
+      0.12375888973474503,
+      0.16692711412906647,
+      0.12466469407081604,
+      0.13749970495700836,
+      0.1462293118238449,
+      0.13946682214736938,
+      0.13414506614208221,
+      0.14158709347248077,
+      0.11992070078849792,
+      0.13205362856388092,
+      0.10871059447526932,
+      0.14603517949581146,
+      0.12842875719070435,
+      0.10888461023569107,
+      0.13081012666225433,
+      0.1278458833694458,
+      0.1344360113143921,
+      0.1167023554444313,
+      0.1376543492078781,
+      0.11900731176137924,
+      0.1396193951368332,
+      0.1632918417453766,
+      0.10951266437768936,
+      0.1426679641008377,
+      0.12421008199453354,
+      0.09802278876304626,
+      0.12145048379898071,
+      0.13719113171100616,
+      0.1666850596666336,
+      0.1143966093659401,
+      0.13817617297172546,
+      0.12770074605941772,
+      0.1320442110300064,
+      0.156675323843956,
+      0.12263376265764236,
+      0.11993134021759033,
+      0.11233779788017273,
+      0.10352327674627304,
+      0.10694246739149094,
+      0.0978497639298439,
+      0.1038326844573021,
+      0.1087891086935997,
+      0.12045059353113174,
+      0.10178189724683762,
+      0.10108641535043716,
+      0.10563056915998459,
+      0.09969466924667358,
+      0.08930913358926773,
+      0.09622367471456528,
+      0.08112730830907822,
+      0.09679283946752548,
+      0.09852912276983261,
+      0.10154712945222855,
+      0.08536344021558762,
+      0.10110368579626083,
+      0.0877099558711052,
+      0.0935603603720665,
+      0.07935286313295364,
+      0.07173427194356918,
+      0.06842287629842758,
+      0.07871238887310028,
+      0.09305604547262192,
+      0.061133552342653275,
+      0.06750581413507462,
+      0.04733775556087494,
+      0.041590336710214615,
+      0.03440967574715614,
+      0.0448673851788044,
+      0.048636067658662796,
+      0.058997560292482376,
+      0.05349382758140564,
+      0.06911490857601166,
+      0.057384829968214035,
+      0.06524454802274704,
+      0.03691668435931206,
+      0.04632076621055603,
+      0.044899653643369675,
+      0.04991862550377846,
+      0.03430148586630821,
+      0.06102760508656502,
+      0.059088896960020065,
+      0.05033651366829872,
+      0.03766924515366554,
+      0.04401188716292381,
+      0.0495685450732708,
+      0.04394489526748657,
+      0.032825104892253876,
+      0.03533637151122093,
+      0.036821670830249786,
+      0.03871423378586769,
+      0.050813909620046616,
+      0.0320887565612793,
+      0.0323503203690052,
+      0.04162690415978432,
+      0.04652707651257515,
+      0.03234953060746193,
+      0.02907419763505459,
+      0.04330940172076225,
+      0.03034401126205921,
+      0.03533114865422249,
+      0.044240038841962814,
+      0.03451972082257271,
+      0.03138953447341919,
+      0.02100559137761593,
+      0.03484845533967018,
+      0.027008062228560448,
+      0.03034891188144684,
+      0.031032299622893333,
+      0.031608667224645615,
+      0.027965238317847252,
+      0.027770444750785828,
+      0.04124467447400093,
+      0.046497613191604614,
+      0.023359201848506927,
+      0.032218821346759796,
+      0.023809067904949188,
+      0.03547733649611473,
+      0.017117025330662727,
+      0.02833031676709652,
+      0.028712743893265724,
+      0.03794991597533226,
+      0.029193580150604248,
+      0.017572076991200447,
+      0.02578355371952057,
+      0.03570059686899185,
+      0.0297600906342268,
+      0.024734100326895714,
+      0.022516677156090736,
+      0.03747735917568207,
+      0.0201548021286726,
+      0.03178713098168373,
+      0.024044664576649666,
+      0.030839068815112114,
+      0.02763315476477146,
+      0.02342303656041622,
+      0.02100348472595215,
+      0.032144565135240555,
+      0.020648693665862083,
+      0.011358563788235188,
+      0.02996905706822872,
+      0.023514235392212868,
+      0.02822832763195038,
+      0.02316158078610897,
+      0.02231030911207199,
+      0.019833313301205635,
+      0.021736932918429375,
+      0.04444855824112892,
+      0.015920737758278847,
+      0.01798548549413681,
+      0.024343930184841156,
+      0.023111948743462563,
+      0.02028748020529747,
+      0.013007582165300846,
+      0.018035665154457092,
+      0.008614753372967243,
+      0.02372141182422638,
+      0.036382295191287994,
+      0.012034427374601364,
+      0.024076899513602257,
+      0.026708101853728294,
+      0.039029981940984726,
+      0.012135415337979794,
+      0.021434441208839417,
+      0.03262598067522049,
+      0.02024141512811184,
+      0.020032141357660294,
+      0.026833519339561462,
