Upload add_sub_sorl_v1_abs50_100K

add_sub_sorl_v1_abs50_100K/config.json

@@ -0,0 +1,37 @@
+{
+  "architectures": [
+    "SorlModelWrapper"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": null,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 510,
+  "initializer_range": 0.02,
+  "intermediate_size": 2040,
+  "layer_types": [
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 128,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 3,
+  "num_hidden_layers": 2,
+  "num_key_value_heads": 3,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "transformers_version": "5.5.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151694
+}

add_sub_sorl_v1_abs50_100K/generation_config.json

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

add_sub_sorl_v1_abs50_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": [
+      15.902643203735352,
+      11.699418067932129,
+      8.803803443908691,
+      7.566174030303955,
+      4.88871955871582,
+      3.483307123184204,
+      2.9673330783843994,
+      2.5199191570281982,
+      2.359388828277588,
+      1.899479627609253,
+      2.3249804973602295,
+      1.472941279411316,
+      1.3966615200042725,
+      0.7510354518890381,
+      0.8818374872207642,
+      0.6415048837661743,
+      0.2329241931438446,
+      -1.0738698244094849,
+      -2.200684070587158,
+      -3.7577319145202637,
+      -5.699623107910156,
+      -7.358343124389648,
+      -9.194567680358887,
+      -9.076865196228027,
+      -10.468391418457031,
+      -10.008020401000977,
+      -11.121683120727539,
+      -10.967080116271973,
+      -11.859116554260254,
+      -12.660564422607422,
+      -12.601607322692871,
+      -13.966347694396973,
+      -14.086410522460938,
+      -13.412335395812988,
+      -14.228898048400879,
+      -13.165425300598145,
+      -14.058863639831543,
+      -14.419401168823242,
+      -14.229293823242188,
+      -13.882869720458984,
+      -13.9789457321167,
+      -14.555558204650879,
+      -14.0433988571167,
+      -13.979240417480469,
+      -13.299639701843262,
+      -14.608553886413574,
+      -14.117591857910156,
+      -14.673713684082031,
+      -14.202105522155762,
+      -14.54289722442627,
+      -14.768146514892578,
+      -14.255924224853516,
+      -14.07961654663086,
+      -14.165740013122559,
+      -14.195342063903809,
+      -13.844704627990723,
+      -13.554451942443848,
+      -13.777854919433594,
+      -14.790366172790527,
+      -14.977351188659668,
+      -14.2539644241333,
+      -14.452176094055176,
+      -13.604500770568848,
+      -14.904753684997559,
+      -14.828282356262207,
+      -14.090341567993164,
+      -14.200688362121582,
+      -13.650691032409668,
+      -14.073995590209961,
+      -13.802224159240723,
+      -13.814070701599121,
+      -13.819828987121582,
+      -14.0003023147583,
+      -14.25842571258545,
+      -14.026065826416016,
+      -13.91340446472168,
+      -14.058174133300781,
+      -13.821667671203613,
+      -13.962796211242676,
+      -14.399702072143555,
+      -14.518136978149414,
+      -13.377851486206055,
+      -13.583179473876953,
+      -13.494460105895996,
+      -13.96147346496582,
+      -13.965815544128418,
+      -13.858078002929688,
+      -13.814715385437012,
+      -13.289409637451172,
+      -12.859536170959473,
+      -13.226099014282227,
+      -12.458443641662598,
+      -12.99895191192627,
+      -11.597752571105957,
+      -9.364056587219238,
+      -8.954510688781738,
+      -7.378148078918457,
+      -6.393317699432373,
+      -5.539052963256836,
+      -5.486110210418701,
+      -5.265102863311768,
+      -5.159455299377441,
+      -5.0415449142456055,
+      -4.666078567504883,
+      -4.503330707550049,
+      -4.285207748413086,
+      -4.761104106903076,
+      -4.316300392150879,
+      -4.36053991317749,
+      -4.242866039276123,
+      -4.096912384033203,
+      -3.6605536937713623,
+      -4.064852237701416,
+      -3.8849408626556396,
+      -3.7402796745300293,
+      -3.6939871311187744,
+      -3.523607015609741,
+      -3.4979679584503174,
+      -3.4444966316223145,
+      -3.978267192840576,
+      -3.5759990215301514,
+      -4.023746013641357,
+      -3.0218427181243896,
+      -2.863001585006714,
+      -3.637617826461792,
+      -3.0398430824279785,
+      -3.461866855621338,
+      -3.11603045463562,
+      -2.969511032104492,
+      -2.728440761566162,
+      -2.7764720916748047,
+      -2.940053939819336,
+      -3.051244020462036,
+      -2.4871532917022705,
+      -2.6896591186523438,
+      -2.540485382080078,
+      -2.511063575744629,
+      -2.0412261486053467,
+      -2.4017343521118164,
+      -2.4651193618774414,
+      -2.7034363746643066,
+      -2.403578758239746,
+      -2.633699655532837,
+      -2.240712881088257,
+      -2.4593920707702637,
+      -2.367264986038208,
+      -2.406060218811035,
+      -2.497303009033203,
+      -2.8190865516662598,
+      -2.6294970512390137,
+      -2.728332281112671,
+      -2.1717090606689453,
+      -2.995640993118286,
+      -2.4550960063934326,
+      -1.9697117805480957,
+      -2.655184030532837,
+      -2.454906702041626,
+      -2.5678441524505615,
+      -2.4128735065460205,
+      -2.195981740951538,
+      -2.3848819732666016,
+      -2.210343599319458,
+      -2.582068920135498,
+      -2.196033000946045,
+      -1.967943787574768,
+      -1.9972151517868042,
+      -2.089529275894165,
+      -1.7043302059173584,
+      -2.0483813285827637,
+      -1.7416160106658936,
+      -1.8610892295837402,
+      -2.122062921524048,
+      -2.2287797927856445,
+      -2.1510496139526367,
+      -2.003434419631958,
+      -1.7898396253585815,
+      -2.3351781368255615,
+      -2.1391477584838867,
+      -2.124420642852783,
+      -2.0292270183563232,
+      -1.5877554416656494,
+      -1.8878551721572876,
+      -2.184229850769043,
+      -1.8811503648757935,
+      -1.7677620649337769,
+      -1.7471814155578613,
+      -1.7462050914764404,
+      -1.8536919355392456,
+      -2.1378421783447266,
+      -1.875000238418579,
+      -1.9442965984344482,
+      -1.61328125,
+      -1.7730133533477783,
+      -1.60498046875,
+      -1.8977247476577759,
+      -1.7701810598373413,
+      -1.9182641506195068,
+      -1.8594216108322144,
+      -1.627413272857666,
+      -1.6685473918914795,
+      -1.5593777894973755,
+      -1.8531725406646729,
+      -1.741038203239441,
+      -1.72149658203125,
+      -1.6873160600662231,
+      -1.7354228496551514,
+      -1.3678635358810425,
+      -1.6457551717758179,
+      -1.7738306522369385,
+      -1.7181503772735596,
+      -1.7335387468338013,
+      -1.284615397453308,
+      -1.6187137365341187,
+      -1.4650169610977173,
+      -1.4392532110214233,
+      -1.804498314857483,
+      -1.5744978189468384,
+      -1.5620957612991333,
+      -1.311811923980713,
+      -1.3705101013183594,
+      -1.072982668876648,
+      -1.311877727508545,
+      -1.4170998334884644,
+      -1.4101189374923706,
+      -1.3157600164413452,
+      -1.3691258430480957,
+      -1.5042686462402344,
+      -1.7621362209320068,
+      -1.4440076351165771,
+      -1.480931043624878,
+      -1.4486089944839478,
+      -1.3105967044830322,
+      -1.2870160341262817,
+      -1.6739288568496704,
+      -1.3392441272735596,
+      -1.3649474382400513,
+      -1.27271568775177,
+      -1.4094288349151611,
+      -1.6764694452285767,
+      -1.3893120288848877,
+      -1.4240716695785522,
+      -1.4635239839553833,
+      -1.5566202402114868,
+      -1.1928588151931763,
+      -1.3678373098373413,
+      -1.729019045829773,
+      -1.371423602104187,
+      -1.1183242797851562,
+      -1.2645045518875122,
+      -1.2355653047561646,
