9b-132 / trainer_state.json
furproxy's picture
Upload folder using huggingface_hub
4880605 verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 4.0,
"eval_steps": 500,
"global_step": 5284,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.001514004542013626,
"grad_norm": 4.384895324707031,
"learning_rate": 1.1320754716981132e-08,
"loss": 1.7582024335861206,
"step": 2
},
{
"epoch": 0.003028009084027252,
"grad_norm": 8.81789779663086,
"learning_rate": 3.39622641509434e-08,
"loss": 2.091010570526123,
"step": 4
},
{
"epoch": 0.004542013626040878,
"grad_norm": 6.768231391906738,
"learning_rate": 5.660377358490566e-08,
"loss": 2.225084066390991,
"step": 6
},
{
"epoch": 0.006056018168054504,
"grad_norm": 3.0249898433685303,
"learning_rate": 7.924528301886792e-08,
"loss": 1.693494200706482,
"step": 8
},
{
"epoch": 0.00757002271006813,
"grad_norm": 16.95102882385254,
"learning_rate": 1.0188679245283018e-07,
"loss": 1.8492655754089355,
"step": 10
},
{
"epoch": 0.009084027252081756,
"grad_norm": 4.063567161560059,
"learning_rate": 1.2452830188679246e-07,
"loss": 1.4832412004470825,
"step": 12
},
{
"epoch": 0.010598031794095382,
"grad_norm": 31.606918334960938,
"learning_rate": 1.4716981132075472e-07,
"loss": 1.9249378442764282,
"step": 14
},
{
"epoch": 0.012112036336109008,
"grad_norm": 9.38675308227539,
"learning_rate": 1.6981132075471698e-07,
"loss": 1.7762621641159058,
"step": 16
},
{
"epoch": 0.013626040878122634,
"grad_norm": 6.414663314819336,
"learning_rate": 1.9245283018867924e-07,
"loss": 1.5769001245498657,
"step": 18
},
{
"epoch": 0.01514004542013626,
"grad_norm": 2.645232677459717,
"learning_rate": 2.150943396226415e-07,
"loss": 1.8546013832092285,
"step": 20
},
{
"epoch": 0.016654049962149888,
"grad_norm": 3.0886499881744385,
"learning_rate": 2.3773584905660376e-07,
"loss": 1.8705984354019165,
"step": 22
},
{
"epoch": 0.018168054504163512,
"grad_norm": 3.312413454055786,
"learning_rate": 2.60377358490566e-07,
"loss": 1.6591613292694092,
"step": 24
},
{
"epoch": 0.01968205904617714,
"grad_norm": 14.512092590332031,
"learning_rate": 2.8301886792452833e-07,
"loss": 1.6924680471420288,
"step": 26
},
{
"epoch": 0.021196063588190765,
"grad_norm": 4.4867634773254395,
"learning_rate": 3.056603773584906e-07,
"loss": 1.6047476530075073,
"step": 28
},
{
"epoch": 0.022710068130204392,
"grad_norm": 12.943336486816406,
"learning_rate": 3.2830188679245285e-07,
"loss": 1.809732437133789,
"step": 30
},
{
"epoch": 0.024224072672218017,
"grad_norm": 5.657660007476807,
"learning_rate": 3.509433962264151e-07,
"loss": 1.2130705118179321,
"step": 32
},
{
"epoch": 0.025738077214231644,
"grad_norm": 5.186927795410156,
"learning_rate": 3.7358490566037737e-07,
"loss": 1.8291529417037964,
"step": 34
},
{
"epoch": 0.02725208175624527,
"grad_norm": 3.187878370285034,
"learning_rate": 3.9622641509433963e-07,
"loss": 1.8319251537322998,
"step": 36
},
{
"epoch": 0.028766086298258896,
"grad_norm": 21.289854049682617,
"learning_rate": 4.188679245283019e-07,
"loss": 1.7879869937896729,
"step": 38
},
{
"epoch": 0.03028009084027252,
"grad_norm": 2.181772232055664,
"learning_rate": 4.4150943396226415e-07,
"loss": 1.8917243480682373,
"step": 40
},
{
"epoch": 0.03179409538228615,
"grad_norm": 3.3542346954345703,
"learning_rate": 4.641509433962264e-07,
"loss": 1.4650280475616455,
"step": 42
},
{
"epoch": 0.033308099924299776,
"grad_norm": 7.832064151763916,
"learning_rate": 4.867924528301886e-07,
"loss": 1.4264377355575562,
"step": 44
},
{
"epoch": 0.0348221044663134,
"grad_norm": 4.87370491027832,
"learning_rate": 5.094339622641509e-07,
"loss": 0.8579185009002686,
"step": 46
},
{
"epoch": 0.036336109008327025,
"grad_norm": 39.0529670715332,
"learning_rate": 5.320754716981131e-07,
"loss": 1.4417083263397217,
"step": 48
},
{
"epoch": 0.03785011355034065,
"grad_norm": 2.6229774951934814,
"learning_rate": 5.547169811320755e-07,
"loss": 1.778635025024414,
"step": 50
},
{
"epoch": 0.03936411809235428,
"grad_norm": 7.941841125488281,
"learning_rate": 5.773584905660378e-07,
"loss": 1.0647987127304077,
"step": 52
},
{
"epoch": 0.0408781226343679,
"grad_norm": 2.5582528114318848,
"learning_rate": 6.000000000000001e-07,
"loss": 0.8559340834617615,
"step": 54
},
{
"epoch": 0.04239212717638153,
"grad_norm": 5.4595489501953125,
"learning_rate": 6.226415094339623e-07,
"loss": 0.7988621592521667,
"step": 56
},
{
"epoch": 0.04390613171839516,
"grad_norm": 4.560713768005371,
"learning_rate": 6.452830188679246e-07,
"loss": 1.2707360982894897,
"step": 58
},
{
"epoch": 0.045420136260408785,
"grad_norm": 2.874634027481079,
"learning_rate": 6.679245283018868e-07,
"loss": 1.9832544326782227,
"step": 60
},
{
"epoch": 0.046934140802422405,
"grad_norm": 6.723309516906738,
"learning_rate": 6.905660377358491e-07,
"loss": 1.3568415641784668,
"step": 62
},
{
"epoch": 0.04844814534443603,
"grad_norm": 1.7419151067733765,
"learning_rate": 7.132075471698113e-07,
"loss": 1.6522427797317505,
"step": 64
},
{
"epoch": 0.04996214988644966,
"grad_norm": 1.920200228691101,
"learning_rate": 7.358490566037736e-07,
"loss": 1.7422058582305908,
"step": 66
},
{
"epoch": 0.05147615442846329,
"grad_norm": 1.8585798740386963,
"learning_rate": 7.584905660377358e-07,
"loss": 0.8603143095970154,
"step": 68
},
{
"epoch": 0.05299015897047691,
"grad_norm": 2.7313153743743896,
"learning_rate": 7.811320754716982e-07,
"loss": 0.8468101620674133,
"step": 70
},
{
"epoch": 0.05450416351249054,
"grad_norm": 2.7206826210021973,
"learning_rate": 8.037735849056604e-07,
"loss": 0.8229624629020691,
"step": 72
},
{
"epoch": 0.056018168054504165,
"grad_norm": 2.820694923400879,
"learning_rate": 8.264150943396227e-07,
"loss": 0.8030911684036255,
"step": 74
},
{
"epoch": 0.05753217259651779,
"grad_norm": 2.101903200149536,
"learning_rate": 8.490566037735849e-07,
"loss": 0.6831276416778564,
"step": 76
},
{
"epoch": 0.059046177138531414,
"grad_norm": 3.765256881713867,
"learning_rate": 8.716981132075472e-07,
"loss": 0.7411392331123352,
"step": 78
},
{
"epoch": 0.06056018168054504,
"grad_norm": 1.8168247938156128,
"learning_rate": 8.943396226415094e-07,
"loss": 1.0968608856201172,
"step": 80
},
{
"epoch": 0.06207418622255867,
"grad_norm": 8.313712120056152,
"learning_rate": 9.169811320754717e-07,
"loss": 0.8532919883728027,
"step": 82
},
{
"epoch": 0.0635881907645723,
"grad_norm": 1.691605567932129,
"learning_rate": 9.396226415094339e-07,
"loss": 1.2071592807769775,
"step": 84
},
{
"epoch": 0.06510219530658592,
"grad_norm": 1.4280986785888672,
"learning_rate": 9.622641509433961e-07,
"loss": 1.0525227785110474,
"step": 86
},
{
"epoch": 0.06661619984859955,
"grad_norm": 1.750461459159851,
"learning_rate": 9.849056603773586e-07,
"loss": 0.6976984739303589,
"step": 88
},
{
"epoch": 0.06813020439061317,
"grad_norm": 2.1426968574523926,
"learning_rate": 1.0075471698113208e-06,
"loss": 1.06442391872406,
"step": 90
},
{
"epoch": 0.0696442089326268,
"grad_norm": 2.3825888633728027,
"learning_rate": 1.030188679245283e-06,
"loss": 0.9887061715126038,
"step": 92
},
{
"epoch": 0.07115821347464042,
"grad_norm": 5.285498142242432,
"learning_rate": 1.0528301886792452e-06,
"loss": 1.0551021099090576,
"step": 94
},
{
"epoch": 0.07267221801665405,
"grad_norm": 2.1216835975646973,
"learning_rate": 1.0754716981132076e-06,
"loss": 0.7430307865142822,
"step": 96
},
{
"epoch": 0.07418622255866768,
"grad_norm": 1.883683204650879,
"learning_rate": 1.0981132075471698e-06,
"loss": 0.6745927333831787,
"step": 98
},
{
"epoch": 0.0757002271006813,
"grad_norm": 1.9999110698699951,
"learning_rate": 1.120754716981132e-06,
"loss": 1.1162917613983154,
"step": 100
},
{
"epoch": 0.07721423164269493,
"grad_norm": 1.4555745124816895,
"learning_rate": 1.1433962264150944e-06,
"loss": 1.495133399963379,
"step": 102
},
{
"epoch": 0.07872823618470856,
"grad_norm": 2.054089307785034,
"learning_rate": 1.1660377358490566e-06,
"loss": 0.546085774898529,
"step": 104
},
{
"epoch": 0.08024224072672217,
"grad_norm": 2.632761001586914,
"learning_rate": 1.188679245283019e-06,
"loss": 0.5810890197753906,
"step": 106
},
{
"epoch": 0.0817562452687358,
"grad_norm": 3.3831615447998047,
"learning_rate": 1.2113207547169813e-06,
"loss": 0.9427996873855591,
"step": 108
},
{
"epoch": 0.08327024981074943,
"grad_norm": 3.100712776184082,
"learning_rate": 1.2339622641509435e-06,
"loss": 0.9965349435806274,
"step": 110
},
{
"epoch": 0.08478425435276306,
"grad_norm": 2.1988565921783447,
"learning_rate": 1.2566037735849057e-06,
"loss": 0.9902376532554626,
"step": 112
},
{
"epoch": 0.08629825889477669,
"grad_norm": 7.167949199676514,
"learning_rate": 1.279245283018868e-06,
"loss": 0.5847588777542114,
"step": 114
},
{
"epoch": 0.08781226343679031,
"grad_norm": 2.712419271469116,
"learning_rate": 1.3018867924528303e-06,
"loss": 1.145547866821289,
"step": 116
},
{
"epoch": 0.08932626797880394,
"grad_norm": 4.570695877075195,
"learning_rate": 1.3245283018867925e-06,
"loss": 0.8823922276496887,
"step": 118
},
{
"epoch": 0.09084027252081757,
"grad_norm": 1.8274328708648682,
"learning_rate": 1.3471698113207547e-06,
"loss": 0.8956850171089172,
"step": 120
},
{
"epoch": 0.09235427706283118,
"grad_norm": 1.4670203924179077,
"learning_rate": 1.3698113207547171e-06,
"loss": 0.5994776487350464,
"step": 122
},
{
"epoch": 0.09386828160484481,
"grad_norm": 3.0260169506073,
"learning_rate": 1.3924528301886793e-06,
"loss": 0.608551025390625,
"step": 124
},
{
"epoch": 0.09538228614685844,
"grad_norm": 2.6915009021759033,
"learning_rate": 1.4150943396226415e-06,
"loss": 1.432515263557434,
"step": 126
},
{
"epoch": 0.09689629068887207,
"grad_norm": 1.6062544584274292,
"learning_rate": 1.4377358490566038e-06,
"loss": 1.4172536134719849,
"step": 128
},
{
"epoch": 0.0984102952308857,
"grad_norm": 2.7345874309539795,
"learning_rate": 1.4603773584905662e-06,
"loss": 1.1050211191177368,
"step": 130
},
{
"epoch": 0.09992429977289932,
"grad_norm": 1.3847036361694336,
"learning_rate": 1.4830188679245284e-06,
"loss": 0.807343602180481,
"step": 132
},
{
"epoch": 0.10143830431491295,
"grad_norm": 2.7313413619995117,
"learning_rate": 1.5056603773584906e-06,
"loss": 0.752243161201477,
"step": 134
},
{
"epoch": 0.10295230885692658,
"grad_norm": 2.1813056468963623,
"learning_rate": 1.5283018867924528e-06,
"loss": 0.8742433786392212,
"step": 136
},
{
"epoch": 0.10446631339894019,
"grad_norm": 1.965896487236023,
"learning_rate": 1.5509433962264152e-06,
"loss": 1.0023077726364136,
"step": 138
},
{
"epoch": 0.10598031794095382,
"grad_norm": 1.341701626777649,
"learning_rate": 1.5735849056603774e-06,
"loss": 0.4326239824295044,
"step": 140
},
{
"epoch": 0.10749432248296745,
"grad_norm": 2.332487106323242,
"learning_rate": 1.5962264150943396e-06,
"loss": 1.0556713342666626,
"step": 142
},
{
"epoch": 0.10900832702498107,
"grad_norm": 1.6127592325210571,
"learning_rate": 1.6188679245283018e-06,
"loss": 1.4404288530349731,
"step": 144
},
{
"epoch": 0.1105223315669947,
"grad_norm": 1.9773815870285034,
"learning_rate": 1.6415094339622643e-06,
"loss": 0.7526131868362427,
"step": 146
},
{
"epoch": 0.11203633610900833,
"grad_norm": 4.140753746032715,
"learning_rate": 1.6641509433962265e-06,
"loss": 0.8076651096343994,
"step": 148
},
{
"epoch": 0.11355034065102196,
"grad_norm": 2.5584218502044678,
"learning_rate": 1.6867924528301887e-06,
"loss": 1.0231175422668457,
"step": 150
},
{
"epoch": 0.11506434519303559,
"grad_norm": 5.1710524559021,
"learning_rate": 1.7094339622641509e-06,
"loss": 0.7206178903579712,
"step": 152
},
{
"epoch": 0.1165783497350492,
"grad_norm": 1.3863569498062134,
"learning_rate": 1.7320754716981133e-06,
"loss": 0.8312125205993652,
"step": 154
},
{
"epoch": 0.11809235427706283,
"grad_norm": 5.277348518371582,
"learning_rate": 1.7547169811320755e-06,
"loss": 1.2960776090621948,
"step": 156
},
{
"epoch": 0.11960635881907646,
"grad_norm": 2.767324924468994,
"learning_rate": 1.7773584905660377e-06,
"loss": 1.3986912965774536,
"step": 158
},
{
"epoch": 0.12112036336109008,
"grad_norm": 3.0075347423553467,
"learning_rate": 1.8e-06,
"loss": 1.0369937419891357,
"step": 160
},
{
"epoch": 0.12263436790310371,
"grad_norm": 2.4291977882385254,
"learning_rate": 1.8226415094339623e-06,
"loss": 1.1735918521881104,
"step": 162
},
{
"epoch": 0.12414837244511734,
"grad_norm": 10.382956504821777,
"learning_rate": 1.8452830188679245e-06,
"loss": 0.5223852396011353,
"step": 164
},
{
"epoch": 0.12566237698713095,
"grad_norm": 3.3174524307250977,
"learning_rate": 1.8679245283018868e-06,
"loss": 0.9646264314651489,
"step": 166
},
{
"epoch": 0.1271763815291446,
"grad_norm": 3.2446677684783936,
"learning_rate": 1.890566037735849e-06,
"loss": 0.5242027640342712,
"step": 168
},
{
"epoch": 0.1286903860711582,
"grad_norm": 2.5006420612335205,
"learning_rate": 1.913207547169811e-06,
"loss": 1.368085503578186,
"step": 170
},
{
"epoch": 0.13020439061317185,
"grad_norm": 3.4297969341278076,
"learning_rate": 1.9358490566037734e-06,
"loss": 0.5971487164497375,
"step": 172
},
{
"epoch": 0.13171839515518546,
"grad_norm": 2.7067747116088867,
"learning_rate": 1.958490566037736e-06,
"loss": 1.2271755933761597,
"step": 174
},
{
"epoch": 0.1332323996971991,
"grad_norm": 1.2546025514602661,
"learning_rate": 1.981132075471698e-06,
"loss": 1.4489688873291016,
"step": 176
},
{
"epoch": 0.13474640423921272,
"grad_norm": 6.79306697845459,
"learning_rate": 2.0037735849056604e-06,
"loss": 0.7281686663627625,
"step": 178
},
{
"epoch": 0.13626040878122633,
"grad_norm": 2.328155994415283,
"learning_rate": 2.0264150943396226e-06,
"loss": 0.5355978608131409,
"step": 180
},
{
"epoch": 0.13777441332323997,
"grad_norm": 1.7138910293579102,
"learning_rate": 2.049056603773585e-06,
"loss": 1.3829823732376099,
"step": 182
},
{
"epoch": 0.1392884178652536,
"grad_norm": 1.7628329992294312,
"learning_rate": 2.071698113207547e-06,
"loss": 0.6152011156082153,
"step": 184
},
{
"epoch": 0.14080242240726723,
"grad_norm": 5.342782020568848,
"learning_rate": 2.0943396226415092e-06,
"loss": 0.8484565019607544,
"step": 186
},
{
"epoch": 0.14231642694928084,
"grad_norm": 1.8729349374771118,
"learning_rate": 2.1169811320754715e-06,
"loss": 1.1291441917419434,
"step": 188
},
{
"epoch": 0.14383043149129449,
"grad_norm": 2.3892452716827393,
"learning_rate": 2.139622641509434e-06,
"loss": 0.5965837240219116,
"step": 190
},
{
"epoch": 0.1453444360333081,
"grad_norm": 1.9884214401245117,
"learning_rate": 2.1622641509433963e-06,
"loss": 0.948914647102356,
"step": 192
},
{
"epoch": 0.1468584405753217,
"grad_norm": 3.3960092067718506,
"learning_rate": 2.1849056603773585e-06,
"loss": 0.5472453832626343,
"step": 194
},
{
"epoch": 0.14837244511733536,
"grad_norm": 1.738864541053772,
"learning_rate": 2.2075471698113207e-06,
"loss": 1.108245849609375,
"step": 196
},
{
"epoch": 0.14988644965934897,
"grad_norm": 2.3850369453430176,
"learning_rate": 2.230188679245283e-06,
"loss": 0.9389583468437195,
"step": 198
},
{
"epoch": 0.1514004542013626,
"grad_norm": 1.725578784942627,
"learning_rate": 2.2528301886792455e-06,
"loss": 0.8076001405715942,
"step": 200
},
{
"epoch": 0.15291445874337622,
"grad_norm": 2.7566914558410645,
"learning_rate": 2.2754716981132078e-06,
"loss": 0.9999346733093262,
"step": 202
},
{
"epoch": 0.15442846328538987,
"grad_norm": 1.705634355545044,
"learning_rate": 2.29811320754717e-06,
"loss": 0.9279139041900635,
"step": 204
},
{
"epoch": 0.15594246782740348,
"grad_norm": 9.00720500946045,
"learning_rate": 2.320754716981132e-06,
"loss": 0.987282931804657,
"step": 206
},
{
"epoch": 0.15745647236941712,
"grad_norm": 8.06688117980957,
"learning_rate": 2.3433962264150944e-06,
"loss": 0.6445719599723816,
"step": 208
},
{
"epoch": 0.15897047691143074,
"grad_norm": 6.1968183517456055,
"learning_rate": 2.366037735849057e-06,
"loss": 0.8678835034370422,
"step": 210
},
{
"epoch": 0.16048448145344435,
"grad_norm": 10.432062149047852,
"learning_rate": 2.388679245283019e-06,
"loss": 0.7351512312889099,
"step": 212
},
{
"epoch": 0.161998485995458,
"grad_norm": 1.7967216968536377,
"learning_rate": 2.4113207547169814e-06,
"loss": 1.169299840927124,
"step": 214
},
{
"epoch": 0.1635124905374716,
"grad_norm": 3.4995172023773193,
"learning_rate": 2.4339622641509436e-06,
"loss": 0.5779087543487549,
"step": 216
},
{
"epoch": 0.16502649507948525,
"grad_norm": 4.677812099456787,
"learning_rate": 2.456603773584906e-06,
"loss": 0.6416253447532654,
"step": 218
},
{
"epoch": 0.16654049962149886,
"grad_norm": 2.300534725189209,
"learning_rate": 2.479245283018868e-06,
"loss": 1.3942409753799438,
"step": 220
},
{
"epoch": 0.1680545041635125,
"grad_norm": 1.9117436408996582,
"learning_rate": 2.5018867924528302e-06,
"loss": 0.9447011947631836,
"step": 222
},
{
"epoch": 0.16956850870552612,
"grad_norm": 4.164151191711426,
"learning_rate": 2.5245283018867925e-06,
"loss": 0.5729636549949646,
"step": 224
},
{
"epoch": 0.17108251324753973,
"grad_norm": 2.2320473194122314,
"learning_rate": 2.547169811320755e-06,
"loss": 1.0322617292404175,
"step": 226
},
{
"epoch": 0.17259651778955337,
"grad_norm": 1.339289665222168,
"learning_rate": 2.5698113207547173e-06,
"loss": 0.9149308204650879,
"step": 228
},
{
"epoch": 0.17411052233156699,
"grad_norm": 3.060145616531372,
"learning_rate": 2.5924528301886795e-06,
"loss": 0.5762428045272827,
"step": 230
},
{
"epoch": 0.17562452687358063,
"grad_norm": 3.563155174255371,
"learning_rate": 2.6150943396226417e-06,
"loss": 0.9495343565940857,
"step": 232
},
{
"epoch": 0.17713853141559424,
"grad_norm": 2.58048677444458,
"learning_rate": 2.637735849056604e-06,
"loss": 0.556329607963562,
"step": 234
},
{
"epoch": 0.17865253595760788,
"grad_norm": 14.05665397644043,
"learning_rate": 2.660377358490566e-06,
"loss": 0.45048168301582336,
"step": 236
},
{
"epoch": 0.1801665404996215,
"grad_norm": 4.880133628845215,
"learning_rate": 2.6830188679245283e-06,
"loss": 0.6343849897384644,
"step": 238
},
{
"epoch": 0.18168054504163514,
"grad_norm": 1.243223786354065,
"learning_rate": 2.7056603773584905e-06,
"loss": 0.48857587575912476,
"step": 240
},
{
"epoch": 0.18319454958364875,
"grad_norm": 3.3495898246765137,
"learning_rate": 2.728301886792453e-06,
"loss": 0.9088899493217468,
"step": 242
},
{
"epoch": 0.18470855412566237,
"grad_norm": 4.812596321105957,
"learning_rate": 2.7509433962264154e-06,
"loss": 0.5542827844619751,
"step": 244
},
{
"epoch": 0.186222558667676,
"grad_norm": 1.5049998760223389,
"learning_rate": 2.7735849056603776e-06,
"loss": 1.1015349626541138,
"step": 246
},
{
"epoch": 0.18773656320968962,
"grad_norm": 3.4939844608306885,
"learning_rate": 2.7962264150943398e-06,
"loss": 0.7408594489097595,
"step": 248
},
{
"epoch": 0.18925056775170326,
"grad_norm": 2.150743246078491,
"learning_rate": 2.818867924528302e-06,
"loss": 1.0451587438583374,
"step": 250
},
{
"epoch": 0.19076457229371688,
"grad_norm": 2.757932662963867,
"learning_rate": 2.841509433962264e-06,
"loss": 0.9631202816963196,
"step": 252
},
{
"epoch": 0.19227857683573052,
"grad_norm": 2.339393377304077,
"learning_rate": 2.8641509433962264e-06,
"loss": 1.1843230724334717,
"step": 254
},
{
"epoch": 0.19379258137774413,
"grad_norm": 2.819859743118286,
"learning_rate": 2.8867924528301886e-06,
"loss": 1.3307970762252808,
"step": 256
},
{
"epoch": 0.19530658591975775,
"grad_norm": 2.277029514312744,
"learning_rate": 2.9094339622641512e-06,
"loss": 1.314733862876892,
"step": 258
},
{
"epoch": 0.1968205904617714,
"grad_norm": 1.6182972192764282,
"learning_rate": 2.9320754716981135e-06,
"loss": 0.8982797861099243,
"step": 260
},
{
"epoch": 0.198334595003785,
"grad_norm": 0.9466566443443298,
"learning_rate": 2.9547169811320757e-06,
"loss": 0.49237707257270813,
"step": 262
},
{
"epoch": 0.19984859954579864,
"grad_norm": 2.1412227153778076,
"learning_rate": 2.977358490566038e-06,
"loss": 1.3869421482086182,
"step": 264
},
{
"epoch": 0.20136260408781226,
"grad_norm": 4.314519882202148,
"learning_rate": 3e-06,
"loss": 0.5500906705856323,
"step": 266
},
{
"epoch": 0.2028766086298259,
"grad_norm": 2.2210774421691895,
"learning_rate": 2.9999989421378913e-06,
"loss": 1.1445879936218262,
"step": 268
},
{
"epoch": 0.2043906131718395,
"grad_norm": 1.5420582294464111,
"learning_rate": 2.999995768553224e-06,
"loss": 0.8207805156707764,
"step": 270
},
{
"epoch": 0.20590461771385316,
"grad_norm": 2.379532814025879,
"learning_rate": 2.9999904792509703e-06,
"loss": 1.310062289237976,
"step": 272
},
{
"epoch": 0.20741862225586677,
"grad_norm": 4.023805141448975,
"learning_rate": 2.999983074239421e-06,
"loss": 0.9318987131118774,
"step": 274
},
{
"epoch": 0.20893262679788038,
"grad_norm": 2.1359353065490723,
"learning_rate": 2.99997355353018e-06,
"loss": 0.9912564754486084,
"step": 276
},
{
"epoch": 0.21044663133989402,
"grad_norm": 4.8719563484191895,
"learning_rate": 2.9999619171381696e-06,
"loss": 0.9644961953163147,
"step": 278
},
{
"epoch": 0.21196063588190764,
"grad_norm": 4.153341770172119,
"learning_rate": 2.9999481650816256e-06,
"loss": 0.9800684452056885,
"step": 280
},
{
"epoch": 0.21347464042392128,
"grad_norm": 2.0010852813720703,
"learning_rate": 2.9999322973821e-06,
"loss": 1.0485155582427979,
"step": 282
},
{
"epoch": 0.2149886449659349,
"grad_norm": 3.15071702003479,
"learning_rate": 2.9999143140644616e-06,
"loss": 0.6141320466995239,
"step": 284
},
{
"epoch": 0.21650264950794854,
"grad_norm": 2.1563243865966797,
"learning_rate": 2.9998942151568927e-06,
"loss": 0.68123859167099,
"step": 286
},
{
"epoch": 0.21801665404996215,
"grad_norm": 2.4670424461364746,
"learning_rate": 2.9998720006908934e-06,
"loss": 0.880809485912323,
"step": 288
},
{
"epoch": 0.21953065859197576,
"grad_norm": 2.3046791553497314,
"learning_rate": 2.9998476707012776e-06,
"loss": 0.7388229966163635,
"step": 290
},
{
"epoch": 0.2210446631339894,
"grad_norm": 1.8321624994277954,
"learning_rate": 2.999821225226176e-06,
"loss": 0.9509535431861877,
"step": 292
},
{
"epoch": 0.22255866767600302,
"grad_norm": 2.593620777130127,
"learning_rate": 2.9997926643070335e-06,
"loss": 0.8668114542961121,
"step": 294
},
{
"epoch": 0.22407267221801666,
"grad_norm": 1.3647516965866089,
"learning_rate": 2.999761987988611e-06,
"loss": 0.5078874826431274,
"step": 296
},
{
"epoch": 0.22558667676003027,
"grad_norm": 21.98381805419922,
"learning_rate": 2.9997291963189844e-06,
"loss": 0.6921195983886719,
"step": 298
},
{
"epoch": 0.22710068130204392,
"grad_norm": 1.7081471681594849,
"learning_rate": 2.9996942893495453e-06,
"loss": 1.2570655345916748,
"step": 300
},
{
"epoch": 0.22861468584405753,
"grad_norm": 1.7225868701934814,
"learning_rate": 2.999657267135e-06,
"loss": 0.9504136443138123,
"step": 302
},
{
"epoch": 0.23012869038607117,
"grad_norm": 15.755874633789062,
"learning_rate": 2.9996181297333692e-06,
"loss": 0.9759711027145386,
"step": 304
},
{
"epoch": 0.23164269492808479,
"grad_norm": 1.7268153429031372,
"learning_rate": 2.99957687720599e-06,
"loss": 1.3015202283859253,
"step": 306
},
{
"epoch": 0.2331566994700984,
"grad_norm": 2.12369704246521,
"learning_rate": 2.999533509617513e-06,
"loss": 0.6153337955474854,
"step": 308
},
{
"epoch": 0.23467070401211204,
"grad_norm": 3.8299176692962646,
"learning_rate": 2.9994880270359037e-06,
"loss": 0.9555022120475769,
"step": 310
},
{
"epoch": 0.23618470855412566,
"grad_norm": 4.080057144165039,
"learning_rate": 2.999440429532443e-06,
"loss": 0.5967095494270325,
"step": 312
},
{
"epoch": 0.2376987130961393,
"grad_norm": 2.641312837600708,
"learning_rate": 2.9993907171817255e-06,
"loss": 0.5727935433387756,
"step": 314
},
{
"epoch": 0.2392127176381529,
"grad_norm": 2.667450189590454,
"learning_rate": 2.9993388900616615e-06,
"loss": 0.8149162530899048,
"step": 316
},
{
"epoch": 0.24072672218016655,
"grad_norm": 1.8847966194152832,
"learning_rate": 2.9992849482534736e-06,
"loss": 0.8662533760070801,
"step": 318
},
{
"epoch": 0.24224072672218017,
"grad_norm": 3.9809248447418213,
"learning_rate": 2.9992288918417003e-06,
"loss": 0.9989725947380066,
"step": 320
},
{
"epoch": 0.24375473126419378,
"grad_norm": 2.6532821655273438,
"learning_rate": 2.999170720914192e-06,
"loss": 1.308701992034912,
"step": 322
},
{
"epoch": 0.24526873580620742,
"grad_norm": 4.844236373901367,
"learning_rate": 2.999110435562116e-06,
"loss": 0.4661545157432556,
"step": 324
},
{
"epoch": 0.24678274034822104,
"grad_norm": 1.7858662605285645,
"learning_rate": 2.999048035879951e-06,
"loss": 0.8855746984481812,
"step": 326
},
{
"epoch": 0.24829674489023468,
"grad_norm": 1.814077615737915,
"learning_rate": 2.9989835219654903e-06,
"loss": 1.0955607891082764,
"step": 328
},
{
"epoch": 0.2498107494322483,
"grad_norm": 2.4321069717407227,
"learning_rate": 2.99891689391984e-06,
"loss": 1.3380382061004639,
"step": 330
},
{
"epoch": 0.2513247539742619,
"grad_norm": 4.530355453491211,
"learning_rate": 2.99884815184742e-06,
"loss": 0.5787966251373291,
"step": 332
},
{
"epoch": 0.2528387585162756,
"grad_norm": 2.016003131866455,
"learning_rate": 2.9987772958559627e-06,
"loss": 0.6646069288253784,
"step": 334
},
{
"epoch": 0.2543527630582892,
"grad_norm": 2.79805064201355,
"learning_rate": 2.998704326056514e-06,
"loss": 1.3354549407958984,
"step": 336
},
{
"epoch": 0.2558667676003028,
"grad_norm": 6.575490474700928,
"learning_rate": 2.998629242563432e-06,
"loss": 0.3947051763534546,
"step": 338
},
{
"epoch": 0.2573807721423164,
"grad_norm": 4.710763931274414,
"learning_rate": 2.998552045494389e-06,
"loss": 0.4745977222919464,
"step": 340
},
{
"epoch": 0.25889477668433003,
"grad_norm": 2.4593327045440674,
"learning_rate": 2.998472734970367e-06,
"loss": 0.5059724450111389,
"step": 342
},
{
"epoch": 0.2604087812263437,
"grad_norm": 2.6149814128875732,
"learning_rate": 2.998391311115663e-06,
"loss": 0.7527101635932922,
"step": 344
},
{
"epoch": 0.2619227857683573,
"grad_norm": 3.2358217239379883,
"learning_rate": 2.9983077740578835e-06,
"loss": 0.4932746887207031,
"step": 346
},
{
"epoch": 0.2634367903103709,
"grad_norm": 2.8413195610046387,
"learning_rate": 2.998222123927949e-06,
"loss": 0.5023959875106812,
"step": 348
},
{
"epoch": 0.26495079485238454,
"grad_norm": 2.284940719604492,
"learning_rate": 2.9981343608600907e-06,
"loss": 0.530271589756012,
"step": 350
},
{
"epoch": 0.2664647993943982,
"grad_norm": 1.8905975818634033,
"learning_rate": 2.99804448499185e-06,
"loss": 1.058239459991455,
"step": 352
},
{
"epoch": 0.2679788039364118,
"grad_norm": 1.2502957582473755,
"learning_rate": 2.9979524964640826e-06,
"loss": 0.575753390789032,
"step": 354
},
{
"epoch": 0.26949280847842544,
"grad_norm": 1.3402975797653198,
"learning_rate": 2.997858395420951e-06,
"loss": 0.7238829135894775,
"step": 356
},
{
"epoch": 0.27100681302043905,
"grad_norm": 4.392447471618652,
"learning_rate": 2.9977621820099316e-06,
"loss": 1.221725583076477,
"step": 358
},
{
"epoch": 0.27252081756245267,
"grad_norm": 1.6761703491210938,
"learning_rate": 2.997663856381811e-06,
"loss": 0.42734482884407043,
"step": 360
},
{
"epoch": 0.27403482210446634,
"grad_norm": 2.298595905303955,
"learning_rate": 2.997563418690685e-06,
"loss": 0.9713137745857239,
"step": 362
},
{
"epoch": 0.27554882664647995,
"grad_norm": 0.6969879269599915,
"learning_rate": 2.997460869093959e-06,
"loss": 0.31093937158584595,
"step": 364
},
{
"epoch": 0.27706283118849356,
"grad_norm": 2.1157233715057373,
"learning_rate": 2.9973562077523503e-06,
"loss": 0.6549879908561707,
"step": 366
},
{
"epoch": 0.2785768357305072,
"grad_norm": 1.964150071144104,
"learning_rate": 2.9972494348298837e-06,
"loss": 0.8664920330047607,
"step": 368
},
{
"epoch": 0.2800908402725208,
"grad_norm": 2.272841215133667,
"learning_rate": 2.9971405504938943e-06,
"loss": 0.48350414633750916,
"step": 370
},
{
"epoch": 0.28160484481453446,
"grad_norm": 1.9322324991226196,
"learning_rate": 2.9970295549150265e-06,
"loss": 0.9229208827018738,
"step": 372
},
{
"epoch": 0.2831188493565481,
"grad_norm": 1.7196455001831055,
"learning_rate": 2.996916448267232e-06,
"loss": 0.9739328622817993,
"step": 374
},
{
"epoch": 0.2846328538985617,
"grad_norm": 2.918416738510132,
"learning_rate": 2.9968012307277723e-06,
"loss": 0.48950931429862976,
"step": 376
},
{
"epoch": 0.2861468584405753,
"grad_norm": 9.320439338684082,
"learning_rate": 2.9966839024772165e-06,
"loss": 0.8621902465820312,
"step": 378
},
{
"epoch": 0.28766086298258897,
"grad_norm": 1.188773512840271,
"learning_rate": 2.9965644636994427e-06,
"loss": 1.0359840393066406,
"step": 380
},
{
"epoch": 0.2891748675246026,
"grad_norm": 2.3963191509246826,
"learning_rate": 2.996442914581634e-06,
"loss": 0.5880563259124756,
"step": 382
},
{
"epoch": 0.2906888720666162,