+      0.025756537914276123,
+      0.02003246732056141,
+      0.007906577549874783,
+      0.011433945037424564,
+      0.028460770845413208,
+      0.012638579122722149,
+      0.027600837871432304,
+      0.022938931360840797,
+      0.027256881818175316,
+      0.03210896998643875,
+      0.028382470831274986,
+      0.027644798159599304,
+      0.014684910885989666,
+      0.0210502240806818,
+      0.022317929193377495,
+      0.019239502027630806,
+      0.018058478832244873,
+      0.031683724373579025,
+      0.01892039179801941,
+      0.016745159402489662,
+      0.019800638779997826,
+      0.018956510350108147,
+      0.0198680330067873,
+      0.023575974628329277,
+      0.022173406556248665,
+      0.015582915395498276,
+      0.024224141612648964,
+      0.018571188673377037,
+      0.01707600988447666,
+      0.02319365181028843,
+      0.016697650775313377
+    ],
+    "zipf_loss": [
+      9.148764610290527,
+      8.193159103393555,
+      6.072509765625,
+      3.9484927654266357,
+      2.6294360160827637,
+      2.011842966079712,
+      1.7981233596801758,
+      1.7298715114593506,
+      1.7061980962753296,
+      1.6984326839447021,
+      1.6932156085968018,
+      1.6903259754180908,
+      1.6886255741119385,
+      1.6878618001937866,
+      1.686628818511963,
+      1.6883533000946045,
+      1.6861824989318848,
+      1.6868352890014648,
+      1.684238076210022,
+      1.6843323707580566,
+      1.681557297706604,
+      1.6743335723876953,
+      1.6612993478775024,
+      1.6309456825256348,
+      1.5862919092178345,
+      1.5408496856689453,
+      1.4890775680541992,
+      1.4552651643753052,
+      1.3806521892547607,
+      1.3135970830917358,
+      1.2469334602355957,
+      1.1696181297302246,
+      1.0805621147155762,
+      1.0572535991668701,
+      1.0407004356384277,
+      1.0172886848449707,
+      0.9681190848350525,
+      0.8774245977401733,
+      0.8218495845794678,
+      0.7741572260856628,
+      0.6215071082115173,
+      0.4680870771408081,
+      0.38734686374664307,
+      0.30501338839530945,
+      0.2802583575248718,
+      0.276706337928772,
+      0.24773532152175903,
+      0.22969545423984528,
+      0.21012923121452332,
+      0.20455025136470795,
+      0.18192313611507416,
+      0.15067628026008606,
+      0.11456912010908127,
+      0.09580022096633911,
+      0.0668957531452179,
+      0.06967450678348541,
+      0.060803525149822235,
+      0.06049271300435066,
+      0.055438973009586334,
+      0.06107698380947113,
+      0.044147878885269165,
+      0.04942154884338379,
+      0.054558660835027695,
+      0.04915769770741463,
+      0.0500359907746315,
+      0.05347098037600517,
+      0.049553126096725464,
+      0.05055416002869606,
+      0.04518905282020569,
+      0.038480307906866074,
+      0.028954092413187027,
+      0.025756293907761574,
+      0.029039651155471802,
+      0.021393898874521255,
+      0.023127539083361626,
+      0.018351595848798752,
+      0.02461368776857853,
+      0.026984255760908127,
+      0.02383498102426529,
+      0.02516784705221653,
+      0.020445672795176506,
+      0.0184946171939373,
+      0.01568470150232315,
+      0.0335945263504982,
+      0.017658868804574013,
+      0.015548460185527802,
+      0.016619864851236343,
+      0.010704641230404377,
+      0.028467528522014618,
+      0.025375287979841232,
+      0.01649937778711319,
+      0.012676859274506569,
+      0.01538829691708088,
+      0.011839807033538818,
+      0.02237042970955372,
+      0.014730386435985565,
+      0.011545787565410137,
+      0.015047524124383926,
+      0.014527767896652222,
+      0.012202445417642593,
+      0.02355729602277279,
+      0.02182786725461483,
+      0.017060475423932076,
+      0.011250884272158146,
+      0.015658840537071228,
+      0.015689896419644356,
+      0.009464162401854992,
+      0.007833020761609077,
+      0.013375839218497276,
+      0.015904515981674194,
+      0.00788052100688219,
+      0.015139850787818432,
+      0.019742511212825775,
+      0.010081540793180466,
+      0.006598178297281265,
+      0.02656419202685356,
+      0.006301208399236202,
+      0.01116674579679966,
+      0.011517665348947048,
+      0.012123423628509045,
+      0.015418392606079578,
+      0.010212818160653114,
+      0.02571723610162735,
+      0.005934752523899078,
+      0.011283399537205696,
+      0.024015210568904877,
+      0.006147728767246008,
+      0.008858778513967991,
+      0.010002169758081436,
+      0.0111832981929183,
+      0.008553914725780487,
+      0.012515537440776825,
+      0.008607814088463783,
+      0.0060794176533818245,
+      0.013789001852273941,
+      0.007285661995410919,