+      -1.2753121852874756,
+      -1.1917059421539307,
+      -1.4308631420135498,
+      -1.143221139907837,
+      -1.1318707466125488,
+      -1.1786340475082397,
+      -1.1008967161178589,
+      -1.2262732982635498,
+      -1.1759511232376099,
+      -1.2019951343536377,
+      -1.4008299112319946,
+      -1.0337084531784058,
+      -1.3461097478866577,
+      -1.2970634698867798,
+      -1.4298337697982788,
+      -1.3009860515594482,
+      -1.0590935945510864,
+      -1.0305430889129639,
+      -1.357376217842102,
+      -1.109458088874817,
+      -1.2143073081970215,
+      -0.9586305022239685,
+      -1.138513207435608,
+      -1.0083967447280884,
+      -1.3379303216934204,
+      -1.3722052574157715,
+      -1.1853731870651245,
+      -1.3719193935394287,
+      -1.2580769062042236,
+      -0.8599216938018799,
+      -1.0621342658996582,
+      -1.0587685108184814,
+      -1.0981093645095825,
+      -1.1308614015579224,
+      -0.868450403213501,
+      -1.2680256366729736,
+      -1.1257991790771484,
+      -0.8759847283363342,
+      -1.1675376892089844,
+      -1.3457125425338745,
+      -1.151440978050232,
+      -0.9569621086120605,
+      -0.8542989492416382,
+      -1.124998927116394,
+      -1.0205310583114624,
+      -1.2389376163482666,
+      -1.1499571800231934,
+      -0.9360352754592896,
+      -0.9703195691108704,
+      -1.0978248119354248,
+      -0.9414624571800232,
+      -1.074188232421875,
+      -0.8543161153793335,
+      -0.9937012791633606,
+      -1.123839020729065,
+      -0.9685264825820923,
+      -1.0889569520950317,
+      -1.1003398895263672,
+      -1.2492115497589111,
+      -1.1496220827102661
+    ],
+    "base_loss": [
+      10.600613594055176,
+      8.055359840393066,
+      7.324338436126709,
+      6.316225528717041,
+      5.33962345123291,
+      3.5689940452575684,
+      2.1874196529388428,
+      2.047659158706665,
+      1.899756669998169,
+      1.84453284740448,
+      1.83500075340271,
+      1.820839524269104,
+      1.8651269674301147,
+      1.8551760911941528,
+      1.8548781871795654,
+      1.8376959562301636,
+      1.7977161407470703,
+      1.8850847482681274,
+      1.8461447954177856,
+      1.9221431016921997,
+      1.94983971118927,
+      1.9527097940444946,
+      1.997671365737915,
+      1.8204795122146606,
+      1.8524330854415894,
+      1.7538738250732422,
+      1.8139578104019165,
+      1.677846074104309,
+      1.6989243030548096,
+      1.7410989999771118,
+      1.655508041381836,
+      1.7551666498184204,
+      1.7356339693069458,
+      1.6838182210922241,
+      1.7027080059051514,
+      1.6092866659164429,
+      1.7314802408218384,
+      1.715027093887329,
+      1.6868661642074585,
+      1.6269105672836304,
+      1.6311454772949219,
+      1.668233036994934,
+      1.6192152500152588,
+      1.6262046098709106,
+      1.5721343755722046,
+      1.6828041076660156,
+      1.6095987558364868,
+      1.678329586982727,
+      1.6124516725540161,
+      1.656113624572754,
+      1.6720694303512573,
+      1.6209814548492432,
+      1.6137975454330444,
+      1.618564248085022,
+      1.604408860206604,
+      1.602517008781433,
+      1.5463826656341553,
+      1.5854791402816772,
+      1.6620441675186157,
+      1.6873973608016968,
+      1.6140778064727783,
+      1.6244564056396484,
+      1.55022394657135,
+      1.6899614334106445,
+      1.6684008836746216,
+      1.5978590250015259,
+      1.5932568311691284,
+      1.5419341325759888,
+      1.585222840309143,
+      1.549852967262268,
+      1.5714552402496338,
+      1.5638586282730103,
+      1.591749906539917,
+      1.6070278882980347,
+      1.5776277780532837,
+      1.559006929397583,
+      1.5761176347732544,
+      1.55569589138031,
+      1.5638123750686646,
+      1.629416823387146,
+      1.6225910186767578,
+      1.5065295696258545,
+      1.5395022630691528,
+      1.5159441232681274,
+      1.5644861459732056,
+      1.5675691366195679,
+      1.5587130784988403,
+      1.547330617904663,
+      1.4885733127593994,
+      1.4435430765151978,
+      1.489171028137207,
+      1.4163488149642944,
+      1.459105372428894,
+      1.307129144668579,
+      1.0703612565994263,
+      1.0271738767623901,
+      0.837945282459259,
+      0.7230348587036133,
+      0.6405311822891235,
+      0.628436267375946,
+      0.5976390242576599,
+      0.5854808688163757,
+      0.575773298740387,
+      0.5387369394302368,
+      0.5131381750106812,
+      0.5131796598434448,
+      0.5391896367073059,
+      0.5001967549324036,
+      0.5120877027511597,
+      0.486084520816803,
+      0.46478286385536194,
+      0.4414048194885254,
+      0.4803733229637146,
+      0.46711090207099915,
+      0.4299696087837219,
+      0.42117783427238464,
+      0.4076921343803406,
+      0.4038875997066498,
+      0.40442657470703125,
+      0.4533192217350006,
+      0.40610095858573914,
+      0.4571431577205658,
+      0.34926196932792664,
+      0.32839104533195496,
+      0.4149651527404785,
+      0.35457661747932434,
+      0.39466795325279236,
+      0.3604254424571991,
+      0.34129220247268677,
+      0.3307943344116211,
+      0.341112345457077,
+      0.3433581292629242,
+      0.35467222332954407,
+      0.2954230308532715,
+      0.30940061807632446,
+      0.29190659523010254,
+      0.2927117347717285,
+      0.2889794707298279,
+      0.27745234966278076,
+      0.28490832448005676,
+      0.30999067425727844,
+      0.2780962288379669,
+      0.3009397089481354,
+      0.2622397243976593,
+      0.28423628211021423,
+      0.27033689618110657,
+      0.2779816687107086,
+      0.28508269786834717,
+      0.32521364092826843,
+      0.30002689361572266,
+      0.3127082288265228,
+      0.2539328336715698,
+      0.34114691615104675,
+      0.2819333076477051,
+      0.22665226459503174,
+      0.3118877410888672,
+      0.2800125479698181,
+      0.2952449917793274,
+      0.2765387296676636,
+      0.2516744136810303,
+      0.2749289870262146,
+      0.25773903727531433,
+      0.29442644119262695,
+      0.2525728940963745,
+      0.22910907864570618,
+      0.2300143986940384,
+      0.23990151286125183,
+      0.19912955164909363,
+      0.2360854595899582,
+      0.20227625966072083,
+      0.22241690754890442,
+      0.24504487216472626,
+      0.2558191418647766,
+      0.2501135468482971,
+      0.24015279114246368,
+      0.2100088745355606,
+      0.26677006483078003,
+      0.24530638754367828,
+      0.24426667392253876,
+      0.2331373691558838,
+      0.19073544442653656,
+      0.22259521484375,
+      0.24970708787441254,
+      0.2161264270544052,
+      0.2031947523355484,
+      0.21564003825187683,
+      0.20861132442951202,
+      0.21454143524169922,
+      0.24643942713737488,
+      0.21691389381885529,
+      0.22348442673683167,
+      0.18823209404945374,
+      0.2069559544324875,
+      0.18679584562778473,
+      0.21719010174274445,
+      0.20488974452018738,
+      0.2214232236146927,
+      0.21512725949287415,
+      0.19134248793125153,
+      0.19260181486606598,
+      0.18412649631500244,
+      0.21456657350063324,
+      0.2016887664794922,
+      0.19858768582344055,
+      0.19533206522464752,
+      0.20168212056159973,
+      0.16193492710590363,
+      0.19368301331996918,
+      0.2038613110780716,
+      0.19840362668037415,
+      0.19922426342964172,
+      0.1515156775712967,
+      0.18695905804634094,
+      0.16931875050067902,
+      0.166876882314682,
+      0.2077115923166275,
+      0.18261513113975525,
+      0.19456270337104797,
+      0.15236473083496094,
+      0.16070771217346191,
+      0.12959973514080048,
+      0.15200302004814148,
+      0.16590280830860138,