"grad_norm": 2.4038655757904053,
"learning_rate": 2.996319255314284e-06,
"loss": 0.5257418751716614,
"step": 384
},
{
"epoch": 0.2922028766086298,
"grad_norm": 1.5974441766738892,
"learning_rate": 2.996193486091192e-06,
"loss": 1.2693966627120972,
"step": 386
},
{
"epoch": 0.2937168811506434,
"grad_norm": 1.8292707204818726,
"learning_rate": 2.9960656071094628e-06,
"loss": 0.9720443487167358,
"step": 388
},
{
"epoch": 0.2952308856926571,
"grad_norm": 1.8199464082717896,
"learning_rate": 2.9959356185695096e-06,
"loss": 1.135921835899353,
"step": 390
},
{
"epoch": 0.2967448902346707,
"grad_norm": 0.9021064639091492,
"learning_rate": 2.9958035206750504e-06,
"loss": 1.1319067478179932,
"step": 392
},
{
"epoch": 0.2982588947766843,
"grad_norm": 3.2225494384765625,
"learning_rate": 2.9956693136331096e-06,
"loss": 0.9472005367279053,
"step": 394
},
{
"epoch": 0.29977289931869794,
"grad_norm": 4.63291597366333,
"learning_rate": 2.995532997654017e-06,
"loss": 0.99593186378479,
"step": 396
},
{
"epoch": 0.3012869038607116,
"grad_norm": 1.8574267625808716,
"learning_rate": 2.9953945729514073e-06,
"loss": 0.5503549575805664,
"step": 398
},
{
"epoch": 0.3028009084027252,
"grad_norm": 3.5578439235687256,
"learning_rate": 2.99525403974222e-06,
"loss": 0.8387674689292908,
"step": 400
},
{
"epoch": 0.30431491294473884,
"grad_norm": 6.638049602508545,
"learning_rate": 2.9951113982466996e-06,
"loss": 0.5031633377075195,
"step": 402
},
{
"epoch": 0.30582891748675245,
"grad_norm": 2.846966505050659,
"learning_rate": 2.9949666486883937e-06,
"loss": 0.9956486225128174,
"step": 404
},
{
"epoch": 0.30734292202876606,
"grad_norm": 1.3141142129898071,
"learning_rate": 2.9948197912941546e-06,
"loss": 0.288853257894516,
"step": 406
},
{
"epoch": 0.30885692657077973,
"grad_norm": 3.5708980560302734,
"learning_rate": 2.994670826294138e-06,
"loss": 1.5125066041946411,
"step": 408
},
{
"epoch": 0.31037093111279335,
"grad_norm": 7.3045125007629395,
"learning_rate": 2.9945197539218017e-06,
"loss": 0.6546447277069092,
"step": 410
},
{
"epoch": 0.31188493565480696,
"grad_norm": 2.1153337955474854,
"learning_rate": 2.9943665744139075e-06,
"loss": 1.3320523500442505,
"step": 412
},
{
"epoch": 0.3133989401968206,
"grad_norm": 4.322186470031738,
"learning_rate": 2.9942112880105175e-06,
"loss": 0.9307819604873657,
"step": 414
},
{
"epoch": 0.31491294473883424,
"grad_norm": 2.4291980266571045,
"learning_rate": 2.9940538949549984e-06,
"loss": 0.8921429514884949,
"step": 416
},
{
"epoch": 0.31642694928084786,
"grad_norm": 2.061569929122925,
"learning_rate": 2.9938943954940167e-06,
"loss": 0.9342135787010193,
"step": 418
},
{
"epoch": 0.31794095382286147,
"grad_norm": 4.544994831085205,
"learning_rate": 2.9937327898775406e-06,
"loss": 0.5935999751091003,
"step": 420
},
{
"epoch": 0.3194549583648751,
"grad_norm": 3.4589638710021973,
"learning_rate": 2.9935690783588378e-06,
"loss": 0.41867610812187195,
"step": 422
},
{
"epoch": 0.3209689629068887,
"grad_norm": 2.5398411750793457,
"learning_rate": 2.993403261194479e-06,
"loss": 1.2433795928955078,
"step": 424
},
{
"epoch": 0.32248296744890237,
"grad_norm": 1.7882115840911865,
"learning_rate": 2.9932353386443325e-06,
"loss": 0.7358930706977844,
"step": 426
},
{
"epoch": 0.323996971990916,
"grad_norm": 2.736948013305664,
"learning_rate": 2.993065310971568e-06,
"loss": 0.954931378364563,
"step": 428
},
{
"epoch": 0.3255109765329296,
"grad_norm": 10.501503944396973,
"learning_rate": 2.992893178442652e-06,
"loss": 1.289313793182373,
"step": 430
},
{
"epoch": 0.3270249810749432,
"grad_norm": 2.1437489986419678,
"learning_rate": 2.9927189413273517e-06,
"loss": 0.5708879232406616,
"step": 432
},
{
"epoch": 0.3285389856169568,
"grad_norm": 3.5188357830047607,
"learning_rate": 2.9925425998987326e-06,
"loss": 0.5585161447525024,
"step": 434
},
{
"epoch": 0.3300529901589705,
"grad_norm": 1.5307719707489014,
"learning_rate": 2.992364154433157e-06,
"loss": 0.9125571250915527,
"step": 436
},
{
"epoch": 0.3315669947009841,
"grad_norm": 2.313886880874634,
"learning_rate": 2.9921836052102853e-06,
"loss": 1.2311666011810303,
"step": 438
},
{
"epoch": 0.3330809992429977,
"grad_norm": 1.5364906787872314,
"learning_rate": 2.992000952513075e-06,
"loss": 0.5001219511032104,
"step": 440
},
{
"epoch": 0.33459500378501134,
"grad_norm": 2.821791648864746,
"learning_rate": 2.991816196627779e-06,
"loss": 0.5236578583717346,
"step": 442
},
{
"epoch": 0.336109008327025,
"grad_norm": 2.0568583011627197,
"learning_rate": 2.991629337843949e-06,
"loss": 0.5407780408859253,
"step": 444
},
{
"epoch": 0.3376230128690386,
"grad_norm": 1.2596296072006226,
"learning_rate": 2.9914403764544296e-06,
"loss": 1.2255700826644897,
"step": 446
},
{
"epoch": 0.33913701741105223,
"grad_norm": 3.2330868244171143,
"learning_rate": 2.991249312755362e-06,
"loss": 0.9774052500724792,
"step": 448
},
{
"epoch": 0.34065102195306585,
"grad_norm": 2.749786853790283,
"learning_rate": 2.991056147046181e-06,
"loss": 0.5227469205856323,
"step": 450
},
{
"epoch": 0.34216502649507946,
"grad_norm": 1.334883213043213,
"learning_rate": 2.990860879629618e-06,
"loss": 1.231229543685913,
"step": 452
},
{
"epoch": 0.34367903103709313,
"grad_norm": 4.1452860832214355,
"learning_rate": 2.990663510811695e-06,
"loss": 1.0209695100784302,
"step": 454
},
{
"epoch": 0.34519303557910674,
"grad_norm": 1.3384249210357666,
"learning_rate": 2.9904640409017305e-06,
"loss": 0.9044679403305054,
"step": 456
},
{
"epoch": 0.34670704012112036,
"grad_norm": 1.9722930192947388,
"learning_rate": 2.9902624702123334e-06,
"loss": 0.5573891997337341,
"step": 458
},
{
"epoch": 0.34822104466313397,
"grad_norm": 3.713996648788452,
"learning_rate": 2.9900587990594068e-06,
"loss": 0.7835407853126526,
"step": 460
},
{
"epoch": 0.34973504920514764,
"grad_norm": 6.265981197357178,
"learning_rate": 2.989853027762144e-06,
"loss": 0.9101991653442383,
"step": 462
},
{
"epoch": 0.35124905374716126,
"grad_norm": 2.356494665145874,
"learning_rate": 2.989645156643031e-06,
"loss": 1.044858694076538,
"step": 464
},
{
"epoch": 0.35276305828917487,
"grad_norm": 3.8169243335723877,
"learning_rate": 2.9894351860278433e-06,
"loss": 0.38478291034698486,
"step": 466
},
{
"epoch": 0.3542770628311885,
"grad_norm": 3.984137535095215,
"learning_rate": 2.989223116245648e-06,
"loss": 0.4994244873523712,
"step": 468
},
{
"epoch": 0.3557910673732021,
"grad_norm": 4.4269185066223145,
"learning_rate": 2.9890089476288017e-06,
"loss": 0.8007203340530396,
"step": 470
},
{
"epoch": 0.35730507191521577,
"grad_norm": 4.96340799331665,
"learning_rate": 2.98879268051295e-06,
"loss": 0.8296163082122803,
"step": 472
},
{
"epoch": 0.3588190764572294,
"grad_norm": 3.9115028381347656,
"learning_rate": 2.9885743152370267e-06,
"loss": 0.46959009766578674,
"step": 474
},
{
"epoch": 0.360333080999243,
"grad_norm": 2.773907423019409,
"learning_rate": 2.9883538521432557e-06,
"loss": 0.9441350102424622,
"step": 476
},
{
"epoch": 0.3618470855412566,
"grad_norm": 3.812221050262451,
"learning_rate": 2.988131291577147e-06,
"loss": 0.8799582123756409,
"step": 478
},
{
"epoch": 0.3633610900832703,
"grad_norm": 5.634382247924805,
"learning_rate": 2.9879066338874974e-06,
"loss": 0.9772864580154419,
"step": 480
},
{
"epoch": 0.3648750946252839,
"grad_norm": 5.383456707000732,
"learning_rate": 2.9876798794263923e-06,
"loss": 0.4553401470184326,
"step": 482
},
{
"epoch": 0.3663890991672975,
"grad_norm": 1.5636881589889526,
"learning_rate": 2.9874510285492013e-06,
"loss": 1.2522468566894531,
"step": 484
},
{
"epoch": 0.3679031037093111,
"grad_norm": 2.218456506729126,
"learning_rate": 2.9872200816145807e-06,
"loss": 1.2982913255691528,
"step": 486
},
{
"epoch": 0.36941710825132473,
"grad_norm": 2.1322004795074463,
"learning_rate": 2.986987038984471e-06,
"loss": 0.6175529956817627,
"step": 488
},
{
"epoch": 0.3709311127933384,
"grad_norm": 2.873570203781128,
"learning_rate": 2.9867519010240975e-06,
"loss": 0.5848088264465332,
"step": 490
},
{
"epoch": 0.372445117335352,
"grad_norm": 2.3397469520568848,
"learning_rate": 2.9865146681019688e-06,
"loss": 1.2863105535507202,
"step": 492
},
{
"epoch": 0.37395912187736563,
"grad_norm": 1.7062677145004272,
"learning_rate": 2.986275340589877e-06,
"loss": 0.7637373208999634,
"step": 494
},
{
"epoch": 0.37547312641937924,
"grad_norm": 3.172126054763794,
"learning_rate": 2.9860339188628978e-06,
"loss": 1.1434704065322876,
"step": 496
},
{
"epoch": 0.37698713096139286,
"grad_norm": 6.345958232879639,
"learning_rate": 2.985790403299387e-06,
"loss": 0.7667356729507446,
"step": 498
},
{
"epoch": 0.3785011355034065,
"grad_norm": 1.734782099723816,
"learning_rate": 2.9855447942809834e-06,
"loss": 0.8624241352081299,
"step": 500
},
{
"epoch": 0.38001514004542014,
"grad_norm": 7.933587074279785,
"learning_rate": 2.985297092192606e-06,
"loss": 0.9229949116706848,
"step": 502
},
{
"epoch": 0.38152914458743376,
"grad_norm": 2.319852113723755,
"learning_rate": 2.9850472974224557e-06,
"loss": 1.3553284406661987,
"step": 504
},
{
"epoch": 0.38304314912944737,
"grad_norm": 1.4423457384109497,
"learning_rate": 2.9847954103620095e-06,
"loss": 0.8387634754180908,
"step": 506
},
{
"epoch": 0.38455715367146104,
"grad_norm": 4.607706069946289,
"learning_rate": 2.9845414314060265e-06,
"loss": 0.6004860401153564,
"step": 508
},
{
"epoch": 0.38607115821347465,
"grad_norm": 4.358503818511963,
"learning_rate": 2.984285360952544e-06,
"loss": 0.5657727122306824,
"step": 510
},
{
"epoch": 0.38758516275548827,
"grad_norm": 1.6350818872451782,
"learning_rate": 2.9840271994028754e-06,
"loss": 0.4921954274177551,
"step": 512
},
{
"epoch": 0.3890991672975019,
"grad_norm": 2.449856758117676,
"learning_rate": 2.9837669471616124e-06,
"loss": 0.6726453900337219,
"step": 514
},
{
"epoch": 0.3906131718395155,
"grad_norm": 1.7032192945480347,
"learning_rate": 2.9835046046366237e-06,
"loss": 1.297200083732605,
"step": 516
},
{
"epoch": 0.39212717638152916,
"grad_norm": 3.538943290710449,
"learning_rate": 2.983240172239053e-06,
"loss": 0.8581514358520508,
"step": 518
},
{
"epoch": 0.3936411809235428,
"grad_norm": 6.8904194831848145,
"learning_rate": 2.98297365038332e-06,
"loss": 0.5082361698150635,
"step": 520
},
{
"epoch": 0.3951551854655564,
"grad_norm": 3.4525930881500244,
"learning_rate": 2.982705039487118e-06,
"loss": 0.4789521098136902,
"step": 522
},
{
"epoch": 0.39666919000757,
"grad_norm": 1.7367075681686401,
"learning_rate": 2.9824343399714144e-06,
"loss": 0.9578684568405151,
"step": 524
},
{
"epoch": 0.3981831945495837,
"grad_norm": 1.2806156873703003,
"learning_rate": 2.9821615522604515e-06,
"loss": 1.3319694995880127,
"step": 526
},
{
"epoch": 0.3996971990915973,
"grad_norm": 1.590675950050354,
"learning_rate": 2.9818866767817425e-06,
"loss": 0.8608264327049255,
"step": 528
},
{
"epoch": 0.4012112036336109,
"grad_norm": 5.6035943031311035,
"learning_rate": 2.981609713966073e-06,
"loss": 0.46807175874710083,
"step": 530
},
{
"epoch": 0.4027252081756245,
"grad_norm": 1.963212251663208,
"learning_rate": 2.9813306642475005e-06,
"loss": 1.2523926496505737,
"step": 532
},
{
"epoch": 0.40423921271763813,
"grad_norm": 1.9907010793685913,
"learning_rate": 2.9810495280633517e-06,
"loss": 0.8681015372276306,
"step": 534
},
{
"epoch": 0.4057532172596518,
"grad_norm": 3.108275890350342,
"learning_rate": 2.980766305854225e-06,
"loss": 0.8309670090675354,
"step": 536
},
{
"epoch": 0.4072672218016654,
"grad_norm": 2.425701856613159,
"learning_rate": 2.9804809980639865e-06,
"loss": 0.8349297046661377,
"step": 538
},
{
"epoch": 0.408781226343679,
"grad_norm": 2.3050215244293213,
"learning_rate": 2.9801936051397717e-06,
"loss": 0.5291258096694946,
"step": 540
},
{
"epoch": 0.41029523088569264,
"grad_norm": 4.254032611846924,
"learning_rate": 2.979904127531984e-06,
"loss": 0.5817729234695435,
"step": 542
},
{
"epoch": 0.4118092354277063,
"grad_norm": 2.6303694248199463,
"learning_rate": 2.9796125656942925e-06,
"loss": 0.6045780777931213,
"step": 544
},
{
"epoch": 0.4133232399697199,
"grad_norm": 3.902345895767212,
"learning_rate": 2.9793189200836356e-06,
"loss": 0.6001653075218201,
"step": 546
},
{
"epoch": 0.41483724451173354,
"grad_norm": 3.3723676204681396,
"learning_rate": 2.9790231911602143e-06,
"loss": 0.5705008506774902,
"step": 548
},
{
"epoch": 0.41635124905374715,
"grad_norm": 3.0511634349823,
"learning_rate": 2.9787253793874958e-06,
"loss": 0.9024116396903992,
"step": 550
},
{
"epoch": 0.41786525359576077,
"grad_norm": 3.0294415950775146,
"learning_rate": 2.9784254852322125e-06,
"loss": 0.6247788667678833,
"step": 552
},
{
"epoch": 0.41937925813777444,
"grad_norm": 1.308551549911499,
"learning_rate": 2.9781235091643587e-06,
"loss": 0.8290607333183289,
"step": 554
},
{
"epoch": 0.42089326267978805,
"grad_norm": 1.2458851337432861,
"learning_rate": 2.977819451657193e-06,
"loss": 0.5316109657287598,
"step": 556
},
{
"epoch": 0.42240726722180166,
"grad_norm": 2.5829837322235107,
"learning_rate": 2.9775133131872347e-06,
"loss": 1.2767724990844727,
"step": 558
},
{
"epoch": 0.4239212717638153,
"grad_norm": 3.8361194133758545,
"learning_rate": 2.977205094234265e-06,
"loss": 0.9774861335754395,
"step": 560
},
{
"epoch": 0.4254352763058289,
"grad_norm": 2.603776693344116,
"learning_rate": 2.976894795281326e-06,
"loss": 0.85866379737854,
"step": 562
},
{
"epoch": 0.42694928084784256,
"grad_norm": 1.0089654922485352,
"learning_rate": 2.9765824168147184e-06,
"loss": 0.7369099855422974,
"step": 564
},
{
"epoch": 0.4284632853898562,
"grad_norm": 6.3280205726623535,
"learning_rate": 2.9762679593240034e-06,
"loss": 0.5556588768959045,
"step": 566
},
{
"epoch": 0.4299772899318698,
"grad_norm": 1.7390072345733643,
"learning_rate": 2.975951423302e-06,
"loss": 1.0518251657485962,
"step": 568
},
{
"epoch": 0.4314912944738834,
"grad_norm": 2.6690053939819336,
"learning_rate": 2.9756328092447835e-06,
"loss": 1.2721149921417236,
"step": 570
},
{
"epoch": 0.43300529901589707,
"grad_norm": 0.7687424421310425,
"learning_rate": 2.9753121176516876e-06,
"loss": 0.6695957779884338,
"step": 572
},
{
"epoch": 0.4345193035579107,
"grad_norm": 1.9535387754440308,
"learning_rate": 2.9749893490253e-06,
"loss": 1.1010643243789673,
"step": 574
},
{
"epoch": 0.4360333080999243,
"grad_norm": 9.781461715698242,
"learning_rate": 2.974664503871467e-06,
"loss": 1.0317606925964355,
"step": 576
},
{
"epoch": 0.4375473126419379,
"grad_norm": 4.649098873138428,
"learning_rate": 2.9743375826992848e-06,
"loss": 1.0596553087234497,
"step": 578
},
{
"epoch": 0.4390613171839515,
"grad_norm": 2.2979838848114014,
"learning_rate": 2.9740085860211066e-06,
"loss": 1.2803912162780762,
"step": 580
},
{
"epoch": 0.4405753217259652,
"grad_norm": 2.245401382446289,
"learning_rate": 2.9736775143525363e-06,
"loss": 0.9560354948043823,
"step": 582
},
{
"epoch": 0.4420893262679788,
"grad_norm": 3.6574487686157227,
"learning_rate": 2.9733443682124303e-06,
"loss": 0.6721242666244507,
"step": 584
},
{
"epoch": 0.4436033308099924,
"grad_norm": 2.0253052711486816,
"learning_rate": 2.9730091481228967e-06,
"loss": 1.2475272417068481,
"step": 586
},
{
"epoch": 0.44511733535200604,
"grad_norm": 1.5434706211090088,
"learning_rate": 2.9726718546092934e-06,
"loss": 1.0386013984680176,
"step": 588
},
{
"epoch": 0.4466313398940197,
"grad_norm": 3.6868698596954346,
"learning_rate": 2.972332488200228e-06,
"loss": 0.9095458388328552,
"step": 590
},
{
"epoch": 0.4481453444360333,
"grad_norm": 2.4178566932678223,
"learning_rate": 2.971991049427556e-06,
"loss": 1.2065972089767456,
"step": 592
},
{
"epoch": 0.44965934897804694,
"grad_norm": 2.159417152404785,
"learning_rate": 2.9716475388263818e-06,
"loss": 1.3285571336746216,
"step": 594
},
{
"epoch": 0.45117335352006055,
"grad_norm": 0.5650960803031921,
"learning_rate": 2.9713019569350568e-06,
"loss": 0.6450245380401611,
"step": 596
},
{
"epoch": 0.45268735806207416,
"grad_norm": 1.47118079662323,
"learning_rate": 2.9709543042951764e-06,
"loss": 0.5147643685340881,
"step": 598
},
{
"epoch": 0.45420136260408783,
"grad_norm": 1.1347875595092773,
"learning_rate": 2.970604581451585e-06,
"loss": 0.4938359260559082,
"step": 600
},
{
"epoch": 0.45571536714610145,
"grad_norm": 1.2782500982284546,
"learning_rate": 2.970252788952368e-06,
"loss": 1.2716739177703857,
"step": 602
},
{
"epoch": 0.45722937168811506,
"grad_norm": 3.3131613731384277,
"learning_rate": 2.9698989273488553e-06,
"loss": 0.5934652090072632,
"step": 604
},
{
"epoch": 0.4587433762301287,
"grad_norm": 1.5987533330917358,
"learning_rate": 2.9695429971956215e-06,
"loss": 0.8995585441589355,
"step": 606
},
{
"epoch": 0.46025738077214234,
"grad_norm": 1.3289687633514404,
"learning_rate": 2.96918499905048e-06,
"loss": 0.5343512892723083,
"step": 608
},
{
"epoch": 0.46177138531415596,
"grad_norm": 2.6191699504852295,
"learning_rate": 2.968824933474487e-06,
"loss": 1.210081934928894,
"step": 610
},
{
"epoch": 0.46328538985616957,
"grad_norm": 1.5671101808547974,
"learning_rate": 2.968462801031939e-06,
"loss": 0.75421142578125,
"step": 612
},
{
"epoch": 0.4647993943981832,
"grad_norm": 1.4679361581802368,
"learning_rate": 2.9680986022903703e-06,
"loss": 0.8687835931777954,
"step": 614
},
{
"epoch": 0.4663133989401968,
"grad_norm": 2.458467721939087,
"learning_rate": 2.9677323378205546e-06,
"loss": 0.5934958457946777,
"step": 616
},
{
"epoch": 0.46782740348221047,
"grad_norm": 1.3534835577011108,
"learning_rate": 2.967364008196503e-06,
"loss": 0.9021729230880737,
"step": 618
},
{
"epoch": 0.4693414080242241,
"grad_norm": 1.609513282775879,
"learning_rate": 2.966993613995462e-06,
"loss": 0.9018821716308594,
"step": 620
},
{
"epoch": 0.4708554125662377,
"grad_norm": 1.3751788139343262,
"learning_rate": 2.966621155797916e-06,
"loss": 0.7877498269081116,
"step": 622
},
{
"epoch": 0.4723694171082513,
"grad_norm": 1.6474313735961914,
"learning_rate": 2.9662466341875814e-06,
"loss": 1.106297492980957,
"step": 624
},
{
"epoch": 0.4738834216502649,
"grad_norm": 14.291585922241211,
"learning_rate": 2.96587004975141e-06,
"loss": 0.5758960247039795,
"step": 626
},
{
"epoch": 0.4753974261922786,
"grad_norm": 2.3951332569122314,
"learning_rate": 2.965491403079586e-06,
"loss": 0.9572596549987793,
"step": 628
},
{
"epoch": 0.4769114307342922,
"grad_norm": 4.662050724029541,
"learning_rate": 2.965110694765526e-06,
"loss": 0.4738556742668152,
"step": 630
},
{
"epoch": 0.4784254352763058,
"grad_norm": 2.5419774055480957,
"learning_rate": 2.964727925405877e-06,
"loss": 0.5583599805831909,
"step": 632
},
{
"epoch": 0.47993943981831944,
"grad_norm": 3.2088208198547363,
"learning_rate": 2.9643430956005166e-06,
"loss": 0.7850269079208374,
"step": 634
},
{
"epoch": 0.4814534443603331,
"grad_norm": 1.2255338430404663,
"learning_rate": 2.963956205952551e-06,
"loss": 0.5295414924621582,
"step": 636
},
{
"epoch": 0.4829674489023467,
"grad_norm": 15.457282066345215,
"learning_rate": 2.9635672570683145e-06,
"loss": 0.8352210521697998,
"step": 638
},
{
"epoch": 0.48448145344436033,
"grad_norm": 2.2116246223449707,
"learning_rate": 2.9631762495573697e-06,
"loss": 0.9833724498748779,
"step": 640
},
{
"epoch": 0.48599545798637395,
"grad_norm": 1.815476655960083,
"learning_rate": 2.9627831840325043e-06,
"loss": 1.0128812789916992,
"step": 642
},
{
"epoch": 0.48750946252838756,
"grad_norm": 1.7280386686325073,
"learning_rate": 2.9623880611097323e-06,
"loss": 0.9331883788108826,
"step": 644
},
{
"epoch": 0.48902346707040123,
"grad_norm": 1.3087846040725708,
"learning_rate": 2.961990881408291e-06,
"loss": 1.2210347652435303,
"step": 646
},
{
"epoch": 0.49053747161241484,
"grad_norm": 1.7594807147979736,
"learning_rate": 2.9615916455506424e-06,
"loss": 0.8923947215080261,
"step": 648
},
{
"epoch": 0.49205147615442846,
"grad_norm": 7.217441082000732,
"learning_rate": 2.9611903541624695e-06,
"loss": 0.5282926559448242,
"step": 650
},
{
"epoch": 0.49356548069644207,
"grad_norm": 2.2700917720794678,
"learning_rate": 2.960787007872678e-06,
"loss": 0.5946333408355713,
"step": 652
},
{
"epoch": 0.49507948523845574,
"grad_norm": 2.591341018676758,
"learning_rate": 2.960381607313393e-06,
"loss": 0.729837954044342,
"step": 654
},
{
"epoch": 0.49659348978046935,
"grad_norm": 2.293522834777832,
"learning_rate": 2.959974153119959e-06,
"loss": 0.908227801322937,
"step": 656
},
{
"epoch": 0.49810749432248297,
"grad_norm": 1.8502575159072876,
"learning_rate": 2.959564645930941e-06,
"loss": 1.260167121887207,
"step": 658
},
{
"epoch": 0.4996214988644966,
"grad_norm": 22.77242088317871,
"learning_rate": 2.959153086388119e-06,
"loss": 0.9404181241989136,
"step": 660
},
{
"epoch": 0.5011355034065103,
"grad_norm": 1.890344262123108,
"learning_rate": 2.9587394751364895e-06,
"loss": 0.6529523730278015,
"step": 662
},
{
"epoch": 0.5026495079485238,
"grad_norm": 2.7498531341552734,
"learning_rate": 2.9583238128242673e-06,
"loss": 0.7013969421386719,
"step": 664
},
{
"epoch": 0.5041635124905375,
"grad_norm": 2.2211556434631348,
"learning_rate": 2.9579061001028787e-06,
"loss": 0.9186385273933411,
"step": 666
},
{
"epoch": 0.5056775170325511,
"grad_norm": 6.754961967468262,
"learning_rate": 2.9574863376269644e-06,
"loss": 0.5539544820785522,
"step": 668
},
{
"epoch": 0.5071915215745647,
"grad_norm": 3.343362331390381,
"learning_rate": 2.9570645260543773e-06,
"loss": 0.657001793384552,
"step": 670
},
{
"epoch": 0.5087055261165784,
"grad_norm": 2.516889810562134,
"learning_rate": 2.9566406660461816e-06,
"loss": 0.6957874298095703,
"step": 672
},
{
"epoch": 0.5102195306585919,
"grad_norm": 1.4724997282028198,
"learning_rate": 2.956214758266653e-06,
"loss": 1.0336772203445435,
"step": 674
},
{
"epoch": 0.5117335352006056,
"grad_norm": 3.6221024990081787,
"learning_rate": 2.955786803383275e-06,
"loss": 0.8379154205322266,
"step": 676
},
{
"epoch": 0.5132475397426193,
"grad_norm": 1.472934365272522,
"learning_rate": 2.9553568020667393e-06,
"loss": 1.3164961338043213,
"step": 678
},
{
"epoch": 0.5147615442846328,
"grad_norm": 2.5690131187438965,
"learning_rate": 2.9549247549909456e-06,
"loss": 0.7076467275619507,
"step": 680
},
{
"epoch": 0.5162755488266465,
"grad_norm": 2.820261240005493,
"learning_rate": 2.954490662833e-06,
"loss": 1.193648099899292,
"step": 682
},
{
"epoch": 0.5177895533686601,
"grad_norm": 2.8086366653442383,
"learning_rate": 2.954054526273213e-06,
"loss": 0.5091046690940857,
"step": 684
},
{
"epoch": 0.5193035579106737,
"grad_norm": 1.9126583337783813,
"learning_rate": 2.9536163459950984e-06,
"loss": 0.37770235538482666,
"step": 686
},
{
"epoch": 0.5208175624526874,
"grad_norm": 3.101130247116089,
"learning_rate": 2.953176122685374e-06,
"loss": 1.3354110717773438,
"step": 688
},
{
"epoch": 0.522331566994701,
"grad_norm": 2.8612306118011475,
"learning_rate": 2.952733857033959e-06,
"loss": 1.0210174322128296,
"step": 690
},
{
"epoch": 0.5238455715367146,
"grad_norm": 1.8148362636566162,
"learning_rate": 2.9522895497339746e-06,
"loss": 1.0870827436447144,
"step": 692
},
{
"epoch": 0.5253595760787282,
"grad_norm": 9.631368637084961,
"learning_rate": 2.9518432014817396e-06,
"loss": 0.4528180956840515,
"step": 694
},
{
"epoch": 0.5268735806207419,
"grad_norm": 2.0368731021881104,
"learning_rate": 2.951394812976772e-06,
"loss": 0.6437329649925232,
"step": 696
},
{
"epoch": 0.5283875851627555,
"grad_norm": 2.2827751636505127,
"learning_rate": 2.9509443849217887e-06,
"loss": 1.2784912586212158,
"step": 698
},
{
"epoch": 0.5299015897047691,
"grad_norm": 2.5390939712524414,
"learning_rate": 2.9504919180227014e-06,
"loss": 0.5129702687263489,
"step": 700
},
{
"epoch": 0.5314155942467828,
"grad_norm": 4.679416656494141,
"learning_rate": 2.9500374129886178e-06,
"loss": 1.0204864740371704,
"step": 702
},
{
"epoch": 0.5329295987887964,
"grad_norm": 2.2368252277374268,
"learning_rate": 2.9495808705318392e-06,
"loss": 0.8695347309112549,
"step": 704
},
{
"epoch": 0.53444360333081,
"grad_norm": 5.981869697570801,
"learning_rate": 2.9491222913678613e-06,
"loss": 1.1520168781280518,
"step": 706
},
{
"epoch": 0.5359576078728236,
"grad_norm": 2.0948941707611084,
"learning_rate": 2.9486616762153697e-06,
"loss": 1.0977720022201538,
"step": 708
},
{
"epoch": 0.5374716124148372,
"grad_norm": 1.9340217113494873,
"learning_rate": 2.948199025796242e-06,
"loss": 0.8995184302330017,
"step": 710
},
{
"epoch": 0.5389856169568509,
"grad_norm": 1.8215906620025635,
"learning_rate": 2.9477343408355466e-06,
"loss": 0.8296648263931274,
"step": 712
},
{
"epoch": 0.5404996214988645,
"grad_norm": 2.6101953983306885,
"learning_rate": 2.947267622061538e-06,
"loss": 0.4829208254814148,
"step": 714
},
{
"epoch": 0.5420136260408781,
"grad_norm": 8.983611106872559,
"learning_rate": 2.9467988702056594e-06,
"loss": 0.9006999135017395,
"step": 716
},
{
"epoch": 0.5435276305828918,
"grad_norm": 1.1540857553482056,
"learning_rate": 2.946328086002541e-06,
"loss": 0.9905385375022888,
"step": 718
},
{
"epoch": 0.5450416351249053,
"grad_norm": 1.1875252723693848,
"learning_rate": 2.945855270189997e-06,
"loss": 0.9678749442100525,
"step": 720
},
{
"epoch": 0.546555639666919,
"grad_norm": 1.1591203212738037,
"learning_rate": 2.9453804235090253e-06,
"loss": 0.8857312202453613,
"step": 722
},
{
"epoch": 0.5480696442089327,
"grad_norm": 2.282198667526245,
"learning_rate": 2.944903546703807e-06,
"loss": 0.6595865488052368,
"step": 724
},
{
"epoch": 0.5495836487509462,
"grad_norm": 16.777631759643555,
"learning_rate": 2.944424640521706e-06,
"loss": 1.0174167156219482,
"step": 726
},
{
"epoch": 0.5510976532929599,
"grad_norm": 1.7583130598068237,
"learning_rate": 2.9439437057132643e-06,
"loss": 0.9153175354003906,
"step": 728
},
{
"epoch": 0.5526116578349735,
"grad_norm": 1.554577112197876,
"learning_rate": 2.943460743032205e-06,
"loss": 0.7978371381759644,
"step": 730
},
{
"epoch": 0.5541256623769871,
"grad_norm": 6.742075443267822,
"learning_rate": 2.942975753235429e-06,
"loss": 0.8972716331481934,
"step": 732
},
{
"epoch": 0.5556396669190008,
"grad_norm": 6.648641109466553,
"learning_rate": 2.9424887370830134e-06,
"loss": 0.5501348376274109,
"step": 734
},
{
"epoch": 0.5571536714610144,
"grad_norm": 1.9982460737228394,
"learning_rate": 2.9419996953382116e-06,
"loss": 0.5124497413635254,
"step": 736
},
{
"epoch": 0.558667676003028,
"grad_norm": 2.2183444499969482,
"learning_rate": 2.9415086287674513e-06,
"loss": 0.8203888535499573,
"step": 738
},
{
"epoch": 0.5601816805450416,
"grad_norm": 3.337951421737671,
"learning_rate": 2.941015538140334e-06,
"loss": 0.47594571113586426,
"step": 740
},
{
"epoch": 0.5616956850870553,
"grad_norm": 2.2672388553619385,
"learning_rate": 2.940520424229632e-06,
"loss": 0.9030762910842896,
"step": 742
},
{
"epoch": 0.5632096896290689,
"grad_norm": 2.4289283752441406,
"learning_rate": 2.94002328781129e-06,
"loss": 1.2035784721374512,
"step": 744
},
{
"epoch": 0.5647236941710825,
"grad_norm": 3.540351152420044,
"learning_rate": 2.9395241296644217e-06,
"loss": 1.004518747329712,
"step": 746
},
{
"epoch": 0.5662376987130961,
"grad_norm": 1.4717613458633423,
"learning_rate": 2.939022950571309e-06,
"loss": 1.2319071292877197,
"step": 748
},
{
"epoch": 0.5677517032551098,
"grad_norm": 1.7184895277023315,
"learning_rate": 2.938519751317402e-06,
"loss": 1.0665231943130493,
"step": 750
},
{
"epoch": 0.5692657077971234,
"grad_norm": 1.8175034523010254,
"learning_rate": 2.938014532691315e-06,
"loss": 0.9953513145446777,
"step": 752
},
{
"epoch": 0.570779712339137,
"grad_norm": 2.043348789215088,
"learning_rate": 2.937507295484829e-06,
"loss": 0.9582357406616211,
"step": 754
},
{
"epoch": 0.5722937168811506,
"grad_norm": 3.9826252460479736,
"learning_rate": 2.936998040492888e-06,
"loss": 0.6116023063659668,
"step": 756
},
{
"epoch": 0.5738077214231643,
"grad_norm": 1.8964965343475342,
"learning_rate": 2.936486768513597e-06,
"loss": 1.2680330276489258,
"step": 758
},
{
"epoch": 0.5753217259651779,
"grad_norm": 2.7118351459503174,
"learning_rate": 2.9359734803482244e-06,
"loss": 0.7763157486915588,
"step": 760
},
{
"epoch": 0.5768357305071915,
"grad_norm": 1.8754926919937134,
"learning_rate": 2.935458176801196e-06,
"loss": 0.5886606574058533,
"step": 762
},
{
"epoch": 0.5783497350492052,
"grad_norm": 1.2424232959747314,
"learning_rate": 2.9349408586800974e-06,
"loss": 0.8996439576148987,
"step": 764
},
{
"epoch": 0.5798637395912187,
"grad_norm": 4.870052337646484,
"learning_rate": 2.9344215267956716e-06,
"loss": 0.9547120332717896,
"step": 766
},
{
"epoch": 0.5813777441332324,
"grad_norm": 1.3577274084091187,
"learning_rate": 2.933900181961816e-06,
"loss": 0.6967052221298218,
