Upload add_sub_sorl_v1_abs10_100K

add_sub_sorl_v1_abs10_100K/config.json

@@ -0,0 +1,37 @@
+{
+  "architectures": [
+    "SorlModelWrapper"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": null,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 510,
+  "initializer_range": 0.02,
+  "intermediate_size": 2040,
+  "layer_types": [
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 128,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 3,
+  "num_hidden_layers": 2,
+  "num_key_value_heads": 3,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "transformers_version": "5.5.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151654
+}

add_sub_sorl_v1_abs10_100K/generation_config.json

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

add_sub_sorl_v1_abs10_100K/metrics.json

@@ -0,0 +1,3537 @@
+{
+  "history": {
+    "step": [
+      50,
+      100,
+      150,
+      200,
+      250,
+      300,
+      350,
+      400,
+      450,
+      500,
+      550,
+      600,
+      650,
+      700,
+      750,
+      800,
+      850,
+      900,
+      950,
+      1000,
+      1050,
+      1100,
+      1150,
+      1200,
+      1250,
+      1300,
+      1350,
+      1400,
+      1450,
+      1500,
+      1550,
+      1613,
+      1663,
+      1713,
+      1763,
+      1813,
+      1863,
+      1913,
+      1963,
+      2013,
+      2063,
+      2113,
+      2163,
+      2213,
+      2263,
+      2313,
+      2363,
+      2413,
+      2463,
+      2513,
+      2563,
+      2613,
+      2663,
+      2713,
+      2763,
+      2813,
+      2863,
+      2913,
+      2963,
+      3013,
+      3063,
+      3113,
+      3176,
+      3226,
+      3276,
+      3326,
+      3376,
+      3426,
+      3476,
+      3526,
+      3576,
+      3626,
+      3676,
+      3726,
+      3776,
+      3826,
+      3876,
+      3926,
+      3976,
+      4026,
+      4076,
+      4126,
+      4176,
+      4226,
+      4276,
+      4326,
+      4376,
+      4426,
+      4476,
+      4526,
+      4576,
+      4626,
+      4676,
+      4739,
+      4789,
+      4839,
+      4889,
+      4939,
+      4989,
+      5039,
+      5089,
+      5139,
+      5189,
+      5239,
+      5289,
+      5339,
+      5389,
+      5439,
+      5489,
+      5539,
+      5589,
+      5639,
+      5689,
+      5739,
+      5789,
+      5839,
+      5889,
+      5939,
+      5989,
+      6039,
+      6089,
+      6139,
+      6189,
+      6239,
+      6302,
+      6352,
+      6402,
+      6452,
+      6502,
+      6552,
+      6602,
+      6652,
+      6702,
+      6752,
+      6802,
+      6852,
+      6902,
+      6952,
+      7002,
+      7052,
+      7102,
+      7152,
+      7202,
+      7252,
+      7302,
+      7352,
+      7402,
+      7452,
+      7502,
+      7552,
+      7602,
+      7652,
+      7702,
+      7752,
+      7802,
+      7865,
+      7915,
+      7965,
+      8015,
+      8065,
+      8115,
+      8165,
+      8215,
+      8265,
+      8315,
+      8365,
+      8415,
+      8465,
+      8515,
+      8565,
+      8615,
+      8665,
+      8715,
+      8765,
+      8815,
+      8865,
+      8915,
+      8965,
+      9015,
+      9065,
+      9115,
+      9165,
+      9215,
+      9265,
+      9315,
+      9365,
+      9428,
+      9478,
+      9528,
+      9578,
+      9628,
+      9678,
+      9728,
+      9778,
+      9828,
+      9878,
+      9928,
+      9978,
+      10028,
+      10078,
+      10128,
+      10178,
+      10228,
+      10278,
+      10328,
+      10378,
+      10428,
+      10478,
+      10528,
+      10578,
+      10628,
+      10678,
+      10728,
+      10778,
+      10828,
+      10878,
+      10928,
+      10991,
+      11041,
+      11091,
+      11141,
+      11191,
+      11241,
+      11291,
+      11341,
+      11391,
+      11441,
+      11491,
+      11541,
+      11591,
+      11641,
+      11691,
+      11741,
+      11791,
+      11841,
+      11891,
+      11941,
+      11991,
+      12041,
+      12091,
+      12141,
+      12191,
+      12241,
+      12291,
+      12341,
+      12391,
+      12441,
+      12491,
+      12554,
+      12604,
+      12654,
+      12704,
+      12754,
+      12804,
+      12854,
+      12904,
+      12954,
+      13004,
+      13054,
+      13104,
+      13154,
+      13204,
+      13254,
+      13304,
+      13354,
+      13404,
+      13454,
+      13504,
+      13554,
+      13604,
+      13654,
+      13704,
+      13754,
+      13804,
+      13854,
+      13904,
+      13954,
+      14004,
+      14054,
+      14117,
+      14167,
+      14217,
+      14267,
+      14317,
+      14367,
+      14417,
+      14467,
+      14517,
+      14567,
+      14617,
+      14667,
+      14717,
+      14767,
+      14817,
+      14867,
+      14917,
+      14967,
+      15017,
+      15067,
+      15117,
+      15167,
+      15217,
+      15267,
+      15317,
+      15367,
+      15417,
+      15467,
+      15517,
+      15567,
+      15617
+    ],
+    "loss": [
+      17.898242950439453,
+      12.17095947265625,
+      10.000726699829102,
+      8.469908714294434,
+      5.317510604858398,
+      4.32233190536499,
+      3.622828960418701,
+      2.805741786956787,
+      3.010690689086914,
+      2.483719825744629,
+      2.4329562187194824,
+      2.0066065788269043,
+      2.3034160137176514,
+      1.6520445346832275,
+      1.269864559173584,
+      1.2628942728042603,
+      0.8415369987487793,
+      0.4964118003845215,
+      -0.6137298345565796,
+      -1.0917762517929077,
+      -3.30086088180542,
+      -5.9570698738098145,
+      -8.474160194396973,
+      -9.07567024230957,
+      -10.399657249450684,
+      -10.898626327514648,
+      -11.817375183105469,
+      -11.275421142578125,
+      -13.070290565490723,
+      -13.372167587280273,
+      -13.27103328704834,
+      -14.676581382751465,
+      -14.48254108428955,
+      -14.171213150024414,
+      -14.453689575195312,
+      -14.307726860046387,
+      -14.525967597961426,
+      -14.91265869140625,
+      -14.669103622436523,
+      -14.205328941345215,
+      -14.213898658752441,
+      -14.720457077026367,
+      -14.26444149017334,
+      -14.46597671508789,
+      -13.868637084960938,
+      -14.853498458862305,
+      -14.440397262573242,
+      -14.689319610595703,
+      -14.972709655761719,
+      -14.901583671569824,
+      -15.340761184692383,
+      -15.013080596923828,
+      -14.612590789794922,
+      -14.974559783935547,
+      -14.508498191833496,
+      -14.355701446533203,
+      -12.855260848999023,
+      -12.074932098388672,
+      -11.211496353149414,
+      -10.11567497253418,
+      -8.155570030212402,
+      -7.4798760414123535,
+      -6.254654407501221,
+      -6.687215805053711,
+      -6.552494049072266,
+      -5.425172328948975,
+      -5.611931324005127,
+      -4.553066253662109,
+      -4.747787952423096,
+      -4.689442157745361,
+      -4.176784038543701,
+      -4.077182769775391,
+      -4.381514072418213,
+      -4.524280071258545,
+      -5.140686988830566,
+      -3.7790939807891846,
+      -3.596569299697876,
+      -3.590451240539551,
+      -4.132176876068115,
+      -3.858391761779785,
+      -3.7297918796539307,
+      -3.5094447135925293,
+      -3.3534436225891113,
+      -3.800368070602417,
+      -3.373728036880493,
+      -3.301117420196533,
+      -3.8689463138580322,
+      -2.935724973678589,
+      -3.1455042362213135,
+      -3.1886491775512695,
+      -3.0887057781219482,
+      -2.891690254211426,
+      -3.128448486328125,
+      -2.6364502906799316,
+      -2.9243524074554443,
+      -2.4197757244110107,
+      -2.991380214691162,
+      -2.23624324798584,
+      -2.7561862468719482,
+      -2.6544461250305176,
+      -2.244197368621826,
+      -2.360046625137329,
+      -2.3621275424957275,
+      -2.2447409629821777,
+      -2.5257372856140137,
+      -2.61136794090271,
+      -2.6983206272125244,
+      -2.5952131748199463,
+      -2.236645221710205,
+      -2.2343759536743164,
+      -2.2564358711242676,
+      -2.0840580463409424,
+      -2.0381150245666504,
+      -3.134915828704834,
+      -2.1949312686920166,
+      -2.674781322479248,
+      -1.872277855873108,
+      -2.0133588314056396,
+      -2.323359966278076,