+      0.16575734317302704,
+      0.15350034832954407,
+      0.1595541387796402,
+      0.17597787082195282,
+      0.20311801135540009,
+      0.1688549518585205,
+      0.1727948933839798,
+      0.1690131574869156,
+      0.15285637974739075,
+      0.16618596017360687,
+      0.19473372399806976,
+      0.15762819349765778,
+      0.1584198772907257,
+      0.1480090171098709,
+      0.16284768283367157,
+      0.19359400868415833,
+      0.1614084392786026,
+      0.16409209370613098,
+      0.17155098915100098,
+      0.18065497279167175,
+      0.14004693925380707,
+      0.15952065587043762,
+      0.20549999177455902,
+      0.15970775485038757,
+      0.13161496818065643,
+      0.14953382313251495,
+      0.14466509222984314,
+      0.1497855931520462,
+      0.13885840773582458,
+      0.16562673449516296,
+      0.13649068772792816,
+      0.13283535838127136,
+      0.13889345526695251,
+      0.12941350042819977,
+      0.1436983197927475,
+      0.13811363279819489,
+      0.1401999294757843,
+      0.1643994301557541,
+      0.12299828976392746,
+      0.1573929339647293,
+      0.15119539201259613,
+      0.1654321849346161,
+      0.1522522121667862,
+      0.1256517618894577,
+      0.12073888629674911,
+      0.15730193257331848,
+      0.130019873380661,
+      0.14194072782993317,
+      0.11307316273450851,
+      0.13339151442050934,
+      0.11858965456485748,
+      0.15607216954231262,
+      0.16128577291965485,
+      0.13758909702301025,
+      0.15931521356105804,
+      0.14577162265777588,
+      0.10462959855794907,
+      0.12486211210489273,
+      0.1258380115032196,
+      0.12735985219478607,
+      0.1331641972064972,
+      0.10633667558431625,
+      0.14881311357021332,
+      0.13217094540596008,
+      0.10351573675870895,
+      0.13661159574985504,
+      0.15756379067897797,
+      0.13544346392154694,
+      0.1127428486943245,
+      0.1042342409491539,
+      0.13166454434394836,
+      0.1192726120352745,
+      0.14546307921409607,
+      0.1336200088262558,
+      0.11115437000989914,
+      0.11539566516876221,
+      0.12923398613929749,
+      0.11169678717851639,
+      0.12668605148792267,
+      0.10387875139713287,
+      0.11678438633680344,
+      0.1315682828426361,
+      0.11594482511281967,
+      0.12925221025943756,
+      0.12967567145824432,
+      0.14690859615802765,
+      0.1365702897310257
+    ],
+    "info_loss": [
+      -0.13031387329101562,
+      -0.2066669464111328,
+      -0.2755417823791504,
+      -0.17154979705810547,
+      -0.24880743026733398,
+      -0.13640594482421875,
+      -0.026891708374023438,
+      -0.05095040798187256,
+      -0.049540162086486816,
+      -0.0891730785369873,
+      -0.045008301734924316,
+      -0.12970876693725586,
+      -0.14089536666870117,
+      -0.20433330535888672,
+      -0.19143426418304443,
+      -0.21313905715942383,
+      -0.2502744197845459,
+      -0.38831961154937744,
+      -0.49465155601501465,
+      -0.6518208980560303,
+      -0.8392997980117798,
+      -0.9978586435317993,
+      -1.17745840549469,
+      -1.1390786170959473,
+      -1.2744295597076416,
+      -1.211988925933838,
+      -1.3272581100463867,
+      -1.2952699661254883,
+      -1.3856745958328247,
+      -1.4671963453292847,
+      -1.4519892930984497,
+      -1.597023367881775,
+      -1.6061583757400513,
+      -1.5312066078186035,
+      -1.6143242120742798,
+      -1.4986706972122192,
+      -1.5996477603912354,
+      -1.6335408687591553,
+      -1.6122527122497559,
+      -1.570066213607788,
+      -1.579962968826294,
+      -1.6396890878677368,
+      -1.5841457843780518,
+      -1.575208306312561,
+      -1.5022691488265991,
+      -1.6440784931182861,
+      -1.5869543552398682,
+      -1.6497031450271606,
+      -1.5939735174179077,
+      -1.6334444284439087,
+      -1.6568560600280762,
+      -1.5998594760894775,
+      -1.581986665725708,
+      -1.5905956029891968,
+      -1.5907115936279297,
+      -1.5560940504074097,
+      -1.5234050750732422,
+      -1.5467263460159302,
+      -1.6563700437545776,
+      -1.6779274940490723,
+      -1.5964152812957764,
+      -1.6166232824325562,
+      -1.5258904695510864,
+      -1.6699669361114502,
+      -1.6600831747055054,
+      -1.5807301998138428,
+      -1.5883736610412598,
+      -1.5307484865188599,
+      -1.5745688676834106,
+      -1.5450665950775146,
+      -1.5481033325195312,
+      -1.547301173210144,
+      -1.5675588846206665,
+      -1.5962055921554565,
+      -1.5689582824707031,
+      -1.5564454793930054,
+      -1.572546124458313,
+      -1.5471404790878296,
+      -1.5624316930770874,
+      -1.6135194301605225,
+      -1.6212087869644165,
+      -1.496882438659668,
+      -1.520615816116333,
+      -1.5109580755233765,
+      -1.5607216358184814,
+      -1.5615829229354858,
+      -1.550378680229187,
+      -1.544190764427185,
+      -1.4874019622802734,
+      -1.4380072355270386,
+      -1.480360984802246,
+      -1.394813060760498,
+      -1.453619360923767,
+      -1.297343373298645,
+      -1.0534307956695557,
+      -1.005746603012085,
+      -0.8305511474609375,
+      -0.7186072468757629,
+      -0.6269640326499939,
+      -0.619566798210144,
+      -0.5929027795791626,
+      -0.5840208530426025,
+      -0.5696229934692383,
+      -0.5284737944602966,
+      -0.5102920532226562,
+      -0.4880736172199249,
+      -0.5378124713897705,
+      -0.48980799317359924,
+      -0.49561572074890137,
+      -0.48140832781791687,
+      -0.46326401829719543,
+      -0.4183923304080963,
+      -0.46459925174713135,
+      -0.44236481189727783,
+      -0.42633071541786194,
+      -0.41896840929985046,
+      -0.40066781640052795,
+      -0.39869606494903564,
+      -0.39415228366851807,
+      -0.45243293046951294,
+      -0.4048759937286377,
+      -0.4549127221107483,
+      -0.34433242678642273,
+      -0.32652151584625244,
+      -0.4136870205402374,
+      -0.3449402451515198,
+      -0.39311540126800537,
+      -0.3551430106163025,
+      -0.33927783370018005,
+      -0.3140522241592407,
+      -0.32230740785598755,
+      -0.338846892118454,
+      -0.3474924564361572,
+      -0.2882997691631317,
+      -0.30800995230674744,
+      -0.29021155834198,
+      -0.2878582179546356,
+      -0.24030187726020813,
+      -0.2739948034286499,
+      -0.2833787798881531,
+      -0.3085416555404663,
+      -0.27547958493232727,
+      -0.3008471131324768,
+      -0.2574741542339325,
+      -0.2813914120197296,
+      -0.26996612548828125,
+      -0.27560850977897644,
+      -0.2844357490539551,
+      -0.32192084193229675,
+      -0.29908621311187744,
+      -0.31156912446022034,
+      -0.24919772148132324,
+      -0.3408443033695221,
+      -0.2812892198562622,
+      -0.22597096860408783,
+      -0.3030546009540558,
+      -0.2796410024166107,
+      -0.2941798269748688,
+      -0.275758296251297,
+      -0.25148695707321167,
+      -0.2745201885700226,
+      -0.2549305260181427,
+      -0.2942931354045868,
+      -0.25231873989105225,
+      -0.2267131805419922,
+      -0.22994287312030792,
+      -0.23979350924491882,
+      -0.19787666201591492,
+      -0.2359994351863861,
+      -0.20195798575878143,
+      -0.21586523950099945,
+      -0.24478033185005188,
+      -0.2553216218948364,
+      -0.24894730746746063,
+      -0.23246027529239655,
+      -0.2087734043598175,
+      -0.26657262444496155,
+      -0.2449118047952652,
+      -0.24412254989147186,
+      -0.232756569981575,
+      -0.18617703020572662,
+      -0.22069711983203888,
+      -0.24964895844459534,
+      -0.21589621901512146,
+      -0.2029203623533249,
+      -0.20315486192703247,
+      -0.2021145224571228,