"step": 768
},
{
"epoch": 0.5828917486752461,
"grad_norm": 1.5694098472595215,
"learning_rate": 2.933376824995585e-06,
"loss": 0.8374803066253662,
"step": 770
},
{
"epoch": 0.5844057532172596,
"grad_norm": 3.961740255355835,
"learning_rate": 2.932851456717185e-06,
"loss": 1.0225903987884521,
"step": 772
},
{
"epoch": 0.5859197577592733,
"grad_norm": 4.641229629516602,
"learning_rate": 2.9323240779499744e-06,
"loss": 0.866554319858551,
"step": 774
},
{
"epoch": 0.5874337623012869,
"grad_norm": 2.3968756198883057,
"learning_rate": 2.9317946895204634e-06,
"loss": 0.960785984992981,
"step": 776
},
{
"epoch": 0.5889477668433005,
"grad_norm": 1.2767549753189087,
"learning_rate": 2.9312632922583108e-06,
"loss": 1.190723180770874,
"step": 778
},
{
"epoch": 0.5904617713853142,
"grad_norm": 1.090358018875122,
"learning_rate": 2.930729886996324e-06,
"loss": 0.5099388957023621,
"step": 780
},
{
"epoch": 0.5919757759273278,
"grad_norm": 1.057666301727295,
"learning_rate": 2.930194474570458e-06,
"loss": 0.9268434047698975,
"step": 782
},
{
"epoch": 0.5934897804693414,
"grad_norm": 1.2038636207580566,
"learning_rate": 2.929657055819812e-06,
"loss": 0.7505081295967102,
"step": 784
},
{
"epoch": 0.595003785011355,
"grad_norm": 1.194632649421692,
"learning_rate": 2.9291176315866315e-06,
"loss": 0.45634379982948303,
"step": 786
},
{
"epoch": 0.5965177895533686,
"grad_norm": 1.2533210515975952,
"learning_rate": 2.928576202716302e-06,
"loss": 0.7255032062530518,
"step": 788
},
{
"epoch": 0.5980317940953823,
"grad_norm": 1.046190619468689,
"learning_rate": 2.9280327700573545e-06,
"loss": 0.9680758714675903,
"step": 790
},
{
"epoch": 0.5995457986373959,
"grad_norm": 1.2196940183639526,
"learning_rate": 2.9274873344614567e-06,
"loss": 0.901178777217865,
"step": 792
},
{
"epoch": 0.6010598031794095,
"grad_norm": 1.716521978378296,
"learning_rate": 2.926939896783418e-06,
"loss": 0.8521221876144409,
"step": 794
},
{
"epoch": 0.6025738077214232,
"grad_norm": 1.507444143295288,
"learning_rate": 2.926390457881185e-06,
"loss": 0.8391146659851074,
"step": 796
},
{
"epoch": 0.6040878122634368,
"grad_norm": 1.9206948280334473,
"learning_rate": 2.9258390186158377e-06,
"loss": 0.8905218839645386,
"step": 798
},
{
"epoch": 0.6056018168054504,
"grad_norm": 1.6404706239700317,
"learning_rate": 2.925285579851596e-06,
"loss": 0.6892648339271545,
"step": 800
},
{
"epoch": 0.607115821347464,
"grad_norm": 2.493842840194702,
"learning_rate": 2.924730142455809e-06,
"loss": 0.8925967812538147,
"step": 802
},
{
"epoch": 0.6086298258894777,
"grad_norm": 10.76797103881836,
"learning_rate": 2.924172707298962e-06,
"loss": 0.9718807935714722,
"step": 804
},
{
"epoch": 0.6101438304314913,
"grad_norm": 7.091102123260498,
"learning_rate": 2.9236132752546675e-06,
"loss": 0.7837668061256409,
"step": 806
},
{
"epoch": 0.6116578349735049,
"grad_norm": 2.373680353164673,
"learning_rate": 2.9230518471996693e-06,
"loss": 0.8521146774291992,
"step": 808
},
{
"epoch": 0.6131718395155186,
"grad_norm": 4.460805892944336,
"learning_rate": 2.922488424013839e-06,
"loss": 0.7885128259658813,
"step": 810
},
{
"epoch": 0.6146858440575321,
"grad_norm": 1.3457956314086914,
"learning_rate": 2.9219230065801762e-06,
"loss": 0.9336018562316895,
"step": 812
},
{
"epoch": 0.6161998485995458,
"grad_norm": 12.52235221862793,
"learning_rate": 2.9213555957848045e-06,
"loss": 0.625677227973938,
"step": 814
},
{
"epoch": 0.6177138531415595,
"grad_norm": 1.0090450048446655,
"learning_rate": 2.920786192516971e-06,
"loss": 0.4627359211444855,
"step": 816
},
{
"epoch": 0.619227857683573,
"grad_norm": 6.27386999130249,
"learning_rate": 2.920214797669047e-06,
"loss": 0.48187127709388733,
"step": 818
},
{
"epoch": 0.6207418622255867,
"grad_norm": 1.8286688327789307,
"learning_rate": 2.9196414121365246e-06,
"loss": 0.9612034559249878,
"step": 820
},
{
"epoch": 0.6222558667676003,
"grad_norm": 2.0253140926361084,
"learning_rate": 2.919066036818015e-06,
"loss": 0.37848153710365295,
"step": 822
},
{
"epoch": 0.6237698713096139,
"grad_norm": 2.8019373416900635,
"learning_rate": 2.918488672615248e-06,
"loss": 0.48960334062576294,
"step": 824
},
{
"epoch": 0.6252838758516276,
"grad_norm": 3.38137149810791,
"learning_rate": 2.9179093204330706e-06,
"loss": 1.0333915948867798,
"step": 826
},
{
"epoch": 0.6267978803936411,
"grad_norm": 3.585141658782959,
"learning_rate": 2.917327981179446e-06,
"loss": 0.8286981582641602,
"step": 828
},
{
"epoch": 0.6283118849356548,
"grad_norm": 1.340408444404602,
"learning_rate": 2.91674465576545e-06,
"loss": 0.592296838760376,
"step": 830
},
{
"epoch": 0.6298258894776685,
"grad_norm": 2.1734678745269775,
"learning_rate": 2.9161593451052726e-06,
"loss": 0.6662003397941589,
"step": 832
},
{
"epoch": 0.631339894019682,
"grad_norm": 1.7065621614456177,
"learning_rate": 2.9155720501162143e-06,
"loss": 1.1774479150772095,
"step": 834
},
{
"epoch": 0.6328538985616957,
"grad_norm": 0.9802259206771851,
"learning_rate": 2.9149827717186858e-06,
"loss": 0.7125468254089355,
"step": 836
},
{
"epoch": 0.6343679031037093,
"grad_norm": 3.3598525524139404,
"learning_rate": 2.914391510836206e-06,
"loss": 1.0522104501724243,
"step": 838
},
{
"epoch": 0.6358819076457229,
"grad_norm": 4.160059928894043,
"learning_rate": 2.9137982683954005e-06,
"loss": 0.7905111908912659,
"step": 840
},
{
"epoch": 0.6373959121877366,
"grad_norm": 2.1964306831359863,
"learning_rate": 2.9132030453260012e-06,
"loss": 0.8976717591285706,
"step": 842
},
{
"epoch": 0.6389099167297502,
"grad_norm": 1.4463465213775635,
"learning_rate": 2.912605842560843e-06,
"loss": 1.174971342086792,
"step": 844
},
{
"epoch": 0.6404239212717638,
"grad_norm": 1.9015480279922485,
"learning_rate": 2.9120066610358644e-06,
"loss": 1.2245635986328125,
"step": 846
},
{
"epoch": 0.6419379258137774,
"grad_norm": 3.132871150970459,
"learning_rate": 2.911405501690104e-06,
"loss": 0.5686792135238647,
"step": 848
},
{
"epoch": 0.6434519303557911,
"grad_norm": 3.358302354812622,
"learning_rate": 2.910802365465702e-06,
"loss": 0.8344794511795044,
"step": 850
},
{
"epoch": 0.6449659348978047,
"grad_norm": 2.254426956176758,
"learning_rate": 2.9101972533078937e-06,
"loss": 1.2486826181411743,
"step": 852
},
{
"epoch": 0.6464799394398183,
"grad_norm": 1.7597107887268066,
"learning_rate": 2.909590166165013e-06,
"loss": 0.4052983820438385,
"step": 854
},
{
"epoch": 0.647993943981832,
"grad_norm": 4.035180568695068,
"learning_rate": 2.9089811049884896e-06,
"loss": 0.548038125038147,
"step": 856
},
{
"epoch": 0.6495079485238455,
"grad_norm": 2.2516682147979736,
"learning_rate": 2.908370070732846e-06,
"loss": 1.1047321557998657,
"step": 858
},
{
"epoch": 0.6510219530658592,
"grad_norm": 1.6817784309387207,
"learning_rate": 2.9077570643556967e-06,
"loss": 0.9021655917167664,
"step": 860
},
{
"epoch": 0.6525359576078729,
"grad_norm": 1.2909799814224243,
"learning_rate": 2.9071420868177464e-06,
"loss": 0.9059284925460815,
"step": 862
},
{
"epoch": 0.6540499621498864,
"grad_norm": 2.6258788108825684,
"learning_rate": 2.906525139082791e-06,
"loss": 0.8191069960594177,
"step": 864
},
{
"epoch": 0.6555639666919001,
"grad_norm": 1.954457402229309,
"learning_rate": 2.9059062221177133e-06,
"loss": 0.8419861793518066,
"step": 866
},
{
"epoch": 0.6570779712339136,
"grad_norm": 2.6133012771606445,
"learning_rate": 2.90528533689248e-06,
"loss": 0.9011951088905334,
"step": 868
},
{
"epoch": 0.6585919757759273,
"grad_norm": 2.973444700241089,
"learning_rate": 2.904662484380146e-06,
"loss": 0.5388523936271667,
"step": 870
},
{
"epoch": 0.660105980317941,
"grad_norm": 3.0186350345611572,
"learning_rate": 2.9040376655568473e-06,
"loss": 0.9145660400390625,
"step": 872
},
{
"epoch": 0.6616199848599545,
"grad_norm": 2.547288179397583,
"learning_rate": 2.903410881401801e-06,
"loss": 1.2741073369979858,
"step": 874
},
{
"epoch": 0.6631339894019682,
"grad_norm": 1.5956040620803833,
"learning_rate": 2.9027821328973073e-06,
"loss": 0.88572096824646,
"step": 876
},
{
"epoch": 0.6646479939439819,
"grad_norm": 2.2501113414764404,
"learning_rate": 2.9021514210287405e-06,
"loss": 0.5038325190544128,
"step": 878
},
{
"epoch": 0.6661619984859954,
"grad_norm": 2.4447531700134277,
"learning_rate": 2.9015187467845555e-06,
"loss": 0.9085601568222046,
"step": 880
},
{
"epoch": 0.6676760030280091,
"grad_norm": 1.6331331729888916,
"learning_rate": 2.900884111156281e-06,
"loss": 0.9052636027336121,
"step": 882
},
{
"epoch": 0.6691900075700227,
"grad_norm": 1.715603232383728,
"learning_rate": 2.90024751513852e-06,
"loss": 0.9259544014930725,
"step": 884
},
{
"epoch": 0.6707040121120363,
"grad_norm": 2.190739870071411,
"learning_rate": 2.8996089597289474e-06,
"loss": 1.2172551155090332,
"step": 886
},
{
"epoch": 0.67221801665405,
"grad_norm": 1.130866527557373,
"learning_rate": 2.8989684459283103e-06,
"loss": 0.9008451700210571,
"step": 888
},
{
"epoch": 0.6737320211960636,
"grad_norm": 1.3909481763839722,
"learning_rate": 2.898325974740423e-06,
"loss": 0.8698288202285767,
"step": 890
},
{
"epoch": 0.6752460257380772,
"grad_norm": 1.2574244737625122,
"learning_rate": 2.8976815471721686e-06,
"loss": 0.7217528820037842,
"step": 892
},
{
"epoch": 0.6767600302800908,
"grad_norm": 12.1749849319458,
"learning_rate": 2.8970351642334953e-06,
"loss": 0.6941770911216736,
"step": 894
},
{
"epoch": 0.6782740348221045,
"grad_norm": 1.5021611452102661,
"learning_rate": 2.896386826937418e-06,
"loss": 1.2301819324493408,
"step": 896
},
{
"epoch": 0.6797880393641181,
"grad_norm": 3.2311809062957764,
"learning_rate": 2.895736536300012e-06,
"loss": 0.5155861973762512,
"step": 898
},
{
"epoch": 0.6813020439061317,
"grad_norm": 2.549818992614746,
"learning_rate": 2.895084293340415e-06,
"loss": 1.0740303993225098,
"step": 900
},
{
"epoch": 0.6828160484481454,
"grad_norm": 2.2381906509399414,
"learning_rate": 2.8944300990808232e-06,
"loss": 0.7758005261421204,
"step": 902
},
{
"epoch": 0.6843300529901589,
"grad_norm": 2.3603358268737793,
"learning_rate": 2.8937739545464937e-06,
"loss": 0.5189546346664429,
"step": 904
},
{
"epoch": 0.6858440575321726,
"grad_norm": 2.0805881023406982,
"learning_rate": 2.8931158607657375e-06,
"loss": 0.6191146969795227,
"step": 906
},
{
"epoch": 0.6873580620741863,
"grad_norm": 1.9956350326538086,
"learning_rate": 2.89245581876992e-06,
"loss": 0.506104588508606,
"step": 908
},
{
"epoch": 0.6888720666161998,
"grad_norm": 3.0931084156036377,
"learning_rate": 2.891793829593463e-06,
"loss": 0.9202035069465637,
"step": 910
},
{
"epoch": 0.6903860711582135,
"grad_norm": 1.566701054573059,
"learning_rate": 2.8911298942738372e-06,
"loss": 1.0531013011932373,
"step": 912
},
{
"epoch": 0.691900075700227,
"grad_norm": 1.253186821937561,
"learning_rate": 2.8904640138515636e-06,
"loss": 0.8252142071723938,
"step": 914
},
{
"epoch": 0.6934140802422407,
"grad_norm": 1.041472315788269,
"learning_rate": 2.889796189370213e-06,
"loss": 0.5106720924377441,
"step": 916
},
{
"epoch": 0.6949280847842544,
"grad_norm": 2.2845821380615234,
"learning_rate": 2.8891264218764013e-06,
"loss": 0.8980287313461304,
"step": 918
},
{
"epoch": 0.6964420893262679,
"grad_norm": 1.5488086938858032,
"learning_rate": 2.8884547124197904e-06,
"loss": 0.5322882533073425,
"step": 920
},
{
"epoch": 0.6979560938682816,
"grad_norm": 0.5618732571601868,
"learning_rate": 2.887781062053086e-06,
"loss": 1.0926649570465088,
"step": 922
},
{
"epoch": 0.6994700984102953,
"grad_norm": 1.8515206575393677,
"learning_rate": 2.887105471832034e-06,
"loss": 0.9032957553863525,
"step": 924
},
{
"epoch": 0.7009841029523088,
"grad_norm": 3.249413251876831,
"learning_rate": 2.8864279428154225e-06,
"loss": 0.5452030897140503,
"step": 926
},
{
"epoch": 0.7024981074943225,
"grad_norm": 5.527769088745117,
"learning_rate": 2.8857484760650765e-06,
"loss": 0.6838735938072205,
"step": 928
},
{
"epoch": 0.7040121120363361,
"grad_norm": 1.556622862815857,
"learning_rate": 2.885067072645859e-06,
"loss": 0.7241641879081726,
"step": 930
},
{
"epoch": 0.7055261165783497,
"grad_norm": 5.3245463371276855,
"learning_rate": 2.8843837336256664e-06,
"loss": 0.8361733555793762,
"step": 932
},
{
"epoch": 0.7070401211203634,
"grad_norm": 2.6056675910949707,
"learning_rate": 2.883698460075431e-06,
"loss": 0.9540938138961792,
"step": 934
},
{
"epoch": 0.708554125662377,
"grad_norm": 1.26792311668396,
"learning_rate": 2.8830112530691147e-06,
"loss": 1.242187261581421,
"step": 936
},
{
"epoch": 0.7100681302043906,
"grad_norm": 5.294063091278076,
"learning_rate": 2.8823221136837114e-06,
"loss": 0.4748896062374115,
"step": 938
},
{
"epoch": 0.7115821347464042,
"grad_norm": 3.2267799377441406,
"learning_rate": 2.881631042999242e-06,
"loss": 0.8570255637168884,
"step": 940
},
{
"epoch": 0.7130961392884179,
"grad_norm": 1.6323168277740479,
"learning_rate": 2.8809380420987547e-06,
"loss": 0.8971255421638489,
"step": 942
},
{
"epoch": 0.7146101438304315,
"grad_norm": 1.6141637563705444,
"learning_rate": 2.8802431120683226e-06,
"loss": 0.7287229299545288,
"step": 944
},
{
"epoch": 0.7161241483724451,
"grad_norm": 2.4479527473449707,
"learning_rate": 2.8795462539970417e-06,
"loss": 1.0156718492507935,
"step": 946
},
{
"epoch": 0.7176381529144588,
"grad_norm": 2.234118938446045,
"learning_rate": 2.878847468977032e-06,
"loss": 0.8655732870101929,
"step": 948
},
{
"epoch": 0.7191521574564723,
"grad_norm": 1.285766839981079,
"learning_rate": 2.8781467581034287e-06,
"loss": 1.0937553644180298,
"step": 950
},
{
"epoch": 0.720666161998486,
"grad_norm": 3.193481922149658,
"learning_rate": 2.877444122474391e-06,
"loss": 1.252207636833191,
"step": 952
},
{
"epoch": 0.7221801665404997,
"grad_norm": 2.450623035430908,
"learning_rate": 2.8767395631910893e-06,
"loss": 1.2393505573272705,
"step": 954
},
{
"epoch": 0.7236941710825132,
"grad_norm": 2.2160840034484863,
"learning_rate": 2.8760330813577127e-06,
"loss": 0.8748006820678711,
"step": 956
},
{
"epoch": 0.7252081756245269,
"grad_norm": 9.748912811279297,
"learning_rate": 2.8753246780814606e-06,
"loss": 0.8926846981048584,
"step": 958
},
{
"epoch": 0.7267221801665406,
"grad_norm": 1.8308805227279663,
"learning_rate": 2.8746143544725454e-06,
"loss": 0.5339508652687073,
"step": 960
},
{
"epoch": 0.7282361847085541,
"grad_norm": 2.4011330604553223,
"learning_rate": 2.8739021116441887e-06,
"loss": 0.9131770133972168,
"step": 962
},
{
"epoch": 0.7297501892505678,
"grad_norm": 1.7618752717971802,
"learning_rate": 2.8731879507126194e-06,
"loss": 0.8701857328414917,
"step": 964
},
{
"epoch": 0.7312641937925813,
"grad_norm": 2.0772533416748047,
"learning_rate": 2.872471872797073e-06,
"loss": 0.7838905453681946,
"step": 966
},
{
"epoch": 0.732778198334595,
"grad_norm": 1.310855746269226,
"learning_rate": 2.8717538790197887e-06,
"loss": 0.7258808016777039,
"step": 968
},
{
"epoch": 0.7342922028766087,
"grad_norm": 3.5756638050079346,
"learning_rate": 2.8710339705060085e-06,
"loss": 0.7574349045753479,
"step": 970
},
{
"epoch": 0.7358062074186222,
"grad_norm": 2.8669235706329346,
"learning_rate": 2.870312148383976e-06,
"loss": 0.857857346534729,
"step": 972
},
{
"epoch": 0.7373202119606359,
"grad_norm": 5.559253215789795,
"learning_rate": 2.8695884137849317e-06,
"loss": 0.4777136445045471,
"step": 974
},
{
"epoch": 0.7388342165026495,
"grad_norm": 7.590475082397461,
"learning_rate": 2.868862767843116e-06,
"loss": 1.1985186338424683,
"step": 976
},
{
"epoch": 0.7403482210446631,
"grad_norm": 1.6530134677886963,
"learning_rate": 2.868135211695763e-06,
"loss": 1.2720890045166016,
"step": 978
},
{
"epoch": 0.7418622255866768,
"grad_norm": 2.517655372619629,
"learning_rate": 2.8674057464831016e-06,
"loss": 0.8914427757263184,
"step": 980
},
{
"epoch": 0.7433762301286904,
"grad_norm": 2.6291592121124268,
"learning_rate": 2.866674373348351e-06,
"loss": 0.7555364966392517,
"step": 982
},
{
"epoch": 0.744890234670704,
"grad_norm": 2.1514620780944824,
"learning_rate": 2.865941093437721e-06,
"loss": 0.8511964678764343,
"step": 984
},
{
"epoch": 0.7464042392127176,
"grad_norm": 1.976210355758667,
"learning_rate": 2.865205907900412e-06,
"loss": 0.9552498459815979,
"step": 986
},
{
"epoch": 0.7479182437547313,
"grad_norm": 3.423999547958374,
"learning_rate": 2.864468817888608e-06,
"loss": 1.2477160692214966,
"step": 988
},
{
"epoch": 0.7494322482967449,
"grad_norm": 2.068427562713623,
"learning_rate": 2.863729824557479e-06,
"loss": 1.0305664539337158,
"step": 990
},
{
"epoch": 0.7509462528387585,
"grad_norm": 1.6632643938064575,
"learning_rate": 2.862988929065177e-06,
"loss": 1.1914719343185425,
"step": 992
},
{
"epoch": 0.7524602573807722,
"grad_norm": 2.210796356201172,
"learning_rate": 2.862246132572837e-06,
"loss": 0.8302785158157349,
"step": 994
},
{
"epoch": 0.7539742619227857,
"grad_norm": 7.937429904937744,
"learning_rate": 2.8615014362445708e-06,
"loss": 0.8746134638786316,
"step": 996
},
{
"epoch": 0.7554882664647994,
"grad_norm": 1.960372805595398,
"learning_rate": 2.860754841247469e-06,
"loss": 1.107035756111145,
"step": 998
},
{
"epoch": 0.757002271006813,
"grad_norm": 6.556978702545166,
"learning_rate": 2.860006348751598e-06,
"loss": 0.48335692286491394,
"step": 1000
},
{
"epoch": 0.7585162755488266,
"grad_norm": 2.474748373031616,
"learning_rate": 2.8592559599299976e-06,
"loss": 0.5073326826095581,
"step": 1002
},
{
"epoch": 0.7600302800908403,
"grad_norm": 1.2302364110946655,
"learning_rate": 2.8585036759586796e-06,
"loss": 0.6908177733421326,
"step": 1004
},
{
"epoch": 0.761544284632854,
"grad_norm": 1.1481744050979614,
"learning_rate": 2.857749498016625e-06,
"loss": 0.6242103576660156,
"step": 1006
},
{
"epoch": 0.7630582891748675,
"grad_norm": 1.667983889579773,
"learning_rate": 2.856993427285784e-06,
"loss": 0.5696890950202942,
"step": 1008
},
{
"epoch": 0.7645722937168812,
"grad_norm": 2.3057103157043457,
"learning_rate": 2.856235464951074e-06,
"loss": 0.43427762389183044,
"step": 1010
},
{
"epoch": 0.7660862982588947,
"grad_norm": 2.363940715789795,
"learning_rate": 2.8554756122003755e-06,
"loss": 0.8118288516998291,
"step": 1012
},
{
"epoch": 0.7676003028009084,
"grad_norm": 2.5736377239227295,
"learning_rate": 2.8547138702245316e-06,
"loss": 1.203795313835144,
"step": 1014
},
{
"epoch": 0.7691143073429221,
"grad_norm": 4.996185779571533,
"learning_rate": 2.853950240217347e-06,
"loss": 0.5433783531188965,
"step": 1016
},
{
"epoch": 0.7706283118849356,
"grad_norm": 1.1869832277297974,
"learning_rate": 2.8531847233755852e-06,
"loss": 1.1779699325561523,
"step": 1018
},
{
"epoch": 0.7721423164269493,
"grad_norm": 1.2899645566940308,
"learning_rate": 2.852417320898966e-06,
"loss": 1.1803244352340698,
"step": 1020
},
{
"epoch": 0.7736563209689629,
"grad_norm": 1.5888104438781738,
"learning_rate": 2.8516480339901663e-06,
"loss": 0.6052057147026062,
"step": 1022
},
{
"epoch": 0.7751703255109765,
"grad_norm": 1.4890227317810059,
"learning_rate": 2.8508768638548135e-06,
"loss": 1.1067696809768677,
"step": 1024
},
{
"epoch": 0.7766843300529902,
"grad_norm": 4.891721248626709,
"learning_rate": 2.850103811701488e-06,
"loss": 0.5788875222206116,
"step": 1026
},
{
"epoch": 0.7781983345950038,
"grad_norm": 2.0971038341522217,
"learning_rate": 2.849328878741721e-06,
"loss": 0.8433497548103333,
"step": 1028
},
{
"epoch": 0.7797123391370174,
"grad_norm": 3.295856237411499,
"learning_rate": 2.8485520661899875e-06,
"loss": 0.9592798352241516,
"step": 1030
},
{
"epoch": 0.781226343679031,
"grad_norm": 1.9466731548309326,
"learning_rate": 2.8477733752637124e-06,
"loss": 0.4877493977546692,
"step": 1032
},
{
"epoch": 0.7827403482210447,
"grad_norm": 5.624363899230957,
"learning_rate": 2.8469928071832622e-06,
"loss": 0.5387091040611267,
"step": 1034
},
{
"epoch": 0.7842543527630583,
"grad_norm": 1.9813792705535889,
"learning_rate": 2.846210363171945e-06,
"loss": 0.9394605159759521,
"step": 1036
},
{
"epoch": 0.7857683573050719,
"grad_norm": 2.6155548095703125,
"learning_rate": 2.8454260444560107e-06,
"loss": 0.745268702507019,
"step": 1038
},
{
"epoch": 0.7872823618470856,
"grad_norm": 3.142620325088501,
"learning_rate": 2.8446398522646444e-06,
"loss": 0.3901050388813019,
"step": 1040
},
{
"epoch": 0.7887963663890991,
"grad_norm": 1.0682148933410645,
"learning_rate": 2.8438517878299704e-06,
"loss": 0.8102407455444336,
"step": 1042
},
{
"epoch": 0.7903103709311128,
"grad_norm": 2.6520161628723145,
"learning_rate": 2.8430618523870454e-06,
"loss": 1.0556105375289917,
"step": 1044
},
{
"epoch": 0.7918243754731265,
"grad_norm": 6.871322154998779,
"learning_rate": 2.842270047173858e-06,
"loss": 1.212335467338562,
"step": 1046
},
{
"epoch": 0.79333838001514,
"grad_norm": 1.3800029754638672,
"learning_rate": 2.841476373431328e-06,
"loss": 1.054120421409607,
"step": 1048
},
{
"epoch": 0.7948523845571537,
"grad_norm": 2.970756769180298,
"learning_rate": 2.8406808324033043e-06,
"loss": 1.2394263744354248,
"step": 1050
},
{
"epoch": 0.7963663890991673,
"grad_norm": 2.5240118503570557,
"learning_rate": 2.83988342533656e-06,
"loss": 0.913371741771698,
"step": 1052
},
{
"epoch": 0.7978803936411809,
"grad_norm": 1.9976874589920044,
"learning_rate": 2.839084153480795e-06,
"loss": 1.2547472715377808,
"step": 1054
},
{
"epoch": 0.7993943981831946,
"grad_norm": 3.423712968826294,
"learning_rate": 2.83828301808863e-06,
"loss": 0.435365229845047,
"step": 1056
},
{
"epoch": 0.8009084027252081,
"grad_norm": 5.868340015411377,
"learning_rate": 2.837480020415607e-06,
"loss": 1.0489962100982666,
"step": 1058
},
{
"epoch": 0.8024224072672218,
"grad_norm": 1.535408854484558,
"learning_rate": 2.836675161720187e-06,
"loss": 1.2446261644363403,
"step": 1060
},
{
"epoch": 0.8039364118092355,
"grad_norm": 1.8610411882400513,
"learning_rate": 2.8358684432637464e-06,
"loss": 1.2359105348587036,
"step": 1062
},
{
"epoch": 0.805450416351249,
"grad_norm": 1.2122951745986938,
"learning_rate": 2.8350598663105774e-06,
"loss": 0.500268816947937,
"step": 1064
},
{
"epoch": 0.8069644208932627,
"grad_norm": 1.37546706199646,
"learning_rate": 2.834249432127884e-06,
"loss": 0.9466779232025146,
"step": 1066
},
{
"epoch": 0.8084784254352763,
"grad_norm": 5.66973876953125,
"learning_rate": 2.833437141985781e-06,
"loss": 0.6052334904670715,
"step": 1068
},
{
"epoch": 0.8099924299772899,
"grad_norm": 2.347961902618408,
"learning_rate": 2.832622997157292e-06,
"loss": 0.858746349811554,
"step": 1070
},
{
"epoch": 0.8115064345193036,
"grad_norm": 1.3671703338623047,
"learning_rate": 2.831806998918348e-06,
"loss": 1.2333590984344482,
"step": 1072
},
{
"epoch": 0.8130204390613172,
"grad_norm": 2.0911505222320557,
"learning_rate": 2.8309891485477835e-06,
"loss": 0.7037978172302246,
"step": 1074
},
{
"epoch": 0.8145344436033308,
"grad_norm": 1.9133065938949585,
"learning_rate": 2.830169447327336e-06,
"loss": 0.5428035259246826,
"step": 1076
},
{
"epoch": 0.8160484481453444,
"grad_norm": 4.97536563873291,
"learning_rate": 2.8293478965416444e-06,
"loss": 0.8640028238296509,
"step": 1078
},
{
"epoch": 0.817562452687358,
"grad_norm": 2.3419647216796875,
"learning_rate": 2.8285244974782453e-06,
"loss": 0.8723977208137512,
"step": 1080
},
{
"epoch": 0.8190764572293717,
"grad_norm": 7.133307933807373,
"learning_rate": 2.827699251427572e-06,
"loss": 0.9312586784362793,
"step": 1082
},
{
"epoch": 0.8205904617713853,
"grad_norm": 3.1311888694763184,
"learning_rate": 2.8268721596829532e-06,
"loss": 1.2033246755599976,
"step": 1084
},
{
"epoch": 0.822104466313399,
"grad_norm": 1.5448412895202637,
"learning_rate": 2.8260432235406094e-06,
"loss": 0.7385362386703491,
"step": 1086
},
{
"epoch": 0.8236184708554126,
"grad_norm": 2.094623565673828,
"learning_rate": 2.825212444299652e-06,
"loss": 0.4562510550022125,
"step": 1088
},
{
"epoch": 0.8251324753974262,
"grad_norm": 1.6329617500305176,
"learning_rate": 2.8243798232620807e-06,
"loss": 0.9182811975479126,
"step": 1090
},
{
"epoch": 0.8266464799394398,
"grad_norm": 4.220635890960693,
"learning_rate": 2.823545361732782e-06,
"loss": 0.520475447177887,
"step": 1092
},
{
"epoch": 0.8281604844814534,
"grad_norm": 1.5116798877716064,
"learning_rate": 2.8227090610195265e-06,
"loss": 0.7769816517829895,
"step": 1094
},
{
"epoch": 0.8296744890234671,
"grad_norm": 6.517539978027344,
"learning_rate": 2.821870922432967e-06,
"loss": 0.8503794074058533,
"step": 1096
},
{
"epoch": 0.8311884935654807,
"grad_norm": 2.3296096324920654,
"learning_rate": 2.8210309472866375e-06,
"loss": 0.8540775179862976,
"step": 1098
},
{
"epoch": 0.8327024981074943,
"grad_norm": 3.2018163204193115,
"learning_rate": 2.820189136896949e-06,
"loss": 1.1501890420913696,
"step": 1100
},
{
"epoch": 0.834216502649508,
"grad_norm": 4.151829719543457,
"learning_rate": 2.8193454925831903e-06,
"loss": 0.52679842710495,
"step": 1102
},
{
"epoch": 0.8357305071915215,
"grad_norm": 2.1522796154022217,
"learning_rate": 2.818500015667523e-06,
"loss": 0.6993886232376099,
"step": 1104
},
{
"epoch": 0.8372445117335352,
"grad_norm": 3.550656318664551,
"learning_rate": 2.8176527074749803e-06,
"loss": 0.8634568452835083,
"step": 1106
},
{
"epoch": 0.8387585162755489,
"grad_norm": 2.264739990234375,
"learning_rate": 2.816803569333467e-06,
"loss": 0.3652278780937195,
"step": 1108
},
{
"epoch": 0.8402725208175624,
"grad_norm": 1.6804509162902832,
"learning_rate": 2.815952602573755e-06,
"loss": 0.8796547651290894,
"step": 1110
},
{
"epoch": 0.8417865253595761,
"grad_norm": 2.1686923503875732,
"learning_rate": 2.815099808529482e-06,
"loss": 1.461799144744873,
"step": 1112
},
{
"epoch": 0.8433005299015897,
"grad_norm": 11.775230407714844,
"learning_rate": 2.81424518853715e-06,
"loss": 0.8035557270050049,
"step": 1114
},
{
"epoch": 0.8448145344436033,
"grad_norm": 1.5292541980743408,
"learning_rate": 2.8133887439361213e-06,
"loss": 0.8822787404060364,
"step": 1116
},
{
"epoch": 0.846328538985617,
"grad_norm": 1.0312873125076294,
"learning_rate": 2.8125304760686196e-06,
"loss": 0.7623322010040283,
"step": 1118
},
{
"epoch": 0.8478425435276306,
"grad_norm": 2.891306161880493,
"learning_rate": 2.811670386279724e-06,
"loss": 0.9173418879508972,
"step": 1120
},
{
"epoch": 0.8493565480696442,
"grad_norm": 3.8095450401306152,
"learning_rate": 2.8108084759173695e-06,
"loss": 1.0241823196411133,
"step": 1122
},
{
"epoch": 0.8508705526116578,
"grad_norm": 2.514314889907837,
"learning_rate": 2.809944746332346e-06,
"loss": 0.8152939081192017,
"step": 1124
},
{
"epoch": 0.8523845571536715,
"grad_norm": 5.4192609786987305,
"learning_rate": 2.8090791988782928e-06,
"loss": 0.44812607765197754,
"step": 1126
},
{
"epoch": 0.8538985616956851,
"grad_norm": 1.8970630168914795,
"learning_rate": 2.808211834911698e-06,
"loss": 0.7814205288887024,
"step": 1128
},
{
"epoch": 0.8554125662376987,
"grad_norm": 1.624800205230713,
"learning_rate": 2.8073426557918975e-06,
"loss": 0.8644953966140747,
"step": 1130
},
{
"epoch": 0.8569265707797123,
"grad_norm": 2.4981987476348877,
"learning_rate": 2.806471662881072e-06,
"loss": 0.5381126999855042,
"step": 1132
},
{
"epoch": 0.858440575321726,
"grad_norm": 1.756712794303894,
"learning_rate": 2.8055988575442435e-06,
"loss": 0.5988097190856934,
"step": 1134
},
{
"epoch": 0.8599545798637396,
"grad_norm": 3.4914777278900146,
"learning_rate": 2.804724241149276e-06,
"loss": 0.6631752848625183,
"step": 1136
},
{
"epoch": 0.8614685844057532,
"grad_norm": 4.860958099365234,
"learning_rate": 2.8038478150668704e-06,
"loss": 1.2718474864959717,
"step": 1138
},
{
"epoch": 0.8629825889477668,
"grad_norm": 2.0775246620178223,
"learning_rate": 2.8029695806705645e-06,
"loss": 0.8018662929534912,
"step": 1140
},
{
"epoch": 0.8644965934897805,
"grad_norm": 2.704127311706543,
"learning_rate": 2.8020895393367304e-06,
"loss": 0.4317402243614197,
"step": 1142
},
{
"epoch": 0.8660105980317941,
"grad_norm": 3.2498629093170166,
"learning_rate": 2.8012076924445715e-06,
"loss": 1.2137093544006348,
"step": 1144
},
{
"epoch": 0.8675246025738077,
"grad_norm": 2.4548707008361816,
"learning_rate": 2.8003240413761203e-06,
"loss": 0.7217168211936951,
"step": 1146
},
{
"epoch": 0.8690386071158214,
"grad_norm": 2.6822304725646973,
"learning_rate": 2.7994385875162384e-06,
"loss": 0.8042414784431458,
"step": 1148
},
{
"epoch": 0.8705526116578349,
"grad_norm": 2.950601577758789,
"learning_rate": 2.7985513322526113e-06,
"loss": 1.1039682626724243,
"step": 1150
},
{
"epoch": 0.8720666161998486,
"grad_norm": 1.850177526473999,
"learning_rate": 2.797662276975749e-06,
"loss": 0.7056236267089844,
"step": 1152
},
{
"epoch": 0.8735806207418623,
"grad_norm": 6.22799825668335,
"learning_rate": 2.7967714230789814e-06,