+      -2.209495782852173,
+      -2.056077241897583,
+      -2.3222715854644775,
+      -1.850269079208374,
+      -2.011166572570801,
+      -2.423705816268921,
+      -1.995131254196167,
+      -2.2757530212402344,
+      -2.1609740257263184,
+      -1.8811988830566406,
+      -2.178713083267212,
+      -2.0915629863739014,
+      -2.2504334449768066,
+      -1.9422235488891602,
+      -1.9811478853225708,
+      -1.689926028251648,
+      -1.8429358005523682,
+      -2.013540267944336,
+      -2.1036672592163086,
+      -1.8045049905776978,
+      -1.5718125104904175,
+      -2.1454601287841797,
+      -1.6350609064102173,
+      -1.978631615638733,
+      -1.6527613401412964,
+      -1.8721427917480469,
+      -1.604819655418396,
+      -1.5306353569030762,
+      -2.0175392627716064,
+      -1.8658630847930908,
+      -1.6680402755737305,
+      -1.6254310607910156,
+      -1.4316753149032593,
+      -1.957127332687378,
+      -1.3733186721801758,
+      -1.5614428520202637,
+      -1.6985728740692139,
+      -1.528280258178711,
+      -1.4709199666976929,
+      -1.6057331562042236,
+      -1.5918432474136353,
+      -1.8127381801605225,
+      -1.3713455200195312,
+      -2.157377004623413,
+      -1.8324764966964722,
+      -1.5383961200714111,
+      -1.2715643644332886,
+      -1.43584144115448,
+      -1.363757610321045,
+      -1.2672580480575562,
+      -1.0625731945037842,
+      -1.0825140476226807,
+      -1.4479258060455322,
+      -1.4802824258804321,
+      -1.6093518733978271,
+      -1.3037806749343872,
+      -1.1348497867584229,
+      -1.1780712604522705,
+      -1.3891611099243164,
+      -1.2522574663162231,
+      -1.2620816230773926,
+      -1.1609874963760376,
+      -1.3132126331329346,
+      -1.2223495244979858,
+      -0.9222760200500488,
+      -0.9897661805152893,
+      -1.2174392938613892,
+      -1.2716630697250366,
+      -0.9742103219032288,
+      -1.4609808921813965,
+      -0.9385769367218018,
+      -1.05704927444458,
+      -0.7617784738540649,
+      -0.7693957090377808,
+      -0.8679118752479553,
+      -0.9695261716842651,
+      -0.7272455096244812,
+      -1.128410816192627,
+      -0.5502809286117554,
+      -0.6828831434249878,
+      -0.6310953497886658,
+      -0.6145992875099182,
+      -1.0135717391967773,
+      -0.6107086539268494,
+      -1.1609718799591064,
+      -0.9355491399765015,
+      -1.138651967048645,
+      -0.7039048075675964,
+      -0.7265371680259705,
+      -0.7398449778556824,
+      -1.1441463232040405,
+      -0.716513991355896,
+      -0.549831748008728,
+      -0.6496762037277222,
+      -0.5133782029151917,
+      -0.7706650495529175,
+      -0.6165258884429932,
+      -0.5687806606292725,
+      -0.4623892605304718,
+      -0.5749399065971375,
+      -0.4525548219680786,
+      -0.5484775900840759,
+      -0.39739224314689636,
+      -0.2298678755760193,
+      -0.5226291418075562,
+      -0.44217923283576965,
+      -0.45984476804733276,
+      -0.3971169888973236,
+      -0.2655782103538513,
+      -0.30226850509643555,
+      -0.5683307647705078,
+      -0.7837846875190735,
+      -0.2775364816188812,
+      -0.40293148159980774,
+      -0.39146527647972107,
+      -0.3872049152851105,
+      -0.2855488955974579,
+      -0.3582993745803833,
+      -0.29680517315864563,
+      -0.3545703887939453,
+      -0.5397948622703552,
+      -0.43307721614837646,
+      -0.4876730442047119,
+      -0.5343471169471741,
+      -0.21957853436470032,
+      -0.39358770847320557,
+      0.015100548975169659,
+      -0.42391523718833923,
+      -0.4264887571334839,
+      -0.14468659460544586,
+      -0.20275095105171204,
+      -0.35929185152053833,
+      -0.34941935539245605,
+      -0.48810499906539917,
+      -0.08532863110303879,
+      -0.3265494108200073,
+      -0.3058772683143616,
+      -0.39665207266807556,
+      -0.2523774802684784,
+      -0.20658715069293976,
+      -0.21619853377342224,
+      -0.18327531218528748,
+      -0.23171119391918182,
+      -0.21158039569854736,
+      -0.275308758020401,
+      -0.18174022436141968,
+      -0.27848175168037415,
+      -0.16791534423828125,
+      -0.21869049966335297,
+      -0.12352351099252701,
+      -0.2989782691001892,
+      -0.2069055289030075,
+      -0.13711804151535034,
+      -0.04984921216964722,
+      -0.04339553415775299,
+      -0.14695435762405396,
+      -0.16203084588050842,
+      -0.1392071545124054,
+      -0.13278606534004211,
+      -0.11039897799491882,
+      -0.07540819048881531,
+      -0.1325184851884842,
+      -0.21475887298583984,
+      -0.10040821135044098,
+      -0.15006886422634125,
+      -0.03858878090977669,
+      -0.15046735107898712,
+      -0.08435267210006714,
+      -0.10838817805051804,
+      -0.15623150765895844,
+      -0.18306495249271393,
+      -0.0787075012922287,
+      -0.0882338136434555,
+      -0.1638314425945282,
+      -0.12283475697040558,
+      -0.11741890758275986,
+      -0.08260732144117355,
+      -0.04303062707185745,
+      -0.10231343656778336,
+      -0.07560180127620697,
+      -0.052348893135786057,
+      -0.10580295324325562,
+      -0.06423408538103104,
+      -0.15607650578022003,
+      -0.02391888201236725,
+      -0.13538967072963715,
+      -0.08471515774726868,
+      -0.09317309409379959,
+      -0.048827387392520905,
+      -0.16070179641246796,
+      -0.07308726012706757
+    ],
+    "base_loss": [
+      10.568315505981445,
+      8.143197059631348,
+      7.364155292510986,
+      6.420115947723389,
+      5.458926200866699,
+      3.896120548248291,
+      2.368774890899658,
+      2.12823224067688,
+      1.9438966512680054,
+      1.8385487794876099,
+      1.8635467290878296,
+      1.8358601331710815,
+      1.8607227802276611,
+      1.867401361465454,
+      1.8732596635818481,
+      1.8480777740478516,
+      1.8224256038665771,
+      1.887178659439087,
+      1.8829847574234009,
+      1.9112404584884644,
+      1.8766212463378906,
+      1.9506890773773193,
+      1.9369127750396729,
+      1.809589147567749,
+      1.8270591497421265,
+      1.7747080326080322,
+      1.804424524307251,
+      1.6654764413833618,
+      1.751176118850708,
+      1.7893718481063843,
+      1.712988257408142,
+      1.8015085458755493,
+      1.7446907758712769,
+      1.6956506967544556,
+      1.7068536281585693,
+      1.650771141052246,
+      1.6782304048538208,
+      1.6954421997070312,
+      1.6798969507217407,
+      1.628057837486267,
+      1.61090886592865,
+      1.6740840673446655,
+      1.6245861053466797,
+      1.63609778881073,
+      1.56288743019104,
+      1.6816425323486328,
+      1.6284650564193726,
+      1.6451350450515747,
+      1.6970590353012085,
+      1.6629184484481812,
+      1.716318964958191,
+      1.6742541790008545,
+      1.6343353986740112,
+      1.6724300384521484,
+      1.6198190450668335,
+      1.6007970571517944,
+      1.451180100440979,
+      1.3689472675323486,
+      1.2527589797973633,
+      1.1551610231399536,
+      0.9227293133735657,
+      0.8394975662231445,
+      0.7029710412025452,
+      0.7575051188468933,
+      0.7359369993209839,
+      0.6122614741325378,
+      0.6277788877487183,
+      0.5123354196548462,
+      0.5338669419288635,
+      0.5280805826187134,
+      0.46961620450019836,
+      0.46630558371543884,
+      0.49189749360084534,
+      0.5055493712425232,
+      0.5758814215660095,
+      0.4242982268333435,
+      0.40284016728401184,
+      0.4024098217487335,
+      0.46268826723098755,
+      0.43445953726768494,
+      0.4170833230018616,
+      0.3946475386619568,
+      0.37832775712013245,
+      0.426046222448349,
+      0.37859299778938293,
+      0.3699304163455963,
+      0.4385518431663513,
+      0.3466642200946808,
+      0.3540719449520111,
+      0.356463223695755,
+      0.3454909324645996,
+      0.32347074151039124,
+      0.3511905074119568,
+      0.2992519736289978,
+      0.3298533260822296,
+      0.27088990807533264,
+      0.33507975935935974,
+      0.25134751200675964,
+      0.3087560534477234,
+      0.29757770895957947,
+      0.251931369304657,
+      0.26579374074935913,
+      0.2649250626564026,
+      0.25241801142692566,
+      0.28240329027175903,
+      0.2957579791545868,