+      -0.2130969911813736,
+      -0.24579359591007233,
+      -0.21634463965892792,
+      -0.2222777009010315,
+      -0.1872347742319107,
+      -0.2061428278684616,
+      -0.18646681308746338,
+      -0.2170478254556656,
+      -0.2047279179096222,
+      -0.2211330085992813,
+      -0.21506346762180328,
+      -0.18883828818798065,
+      -0.19202345609664917,
+      -0.18082721531391144,
+      -0.2144140899181366,
+      -0.20128116011619568,
+      -0.1985008269548416,
+      -0.19459839165210724,
+      -0.20164254307746887,
+      -0.16043563187122345,
+      -0.1908424198627472,
+      -0.20383040606975555,
+      -0.19826574623584747,
+      -0.19916637241840363,
+      -0.1495133340358734,
+      -0.18693779408931732,
+      -0.1692177951335907,
+      -0.16684888303279877,
+      -0.20739074051380157,
+      -0.1820777952671051,
+      -0.18337459862232208,
+      -0.15228356420993805,
+      -0.16066615283489227,
+      -0.12708906829357147,
+      -0.1518956869840622,
+      -0.16585469245910645,
+      -0.16565538942813873,
+      -0.15341567993164062,
+      -0.159390389919281,
+      -0.17527149617671967,
+      -0.2030913531780243,
+      -0.16879497468471527,
+      -0.1726958453655243,
+      -0.16900011897087097,
+      -0.15276168286800385,
+      -0.15299995243549347,
+      -0.1930725872516632,
+      -0.1576014757156372,
+      -0.15811686217784882,
+      -0.14795033633708954,
+      -0.16282221674919128,
+      -0.1934397965669632,
+      -0.1610974222421646,
+      -0.16405907273292542,
+      -0.17148086428642273,
+      -0.18048837780952454,
+      -0.13926289975643158,
+      -0.1594965010881424,
+      -0.19997486472129822,
+      -0.1596875786781311,
+      -0.13160035014152527,
+      -0.14948993921279907,
+      -0.14455373585224152,
+      -0.14896586537361145,
+      -0.13855019211769104,
+      -0.16561074554920197,
+      -0.13647276163101196,
+      -0.13281981647014618,
+      -0.13887101411819458,
+      -0.12921711802482605,
+      -0.14366890490055084,
+      -0.13803906738758087,
+      -0.14018695056438446,
+      -0.16434404253959656,
+      -0.12230131030082703,
+      -0.15738020837306976,
+      -0.1511777937412262,
+      -0.16526386141777039,
+      -0.1522277593612671,
+      -0.1256326287984848,
+      -0.12071990221738815,
+      -0.15727737545967102,
+      -0.12999925017356873,
+      -0.14192911982536316,
+      -0.11304624378681183,
+      -0.13336944580078125,
+      -0.11857951432466507,
+      -0.15606282651424408,
+      -0.16127531230449677,
+      -0.13757632672786713,
+      -0.1592911183834076,
+      -0.14575687050819397,
+      -0.10454907268285751,
+      -0.12483809143304825,
+      -0.12575353682041168,
+      -0.1273241937160492,
+      -0.13314658403396606,
+      -0.106038898229599,
+      -0.1488012969493866,
+      -0.13211604952812195,
+      -0.10350042581558228,
+      -0.1365896314382553,
+      -0.15755146741867065,
+      -0.13538393378257751,
+      -0.11269137263298035,
+      -0.1029544547200203,
+      -0.13163983821868896,
+      -0.11926034837961197,
+      -0.14545533061027527,
+      -0.13361108303070068,
+      -0.11113712936639786,
+      -0.11538238078355789,
+      -0.1292199194431305,
+      -0.11164633929729462,
+      -0.12667834758758545,
+      -0.10387279838323593,
+      -0.11676009744405746,
+      -0.13150247931480408,
+      -0.1159355640411377,
+      -0.1292443424463272,
+      -0.12966513633728027,
+      -0.14678968489170074,
+      -0.13655707240104675
+    ],
+    "abs_loss": [
+      3.9078216552734375,
+      3.8666019439697266,
+      3.7847700119018555,
+      3.7433247566223145,
+      3.5915701389312744,
+      3.1314046382904053,
+      2.939772605895996,
+      2.8470466136932373,
+      2.801816940307617,
+      2.7607271671295166,
+      2.7303786277770996,
+      2.8693246841430664,
+      2.7541182041168213,
+      2.786644220352173,
+      2.819948196411133,
+      2.7823843955993652,
+      2.878891944885254,
+      2.85870099067688,
+      2.8717331886291504,
+      2.7764925956726074,
+      2.707761287689209,
+      2.735921859741211,
+      2.5215401649475098,
+      2.1574792861938477,
+      1.8795905113220215,
+      1.5430264472961426,
+      1.5506346225738525,
+      1.5533703565597534,
+      1.527092695236206,
+      1.2919174432754517,
+      1.4574626684188843,
+      1.4619550704956055,
+      1.3512728214263916,
+      1.2696706056594849,
+      1.2495325803756714,
+      1.2649794816970825,
+      1.1889044046401978,
+      1.3659778833389282,
+      1.4574460983276367,
+      1.2646446228027344,
+      1.1096229553222656,
+      1.1550531387329102,
+      1.2933039665222168,
+      0.9491195678710938,
+      1.0319973230361938,
+      0.7447066307067871,
+      0.9592666625976562,
+      1.0832123756408691,
+      0.9912242889404297,
+      0.8530040979385376,
+      0.8741682767868042,
+      0.7585433721542358,
+      0.9221714735031128,
+      0.6913226842880249,
+      0.8177131414413452,
+      0.8079922199249268,
+      0.691746711730957,
+      0.7857188582420349,
+      0.6209266185760498,
+      0.7075135707855225,
+      0.4876062870025635,
+      0.604632556438446,
+      0.6973446607589722,
+      0.6192052364349365,
+      0.6596052646636963,
+      0.44709309935569763,
+      0.6409910917282104,
+      0.6577365398406982,
+      0.6532579660415649,
+      0.5378486514091492,
+      0.572986364364624,
+      0.5970069169998169,
+      0.4469451308250427,
+      0.48649829626083374,
+      0.5458441972732544,
+      0.5435513257980347,
+      0.5201212167739868,
+      0.45395076274871826,
+      0.7439022660255432,
+      0.41333064436912537,
+      0.521793007850647,
+      0.5093913078308105,
+      0.5536577701568604,
+      0.5998554229736328,
+      0.48121434450149536,
+      0.4819636940956116,
+      0.403691828250885,
+      0.5651859641075134,
+      0.44529423117637634,
+      0.4128825068473816,
+      0.4012378752231598,
+      0.3511068820953369,
+      0.45841652154922485,
+      0.38696128129959106,
+      0.39995282888412476,
+      0.49512308835983276,
+      0.5003008842468262,
+      0.44825389981269836,
+      0.474168598651886,
+      0.40235957503318787,
+      0.4273836016654968,
+      0.3681586980819702,
+      0.43717366456985474,
+      0.3908582925796509,
+      0.41014623641967773,
+      0.520478367805481,
+      0.30796822905540466,
+      0.412836492061615,
+      0.5451231002807617,
+      0.3523538112640381,
+      0.4003564119338989,
+      0.4728928804397583,
+      0.432488352060318,
+      0.4737667441368103,
+      0.4488798975944519,
+      0.37335819005966187,
+      0.29797255992889404,
+      0.48575296998023987,
+      0.3987618684768677,
+      0.4103950262069702,
+      0.4353116452693939,
+      0.29200854897499084,
+      0.4710025191307068,
+      0.40061938762664795,
+      0.448482871055603,
+      0.2979411482810974,
+      0.3269108533859253,
+      0.42152369022369385,
+      0.3489609360694885,
+      0.4247787594795227,
+      0.2711609899997711,
+      0.36050915718078613,
+      0.47094184160232544,
+      0.3136920928955078,
+      0.42449045181274414,
+      0.35395294427871704,
+      0.3839690089225769,
+      0.24007722735404968,
+      0.35046425461769104,
+      0.37765806913375854,
+      0.3609406650066376,
+      0.32252731919288635,
+      0.3516649603843689,
+      0.2885376513004303,
+      0.39194509387016296,
+      0.32523733377456665,
+      0.41631102561950684,
+      0.3914733827114105,
+      0.30636662244796753,
+      0.3401767611503601,
+      0.3878169655799866,
+      0.24963311851024628,
+      0.42236530780792236,
+      0.39474695920944214,