"loss": 0.8468853235244751,
"step": 1154
},
{
"epoch": 0.8750946252838758,
"grad_norm": 1.8293390274047852,
"learning_rate": 2.7958787719584563e-06,
"loss": 0.8229158520698547,
"step": 1156
},
{
"epoch": 0.8766086298258895,
"grad_norm": 1.768035888671875,
"learning_rate": 2.794984325013141e-06,
"loss": 0.47428685426712036,
"step": 1158
},
{
"epoch": 0.878122634367903,
"grad_norm": 2.19268798828125,
"learning_rate": 2.7940880836448146e-06,
"loss": 0.8314292430877686,
"step": 1160
},
{
"epoch": 0.8796366389099167,
"grad_norm": 2.4109113216400146,
"learning_rate": 2.7931900492580693e-06,
"loss": 1.1764559745788574,
"step": 1162
},
{
"epoch": 0.8811506434519304,
"grad_norm": 1.2788642644882202,
"learning_rate": 2.7922902232603086e-06,
"loss": 0.30559080839157104,
"step": 1164
},
{
"epoch": 0.882664647993944,
"grad_norm": 2.033473014831543,
"learning_rate": 2.7913886070617414e-06,
"loss": 0.8132178783416748,
"step": 1166
},
{
"epoch": 0.8841786525359576,
"grad_norm": 5.624773025512695,
"learning_rate": 2.7904852020753835e-06,
"loss": 0.8571276664733887,
"step": 1168
},
{
"epoch": 0.8856926570779712,
"grad_norm": 1.4092870950698853,
"learning_rate": 2.789580009717054e-06,
"loss": 0.8549548983573914,
"step": 1170
},
{
"epoch": 0.8872066616199848,
"grad_norm": 2.9512217044830322,
"learning_rate": 2.788673031405374e-06,
"loss": 0.5353968739509583,
"step": 1172
},
{
"epoch": 0.8887206661619985,
"grad_norm": 4.28328800201416,
"learning_rate": 2.787764268561762e-06,
"loss": 0.731753408908844,
"step": 1174
},
{
"epoch": 0.8902346707040121,
"grad_norm": 2.1515896320343018,
"learning_rate": 2.7868537226104346e-06,
"loss": 0.7792448997497559,
"step": 1176
},
{
"epoch": 0.8917486752460257,
"grad_norm": 7.964438438415527,
"learning_rate": 2.7859413949784013e-06,
"loss": 0.47553539276123047,
"step": 1178
},
{
"epoch": 0.8932626797880394,
"grad_norm": 4.39863395690918,
"learning_rate": 2.7850272870954657e-06,
"loss": 1.2436693906784058,
"step": 1180
},
{
"epoch": 0.894776684330053,
"grad_norm": 1.8371373414993286,
"learning_rate": 2.78411140039422e-06,
"loss": 1.127423882484436,
"step": 1182
},
{
"epoch": 0.8962906888720666,
"grad_norm": 2.7225122451782227,
"learning_rate": 2.783193736310045e-06,
"loss": 0.8037985563278198,
"step": 1184
},
{
"epoch": 0.8978046934140802,
"grad_norm": 2.108687400817871,
"learning_rate": 2.782274296281107e-06,
"loss": 1.2393871545791626,
"step": 1186
},
{
"epoch": 0.8993186979560939,
"grad_norm": 1.637842059135437,
"learning_rate": 2.7813530817483537e-06,
"loss": 1.1900055408477783,
"step": 1188
},
{
"epoch": 0.9008327024981075,
"grad_norm": 4.827752113342285,
"learning_rate": 2.780430094155517e-06,
"loss": 0.8907873630523682,
"step": 1190
},
{
"epoch": 0.9023467070401211,
"grad_norm": 4.129854679107666,
"learning_rate": 2.7795053349491062e-06,
"loss": 0.30467483401298523,
"step": 1192
},
{
"epoch": 0.9038607115821348,
"grad_norm": 5.735848426818848,
"learning_rate": 2.778578805578406e-06,
"loss": 0.8214407563209534,
"step": 1194
},
{
"epoch": 0.9053747161241483,
"grad_norm": 1.920812964439392,
"learning_rate": 2.7776505074954756e-06,
"loss": 0.8967646956443787,
"step": 1196
},
{
"epoch": 0.906888720666162,
"grad_norm": 2.8142926692962646,
"learning_rate": 2.776720442155148e-06,
"loss": 0.5961228609085083,
"step": 1198
},
{
"epoch": 0.9084027252081757,
"grad_norm": 1.3868736028671265,
"learning_rate": 2.7757886110150234e-06,
"loss": 1.00760817527771,
"step": 1200
},
{
"epoch": 0.9099167297501892,
"grad_norm": 2.524928331375122,
"learning_rate": 2.774855015535471e-06,
"loss": 1.0168966054916382,
"step": 1202
},
{
"epoch": 0.9114307342922029,
"grad_norm": 2.3753180503845215,
"learning_rate": 2.7739196571796242e-06,
"loss": 1.2498538494110107,
"step": 1204
},
{
"epoch": 0.9129447388342165,
"grad_norm": 1.77561616897583,
"learning_rate": 2.7729825374133805e-06,
"loss": 0.7145369648933411,
"step": 1206
},
{
"epoch": 0.9144587433762301,
"grad_norm": 2.117807149887085,
"learning_rate": 2.7720436577053957e-06,
"loss": 0.9042050838470459,
"step": 1208
},
{
"epoch": 0.9159727479182438,
"grad_norm": 5.781267166137695,
"learning_rate": 2.7711030195270854e-06,
"loss": 0.7323153614997864,
"step": 1210
},
{
"epoch": 0.9174867524602573,
"grad_norm": 3.5370779037475586,
"learning_rate": 2.770160624352621e-06,
"loss": 0.8055700063705444,
"step": 1212
},
{
"epoch": 0.919000757002271,
"grad_norm": 1.6705321073532104,
"learning_rate": 2.7692164736589273e-06,
"loss": 0.850743293762207,
"step": 1214
},
{
"epoch": 0.9205147615442847,
"grad_norm": 1.9116226434707642,
"learning_rate": 2.7682705689256797e-06,
"loss": 1.4134933948516846,
"step": 1216
},
{
"epoch": 0.9220287660862982,
"grad_norm": 2.118795871734619,
"learning_rate": 2.767322911635303e-06,
"loss": 0.42531701922416687,
"step": 1218
},
{
"epoch": 0.9235427706283119,
"grad_norm": 1.391668677330017,
"learning_rate": 2.7663735032729687e-06,
"loss": 0.945673406124115,
"step": 1220
},
{
"epoch": 0.9250567751703255,
"grad_norm": 2.9070723056793213,
"learning_rate": 2.7654223453265933e-06,
"loss": 0.922468364238739,
"step": 1222
},
{
"epoch": 0.9265707797123391,
"grad_norm": 1.8511205911636353,
"learning_rate": 2.7644694392868335e-06,
"loss": 0.6354098916053772,
"step": 1224
},
{
"epoch": 0.9280847842543528,
"grad_norm": 4.253298282623291,
"learning_rate": 2.7635147866470874e-06,
"loss": 0.19364294409751892,
"step": 1226
},
{
"epoch": 0.9295987887963664,
"grad_norm": 1.7090325355529785,
"learning_rate": 2.7625583889034892e-06,
"loss": 0.8719943761825562,
"step": 1228
},
{
"epoch": 0.93111279333838,
"grad_norm": 3.77422833442688,
"learning_rate": 2.7616002475549083e-06,
"loss": 0.42701461911201477,
"step": 1230
},
{
"epoch": 0.9326267978803936,
"grad_norm": 2.711357593536377,
"learning_rate": 2.7606403641029477e-06,
"loss": 1.1919710636138916,
"step": 1232
},
{
"epoch": 0.9341408024224073,
"grad_norm": 1.9754915237426758,
"learning_rate": 2.7596787400519383e-06,
"loss": 1.2326306104660034,
"step": 1234
},
{
"epoch": 0.9356548069644209,
"grad_norm": 11.081687927246094,
"learning_rate": 2.7587153769089416e-06,
"loss": 0.7123717069625854,
"step": 1236
},
{
"epoch": 0.9371688115064345,
"grad_norm": 2.403883218765259,
"learning_rate": 2.757750276183743e-06,
"loss": 1.3490595817565918,
"step": 1238
},
{
"epoch": 0.9386828160484482,
"grad_norm": 3.322733163833618,
"learning_rate": 2.756783439388851e-06,
"loss": 0.9039950370788574,
"step": 1240
},
{
"epoch": 0.9401968205904617,
"grad_norm": 5.353653907775879,
"learning_rate": 2.7558148680394947e-06,
"loss": 0.4231591522693634,
"step": 1242
},
{
"epoch": 0.9417108251324754,
"grad_norm": 2.372028350830078,
"learning_rate": 2.7548445636536234e-06,
"loss": 0.6393493413925171,
"step": 1244
},
{
"epoch": 0.9432248296744891,
"grad_norm": 2.0998072624206543,
"learning_rate": 2.753872527751901e-06,
"loss": 1.0599191188812256,
"step": 1246
},
{
"epoch": 0.9447388342165026,
"grad_norm": 1.9805744886398315,
"learning_rate": 2.7528987618577047e-06,
"loss": 1.1206618547439575,
"step": 1248
},
{
"epoch": 0.9462528387585163,
"grad_norm": 1.0623162984848022,
"learning_rate": 2.7519232674971233e-06,
"loss": 0.6939067840576172,
"step": 1250
},
{
"epoch": 0.9477668433005298,
"grad_norm": 4.080114841461182,
"learning_rate": 2.750946046198955e-06,
"loss": 0.902068555355072,
"step": 1252
},
{
"epoch": 0.9492808478425435,
"grad_norm": 5.1340227127075195,
"learning_rate": 2.749967099494704e-06,
"loss": 0.8372121453285217,
"step": 1254
},
{
"epoch": 0.9507948523845572,
"grad_norm": 3.9078993797302246,
"learning_rate": 2.7489864289185786e-06,
"loss": 0.8124945163726807,
"step": 1256
},
{
"epoch": 0.9523088569265707,
"grad_norm": 7.938327789306641,
"learning_rate": 2.7480040360074886e-06,
"loss": 0.9052401185035706,
"step": 1258
},
{
"epoch": 0.9538228614685844,
"grad_norm": 2.190640449523926,
"learning_rate": 2.7470199223010446e-06,
"loss": 0.9121738076210022,
"step": 1260
},
{
"epoch": 0.9553368660105981,
"grad_norm": 2.5523154735565186,
"learning_rate": 2.7460340893415503e-06,
"loss": 0.682637631893158,
"step": 1262
},
{
"epoch": 0.9568508705526116,
"grad_norm": 2.231675624847412,
"learning_rate": 2.7450465386740074e-06,
"loss": 0.8745209574699402,
"step": 1264
},
{
"epoch": 0.9583648750946253,
"grad_norm": 1.344656229019165,
"learning_rate": 2.744057271846109e-06,
"loss": 1.2278828620910645,
"step": 1266
},
{
"epoch": 0.9598788796366389,
"grad_norm": 1.938359260559082,
"learning_rate": 2.743066290408236e-06,
"loss": 0.76249760389328,
"step": 1268
},
{
"epoch": 0.9613928841786525,
"grad_norm": 1.5976496934890747,
"learning_rate": 2.7420735959134585e-06,
"loss": 0.5155737400054932,
"step": 1270
},
{
"epoch": 0.9629068887206662,
"grad_norm": 1.5426353216171265,
"learning_rate": 2.74107918991753e-06,
"loss": 1.2777304649353027,
"step": 1272
},
{
"epoch": 0.9644208932626798,
"grad_norm": 3.2231552600860596,
"learning_rate": 2.740083073978887e-06,
"loss": 0.5381708741188049,
"step": 1274
},
{
"epoch": 0.9659348978046934,
"grad_norm": 2.456082344055176,
"learning_rate": 2.739085249658645e-06,
"loss": 0.7773256301879883,
"step": 1276
},
{
"epoch": 0.967448902346707,
"grad_norm": 6.216519832611084,
"learning_rate": 2.738085718520598e-06,
"loss": 0.39604809880256653,
"step": 1278
},
{
"epoch": 0.9689629068887207,
"grad_norm": 2.6285789012908936,
"learning_rate": 2.7370844821312133e-06,
"loss": 0.7192943096160889,
"step": 1280
},
{
"epoch": 0.9704769114307343,
"grad_norm": 1.3127273321151733,
"learning_rate": 2.736081542059633e-06,
"loss": 1.2801799774169922,
"step": 1282
},
{
"epoch": 0.9719909159727479,
"grad_norm": 1.9447704553604126,
"learning_rate": 2.7350768998776668e-06,
"loss": 1.1748425960540771,
"step": 1284
},
{
"epoch": 0.9735049205147616,
"grad_norm": 1.5830963850021362,
"learning_rate": 2.7340705571597945e-06,
"loss": 0.8880859017372131,
"step": 1286
},
{
"epoch": 0.9750189250567751,
"grad_norm": 1.9534872770309448,
"learning_rate": 2.733062515483158e-06,
"loss": 0.6956273317337036,
"step": 1288
},
{
"epoch": 0.9765329295987888,
"grad_norm": 1.4043753147125244,
"learning_rate": 2.732052776427564e-06,
"loss": 0.8173001408576965,
"step": 1290
},
{
"epoch": 0.9780469341408025,
"grad_norm": 1.894389033317566,
"learning_rate": 2.7310413415754785e-06,
"loss": 0.7789499759674072,
"step": 1292
},
{
"epoch": 0.979560938682816,
"grad_norm": 3.465914487838745,
"learning_rate": 2.7300282125120255e-06,
"loss": 0.5313485860824585,
"step": 1294
},
{
"epoch": 0.9810749432248297,
"grad_norm": 9.61059856414795,
"learning_rate": 2.7290133908249844e-06,
"loss": 0.9159805774688721,
"step": 1296
},
{
"epoch": 0.9825889477668432,
"grad_norm": 4.936594486236572,
"learning_rate": 2.7279968781047867e-06,
"loss": 0.8132763504981995,
"step": 1298
},
{
"epoch": 0.9841029523088569,
"grad_norm": 2.389509916305542,
"learning_rate": 2.726978675944514e-06,
"loss": 1.0288949012756348,
"step": 1300
},
{
"epoch": 0.9856169568508706,
"grad_norm": 1.7264068126678467,
"learning_rate": 2.7259587859398954e-06,
"loss": 0.837356686592102,
"step": 1302
},
{
"epoch": 0.9871309613928841,
"grad_norm": 1.2041871547698975,
"learning_rate": 2.7249372096893075e-06,
"loss": 0.444491982460022,
"step": 1304
},
{
"epoch": 0.9886449659348978,
"grad_norm": 1.9402670860290527,
"learning_rate": 2.723913948793766e-06,
"loss": 0.8788086175918579,
"step": 1306
},
{
"epoch": 0.9901589704769115,
"grad_norm": 8.380989074707031,
"learning_rate": 2.7228890048569294e-06,
"loss": 0.5033178329467773,
"step": 1308
},
{
"epoch": 0.991672975018925,
"grad_norm": 2.616488218307495,
"learning_rate": 2.721862379485094e-06,
"loss": 0.751565158367157,
"step": 1310
},
{
"epoch": 0.9931869795609387,
"grad_norm": 1.3802661895751953,
"learning_rate": 2.720834074287188e-06,
"loss": 0.7129951119422913,
"step": 1312
},
{
"epoch": 0.9947009841029523,
"grad_norm": 1.6147418022155762,
"learning_rate": 2.719804090874776e-06,
"loss": 0.7851198315620422,
"step": 1314
},
{
"epoch": 0.9962149886449659,
"grad_norm": 1.4303407669067383,
"learning_rate": 2.7187724308620507e-06,
"loss": 0.8599988222122192,
"step": 1316
},
{
"epoch": 0.9977289931869796,
"grad_norm": 3.953418254852295,
"learning_rate": 2.7177390958658336e-06,
"loss": 0.9210554957389832,
"step": 1318
},
{
"epoch": 0.9992429977289932,
"grad_norm": 1.302258849143982,
"learning_rate": 2.71670408750557e-06,
"loss": 0.38900431990623474,
"step": 1320
},
{
"epoch": 1.0007570022710068,
"grad_norm": 2.8431973457336426,
"learning_rate": 2.715667407403328e-06,
"loss": 0.6563358306884766,
"step": 1322
},
{
"epoch": 1.0022710068130205,
"grad_norm": 1.8707140684127808,
"learning_rate": 2.7146290571837965e-06,
"loss": 0.7056888341903687,
"step": 1324
},
{
"epoch": 1.0037850113550342,
"grad_norm": 1.6970131397247314,
"learning_rate": 2.7135890384742804e-06,
"loss": 0.8172030448913574,
"step": 1326
},
{
"epoch": 1.0052990158970476,
"grad_norm": 2.909482717514038,
"learning_rate": 2.712547352904701e-06,
"loss": 0.4417012631893158,
"step": 1328
},
{
"epoch": 1.0068130204390613,
"grad_norm": 1.567160964012146,
"learning_rate": 2.7115040021075915e-06,
"loss": 1.1346977949142456,
"step": 1330
},
{
"epoch": 1.008327024981075,
"grad_norm": 1.576781988143921,
"learning_rate": 2.710458987718094e-06,
"loss": 0.7643612027168274,
"step": 1332
},
{
"epoch": 1.0098410295230886,
"grad_norm": 2.6382930278778076,
"learning_rate": 2.7094123113739576e-06,
"loss": 0.6715695858001709,
"step": 1334
},
{
"epoch": 1.0113550340651023,
"grad_norm": 2.5436923503875732,
"learning_rate": 2.7083639747155376e-06,
"loss": 0.3763667047023773,
"step": 1336
},
{
"epoch": 1.0128690386071157,
"grad_norm": 3.2012388706207275,
"learning_rate": 2.7073139793857908e-06,
"loss": 0.6741540431976318,
"step": 1338
},
{
"epoch": 1.0143830431491294,
"grad_norm": 3.748810291290283,
"learning_rate": 2.706262327030272e-06,
"loss": 0.6608448624610901,
"step": 1340
},
{
"epoch": 1.015897047691143,
"grad_norm": 1.3386434316635132,
"learning_rate": 2.705209019297135e-06,
"loss": 0.47587254643440247,
"step": 1342
},
{
"epoch": 1.0174110522331568,
"grad_norm": 2.3118233680725098,
"learning_rate": 2.7041540578371273e-06,
"loss": 0.801154375076294,
"step": 1344
},
{
"epoch": 1.0189250567751704,
"grad_norm": 20.85770034790039,
"learning_rate": 2.7030974443035864e-06,
"loss": 0.8961002826690674,
"step": 1346
},
{
"epoch": 1.0204390613171839,
"grad_norm": 1.6689656972885132,
"learning_rate": 2.7020391803524415e-06,
"loss": 0.6929575800895691,
"step": 1348
},
{
"epoch": 1.0219530658591975,
"grad_norm": 1.3510750532150269,
"learning_rate": 2.7009792676422067e-06,
"loss": 0.7450751066207886,
"step": 1350
},
{
"epoch": 1.0234670704012112,
"grad_norm": 0.8317976593971252,
"learning_rate": 2.6999177078339807e-06,
"loss": 0.6293408870697021,
"step": 1352
},
{
"epoch": 1.0249810749432249,
"grad_norm": 3.2913503646850586,
"learning_rate": 2.6988545025914437e-06,
"loss": 0.841137707233429,
"step": 1354
},
{
"epoch": 1.0264950794852385,
"grad_norm": 3.2313413619995117,
"learning_rate": 2.697789653580853e-06,
"loss": 0.7870294451713562,
"step": 1356
},
{
"epoch": 1.028009084027252,
"grad_norm": 1.451443076133728,
"learning_rate": 2.6967231624710447e-06,
"loss": 0.9140100479125977,
"step": 1358
},
{
"epoch": 1.0295230885692657,
"grad_norm": 2.742377996444702,
"learning_rate": 2.695655030933426e-06,
"loss": 0.9199535250663757,
"step": 1360
},
{
"epoch": 1.0310370931112793,
"grad_norm": 9.10448169708252,
"learning_rate": 2.694585260641977e-06,
"loss": 0.6621928811073303,
"step": 1362
},
{
"epoch": 1.032551097653293,
"grad_norm": 1.4048140048980713,
"learning_rate": 2.6935138532732442e-06,
"loss": 0.8045563697814941,
"step": 1364
},
{
"epoch": 1.0340651021953067,
"grad_norm": 2.4908690452575684,
"learning_rate": 2.6924408105063405e-06,
"loss": 0.8018115758895874,
"step": 1366
},
{
"epoch": 1.0355791067373201,
"grad_norm": 0.5607267022132874,
"learning_rate": 2.6913661340229423e-06,
"loss": 0.45020002126693726,
"step": 1368
},
{
"epoch": 1.0370931112793338,
"grad_norm": 4.257814884185791,
"learning_rate": 2.6902898255072865e-06,
"loss": 0.6275112628936768,
"step": 1370
},
{
"epoch": 1.0386071158213475,
"grad_norm": 3.537292242050171,
"learning_rate": 2.6892118866461664e-06,
"loss": 0.3463631570339203,
"step": 1372
},
{
"epoch": 1.0401211203633611,
"grad_norm": 3.6382415294647217,
"learning_rate": 2.6881323191289305e-06,
"loss": 0.791259765625,
"step": 1374
},
{
"epoch": 1.0416351249053748,
"grad_norm": 2.5053513050079346,
"learning_rate": 2.6870511246474815e-06,
"loss": 0.9765958786010742,
"step": 1376
},
{
"epoch": 1.0431491294473882,
"grad_norm": 2.3160805702209473,
"learning_rate": 2.685968304896271e-06,
"loss": 0.7225064039230347,
"step": 1378
},
{
"epoch": 1.044663133989402,
"grad_norm": 2.5423390865325928,
"learning_rate": 2.6848838615722962e-06,
"loss": 1.1174613237380981,
"step": 1380
},
{
"epoch": 1.0461771385314156,
"grad_norm": 1.6330304145812988,
"learning_rate": 2.683797796375101e-06,
"loss": 0.741710364818573,
"step": 1382
},
{
"epoch": 1.0476911430734293,
"grad_norm": 1.4876264333724976,
"learning_rate": 2.68271011100677e-06,
"loss": 0.7462827563285828,
"step": 1384
},
{
"epoch": 1.049205147615443,
"grad_norm": 2.399090051651001,
"learning_rate": 2.6816208071719274e-06,
"loss": 1.0535266399383545,
"step": 1386
},
{
"epoch": 1.0507191521574564,
"grad_norm": 1.310610055923462,
"learning_rate": 2.680529886577733e-06,
"loss": 1.0984599590301514,
"step": 1388
},
{
"epoch": 1.05223315669947,
"grad_norm": 20.521469116210938,
"learning_rate": 2.6794373509338812e-06,
"loss": 0.746917724609375,
"step": 1390
},
{
"epoch": 1.0537471612414837,
"grad_norm": 19.680225372314453,
"learning_rate": 2.6783432019525967e-06,
"loss": 0.4829588532447815,
"step": 1392
},
{
"epoch": 1.0552611657834974,
"grad_norm": 1.8437694311141968,
"learning_rate": 2.6772474413486345e-06,
"loss": 0.7734227180480957,
"step": 1394
},
{
"epoch": 1.056775170325511,
"grad_norm": 1.524431586265564,
"learning_rate": 2.6761500708392727e-06,
"loss": 0.7006011009216309,
"step": 1396
},
{
"epoch": 1.0582891748675245,
"grad_norm": 5.695611000061035,
"learning_rate": 2.675051092144315e-06,
"loss": 0.2834855914115906,
"step": 1398
},
{
"epoch": 1.0598031794095382,
"grad_norm": 2.3736777305603027,
"learning_rate": 2.6739505069860835e-06,
"loss": 0.8014317750930786,
"step": 1400
},
{
"epoch": 1.0613171839515518,
"grad_norm": 1.2972900867462158,
"learning_rate": 2.672848317089419e-06,
"loss": 0.43378323316574097,
"step": 1402
},
{
"epoch": 1.0628311884935655,
"grad_norm": 7.267664909362793,
"learning_rate": 2.6717445241816766e-06,
"loss": 0.41693636775016785,
"step": 1404
},
{
"epoch": 1.0643451930355792,
"grad_norm": 2.0065371990203857,
"learning_rate": 2.670639129992724e-06,
"loss": 0.8223442435264587,
"step": 1406
},
{
"epoch": 1.0658591975775926,
"grad_norm": 1.2673693895339966,
"learning_rate": 2.669532136254939e-06,
"loss": 0.8150830864906311,
"step": 1408
},
{
"epoch": 1.0673732021196063,
"grad_norm": 4.826721668243408,
"learning_rate": 2.668423544703205e-06,
"loss": 1.1558277606964111,
"step": 1410
},
{
"epoch": 1.06888720666162,
"grad_norm": 2.484870433807373,
"learning_rate": 2.6673133570749107e-06,
"loss": 0.986355721950531,
"step": 1412
},
{
"epoch": 1.0704012112036336,
"grad_norm": 1.8060321807861328,
"learning_rate": 2.666201575109945e-06,
"loss": 1.1337124109268188,
"step": 1414
},
{
"epoch": 1.0719152157456473,
"grad_norm": 2.5740833282470703,
"learning_rate": 2.665088200550697e-06,
"loss": 0.8420611619949341,
"step": 1416
},
{
"epoch": 1.073429220287661,
"grad_norm": 3.167254686355591,
"learning_rate": 2.6639732351420494e-06,
"loss": 0.7230285406112671,
"step": 1418
},
{
"epoch": 1.0749432248296744,
"grad_norm": 2.329456329345703,
"learning_rate": 2.66285668063138e-06,
"loss": 0.7629521489143372,
"step": 1420
},
{
"epoch": 1.076457229371688,
"grad_norm": 3.7189221382141113,
"learning_rate": 2.6617385387685576e-06,
"loss": 0.4574144780635834,
"step": 1422
},
{
"epoch": 1.0779712339137018,
"grad_norm": 3.426736354827881,
"learning_rate": 2.6606188113059366e-06,
"loss": 0.37381836771965027,
"step": 1424
},
{
"epoch": 1.0794852384557154,
"grad_norm": 4.68638801574707,
"learning_rate": 2.6594974999983575e-06,
"loss": 0.6116821765899658,
"step": 1426
},
{
"epoch": 1.080999242997729,
"grad_norm": 3.7828402519226074,
"learning_rate": 2.6583746066031428e-06,
"loss": 0.3239648938179016,
"step": 1428
},
{
"epoch": 1.0825132475397425,
"grad_norm": 2.0054421424865723,
"learning_rate": 2.6572501328800947e-06,
"loss": 0.774564802646637,
"step": 1430
},
{
"epoch": 1.0840272520817562,
"grad_norm": 2.6869266033172607,
"learning_rate": 2.656124080591492e-06,
"loss": 0.4418902099132538,
"step": 1432
},
{
"epoch": 1.0855412566237699,
"grad_norm": 3.1833698749542236,
"learning_rate": 2.654996451502087e-06,
"loss": 0.5636769533157349,
"step": 1434
},
{
"epoch": 1.0870552611657835,
"grad_norm": 1.5770076513290405,
"learning_rate": 2.653867247379104e-06,
"loss": 1.1258914470672607,
"step": 1436
},
{
"epoch": 1.0885692657077972,
"grad_norm": 1.492074728012085,
"learning_rate": 2.6527364699922356e-06,
"loss": 0.6968456506729126,
"step": 1438
},
{
"epoch": 1.0900832702498107,
"grad_norm": 2.156663656234741,
"learning_rate": 2.651604121113639e-06,
"loss": 0.42688336968421936,
"step": 1440
},
{
"epoch": 1.0915972747918243,
"grad_norm": 1.730763554573059,
"learning_rate": 2.6504702025179344e-06,
"loss": 0.7748847007751465,
"step": 1442
},
{
"epoch": 1.093111279333838,
"grad_norm": 4.515677452087402,
"learning_rate": 2.6493347159822033e-06,
"loss": 0.3128478527069092,
"step": 1444
},
{
"epoch": 1.0946252838758517,
"grad_norm": 1.7783712148666382,
"learning_rate": 2.6481976632859837e-06,
"loss": 0.5876952409744263,
"step": 1446
},
{
"epoch": 1.0961392884178653,
"grad_norm": 2.324995517730713,
"learning_rate": 2.647059046211268e-06,
"loss": 1.1297352313995361,
"step": 1448
},
{
"epoch": 1.0976532929598788,
"grad_norm": 2.034179925918579,
"learning_rate": 2.645918866542501e-06,
"loss": 0.8161102533340454,
"step": 1450
},
{
"epoch": 1.0991672975018925,
"grad_norm": 2.0529401302337646,
"learning_rate": 2.6447771260665746e-06,
"loss": 0.7258488535881042,
"step": 1452
},
{
"epoch": 1.1006813020439061,
"grad_norm": 2.404632806777954,
"learning_rate": 2.643633826572829e-06,
"loss": 0.8145933747291565,
"step": 1454
},
{
"epoch": 1.1021953065859198,
"grad_norm": 2.9968576431274414,
"learning_rate": 2.6424889698530463e-06,
"loss": 0.7694134712219238,
"step": 1456
},
{
"epoch": 1.1037093111279335,
"grad_norm": 3.773648977279663,
"learning_rate": 2.64134255770145e-06,
"loss": 0.6372519135475159,
"step": 1458
},
{
"epoch": 1.105223315669947,
"grad_norm": 1.944852590560913,
"learning_rate": 2.6401945919147e-06,
"loss": 0.7862814664840698,
"step": 1460
},
{
"epoch": 1.1067373202119606,
"grad_norm": 1.8397468328475952,
"learning_rate": 2.6390450742918934e-06,
"loss": 0.852626621723175,
"step": 1462
},
{
"epoch": 1.1082513247539743,
"grad_norm": 2.014392375946045,
"learning_rate": 2.6378940066345563e-06,
"loss": 0.8128108382225037,
"step": 1464
},
{
"epoch": 1.109765329295988,
"grad_norm": 7.674124240875244,
"learning_rate": 2.636741390746646e-06,
"loss": 0.812615156173706,
"step": 1466
},
{
"epoch": 1.1112793338380016,
"grad_norm": 1.6925760507583618,
"learning_rate": 2.635587228434546e-06,
"loss": 0.4845889210700989,
"step": 1468
},
{
"epoch": 1.112793338380015,
"grad_norm": 1.3222905397415161,
"learning_rate": 2.6344315215070623e-06,
"loss": 0.6777662038803101,
"step": 1470
},
{
"epoch": 1.1143073429220287,
"grad_norm": 5.2306013107299805,
"learning_rate": 2.633274271775423e-06,
"loss": 0.6973274350166321,
"step": 1472
},
{
"epoch": 1.1158213474640424,
"grad_norm": 28.449199676513672,
"learning_rate": 2.6321154810532736e-06,
"loss": 1.161055564880371,
"step": 1474
},
{
"epoch": 1.117335352006056,
"grad_norm": 6.230593204498291,
"learning_rate": 2.6309551511566748e-06,
"loss": 0.7826764583587646,
"step": 1476
},
{
"epoch": 1.1188493565480697,
"grad_norm": 2.1054954528808594,
"learning_rate": 2.629793283904098e-06,
"loss": 0.6974689364433289,
"step": 1478
},
{
"epoch": 1.1203633610900834,
"grad_norm": 10.107316970825195,
"learning_rate": 2.628629881116427e-06,
"loss": 0.9081476926803589,
"step": 1480
},
{
"epoch": 1.1218773656320968,
"grad_norm": 3.2295544147491455,
"learning_rate": 2.6274649446169484e-06,
"loss": 0.28264302015304565,
"step": 1482
},
{
"epoch": 1.1233913701741105,
"grad_norm": 3.182004690170288,
"learning_rate": 2.6262984762313558e-06,
"loss": 0.5937165021896362,
"step": 1484
},
{
"epoch": 1.1249053747161242,
"grad_norm": 1.4822916984558105,
"learning_rate": 2.625130477787741e-06,
"loss": 0.6934922933578491,
"step": 1486
},
{
"epoch": 1.1264193792581378,
"grad_norm": 5.625181674957275,
"learning_rate": 2.623960951116596e-06,
"loss": 0.41950473189353943,
"step": 1488
},
{
"epoch": 1.1279333838001513,
"grad_norm": 1.72547447681427,
"learning_rate": 2.622789898050805e-06,
"loss": 0.5038779973983765,
"step": 1490
},
{
"epoch": 1.129447388342165,
"grad_norm": 3.188124895095825,
"learning_rate": 2.6216173204256478e-06,
"loss": 0.6977412104606628,
"step": 1492
},
{
"epoch": 1.1309613928841786,
"grad_norm": 1.6635669469833374,
"learning_rate": 2.6204432200787906e-06,
"loss": 0.36403128504753113,
"step": 1494
},
{
"epoch": 1.1324753974261923,
"grad_norm": 2.5082645416259766,
"learning_rate": 2.6192675988502873e-06,
"loss": 0.5870634317398071,
"step": 1496
},
{
"epoch": 1.133989401968206,
"grad_norm": 1.4929744005203247,
"learning_rate": 2.6180904585825756e-06,
"loss": 0.5060266256332397,
"step": 1498
},
{
"epoch": 1.1355034065102196,
"grad_norm": 3.6979591846466064,
"learning_rate": 2.6169118011204733e-06,
"loss": 0.4492890536785126,
"step": 1500
},
{
"epoch": 1.137017411052233,
"grad_norm": 1.9358099699020386,
"learning_rate": 2.615731628311175e-06,
"loss": 0.6397398114204407,
"step": 1502
},
{
"epoch": 1.1385314155942468,
"grad_norm": 2.5439043045043945,
"learning_rate": 2.614549942004253e-06,
"loss": 0.7282018661499023,
"step": 1504
},
{
"epoch": 1.1400454201362604,
"grad_norm": 8.694692611694336,
"learning_rate": 2.613366744051648e-06,
"loss": 0.6934101581573486,
"step": 1506
},
{
"epoch": 1.141559424678274,
"grad_norm": 7.270405292510986,
"learning_rate": 2.612182036307673e-06,
"loss": 0.8005611300468445,
"step": 1508
},
{
"epoch": 1.1430734292202875,
"grad_norm": 1.9914963245391846,
"learning_rate": 2.6109958206290047e-06,
"loss": 0.7519674897193909,
"step": 1510
},
{
"epoch": 1.1445874337623012,
"grad_norm": 1.2573553323745728,
"learning_rate": 2.6098080988746847e-06,
"loss": 0.7651998996734619,
"step": 1512
},
{
"epoch": 1.1461014383043149,
"grad_norm": 13.637552261352539,
"learning_rate": 2.6086188729061137e-06,
"loss": 0.3738044500350952,
"step": 1514
},
{
"epoch": 1.1476154428463285,
"grad_norm": 2.9253766536712646,
"learning_rate": 2.6074281445870504e-06,
"loss": 0.7124196290969849,
"step": 1516
},
{
"epoch": 1.1491294473883422,
"grad_norm": 2.5097551345825195,
"learning_rate": 2.6062359157836085e-06,
"loss": 1.0747299194335938,
"step": 1518
},
{
"epoch": 1.1506434519303559,
"grad_norm": 2.58630633354187,
"learning_rate": 2.6050421883642523e-06,
"loss": 0.9966959357261658,
"step": 1520
},
{
"epoch": 1.1521574564723693,
"grad_norm": 2.2582290172576904,
"learning_rate": 2.603846964199795e-06,
"loss": 0.5746228694915771,
"step": 1522
},
{
"epoch": 1.153671461014383,
"grad_norm": 2.2150135040283203,
"learning_rate": 2.602650245163396e-06,
"loss": 1.1288074254989624,
"step": 1524
},
{
"epoch": 1.1551854655563967,
"grad_norm": 1.7310847043991089,
"learning_rate": 2.601452033130556e-06,
"loss": 1.1471965312957764,
"step": 1526
},
{
"epoch": 1.1566994700984103,
"grad_norm": 1.5647629499435425,
"learning_rate": 2.6002523299791185e-06,
"loss": 0.828460693359375,
"step": 1528
},
{
"epoch": 1.158213474640424,
"grad_norm": 4.7715229988098145,
"learning_rate": 2.599051137589261e-06,
"loss": 0.9581698179244995,
"step": 1530
},
{
"epoch": 1.1597274791824375,
"grad_norm": 2.2824630737304688,
"learning_rate": 2.5978484578434956e-06,
"loss": 0.3696361482143402,
"step": 1532
},
{
"epoch": 1.1612414837244511,
"grad_norm": 2.047269344329834,
"learning_rate": 2.5966442926266667e-06,
"loss": 0.42837634682655334,
"step": 1534
},
{
"epoch": 1.1627554882664648,
"grad_norm": 1.830245018005371,
"learning_rate": 2.5954386438259446e-06,
"loss": 0.3391342759132385,
"step": 1536
},
{
"epoch": 1.1642694928084785,
"grad_norm": 4.029056549072266,
"learning_rate": 2.5942315133308264e-06,