+      0.30345407128334045,
+      0.2916712760925293,
+      0.25145572423934937,
+      0.2508265972137451,
+      0.2539704442024231,
+      0.23422227799892426,
+      0.23018407821655273,
+      0.35051247477531433,
+      0.25069090723991394,
+      0.32541435956954956,
+      0.21303556859493256,
+      0.22561796009540558,
+      0.2612958550453186,
+      0.24799372255802155,
+      0.23154176771640778,
+      0.2606547474861145,
+      0.21254435181617737,
+      0.22772814333438873,
+      0.27201953530311584,
+      0.22492162883281708,
+      0.2579091191291809,
+      0.2427002340555191,
+      0.22184371948242188,
+      0.24444580078125,
+      0.23784099519252777,
+      0.2554014027118683,
+      0.21875402331352234,
+      0.22619032859802246,
+      0.19057056307792664,
+      0.20802749693393707,
+      0.22737279534339905,
+      0.23751647770404816,
+      0.20238709449768066,
+      0.18440593779087067,
+      0.24258053302764893,
+      0.1829206496477127,
+      0.22208870947360992,
+      0.18690092861652374,
+      0.20956198871135712,
+      0.18001887202262878,
+      0.1723724603652954,
+      0.22669336199760437,
+      0.2112317830324173,
+      0.18780024349689484,
+      0.18407921493053436,
+      0.1614924669265747,
+      0.2193380743265152,
+      0.15642760694026947,
+      0.1785077303647995,
+      0.19131197035312653,
+      0.1719256490468979,
+      0.16998495161533356,
+      0.18009281158447266,
+      0.1786258965730667,
+      0.20350196957588196,
+      0.15445144474506378,
+      0.24213048815727234,
+      0.20611858367919922,
+      0.1730114370584488,
+      0.14370106160640717,
+      0.16377748548984528,
+      0.15389804542064667,
+      0.14356811344623566,
+      0.14054524898529053,
+      0.12244506925344467,
+      0.1635432243347168,
+      0.16939735412597656,
+      0.18151232600212097,
+      0.1470525562763214,
+      0.1271251142024994,
+      0.13306169211864471,
+      0.15618088841438293,
+      0.1405344009399414,
+      0.14169029891490936,
+      0.1312611699104309,
+      0.15000268816947937,
+      0.13788385689258575,
+      0.10699386149644852,
+      0.11203130334615707,
+      0.1377689689397812,
+      0.14300259947776794,
+      0.11074953526258469,
+      0.16385547816753387,
+      0.106101393699646,
+      0.11837797611951828,
+      0.08841868489980698,
+      0.08911939710378647,
+      0.09771926701068878,
+      0.10882412642240524,
+      0.08234436064958572,
+      0.12939295172691345,
+      0.0629371628165245,
+      0.07891169935464859,
+      0.07331836223602295,
+      0.06941264122724533,
+      0.1135721355676651,
+      0.06897144019603729,
+      0.13211266696453094,
+      0.10580595582723618,
+      0.12884125113487244,
+      0.07908945530653,
+      0.08248370885848999,
+      0.0836305096745491,
+      0.12909452617168427,
+      0.08136750757694244,
+      0.06289603561162949,
+      0.07312337309122086,
+      0.0587802417576313,
+      0.0881800428032875,
+      0.06981097906827927,
+      0.0647372454404831,
+      0.05323735252022743,
+      0.0651235431432724,
+      0.05145469680428505,
+      0.06290210038423538,
+      0.045060139149427414,
+      0.02635861560702324,
+      0.06054062023758888,
+      0.05092387646436691,
+      0.051927804946899414,
+      0.04568718746304512,
+      0.05043186619877815,
+      0.03472381457686424,
+      0.06507039815187454,
+      0.08881647139787674,
+      0.03155577555298805,
+      0.04574210196733475,
+      0.04629113897681236,
+      0.04489707201719284,
+      0.03378498926758766,
+      0.04083876684308052,
+      0.04593208059668541,
+      0.04099803417921066,
+      0.061005882918834686,
+      0.050650376826524734,
+      0.057167086750268936,
+      0.06097138673067093,
+      0.026157842949032784,
+      0.04480227455496788,
+      0.03374704346060753,
+      0.04900706559419632,
+      0.048461027443408966,
+      0.017718875780701637,
+      0.023490795865654945,
+      0.04097198694944382,
+      0.04008692502975464,
+      0.055264316499233246,
+      0.010471915826201439,
+      0.03744228929281235,
+      0.035062629729509354,
+      0.0454421266913414,
+      0.03002260811626911,
+      0.023973463103175163,
+      0.025306016206741333,
+      0.02189379557967186,
+      0.027451980859041214,
+      0.025912059471011162,
+      0.03169391676783562,
+      0.021139992401003838,
+      0.032432056963443756,
+      0.019880054518580437,
+      0.02606010250747204,
+      0.014689230360090733,
+      0.03407203033566475,
+      0.02402709797024727,
+      0.017139960080385208,
+      0.009125121869146824,
+      0.006708136759698391,
+      0.01757701113820076,
+      0.019256500527262688,
+      0.017534470185637474,
+      0.015949562191963196,
+      0.01345624215900898,
+      0.009886420331895351,
+      0.01634773425757885,
+      0.024979816749691963,
+      0.013034196570515633,
+      0.018518775701522827,
+      0.007457684725522995,
+      0.01740308478474617,
+      0.011986021883785725,
+      0.01300106942653656,
+      0.01853434555232525,
+      0.021597137674689293,
+      0.01134154386818409,
+      0.011389104649424553,
+      0.01879272796213627,
+      0.014911564067006111,
+      0.013964526355266571,
+      0.011024332605302334,
+      0.006991051137447357,
+      0.012699066661298275,
+      0.009297139011323452,
+      0.007379917893558741,
+      0.013352419249713421,
+      0.008603994734585285,
+      0.018944447860121727,
+      0.00466900784522295,
+      0.016725314781069756,
+      0.01058161724358797,
+      0.01205870695412159,
+      0.007595142815262079,
+      0.019408103078603745,
+      0.009309765882790089
+    ],
+    "info_loss": [
+      -0.1958761215209961,
+      -0.408872127532959,
+      -0.33576154708862305,
+      -0.22092151641845703,
+      -0.31899309158325195,
+      -0.19400596618652344,
+      -0.08042073249816895,
+      -0.12741851806640625,
+      -0.08406126499176025,
+      -0.12506330013275146,
+      -0.13188695907592773,
+      -0.1713271141052246,
+      -0.14365017414093018,
+      -0.20933854579925537,
+      -0.2477884292602539,
+      -0.24557054042816162,
+      -0.2851436138153076,
+      -0.3258519172668457,
+      -0.4353005886077881,
+      -0.4823911190032959,
+      -0.6876966953277588,
+      -0.9192705154418945,
+      -1.093753457069397,
+      -1.1159758567810059,
+      -1.2424697875976562,
+      -1.283799648284912,
+      -1.3774747848510742,
+      -1.3076419830322266,
+      -1.4936635494232178,
+      -1.5272959470748901,
+      -1.509727120399475,
+      -1.65697181224823,
+      -1.6324580907821655,
+      -1.5960543155670166,
+      -1.62422776222229,
+      -1.6046651601791382,
+      -1.6272552013397217,
+      -1.668359398841858,
+      -1.6423128843307495,
+      -1.5906425714492798,
+      -1.58965265750885,
+      -1.6462726593017578,
+      -1.5967046022415161,
+      -1.6165791749954224,
+      -1.5496994256973267,
+      -1.6593738794326782,
+      -1.6130362749099731,
+      -1.6383200883865356,
+      -1.6712359189987183,
+      -1.6604275703430176,
+      -1.7099218368530273,
+      -1.6720092296600342,
+      -1.6285072565078735,
+      -1.6703423261642456,
+      -1.616640329360962,
+      -1.5992320775985718,
+      -1.4349874258041382,
+      -1.347501516342163,
+      -1.2491847276687622,
+      -1.1298017501831055,
+      -0.9112951755523682,
+      -0.8348812460899353,
+      -0.7004478573799133,
+      -0.7471547722816467,
+      -0.7329674959182739,
+      -0.6061534285545349,
+      -0.6268084645271301,
+      -0.5092931389808655,
+      -0.5315150022506714,
+      -0.5259316563606262,
+      -0.46720921993255615,
+      -0.4566917419433594,
+      -0.48945683240890503,
+      -0.5048118829727173,
+      -0.5754508376121521,
+      -0.42379122972488403,
+      -0.4020898640155792,
+      -0.4019421637058258,
+      -0.46214649081230164,
+      -0.433444082736969,
+      -0.416523814201355,
+      -0.3938884437084198,
+      -0.3746926188468933,
+      -0.42470821738243103,
+      -0.3775221109390259,
+      -0.3694234788417816,
+      -0.4327273666858673,
+      -0.33102893829345703,
+      -0.35362643003463745,
+      -0.35618698596954346,