+      0.007553542964160442,
+      0.007395597640424967,
+      0.019424868747591972,
+      0.007891308516263962,
+      0.013147450983524323,
+      0.006194372661411762,
+      0.013805458322167397,
+      0.008373916149139404,
+      0.005300499498844147,
+      0.006729935761541128,
+      0.007942927069962025,
+      0.00961179006844759,
+      0.011826831847429276,
+      0.012580672279000282,
+      0.011939752846956253,
+      0.009729165583848953,
+      0.011491796933114529,
+      0.008557336404919624,
+      0.006641690619289875,
+      0.01521589420735836,
+      0.018641222268342972,
+      0.014188352972269058,
+      0.004948650486767292,
+      0.011573015712201595,
+      0.00892910361289978,
+      0.011337338015437126,
+      0.004030641634017229,
+      0.014527570456266403,
+      0.0057725682854652405,
+      0.008571498095989227,
+      0.008519389666616917,
+      0.010961297899484634,
+      0.005612564273178577,
+      0.003417621599510312,
+      0.004866757430136204,
+      0.007937588728964329,
+      0.006038891151547432,
+      0.006155673414468765,
+      0.007668308913707733,
+      0.006662166677415371,
+      0.010678012855350971,
+      0.008212166838347912,
+      0.009509880095720291,
+      0.007193836383521557,
+      0.003942061215639114,
+      0.003548411652445793,
+      0.004174727946519852,
+      0.005232849158346653,
+      0.00789184682071209,
+      0.009482812136411667,
+      0.008240800350904465,
+      0.005656293593347073,
+      0.008366700261831284,
+      0.008680871687829494,
+      0.005021273624151945,
+      0.009234117344021797,
+      0.009519586339592934,
+      0.00818641111254692,
+      0.0075712306424975395,
+      0.012455126270651817,
+      0.007254922296851873,
+      0.005930674262344837,
+      0.0052425069734454155,
+      0.0026599280536174774,
+      0.0042840042151510715,
+      0.004441949538886547,
+      0.006021777167916298,
+      0.002922310493886471,
+      0.009024055674672127,
+      0.0037710736505687237,
+      0.006683184765279293,
+      0.0072190845385193825,
+      0.004481279291212559,
+      0.003004686674103141,
+      0.004501868039369583,
+      0.0038021784275770187,
+      0.0027652522549033165,
+      0.006458859890699387,
+      0.005650538019835949,
+      0.0054639726877212524,
+      0.007049696519970894,
+      0.013013076037168503,
+      0.004686283878982067,
+      0.011087823659181595,
+      0.004652542062103748,
+      0.004659920930862427,
+      0.00613184692338109,
+      0.006299192551523447,
+      0.005826734937727451,
+      0.005277771968394518,
+      0.0111530851572752,
+      0.010877054184675217,
+      0.003169892355799675,
+      0.005685212090611458,
+      0.005942808464169502,
+      0.006910073570907116,
+      0.007615677081048489,
+      0.010119684040546417,
+      0.007940967567265034,
+      0.007782436441630125,
+      0.005521237850189209,
+      0.009301232174038887,
+      0.005780254025012255,
+      0.009563151746988297,
+      0.005120998248457909,
+      0.011876009404659271,
+      0.004899022169411182,
+      0.00695391558110714,
+      0.005302120000123978,
+      0.0042844172567129135,
+      0.004938999190926552,
+      0.008059222251176834,
+      0.006184779107570648,
+      0.0042833685874938965,
+      0.008083581924438477,
+      0.009481126442551613,
+      0.0069039976224303246,
+      0.007911891676485538,
+      0.0036851181648671627,
+      0.008952926844358444,
+      0.00426842924207449,
+      0.00578036205843091,
+      0.0034461834002286196,
+      0.005975274369120598,
+      0.007607896812260151,
+      0.0090834591537714,
+      0.006502381060272455,
+      0.004667104221880436,
+      0.005495388060808182,
+      0.0038040021900087595,
+      0.009738792665302753,
+      0.005345105193555355,
+      0.008480982854962349,
+      0.005129683762788773,
+      0.00557674840092659,
+      0.007796278223395348,
+      0.00514574721455574,
+      0.0045755114406347275,
+      0.00829264335334301,
+      0.013153402134776115,
+      0.0040350379422307014,
+      0.005236027296632528,
+      0.004422519356012344,
+      0.007094629108905792,
+      0.0058034141547977924,
+      0.006710957735776901,
+      0.008961949497461319,
+      0.006426266394555569,
+      0.006725172512233257,
+      0.008542411029338837,
+      0.0040588038973510265,
+      0.009948624297976494,
+      0.004500346723943949,
+      0.003111586906015873,
+      0.005494164768606424,
+      0.0037629390135407448,
+      0.008611260913312435,
+      0.0050796265713870525,
+      0.012169968336820602,
+      0.0052435919642448425,
+      0.005147482268512249,
+      0.00429452583193779,