+      0.400269091129303,
+      0.27627962827682495,
+      0.3362012207508087,
+      0.4149481952190399,
+      0.34324753284454346,
+      0.26619914174079895,
+      0.3912666440010071,
+      0.32785266637802124,
+      0.36352622509002686,
+      0.32938316464424133,
+      0.3147716522216797,
+      0.3593710958957672,
+      0.36116552352905273,
+      0.4445563852787018,
+      0.291462779045105,
+      0.3341894745826721,
+      0.2950514554977417,
+      0.47399190068244934,
+      0.28137409687042236,
+      0.3779450058937073,
+      0.3186456561088562,
+      0.2613189220428467,
+      0.29720720648765564,
+      0.3354172110557556,
+      0.36834996938705444,
+      0.28894710540771484,
+      0.40493836998939514,
+      0.3791912794113159,
+      0.2588992118835449,
+      0.3034725785255432,
+      0.2535718083381653,
+      0.27278196811676025,
+      0.28785043954849243,
+      0.3741530478000641,
+      0.3258816599845886,
+      0.2943769097328186,
+      0.2505449950695038,
+      0.3018356263637543,
+      0.3233311176300049,
+      0.2627747058868408,
+      0.3656883239746094,
+      0.27431416511535645,
+      0.319694459438324,
+      0.35664820671081543,
+      0.24510787427425385,
+      0.21204662322998047,
+      0.3086380660533905,
+      0.22820687294006348,
+      0.3071877360343933,
+      0.24265898764133453,
+      0.30144238471984863,
+      0.36596065759658813,
+      0.302855908870697,
+      0.3670097589492798,
+      0.286008358001709,
+      0.3439257740974426,
+      0.36988365650177,
+      0.23277288675308228,
+      0.22205711901187897,
+      0.31526389718055725,
+      0.3289754390716553,
+      0.2893524169921875,
+      0.2731555700302124,
+      0.30137956142425537,
+      0.3124421238899231,
+      0.3032505512237549,
+      0.28976911306381226,
+      0.3287695050239563,
+      0.3032025098800659,
+      0.30797237157821655,
+      0.25027644634246826,
+      0.37846702337265015,
+      0.2393101304769516,
+      0.36551088094711304,
+      0.29606664180755615,
+      0.2943345010280609,
+      0.3079445958137512,
+      0.3069365918636322,
+      0.18928225338459015,
+      0.31161224842071533,
+      0.26681971549987793,
+      0.34095317125320435,
+      0.2580564320087433,
+      0.37255945801734924,
+      0.2797882556915283,
+      0.2936325669288635,
+      0.21242846548557281,
+      0.37606096267700195,
+      0.32313060760498047,
+      0.28985854983329773,
+      0.30060988664627075,
+      0.26349306106567383,
+      0.23728032410144806,
+      0.24237170815467834,
+      0.3466225862503052,
+      0.262248158454895,
+      0.3313555121421814,
+      0.24145424365997314,
+      0.2583176791667938,
+      0.35590845346450806,
+      0.22335973381996155,
+      0.44185110926628113,
+      0.27968838810920715,
+      0.3198820650577545,
+      0.28301069140434265,
+      0.2559143304824829,
+      0.2832440733909607,
+      0.3193395137786865,
+      0.27422747015953064,
+      0.2709762454032898,
+      0.21882885694503784,
+      0.22499187290668488,
+      0.24804788827896118,
+      0.2699471414089203,
+      0.28384077548980713,
+      0.2389158308506012,
+      0.2748531699180603,
+      0.2611979842185974,
+      0.26433876156806946,
+      0.26748785376548767,
+      0.27537328004837036,
+      0.3269525170326233,
+      0.2332027107477188,
+      0.3006930649280548,
+      0.23841100931167603,
+      0.2991859018802643,
+      0.29406994581222534,
+      0.2334449291229248,
+      0.2323375940322876,
+      0.2632378339767456,
+      0.2660562992095947,
+      0.2756766676902771,
+      0.2961580157279968,
+      0.2223338484764099,
+      0.2187422513961792,
+      0.21723613142967224,
+      0.24943313002586365,
+      0.2962314784526825,
+      0.2896310091018677,
+      0.24327483773231506,
+      0.2072056531906128,
+      0.337163507938385,
+      0.20089191198349,
+      0.3238508403301239,
+      0.29043346643447876,
+      0.3112029433250427,
+      0.2868446111679077,
+      0.20166978240013123,
+      0.2161874771118164,
+      0.23737166821956635,
+      0.2567029893398285,
+      0.22636567056179047,
+      0.333919882774353,
+      0.19082900881767273,
+      0.38755038380622864,
+      0.3178732097148895
+    ],
+    "zipf_loss": [
+      6.214385986328125,
+      5.32406759262085,
+      3.8564059734344482,
+      2.591114044189453,
+      1.678013563156128,
+      0.9652321934700012,
+      0.7548532485961914,
+      0.6970594525337219,
+      0.6748520731925964,
+      0.6706048250198364,
+      0.6670247912406921,
+      0.6622569561004639,
+      0.6650763750076294,
+      0.6605280041694641,
+      0.6593071222305298,
+      0.6569610834121704,
+      0.6500630378723145,
+      0.638371467590332,
+      0.6125134229660034,
+      0.5606845617294312,
+      0.47275885939598083,
+      0.3939410448074341,
+      0.3301907777786255,
+      0.27769359946250916,
+      0.2355114072561264,
+      0.20369206368923187,
+      0.18187572062015533,
+      0.15243616700172424,
+      0.1459963172674179,
+      0.14110854268074036,
+      0.1170307844877243,
+      0.10252369940280914,
+      0.10441293567419052,
+      0.08894531428813934,
+      0.0866827666759491,
+      0.08549686521291733,
+      0.08724188059568405,
+      0.06438370048999786,
+      0.06062450259923935,
+      0.06441779434680939,
+      0.07857652753591537,
+      0.05759456753730774,
+      0.04951252043247223,
+      0.05172610282897949,
+      0.04771823436021805,
+      0.07495627552270889,
+      0.04642573744058609,
+      0.036667607724666595,
+      0.026056334376335144,
+      0.05013240873813629,
+      0.0409291610121727,
+      0.04583451896905899,
+      0.03423535078763962,
+      0.05251985043287277,
+      0.025593411177396774,
+      0.032919563353061676,
+      0.06404099613428116,
+      0.02535686269402504,
+      0.04919612035155296,
+      0.04377426952123642,
+      0.04735066741704941,
+      0.029137253761291504,
+      0.03444506227970123,
+      0.043033648282289505,
+      0.03818806633353233,
+      0.07439252734184265,
+      0.025692259892821312,
+      0.04908574000000954,
+      0.021144278347492218,
+      0.04480339586734772,
+      0.03820859268307686,
+      0.029622916132211685,
+      0.03884103149175644,
+      0.04795222356915474,
+      0.03130391985177994,
+      0.037689074873924255,
+      0.03915674239397049,
+      0.04864725470542908,
+      0.023318255320191383,
+      0.06474263966083527,
+      0.01918068714439869,
+      0.033503539860248566,
+      0.028110641986131668,
+      0.03919079899787903,
+      0.03313414752483368,
+      0.034248415380716324,
+      0.04662713035941124,
+      0.0233431588858366,
+      0.05150822177529335,
+      0.03570423647761345,
+      0.04821627587080002,
+      0.03822769969701767,
+      0.03229536488652229,
+      0.029855959117412567,
+      0.05989418923854828,
+      0.02627028524875641,
+      0.039387792348861694,
+      0.024894418194890022,
+      0.042639803141355515,
+      0.040886059403419495,
+      0.023547116667032242,
+      0.05845678225159645,
+      0.03519462049007416,
+      0.040836598724126816,
+      0.04543702304363251,
+      0.030301017686724663,
+      0.04703367501497269,
+      0.040298864245414734,
+      0.02901679463684559,
+      0.04989728331565857,
+      0.030909691005945206,
+      0.03467531502246857,
+      0.0575178861618042,
+      0.024219483137130737,
+      0.048169784247875214,
+      0.03718308359384537,
+      0.04558166116476059,
+      0.036529719829559326,
+      0.05272334814071655,
+      0.051703374832868576,
+      0.023128744214773178,
+      0.0390365906059742,