+      -0.3451445698738098,
+      -0.3230594992637634,
+      -0.3498040735721588,
+      -0.2983073890209198,
+      -0.328479528427124,
+      -0.2706272006034851,
+      -0.33466997742652893,
+      -0.25122931599617004,
+      -0.30850228667259216,
+      -0.29735299944877625,
+      -0.2517610788345337,
+      -0.26403889060020447,
+      -0.26419973373413086,
+      -0.2520931363105774,
+      -0.2822090983390808,
+      -0.2948746681213379,
+      -0.30327367782592773,
+      -0.29149749875068665,
+      -0.25109487771987915,
+      -0.25057077407836914,
+      -0.2531075179576874,
+      -0.23408158123493195,
+      -0.2298281192779541,
+      -0.3504043221473694,
+      -0.24677129089832306,
+      -0.3017372190952301,
+      -0.2120485156774521,
+      -0.22554655373096466,
+      -0.26121786236763,
+      -0.24792437255382538,
+      -0.2300400584936142,
+      -0.2605368196964264,
+      -0.20973044633865356,
+      -0.22753840684890747,
+      -0.27192503213882446,
+      -0.2248055338859558,
+      -0.2550696134567261,
+      -0.24217553436756134,
+      -0.21174460649490356,
+      -0.24400071799755096,
+      -0.23711474239826202,
+      -0.25497865676879883,
+      -0.2180212289094925,
+      -0.2259577214717865,
+      -0.1904374063014984,
+      -0.20753498375415802,
+      -0.22714219987392426,
+      -0.23731212317943573,
+      -0.20226487517356873,
+      -0.17811334133148193,
+      -0.24165070056915283,
+      -0.18276356160640717,
+      -0.22199979424476624,
+      -0.18578781187534332,
+      -0.20945100486278534,
+      -0.17994889616966248,
+      -0.17219844460487366,
+      -0.22660747170448303,
+      -0.20932692289352417,
+      -0.18746373057365417,
+      -0.18371479213237762,
+      -0.1613215059041977,
+      -0.2192639261484146,
+      -0.15637393295764923,
+      -0.17581233382225037,
+      -0.19124145805835724,
+      -0.17119184136390686,
+      -0.16722899675369263,
+      -0.17996202409267426,
+      -0.1783156394958496,
+      -0.20310139656066895,
+      -0.15436813235282898,
+      -0.24130144715309143,
+      -0.20557448267936707,
+      -0.17239594459533691,
+      -0.14307594299316406,
+      -0.16181223094463348,
+      -0.1532348245382309,
+      -0.14306606352329254,
+      -0.12188619375228882,
+      -0.12194667011499405,
+      -0.16347120702266693,
+      -0.1671372354030609,
+      -0.18076840043067932,
+      -0.1464703232049942,
+      -0.12702254951000214,
+      -0.13300567865371704,
+      -0.15573778748512268,
+      -0.14048932492733002,
+      -0.1415444314479828,
+      -0.131224125623703,
+      -0.1496949940919876,
+      -0.1377827525138855,
+      -0.10693496465682983,
+      -0.11159024387598038,
+      -0.13759806752204895,
+      -0.14285747706890106,
+      -0.10937677323818207,
+      -0.16366244852542877,
+      -0.10602877289056778,
+      -0.1183440312743187,
+      -0.08684682101011276,
+      -0.08878514170646667,
+      -0.097652867436409,
+      -0.10875992476940155,
+      -0.0822797417640686,
+      -0.12756849825382233,
+      -0.0629057064652443,
+      -0.0781344622373581,
+      -0.0717904195189476,
+      -0.0689859539270401,
+      -0.11354324966669083,
+      -0.0689539909362793,
+      -0.13209789991378784,
+      -0.10523242503404617,
+      -0.12857557833194733,
+      -0.07905884832143784,
+      -0.08211970329284668,
+      -0.08358220756053925,
+      -0.12888899445533752,
+      -0.08076901733875275,
+      -0.06234963238239288,
+      -0.07308661192655563,
+      -0.05873045697808266,
+      -0.0879305899143219,
+      -0.06941428035497665,
+      -0.06471981108188629,
+      -0.05259700119495392,
+      -0.06507353484630585,
+      -0.05134617164731026,
+      -0.062194351106882095,
+      -0.04503596946597099,
+      -0.026294687762856483,
+      -0.060498833656311035,
+      -0.05090045928955078,
+      -0.05191384255886078,
+      -0.0455852746963501,
+      -0.03276360034942627,
+      -0.034416913986206055,
+      -0.06487513333559036,
+      -0.08880289644002914,
+      -0.03150329738855362,
+      -0.045720167458057404,
+      -0.04626934602856636,
+      -0.044708702713251114,
+      -0.0337369404733181,
+      -0.040805280208587646,
+      -0.03534899652004242,
+      -0.040939390659332275,
+      -0.06097675859928131,
+      -0.05063151195645332,
+      -0.05713341385126114,
+      -0.060942329466342926,
+      -0.02614479325711727,
+      -0.04479479789733887,
+      -0.0025909747928380966,
+      -0.04899991303682327,
+      -0.04843590781092644,
+      -0.017695382237434387,
+      -0.023478390648961067,
+      -0.0409560464322567,
+      -0.040065813809633255,
+      -0.05523411184549332,
+      -0.010462231934070587,
+      -0.037393707782030106,
+      -0.03504030406475067,
+      -0.04541987553238869,
+      -0.030009526759386063,
+      -0.023965151980519295,
+      -0.025134364143013954,
+      -0.021883761510252953,
+      -0.027442272752523422,
+      -0.025906745344400406,
+      -0.031664200127124786,
+      -0.02112094685435295,
+      -0.03222130611538887,
+      -0.019869931042194366,
+      -0.02602902054786682,
+      -0.014678400941193104,
+      -0.03405144065618515,
+      -0.024011127650737762,
+      -0.017130805179476738,
+      -0.00910995900630951,
+      -0.006597171537578106,
+      -0.017515825107693672,
+      -0.019251558929681778,
+      -0.017495347186923027,
+      -0.015909546986222267,
+      -0.013114679604768753,
+      -0.009829901158809662,
+      -0.01633986458182335,
+      -0.024949651211500168,
+      -0.012981235980987549,
+      -0.018510917201638222,
+      -0.007451067212969065,
+      -0.01737871952354908,
+      -0.011973484419286251,
+      -0.01299270335584879,
+      -0.01852962002158165,
+      -0.021585263311862946,
+      -0.011337866075336933,
+      -0.01138055045157671,
+      -0.01878957450389862,
+      -0.014898382127285004,
+      -0.013961542397737503,
+      -0.011020154692232609,
+      -0.0068219625391066074,
+      -0.01268070936203003,
+      -0.009293211624026299,
+      -0.007372069638222456,
+      -0.01334339939057827,
+      -0.008575188927352428,
+      -0.01893424242734909,
+      -0.004299005027860403,
+      -0.01672191545367241,
+      -0.010577057488262653,
+      -0.012055435217916965,
+      -0.007592372130602598,
+      -0.01940261386334896,
+      -0.0093060452491045
+    ],
+    "abs_loss": [
+      2.299344062805176,
+      2.260986328125,
+      2.1556284427642822,
+      2.104736328125,
+      1.9904539585113525,
+      1.9295542240142822,
+      1.8402255773544312,
+      1.8242886066436768,
+      1.7795265913009644,
+      1.8407790660858154,
+      1.8398520946502686,
+      1.849395990371704,
+      1.843975305557251,
+      1.8516335487365723,
+      1.7879228591918945,
+      1.788717269897461,
+      1.759417176246643,
+      1.8009793758392334,
+      1.7452125549316406,
+      1.7425484657287598,
+      1.555834412574768,
+      1.1474618911743164,
+      0.8454347848892212,
+      0.704007625579834,
+      0.6705518364906311,
+      0.609591007232666,
+      0.6524008512496948,
+      0.6311492919921875,
+      0.5775978565216064,
+      0.5621682405471802,
+      0.5438365936279297,
+      0.44852185249328613,
+      0.5401815176010132,
+      0.4914752244949341,
+      0.4704228639602661,
+      0.4958714246749878,
+      0.3765189051628113,
+      0.4290388226509094,
+      0.4278205335140228,
+      0.4029718339443207,
+      0.363197386264801,
+      0.3876047730445862,
+      0.35308271646499634,
+      0.4169761538505554,
+      0.4003360867500305,
+      0.36092817783355713,
+      0.2622601389884949,
+      0.21992714703083038,
+      0.22185902297496796,
+      0.22628647089004517,
+      0.1717938482761383,
+      0.13614629209041595,
+      0.23006287217140198,
+      0.12394532561302185,
+      0.19495609402656555,
+      0.11710832267999649,
+      0.12505380809307098,
+      0.1392887979745865,
+      0.13805821537971497,
+      0.11470522731542587,
+      0.12912991642951965,
+      0.1577836573123932,
+      0.10746321082115173,
+      0.12499044835567474,
+      0.08797664940357208,
+      0.15068618953227997,
+      0.09933540225028992,
+      0.08326271176338196,
+      0.1302133947610855,