+      0.005704828538000584,
+      0.007688248064368963,
+      0.011776888743042946,
+      0.005811269395053387,
+      0.005097261629998684,
+      0.008375940844416618,
+      0.007997730746865273,
+      0.0063632940873503685,
+      0.007536925375461578,
+      0.012007943354547024,
+      0.018611159175634384,
+      0.005732273682951927
+    ],
+    "denoise_loss": [],
+    "ortho_loss": [
+      0.8885435461997986,
+      0.7170970439910889,
+      0.5767908692359924,
+      0.4600399434566498,
+      0.43550196290016174,
+      0.41573062539100647,
+      0.34819266200065613,
+      0.2961730659008026,
+      0.2692863643169403,
+      0.28603875637054443,
+      0.25260230898857117,
+      0.2519712746143341,
+      0.25865286588668823,
+      0.25377777218818665,
+      0.2461307942867279,
+      0.22820307314395905,
+      0.21362219750881195,
+      0.22701697051525116,
+      0.2701369524002075,
+      0.3389910161495209,
+      0.34586986899375916,
+      0.36707010865211487,
+      0.3889461159706116,
+      0.3966933488845825,
+      0.39621293544769287,
+      0.3982795476913452,
+      0.3988940119743347,
+      0.40131956338882446,
+      0.40211835503578186,
+      0.40279901027679443,
+      0.40056779980659485,
+      0.4035308063030243,
+      0.40373891592025757,
+      0.40222078561782837,
+      0.39735692739486694,
+      0.39471435546875,
+      0.3861656188964844,
+      0.3764215409755707,
+      0.37386637926101685,
+      0.37017256021499634,
+      0.3712293803691864,
+      0.371606707572937,
+      0.35931164026260376,
+      0.3534904718399048,
+      0.3533676266670227,
+      0.35187551379203796,
+      0.35193905234336853,
+      0.34886762499809265,
+      0.34502607583999634,
+      0.3453461825847626,
+      0.3458319306373596,
+      0.33560043573379517,
+      0.3396965563297272,
+      0.33749768137931824,
+      0.3369745910167694,
+      0.3322361409664154,
+      0.3279988467693329,
+      0.32231560349464417,
+      0.3140047788619995,
+      0.3148951828479767,
+      0.30706074833869934,
+      0.3004221022129059,
+      0.2957034409046173,
+      0.291050523519516,
+      0.28938552737236023,
+      0.28469592332839966,
+      0.2833675444126129,
+      0.27923813462257385,
+      0.2769160866737366,
+      0.27769383788108826,
+      0.27265939116477966,
+      0.27549681067466736,
+      0.26848623156547546,
+      0.27251937985420227,
+      0.2696530520915985,
+      0.2705014646053314,
+      0.26794448494911194,
+      0.26621025800704956,
+      0.26695698499679565,
+      0.2677346467971802,
+      0.2682175934314728,
+      0.26711276173591614,
+      0.2629030644893646,
+      0.26805412769317627,
+      0.26911965012550354,
+      0.27012115716934204,
+      0.27012962102890015,
+      0.26948150992393494,
+      0.2718905210494995,
+      0.2693946361541748,
+      0.2690868377685547,
+      0.26841506361961365,
+      0.2688615322113037,
+      0.26440539956092834,
+      0.2661382257938385,
+      0.26457643508911133,
+      0.2629278898239136,
+      0.2636287212371826,
+      0.26622527837753296,
+      0.26416337490081787,
+      0.2650313973426819,
+      0.2698652446269989,
+      0.26900938153266907,
+      0.2571031451225281,
+      0.2569356858730316,
+      0.2564140260219574,
+      0.2546826899051666,
+      0.2549506425857544,
+      0.2541828751564026,
+      0.25547727942466736,
+      0.2605178952217102,
+      0.26146799325942993,
+      0.25865280628204346,
+      0.2535390853881836,
+      0.24878275394439697,
+      0.2532852292060852,
+      0.25691404938697815,
+      0.25607773661613464,
+      0.2578878700733185,
+      0.2600436508655548,
+      0.2597696781158447,
+      0.25679489970207214,
+      0.25775811076164246,
+      0.2537018656730652,
+      0.25513455271720886,
+      0.25609615445137024,
+      0.2565081715583801,
+      0.2563503384590149,
+      0.25798508524894714,
+      0.25841814279556274,
+      0.2571648061275482,
+      0.2571255564689636,
+      0.2550588846206665,
+      0.25339433550834656,
+      0.2531038522720337,
+      0.2531052231788635,
+      0.2499697357416153,
+      0.2552987337112427,
+      0.2573709487915039,
+      0.25619930028915405,
+      0.24927644431591034,
+      0.2494434416294098,
+      0.2452142983675003,
+      0.24594520032405853,
+      0.24564315378665924,
+      0.24543415009975433,
+      0.2530146837234497,
+      0.255063533782959,
+      0.2588399350643158,
+      0.25854992866516113,
+      0.25709789991378784,
+      0.2601565420627594,
+      0.2589494585990906,
+      0.2594975531101227,
+      0.2603684663772583,
+      0.25523388385772705,
+      0.2566591501235962,