"loss": 0.5852560997009277,
"step": 1538
},
{
"epoch": 1.1657834973504921,
"grad_norm": 2.035017251968384,
"learning_rate": 2.5930229030331323e-06,
"loss": 0.8092902302742004,
"step": 1540
},
{
"epoch": 1.1672975018925056,
"grad_norm": 9.123196601867676,
"learning_rate": 2.5918128148269977e-06,
"loss": 0.7252377271652222,
"step": 1542
},
{
"epoch": 1.1688115064345193,
"grad_norm": 3.822357416152954,
"learning_rate": 2.590601250608878e-06,
"loss": 0.7166582942008972,
"step": 1544
},
{
"epoch": 1.170325510976533,
"grad_norm": 2.210566997528076,
"learning_rate": 2.58938821227754e-06,
"loss": 0.9769903421401978,
"step": 1546
},
{
"epoch": 1.1718395155185466,
"grad_norm": 8.526498794555664,
"learning_rate": 2.588173701734061e-06,
"loss": 0.3668955862522125,
"step": 1548
},
{
"epoch": 1.1733535200605603,
"grad_norm": 3.4576306343078613,
"learning_rate": 2.5869577208818264e-06,
"loss": 0.7129625678062439,
"step": 1550
},
{
"epoch": 1.1748675246025737,
"grad_norm": 2.7367055416107178,
"learning_rate": 2.585740271626525e-06,
"loss": 0.7673206329345703,
"step": 1552
},
{
"epoch": 1.1763815291445874,
"grad_norm": 1.9763902425765991,
"learning_rate": 2.5845213558761464e-06,
"loss": 0.5425819158554077,
"step": 1554
},
{
"epoch": 1.177895533686601,
"grad_norm": 1.5533753633499146,
"learning_rate": 2.5833009755409798e-06,
"loss": 0.8152801990509033,
"step": 1556
},
{
"epoch": 1.1794095382286147,
"grad_norm": 1.6180309057235718,
"learning_rate": 2.582079132533609e-06,
"loss": 0.7211406826972961,
"step": 1558
},
{
"epoch": 1.1809235427706284,
"grad_norm": 2.6173770427703857,
"learning_rate": 2.58085582876891e-06,
"loss": 0.7489441633224487,
"step": 1560
},
{
"epoch": 1.182437547312642,
"grad_norm": 3.008863925933838,
"learning_rate": 2.579631066164048e-06,
"loss": 0.8485870957374573,
"step": 1562
},
{
"epoch": 1.1839515518546555,
"grad_norm": 7.870699405670166,
"learning_rate": 2.5784048466384754e-06,
"loss": 0.46112170815467834,
"step": 1564
},
{
"epoch": 1.1854655563966692,
"grad_norm": 3.498415470123291,
"learning_rate": 2.577177172113927e-06,
"loss": 0.39019960165023804,
"step": 1566
},
{
"epoch": 1.1869795609386828,
"grad_norm": 3.178898572921753,
"learning_rate": 2.5759480445144183e-06,
"loss": 0.39782366156578064,
"step": 1568
},
{
"epoch": 1.1884935654806965,
"grad_norm": 2.614898443222046,
"learning_rate": 2.5747174657662415e-06,
"loss": 0.33162638545036316,
"step": 1570
},
{
"epoch": 1.19000757002271,
"grad_norm": 3.082942008972168,
"learning_rate": 2.5734854377979643e-06,
"loss": 0.557581901550293,
"step": 1572
},
{
"epoch": 1.1915215745647236,
"grad_norm": 1.9442788362503052,
"learning_rate": 2.572251962540424e-06,
"loss": 0.8836531639099121,
"step": 1574
},
{
"epoch": 1.1930355791067373,
"grad_norm": 2.360757827758789,
"learning_rate": 2.571017041926727e-06,
"loss": 0.37596502900123596,
"step": 1576
},
{
"epoch": 1.194549583648751,
"grad_norm": 4.975614547729492,
"learning_rate": 2.5697806778922442e-06,
"loss": 0.39887678623199463,
"step": 1578
},
{
"epoch": 1.1960635881907646,
"grad_norm": 1.4703550338745117,
"learning_rate": 2.5685428723746106e-06,
"loss": 0.6991298198699951,
"step": 1580
},
{
"epoch": 1.1975775927327783,
"grad_norm": 2.316718816757202,
"learning_rate": 2.567303627313718e-06,
"loss": 0.671280562877655,
"step": 1582
},
{
"epoch": 1.1990915972747918,
"grad_norm": 5.537595272064209,
"learning_rate": 2.566062944651715e-06,
"loss": 0.3257960081100464,
"step": 1584
},
{
"epoch": 1.2006056018168054,
"grad_norm": 4.8405609130859375,
"learning_rate": 2.5648208263330033e-06,
"loss": 0.37777429819107056,
"step": 1586
},
{
"epoch": 1.202119606358819,
"grad_norm": 2.0979316234588623,
"learning_rate": 2.5635772743042354e-06,
"loss": 0.8450133800506592,
"step": 1588
},
{
"epoch": 1.2036336109008328,
"grad_norm": 3.63385272026062,
"learning_rate": 2.562332290514309e-06,
"loss": 0.4902435839176178,
"step": 1590
},
{
"epoch": 1.2051476154428462,
"grad_norm": 2.6919708251953125,
"learning_rate": 2.5610858769143673e-06,
"loss": 0.7547534704208374,
"step": 1592
},
{
"epoch": 1.2066616199848599,
"grad_norm": 3.5736734867095947,
"learning_rate": 2.5598380354577934e-06,
"loss": 1.0713350772857666,
"step": 1594
},
{
"epoch": 1.2081756245268735,
"grad_norm": 2.7801170349121094,
"learning_rate": 2.5585887681002077e-06,
"loss": 1.0252490043640137,
"step": 1596
},
{
"epoch": 1.2096896290688872,
"grad_norm": 5.497828960418701,
"learning_rate": 2.5573380767994667e-06,
"loss": 0.6267985105514526,
"step": 1598
},
{
"epoch": 1.2112036336109009,
"grad_norm": 2.441645383834839,
"learning_rate": 2.556085963515657e-06,
"loss": 0.7817872166633606,
"step": 1600
},
{
"epoch": 1.2127176381529146,
"grad_norm": 10.523898124694824,
"learning_rate": 2.554832430211095e-06,
"loss": 0.6640578508377075,
"step": 1602
},
{
"epoch": 1.214231642694928,
"grad_norm": 5.218235015869141,
"learning_rate": 2.5535774788503226e-06,
"loss": 0.7082098722457886,
"step": 1604
},
{
"epoch": 1.2157456472369417,
"grad_norm": 1.9254984855651855,
"learning_rate": 2.552321111400102e-06,
"loss": 1.0181281566619873,
"step": 1606
},
{
"epoch": 1.2172596517789553,
"grad_norm": 2.4773366451263428,
"learning_rate": 2.551063329829417e-06,
"loss": 0.36115017533302307,
"step": 1608
},
{
"epoch": 1.218773656320969,
"grad_norm": 3.622504234313965,
"learning_rate": 2.5498041361094675e-06,
"loss": 0.4864136278629303,
"step": 1610
},
{
"epoch": 1.2202876608629827,
"grad_norm": 6.712879657745361,
"learning_rate": 2.548543532213664e-06,
"loss": 0.6208847165107727,
"step": 1612
},
{
"epoch": 1.2218016654049961,
"grad_norm": 1.3727296590805054,
"learning_rate": 2.5472815201176305e-06,
"loss": 0.43101292848587036,
"step": 1614
},
{
"epoch": 1.2233156699470098,
"grad_norm": 1.4034091234207153,
"learning_rate": 2.5460181017991965e-06,
"loss": 0.369363933801651,
"step": 1616
},
{
"epoch": 1.2248296744890235,
"grad_norm": 1.6807384490966797,
"learning_rate": 2.5447532792383934e-06,
"loss": 1.2105756998062134,
"step": 1618
},
{
"epoch": 1.2263436790310371,
"grad_norm": 4.796979904174805,
"learning_rate": 2.5434870544174565e-06,
"loss": 0.9204561114311218,
"step": 1620
},
{
"epoch": 1.2278576835730508,
"grad_norm": 3.0788776874542236,
"learning_rate": 2.542219429320816e-06,
"loss": 0.6513817310333252,
"step": 1622
},
{
"epoch": 1.2293716881150643,
"grad_norm": 2.194180488586426,
"learning_rate": 2.5409504059350997e-06,
"loss": 0.48548203706741333,
"step": 1624
},
{
"epoch": 1.230885692657078,
"grad_norm": 1.6935890913009644,
"learning_rate": 2.5396799862491234e-06,
"loss": 0.8554729223251343,
"step": 1626
},
{
"epoch": 1.2323996971990916,
"grad_norm": 1.9916133880615234,
"learning_rate": 2.5384081722538944e-06,
"loss": 0.9386757612228394,
"step": 1628
},
{
"epoch": 1.2339137017411053,
"grad_norm": 1.5255463123321533,
"learning_rate": 2.537134965942602e-06,
"loss": 0.7088950276374817,
"step": 1630
},
{
"epoch": 1.235427706283119,
"grad_norm": 2.5169317722320557,
"learning_rate": 2.53586036931062e-06,
"loss": 0.6848646998405457,
"step": 1632
},
{
"epoch": 1.2369417108251324,
"grad_norm": 1.438494324684143,
"learning_rate": 2.5345843843554997e-06,
"loss": 0.7582930326461792,
"step": 1634
},
{
"epoch": 1.238455715367146,
"grad_norm": 1.4521311521530151,
"learning_rate": 2.5333070130769693e-06,
"loss": 0.5319505929946899,
"step": 1636
},
{
"epoch": 1.2399697199091597,
"grad_norm": 2.501084089279175,
"learning_rate": 2.5320282574769286e-06,
"loss": 0.7813730239868164,
"step": 1638
},
{
"epoch": 1.2414837244511734,
"grad_norm": 1.1438665390014648,
"learning_rate": 2.530748119559447e-06,
"loss": 0.8257892727851868,
"step": 1640
},
{
"epoch": 1.242997728993187,
"grad_norm": 16.8778076171875,
"learning_rate": 2.5294666013307625e-06,
"loss": 0.7767748832702637,
"step": 1642
},
{
"epoch": 1.2445117335352007,
"grad_norm": 6.9293084144592285,
"learning_rate": 2.528183704799272e-06,
"loss": 0.3963207006454468,
"step": 1644
},
{
"epoch": 1.2460257380772142,
"grad_norm": 4.275975227355957,
"learning_rate": 2.5268994319755364e-06,
"loss": 0.5166948437690735,
"step": 1646
},
{
"epoch": 1.2475397426192278,
"grad_norm": 7.330279350280762,
"learning_rate": 2.5256137848722716e-06,
"loss": 0.4638979434967041,
"step": 1648
},
{
"epoch": 1.2490537471612415,
"grad_norm": 1.5681062936782837,
"learning_rate": 2.5243267655043485e-06,
"loss": 0.7441158294677734,
"step": 1650
},
{
"epoch": 1.2505677517032552,
"grad_norm": 8.350687980651855,
"learning_rate": 2.523038375888787e-06,
"loss": 0.7322012782096863,
"step": 1652
},
{
"epoch": 1.2520817562452686,
"grad_norm": 1.7311092615127563,
"learning_rate": 2.521748618044755e-06,
"loss": 0.7636613249778748,
"step": 1654
},
{
"epoch": 1.2535957607872823,
"grad_norm": 2.5470802783966064,
"learning_rate": 2.520457493993566e-06,
"loss": 0.714220404624939,
"step": 1656
},
{
"epoch": 1.255109765329296,
"grad_norm": 2.575735330581665,
"learning_rate": 2.519165005758674e-06,
"loss": 0.8002930879592896,
"step": 1658
},
{
"epoch": 1.2566237698713096,
"grad_norm": 2.6097710132598877,
"learning_rate": 2.5178711553656694e-06,
"loss": 1.2010271549224854,
"step": 1660
},
{
"epoch": 1.2581377744133233,
"grad_norm": 1.4934430122375488,
"learning_rate": 2.5165759448422783e-06,
"loss": 0.2964455783367157,
"step": 1662
},
{
"epoch": 1.259651778955337,
"grad_norm": 1.8192704916000366,
"learning_rate": 2.5152793762183605e-06,
"loss": 0.3605591356754303,
"step": 1664
},
{
"epoch": 1.2611657834973504,
"grad_norm": 1.903306245803833,
"learning_rate": 2.5139814515259005e-06,
"loss": 0.2693943381309509,
"step": 1666
},
{
"epoch": 1.262679788039364,
"grad_norm": 1.477151870727539,
"learning_rate": 2.5126821727990115e-06,
"loss": 1.16646409034729,
"step": 1668
},
{
"epoch": 1.2641937925813778,
"grad_norm": 2.5828359127044678,
"learning_rate": 2.5113815420739265e-06,
"loss": 0.6955753564834595,
"step": 1670
},
{
"epoch": 1.2657077971233914,
"grad_norm": 2.8693950176239014,
"learning_rate": 2.5100795613889975e-06,
"loss": 1.1036823987960815,
"step": 1672
},
{
"epoch": 1.2672218016654049,
"grad_norm": 4.083218097686768,
"learning_rate": 2.5087762327846932e-06,
"loss": 0.8175725936889648,
"step": 1674
},
{
"epoch": 1.2687358062074185,
"grad_norm": 4.969780445098877,
"learning_rate": 2.5074715583035945e-06,
"loss": 0.5654513835906982,
"step": 1676
},
{
"epoch": 1.2702498107494322,
"grad_norm": 2.3143811225891113,
"learning_rate": 2.506165539990391e-06,
"loss": 0.4701712131500244,
"step": 1678
},
{
"epoch": 1.2717638152914459,
"grad_norm": 1.1072319746017456,
"learning_rate": 2.5048581798918786e-06,
"loss": 0.7916725277900696,
"step": 1680
},
{
"epoch": 1.2732778198334596,
"grad_norm": 7.131253719329834,
"learning_rate": 2.5035494800569568e-06,
"loss": 0.34638112783432007,
"step": 1682
},
{
"epoch": 1.2747918243754732,
"grad_norm": 4.370595455169678,
"learning_rate": 2.5022394425366226e-06,
"loss": 0.29949814081192017,
"step": 1684
},
{
"epoch": 1.2763058289174867,
"grad_norm": 3.430283546447754,
"learning_rate": 2.5009280693839733e-06,
"loss": 0.37807488441467285,
"step": 1686
},
{
"epoch": 1.2778198334595003,
"grad_norm": 4.1393208503723145,
"learning_rate": 2.4996153626541942e-06,
"loss": 0.4743579924106598,
"step": 1688
},
{
"epoch": 1.279333838001514,
"grad_norm": 1.6699633598327637,
"learning_rate": 2.498301324404565e-06,
"loss": 0.518422544002533,
"step": 1690
},
{
"epoch": 1.2808478425435277,
"grad_norm": 1.5848231315612793,
"learning_rate": 2.4969859566944504e-06,
"loss": 0.5937182903289795,
"step": 1692
},
{
"epoch": 1.2823618470855411,
"grad_norm": 4.64746618270874,
"learning_rate": 2.495669261585299e-06,
"loss": 0.5423171520233154,
"step": 1694
},
{
"epoch": 1.2838758516275548,
"grad_norm": 2.7205491065979004,
"learning_rate": 2.494351241140639e-06,
"loss": 0.8974373936653137,
"step": 1696
},
{
"epoch": 1.2853898561695685,
"grad_norm": 2.0516796112060547,
"learning_rate": 2.493031897426076e-06,
"loss": 0.3728986382484436,
"step": 1698
},
{
"epoch": 1.2869038607115821,
"grad_norm": 3.630496025085449,
"learning_rate": 2.4917112325092903e-06,
"loss": 0.3708462715148926,
"step": 1700
},
{
"epoch": 1.2884178652535958,
"grad_norm": 7.670022964477539,
"learning_rate": 2.4903892484600315e-06,
"loss": 0.45419633388519287,
"step": 1702
},
{
"epoch": 1.2899318697956095,
"grad_norm": 3.7088091373443604,
"learning_rate": 2.489065947350117e-06,
"loss": 0.3158243000507355,
"step": 1704
},
{
"epoch": 1.2914458743376231,
"grad_norm": 2.961735486984253,
"learning_rate": 2.4877413312534293e-06,
"loss": 0.33370882272720337,
"step": 1706
},
{
"epoch": 1.2929598788796366,
"grad_norm": 2.4538347721099854,
"learning_rate": 2.4864154022459104e-06,
"loss": 0.794564962387085,
"step": 1708
},
{
"epoch": 1.2944738834216503,
"grad_norm": 2.663095235824585,
"learning_rate": 2.48508816240556e-06,
"loss": 0.7713342905044556,
"step": 1710
},
{
"epoch": 1.295987887963664,
"grad_norm": 2.7029004096984863,
"learning_rate": 2.4837596138124327e-06,
"loss": 0.6392582654953003,
"step": 1712
},
{
"epoch": 1.2975018925056774,
"grad_norm": 2.8968873023986816,
"learning_rate": 2.4824297585486346e-06,
"loss": 0.5847026109695435,
"step": 1714
},
{
"epoch": 1.299015897047691,
"grad_norm": 8.163978576660156,
"learning_rate": 2.4810985986983177e-06,
"loss": 0.6873840689659119,
"step": 1716
},
{
"epoch": 1.3005299015897047,
"grad_norm": 4.473104953765869,
"learning_rate": 2.479766136347682e-06,
"loss": 0.792141318321228,
"step": 1718
},
{
"epoch": 1.3020439061317184,
"grad_norm": 34.86464309692383,
"learning_rate": 2.478432373584964e-06,
"loss": 0.27867889404296875,
"step": 1720
},
{
"epoch": 1.303557910673732,
"grad_norm": 4.670142650604248,
"learning_rate": 2.477097312500444e-06,
"loss": 0.7536966800689697,
"step": 1722
},
{
"epoch": 1.3050719152157457,
"grad_norm": 1.702530026435852,
"learning_rate": 2.4757609551864307e-06,
"loss": 1.0583717823028564,
"step": 1724
},
{
"epoch": 1.3065859197577594,
"grad_norm": 2.3609373569488525,
"learning_rate": 2.4744233037372697e-06,
"loss": 0.28235501050949097,
"step": 1726
},
{
"epoch": 1.3080999242997728,
"grad_norm": 3.98068904876709,
"learning_rate": 2.4730843602493315e-06,
"loss": 1.0268464088439941,
"step": 1728
},
{
"epoch": 1.3096139288417865,
"grad_norm": 2.23313307762146,
"learning_rate": 2.4717441268210127e-06,
"loss": 0.5403386354446411,
"step": 1730
},
{
"epoch": 1.3111279333838002,
"grad_norm": 6.894443035125732,
"learning_rate": 2.4704026055527315e-06,
"loss": 0.6712117791175842,
"step": 1732
},
{
"epoch": 1.3126419379258139,
"grad_norm": 1.8715323209762573,
"learning_rate": 2.469059798546924e-06,
"loss": 1.0742299556732178,
"step": 1734
},
{
"epoch": 1.3141559424678273,
"grad_norm": 4.290128231048584,
"learning_rate": 2.4677157079080418e-06,
"loss": 1.2617570161819458,
"step": 1736
},
{
"epoch": 1.315669947009841,
"grad_norm": 3.091566324234009,
"learning_rate": 2.4663703357425475e-06,
"loss": 1.0468664169311523,
"step": 1738
},
{
"epoch": 1.3171839515518546,
"grad_norm": 3.078235149383545,
"learning_rate": 2.4650236841589136e-06,
"loss": 0.8127296566963196,
"step": 1740
},
{
"epoch": 1.3186979560938683,
"grad_norm": 3.087818145751953,
"learning_rate": 2.4636757552676146e-06,
"loss": 1.1390008926391602,
"step": 1742
},
{
"epoch": 1.320211960635882,
"grad_norm": 2.46710467338562,
"learning_rate": 2.4623265511811316e-06,
"loss": 0.5823556184768677,
"step": 1744
},
{
"epoch": 1.3217259651778956,
"grad_norm": 3.8972294330596924,
"learning_rate": 2.4609760740139393e-06,
"loss": 0.612288236618042,
"step": 1746
},
{
"epoch": 1.323239969719909,
"grad_norm": 3.4614830017089844,
"learning_rate": 2.4596243258825107e-06,
"loss": 0.3169248700141907,
"step": 1748
},
{
"epoch": 1.3247539742619228,
"grad_norm": 1.4857923984527588,
"learning_rate": 2.458271308905309e-06,
"loss": 0.7403193712234497,
"step": 1750
},
{
"epoch": 1.3262679788039364,
"grad_norm": 2.9274091720581055,
"learning_rate": 2.4569170252027877e-06,
"loss": 0.29984942078590393,
"step": 1752
},
{
"epoch": 1.32778198334595,
"grad_norm": 2.223027467727661,
"learning_rate": 2.4555614768973842e-06,
"loss": 0.39937183260917664,
"step": 1754
},
{
"epoch": 1.3292959878879635,
"grad_norm": 1.9060097932815552,
"learning_rate": 2.454204666113517e-06,
"loss": 0.1987786591053009,
"step": 1756
},
{
"epoch": 1.3308099924299772,
"grad_norm": 2.1526589393615723,
"learning_rate": 2.452846594977585e-06,
"loss": 0.5061262845993042,
"step": 1758
},
{
"epoch": 1.3323239969719909,
"grad_norm": 2.4836909770965576,
"learning_rate": 2.451487265617962e-06,
"loss": 0.9262844324111938,
"step": 1760
},
{
"epoch": 1.3338380015140046,
"grad_norm": 3.6183834075927734,
"learning_rate": 2.450126680164992e-06,
"loss": 0.4936198890209198,
"step": 1762
},
{
"epoch": 1.3353520060560182,
"grad_norm": 25.579692840576172,
"learning_rate": 2.4487648407509897e-06,
"loss": 0.7663915157318115,
"step": 1764
},
{
"epoch": 1.336866010598032,
"grad_norm": 3.81601619720459,
"learning_rate": 2.447401749510234e-06,
"loss": 0.7129659056663513,
"step": 1766
},
{
"epoch": 1.3383800151400453,
"grad_norm": 2.3081166744232178,
"learning_rate": 2.446037408578965e-06,
"loss": 0.358845055103302,
"step": 1768
},
{
"epoch": 1.339894019682059,
"grad_norm": 3.634019136428833,
"learning_rate": 2.444671820095383e-06,
"loss": 0.8582119941711426,
"step": 1770
},
{
"epoch": 1.3414080242240727,
"grad_norm": 2.7738685607910156,
"learning_rate": 2.443304986199642e-06,
"loss": 0.7216231822967529,
"step": 1772
},
{
"epoch": 1.3429220287660864,
"grad_norm": 2.8546862602233887,
"learning_rate": 2.4419369090338485e-06,
"loss": 0.5570340156555176,
"step": 1774
},
{
"epoch": 1.3444360333080998,
"grad_norm": 1.778089165687561,
"learning_rate": 2.4405675907420575e-06,
"loss": 1.1104695796966553,
"step": 1776
},
{
"epoch": 1.3459500378501135,
"grad_norm": 2.5244617462158203,
"learning_rate": 2.439197033470269e-06,
"loss": 1.0705184936523438,
"step": 1778
},
{
"epoch": 1.3474640423921271,
"grad_norm": 4.3281941413879395,
"learning_rate": 2.437825239366424e-06,
"loss": 0.7556070685386658,
"step": 1780
},
{
"epoch": 1.3489780469341408,
"grad_norm": 2.8140037059783936,
"learning_rate": 2.4364522105804026e-06,
"loss": 1.0987634658813477,
"step": 1782
},
{
"epoch": 1.3504920514761545,
"grad_norm": 9.171343803405762,
"learning_rate": 2.4350779492640203e-06,
"loss": 0.33399084210395813,
"step": 1784
},
{
"epoch": 1.3520060560181681,
"grad_norm": 4.052097320556641,
"learning_rate": 2.433702457571024e-06,
"loss": 0.598379373550415,
"step": 1786
},
{
"epoch": 1.3535200605601818,
"grad_norm": 2.5520999431610107,
"learning_rate": 2.432325737657087e-06,
"loss": 0.4708116054534912,
"step": 1788
},
{
"epoch": 1.3550340651021953,
"grad_norm": 27.633031845092773,
"learning_rate": 2.430947791679811e-06,
"loss": 0.9031903147697449,
"step": 1790
},
{
"epoch": 1.356548069644209,
"grad_norm": 3.972987651824951,
"learning_rate": 2.4295686217987165e-06,
"loss": 0.3332988917827606,
"step": 1792
},
{
"epoch": 1.3580620741862226,
"grad_norm": 0.5608827471733093,
"learning_rate": 2.4281882301752424e-06,
"loss": 0.4873623251914978,
"step": 1794
},
{
"epoch": 1.359576078728236,
"grad_norm": 2.9013092517852783,
"learning_rate": 2.4268066189727427e-06,
"loss": 0.40898972749710083,
"step": 1796
},
{
"epoch": 1.3610900832702497,
"grad_norm": 2.2795474529266357,
"learning_rate": 2.4254237903564834e-06,
"loss": 1.0034469366073608,
"step": 1798
},
{
"epoch": 1.3626040878122634,
"grad_norm": 9.1624755859375,
"learning_rate": 2.424039746493638e-06,
"loss": 0.3764766454696655,
"step": 1800
},
{
"epoch": 1.364118092354277,
"grad_norm": 13.661250114440918,
"learning_rate": 2.4226544895532837e-06,
"loss": 0.7672209143638611,
"step": 1802
},
{
"epoch": 1.3656320968962907,
"grad_norm": 1.79072904586792,
"learning_rate": 2.4212680217063996e-06,
"loss": 0.6887940764427185,
"step": 1804
},
{
"epoch": 1.3671461014383044,
"grad_norm": 1.4073220491409302,
"learning_rate": 2.4198803451258624e-06,
"loss": 0.26766905188560486,
"step": 1806
},
{
"epoch": 1.368660105980318,
"grad_norm": 3.8320226669311523,
"learning_rate": 2.418491461986444e-06,
"loss": 0.5123397707939148,
"step": 1808
},
{
"epoch": 1.3701741105223315,
"grad_norm": 5.280299663543701,
"learning_rate": 2.4171013744648053e-06,
"loss": 0.7153005599975586,
"step": 1810
},
{
"epoch": 1.3716881150643452,
"grad_norm": 1.7055562734603882,
"learning_rate": 2.4157100847394964e-06,
"loss": 0.8119074702262878,
"step": 1812
},
{
"epoch": 1.3732021196063589,
"grad_norm": 2.1829960346221924,
"learning_rate": 2.414317594990951e-06,
"loss": 0.3074583113193512,
"step": 1814
},
{
"epoch": 1.3747161241483725,
"grad_norm": 1.3141074180603027,
"learning_rate": 2.412923907401483e-06,
"loss": 1.1208783388137817,
"step": 1816
},
{
"epoch": 1.376230128690386,
"grad_norm": 4.528420925140381,
"learning_rate": 2.411529024155284e-06,
"loss": 0.7128023505210876,
"step": 1818
},
{
"epoch": 1.3777441332323996,
"grad_norm": 1.3782234191894531,
"learning_rate": 2.41013294743842e-06,
"loss": 0.7611986994743347,
"step": 1820
},
{
"epoch": 1.3792581377744133,
"grad_norm": 4.164299964904785,
"learning_rate": 2.408735679438825e-06,
"loss": 0.16963380575180054,
"step": 1822
},
{
"epoch": 1.380772142316427,
"grad_norm": 11.236958503723145,
"learning_rate": 2.4073372223463043e-06,
"loss": 0.8310223817825317,
"step": 1824
},
{
"epoch": 1.3822861468584406,
"grad_norm": 5.005019187927246,
"learning_rate": 2.405937578352523e-06,
"loss": 0.8467041850090027,
"step": 1826
},
{
"epoch": 1.3838001514004543,
"grad_norm": 2.2640397548675537,
"learning_rate": 2.404536749651007e-06,
"loss": 1.099280595779419,
"step": 1828
},
{
"epoch": 1.3853141559424678,
"grad_norm": 2.679772138595581,
"learning_rate": 2.40313473843714e-06,
"loss": 1.1264312267303467,
"step": 1830
},
{
"epoch": 1.3868281604844814,
"grad_norm": 1.491464376449585,
"learning_rate": 2.4017315469081583e-06,
"loss": 1.1441285610198975,
"step": 1832
},
{
"epoch": 1.388342165026495,
"grad_norm": 4.6747541427612305,
"learning_rate": 2.400327177263148e-06,
"loss": 0.6041007041931152,
"step": 1834
},
{
"epoch": 1.3898561695685088,
"grad_norm": 4.323363304138184,
"learning_rate": 2.3989216317030422e-06,
"loss": 0.7551705241203308,
"step": 1836
},
{
"epoch": 1.3913701741105222,
"grad_norm": 1.805139422416687,
"learning_rate": 2.3975149124306153e-06,
"loss": 0.9166613817214966,
"step": 1838
},
{
"epoch": 1.3928841786525359,
"grad_norm": 1.763720989227295,
"learning_rate": 2.396107021650482e-06,
"loss": 0.3972744047641754,
"step": 1840
},
{
"epoch": 1.3943981831945496,
"grad_norm": 4.463486194610596,
"learning_rate": 2.3946979615690946e-06,
"loss": 0.8842801451683044,
"step": 1842
},
{
"epoch": 1.3959121877365632,
"grad_norm": 2.2309823036193848,
"learning_rate": 2.393287734394735e-06,
"loss": 0.8152608275413513,
"step": 1844
},
{
"epoch": 1.397426192278577,
"grad_norm": 1.9330055713653564,
"learning_rate": 2.3918763423375162e-06,
"loss": 0.7752976417541504,
"step": 1846
},
{
"epoch": 1.3989401968205906,
"grad_norm": 2.6784393787384033,
"learning_rate": 2.3904637876093765e-06,
"loss": 0.4102567136287689,
"step": 1848
},
{
"epoch": 1.400454201362604,
"grad_norm": 1.4778658151626587,
"learning_rate": 2.3890500724240754e-06,
"loss": 0.7943294048309326,
"step": 1850
},
{
"epoch": 1.4019682059046177,
"grad_norm": 2.401735305786133,
"learning_rate": 2.387635198997193e-06,
"loss": 0.407833993434906,
"step": 1852
},
{
"epoch": 1.4034822104466314,
"grad_norm": 2.32517409324646,
"learning_rate": 2.386219169546122e-06,
"loss": 1.1640748977661133,
"step": 1854
},
{
"epoch": 1.404996214988645,
"grad_norm": 7.444009304046631,
"learning_rate": 2.3848019862900684e-06,
"loss": 0.5029972195625305,
"step": 1856
},
{
"epoch": 1.4065102195306585,
"grad_norm": 2.249410390853882,
"learning_rate": 2.383383651450047e-06,
"loss": 0.738408088684082,
"step": 1858
},
{
"epoch": 1.4080242240726721,
"grad_norm": 6.936520576477051,
"learning_rate": 2.3819641672488756e-06,
"loss": 0.6538509726524353,
"step": 1860
},
{
"epoch": 1.4095382286146858,
"grad_norm": 3.3028481006622314,
"learning_rate": 2.3805435359111753e-06,
"loss": 0.5028783082962036,
"step": 1862
},
{
"epoch": 1.4110522331566995,
"grad_norm": 4.275155544281006,
"learning_rate": 2.379121759663363e-06,
"loss": 0.8367002010345459,
"step": 1864
},
{
"epoch": 1.4125662376987131,
"grad_norm": 2.1487843990325928,
"learning_rate": 2.377698840733652e-06,
"loss": 0.7150232791900635,
"step": 1866
},
{
"epoch": 1.4140802422407268,
"grad_norm": 2.527843475341797,
"learning_rate": 2.3762747813520437e-06,
"loss": 0.8052014112472534,
"step": 1868
},
{
"epoch": 1.4155942467827405,
"grad_norm": 1.5240310430526733,
"learning_rate": 2.3748495837503302e-06,
"loss": 1.0286638736724854,
"step": 1870
},
{
"epoch": 1.417108251324754,
"grad_norm": 4.434390068054199,
"learning_rate": 2.3734232501620843e-06,
"loss": 1.0798012018203735,
"step": 1872
},
{
"epoch": 1.4186222558667676,
"grad_norm": 13.43934440612793,
"learning_rate": 2.371995782822661e-06,
"loss": 0.3294588625431061,
"step": 1874
},
{
"epoch": 1.4201362604087813,
"grad_norm": 3.540721893310547,
"learning_rate": 2.3705671839691915e-06,
"loss": 1.0713666677474976,
"step": 1876
},
{
"epoch": 1.4216502649507947,
"grad_norm": 2.280219078063965,
"learning_rate": 2.3691374558405806e-06,
"loss": 0.7495984435081482,
"step": 1878
},
{
"epoch": 1.4231642694928084,
"grad_norm": 1.9954382181167603,
"learning_rate": 2.3677066006775023e-06,
"loss": 0.6960147023200989,
"step": 1880
},
{
"epoch": 1.424678274034822,
"grad_norm": 7.99771785736084,
"learning_rate": 2.3662746207223975e-06,
"loss": 1.105854868888855,
"step": 1882
},
{
"epoch": 1.4261922785768357,
"grad_norm": 1.6709688901901245,
"learning_rate": 2.36484151821947e-06,
"loss": 0.3433768153190613,
"step": 1884
},
{
"epoch": 1.4277062831188494,
"grad_norm": 2.90567684173584,
"learning_rate": 2.363407295414681e-06,
"loss": 0.28263920545578003,
"step": 1886
},
{
"epoch": 1.429220287660863,
"grad_norm": 6.311897277832031,
"learning_rate": 2.361971954555751e-06,
"loss": 0.5564744472503662,
"step": 1888
},
{
"epoch": 1.4307342922028767,
"grad_norm": 1.9530609846115112,
"learning_rate": 2.3605354978921497e-06,
"loss": 0.7947676777839661,
"step": 1890
},
{
"epoch": 1.4322482967448902,
"grad_norm": 2.0929741859436035,
"learning_rate": 2.359097927675097e-06,
"loss": 1.120530366897583,
"step": 1892
},
{
"epoch": 1.4337623012869039,
"grad_norm": 2.99371600151062,
"learning_rate": 2.3576592461575562e-06,
"loss": 0.656132698059082,
"step": 1894
},
{
"epoch": 1.4352763058289175,
"grad_norm": 1.237682580947876,
"learning_rate": 2.356219455594234e-06,
"loss": 1.1159588098526,
"step": 1896
},
{
"epoch": 1.4367903103709312,
"grad_norm": 2.7394959926605225,
"learning_rate": 2.3547785582415757e-06,
"loss": 0.6585239171981812,
"step": 1898
},
{
"epoch": 1.4383043149129446,
"grad_norm": 3.580203056335449,
"learning_rate": 2.353336556357759e-06,
"loss": 0.28547555208206177,
"step": 1900
},
{
"epoch": 1.4398183194549583,
"grad_norm": 1.8582044839859009,
"learning_rate": 2.351893452202694e-06,
"loss": 1.0720411539077759,
"step": 1902
},
{
"epoch": 1.441332323996972,
"grad_norm": 1.8799811601638794,
"learning_rate": 2.350449248038018e-06,
"loss": 0.7256011366844177,
"step": 1904
},
{
"epoch": 1.4428463285389856,
"grad_norm": 1.9906892776489258,
"learning_rate": 2.349003946127093e-06,
"loss": 0.6670038104057312,
"step": 1906
},
{
"epoch": 1.4443603330809993,
"grad_norm": 7.704482078552246,
"learning_rate": 2.3475575487349996e-06,
"loss": 0.959648609161377,
"step": 1908
},
{
"epoch": 1.445874337623013,
"grad_norm": 1.963994026184082,
"learning_rate": 2.3461100581285374e-06,
"loss": 0.7701181769371033,
"step": 1910
},
{
"epoch": 1.4473883421650264,
"grad_norm": 3.128945827484131,
"learning_rate": 2.344661476576217e-06,
"loss": 0.6871995329856873,
"step": 1912
},
{
"epoch": 1.44890234670704,
"grad_norm": 7.108216285705566,
"learning_rate": 2.343211806348261e-06,
"loss": 0.4328969120979309,
"step": 1914
},
{
"epoch": 1.4504163512490538,
"grad_norm": 4.275369644165039,
"learning_rate": 2.3417610497165965e-06,
"loss": 0.42519325017929077,
"step": 1916
},
{
"epoch": 1.4519303557910674,
"grad_norm": 2.9187445640563965,
"learning_rate": 2.3403092089548533e-06,
"loss": 0.41934680938720703,
"step": 1918
},
{
"epoch": 1.4534443603330809,
"grad_norm": 2.5698537826538086,
"learning_rate": 2.3388562863383623e-06,
"loss": 0.6157204508781433,
"step": 1920
},
{
"epoch": 1.4549583648750946,
"grad_norm": 2.6239511966705322,
"learning_rate": 2.3374022841441473e-06,