+      0.09028597921133041,
+      0.11054735630750656,
+      0.0894804447889328,
+      0.09026975929737091,
+      0.07138688862323761,
+      0.06727860867977142,
+      0.07304547727108002,
+      0.06594955921173096,
+      0.07161735743284225,
+      0.062283098697662354,
+      0.06454157084226608,
+      0.04927849397063255,
+      0.06570068001747131,
+      0.056670963764190674,
+      0.05045980215072632,
+      0.05004817619919777,
+      0.06310935318470001,
+      0.06392456591129303,
+      0.06564883887767792,
+      0.05773782730102539,
+      0.061772946268320084,
+      0.05797974392771721,
+      0.07630695402622223,
+      0.06167856603860855,
+      0.07231593132019043,
+      0.11075171828269958,
+      0.04661320522427559,
+      0.0489351823925972,
+      0.0893353521823883,
+      0.060487184673547745,
+      0.09393326193094254,
+      0.03158995509147644,
+      0.055979665368795395,
+      0.06557424366474152,
+      0.05441770330071449,
+      0.04168401658535004,
+      0.0773334950208664,
+      0.07367226481437683,
+      0.04429794102907181,
+      0.0793289840221405,
+      0.052316803485155106,
+      0.06733226776123047,
+      0.03158876299858093,
+      0.034266795963048935,
+      0.054954931139945984,
+      0.04762925207614899,
+      0.043583113700151443,
+      0.04020775109529495,
+      0.03741905838251114,
+      0.035040151327848434,
+      0.04534371942281723,
+      0.03905560076236725,
+      0.037578366696834564,
+      0.032150596380233765,
+      0.029294807463884354,
+      0.024255525320768356,
+      0.027897939085960388,
+      0.023556236177682877,
+      0.022768860682845116,
+      0.038224779069423676,
+      0.023875029757618904,
+      0.020412296056747437,
+      0.02970881760120392,
+      0.022083569318056107,
+      0.04237690567970276,
+      0.04460488259792328,
+      0.030337054282426834,
+      0.027709612622857094,
+      0.06888288259506226,
+      0.039946548640728,
+      0.042287010699510574,
+      0.029126375913619995,
+      0.025236384943127632,
+      0.02371005341410637,
+      0.027542464435100555,
+      0.025641027837991714,
+      0.021950792521238327,
+      0.036771371960639954,
+      0.025186948478221893,
+      0.02962624467909336,
+      0.053835876286029816,
+      0.03475363925099373,
+      0.01969662308692932,
+      0.022654570639133453,
+      0.03054218366742134,
+      0.037232786417007446,
+      0.028109990060329437,
+      0.03265418857336044,
+      0.02478908933699131,
+      0.02990144118666649,
+      0.03199092298746109,
+      0.02706174924969673,
+      0.028437677770853043,
+      0.060366686433553696,
+      0.022765610367059708,
+      0.015636730939149857,
+      0.027268197387456894,
+      0.023486804217100143,
+      0.038214731961488724,
+      0.01846570149064064,
+      0.03504270315170288,
+      0.021792110055685043,
+      0.017945710569620132,
+      0.011015732772648335,
+      0.0115514462813735,
+      0.016857855021953583,
+      0.009889829903841019,
+      0.017013514414429665,
+      0.0045443433336913586,
+      0.01444745808839798,
+      0.014259792864322662,
+      0.014394810423254967,
+      0.009073961526155472,
+      0.013447131961584091,
+      0.004474250599741936,
+      0.011651335284113884,
+      0.005200245417654514,
+      0.004128721076995134,
+      0.00229267543181777,
+      0.00947949942201376,
+      0.007636476308107376,
+      0.009588472545146942,
+      0.0010691049974411726,
+      0.004199020564556122,
+      0.006084855180233717,
+      0.02109791524708271,
+      0.005467078648507595,
+      0.02420104667544365,
+      0.002727311570197344,
+      0.008652322925627232,
+      0.011295134201645851,
+      0.0014467908767983317,
+      0.0038114560302346945,
+      0.005532393231987953,
+      0.00694839796051383,
+      0.004151223227381706,
+      0.0009016941185109317,
+      0.0006601667264476418,
+      0.004251205362379551,
+      0.00760949170216918,
+      0.012696701101958752,
+      0.008532650768756866,
+      0.001909587299451232,
+      0.0026952014304697514,
+      0.004836875945329666,
+      0.004397994838654995,
+      0.0006370687624439597,
+      0.0008205081103369594,
+      0.00705351447686553,
+      0.0013074338203296065,
+      0.022862592712044716,
+      0.002685739891603589,
+      0.0011106947204098105,
+      0.0010044374503195286,
+      0.011497174389660358,
+      0.0011244597844779491,
+      0.0008501890697516501,
+      0.01461706217378378,
+      0.0003344400611240417,
+      0.00048674835124984384,
+      0.0014398308703675866,
+      0.004889882635325193,
+      0.0050505525432527065,
+      0.00018931939848698676,
+      0.0003458811843302101,
+      0.002722643781453371,
+      0.0002332775911781937,
+      0.0013001272454857826,
+      0.00021184541401453316,
+      0.0009218329796567559,
+      0.0027250817511230707,
+      0.00031440556631423533,
+      0.0002346429682802409,
+      0.001220405800268054,
+      0.0018606358207762241,
+      0.0035638727713376284,
+      0.0016847129445523024,
+      0.014343084767460823,
+      0.0011122154537588358,
+      0.00010775879491120577,
+      0.002860118169337511,
+      0.0009351018234156072,
+      0.0001049952843459323,
+      0.0015063899336382747,
+      0.0004531373269855976,
+      0.0006659829523414373,
+      0.0001711525401333347,
+      0.0008639515144750476,
+      0.0004461331118363887,
+      0.0006608754047192633,
+      0.0036144552286714315,
+      0.0006740841781720519,
+      0.032447319477796555,
+      0.0007009927649050951,
+      0.00489542493596673,
+      0.00434274785220623,
+      8.473070920445025e-05,
+      0.000703380152117461,
+      0.006214356981217861,
+      0.00016676468658261,
+      0.0009907478233799338,
+      0.0025063965003937483,
+      0.0011624305043369532,
+      0.00016601078095845878,
+      0.0006969748064875603,
+      0.007066912017762661,
+      0.003668984165415168,
+      7.324820762732998e-05,
+      0.0028977978508919477,
+      0.005441818851977587,
+      0.0004984366241842508,
+      0.0017339475452899933,
+      0.00023327581584453583,
+      0.0001926865370478481,
+      0.0012457568664103746,
+      0.0007202249835245311,
+      0.000402639911044389,
+      0.0013670185580849648,
+      3.5127817682223395e-05,
+      0.00021311761520337313,
+      0.002809969475492835,
+      4.842525959247723e-05,
+      0.000630392343737185,
+      9.536967263557017e-05,
+      0.00013890725676901639,
+      0.0004332676762714982,
+      0.0023571839556097984,
+      0.00031570333521813154,
+      4.98054432682693e-05,
+      9.038536518346518e-05,
+      8.741506462683901e-05,
+      0.0017366388346999884,
+      0.015279402956366539,
+      0.009888592176139355,
+      0.0007309710490517318,
+      0.0001263176091015339,
+      0.0015627600951120257,
+      3.972904960392043e-05,
+      0.00020485775894485414,
+      0.0020928476005792618,
+      0.01206766813993454
+    ],
+    "zipf_loss": [
+      9.058754920959473,
+      7.890384674072266,
+      5.778624534606934,
+      4.048534393310547,
+      2.8494701385498047,
+      2.1733155250549316,
+      1.8742389678955078,
+      1.76926589012146,
+      1.7294540405273438,
+      1.7117260694503784,
+      1.7042937278747559,
+      1.699077844619751,
+      1.694797396659851,
+      1.692865252494812,
+      1.6956968307495117,
+      1.691650152206421,
+      1.694605827331543,
+      1.6876543760299683,
+      1.6817702054977417,
+      1.6466394662857056,
+      1.5439013242721558,
+      1.1701998710632324,
+      0.44191816449165344,
+      0.2040979564189911,
+      0.1309266984462738,
+      0.10370349138975143,
+      0.08770886063575745,
+      0.07240704447031021,
+      0.0574091337621212,
+      0.0552031472325325,
+      0.05886688455939293,
+      0.04677460715174675,
+      0.04333113506436348,
+      0.04453083872795105,
+      0.034692052751779556,
+      0.03856681287288666,
+      0.0307029876857996,
+      0.032588567584753036,
+      0.03134528174996376,
+      0.03274175524711609,
+      0.03539920970797539,
+      0.029425304383039474,
+      0.04271072894334793,
+      0.022020675241947174,
+      0.025436237454414368,