+      0.2613082528114319,
+      0.26063770055770874,
+      0.2606363296508789,
+      0.25953832268714905,
+      0.2609032988548279,
+      0.26214295625686646,
+      0.2610551118850708,
+      0.25964516401290894,
+      0.25923821330070496,
+      0.25974103808403015,
+      0.2597079873085022,
+      0.2560349404811859,
+      0.25539258122444153,
+      0.2563842535018921,
+      0.25771430134773254,
+      0.25725558400154114,
+      0.25808581709861755,
+      0.2630041241645813,
+      0.268037885427475,
+      0.26821357011795044,
+      0.26748761534690857,
+      0.2678012549877167,
+      0.2589096128940582,
+      0.26054227352142334,
+      0.2608339488506317,
+      0.2611874043941498,
+      0.2599111497402191,
+      0.25688591599464417,
+      0.26534929871559143,
+      0.2643160820007324,
+      0.2675628364086151,
+      0.2699528932571411,
+      0.2705341577529907,
+      0.2709191143512726,
+      0.2723911702632904,
+      0.2701484262943268,
+      0.2669573128223419,
+      0.266337513923645,
+      0.26788681745529175,
+      0.26726365089416504,
+      0.26725485920906067,
+      0.26754555106163025,
+      0.2686413824558258,
+      0.2755902111530304,
+      0.27544665336608887,
+      0.27594098448753357,
+      0.2768699824810028,
+      0.2770766019821167,
+      0.275217741727829,
+      0.27790069580078125,
+      0.2772781550884247,
+      0.27898573875427246,
+      0.27966248989105225,
+      0.2800298035144806,
+      0.278813898563385,
+      0.2794055640697479,
+      0.2796669900417328,
+      0.2775470018386841,
+      0.27806222438812256,
+      0.2780100703239441,
+      0.27965667843818665,
+      0.2785974442958832,
+      0.27775922417640686,
+      0.2781241238117218,
+      0.27801811695098877,
+      0.27891024947166443,
+      0.2812408208847046,
+      0.28048190474510193,
+      0.28028103709220886,
+      0.2798750698566437,
+      0.2788301408290863,
+      0.27951404452323914,
+      0.2783767879009247,
+      0.27976107597351074,
+      0.2796590328216553,
+      0.28010427951812744,
+      0.27874061465263367,
+      0.27861717343330383,
+      0.27829408645629883,
+      0.2787402272224426,
+      0.27952054142951965,
+      0.2806849777698517,
+      0.2853986918926239,
+      0.28703945875167847,
+      0.28547102212905884,
+      0.28476831316947937,
+      0.28264766931533813,
+      0.28226372599601746,
+      0.2809774577617645,
+      0.2800523638725281,
+      0.280084490776062,
+      0.28102418780326843,
+      0.2757742702960968,
+      0.2758842408657074,
+      0.2775470018386841,
+      0.2777790427207947,
+      0.27810153365135193,
+      0.2768770158290863,
+      0.2779047191143036,
+      0.27850082516670227,
+      0.27749329805374146,
+      0.2764735519886017,
+      0.27712079882621765,
+      0.2759023606777191,
+      0.27512118220329285,
+      0.27608150243759155,
+      0.2755919098854065,
+      0.2751915156841278,
+      0.27574747800827026,
+      0.27595052123069763,
+      0.275706022977829,
+      0.27671322226524353,
+      0.2758370041847229,
+      0.2761959433555603,
+      0.27639609575271606,
+      0.2769555151462555,
+      0.27667707204818726,
+      0.2767086923122406,
+      0.27628183364868164,
+      0.2764381170272827,
+      0.27710217237472534,
+      0.2773119807243347,
+      0.2782318890094757,
+      0.2790510952472687,
+      0.2790542244911194,
+      0.2792282998561859,
+      0.27902498841285706,
+      0.27822360396385193,
+      0.27823856472969055,
+      0.2783638834953308,
+      0.27829623222351074,
+      0.2803868353366852,
+      0.2808542549610138,
+      0.28132185339927673,
+      0.28175821900367737,
+      0.2818266749382019,
+      0.2817915380001068,
+      0.2818235754966736,
+      0.2817734181880951,
+      0.2814966142177582,
+      0.2815135717391968,
+      0.28160402178764343,
+      0.2822325825691223,
+      0.28264033794403076,
+      0.28252533078193665,
+      0.2823033034801483,
+      0.2821318507194519,
+      0.28197363018989563,
+      0.28218743205070496,
+      0.2823140621185303,
+      0.2824079394340515,
+      0.2824495732784271,
+      0.28276515007019043
+    ],
+    "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": 0.70625,
+  "sft_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": null,
+      "mode": "sft",
+      "n_digits": 6,
+      "n_per_split": 100
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 0.94,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9917355371900827,
+            "count": 605
+          },
+          "SS": {
+            "accuracy": 0.9894736842105263,
+            "count": 95