"loss": 0.7134471535682678,
"step": 1922
},
{
"epoch": 1.4564723694171082,
"grad_norm": 1.3179781436920166,
"learning_rate": 2.3359472046509254e-06,
"loss": 0.6652729511260986,
"step": 1924
},
{
"epoch": 1.457986373959122,
"grad_norm": 1.326392412185669,
"learning_rate": 2.3344910501391012e-06,
"loss": 0.306835412979126,
"step": 1926
},
{
"epoch": 1.4595003785011356,
"grad_norm": 1.8047044277191162,
"learning_rate": 2.3330338228907653e-06,
"loss": 0.7175745964050293,
"step": 1928
},
{
"epoch": 1.4610143830431492,
"grad_norm": 1.8204708099365234,
"learning_rate": 2.3315755251896883e-06,
"loss": 0.760126531124115,
"step": 1930
},
{
"epoch": 1.4625283875851627,
"grad_norm": 4.059135913848877,
"learning_rate": 2.3301161593213196e-06,
"loss": 0.6605585813522339,
"step": 1932
},
{
"epoch": 1.4640423921271764,
"grad_norm": 2.046104669570923,
"learning_rate": 2.328655727572781e-06,
"loss": 0.6077765226364136,
"step": 1934
},
{
"epoch": 1.46555639666919,
"grad_norm": 7.639178276062012,
"learning_rate": 2.327194232232866e-06,
"loss": 0.2979111671447754,
"step": 1936
},
{
"epoch": 1.4670704012112037,
"grad_norm": 6.912994861602783,
"learning_rate": 2.3257316755920356e-06,
"loss": 1.1609059572219849,
"step": 1938
},
{
"epoch": 1.4685844057532171,
"grad_norm": 1.4372037649154663,
"learning_rate": 2.3242680599424116e-06,
"loss": 0.6751406192779541,
"step": 1940
},
{
"epoch": 1.4700984102952308,
"grad_norm": 1.8448315858840942,
"learning_rate": 2.3228033875777787e-06,
"loss": 0.6058751344680786,
"step": 1942
},
{
"epoch": 1.4716124148372445,
"grad_norm": 1.561957597732544,
"learning_rate": 2.321337660793574e-06,
"loss": 0.688287079334259,
"step": 1944
},
{
"epoch": 1.4731264193792581,
"grad_norm": 3.6917073726654053,
"learning_rate": 2.319870881886891e-06,
"loss": 0.671501100063324,
"step": 1946
},
{
"epoch": 1.4746404239212718,
"grad_norm": 2.6265804767608643,
"learning_rate": 2.318403053156469e-06,
"loss": 1.0544801950454712,
"step": 1948
},
{
"epoch": 1.4761544284632855,
"grad_norm": 2.026118755340576,
"learning_rate": 2.316934176902694e-06,
"loss": 0.6782279014587402,
"step": 1950
},
{
"epoch": 1.4776684330052992,
"grad_norm": 2.275644540786743,
"learning_rate": 2.3154642554275942e-06,
"loss": 0.3071862459182739,
"step": 1952
},
{
"epoch": 1.4791824375473126,
"grad_norm": 1.539597988128662,
"learning_rate": 2.313993291034834e-06,
"loss": 0.6210603713989258,
"step": 1954
},
{
"epoch": 1.4806964420893263,
"grad_norm": 1.7491732835769653,
"learning_rate": 2.312521286029714e-06,
"loss": 0.663459300994873,
"step": 1956
},
{
"epoch": 1.48221044663134,
"grad_norm": 0.9735066890716553,
"learning_rate": 2.3110482427191647e-06,
"loss": 0.3123417794704437,
"step": 1958
},
{
"epoch": 1.4837244511733534,
"grad_norm": 2.4855947494506836,
"learning_rate": 2.309574163411745e-06,
"loss": 0.6801527738571167,
"step": 1960
},
{
"epoch": 1.485238455715367,
"grad_norm": 3.174555540084839,
"learning_rate": 2.308099050417636e-06,
"loss": 0.6908773183822632,
"step": 1962
},
{
"epoch": 1.4867524602573807,
"grad_norm": 1.6514712572097778,
"learning_rate": 2.3066229060486395e-06,
"loss": 1.0852717161178589,
"step": 1964
},
{
"epoch": 1.4882664647993944,
"grad_norm": 5.090445041656494,
"learning_rate": 2.3051457326181727e-06,
"loss": 0.5876284837722778,
"step": 1966
},
{
"epoch": 1.489780469341408,
"grad_norm": 1.7408846616744995,
"learning_rate": 2.303667532441268e-06,
"loss": 0.6825834512710571,
"step": 1968
},
{
"epoch": 1.4912944738834217,
"grad_norm": 1.8605716228485107,
"learning_rate": 2.3021883078345644e-06,
"loss": 1.0978846549987793,
"step": 1970
},
{
"epoch": 1.4928084784254354,
"grad_norm": 3.411088466644287,
"learning_rate": 2.3007080611163075e-06,
"loss": 0.63323974609375,
"step": 1972
},
{
"epoch": 1.4943224829674489,
"grad_norm": 1.4026780128479004,
"learning_rate": 2.2992267946063442e-06,
"loss": 0.6705458760261536,
"step": 1974
},
{
"epoch": 1.4958364875094625,
"grad_norm": 1.7022696733474731,
"learning_rate": 2.2977445106261203e-06,
"loss": 0.6665899157524109,
"step": 1976
},
{
"epoch": 1.4973504920514762,
"grad_norm": 2.7791049480438232,
"learning_rate": 2.2962612114986766e-06,
"loss": 1.0865336656570435,
"step": 1978
},
{
"epoch": 1.4988644965934899,
"grad_norm": 1.447414517402649,
"learning_rate": 2.2947768995486425e-06,
"loss": 0.29112380743026733,
"step": 1980
},
{
"epoch": 1.5003785011355033,
"grad_norm": 1.5562597513198853,
"learning_rate": 2.293291577102238e-06,
"loss": 1.0406157970428467,
"step": 1982
},
{
"epoch": 1.501892505677517,
"grad_norm": 2.4863104820251465,
"learning_rate": 2.291805246487264e-06,
"loss": 0.8998382687568665,
"step": 1984
},
{
"epoch": 1.5034065102195306,
"grad_norm": 2.075108289718628,
"learning_rate": 2.2903179100331036e-06,
"loss": 1.0479289293289185,
"step": 1986
},
{
"epoch": 1.5049205147615443,
"grad_norm": 3.2120509147644043,
"learning_rate": 2.2888295700707136e-06,
"loss": 0.2061680257320404,
"step": 1988
},
{
"epoch": 1.506434519303558,
"grad_norm": 1.16193425655365,
"learning_rate": 2.287340228932626e-06,
"loss": 0.8722069263458252,
"step": 1990
},
{
"epoch": 1.5079485238455717,
"grad_norm": 4.635110378265381,
"learning_rate": 2.2858498889529404e-06,
"loss": 0.6780929565429688,
"step": 1992
},
{
"epoch": 1.5094625283875853,
"grad_norm": 1.9462432861328125,
"learning_rate": 2.284358552467323e-06,
"loss": 1.0604909658432007,
"step": 1994
},
{
"epoch": 1.5109765329295988,
"grad_norm": 3.5864200592041016,
"learning_rate": 2.282866221813001e-06,
"loss": 0.7383281588554382,
"step": 1996
},
{
"epoch": 1.5124905374716124,
"grad_norm": 1.6705453395843506,
"learning_rate": 2.2813728993287584e-06,
"loss": 1.1641281843185425,
"step": 1998
},
{
"epoch": 1.5140045420136259,
"grad_norm": 4.274797439575195,
"learning_rate": 2.279878587354936e-06,
"loss": 0.7858713865280151,
"step": 2000
},
{
"epoch": 1.5155185465556396,
"grad_norm": 3.163729429244995,
"learning_rate": 2.2783832882334237e-06,
"loss": 0.5768117308616638,
"step": 2002
},
{
"epoch": 1.5170325510976532,
"grad_norm": 1.7296491861343384,
"learning_rate": 2.2768870043076593e-06,
"loss": 0.46675315499305725,
"step": 2004
},
{
"epoch": 1.518546555639667,
"grad_norm": 1.1777745485305786,
"learning_rate": 2.2753897379226236e-06,
"loss": 1.0568435192108154,
"step": 2006
},
{
"epoch": 1.5200605601816806,
"grad_norm": 2.120549201965332,
"learning_rate": 2.2738914914248375e-06,
"loss": 0.3315390944480896,
"step": 2008
},
{
"epoch": 1.5215745647236942,
"grad_norm": 1.4694416522979736,
"learning_rate": 2.272392267162356e-06,
"loss": 0.9205579161643982,
"step": 2010
},
{
"epoch": 1.523088569265708,
"grad_norm": 4.467464447021484,
"learning_rate": 2.27089206748477e-06,
"loss": 0.2087993174791336,
"step": 2012
},
{
"epoch": 1.5246025738077216,
"grad_norm": 1.785783290863037,
"learning_rate": 2.269390894743196e-06,
"loss": 0.7023442387580872,
"step": 2014
},
{
"epoch": 1.526116578349735,
"grad_norm": 9.867462158203125,
"learning_rate": 2.2678887512902772e-06,
"loss": 0.27967381477355957,
"step": 2016
},
{
"epoch": 1.5276305828917487,
"grad_norm": 1.9199988842010498,
"learning_rate": 2.266385639480177e-06,
"loss": 0.7091067433357239,
"step": 2018
},
{
"epoch": 1.5291445874337621,
"grad_norm": 33.07830810546875,
"learning_rate": 2.264881561668577e-06,
"loss": 0.6637606620788574,
"step": 2020
},
{
"epoch": 1.5306585919757758,
"grad_norm": 2.0145633220672607,
"learning_rate": 2.263376520212673e-06,
"loss": 0.7130411863327026,
"step": 2022
},
{
"epoch": 1.5321725965177895,
"grad_norm": 2.2729172706604004,
"learning_rate": 2.261870517471171e-06,
"loss": 1.0831512212753296,
"step": 2024
},
{
"epoch": 1.5336866010598031,
"grad_norm": 6.332074165344238,
"learning_rate": 2.260363555804282e-06,
"loss": 0.20880655944347382,
"step": 2026
},
{
"epoch": 1.5352006056018168,
"grad_norm": 2.0282723903656006,
"learning_rate": 2.2588556375737217e-06,
"loss": 0.6106334328651428,
"step": 2028
},
{
"epoch": 1.5367146101438305,
"grad_norm": 1.8590947389602661,
"learning_rate": 2.2573467651427044e-06,
"loss": 0.24480338394641876,
"step": 2030
},
{
"epoch": 1.5382286146858442,
"grad_norm": 2.0381529331207275,
"learning_rate": 2.2558369408759395e-06,
"loss": 0.40811577439308167,
"step": 2032
},
{
"epoch": 1.5397426192278578,
"grad_norm": 2.440659761428833,
"learning_rate": 2.254326167139628e-06,
"loss": 0.9319549798965454,
"step": 2034
},
{
"epoch": 1.5412566237698713,
"grad_norm": 6.197136878967285,
"learning_rate": 2.2528144463014607e-06,
"loss": 0.6879948377609253,
"step": 2036
},
{
"epoch": 1.542770628311885,
"grad_norm": 4.5604472160339355,
"learning_rate": 2.2513017807306087e-06,
"loss": 0.5867362022399902,
"step": 2038
},
{
"epoch": 1.5442846328538986,
"grad_norm": 5.366823196411133,
"learning_rate": 2.2497881727977283e-06,
"loss": 1.0565264225006104,
"step": 2040
},
{
"epoch": 1.545798637395912,
"grad_norm": 1.2136204242706299,
"learning_rate": 2.24827362487495e-06,
"loss": 0.34370917081832886,
"step": 2042
},
{
"epoch": 1.5473126419379257,
"grad_norm": 2.217613697052002,
"learning_rate": 2.246758139335878e-06,
"loss": 0.501958966255188,
"step": 2044
},
{
"epoch": 1.5488266464799394,
"grad_norm": 43.41156768798828,
"learning_rate": 2.245241718555586e-06,
"loss": 0.3205106854438782,
"step": 2046
},
{
"epoch": 1.550340651021953,
"grad_norm": 8.517773628234863,
"learning_rate": 2.2437243649106126e-06,
"loss": 0.7185165286064148,
"step": 2048
},
{
"epoch": 1.5518546555639667,
"grad_norm": 1.816855549812317,
"learning_rate": 2.2422060807789602e-06,
"loss": 0.5383141040802002,
"step": 2050
},
{
"epoch": 1.5533686601059804,
"grad_norm": 1.642162561416626,
"learning_rate": 2.240686868540088e-06,
"loss": 0.3428835868835449,
"step": 2052
},
{
"epoch": 1.554882664647994,
"grad_norm": 3.3189697265625,
"learning_rate": 2.23916673057491e-06,
"loss": 0.16990278661251068,
"step": 2054
},
{
"epoch": 1.5563966691900075,
"grad_norm": 2.2869858741760254,
"learning_rate": 2.2376456692657917e-06,
"loss": 0.4399016499519348,
"step": 2056
},
{
"epoch": 1.5579106737320212,
"grad_norm": 2.9362711906433105,
"learning_rate": 2.2361236869965447e-06,
"loss": 0.19548234343528748,
"step": 2058
},
{
"epoch": 1.5594246782740349,
"grad_norm": 3.8211545944213867,
"learning_rate": 2.234600786152425e-06,
"loss": 1.248904824256897,
"step": 2060
},
{
"epoch": 1.5609386828160483,
"grad_norm": 6.065986156463623,
"learning_rate": 2.2330769691201267e-06,
"loss": 0.7616989612579346,
"step": 2062
},
{
"epoch": 1.562452687358062,
"grad_norm": 9.370930671691895,
"learning_rate": 2.231552238287781e-06,
"loss": 0.3147810399532318,
"step": 2064
},
{
"epoch": 1.5639666919000756,
"grad_norm": 3.1894898414611816,
"learning_rate": 2.230026596044951e-06,
"loss": 0.34935462474823,
"step": 2066
},
{
"epoch": 1.5654806964420893,
"grad_norm": 2.467068672180176,
"learning_rate": 2.2285000447826276e-06,
"loss": 0.7595210671424866,
"step": 2068
},
{
"epoch": 1.566994700984103,
"grad_norm": 2.4427568912506104,
"learning_rate": 2.2269725868932266e-06,
"loss": 0.4047495424747467,
"step": 2070
},
{
"epoch": 1.5685087055261167,
"grad_norm": 4.277834892272949,
"learning_rate": 2.2254442247705855e-06,
"loss": 0.626862645149231,
"step": 2072
},
{
"epoch": 1.5700227100681303,
"grad_norm": 1.623204231262207,
"learning_rate": 2.223914960809958e-06,
"loss": 1.1122664213180542,
"step": 2074
},
{
"epoch": 1.571536714610144,
"grad_norm": 1.6308976411819458,
"learning_rate": 2.222384797408011e-06,
"loss": 1.063917875289917,
"step": 2076
},
{
"epoch": 1.5730507191521574,
"grad_norm": 3.854956865310669,
"learning_rate": 2.220853736962821e-06,
"loss": 0.5665196776390076,
"step": 2078
},
{
"epoch": 1.574564723694171,
"grad_norm": 3.0789151191711426,
"learning_rate": 2.2193217818738714e-06,
"loss": 0.5702639818191528,
"step": 2080
},
{
"epoch": 1.5760787282361846,
"grad_norm": 2.537348508834839,
"learning_rate": 2.217788934542047e-06,
"loss": 0.32417529821395874,
"step": 2082
},
{
"epoch": 1.5775927327781982,
"grad_norm": 3.38523530960083,
"learning_rate": 2.2162551973696306e-06,
"loss": 0.35859784483909607,
"step": 2084
},
{
"epoch": 1.579106737320212,
"grad_norm": 1.4633924961090088,
"learning_rate": 2.2147205727603e-06,
"loss": 0.6509751081466675,
"step": 2086
},
{
"epoch": 1.5806207418622256,
"grad_norm": 1.6090911626815796,
"learning_rate": 2.2131850631191237e-06,
"loss": 0.7736411094665527,
"step": 2088
},
{
"epoch": 1.5821347464042392,
"grad_norm": 2.0669453144073486,
"learning_rate": 2.2116486708525576e-06,
"loss": 0.7473424673080444,
"step": 2090
},
{
"epoch": 1.583648750946253,
"grad_norm": 7.861624717712402,
"learning_rate": 2.2101113983684397e-06,
"loss": 0.3567483127117157,
"step": 2092
},
{
"epoch": 1.5851627554882666,
"grad_norm": 2.284384250640869,
"learning_rate": 2.208573248075989e-06,
"loss": 0.7789891958236694,
"step": 2094
},
{
"epoch": 1.5866767600302802,
"grad_norm": 5.964828968048096,
"learning_rate": 2.2070342223857986e-06,
"loss": 0.69113689661026,
"step": 2096
},
{
"epoch": 1.5881907645722937,
"grad_norm": 1.429634690284729,
"learning_rate": 2.205494323709835e-06,
"loss": 0.4352221190929413,
"step": 2098
},
{
"epoch": 1.5897047691143074,
"grad_norm": 10.728472709655762,
"learning_rate": 2.2039535544614325e-06,
"loss": 0.5330414772033691,
"step": 2100
},
{
"epoch": 1.5912187736563208,
"grad_norm": 2.230041980743408,
"learning_rate": 2.2024119170552886e-06,
"loss": 0.15315233170986176,
"step": 2102
},
{
"epoch": 1.5927327781983345,
"grad_norm": 2.4584739208221436,
"learning_rate": 2.2008694139074623e-06,
"loss": 0.2879572808742523,
"step": 2104
},
{
"epoch": 1.5942467827403481,
"grad_norm": 8.412919998168945,
"learning_rate": 2.19932604743537e-06,
"loss": 0.6374467015266418,
"step": 2106
},
{
"epoch": 1.5957607872823618,
"grad_norm": 1.1442950963974,
"learning_rate": 2.19778182005778e-06,
"loss": 1.0432395935058594,
"step": 2108
},
{
"epoch": 1.5972747918243755,
"grad_norm": 2.137449026107788,
"learning_rate": 2.1962367341948103e-06,
"loss": 1.115038275718689,
"step": 2110
},
{
"epoch": 1.5987887963663892,
"grad_norm": 5.937266826629639,
"learning_rate": 2.194690792267925e-06,
"loss": 0.8900066018104553,
"step": 2112
},
{
"epoch": 1.6003028009084028,
"grad_norm": 3.706021785736084,
"learning_rate": 2.1931439966999285e-06,
"loss": 0.7339775562286377,
"step": 2114
},
{
"epoch": 1.6018168054504165,
"grad_norm": 1.7120450735092163,
"learning_rate": 2.191596349914964e-06,
"loss": 0.5568338632583618,
"step": 2116
},
{
"epoch": 1.60333080999243,
"grad_norm": 4.790015697479248,
"learning_rate": 2.1900478543385073e-06,
"loss": 0.6627650260925293,
"step": 2118
},
{
"epoch": 1.6048448145344436,
"grad_norm": 1.9481124877929688,
"learning_rate": 2.188498512397367e-06,
"loss": 0.8415189981460571,
"step": 2120
},
{
"epoch": 1.6063588190764573,
"grad_norm": 6.659837245941162,
"learning_rate": 2.186948326519675e-06,
"loss": 0.2886730134487152,
"step": 2122
},
{
"epoch": 1.6078728236184707,
"grad_norm": 3.406694173812866,
"learning_rate": 2.1853972991348895e-06,
"loss": 0.1624942421913147,
"step": 2124
},
{
"epoch": 1.6093868281604844,
"grad_norm": 1.9080997705459595,
"learning_rate": 2.1838454326737836e-06,
"loss": 0.4276244044303894,
"step": 2126
},
{
"epoch": 1.610900832702498,
"grad_norm": 5.854907035827637,
"learning_rate": 2.182292729568448e-06,
"loss": 0.4545116424560547,
"step": 2128
},
{
"epoch": 1.6124148372445117,
"grad_norm": 2.8690412044525146,
"learning_rate": 2.180739192252284e-06,
"loss": 0.7988955974578857,
"step": 2130
},
{
"epoch": 1.6139288417865254,
"grad_norm": 8.291343688964844,
"learning_rate": 2.1791848231600002e-06,
"loss": 0.6213572025299072,
"step": 2132
},
{
"epoch": 1.615442846328539,
"grad_norm": 0.862676203250885,
"learning_rate": 2.1776296247276077e-06,
"loss": 0.4094568192958832,
"step": 2134
},
{
"epoch": 1.6169568508705527,
"grad_norm": 2.8311262130737305,
"learning_rate": 2.1760735993924196e-06,
"loss": 0.7716742157936096,
"step": 2136
},
{
"epoch": 1.6184708554125662,
"grad_norm": 3.6992971897125244,
"learning_rate": 2.174516749593044e-06,
"loss": 0.7096973061561584,
"step": 2138
},
{
"epoch": 1.6199848599545799,
"grad_norm": 2.3221044540405273,
"learning_rate": 2.172959077769379e-06,
"loss": 0.8890570402145386,
"step": 2140
},
{
"epoch": 1.6214988644965935,
"grad_norm": 2.638880491256714,
"learning_rate": 2.1714005863626143e-06,
"loss": 0.5222177505493164,
"step": 2142
},
{
"epoch": 1.623012869038607,
"grad_norm": 2.155319929122925,
"learning_rate": 2.169841277815221e-06,
"loss": 0.7492667436599731,
"step": 2144
},
{
"epoch": 1.6245268735806206,
"grad_norm": 2.3153676986694336,
"learning_rate": 2.168281154570954e-06,
"loss": 0.9271564483642578,
"step": 2146
},
{
"epoch": 1.6260408781226343,
"grad_norm": 3.7721846103668213,
"learning_rate": 2.1667202190748423e-06,
"loss": 0.3474150002002716,
"step": 2148
},
{
"epoch": 1.627554882664648,
"grad_norm": 1.7081248760223389,
"learning_rate": 2.165158473773189e-06,
"loss": 0.2777302861213684,
"step": 2150
},
{
"epoch": 1.6290688872066617,
"grad_norm": 3.3190102577209473,
"learning_rate": 2.163595921113567e-06,
"loss": 0.4995487928390503,
"step": 2152
},
{
"epoch": 1.6305828917486753,
"grad_norm": 8.794017791748047,
"learning_rate": 2.1620325635448127e-06,
"loss": 0.7314831614494324,
"step": 2154
},
{
"epoch": 1.632096896290689,
"grad_norm": 0.8514871001243591,
"learning_rate": 2.1604684035170253e-06,
"loss": 0.34603098034858704,
"step": 2156
},
{
"epoch": 1.6336109008327027,
"grad_norm": 3.585616111755371,
"learning_rate": 2.158903443481561e-06,
"loss": 0.6739398837089539,
"step": 2158
},
{
"epoch": 1.635124905374716,
"grad_norm": 4.3569865226745605,
"learning_rate": 2.157337685891031e-06,
"loss": 0.7613204717636108,
"step": 2160
},
{
"epoch": 1.6366389099167298,
"grad_norm": 5.168086051940918,
"learning_rate": 2.155771133199294e-06,
"loss": 0.5170891880989075,
"step": 2162
},
{
"epoch": 1.6381529144587432,
"grad_norm": 1.6620736122131348,
"learning_rate": 2.154203787861458e-06,
"loss": 0.5840808153152466,
"step": 2164
},
{
"epoch": 1.639666919000757,
"grad_norm": 3.4618961811065674,
"learning_rate": 2.1526356523338704e-06,
"loss": 0.6698583364486694,
"step": 2166
},
{
"epoch": 1.6411809235427706,
"grad_norm": 3.096137523651123,
"learning_rate": 2.1510667290741183e-06,
"loss": 0.43474137783050537,
"step": 2168
},
{
"epoch": 1.6426949280847842,
"grad_norm": 13.721327781677246,
"learning_rate": 2.149497020541023e-06,
"loss": 0.6251839399337769,
"step": 2170
},
{
"epoch": 1.644208932626798,
"grad_norm": 1.7023292779922485,
"learning_rate": 2.1479265291946365e-06,
"loss": 0.8443311452865601,
"step": 2172
},
{
"epoch": 1.6457229371688116,
"grad_norm": 2.377385377883911,
"learning_rate": 2.146355257496239e-06,
"loss": 0.7048377990722656,
"step": 2174
},
{
"epoch": 1.6472369417108252,
"grad_norm": 6.913028717041016,
"learning_rate": 2.1447832079083306e-06,
"loss": 0.3259809911251068,
"step": 2176
},
{
"epoch": 1.648750946252839,
"grad_norm": 1.9534987211227417,
"learning_rate": 2.1432103828946335e-06,
"loss": 1.068019986152649,
"step": 2178
},
{
"epoch": 1.6502649507948524,
"grad_norm": 2.9702231884002686,
"learning_rate": 2.141636784920083e-06,
"loss": 0.7385921478271484,
"step": 2180
},
{
"epoch": 1.651778955336866,
"grad_norm": 3.8918232917785645,
"learning_rate": 2.1400624164508283e-06,
"loss": 0.6695861220359802,
"step": 2182
},
{
"epoch": 1.6532929598788795,
"grad_norm": 3.2664449214935303,
"learning_rate": 2.1384872799542233e-06,
"loss": 0.5747061371803284,
"step": 2184
},
{
"epoch": 1.6548069644208931,
"grad_norm": 9.48470687866211,
"learning_rate": 2.1369113778988272e-06,
"loss": 0.6012070178985596,
"step": 2186
},
{
"epoch": 1.6563209689629068,
"grad_norm": 2.4145398139953613,
"learning_rate": 2.135334712754399e-06,
"loss": 0.2863891124725342,
"step": 2188
},
{
"epoch": 1.6578349735049205,
"grad_norm": 4.188508033752441,
"learning_rate": 2.1337572869918926e-06,
"loss": 1.1720845699310303,
"step": 2190
},
{
"epoch": 1.6593489780469342,
"grad_norm": 2.1696295738220215,
"learning_rate": 2.132179103083455e-06,
"loss": 1.126940131187439,
"step": 2192
},
{
"epoch": 1.6608629825889478,
"grad_norm": 2.3900654315948486,
"learning_rate": 2.1306001635024204e-06,
"loss": 1.0954455137252808,
"step": 2194
},
{
"epoch": 1.6623769871309615,
"grad_norm": 2.878375768661499,
"learning_rate": 2.129020470723309e-06,
"loss": 0.6001009345054626,
"step": 2196
},
{
"epoch": 1.6638909916729752,
"grad_norm": 26.621519088745117,
"learning_rate": 2.127440027221819e-06,
"loss": 0.47328245639801025,
"step": 2198
},
{
"epoch": 1.6654049962149886,
"grad_norm": 2.122884511947632,
"learning_rate": 2.1258588354748273e-06,
"loss": 0.3275623619556427,
"step": 2200
},
{
"epoch": 1.6669190007570023,
"grad_norm": 1.6605173349380493,
"learning_rate": 2.1242768979603817e-06,
"loss": 0.28766143321990967,
"step": 2202
},
{
"epoch": 1.668433005299016,
"grad_norm": 4.779479026794434,
"learning_rate": 2.1226942171577e-06,
"loss": 0.6301637887954712,
"step": 2204
},
{
"epoch": 1.6699470098410294,
"grad_norm": 1.1953977346420288,
"learning_rate": 2.1211107955471653e-06,
"loss": 0.7062985897064209,
"step": 2206
},
{
"epoch": 1.671461014383043,
"grad_norm": 2.85499906539917,
"learning_rate": 2.1195266356103194e-06,
"loss": 0.2579890191555023,
"step": 2208
},
{
"epoch": 1.6729750189250567,
"grad_norm": 6.130537033081055,
"learning_rate": 2.117941739829864e-06,
"loss": 0.2943982183933258,
"step": 2210
},
{
"epoch": 1.6744890234670704,
"grad_norm": 2.1135518550872803,
"learning_rate": 2.116356110689652e-06,
"loss": 0.5594744086265564,
"step": 2212
},
{
"epoch": 1.676003028009084,
"grad_norm": 2.256420850753784,
"learning_rate": 2.1147697506746865e-06,
"loss": 0.6435813307762146,
"step": 2214
},
{
"epoch": 1.6775170325510977,
"grad_norm": 430.08251953125,
"learning_rate": 2.1131826622711157e-06,
"loss": 0.3069487512111664,
"step": 2216
},
{
"epoch": 1.6790310370931114,
"grad_norm": 1.5891367197036743,
"learning_rate": 2.1115948479662303e-06,
"loss": 0.7901256680488586,
"step": 2218
},
{
"epoch": 1.6805450416351249,
"grad_norm": 3.6547927856445312,
"learning_rate": 2.1100063102484567e-06,
"loss": 0.4188877046108246,
"step": 2220
},
{
"epoch": 1.6820590461771385,
"grad_norm": 1.7542855739593506,
"learning_rate": 2.1084170516073566e-06,
"loss": 0.2679920792579651,
"step": 2222
},
{
"epoch": 1.6835730507191522,
"grad_norm": 5.140130996704102,
"learning_rate": 2.106827074533622e-06,
"loss": 0.24302098155021667,
"step": 2224
},
{
"epoch": 1.6850870552611656,
"grad_norm": 3.187342405319214,
"learning_rate": 2.1052363815190685e-06,
"loss": 0.1624186784029007,
"step": 2226
},
{
"epoch": 1.6866010598031793,
"grad_norm": 6.9088053703308105,
"learning_rate": 2.1036449750566363e-06,
"loss": 0.6162082552909851,
"step": 2228
},
{
"epoch": 1.688115064345193,
"grad_norm": 1.30314040184021,
"learning_rate": 2.102052857640381e-06,
"loss": 0.2861323654651642,
"step": 2230
},
{
"epoch": 1.6896290688872067,
"grad_norm": 5.282270908355713,
"learning_rate": 2.1004600317654764e-06,
"loss": 0.7010675668716431,
"step": 2232
},
{
"epoch": 1.6911430734292203,
"grad_norm": 6.541689872741699,
"learning_rate": 2.0988664999282025e-06,
"loss": 0.17050687968730927,
"step": 2234
},
{
"epoch": 1.692657077971234,
"grad_norm": 2.042179584503174,
"learning_rate": 2.0972722646259476e-06,
"loss": 1.1382827758789062,
"step": 2236
},
{
"epoch": 1.6941710825132477,
"grad_norm": 2.3376784324645996,
"learning_rate": 2.095677328357202e-06,
"loss": 0.23813597857952118,
"step": 2238
},
{
"epoch": 1.6956850870552613,
"grad_norm": 1.6474254131317139,
"learning_rate": 2.0940816936215553e-06,
"loss": 0.5763483047485352,
"step": 2240
},
{
"epoch": 1.6971990915972748,
"grad_norm": 2.9063918590545654,
"learning_rate": 2.0924853629196918e-06,
"loss": 0.34474313259124756,
"step": 2242
},
{
"epoch": 1.6987130961392884,
"grad_norm": 2.5557656288146973,
"learning_rate": 2.090888338753385e-06,
"loss": 0.21587520837783813,
"step": 2244
},
{
"epoch": 1.700227100681302,
"grad_norm": 5.877704620361328,
"learning_rate": 2.0892906236254966e-06,
"loss": 0.621183454990387,
"step": 2246
},
{
"epoch": 1.7017411052233156,
"grad_norm": 4.238709926605225,
"learning_rate": 2.08769222003997e-06,
"loss": 1.0777775049209595,
"step": 2248
},
{
"epoch": 1.7032551097653292,
"grad_norm": 2.4976248741149902,
"learning_rate": 2.08609313050183e-06,
"loss": 0.6810624599456787,
"step": 2250
},
{
"epoch": 1.704769114307343,
"grad_norm": 5.799039363861084,
"learning_rate": 2.0844933575171725e-06,
"loss": 0.6343029141426086,
"step": 2252
},
{
"epoch": 1.7062831188493566,
"grad_norm": 1.8170959949493408,
"learning_rate": 2.0828929035931685e-06,
"loss": 0.7549158930778503,
"step": 2254
},
{
"epoch": 1.7077971233913702,
"grad_norm": 4.467005252838135,
"learning_rate": 2.0812917712380533e-06,
"loss": 0.6830211281776428,
"step": 2256
},
{
"epoch": 1.709311127933384,
"grad_norm": 0.8398280739784241,
"learning_rate": 2.0796899629611274e-06,
"loss": 0.6890422105789185,
"step": 2258
},
{
"epoch": 1.7108251324753976,
"grad_norm": 1.046366572380066,
"learning_rate": 2.078087481272749e-06,
"loss": 0.4800116717815399,
"step": 2260
},
{
"epoch": 1.712339137017411,
"grad_norm": 1.4331594705581665,
"learning_rate": 2.0764843286843326e-06,
"loss": 0.8580260276794434,
"step": 2262
},
{
"epoch": 1.7138531415594247,
"grad_norm": 19.763986587524414,
"learning_rate": 2.0748805077083444e-06,
"loss": 0.6199619174003601,
"step": 2264
},
{
"epoch": 1.7153671461014381,
"grad_norm": 17.90291976928711,
"learning_rate": 2.0732760208582967e-06,
"loss": 0.7525902390480042,
"step": 2266
},
{
"epoch": 1.7168811506434518,
"grad_norm": 4.046377658843994,
"learning_rate": 2.0716708706487476e-06,
"loss": 0.7198786735534668,
"step": 2268
},
{
"epoch": 1.7183951551854655,
"grad_norm": 9.317425727844238,
"learning_rate": 2.0700650595952925e-06,
"loss": 0.30152690410614014,
"step": 2270
},
{
"epoch": 1.7199091597274792,
"grad_norm": 1.2944056987762451,
"learning_rate": 2.0684585902145637e-06,
"loss": 0.6058851480484009,
"step": 2272
},
{
"epoch": 1.7214231642694928,
"grad_norm": 2.4493825435638428,
"learning_rate": 2.0668514650242252e-06,
"loss": 0.6572599411010742,
"step": 2274
},
{
"epoch": 1.7229371688115065,
"grad_norm": 2.4681243896484375,
"learning_rate": 2.0652436865429685e-06,
"loss": 0.14060214161872864,
"step": 2276
},
{
"epoch": 1.7244511733535202,
"grad_norm": 3.9299867153167725,
"learning_rate": 2.0636352572905093e-06,
"loss": 0.6264467239379883,
"step": 2278
},
{
"epoch": 1.7259651778955338,
"grad_norm": 4.141777992248535,
"learning_rate": 2.0620261797875824e-06,
"loss": 0.2831032872200012,
"step": 2280
},
{
"epoch": 1.7274791824375473,
"grad_norm": 2.3005599975585938,
"learning_rate": 2.06041645655594e-06,
"loss": 0.6381456851959229,
"step": 2282
},
{
"epoch": 1.728993186979561,
"grad_norm": 1.5281527042388916,
"learning_rate": 2.0588060901183444e-06,
"loss": 0.6487545371055603,
"step": 2284
},
{
"epoch": 1.7305071915215746,
"grad_norm": 2.2702927589416504,
"learning_rate": 2.0571950829985673e-06,
"loss": 0.810280978679657,
"step": 2286
},
{
"epoch": 1.732021196063588,
"grad_norm": 1.50748610496521,
"learning_rate": 2.0555834377213843e-06,
"loss": 0.3577505648136139,
"step": 2288
},
{
"epoch": 1.7335352006056017,
"grad_norm": 2.3487930297851562,
"learning_rate": 2.0539711568125707e-06,
"loss": 0.3521404564380646,
"step": 2290
},
{
"epoch": 1.7350492051476154,
"grad_norm": 8.966695785522461,
"learning_rate": 2.052358242798898e-06,
"loss": 0.2824268341064453,
"step": 2292
},
{
"epoch": 1.736563209689629,
"grad_norm": 1.4676111936569214,
"learning_rate": 2.050744698208131e-06,
"loss": 0.4738492965698242,
"step": 2294
},
{
"epoch": 1.7380772142316427,
"grad_norm": 3.0044350624084473,
"learning_rate": 2.0491305255690207e-06,
"loss": 0.7092089653015137,
"step": 2296
},
{
"epoch": 1.7395912187736564,
"grad_norm": 1.4396655559539795,
"learning_rate": 2.047515727411304e-06,
"loss": 0.6721881628036499,
"step": 2298
},
{
"epoch": 1.74110522331567,
"grad_norm": 8.363606452941895,
"learning_rate": 2.0459003062656975e-06,
"loss": 0.2827519178390503,
"step": 2300
},
{
"epoch": 1.7426192278576835,
"grad_norm": 2.10673451423645,
"learning_rate": 2.0442842646638944e-06,
"loss": 0.7053329944610596,
"step": 2302
},
{
"epoch": 1.7441332323996972,
"grad_norm": 2.727703809738159,
"learning_rate": 2.0426676051385603e-06,
"loss": 0.22457894682884216,
"step": 2304
},
{
"epoch": 1.7456472369417109,
"grad_norm": 3.643404960632324,
"learning_rate": 2.041050330223328e-06,