+      0.0225045308470726,
+      0.0352756530046463,
+      0.02675449475646019,
+      0.020403895527124405,
+      0.017143985256552696,
+      0.024958601221442223,
+      0.01914350688457489,
+      0.015139289200305939,
+      0.04403791204094887,
+      0.018589705228805542,
+      0.024111982434988022,
+      0.0309278666973114,
+      0.01720796898007393,
+      0.013786230236291885,
+      0.015711206942796707,
+      0.02173854038119316,
+      0.013659925200045109,
+      0.036106519401073456,
+      0.014328142628073692,
+      0.03244594484567642,
+      0.009032469242811203,
+      0.018440525978803635,
+      0.01920383982360363,
+      0.020473865792155266,
+      0.032765328884124756,
+      0.014637406915426254,
+      0.01448108535259962,
+      0.012129629030823708,
+      0.011150146834552288,
+      0.031212156638503075,
+      0.02721549943089485,
+      0.014894500374794006,
+      0.019399020820856094,
+      0.020370926707983017,
+      0.03513526916503906,
+      0.013435397297143936,
+      0.02822204679250717,
+      0.009487655013799667,
+      0.015622190199792385,
+      0.017895258963108063,
+      0.016876041889190674,
+      0.013383161276578903,
+      0.021335359662771225,
+      0.030914466828107834,
+      0.010580360889434814,
+      0.011451056227087975,
+      0.007803402841091156,
+      0.012233724817633629,
+      0.040140219032764435,
+      0.019514363259077072,
+      0.010945022106170654,
+      0.015346281230449677,
+      0.015768669545650482,
+      0.014031963422894478,
+      0.012113187462091446,
+      0.018322959542274475,
+      0.008950436487793922,
+      0.008387358859181404,
+      0.01833055168390274,
+      0.009781958535313606,
+      0.03388731926679611,
+      0.023594915866851807,
+      0.02366054803133011,
+      0.014914799481630325,
+      0.01527368277311325,
+      0.013935800641775131,
+      0.01937665417790413,
+      0.026555389165878296,
+      0.013119127601385117,
+      0.017327692359685898,
+      0.01281806081533432,
+      0.03115082159638405,
+      0.01274685375392437,
+      0.024018671363592148,
+      0.01721969060599804,
+      0.008875923231244087,
+      0.0186840258538723,
+      0.03127584606409073,
+      0.03355969488620758,
+      0.02109907940030098,
+      0.02521275356411934,
+      0.014678552746772766,
+      0.015803975984454155,
+      0.010580860078334808,
+      0.01446072943508625,
+      0.039702266454696655,
+      0.04098086804151535,
+      0.017026258632540703,
+      0.04800141602754593,
+      0.019417084753513336,
+      0.021352805197238922,
+      0.0277378149330616,
+      0.025049306452274323,
+      0.011761985719203949,
+      0.02068621665239334,
+      0.025553595274686813,
+      0.0071304067969322205,
+      0.016906581819057465,
+      0.01546161063015461,
+      0.010241236537694931,
+      0.012455351650714874,
+      0.015299420803785324,
+      0.019323330372571945,
+      0.013211840763688087,
+      0.013413187116384506,
+      0.024162352085113525,
+      0.01807761937379837,
+      0.013908561319112778,
+      0.03093879669904709,
+      0.0144495889544487,
+      0.019718807190656662,
+      0.00844704918563366,
+      0.028906114399433136,
+      0.010804221034049988,
+      0.009488066658377647,
+      0.012067637406289577,
+      0.015040576457977295,
+      0.007470171898603439,
+      0.014873256906867027,
+      0.010988231748342514,
+      0.012767143547534943,
+      0.016154780983924866,
+      0.010871181264519691,
+      0.017987877130508423,
+      0.012239239178597927,
+      0.012328235432505608,
+      0.021448515355587006,
+      0.02059098333120346,
+      0.015664558857679367,
+      0.01218429021537304,
+      0.007261655293405056,
+      0.017222527414560318,
+      0.011581387370824814,
+      0.010656767524778843,
+      0.010246491059660912,
+      0.018553227186203003,
+      0.03282719478011131,
+      0.0162493996322155,
+      0.03963233903050423,
+      0.012939762324094772,
+      0.020252341404557228,
+      0.013496289029717445,
+      0.00857863575220108,
+      0.010840233415365219,
+      0.014845713973045349,
+      0.007054360583424568,
+      0.01816418580710888,
+      0.02891641855239868,
+      0.010288992896676064,
+      0.00713920034468174,
+      0.012660834938287735,
+      0.015461052767932415,
+      0.015566255897283554,
+      0.018684573471546173,
+      0.012360978871583939,
+      0.005702901631593704,
+      0.007907480001449585,
+      0.009306533262133598,
+      0.02719968557357788,
+      0.010554085485637188,
+      0.018172377720475197,
+      0.007528157904744148,
+      0.01175107154995203,
+      0.011585632339119911,
+      0.014379439875483513,
+      0.008955443277955055,
+      0.010577559471130371,
+      0.007796976715326309,
+      0.014662452042102814,
+      0.020020950585603714,
+      0.0077422186732292175,
+      0.013598084449768066,
+      0.009638023562729359,
+      0.010541138239204884,
+      0.007165942341089249,
+      0.010295277461409569,
+      0.007796257734298706,
+      0.006619937717914581,
+      0.020668882876634598,
+      0.01578902080655098,
+      0.007280856370925903,
+      0.011586855165660381,
+      0.011592477560043335,
+      0.007128148805350065,
+      0.015206173062324524,
+      0.014938851818442345,
+      0.00543565209954977,
+      0.008509164676070213,
+      0.024902423843741417,
+      0.014712754637002945,
+      0.018012208864092827,
+      0.008784620091319084,
+      0.010731510818004608,
+      0.013733292929828167,
+      0.00869428925216198,
+      0.022556070238351822,
+      0.02647048607468605,
+      0.013982734642922878,
+      0.015525490045547485,
+      0.009201617911458015,
+      0.007094781845808029,
+      0.015642520040273666,
+      0.009298092685639858,
+      0.014537567272782326,
+      0.00825614295899868,
+      0.009203104302287102,
+      0.011141359806060791,
+      0.008821116760373116,
+      0.008776454254984856,
+      0.009878752753138542,
+      0.009446032345294952,
+      0.012018155306577682,
+      0.017650537192821503,
+      0.009024819359183311,
+      0.009477643296122551,
+      0.013601088896393776,
+      0.012014838866889477,
+      0.021504897624254227,
+      0.009149780496954918,
+      0.007894977927207947,
+      0.01129077933728695,
+      0.010833574458956718,
+      0.014918159693479538,
+      0.00855458527803421,
+      0.00736504141241312,
+      0.008928017690777779,
+      0.016933809965848923,
+      0.03210865706205368,
+      0.01579834707081318,
+      0.009920194745063782,
+      0.010861337184906006,
+      0.018204521387815475,
+      0.010070055723190308,
+      0.006747391074895859,
+      0.012954551726579666,
+      0.014359045773744583,
+      0.009734511375427246,
+      0.01635069027543068,
+      0.016396958380937576,
+      0.028392184525728226,
+      0.005876493640244007,
+      0.023259446024894714,
+      0.008534284308552742,
+      0.010509041137993336,
+      0.010909554548561573,
+      0.023324768990278244,
+      0.014119546860456467,
+      0.005262024700641632,
+      0.01122361421585083,
+      0.008188644424080849,
+      0.016334168612957,
+      0.018166374415159225,
+      0.011789614334702492,
+      0.008024140261113644,
+      0.013983141630887985,
+      0.014104964211583138,
+      0.01138586550951004,
+      0.013332624919712543,
+      0.014329064637422562,
+      0.015091532841324806,
+      0.010317517444491386,
+      0.015318581834435463,
+      0.019480697810649872,
+      0.013706950470805168,
+      0.00945664569735527
+    ],
+    "denoise_loss": [],
+    "ortho_loss": [
+      0.7847321629524231,
+      0.6201837658882141,
+      0.485784113407135,
+      0.35137468576431274,
+      0.2879585325717926,
+      0.24915075302124023,
+      0.21809658408164978,
+      0.16125144064426422,
+      0.13464920222759247,
+      0.11504802852869034,
+      0.10673171281814575,
+      0.1009758859872818,
+      0.10426314175128937,
+      0.10648877918720245,
+      0.10758698731660843,
+      0.10554790496826172,
+      0.09970692545175552,
+      0.10068698972463608,
+      0.1037362664937973,
+      0.11161494255065918,
+      0.11507049202919006,
+      0.1317492574453354,