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 0.86,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9901960784313726,
+            "count": 204
+          },
+          "SC": {
+            "accuracy": 0.9822485207100592,
+            "count": 169
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 31
+          },
+          "UC": {
+            "accuracy": 0.9695945945945946,
+            "count": 296
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 0.83,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9938650306748467,
+            "count": 163
+          },
+          "SC": {
+            "accuracy": 0.9615384615384616,
+            "count": 130
+          },
+          "SS": {
+            "accuracy": 0.9540229885057471,
+            "count": 87
+          },
+          "UC": {
+            "accuracy": 0.9655172413793104,
+            "count": 203
+          },
+          "US": {
+            "accuracy": 0.9914529914529915,
+            "count": 117
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 0.53,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9752066115702479,
+            "count": 121
+          },
+          "SC": {
+            "accuracy": 0.9917355371900827,
+            "count": 121
+          },
+          "SS": {
+            "accuracy": 0.9795918367346939,
+            "count": 49
+          },
+          "UC": {
+            "accuracy": 0.7849462365591398,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 0.9551569506726457,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 0.32,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9903846153846154,
+            "count": 104
+          },
+          "SC": {
+            "accuracy": 0.9811320754716981,
+            "count": 106
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 23
+          },
+          "UC": {
+            "accuracy": 0.7,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 0.755700325732899,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 0.13,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.33,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.4825,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 0.48,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.65,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.7,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 0.86,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9798657718120806,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 0.9875,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 0.9697542533081286,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 0.57,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.7927461139896373,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 0.9065420560747663,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 0.66,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.890625,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 0.8819444444444444,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 0.47,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.8169934640522876,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 0.845360824742268,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 0.61,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.8934426229508197,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 0.9401709401709402,
+            "count": 234
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 0.89,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9816971713810316,
+            "count": 601
+          },
+          "ME": {
+            "accuracy": 0.98989898989899,
+            "count": 99
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 0.85,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.985663082437276,
+            "count": 279
+          },
+          "MB": {
+            "accuracy": 0.9724137931034482,
+            "count": 145
+          },
+          "ME": {
+            "accuracy": 0.9583333333333334,
+            "count": 24
+          },
+          "UB": {
+            "accuracy": 0.9722222222222222,
+            "count": 252
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 0.67,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9812206572769953,
+            "count": 213
+          },
+          "MB": {
+            "accuracy": 0.9557522123893806,
+            "count": 113
+          },
+          "ME": {
+            "accuracy": 0.9647058823529412,
+            "count": 85
+          },
+          "UB": {
+            "accuracy": 0.8895027624309392,
+            "count": 181
+          },
+          "UD": {
+            "accuracy": 0.9629629629629629,
+            "count": 108