+      0.025119362398982048,
+      0.033760469406843185,
+      0.03943914920091629,
+      0.02518836222589016,
+      0.041928377002477646,
+      0.032821983098983765,
+      0.047079045325517654,
+      0.03880929946899414,
+      0.07837331295013428,
+      0.06900592893362045,
+      0.021914267912507057,
+      0.06905221939086914,
+      0.03859081119298935,
+      0.034328341484069824,
+      0.03641020506620407,
+      0.04880528897047043,
+      0.025715112686157227,
+      0.04599444568157196,
+      0.03589567169547081,
+      0.04086793214082718,
+      0.038665153086185455,
+      0.04293541610240936,
+      0.03109125979244709,
+      0.02953575924038887,
+      0.030412163585424423,
+      0.022824294865131378,
+      0.04427170008420944,
+      0.027320725843310356,
+      0.03586902096867561,
+      0.041371818631887436,
+      0.029418550431728363,
+      0.03638826310634613,
+      0.023318560793995857,
+      0.03584618121385574,
+      0.027870723977684975,
+      0.03721432760357857,
+      0.03384622931480408,
+      0.04059351608157158,
+      0.04626419022679329,
+      0.048437267541885376,
+      0.03008340299129486,
+      0.04164348542690277,
+      0.03860174119472504,
+      0.036262135952711105,
+      0.032387636601924896,
+      0.030851230025291443,
+      0.046381473541259766,
+      0.04226874187588692,
+      0.04564111679792404,
+      0.03329605609178543,
+      0.04047981649637222,
+      0.05051535367965698,
+      0.04915095865726471,
+      0.06175360083580017,
+      0.03405715897679329,
+      0.03112226165831089,
+      0.03570321202278137,
+      0.036306604743003845,
+      0.042785484343767166,
+      0.05860166251659393,
+      0.036662742495536804,
+      0.03133822977542877,
+      0.032889679074287415,
+      0.041448868811130524,
+      0.03754384070634842,
+      0.025321388617157936,
+      0.04106614738702774,
+      0.042094629257917404,
+      0.02994142472743988,
+      0.040650829672813416,
+      0.04912589117884636,
+      0.0466143824160099,
+      0.018994512036442757,
+      0.04477697238326073,
+      0.03967338800430298,
+      0.04042107239365578,
+      0.04511641338467598,
+      0.03788059949874878,
+      0.0339040532708168,
+      0.05358113348484039,
+      0.039365921169519424,
+      0.040658168494701385,
+      0.033191531896591187,
+      0.04272434860467911,
+      0.04427221789956093,
+      0.032284997403621674,
+      0.0320112444460392,
+      0.0317109152674675,
+      0.0219123512506485,
+      0.03572501614689827,
+      0.04149950295686722,
+      0.026315785944461823,
+      0.029461124911904335,
+      0.03276225924491882,
+      0.03634945675730705,
+      0.04694950208067894,
+      0.027414850890636444,
+      0.04511871933937073,
+      0.03933146595954895,
+      0.022199109196662903,
+      0.04522397369146347,
+      0.04988035932183266,
+      0.03986883535981178,
+      0.027377299964427948,
+      0.0485374853014946,
+      0.029108263552188873,
+      0.045480526983737946,
+      0.04379906505346298,
+      0.04158465191721916,
+      0.03347010537981987,
+      0.05786935240030289,
+      0.030902042984962463,
+      0.05246037244796753,
+      0.023706108331680298,
+      0.03297300636768341,
+      0.018689725548028946,
+      0.03635568171739578,
+      0.03089042566716671,
+      0.031184107065200806,
+      0.042127594351768494,
+      0.03529534116387367,
+      0.030737370252609253,
+      0.03754604235291481,
+      0.038880299776792526,
+      0.04201637953519821,
+      0.041827209293842316,
+      0.04619868844747543,
+      0.03908207267522812,
+      0.03142523765563965,
+      0.030792005360126495,
+      0.03378566727042198,
+      0.049424830824136734,
+      0.041156038641929626,
+      0.026997443288564682,
+      0.033892180770635605,
+      0.03472912311553955,
+      0.03802477568387985,
+      0.03408318758010864,
+      0.04988682270050049,
+      0.03437255322933197,
+      0.042876578867435455,
+      0.03642142564058304,
+      0.0354897603392601,
+      0.04654018580913544,
+      0.04677622392773628,
+      0.02892228029668331,
+      0.02971148118376732,
+      0.03662301227450371,
+      0.035557832568883896,
+      0.03263894468545914,
+      0.03535591810941696,
+      0.03206000477075577,
+      0.039088472723960876,
+      0.04656688868999481,
+      0.029480742290616035,
+      0.03160722553730011,
+      0.029878996312618256,
+      0.05102085694670677,
+      0.03197754546999931,
+      0.049584321677684784,
+      0.024539010599255562,
+      0.04111644625663757,
+      0.05899637192487717,
+      0.04360661655664444,
+      0.0335746705532074,
+      0.03327043727040291,
+      0.039872907102108,
+      0.05051466077566147,
+      0.04201159626245499,
+      0.027585603296756744,
+      0.04204836115241051,
+      0.0354074202477932,
+      0.03207923471927643,
+      0.036436282098293304,
+      0.032444484531879425,
+      0.03179655596613884,
+      0.03906530886888504,
+      0.03402011841535568,
+      0.034619688987731934,
+      0.0457422100007534,
+      0.05891440436244011,
+      0.03337816148996353,
+      0.03394727408885956,
+      0.05224776268005371,
+      0.040842294692993164,
+      0.0475529283285141,
+      0.033021535724401474,
+      0.047591038048267365
+    ],
+    "denoise_loss": [],
+    "ortho_loss": [
+      0.8366642594337463,
+      0.6618494391441345,
+      0.4207803010940552,
+      0.2298196405172348,
+      0.14970402419567108,
+      0.1140080913901329,
+      0.10427522659301758,
+      0.0827057734131813,
+      0.06960917264223099,
+      0.05503741651773453,
+      0.05058750510215759,
+      0.05038199946284294,
+      0.05585004389286041,
+      0.06042852997779846,
+      0.06079588457942009,
+      0.0625748559832573,
+      0.06289029866456985,
+      0.0685804933309555,
+      0.07292303442955017,
+      0.07979366183280945,
+      0.08727797120809555,
+      0.0949409008026123,
+      0.10431933403015137,
+      0.11506061255931854,
+      0.12235318869352341,
+      0.12447769939899445,
+      0.1303531378507614,
+      0.13381065428256989,
+      0.13906820118427277,
+      0.1462385207414627,
+      0.1518447995185852,
+      0.1605212688446045,
+      0.16572725772857666,
+      0.17025168240070343,
+      0.17526577413082123,
+      0.17968608438968658,
+      0.18079596757888794,
+      0.18310832977294922,
+      0.18538832664489746,
+      0.18738751113414764,
+      0.1883198767900467,
+      0.19103409349918365,
+      0.19147978723049164,
+      0.19242602586746216,
+      0.19312480092048645,
+      0.1932714581489563,
+      0.19615623354911804,
+      0.19566021859645844,
+      0.19523394107818604,
+      0.1961299479007721,
+      0.1973007321357727,
+      0.1989651769399643,
+      0.19842150807380676,
+      0.1993802934885025,
+      0.19886934757232666,
+      0.198268860578537,
+      0.19719672203063965,
+      0.19572870433330536,
+      0.1973220258951187,
+      0.20049990713596344,
+      0.20039710402488708,
+      0.2033097892999649,
+      0.2010970264673233,
+      0.20181842148303986,
+      0.20380006730556488,
+      0.2043500393629074,
+      0.2046346813440323,
+      0.20441186428070068,
+      0.20420598983764648,
+      0.2034817487001419,
+      0.20542271435260773,
+      0.20938640832901,
+      0.21274727582931519,
+      0.21205349266529083,
+      0.21244817972183228,
+      0.21338209509849548,
+      0.21705076098442078,
+      0.2178974449634552,
+      0.21980583667755127,
+      0.2239789366722107,
+      0.22352708876132965,
+      0.22076594829559326,
+      0.22033487260341644,
+      0.22331355512142181,
+      0.22360210120677948,
+      0.2238202840089798,