+      0.14284595847129822,
+      0.14627856016159058,
+      0.15046416223049164,
+      0.1489173024892807,
+      0.15455691516399384,
+      0.15504224598407745,
+      0.15456973016262054,
+      0.15529467165470123,
+      0.15482097864151,
+      0.1637507677078247,
+      0.16011153161525726,
+      0.1651749312877655,
+      0.1716211438179016,
+      0.17574329674243927,
+      0.15954194962978363,
+      0.1582264006137848,
+      0.15588511526584625,
+      0.1600116342306137,
+      0.15758398175239563,
+      0.1619112193584442,
+      0.16302575170993805,
+      0.15522991120815277,
+      0.16260525584220886,
+      0.15874461829662323,
+      0.15806731581687927,
+      0.16073763370513916,
+      0.16050544381141663,
+      0.16037650406360626,
+      0.1638392060995102,
+      0.16640426218509674,
+      0.16966049373149872,
+      0.17420445382595062,
+      0.17574597895145416,
+      0.1799435168504715,
+      0.18488837778568268,
+      0.19086268544197083,
+      0.19754229485988617,
+      0.20065085589885712,
+      0.20717096328735352,
+      0.22641225159168243,
+      0.22547413408756256,
+      0.2379624992609024,
+      0.24835960566997528,
+      0.2551441490650177,
+      0.26230356097221375,
+      0.26771965622901917,
+      0.27327385544776917,
+      0.27560171484947205,
+      0.2836180627346039,
+      0.28607603907585144,
+      0.28698083758354187,
+      0.28780463337898254,
+      0.28826531767845154,
+      0.29221728444099426,
+      0.29222774505615234,
+      0.3024168312549591,
+      0.3035641312599182,
+      0.30658817291259766,
+      0.3131972849369049,
+      0.3128298223018646,
+      0.3272259831428528,
+      0.3372877538204193,
+      0.344853013753891,
+      0.347524493932724,
+      0.34524956345558167,
+      0.3452718257904053,
+      0.34162092208862305,
+      0.34223735332489014,
+      0.3512662351131439,
+      0.349275141954422,
+      0.35955575108528137,
+      0.36343318223953247,
+      0.3586910367012024,
+      0.35721367597579956,
+      0.3625499904155731,
+      0.3687310218811035,
+      0.37282317876815796,
+      0.37235260009765625,
+      0.3716725707054138,
+      0.37703198194503784,
+      0.38177719712257385,
+      0.38606080412864685,
+      0.39012834429740906,
+      0.39147788286209106,
+      0.3895938992500305,
+      0.39241254329681396,
+      0.3979438841342926,
+      0.4000576436519623,
+      0.39591819047927856,
+      0.3956204056739807,
+      0.3989535868167877,
+      0.400198370218277,
+      0.4030740261077881,
+      0.4027870297431946,
+      0.4019368290901184,
+      0.4043729305267334,
+      0.40525487065315247,
+      0.4084034562110901,
+      0.4060479998588562,
+      0.40902647376060486,
+      0.40762025117874146,
+      0.40840235352516174,
+      0.41167858242988586,
+      0.4119378924369812,
+      0.4134664833545685,
+      0.41601672768592834,
+      0.42163756489753723,
+      0.42329949140548706,
+      0.4231146574020386,
+      0.4070468544960022,
+      0.410058856010437,
+      0.41018280386924744,
+      0.4142322242259979,
+      0.4147290587425232,
+      0.41209307312965393,
+      0.41136229038238525,
+      0.41243118047714233,
+      0.41491344571113586,
+      0.4101180136203766,
+      0.41006335616111755,
+      0.40823736786842346,
+      0.4069826006889343,
+      0.409622460603714,
+      0.41260066628456116,
+      0.4101448953151703,
+      0.41018062829971313,
+      0.4115578234195709,
+      0.4015924334526062,
+      0.40146732330322266,
+      0.4005260467529297,
+      0.40929973125457764,
+      0.41000333428382874,
+      0.4101176857948303,
+      0.41058748960494995,
+      0.41650286316871643,
+      0.41580337285995483,
+      0.4170321226119995,
+      0.4204050302505493,
+      0.4221915900707245,
+      0.41944631934165955,
+      0.41892117261886597,
+      0.4185391068458557,
+      0.4206126928329468,
+      0.4234926998615265,
+      0.419546514749527,
+      0.4183448553085327,
+      0.4203522801399231,
+      0.42284834384918213,
+      0.42296749353408813,
+      0.42207303643226624,
+      0.42741724848747253,
+      0.43038639426231384,
+      0.4332205653190613,
+      0.43608179688453674,
+      0.43772828578948975,
+      0.4389691650867462,
+      0.4399281144142151,
+      0.43813154101371765,
+      0.43532952666282654,
+      0.43657514452934265,
+      0.43844500184059143,
+      0.4404738247394562,
+      0.44257253408432007,
+      0.4390065670013428,
+      0.4419112503528595,
+      0.44427263736724854,
+      0.4463488757610321,
+      0.4405331611633301,
+      0.4419408142566681,
+      0.44285067915916443,
+      0.443846732378006,
+      0.4430800676345825,
+      0.44508397579193115,
+      0.4468174874782562,
+      0.4474944770336151,
+      0.4521268308162689,
+      0.4507824778556824,
+      0.4480542242527008,
+      0.4508533179759979,
+      0.45075565576553345,
+      0.4510415196418762,
+      0.45447415113449097,
+      0.45526012778282166,
+      0.4589647948741913,
+      0.45821577310562134,
+      0.4597731828689575,
+      0.4576256275177002,
+      0.458015501499176,
+      0.4577544033527374,
+      0.45788660645484924,
+      0.46253782510757446,
+      0.4643017053604126,
+      0.46533340215682983,
+      0.4694773554801941,
+      0.4710234999656677,
+      0.474755197763443,
+      0.4747978448867798,
+      0.47470852732658386,
+      0.4768957793712616,
+      0.4764065742492676,
+      0.4771294593811035,
+      0.4750293493270874,
+      0.4773629903793335,
+      0.4821653664112091,
+      0.4824097156524658,
+      0.48181289434432983,
+      0.48094648122787476,
+      0.48161426186561584,
+      0.4826149046421051,
+      0.48084303736686707,
+      0.4804546535015106,
+      0.4814865291118622,
+      0.48577627539634705,
+      0.48592525720596313,
+      0.48632577061653137,
+      0.4861513376235962,
+      0.48470595479011536,
+      0.4860209822654724,
+      0.48682039976119995,
+      0.4872964024543762,
+      0.4871342182159424,
+      0.48852646350860596,
+      0.4895513355731964,
+      0.48887261748313904,
+      0.4911658465862274,
+      0.4918389618396759,
+      0.49207761883735657,
+      0.4916519820690155,
+      0.49232858419418335,
+      0.49125486612319946,
+      0.49178799986839294,
+      0.4918094277381897,
+      0.4970461130142212,
+      0.4971783757209778,
+      0.4979783892631531,
+      0.4978001117706299,
+      0.49840614199638367,
+      0.4982592761516571,
+      0.4988299310207367,
+      0.4990873634815216,
+      0.4977569282054901,
+      0.4987112283706665,
+      0.4936622381210327,
+      0.4934949278831482,
+      0.4944172501564026,
+      0.4960152506828308,
+      0.4952702224254608,
+      0.49674108624458313,
+      0.4976678788661957,
+      0.49840182065963745,
+      0.5009401440620422,
+      0.5002650618553162,
+      0.5008472800254822,
+      0.5004611015319824,
+      0.500277042388916,
+      0.5007216930389404,
+      0.5007331371307373,
+      0.5008963942527771,
+      0.5007463097572327,
+      0.5002378821372986,
+      0.5010074377059937,
+      0.5006035566329956,
+      0.5012862682342529,
+      0.5009413957595825,
+      0.5010911822319031,
+      0.5015745759010315,
+      0.501132607460022,
+      0.5010624527931213,
+      0.501613199710846,
+      0.5014798045158386,
+      0.5017808675765991,
+      0.5021051168441772,
+      0.5025299191474915,
+      0.5023022890090942,
+      0.5030853748321533,
+      0.5032898783683777,
+      0.5033749938011169,
+      0.5042749643325806,
+      0.5036572813987732,
+      0.5036152601242065,
+      0.5036109089851379,
+      0.504612386226654,
+      0.5043815970420837,
+      0.5043673515319824,
+      0.5048116445541382,
+      0.5051637291908264,
+      0.5053876638412476,
+      0.5054917335510254
+    ],
+    "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": [
+      0.6,
+      1.0,
+      1.0,
+      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": 100
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 605
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 95
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 204
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 169