"loss": 0.8219473361968994,
"step": 2306
},
{
"epoch": 1.7471612414837243,
"grad_norm": 3.9443776607513428,
"learning_rate": 2.0394324424527976e-06,
"loss": 0.2641485929489136,
"step": 2308
},
{
"epoch": 1.748675246025738,
"grad_norm": 0.7181191444396973,
"learning_rate": 2.0378139443625263e-06,
"loss": 0.7425059080123901,
"step": 2310
},
{
"epoch": 1.7501892505677517,
"grad_norm": 6.626009464263916,
"learning_rate": 2.03619483848903e-06,
"loss": 0.7613123655319214,
"step": 2312
},
{
"epoch": 1.7517032551097653,
"grad_norm": 1.737387776374817,
"learning_rate": 2.034575127369776e-06,
"loss": 0.9189949035644531,
"step": 2314
},
{
"epoch": 1.753217259651779,
"grad_norm": 3.098576068878174,
"learning_rate": 2.0329548135431816e-06,
"loss": 1.189211130142212,
"step": 2316
},
{
"epoch": 1.7547312641937927,
"grad_norm": 2.3874411582946777,
"learning_rate": 2.0313338995486073e-06,
"loss": 0.3131447732448578,
"step": 2318
},
{
"epoch": 1.7562452687358063,
"grad_norm": 2.1706600189208984,
"learning_rate": 2.0297123879263546e-06,
"loss": 1.1029870510101318,
"step": 2320
},
{
"epoch": 1.7577592732778198,
"grad_norm": 1.2848676443099976,
"learning_rate": 2.0280902812176607e-06,
"loss": 0.7795432209968567,
"step": 2322
},
{
"epoch": 1.7592732778198334,
"grad_norm": 1.6532896757125854,
"learning_rate": 2.0264675819646977e-06,
"loss": 1.0654683113098145,
"step": 2324
},
{
"epoch": 1.7607872823618471,
"grad_norm": 1.1202808618545532,
"learning_rate": 2.0248442927105635e-06,
"loss": 1.0802593231201172,
"step": 2326
},
{
"epoch": 1.7623012869038606,
"grad_norm": 2.4864604473114014,
"learning_rate": 2.0232204159992833e-06,
"loss": 0.6008468270301819,
"step": 2328
},
{
"epoch": 1.7638152914458742,
"grad_norm": 1.4563044309616089,
"learning_rate": 2.021595954375801e-06,
"loss": 0.6921793222427368,
"step": 2330
},
{
"epoch": 1.765329295987888,
"grad_norm": 1.7934669256210327,
"learning_rate": 2.0199709103859784e-06,
"loss": 0.28670886158943176,
"step": 2332
},
{
"epoch": 1.7668433005299016,
"grad_norm": 1.6941759586334229,
"learning_rate": 2.0183452865765893e-06,
"loss": 1.050851821899414,
"step": 2334
},
{
"epoch": 1.7683573050719152,
"grad_norm": 2.231344223022461,
"learning_rate": 2.0167190854953167e-06,
"loss": 0.1752273291349411,
"step": 2336
},
{
"epoch": 1.769871309613929,
"grad_norm": 1.5495513677597046,
"learning_rate": 2.0150923096907473e-06,
"loss": 0.6756109595298767,
"step": 2338
},
{
"epoch": 1.7713853141559426,
"grad_norm": 3.340039014816284,
"learning_rate": 2.0134649617123697e-06,
"loss": 1.0384888648986816,
"step": 2340
},
{
"epoch": 1.7728993186979563,
"grad_norm": 7.249871253967285,
"learning_rate": 2.011837044110569e-06,
"loss": 0.9125099182128906,
"step": 2342
},
{
"epoch": 1.7744133232399697,
"grad_norm": 2.0098390579223633,
"learning_rate": 2.0102085594366227e-06,
"loss": 1.1059890985488892,
"step": 2344
},
{
"epoch": 1.7759273277819834,
"grad_norm": 2.3969979286193848,
"learning_rate": 2.0085795102426962e-06,
"loss": 0.3932843804359436,
"step": 2346
},
{
"epoch": 1.7774413323239968,
"grad_norm": 3.099766254425049,
"learning_rate": 2.0069498990818417e-06,
"loss": 0.5124934911727905,
"step": 2348
},
{
"epoch": 1.7789553368660105,
"grad_norm": 1.7208963632583618,
"learning_rate": 2.00531972850799e-06,
"loss": 0.593222439289093,
"step": 2350
},
{
"epoch": 1.7804693414080242,
"grad_norm": 4.858328819274902,
"learning_rate": 2.00368900107595e-06,
"loss": 0.2706722617149353,
"step": 2352
},
{
"epoch": 1.7819833459500378,
"grad_norm": 3.2357375621795654,
"learning_rate": 2.0020577193414025e-06,
"loss": 1.0699340105056763,
"step": 2354
},
{
"epoch": 1.7834973504920515,
"grad_norm": 2.757415771484375,
"learning_rate": 2.0004258858608973e-06,
"loss": 0.650534451007843,
"step": 2356
},
{
"epoch": 1.7850113550340652,
"grad_norm": 1.8998754024505615,
"learning_rate": 1.9987935031918496e-06,
"loss": 0.7294416427612305,
"step": 2358
},
{
"epoch": 1.7865253595760788,
"grad_norm": 1.2810745239257812,
"learning_rate": 1.997160573892534e-06,
"loss": 0.6762998104095459,
"step": 2360
},
{
"epoch": 1.7880393641180925,
"grad_norm": 1.3143075704574585,
"learning_rate": 1.9955271005220826e-06,
"loss": 1.1342246532440186,
"step": 2362
},
{
"epoch": 1.789553368660106,
"grad_norm": 18.63542938232422,
"learning_rate": 1.9938930856404796e-06,
"loss": 0.4164232015609741,
"step": 2364
},
{
"epoch": 1.7910673732021196,
"grad_norm": 7.862457752227783,
"learning_rate": 1.9922585318085586e-06,
"loss": 0.2773081958293915,
"step": 2366
},
{
"epoch": 1.7925813777441333,
"grad_norm": 0.9495812058448792,
"learning_rate": 1.990623441587998e-06,
"loss": 0.15149936079978943,
"step": 2368
},
{
"epoch": 1.7940953822861467,
"grad_norm": 2.372420072555542,
"learning_rate": 1.988987817541315e-06,
"loss": 1.0846363306045532,
"step": 2370
},
{
"epoch": 1.7956093868281604,
"grad_norm": 1.3067657947540283,
"learning_rate": 1.9873516622318655e-06,
"loss": 0.5807230472564697,
"step": 2372
},
{
"epoch": 1.797123391370174,
"grad_norm": 3.343245029449463,
"learning_rate": 1.9857149782238376e-06,
"loss": 0.45187991857528687,
"step": 2374
},
{
"epoch": 1.7986373959121877,
"grad_norm": 1.814562439918518,
"learning_rate": 1.9840777680822465e-06,
"loss": 1.1235315799713135,
"step": 2376
},
{
"epoch": 1.8001514004542014,
"grad_norm": 1.2937862873077393,
"learning_rate": 1.982440034372934e-06,
"loss": 0.9832783341407776,
"step": 2378
},
{
"epoch": 1.801665404996215,
"grad_norm": 3.880544900894165,
"learning_rate": 1.9808017796625614e-06,
"loss": 0.31947067379951477,
"step": 2380
},
{
"epoch": 1.8031794095382288,
"grad_norm": 2.1368255615234375,
"learning_rate": 1.979163006518606e-06,
"loss": 0.7589985728263855,
"step": 2382
},
{
"epoch": 1.8046934140802422,
"grad_norm": 1.9387202262878418,
"learning_rate": 1.977523717509359e-06,
"loss": 0.7736470699310303,
"step": 2384
},
{
"epoch": 1.8062074186222559,
"grad_norm": 2.6687638759613037,
"learning_rate": 1.9758839152039183e-06,
"loss": 0.5925498604774475,
"step": 2386
},
{
"epoch": 1.8077214231642695,
"grad_norm": 1.7037615776062012,
"learning_rate": 1.974243602172188e-06,
"loss": 0.707631528377533,
"step": 2388
},
{
"epoch": 1.809235427706283,
"grad_norm": 2.250582456588745,
"learning_rate": 1.972602780984871e-06,
"loss": 0.6910974979400635,
"step": 2390
},
{
"epoch": 1.8107494322482967,
"grad_norm": 4.812062740325928,
"learning_rate": 1.9709614542134684e-06,
"loss": 0.6874426603317261,
"step": 2392
},
{
"epoch": 1.8122634367903103,
"grad_norm": 1.8871315717697144,
"learning_rate": 1.969319624430272e-06,
"loss": 0.24301689863204956,
"step": 2394
},
{
"epoch": 1.813777441332324,
"grad_norm": 12.313353538513184,
"learning_rate": 1.9676772942083627e-06,
"loss": 0.32100293040275574,
"step": 2396
},
{
"epoch": 1.8152914458743377,
"grad_norm": 1.4709579944610596,
"learning_rate": 1.9660344661216058e-06,
"loss": 0.7021037340164185,
"step": 2398
},
{
"epoch": 1.8168054504163513,
"grad_norm": 4.930418014526367,
"learning_rate": 1.9643911427446458e-06,
"loss": 0.22974015772342682,
"step": 2400
},
{
"epoch": 1.818319454958365,
"grad_norm": 1.7876191139221191,
"learning_rate": 1.9627473266529055e-06,
"loss": 0.1252487748861313,
"step": 2402
},
{
"epoch": 1.8198334595003784,
"grad_norm": 4.775298595428467,
"learning_rate": 1.9611030204225776e-06,
"loss": 0.6840951442718506,
"step": 2404
},
{
"epoch": 1.8213474640423921,
"grad_norm": 3.2811765670776367,
"learning_rate": 1.9594582266306244e-06,
"loss": 0.8643097877502441,
"step": 2406
},
{
"epoch": 1.8228614685844058,
"grad_norm": 3.3247106075286865,
"learning_rate": 1.957812947854771e-06,
"loss": 0.342410683631897,
"step": 2408
},
{
"epoch": 1.8243754731264192,
"grad_norm": 14.739418029785156,
"learning_rate": 1.9561671866735053e-06,
"loss": 0.4793134331703186,
"step": 2410
},
{
"epoch": 1.825889477668433,
"grad_norm": 1.838046669960022,
"learning_rate": 1.954520945666068e-06,
"loss": 0.6665730476379395,
"step": 2412
},
{
"epoch": 1.8274034822104466,
"grad_norm": 6.243978500366211,
"learning_rate": 1.9528742274124527e-06,
"loss": 0.8040502071380615,
"step": 2414
},
{
"epoch": 1.8289174867524602,
"grad_norm": 2.532534599304199,
"learning_rate": 1.9512270344934027e-06,
"loss": 0.8092038631439209,
"step": 2416
},
{
"epoch": 1.830431491294474,
"grad_norm": 2.357820987701416,
"learning_rate": 1.949579369490403e-06,
"loss": 0.5882593393325806,
"step": 2418
},
{
"epoch": 1.8319454958364876,
"grad_norm": 1.8288893699645996,
"learning_rate": 1.9479312349856796e-06,
"loss": 0.6844436526298523,
"step": 2420
},
{
"epoch": 1.8334595003785013,
"grad_norm": 1.6580768823623657,
"learning_rate": 1.946282633562194e-06,
"loss": 0.8981972336769104,
"step": 2422
},
{
"epoch": 1.834973504920515,
"grad_norm": 0.8140391111373901,
"learning_rate": 1.94463356780364e-06,
"loss": 0.7706785202026367,
"step": 2424
},
{
"epoch": 1.8364875094625284,
"grad_norm": 4.756402969360352,
"learning_rate": 1.942984040294438e-06,
"loss": 0.7183281779289246,
"step": 2426
},
{
"epoch": 1.838001514004542,
"grad_norm": 0.7826236486434937,
"learning_rate": 1.9413340536197326e-06,
"loss": 0.5470402240753174,
"step": 2428
},
{
"epoch": 1.8395155185465555,
"grad_norm": 5.349993705749512,
"learning_rate": 1.9396836103653883e-06,
"loss": 0.6334415078163147,
"step": 2430
},
{
"epoch": 1.8410295230885692,
"grad_norm": 1.5277384519577026,
"learning_rate": 1.938032713117985e-06,
"loss": 0.6904246211051941,
"step": 2432
},
{
"epoch": 1.8425435276305828,
"grad_norm": 2.268113136291504,
"learning_rate": 1.936381364464814e-06,
"loss": 0.27258798480033875,
"step": 2434
},
{
"epoch": 1.8440575321725965,
"grad_norm": 2.470331907272339,
"learning_rate": 1.934729566993874e-06,
"loss": 0.5896321535110474,
"step": 2436
},
{
"epoch": 1.8455715367146102,
"grad_norm": 6.7304558753967285,
"learning_rate": 1.9330773232938673e-06,
"loss": 0.6072213053703308,
"step": 2438
},
{
"epoch": 1.8470855412566238,
"grad_norm": 1.9519010782241821,
"learning_rate": 1.931424635954195e-06,
"loss": 0.8895407319068909,
"step": 2440
},
{
"epoch": 1.8485995457986375,
"grad_norm": 1.34812593460083,
"learning_rate": 1.9297715075649543e-06,
"loss": 1.1164705753326416,
"step": 2442
},
{
"epoch": 1.8501135503406512,
"grad_norm": 2.2477667331695557,
"learning_rate": 1.928117940716933e-06,
"loss": 0.34229204058647156,
"step": 2444
},
{
"epoch": 1.8516275548826646,
"grad_norm": 4.346146106719971,
"learning_rate": 1.9264639380016063e-06,
"loss": 0.7773399949073792,
"step": 2446
},
{
"epoch": 1.8531415594246783,
"grad_norm": 1.9627861976623535,
"learning_rate": 1.9248095020111323e-06,
"loss": 0.7164353728294373,
"step": 2448
},
{
"epoch": 1.8546555639666917,
"grad_norm": 1.1922169923782349,
"learning_rate": 1.923154635338348e-06,
"loss": 0.7517849206924438,
"step": 2450
},
{
"epoch": 1.8561695685087054,
"grad_norm": 3.457965850830078,
"learning_rate": 1.921499340576766e-06,
"loss": 0.5871737599372864,
"step": 2452
},
{
"epoch": 1.857683573050719,
"grad_norm": 2.036007881164551,
"learning_rate": 1.9198436203205694e-06,
"loss": 0.6375160217285156,
"step": 2454
},
{
"epoch": 1.8591975775927327,
"grad_norm": 3.6813805103302,
"learning_rate": 1.9181874771646085e-06,
"loss": 1.134217619895935,
"step": 2456
},
{
"epoch": 1.8607115821347464,
"grad_norm": 3.64957594871521,
"learning_rate": 1.916530913704395e-06,
"loss": 0.7019612193107605,
"step": 2458
},
{
"epoch": 1.86222558667676,
"grad_norm": 1.4620537757873535,
"learning_rate": 1.9148739325361015e-06,
"loss": 1.0680433511734009,
"step": 2460
},
{
"epoch": 1.8637395912187738,
"grad_norm": 3.8369688987731934,
"learning_rate": 1.913216536256553e-06,
"loss": 0.7128046751022339,
"step": 2462
},
{
"epoch": 1.8652535957607874,
"grad_norm": 5.886375427246094,
"learning_rate": 1.9115587274632274e-06,
"loss": 0.282992959022522,
"step": 2464
},
{
"epoch": 1.8667676003028009,
"grad_norm": 1.4305994510650635,
"learning_rate": 1.9099005087542467e-06,
"loss": 0.9355074763298035,
"step": 2466
},
{
"epoch": 1.8682816048448145,
"grad_norm": 5.087035179138184,
"learning_rate": 1.9082418827283766e-06,
"loss": 0.6677455902099609,
"step": 2468
},
{
"epoch": 1.8697956093868282,
"grad_norm": 1.5539398193359375,
"learning_rate": 1.9065828519850212e-06,
"loss": 0.6931514739990234,
"step": 2470
},
{
"epoch": 1.8713096139288417,
"grad_norm": 6.549385070800781,
"learning_rate": 1.9049234191242185e-06,
"loss": 0.5165091753005981,
"step": 2472
},
{
"epoch": 1.8728236184708553,
"grad_norm": 1.6973838806152344,
"learning_rate": 1.9032635867466376e-06,
"loss": 1.1013144254684448,
"step": 2474
},
{
"epoch": 1.874337623012869,
"grad_norm": 1.8635910749435425,
"learning_rate": 1.9016033574535719e-06,
"loss": 0.8190028667449951,
"step": 2476
},
{
"epoch": 1.8758516275548827,
"grad_norm": 18.24675750732422,
"learning_rate": 1.8999427338469386e-06,
"loss": 0.5483837127685547,
"step": 2478
},
{
"epoch": 1.8773656320968963,
"grad_norm": 4.019062519073486,
"learning_rate": 1.8982817185292722e-06,
"loss": 0.7344475984573364,
"step": 2480
},
{
"epoch": 1.87887963663891,
"grad_norm": 7.511033535003662,
"learning_rate": 1.896620314103721e-06,
"loss": 0.28435125946998596,
"step": 2482
},
{
"epoch": 1.8803936411809237,
"grad_norm": 5.958474636077881,
"learning_rate": 1.894958523174043e-06,
"loss": 0.6226410269737244,
"step": 2484
},
{
"epoch": 1.8819076457229371,
"grad_norm": 2.613577365875244,
"learning_rate": 1.8932963483446027e-06,
"loss": 0.26969021558761597,
"step": 2486
},
{
"epoch": 1.8834216502649508,
"grad_norm": 2.083977222442627,
"learning_rate": 1.8916337922203647e-06,
"loss": 0.6489355564117432,
"step": 2488
},
{
"epoch": 1.8849356548069645,
"grad_norm": 1.8885645866394043,
"learning_rate": 1.8899708574068928e-06,
"loss": 0.8442902565002441,
"step": 2490
},
{
"epoch": 1.886449659348978,
"grad_norm": 2.0844578742980957,
"learning_rate": 1.8883075465103431e-06,
"loss": 0.4130919277667999,
"step": 2492
},
{
"epoch": 1.8879636638909916,
"grad_norm": 5.105222702026367,
"learning_rate": 1.8866438621374628e-06,
"loss": 0.18109102547168732,
"step": 2494
},
{
"epoch": 1.8894776684330052,
"grad_norm": 12.537839889526367,
"learning_rate": 1.8849798068955823e-06,
"loss": 0.25919821858406067,
"step": 2496
},
{
"epoch": 1.890991672975019,
"grad_norm": 1.3052566051483154,
"learning_rate": 1.8833153833926146e-06,
"loss": 0.6028334498405457,
"step": 2498
},
{
"epoch": 1.8925056775170326,
"grad_norm": 1.4219228029251099,
"learning_rate": 1.8816505942370496e-06,
"loss": 0.47099339962005615,
"step": 2500
},
{
"epoch": 1.8940196820590463,
"grad_norm": 1.4984897375106812,
"learning_rate": 1.8799854420379487e-06,
"loss": 0.6924835443496704,
"step": 2502
},
{
"epoch": 1.89553368660106,
"grad_norm": 3.9320881366729736,
"learning_rate": 1.8783199294049453e-06,
"loss": 0.2342991828918457,
"step": 2504
},
{
"epoch": 1.8970476911430736,
"grad_norm": 1.5038467645645142,
"learning_rate": 1.8766540589482351e-06,
"loss": 1.052242398262024,
"step": 2506
},
{
"epoch": 1.898561695685087,
"grad_norm": 1.7633041143417358,
"learning_rate": 1.8749878332785754e-06,
"loss": 0.4072822034358978,
"step": 2508
},
{
"epoch": 1.9000757002271007,
"grad_norm": 6.334680080413818,
"learning_rate": 1.8733212550072807e-06,
"loss": 0.6254635453224182,
"step": 2510
},
{
"epoch": 1.9015897047691142,
"grad_norm": 4.005409240722656,
"learning_rate": 1.8716543267462177e-06,
"loss": 0.20865847170352936,
"step": 2512
},
{
"epoch": 1.9031037093111278,
"grad_norm": 3.80173921585083,
"learning_rate": 1.8699870511078017e-06,
"loss": 0.22845837473869324,
"step": 2514
},
{
"epoch": 1.9046177138531415,
"grad_norm": 6.341163158416748,
"learning_rate": 1.8683194307049918e-06,
"loss": 0.29452162981033325,
"step": 2516
},
{
"epoch": 1.9061317183951552,
"grad_norm": 2.0832197666168213,
"learning_rate": 1.866651468151288e-06,
"loss": 0.9766835570335388,
"step": 2518
},
{
"epoch": 1.9076457229371688,
"grad_norm": 1.7201058864593506,
"learning_rate": 1.8649831660607262e-06,
"loss": 0.7189306616783142,
"step": 2520
},
{
"epoch": 1.9091597274791825,
"grad_norm": 12.217644691467285,
"learning_rate": 1.8633145270478757e-06,
"loss": 0.18444077670574188,
"step": 2522
},
{
"epoch": 1.9106737320211962,
"grad_norm": 2.312544822692871,
"learning_rate": 1.8616455537278318e-06,
"loss": 0.6718014478683472,
"step": 2524
},
{
"epoch": 1.9121877365632098,
"grad_norm": 4.253828048706055,
"learning_rate": 1.8599762487162146e-06,
"loss": 0.21305717527866364,
"step": 2526
},
{
"epoch": 1.9137017411052233,
"grad_norm": 1.2430156469345093,
"learning_rate": 1.858306614629165e-06,
"loss": 1.0205436944961548,
"step": 2528
},
{
"epoch": 1.915215745647237,
"grad_norm": 2.997448682785034,
"learning_rate": 1.856636654083338e-06,
"loss": 0.25120025873184204,
"step": 2530
},
{
"epoch": 1.9167297501892504,
"grad_norm": 8.223458290100098,
"learning_rate": 1.8549663696959016e-06,
"loss": 0.7833942174911499,
"step": 2532
},
{
"epoch": 1.918243754731264,
"grad_norm": 3.068028450012207,
"learning_rate": 1.8532957640845296e-06,
"loss": 1.1276060342788696,
"step": 2534
},
{
"epoch": 1.9197577592732777,
"grad_norm": 8.644725799560547,
"learning_rate": 1.851624839867402e-06,
"loss": 0.21417191624641418,
"step": 2536
},
{
"epoch": 1.9212717638152914,
"grad_norm": 2.2275826930999756,
"learning_rate": 1.8499535996631946e-06,
"loss": 1.0725288391113281,
"step": 2538
},
{
"epoch": 1.922785768357305,
"grad_norm": 3.2605981826782227,
"learning_rate": 1.8482820460910817e-06,
"loss": 0.3155927062034607,
"step": 2540
},
{
"epoch": 1.9242997728993188,
"grad_norm": 1.3210935592651367,
"learning_rate": 1.8466101817707271e-06,
"loss": 0.7417844533920288,
"step": 2542
},
{
"epoch": 1.9258137774413324,
"grad_norm": 4.283865928649902,
"learning_rate": 1.844938009322281e-06,
"loss": 0.8048570156097412,
"step": 2544
},
{
"epoch": 1.927327781983346,
"grad_norm": 1.1630911827087402,
"learning_rate": 1.8432655313663771e-06,
"loss": 1.1231316328048706,
"step": 2546
},
{
"epoch": 1.9288417865253595,
"grad_norm": 1.1668163537979126,
"learning_rate": 1.8415927505241298e-06,
"loss": 1.1409462690353394,
"step": 2548
},
{
"epoch": 1.9303557910673732,
"grad_norm": 3.072653293609619,
"learning_rate": 1.8399196694171252e-06,
"loss": 0.2875867187976837,
"step": 2550
},
{
"epoch": 1.9318697956093869,
"grad_norm": 1.0892410278320312,
"learning_rate": 1.838246290667421e-06,
"loss": 0.7053228616714478,
"step": 2552
},
{
"epoch": 1.9333838001514003,
"grad_norm": 3.109567165374756,
"learning_rate": 1.8365726168975425e-06,
"loss": 0.2398093193769455,
"step": 2554
},
{
"epoch": 1.934897804693414,
"grad_norm": 3.3759360313415527,
"learning_rate": 1.8348986507304757e-06,
"loss": 0.6860714554786682,
"step": 2556
},
{
"epoch": 1.9364118092354277,
"grad_norm": 1.5024205446243286,
"learning_rate": 1.833224394789666e-06,
"loss": 1.1165058612823486,
"step": 2558
},
{
"epoch": 1.9379258137774413,
"grad_norm": 44.66451644897461,
"learning_rate": 1.8315498516990123e-06,
"loss": 0.609163761138916,
"step": 2560
},
{
"epoch": 1.939439818319455,
"grad_norm": 0.7202508449554443,
"learning_rate": 1.8298750240828638e-06,
"loss": 0.5042010545730591,
"step": 2562
},
{
"epoch": 1.9409538228614687,
"grad_norm": 1.71787428855896,
"learning_rate": 1.828199914566016e-06,
"loss": 0.5848456621170044,
"step": 2564
},
{
"epoch": 1.9424678274034823,
"grad_norm": 1.383819341659546,
"learning_rate": 1.826524525773705e-06,
"loss": 1.1039332151412964,
"step": 2566
},
{
"epoch": 1.9439818319454958,
"grad_norm": 3.2185161113739014,
"learning_rate": 1.8248488603316063e-06,
"loss": 0.8656116724014282,
"step": 2568
},
{
"epoch": 1.9454958364875095,
"grad_norm": 1.7334600687026978,
"learning_rate": 1.8231729208658271e-06,
"loss": 0.597359299659729,
"step": 2570
},
{
"epoch": 1.9470098410295231,
"grad_norm": 0.8499669432640076,
"learning_rate": 1.821496710002905e-06,
"loss": 0.66790372133255,
"step": 2572
},
{
"epoch": 1.9485238455715366,
"grad_norm": 2.160719871520996,
"learning_rate": 1.8198202303698038e-06,
"loss": 0.7572706937789917,
"step": 2574
},
{
"epoch": 1.9500378501135502,
"grad_norm": 2.4661171436309814,
"learning_rate": 1.8181434845939077e-06,
"loss": 0.7103179693222046,
"step": 2576
},
{
"epoch": 1.951551854655564,
"grad_norm": 1.7464579343795776,
"learning_rate": 1.8164664753030164e-06,
"loss": 0.7086505889892578,
"step": 2578
},
{
"epoch": 1.9530658591975776,
"grad_norm": 2.0125958919525146,
"learning_rate": 1.8147892051253455e-06,
"loss": 0.3641662001609802,
"step": 2580
},
{
"epoch": 1.9545798637395913,
"grad_norm": 2.989818811416626,
"learning_rate": 1.8131116766895169e-06,
"loss": 0.700904369354248,
"step": 2582
},
{
"epoch": 1.956093868281605,
"grad_norm": 1.8995088338851929,
"learning_rate": 1.8114338926245596e-06,
"loss": 0.6467546224594116,
"step": 2584
},
{
"epoch": 1.9576078728236186,
"grad_norm": 1.562253713607788,
"learning_rate": 1.8097558555599016e-06,
"loss": 1.0711427927017212,
"step": 2586
},
{
"epoch": 1.9591218773656323,
"grad_norm": 3.0294432640075684,
"learning_rate": 1.8080775681253673e-06,
"loss": 0.7762032747268677,
"step": 2588
},
{
"epoch": 1.9606358819076457,
"grad_norm": 4.3492751121521,
"learning_rate": 1.8063990329511749e-06,
"loss": 0.41252315044403076,
"step": 2590
},
{
"epoch": 1.9621498864496594,
"grad_norm": 1.1631908416748047,
"learning_rate": 1.8047202526679291e-06,
"loss": 0.5813542604446411,
"step": 2592
},
{
"epoch": 1.9636638909916728,
"grad_norm": 9.60074520111084,
"learning_rate": 1.8030412299066201e-06,
"loss": 0.2214062511920929,
"step": 2594
},
{
"epoch": 1.9651778955336865,
"grad_norm": 2.970233201980591,
"learning_rate": 1.8013619672986173e-06,
"loss": 0.24248868227005005,
"step": 2596
},
{
"epoch": 1.9666919000757002,
"grad_norm": 2.1815083026885986,
"learning_rate": 1.799682467475667e-06,
"loss": 0.28807175159454346,
"step": 2598
},
{
"epoch": 1.9682059046177138,
"grad_norm": 6.088417053222656,
"learning_rate": 1.798002733069886e-06,
"loss": 0.3555501699447632,
"step": 2600
},
{
"epoch": 1.9697199091597275,
"grad_norm": 3.4696614742279053,
"learning_rate": 1.796322766713759e-06,
"loss": 0.2768677771091461,
"step": 2602
},
{
"epoch": 1.9712339137017412,
"grad_norm": 4.333212852478027,
"learning_rate": 1.7946425710401357e-06,
"loss": 0.7167282700538635,
"step": 2604
},
{
"epoch": 1.9727479182437548,
"grad_norm": 2.5988035202026367,
"learning_rate": 1.7929621486822223e-06,
"loss": 0.6456173062324524,
"step": 2606
},
{
"epoch": 1.9742619227857685,
"grad_norm": 3.324312448501587,
"learning_rate": 1.7912815022735837e-06,
"loss": 0.1809239685535431,
"step": 2608
},
{
"epoch": 1.975775927327782,
"grad_norm": 1.678682804107666,
"learning_rate": 1.789600634448133e-06,
"loss": 1.049579381942749,
"step": 2610
},
{
"epoch": 1.9772899318697956,
"grad_norm": 2.530160665512085,
"learning_rate": 1.7879195478401319e-06,
"loss": 0.7081025838851929,
"step": 2612
},
{
"epoch": 1.978803936411809,
"grad_norm": 2.570563793182373,
"learning_rate": 1.7862382450841844e-06,
"loss": 0.6838253140449524,
"step": 2614
},
{
"epoch": 1.9803179409538227,
"grad_norm": 6.012343883514404,
"learning_rate": 1.784556728815234e-06,
"loss": 0.6574077010154724,
"step": 2616
},
{
"epoch": 1.9818319454958364,
"grad_norm": 8.046130180358887,
"learning_rate": 1.7828750016685576e-06,
"loss": 0.17780661582946777,
"step": 2618
},
{
"epoch": 1.98334595003785,
"grad_norm": 3.787870168685913,
"learning_rate": 1.7811930662797638e-06,
"loss": 0.7062051892280579,
"step": 2620
},
{
"epoch": 1.9848599545798638,
"grad_norm": 1.1279329061508179,
"learning_rate": 1.7795109252847867e-06,
"loss": 0.7671316862106323,
"step": 2622
},
{
"epoch": 1.9863739591218774,
"grad_norm": 4.606888294219971,
"learning_rate": 1.7778285813198826e-06,
"loss": 0.31346026062965393,
"step": 2624
},
{
"epoch": 1.987887963663891,
"grad_norm": 2.6706228256225586,
"learning_rate": 1.7761460370216267e-06,
"loss": 0.5862433314323425,
"step": 2626
},
{
"epoch": 1.9894019682059048,
"grad_norm": 2.2766308784484863,
"learning_rate": 1.7744632950269075e-06,
"loss": 1.0763511657714844,
"step": 2628
},
{
"epoch": 1.9909159727479182,
"grad_norm": 5.0256667137146,
"learning_rate": 1.772780357972924e-06,
"loss": 0.13953058421611786,
"step": 2630
},
{
"epoch": 1.9924299772899319,
"grad_norm": 5.517915725708008,
"learning_rate": 1.7710972284971793e-06,
"loss": 1.0178477764129639,
"step": 2632
},
{
"epoch": 1.9939439818319455,
"grad_norm": 2.4863970279693604,
"learning_rate": 1.7694139092374802e-06,
"loss": 0.6464669108390808,
"step": 2634
},
{
"epoch": 1.995457986373959,
"grad_norm": 4.220505714416504,
"learning_rate": 1.7677304028319295e-06,
"loss": 0.8648689985275269,
"step": 2636
},
{
"epoch": 1.9969719909159727,
"grad_norm": 2.9062840938568115,
"learning_rate": 1.7660467119189236e-06,
"loss": 0.16142083704471588,
"step": 2638
},
{
"epoch": 1.9984859954579863,
"grad_norm": 18.753829956054688,
"learning_rate": 1.7643628391371484e-06,
"loss": 0.3382264971733093,
"step": 2640
},
{
"epoch": 2.0,
"grad_norm": 2.7218689918518066,
"learning_rate": 1.762678787125574e-06,
"loss": 0.719879150390625,
"step": 2642
},
{
"epoch": 2.0015140045420137,
"grad_norm": 0.6632734537124634,
"learning_rate": 1.7609945585234533e-06,
"loss": 0.14925611019134521,
"step": 2644
},
{
"epoch": 2.0030280090840273,
"grad_norm": 1.8668553829193115,
"learning_rate": 1.7593101559703132e-06,
"loss": 0.6265012621879578,
"step": 2646
},
{
"epoch": 2.004542013626041,
"grad_norm": 1.4000589847564697,
"learning_rate": 1.7576255821059549e-06,
"loss": 0.5532234311103821,
"step": 2648
},
{
"epoch": 2.0060560181680547,
"grad_norm": 1.3937102556228638,
"learning_rate": 1.7559408395704483e-06,
"loss": 0.12811285257339478,
"step": 2650
},
{
"epoch": 2.0075700227100683,
"grad_norm": 2.235651969909668,
"learning_rate": 1.7542559310041272e-06,
"loss": 0.5081092119216919,
"step": 2652
},
{
"epoch": 2.0090840272520816,
"grad_norm": 1.5715231895446777,
"learning_rate": 1.7525708590475855e-06,
"loss": 0.12974363565444946,
"step": 2654
},
{
"epoch": 2.0105980317940952,
"grad_norm": 3.355414867401123,
"learning_rate": 1.7508856263416728e-06,
"loss": 0.17922165989875793,
"step": 2656
},
{
"epoch": 2.012112036336109,
"grad_norm": 2.6334707736968994,
"learning_rate": 1.7492002355274917e-06,
"loss": 0.26148444414138794,
"step": 2658
},
{
"epoch": 2.0136260408781226,
"grad_norm": 1.8044371604919434,
"learning_rate": 1.7475146892463911e-06,
"loss": 0.6592626571655273,
"step": 2660
},
{
"epoch": 2.0151400454201362,
"grad_norm": 0.6419950127601624,
"learning_rate": 1.7458289901399652e-06,
"loss": 0.5145966410636902,
"step": 2662
},
{
"epoch": 2.01665404996215,
"grad_norm": 3.272899627685547,
"learning_rate": 1.7441431408500469e-06,
"loss": 0.577723503112793,
"step": 2664
},
{
"epoch": 2.0181680545041636,
"grad_norm": 2.162247896194458,
"learning_rate": 1.7424571440187036e-06,
"loss": 0.9184585809707642,
"step": 2666
},
{
"epoch": 2.0196820590461773,
"grad_norm": 4.363552570343018,
"learning_rate": 1.7407710022882353e-06,
"loss": 0.7015078663825989,
"step": 2668
},
{
"epoch": 2.021196063588191,
"grad_norm": 0.746873676776886,
"learning_rate": 1.7390847183011696e-06,
"loss": 0.14988543093204498,
"step": 2670
},
{
"epoch": 2.0227100681302046,
"grad_norm": 1.4385411739349365,
"learning_rate": 1.7373982947002545e-06,
"loss": 0.5507326126098633,
"step": 2672
},
{
"epoch": 2.024224072672218,
"grad_norm": 5.640541076660156,
"learning_rate": 1.7357117341284586e-06,
"loss": 0.5066704750061035,
"step": 2674
},
{
"epoch": 2.0257380772142315,
"grad_norm": 1.6015201807022095,
"learning_rate": 1.7340250392289654e-06,
"loss": 0.58171147108078,
"step": 2676
},
{
"epoch": 2.027252081756245,
"grad_norm": 1.6645236015319824,
"learning_rate": 1.7323382126451683e-06,
"loss": 1.0548933744430542,
"step": 2678
},
{
"epoch": 2.028766086298259,
"grad_norm": 1.4560426473617554,
"learning_rate": 1.7306512570206675e-06,
"loss": 1.017156720161438,
"step": 2680
},
{
"epoch": 2.0302800908402725,
"grad_norm": 2.0511622428894043,
"learning_rate": 1.7289641749992642e-06,
"loss": 0.5105241537094116,
"step": 2682
},
{
"epoch": 2.031794095382286,
"grad_norm": 4.8046464920043945,
"learning_rate": 1.7272769692249596e-06,
"loss": 0.5566238164901733,
"step": 2684
},
{
"epoch": 2.0333080999243,
"grad_norm": 18.07652473449707,
"learning_rate": 1.7255896423419474e-06,
"loss": 0.2835147976875305,
"step": 2686
},
{
"epoch": 2.0348221044663135,
"grad_norm": 2.6613402366638184,
"learning_rate": 1.7239021969946115e-06,
"loss": 0.5674735307693481,
"step": 2688
},
{
"epoch": 2.036336109008327,
"grad_norm": 1.3865885734558105,
"learning_rate": 1.7222146358275214e-06,