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 0.3,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9888268156424581,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 0.970873786407767,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 0.6174496644295302,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 0.8826291079812206,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 0.1,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.57,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.5366666666666666,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 0.09,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.58,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.4375,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 0.84,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9833333333333333,
+            "count": 600
+          },
+          "MB": {
+            "accuracy": 0.9887640449438202,
+            "count": 267
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 53
+          },
+          "UB": {
+            "accuracy": 0.9567198177676538,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 0.975609756097561,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 0.62,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9933333333333333,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.8477157360406091,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 0.7961165048543689,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 0.55,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.8502024291497976,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 0.7908496732026143,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 0.49,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.8489932885906041,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 0.8465346534653465,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 0.5979166666666667,
+      "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": 0.99,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9983471074380166,
+            "count": 605
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 95
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 0.95,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 204
+          },
+          "SC": {
+            "accuracy": 0.9940828402366864,
+            "count": 169
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 31
+          },
+          "UC": {
+            "accuracy": 0.9864864864864865,
+            "count": 296
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 0.87,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9938650306748467,
+            "count": 163
+          },
+          "SC": {
+            "accuracy": 0.9769230769230769,
+            "count": 130
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 87
+          },
+          "UC": {
+            "accuracy": 0.9556650246305419,
+            "count": 203
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 117
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 0.66,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9917355371900827,
+            "count": 121
+          },
+          "SC": {
+            "accuracy": 0.9917355371900827,
+            "count": 121
+          },
+          "SS": {
+            "accuracy": 0.9795918367346939,
+            "count": 49
+          },
+          "UC": {
+            "accuracy": 0.8387096774193549,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 0.9506726457399103,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 0.56,
+        "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": 0.7625,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 0.8697068403908795,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 0.18,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.43,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.5775,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 0.32,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.53,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.632,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 0.92,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9865771812080537,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 0.996875,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 0.9829867674858223,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 0.9791666666666666,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 0.65,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.8290155440414507,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 0.9065420560747663,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 0.72,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.91015625,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 0.8888888888888888,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 0.69,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.9052287581699346,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 0.8969072164948454,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 0.72,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.9234972677595629,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 0.9572649572649573,