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 31
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 296
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 0.99,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 163
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 130
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 87
+          },
+          "UC": {
+            "accuracy": 0.9950738916256158,
+            "count": 203
+          },
+          "US": {
+            "accuracy": 0.9914529914529915,
+            "count": 117
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 0.95,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9917355371900827,
+            "count": 121
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 49
+          },
+          "UC": {
+            "accuracy": 0.978494623655914,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 0.87,
+        "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": 0.91875,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 0.996742671009772,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 0.61,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.64,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.93,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 0.6,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.6,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.878,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 0.99,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9977628635346756,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 0.996875,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 0.98,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.9896373056994818,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 0.9906542056074766,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 0.99,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.99609375,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 0.91,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.9705882352941176,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 0.979381443298969,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 0.99,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.9972677595628415,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 234
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 0.98,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9966722129783694,
+            "count": 601
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 99
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 0.99,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 279
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 145
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 24
+          },
+          "UB": {
+            "accuracy": 0.996031746031746,
+            "count": 252
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 213
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 113
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 85
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 181
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 108
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 0.96,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 0.9731543624161074,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 0.65,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.67,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.95,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 0.48,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.75,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.8725,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 0.995,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 600
+          },
+          "MB": {
+            "accuracy": 0.9962546816479401,
+            "count": 267
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 53
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 0.86,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9966666666666667,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9543147208121827,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 0.9514563106796117,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 0.92,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9676113360323887,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 0.9803921568627451,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 0.98,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9966442953020134,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 0.995049504950495,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 0.9033333333333333,
+      "total_examples": 2400,
+      "n_splits": 22
+    }
+  },
+  "sorl_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": 4,
+      "mode": "sorl",
+      "n_digits": 6,
+      "n_per_split": 100
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 605
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 95
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 204
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 169
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 31
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 296
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 163
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 130
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 87
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 203
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 117
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 49
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 104
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 106
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 23
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 234
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 601
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 99
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 279
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 145
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 24
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 252
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 213
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 113
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 85
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 181
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 108
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 600
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 267
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 53
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 1.0,
+      "total_examples": 2400,
+      "n_splits": 22
+    }
+  },
+  "sorl_overall_accuracy": 1.0,
+  "sft_overall_accuracy": 0.9033333333333333
+}

add_sub_sorl_v1_abs10_100K/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7efc43879466cf8cb61bccdd2456daca0af63923b61b39df6ff053d15d3d7d91
+size 650303660

add_sub_sorl_v1_abs10_100K/train_config.json

@@ -0,0 +1,84 @@
+{
+  "num_rollouts": 4,
+  "K": 4,
+  "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_K4_100K",
+  "eval_K": 4,
+  "alpha_traj": 0.0,
+  "corrupt_method": "shuffle",
+  "corrupt_ratio": 0.3,
+  "alpha_contrastive": 1.0,
+  "gamma_contrastive": 0.5,
+  "alpha_masked_traj": 0.0,
+  "mask_nl_ratio": 0.3,
+  "mask_nl_mode": "fixed",
+  "mask_nl_fixed_id": 0,
+  "use_ste": true,
+  "n_inner": 1,
+  "random_K": null,
+  "strip_suffix": null,
+  "compress_prefix": null,
+  "random_mem_span": null,
+  "warmup_ratio": 0.03,
+  "beta2": 0.999,
+  "seed": 42,
+  "n_digits": 6,
+  "n_layer": 2,
+  "n_head": 3,
+  "n_embd": 510,
+  "ops": "add_sub",
+  "abs_vocab": 10,
+  "dataset_size": 100000,
+  "mode": "sorl",
+  "device": "cuda",
+  "push_to_hub": true,
+  "no_wandb": false,
+  "n_params": 162499262,
+  "run_name": "add_sub_sorl_v1_abs10_100K",
+  "git_commit": "f447da529caceac8c7d256cbb2cd185cbc50feac",
+  "timestamp": "2026-04-12T13:33:48.858792+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": "ym82ypoo",
+  "wandb_url": "https://wandb.ai/nlp_and_interpretability/sorl-arithmetic/runs/ym82ypoo",
+  "final_accuracy": 1.0,
+  "sft_accuracy": 0.9033333333333333,
+  "eval_method": "ArithmeticEvaluator"
+}