+      0.2251146286725998,
+      0.22160764038562775,
+      0.21952784061431885,
+      0.22078844904899597,
+      0.22575822472572327,
+      0.22670726478099823,
+      0.22682377696037292,
+      0.2287573516368866,
+      0.2323237657546997,
+      0.2338673323392868,
+      0.23920448124408722,
+      0.24087820947170258,
+      0.24260224401950836,
+      0.24455730617046356,
+      0.2473645806312561,
+      0.251116007566452,
+      0.2551710903644562,
+      0.25852328538894653,
+      0.26045820116996765,
+      0.2623943090438843,
+      0.2626422345638275,
+      0.2616928517818451,
+      0.26257553696632385,
+      0.2634384036064148,
+      0.26375117897987366,
+      0.26420125365257263,
+      0.2640562653541565,
+      0.2633400559425354,
+      0.2652282118797302,
+      0.26459839940071106,
+      0.26582202315330505,
+      0.2638816237449646,
+      0.2658769488334656,
+      0.2657202482223511,
+      0.2675725817680359,
+      0.26992014050483704,
+      0.26918625831604004,
+      0.27417024970054626,
+      0.27616775035858154,
+      0.2779380679130554,
+      0.2797730267047882,
+      0.27854591608047485,
+      0.27968814969062805,
+      0.2813499867916107,
+      0.2842121124267578,
+      0.2838771343231201,
+      0.28416967391967773,
+      0.28685376048088074,
+      0.2878960967063904,
+      0.2890807092189789,
+      0.28659626841545105,
+      0.287469744682312,
+      0.28806453943252563,
+      0.28970998525619507,
+      0.2926311194896698,
+      0.29450523853302,
+      0.2961150109767914,
+      0.2974989116191864,
+      0.2956659495830536,
+      0.29489144682884216,
+      0.2940598726272583,
+      0.29574665427207947,
+      0.296213835477829,
+      0.29641327261924744,
+      0.2989657521247864,
+      0.2979440987110138,
+      0.3012530505657196,
+      0.30332884192466736,
+      0.30361616611480713,
+      0.3021412789821625,
+      0.30142006278038025,
+      0.30223923921585083,
+      0.3061709403991699,
+      0.30908504128456116,
+      0.30920276045799255,
+      0.3092802166938782,
+      0.31009936332702637,
+      0.311633437871933,
+      0.3131062090396881,
+      0.31248563528060913,
+      0.31225261092185974,
+      0.31379827857017517,
+      0.3157505393028259,
+      0.31856241822242737,
+      0.3162039518356323,
+      0.3176734745502472,
+      0.3194302022457123,
+      0.3206985592842102,
+      0.3208954930305481,
+      0.3191499412059784,
+      0.3219895362854004,
+      0.32294604182243347,
+      0.32299721240997314,
+      0.32279571890830994,
+      0.3272733986377716,
+      0.32830968499183655,
+      0.3312630355358124,
+      0.33160510659217834,
+      0.3297317326068878,
+      0.3290911614894867,
+      0.32943281531333923,
+      0.33006271719932556,
+      0.33111897110939026,
+      0.3323107659816742,
+      0.3330026865005493,
+      0.33500078320503235,
+      0.33705705404281616,
+      0.3376968204975128,
+      0.3371988534927368,
+      0.33931219577789307,
+      0.33796313405036926,
+      0.339562326669693,
+      0.34457653760910034,
+      0.3443189561367035,
+      0.34598836302757263,
+      0.3451504707336426,
+      0.34450599551200867,
+      0.3459540009498596,
+      0.34577634930610657,
+      0.3469638228416443,
+      0.34592297673225403,
+      0.3467676639556885,
+      0.34918463230133057,
+      0.3507835865020752,
+      0.35098275542259216,
+      0.3524107038974762,
+      0.3526849150657654,
+      0.3539811074733734,
+      0.3557855486869812,
+      0.3547760546207428,
+      0.35506218671798706,
+      0.35701730847358704,
+      0.35708531737327576,
+      0.3578985035419464,
+      0.35822007060050964,
+      0.3598053455352783,
+      0.3599563539028168,
+      0.35986557602882385,
+      0.3597034513950348,
+      0.360770046710968,
+      0.36094164848327637,
+      0.3624958395957947,
+      0.3637600243091583,
+      0.3650926351547241,
+      0.3679355978965759,
+      0.36923104524612427,
+      0.37015196681022644,
+      0.3716208040714264,
+      0.37223806977272034,
+      0.3734593093395233,
+      0.3734816312789917,
+      0.3747451603412628,
+      0.37534624338150024,
+      0.37503674626350403,
+      0.3756308853626251,
+      0.3763791024684906,
+      0.3763023316860199,
+      0.3765925467014313,
+      0.374718576669693,
+      0.37464314699172974,
+      0.3744320869445801,
+      0.3761724829673767,
+      0.377485066652298,
+      0.37841469049453735,
+      0.3784253001213074,
+      0.37860018014907837,
+      0.3785088360309601,
+      0.3798028826713562,
+      0.3795837163925171,
+      0.3801801800727844,
+      0.38101083040237427,
+      0.3811332583427429,
+      0.38183483481407166,
+      0.38436123728752136,
+      0.3843342363834381,
+      0.3858630657196045,
+      0.3854551613330841,
+      0.38591232895851135,
+      0.3870542347431183,
+      0.387127161026001,
+      0.3882277309894562,
+      0.3888079524040222,
+      0.38914212584495544,
+      0.3892621397972107,
+      0.38910847902297974,
+      0.38890260457992554,
+      0.38967669010162354,
+      0.3896617293357849,
+      0.39047491550445557,
+      0.3920934200286865,
+      0.3925377428531647,
+      0.3925579786300659,
+      0.39288005232810974,
+      0.39340248703956604,
+      0.39419203996658325,
+      0.3946855962276459,
+      0.3945488929748535,
+      0.3940354585647583,
+      0.39542973041534424,
+      0.39583808183670044,
+      0.3957192599773407,
+      0.3959352970123291,
+      0.39639943838119507,
+      0.39707598090171814,
+      0.3968469500541687,
+      0.39755985140800476,
+      0.3978676199913025,
+      0.39900830388069153,
+      0.3994162678718567,
+      0.3995503783226013,
+      0.40007805824279785,
+      0.4004404842853546,
+      0.4010798931121826,
+      0.40130743384361267,
+      0.40163904428482056,
+      0.4020545184612274,
+      0.4020858407020569,
+      0.4020998477935791,
+      0.40228092670440674,
+      0.40249043703079224,
+      0.4024031460285187,
+      0.4025253653526306,
+      0.40290385484695435,
+      0.4028412997722626
+    ],
+    "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": [
+      0.57,
+      0.94,
+      0.94,
+      0.99,
+      0.97,
+      1.0,
+      0.99,
+      1.0,
+      1.0,
+      1.0
+    ]
+  },
+  "final_accuracy": 0.9995833333333334,
+  "sft_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": null,
+      "mode": "sft",
+      "n_digits": 6,
+      "n_per_split": 100
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 0.84,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9735537190082645,
+            "count": 605
+          },
+          "SS": {
+            "accuracy": 0.968421052631579,
+            "count": 95
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 0.86,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9656862745098039,
+            "count": 204
+          },
+          "SC": {
+            "accuracy": 0.9881656804733728,
+            "count": 169
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 31
+          },
+          "UC": {
+            "accuracy": 0.9763513513513513,
+            "count": 296
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 0.6,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9877300613496932,
+            "count": 163
+          },
+          "SC": {
+            "accuracy": 0.9230769230769231,
+            "count": 130
+          },
+          "SS": {
+            "accuracy": 0.9195402298850575,
+            "count": 87
+          },
+          "UC": {
+            "accuracy": 0.8669950738916257,
+            "count": 203
+          },
+          "US": {