+            "count": 234
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 0.86,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9767054908485857,
+            "count": 601
+          },
+          "ME": {
+            "accuracy": 0.9797979797979798,
+            "count": 99
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 0.91,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.985663082437276,
+            "count": 279
+          },
+          "MB": {
+            "accuracy": 0.9793103448275862,
+            "count": 145
+          },
+          "ME": {
+            "accuracy": 0.9583333333333334,
+            "count": 24
+          },
+          "UB": {
+            "accuracy": 0.9920634920634921,
+            "count": 252
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 0.77,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9812206572769953,
+            "count": 213
+          },
+          "MB": {
+            "accuracy": 0.9823008849557522,
+            "count": 113
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 85
+          },
+          "UB": {
+            "accuracy": 0.9116022099447514,
+            "count": 181
+          },
+          "UD": {
+            "accuracy": 0.9907407407407407,
+            "count": 108
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 0.65,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 0.970873786407767,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 0.9821428571428571,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 0.8322147651006712,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 0.9061032863849765,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 0.42,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.72,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.73,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 0.38,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.67,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.685,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 0.885,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.98,
+            "count": 600
+          },
+          "MB": {
+            "accuracy": 0.9850187265917603,
+            "count": 267
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 53
+          },
+          "UB": {
+            "accuracy": 0.9840546697038725,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 0.76,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.99,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9137055837563451,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 0.883495145631068,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 0.69,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9149797570850202,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 0.9019607843137255,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 0.59,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.8590604026845637,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 0.8811881188118812,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 0.70625,
+      "total_examples": 2400,
+      "n_splits": 22
+    }
+  },
+  "sorl_overall_accuracy": 0.70625,
+  "sft_overall_accuracy": 0.5979166666666667
+}

add_sub_sorl_v1_abs10_K1_100K_1L3H510d/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05915d1d3abf503303872b835b197cd7e1c25afe84a5656f0a395715ce75f417
+size 634679036

add_sub_sorl_v1_abs10_K1_100K_1L3H510d/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_abs10_K1_100K_1L3H510d",
+  "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": 1,
+  "n_head": 3,
+  "n_embd": 510,
+  "ops": "add_sub",
+  "abs_vocab": 10,
+  "dataset_size": 100000,
+  "mode": "sorl",
+  "device": "cuda",
+  "push_to_hub": true,
+  "no_wandb": false,
+  "n_params": 158593426,
+  "run_name": "add_sub_sorl_v1_abs10_K1_100K_1L3H510d",
+  "git_commit": "f447da529caceac8c7d256cbb2cd185cbc50feac",
+  "timestamp": "2026-04-12T09:52:22.660543+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": "u4e5s4m7",
+  "wandb_url": "https://wandb.ai/nlp_and_interpretability/sorl-arithmetic/runs/u4e5s4m7",
+  "final_accuracy": 0.70625,
+  "sft_accuracy": 0.5979166666666667,
+  "eval_method": "ArithmeticEvaluator"
+}