+            "accuracy": 0.9914529914529915,
+            "count": 117
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 0.5,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9917355371900827,
+            "count": 121
+          },
+          "SC": {
+            "accuracy": 0.9834710743801653,
+            "count": 121
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 49
+          },
+          "UC": {
+            "accuracy": 0.8010752688172043,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 0.8430493273542601,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 0.52,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9903846153846154,
+            "count": 104
+          },
+          "SC": {
+            "accuracy": 0.9716981132075472,
+            "count": 106
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 23
+          },
+          "UC": {
+            "accuracy": 0.79375,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 0.752442996742671,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 0.44,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 0.99,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.6,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.56,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 0.51,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.58,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.638,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 0.86,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9821029082774049,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 0.99375,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 0.9821428571428571,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 0.9659735349716446,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 0.67,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.844559585492228,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 0.822429906542056,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 0.53,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.8359375,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 0.8402777777777778,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 0.53,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.8464052287581699,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 0.8247422680412371,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 0.47,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.8387978142076503,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 0.8846153846153846,
+            "count": 234
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 0.87,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.978369384359401,
+            "count": 601
+          },
+          "ME": {
+            "accuracy": 0.9696969696969697,
+            "count": 99
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 0.9,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.982078853046595,
+            "count": 279
+          },
+          "MB": {
+            "accuracy": 0.993103448275862,
+            "count": 145
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 24
+          },
+          "UB": {
+            "accuracy": 0.9841269841269841,
+            "count": 252
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 0.57,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9906103286384976,
+            "count": 213
+          },
+          "MB": {
+            "accuracy": 0.9911504424778761,
+            "count": 113
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 85
+          },
+          "UB": {
+            "accuracy": 0.7624309392265194,
+            "count": 181
+          },
+          "UD": {
+            "accuracy": 0.9814814814814815,
+            "count": 108
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 0.25,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 0.9902912621359223,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 0.6510067114093959,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 0.676056338028169,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 0.18,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.41,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.47333333333333333,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 0.32,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.38,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.465,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 0.83,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9866666666666667,
+            "count": 600
+          },
+          "MB": {
+            "accuracy": 0.9887640449438202,
+            "count": 267
+          },
+          "ME": {
+            "accuracy": 0.9811320754716981,
+            "count": 53
+          },
+          "UB": {
+            "accuracy": 0.9521640091116174,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 0.9024390243902439,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 0.6,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9933333333333333,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.8223350253807107,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 0.7864077669902912,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 0.47,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.8340080971659919,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 0.7058823529411765,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 0.34,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.785234899328859,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 0.7178217821782178,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 0.5975,
+      "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": 0.99,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.99,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 234
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 601
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 99
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 279
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 145
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 24
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 252
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 213
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 113
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 85
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 181
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 108
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 600
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 267
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 53
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 0.9995833333333334,
+      "total_examples": 2400,
+      "n_splits": 22
+    }
+  },
+  "sorl_overall_accuracy": 0.9995833333333334,
+  "sft_overall_accuracy": 0.5975
+}

add_sub_sorl_v1_abs50_100K/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5bd150c526b37100df70f046080caac9d1b4124a923943541ae41a3475fa4044
+size 650466940

add_sub_sorl_v1_abs50_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_abs50_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": 50,
+  "dataset_size": 100000,
+  "mode": "sorl",
+  "device": "cuda",
+  "push_to_hub": true,
+  "no_wandb": false,
+  "n_params": 162540062,
+  "run_name": "add_sub_sorl_v1_abs50_100K",
+  "git_commit": "f447da529caceac8c7d256cbb2cd185cbc50feac",
+  "timestamp": "2026-04-12T13:02:58.777416+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": "jy8qmsel",
+  "wandb_url": "https://wandb.ai/nlp_and_interpretability/sorl-arithmetic/runs/jy8qmsel",
+  "final_accuracy": 0.9995833333333334,
+  "sft_accuracy": 0.5975,
+  "eval_method": "ArithmeticEvaluator"
+}