"loss": 0.15232010185718536,
"step": 2690
},
{
"epoch": 2.037850113550341,
"grad_norm": 1.7241010665893555,
"learning_rate": 1.720526961485429e-06,
"loss": 0.8321764469146729,
"step": 2692
},
{
"epoch": 2.039364118092354,
"grad_norm": 10.220908164978027,
"learning_rate": 1.7188391766132618e-06,
"loss": 0.22783613204956055,
"step": 2694
},
{
"epoch": 2.0408781226343677,
"grad_norm": 1.8045783042907715,
"learning_rate": 1.7171512838561219e-06,
"loss": 0.9568790197372437,
"step": 2696
},
{
"epoch": 2.0423921271763814,
"grad_norm": 1.4294806718826294,
"learning_rate": 1.7154632858592804e-06,
"loss": 0.6187859177589417,
"step": 2698
},
{
"epoch": 2.043906131718395,
"grad_norm": 2.846625804901123,
"learning_rate": 1.7137751852681728e-06,
"loss": 0.5443164110183716,
"step": 2700
},
{
"epoch": 2.0454201362604087,
"grad_norm": 4.185090065002441,
"learning_rate": 1.7120869847283955e-06,
"loss": 0.11699831485748291,
"step": 2702
},
{
"epoch": 2.0469341408024224,
"grad_norm": 5.7305755615234375,
"learning_rate": 1.7103986868857016e-06,
"loss": 0.7156167030334473,
"step": 2704
},
{
"epoch": 2.048448145344436,
"grad_norm": 7.5829267501831055,
"learning_rate": 1.7087102943859973e-06,
"loss": 0.5357162356376648,
"step": 2706
},
{
"epoch": 2.0499621498864498,
"grad_norm": 1.4636871814727783,
"learning_rate": 1.7070218098753363e-06,
"loss": 0.4982175827026367,
"step": 2708
},
{
"epoch": 2.0514761544284634,
"grad_norm": 4.640434741973877,
"learning_rate": 1.7053332359999166e-06,
"loss": 0.6078394055366516,
"step": 2710
},
{
"epoch": 2.052990158970477,
"grad_norm": 1.874154806137085,
"learning_rate": 1.7036445754060766e-06,
"loss": 1.1511926651000977,
"step": 2712
},
{
"epoch": 2.0545041635124903,
"grad_norm": 1.7623599767684937,
"learning_rate": 1.7019558307402901e-06,
"loss": 1.0208662748336792,
"step": 2714
},
{
"epoch": 2.056018168054504,
"grad_norm": 1.8835748434066772,
"learning_rate": 1.7002670046491641e-06,
"loss": 0.5125632286071777,
"step": 2716
},
{
"epoch": 2.0575321725965177,
"grad_norm": 2.5960309505462646,
"learning_rate": 1.6985780997794308e-06,
"loss": 0.35421526432037354,
"step": 2718
},
{
"epoch": 2.0590461771385313,
"grad_norm": 0.24600274860858917,
"learning_rate": 1.696889118777948e-06,
"loss": 0.5041108727455139,
"step": 2720
},
{
"epoch": 2.060560181680545,
"grad_norm": 3.7518346309661865,
"learning_rate": 1.6952000642916918e-06,
"loss": 0.4405469298362732,
"step": 2722
},
{
"epoch": 2.0620741862225587,
"grad_norm": 3.009103536605835,
"learning_rate": 1.6935109389677534e-06,
"loss": 0.5766614079475403,
"step": 2724
},
{
"epoch": 2.0635881907645723,
"grad_norm": 3.843696355819702,
"learning_rate": 1.6918217454533359e-06,
"loss": 0.260686457157135,
"step": 2726
},
{
"epoch": 2.065102195306586,
"grad_norm": 5.3915605545043945,
"learning_rate": 1.6901324863957482e-06,
"loss": 0.25513988733291626,
"step": 2728
},
{
"epoch": 2.0666161998485997,
"grad_norm": 0.4250639081001282,
"learning_rate": 1.6884431644424022e-06,
"loss": 0.16661809384822845,
"step": 2730
},
{
"epoch": 2.0681302043906133,
"grad_norm": 2.3801870346069336,
"learning_rate": 1.6867537822408093e-06,
"loss": 0.5339866876602173,
"step": 2732
},
{
"epoch": 2.069644208932627,
"grad_norm": 2.5381367206573486,
"learning_rate": 1.6850643424385733e-06,
"loss": 0.10374370217323303,
"step": 2734
},
{
"epoch": 2.0711582134746402,
"grad_norm": 1.1726455688476562,
"learning_rate": 1.6833748476833906e-06,
"loss": 0.548637866973877,
"step": 2736
},
{
"epoch": 2.072672218016654,
"grad_norm": 2.1821258068084717,
"learning_rate": 1.6816853006230427e-06,
"loss": 0.8682297468185425,
"step": 2738
},
{
"epoch": 2.0741862225586676,
"grad_norm": 1.3762861490249634,
"learning_rate": 1.6799957039053924e-06,
"loss": 0.619299054145813,
"step": 2740
},
{
"epoch": 2.0757002271006812,
"grad_norm": 4.402048110961914,
"learning_rate": 1.6783060601783816e-06,
"loss": 0.6997363567352295,
"step": 2742
},
{
"epoch": 2.077214231642695,
"grad_norm": 1.7996996641159058,
"learning_rate": 1.6766163720900242e-06,
"loss": 0.43396300077438354,
"step": 2744
},
{
"epoch": 2.0787282361847086,
"grad_norm": 1.5604248046875,
"learning_rate": 1.674926642288406e-06,
"loss": 0.15332743525505066,
"step": 2746
},
{
"epoch": 2.0802422407267223,
"grad_norm": 2.172469139099121,
"learning_rate": 1.6732368734216756e-06,
"loss": 1.0345743894577026,
"step": 2748
},
{
"epoch": 2.081756245268736,
"grad_norm": 3.688363790512085,
"learning_rate": 1.6715470681380446e-06,
"loss": 0.63254714012146,
"step": 2750
},
{
"epoch": 2.0832702498107496,
"grad_norm": 1.5668286085128784,
"learning_rate": 1.6698572290857814e-06,
"loss": 0.5776516199111938,
"step": 2752
},
{
"epoch": 2.0847842543527633,
"grad_norm": 3.8889572620391846,
"learning_rate": 1.6681673589132063e-06,
"loss": 0.4518263638019562,
"step": 2754
},
{
"epoch": 2.0862982588947765,
"grad_norm": 4.702708721160889,
"learning_rate": 1.6664774602686903e-06,
"loss": 0.49554580450057983,
"step": 2756
},
{
"epoch": 2.08781226343679,
"grad_norm": 3.964775323867798,
"learning_rate": 1.6647875358006466e-06,
"loss": 0.5103597044944763,
"step": 2758
},
{
"epoch": 2.089326267978804,
"grad_norm": 3.2854743003845215,
"learning_rate": 1.663097588157531e-06,
"loss": 0.06786801666021347,
"step": 2760
},
{
"epoch": 2.0908402725208175,
"grad_norm": 4.798826694488525,
"learning_rate": 1.661407619987834e-06,
"loss": 0.5872377157211304,
"step": 2762
},
{
"epoch": 2.092354277062831,
"grad_norm": 2.028712272644043,
"learning_rate": 1.6597176339400792e-06,
"loss": 0.6830023527145386,
"step": 2764
},
{
"epoch": 2.093868281604845,
"grad_norm": 2.520421028137207,
"learning_rate": 1.6580276326628184e-06,
"loss": 0.22064897418022156,
"step": 2766
},
{
"epoch": 2.0953822861468585,
"grad_norm": 2.916041851043701,
"learning_rate": 1.6563376188046265e-06,
"loss": 0.1994413435459137,
"step": 2768
},
{
"epoch": 2.096896290688872,
"grad_norm": 3.3804969787597656,
"learning_rate": 1.6546475950140986e-06,
"loss": 0.15971145033836365,
"step": 2770
},
{
"epoch": 2.098410295230886,
"grad_norm": 3.459014654159546,
"learning_rate": 1.6529575639398453e-06,
"loss": 0.2711777985095978,
"step": 2772
},
{
"epoch": 2.0999242997728995,
"grad_norm": 7.535019874572754,
"learning_rate": 1.6512675282304884e-06,
"loss": 0.30359020829200745,
"step": 2774
},
{
"epoch": 2.1014383043149127,
"grad_norm": 2.6062686443328857,
"learning_rate": 1.6495774905346575e-06,
"loss": 0.08550383895635605,
"step": 2776
},
{
"epoch": 2.1029523088569264,
"grad_norm": 2.4432120323181152,
"learning_rate": 1.6478874535009847e-06,
"loss": 1.0165009498596191,
"step": 2778
},
{
"epoch": 2.10446631339894,
"grad_norm": 1.2791420221328735,
"learning_rate": 1.6461974197781015e-06,
"loss": 0.5231562256813049,
"step": 2780
},
{
"epoch": 2.1059803179409537,
"grad_norm": 1.6797175407409668,
"learning_rate": 1.6445073920146336e-06,
"loss": 0.5763256549835205,
"step": 2782
},
{
"epoch": 2.1074943224829674,
"grad_norm": 1.43915855884552,
"learning_rate": 1.6428173728591981e-06,
"loss": 0.9489549994468689,
"step": 2784
},
{
"epoch": 2.109008327024981,
"grad_norm": 3.021160125732422,
"learning_rate": 1.6411273649603988e-06,
"loss": 0.9699093103408813,
"step": 2786
},
{
"epoch": 2.1105223315669948,
"grad_norm": 1.4524022340774536,
"learning_rate": 1.6394373709668207e-06,
"loss": 0.5449516177177429,
"step": 2788
},
{
"epoch": 2.1120363361090084,
"grad_norm": 1.5240285396575928,
"learning_rate": 1.6377473935270272e-06,
"loss": 0.5013771057128906,
"step": 2790
},
{
"epoch": 2.113550340651022,
"grad_norm": 3.620814561843872,
"learning_rate": 1.6360574352895573e-06,
"loss": 0.9098644256591797,
"step": 2792
},
{
"epoch": 2.1150643451930358,
"grad_norm": 1.660154938697815,
"learning_rate": 1.6343674989029185e-06,
"loss": 0.7111346125602722,
"step": 2794
},
{
"epoch": 2.116578349735049,
"grad_norm": 1.55197012424469,
"learning_rate": 1.632677587015584e-06,
"loss": 0.9725939631462097,
"step": 2796
},
{
"epoch": 2.1180923542770627,
"grad_norm": 2.468062400817871,
"learning_rate": 1.6309877022759894e-06,
"loss": 1.0302742719650269,
"step": 2798
},
{
"epoch": 2.1196063588190763,
"grad_norm": 4.183055400848389,
"learning_rate": 1.6292978473325269e-06,
"loss": 0.3166213035583496,
"step": 2800
},
{
"epoch": 2.12112036336109,
"grad_norm": 1.5126471519470215,
"learning_rate": 1.627608024833543e-06,
"loss": 0.48440665006637573,
"step": 2802
},
{
"epoch": 2.1226343679031037,
"grad_norm": 1.1503214836120605,
"learning_rate": 1.6259182374273325e-06,
"loss": 0.5727757215499878,
"step": 2804
},
{
"epoch": 2.1241483724451173,
"grad_norm": 6.117621898651123,
"learning_rate": 1.6242284877621352e-06,
"loss": 0.1801280677318573,
"step": 2806
},
{
"epoch": 2.125662376987131,
"grad_norm": 2.113598585128784,
"learning_rate": 1.6225387784861332e-06,
"loss": 0.8976154327392578,
"step": 2808
},
{
"epoch": 2.1271763815291447,
"grad_norm": 2.00311279296875,
"learning_rate": 1.6208491122474423e-06,
"loss": 0.14283889532089233,
"step": 2810
},
{
"epoch": 2.1286903860711583,
"grad_norm": 16.498226165771484,
"learning_rate": 1.6191594916941145e-06,
"loss": 0.6590249538421631,
"step": 2812
},
{
"epoch": 2.130204390613172,
"grad_norm": 2.0265419483184814,
"learning_rate": 1.6174699194741276e-06,
"loss": 0.6612314581871033,
"step": 2814
},
{
"epoch": 2.1317183951551852,
"grad_norm": 3.4283297061920166,
"learning_rate": 1.6157803982353844e-06,
"loss": 0.6018260717391968,
"step": 2816
},
{
"epoch": 2.133232399697199,
"grad_norm": 1.7489575147628784,
"learning_rate": 1.6140909306257075e-06,
"loss": 0.47798389196395874,
"step": 2818
},
{
"epoch": 2.1347464042392126,
"grad_norm": 1.7350597381591797,
"learning_rate": 1.6124015192928368e-06,
"loss": 0.050837621092796326,
"step": 2820
},
{
"epoch": 2.1362604087812262,
"grad_norm": 3.2042109966278076,
"learning_rate": 1.6107121668844229e-06,
"loss": 0.9809722900390625,
"step": 2822
},
{
"epoch": 2.13777441332324,
"grad_norm": 1.8703489303588867,
"learning_rate": 1.6090228760480233e-06,
"loss": 0.33297398686408997,
"step": 2824
},
{
"epoch": 2.1392884178652536,
"grad_norm": 2.5503528118133545,
"learning_rate": 1.6073336494311e-06,
"loss": 0.5941415429115295,
"step": 2826
},
{
"epoch": 2.1408024224072673,
"grad_norm": 2.0041916370391846,
"learning_rate": 1.605644489681015e-06,
"loss": 0.6204849481582642,
"step": 2828
},
{
"epoch": 2.142316426949281,
"grad_norm": 9.915755271911621,
"learning_rate": 1.6039553994450242e-06,
"loss": 0.3267604410648346,
"step": 2830
},
{
"epoch": 2.1438304314912946,
"grad_norm": 2.206566572189331,
"learning_rate": 1.602266381370275e-06,
"loss": 0.5704283714294434,
"step": 2832
},
{
"epoch": 2.1453444360333083,
"grad_norm": 2.92156982421875,
"learning_rate": 1.6005774381038027e-06,
"loss": 0.5793426036834717,
"step": 2834
},
{
"epoch": 2.146858440575322,
"grad_norm": 3.446460247039795,
"learning_rate": 1.5988885722925236e-06,
"loss": 0.11840743571519852,
"step": 2836
},
{
"epoch": 2.148372445117335,
"grad_norm": 2.32879900932312,
"learning_rate": 1.5971997865832336e-06,
"loss": 0.15414756536483765,
"step": 2838
},
{
"epoch": 2.149886449659349,
"grad_norm": 3.7858409881591797,
"learning_rate": 1.5955110836226026e-06,
"loss": 0.6105750799179077,
"step": 2840
},
{
"epoch": 2.1514004542013625,
"grad_norm": 2.2815473079681396,
"learning_rate": 1.593822466057172e-06,
"loss": 0.11312630772590637,
"step": 2842
},
{
"epoch": 2.152914458743376,
"grad_norm": 9.054616928100586,
"learning_rate": 1.592133936533348e-06,
"loss": 0.3543716073036194,
"step": 2844
},
{
"epoch": 2.15442846328539,
"grad_norm": 1.658771276473999,
"learning_rate": 1.5904454976973997e-06,
"loss": 0.10064493864774704,
"step": 2846
},
{
"epoch": 2.1559424678274035,
"grad_norm": 2.0759246349334717,
"learning_rate": 1.5887571521954526e-06,
"loss": 0.97999507188797,
"step": 2848
},
{
"epoch": 2.157456472369417,
"grad_norm": 2.34515118598938,
"learning_rate": 1.5870689026734887e-06,
"loss": 0.9874440431594849,
"step": 2850
},
{
"epoch": 2.158970476911431,
"grad_norm": 2.011859655380249,
"learning_rate": 1.5853807517773366e-06,
"loss": 0.8331037759780884,
"step": 2852
},
{
"epoch": 2.1604844814534445,
"grad_norm": 2.438368558883667,
"learning_rate": 1.5836927021526724e-06,
"loss": 0.20251956582069397,
"step": 2854
},
{
"epoch": 2.161998485995458,
"grad_norm": 2.275050163269043,
"learning_rate": 1.5820047564450122e-06,
"loss": 1.0513229370117188,
"step": 2856
},
{
"epoch": 2.1635124905374714,
"grad_norm": 1.618593692779541,
"learning_rate": 1.5803169172997105e-06,
"loss": 0.6685404777526855,
"step": 2858
},
{
"epoch": 2.165026495079485,
"grad_norm": 3.4260945320129395,
"learning_rate": 1.578629187361954e-06,
"loss": 0.1660946160554886,
"step": 2860
},
{
"epoch": 2.1665404996214987,
"grad_norm": 1.603978157043457,
"learning_rate": 1.576941569276757e-06,
"loss": 0.15563638508319855,
"step": 2862
},
{
"epoch": 2.1680545041635124,
"grad_norm": 3.278381586074829,
"learning_rate": 1.5752540656889617e-06,
"loss": 0.17102839052677155,
"step": 2864
},
{
"epoch": 2.169568508705526,
"grad_norm": 2.840925455093384,
"learning_rate": 1.5735666792432283e-06,
"loss": 0.14147916436195374,
"step": 2866
},
{
"epoch": 2.1710825132475398,
"grad_norm": 2.1807408332824707,
"learning_rate": 1.5718794125840328e-06,
"loss": 0.5866944193840027,
"step": 2868
},
{
"epoch": 2.1725965177895534,
"grad_norm": 2.3917977809906006,
"learning_rate": 1.570192268355667e-06,
"loss": 0.2330070585012436,
"step": 2870
},
{
"epoch": 2.174110522331567,
"grad_norm": 2.4362566471099854,
"learning_rate": 1.5685052492022274e-06,
"loss": 0.24644072353839874,
"step": 2872
},
{
"epoch": 2.1756245268735808,
"grad_norm": 2.0635030269622803,
"learning_rate": 1.5668183577676157e-06,
"loss": 0.53923499584198,
"step": 2874
},
{
"epoch": 2.1771385314155944,
"grad_norm": 8.901561737060547,
"learning_rate": 1.5651315966955332e-06,
"loss": 0.662903904914856,
"step": 2876
},
{
"epoch": 2.1786525359576077,
"grad_norm": 2.5459957122802734,
"learning_rate": 1.5634449686294778e-06,
"loss": 0.222152441740036,
"step": 2878
},
{
"epoch": 2.1801665404996213,
"grad_norm": 1.7599998712539673,
"learning_rate": 1.561758476212738e-06,
"loss": 0.657718300819397,
"step": 2880
},
{
"epoch": 2.181680545041635,
"grad_norm": 5.629086971282959,
"learning_rate": 1.56007212208839e-06,
"loss": 0.07736900448799133,
"step": 2882
},
{
"epoch": 2.1831945495836487,
"grad_norm": 3.16239070892334,
"learning_rate": 1.5583859088992927e-06,
"loss": 0.4388081431388855,
"step": 2884
},
{
"epoch": 2.1847085541256623,
"grad_norm": 1.404675006866455,
"learning_rate": 1.5566998392880854e-06,
"loss": 0.8538283109664917,
"step": 2886
},
{
"epoch": 2.186222558667676,
"grad_norm": 2.138237714767456,
"learning_rate": 1.5550139158971817e-06,
"loss": 0.2884243428707123,
"step": 2888
},
{
"epoch": 2.1877365632096897,
"grad_norm": 0.9016575813293457,
"learning_rate": 1.553328141368765e-06,
"loss": 0.5784789323806763,
"step": 2890
},
{
"epoch": 2.1892505677517033,
"grad_norm": 2.4022958278656006,
"learning_rate": 1.551642518344788e-06,
"loss": 0.8029235601425171,
"step": 2892
},
{
"epoch": 2.190764572293717,
"grad_norm": 28.471567153930664,
"learning_rate": 1.5499570494669635e-06,
"loss": 0.7160929441452026,
"step": 2894
},
{
"epoch": 2.1922785768357307,
"grad_norm": 37.3333740234375,
"learning_rate": 1.548271737376763e-06,
"loss": 0.49402734637260437,
"step": 2896
},
{
"epoch": 2.1937925813777444,
"grad_norm": 3.0640456676483154,
"learning_rate": 1.5465865847154133e-06,
"loss": 0.5062578320503235,
"step": 2898
},
{
"epoch": 2.1953065859197576,
"grad_norm": 3.4620871543884277,
"learning_rate": 1.5449015941238916e-06,
"loss": 0.10250119119882584,
"step": 2900
},
{
"epoch": 2.1968205904617712,
"grad_norm": 2.5394251346588135,
"learning_rate": 1.5432167682429199e-06,
"loss": 0.15818654000759125,
"step": 2902
},
{
"epoch": 2.198334595003785,
"grad_norm": 0.4949427843093872,
"learning_rate": 1.541532109712962e-06,
"loss": 0.6417489051818848,
"step": 2904
},
{
"epoch": 2.1998485995457986,
"grad_norm": 1.840445637702942,
"learning_rate": 1.5398476211742212e-06,
"loss": 0.6399612426757812,
"step": 2906
},
{
"epoch": 2.2013626040878123,
"grad_norm": 1.615837574005127,
"learning_rate": 1.5381633052666323e-06,
"loss": 0.06168773025274277,
"step": 2908
},
{
"epoch": 2.202876608629826,
"grad_norm": 1.5384738445281982,
"learning_rate": 1.5364791646298612e-06,
"loss": 0.9079161882400513,
"step": 2910
},
{
"epoch": 2.2043906131718396,
"grad_norm": 4.141279697418213,
"learning_rate": 1.5347952019032969e-06,
"loss": 0.12153710424900055,
"step": 2912
},
{
"epoch": 2.2059046177138533,
"grad_norm": 2.1120827198028564,
"learning_rate": 1.533111419726053e-06,
"loss": 0.5576128959655762,
"step": 2914
},
{
"epoch": 2.207418622255867,
"grad_norm": 2.8576767444610596,
"learning_rate": 1.5314278207369572e-06,
"loss": 0.3549292981624603,
"step": 2916
},
{
"epoch": 2.20893262679788,
"grad_norm": 2.535698413848877,
"learning_rate": 1.5297444075745511e-06,
"loss": 0.6953564882278442,
"step": 2918
},
{
"epoch": 2.210446631339894,
"grad_norm": 4.89951229095459,
"learning_rate": 1.5280611828770842e-06,
"loss": 0.5466469526290894,
"step": 2920
},
{
"epoch": 2.2119606358819075,
"grad_norm": 1.158060073852539,
"learning_rate": 1.5263781492825134e-06,
"loss": 0.5347949862480164,
"step": 2922
},
{
"epoch": 2.213474640423921,
"grad_norm": 3.739546775817871,
"learning_rate": 1.524695309428493e-06,
"loss": 0.6225649118423462,
"step": 2924
},
{
"epoch": 2.214988644965935,
"grad_norm": 1.4427951574325562,
"learning_rate": 1.5230126659523748e-06,
"loss": 0.4554247558116913,
"step": 2926
},
{
"epoch": 2.2165026495079485,
"grad_norm": 2.3546080589294434,
"learning_rate": 1.5213302214912033e-06,
"loss": 0.32856038212776184,
"step": 2928
},
{
"epoch": 2.218016654049962,
"grad_norm": 3.2373764514923096,
"learning_rate": 1.5196479786817105e-06,
"loss": 0.57108074426651,
"step": 2930
},
{
"epoch": 2.219530658591976,
"grad_norm": 1.5101537704467773,
"learning_rate": 1.517965940160313e-06,
"loss": 0.4823313355445862,
"step": 2932
},
{
"epoch": 2.2210446631339895,
"grad_norm": 2.8875467777252197,
"learning_rate": 1.5162841085631062e-06,
"loss": 0.12577395141124725,
"step": 2934
},
{
"epoch": 2.222558667676003,
"grad_norm": 2.704087734222412,
"learning_rate": 1.5146024865258626e-06,
"loss": 1.092346429824829,
"step": 2936
},
{
"epoch": 2.224072672218017,
"grad_norm": 1.7303626537322998,
"learning_rate": 1.512921076684025e-06,
"loss": 0.5065004229545593,
"step": 2938
},
{
"epoch": 2.22558667676003,
"grad_norm": 6.826353549957275,
"learning_rate": 1.5112398816727044e-06,
"loss": 0.6223349571228027,
"step": 2940
},
{
"epoch": 2.2271006813020437,
"grad_norm": 1.7305920124053955,
"learning_rate": 1.5095589041266737e-06,
"loss": 0.5957088470458984,
"step": 2942
},
{
"epoch": 2.2286146858440574,
"grad_norm": 2.905937433242798,
"learning_rate": 1.5078781466803683e-06,
"loss": 0.29074761271476746,
"step": 2944
},
{
"epoch": 2.230128690386071,
"grad_norm": 1.5235497951507568,
"learning_rate": 1.5061976119678749e-06,
"loss": 0.5921708345413208,
"step": 2946
},
{
"epoch": 2.2316426949280848,
"grad_norm": 1.6646530628204346,
"learning_rate": 1.5045173026229326e-06,
"loss": 0.4309976398944855,
"step": 2948
},
{
"epoch": 2.2331566994700984,
"grad_norm": 2.0782201290130615,
"learning_rate": 1.502837221278929e-06,
"loss": 0.243109330534935,
"step": 2950
},
{
"epoch": 2.234670704012112,
"grad_norm": 3.6397688388824463,
"learning_rate": 1.5011573705688922e-06,
"loss": 0.8285622596740723,
"step": 2952
},
{
"epoch": 2.2361847085541258,
"grad_norm": 2.441128969192505,
"learning_rate": 1.4994777531254882e-06,
"loss": 0.430575430393219,
"step": 2954
},
{
"epoch": 2.2376987130961394,
"grad_norm": 1.0673562288284302,
"learning_rate": 1.49779837158102e-06,
"loss": 0.040843378752470016,
"step": 2956
},
{
"epoch": 2.239212717638153,
"grad_norm": 1.7394381761550903,
"learning_rate": 1.4961192285674194e-06,
"loss": 0.6127630472183228,
"step": 2958
},
{
"epoch": 2.2407267221801668,
"grad_norm": 1.1732391119003296,
"learning_rate": 1.494440326716245e-06,
"loss": 0.5040069222450256,
"step": 2960
},
{
"epoch": 2.24224072672218,
"grad_norm": 3.5119881629943848,
"learning_rate": 1.4927616686586755e-06,
"loss": 0.4314347505569458,
"step": 2962
},
{
"epoch": 2.2437547312641937,
"grad_norm": 21.22010612487793,
"learning_rate": 1.4910832570255105e-06,
"loss": 0.15195772051811218,
"step": 2964
},
{
"epoch": 2.2452687358062073,
"grad_norm": 1.7871365547180176,
"learning_rate": 1.489405094447162e-06,
"loss": 0.07986627519130707,
"step": 2966
},
{
"epoch": 2.246782740348221,
"grad_norm": 1.7326053380966187,
"learning_rate": 1.4877271835536508e-06,
"loss": 0.163385808467865,
"step": 2968
},
{
"epoch": 2.2482967448902347,
"grad_norm": 4.657560348510742,
"learning_rate": 1.486049526974604e-06,
"loss": 0.07589483261108398,
"step": 2970
},
{
"epoch": 2.2498107494322483,
"grad_norm": 2.992844343185425,
"learning_rate": 1.4843721273392512e-06,
"loss": 0.18186970055103302,
"step": 2972
},
{
"epoch": 2.251324753974262,
"grad_norm": 1.3081936836242676,
"learning_rate": 1.4826949872764181e-06,
"loss": 0.4326886534690857,
"step": 2974
},
{
"epoch": 2.2528387585162757,
"grad_norm": 4.688754081726074,
"learning_rate": 1.4810181094145231e-06,
"loss": 0.5936760306358337,
"step": 2976
},
{
"epoch": 2.2543527630582894,
"grad_norm": 2.676875591278076,
"learning_rate": 1.4793414963815745e-06,
"loss": 0.20641954243183136,
"step": 2978
},
{
"epoch": 2.2558667676003026,
"grad_norm": 2.0084316730499268,
"learning_rate": 1.4776651508051667e-06,
"loss": 0.47862863540649414,
"step": 2980
},
{
"epoch": 2.2573807721423162,
"grad_norm": 1.0666364431381226,
"learning_rate": 1.4759890753124724e-06,
"loss": 0.7472076416015625,
"step": 2982
},
{
"epoch": 2.25889477668433,
"grad_norm": 1.7119903564453125,
"learning_rate": 1.4743132725302427e-06,
"loss": 0.5262482166290283,
"step": 2984
},
{
"epoch": 2.2604087812263436,
"grad_norm": 1.4337292909622192,
"learning_rate": 1.472637745084801e-06,
"loss": 0.4823380410671234,
"step": 2986
},
{
"epoch": 2.2619227857683573,
"grad_norm": 2.332228183746338,
"learning_rate": 1.47096249560204e-06,
"loss": 0.983479917049408,
"step": 2988
},
{
"epoch": 2.263436790310371,
"grad_norm": 9.281291961669922,
"learning_rate": 1.469287526707415e-06,
"loss": 0.6645342111587524,
"step": 2990
},
{
"epoch": 2.2649507948523846,
"grad_norm": 11.439191818237305,
"learning_rate": 1.467612841025942e-06,
"loss": 0.43795791268348694,
"step": 2992
},
{
"epoch": 2.2664647993943983,
"grad_norm": 0.34899380803108215,
"learning_rate": 1.465938441182195e-06,
"loss": 0.486656129360199,
"step": 2994
},
{
"epoch": 2.267978803936412,
"grad_norm": 1.5586745738983154,
"learning_rate": 1.4642643298002977e-06,
"loss": 0.5811508893966675,
"step": 2996
},
{
"epoch": 2.2694928084784256,
"grad_norm": 1.6818510293960571,
"learning_rate": 1.4625905095039232e-06,
"loss": 0.9852191209793091,
"step": 2998
},
{
"epoch": 2.2710068130204393,
"grad_norm": 3.359684467315674,
"learning_rate": 1.4609169829162866e-06,
"loss": 0.5309067368507385,
"step": 3000
},
{
"epoch": 2.2725208175624525,
"grad_norm": 2.7355499267578125,
"learning_rate": 1.4592437526601462e-06,
"loss": 0.03945651277899742,
"step": 3002
},
{
"epoch": 2.274034822104466,
"grad_norm": 1.5781525373458862,
"learning_rate": 1.4575708213577915e-06,
"loss": 0.5300910472869873,
"step": 3004
},
{
"epoch": 2.27554882664648,
"grad_norm": 2.344346523284912,
"learning_rate": 1.4558981916310474e-06,
"loss": 0.08178112655878067,
"step": 3006
},
{
"epoch": 2.2770628311884935,
"grad_norm": 2.0909883975982666,
"learning_rate": 1.4542258661012636e-06,
"loss": 0.5931339859962463,
"step": 3008
},
{
"epoch": 2.278576835730507,
"grad_norm": 4.088789939880371,
"learning_rate": 1.4525538473893138e-06,
"loss": 0.05056603625416756,
"step": 3010
},
{
"epoch": 2.280090840272521,
"grad_norm": 2.683748245239258,
"learning_rate": 1.4508821381155916e-06,
"loss": 0.08106040209531784,
"step": 3012
},
{
"epoch": 2.2816048448145345,
"grad_norm": 9.581048011779785,
"learning_rate": 1.4492107409000037e-06,
"loss": 0.23262442648410797,
"step": 3014
},
{
"epoch": 2.283118849356548,
"grad_norm": 2.917015552520752,
"learning_rate": 1.4475396583619706e-06,
"loss": 0.31293708086013794,
"step": 3016
},
{
"epoch": 2.284632853898562,
"grad_norm": 1.8726041316986084,
"learning_rate": 1.445868893120417e-06,
"loss": 0.9713987112045288,
"step": 3018
},
{
"epoch": 2.286146858440575,
"grad_norm": 2.3735573291778564,
"learning_rate": 1.444198447793772e-06,
"loss": 0.9806248545646667,
"step": 3020
},
{
"epoch": 2.287660862982589,
"grad_norm": 2.190314769744873,
"learning_rate": 1.4425283249999626e-06,
"loss": 0.7395898699760437,
"step": 3022
},
{
"epoch": 2.2891748675246024,
"grad_norm": 1.5809588432312012,
"learning_rate": 1.4408585273564101e-06,
"loss": 0.5897817015647888,
"step": 3024
},
{
"epoch": 2.290688872066616,
"grad_norm": 1.4036309719085693,
"learning_rate": 1.4391890574800273e-06,
"loss": 0.5153738260269165,
"step": 3026
},
{
"epoch": 2.2922028766086298,
"grad_norm": 1.4983464479446411,
"learning_rate": 1.4375199179872111e-06,
"loss": 0.9634343981742859,
"step": 3028
},
{
"epoch": 2.2937168811506434,
"grad_norm": 2.8257312774658203,
"learning_rate": 1.435851111493844e-06,
"loss": 0.27471813559532166,
"step": 3030
},
{
"epoch": 2.295230885692657,
"grad_norm": 1.8934541940689087,
"learning_rate": 1.434182640615284e-06,
"loss": 0.5577218532562256,
"step": 3032
},
{
"epoch": 2.2967448902346708,
"grad_norm": 3.0278127193450928,
"learning_rate": 1.4325145079663634e-06,
"loss": 0.13709710538387299,
"step": 3034
},
{
"epoch": 2.2982588947766844,
"grad_norm": 1.5450032949447632,
"learning_rate": 1.4308467161613854e-06,
"loss": 0.4284347593784332,
"step": 3036
},
{
"epoch": 2.299772899318698,
"grad_norm": 3.1318891048431396,
"learning_rate": 1.4291792678141184e-06,
"loss": 0.810684323310852,
"step": 3038
},
{
"epoch": 2.3012869038607118,
"grad_norm": 6.301877021789551,
"learning_rate": 1.4275121655377932e-06,
"loss": 0.07751720398664474,
"step": 3040
},
{
"epoch": 2.302800908402725,
"grad_norm": 3.4808292388916016,
"learning_rate": 1.4258454119450961e-06,
"loss": 0.61885666847229,
"step": 3042
},
{
"epoch": 2.3043149129447387,
"grad_norm": 3.063957452774048,
"learning_rate": 1.4241790096481704e-06,
"loss": 0.4505324363708496,
"step": 3044
},
{
"epoch": 2.3058289174867523,
"grad_norm": 4.3270039558410645,
"learning_rate": 1.4225129612586064e-06,
"loss": 0.6248122453689575,
"step": 3046
},
{
"epoch": 2.307342922028766,
"grad_norm": 1.7819933891296387,
"learning_rate": 1.4208472693874397e-06,
"loss": 0.4963420331478119,
"step": 3048
},
{
"epoch": 2.3088569265707797,
"grad_norm": 1.4675928354263306,
"learning_rate": 1.4191819366451482e-06,
"loss": 1.0235692262649536,
"step": 3050
},
{
"epoch": 2.3103709311127933,
"grad_norm": 3.1224544048309326,
"learning_rate": 1.4175169656416467e-06,
"loss": 0.530079185962677,
"step": 3052
},
{
"epoch": 2.311884935654807,
"grad_norm": 1.1855491399765015,
"learning_rate": 1.4158523589862829e-06,
"loss": 0.5151659250259399,
"step": 3054
},
{
"epoch": 2.3133989401968207,
"grad_norm": 2.658396005630493,
"learning_rate": 1.4141881192878332e-06,
"loss": 0.17711491882801056,
"step": 3056
},
{
"epoch": 2.3149129447388344,
"grad_norm": 0.5937647819519043,
"learning_rate": 1.4125242491545e-06,
"loss": 0.5193536877632141,
"step": 3058
},
{
"epoch": 2.316426949280848,
"grad_norm": 1.0708814859390259,
"learning_rate": 1.4108607511939053e-06,
"loss": 0.39464443922042847,
"step": 3060
},
{
"epoch": 2.3179409538228617,
"grad_norm": 3.7162861824035645,
"learning_rate": 1.4091976280130884e-06,
"loss": 0.0514480397105217,
"step": 3062
},
{
"epoch": 2.319454958364875,
"grad_norm": 3.0497095584869385,
"learning_rate": 1.4075348822185006e-06,
"loss": 0.14000831544399261,
"step": 3064
},
{
"epoch": 2.3209689629068886,
"grad_norm": 1.7098908424377441,
"learning_rate": 1.4058725164160035e-06,
"loss": 0.2299700379371643,
"step": 3066
},
{
"epoch": 2.3224829674489023,
"grad_norm": 2.4175784587860107,
"learning_rate": 1.404210533210861e-06,
"loss": 0.7841640114784241,
"step": 3068
},
{
"epoch": 2.323996971990916,
"grad_norm": 8.18373966217041,
"learning_rate": 1.4025489352077387e-06,
"loss": 0.6784352660179138,
"step": 3070
},
{
"epoch": 2.3255109765329296,
"grad_norm": 1.799009919166565,
"learning_rate": 1.4008877250106977e-06,
"loss": 0.6317290663719177,
"step": 3072
},
{
"epoch": 2.3270249810749433,
"grad_norm": 3.8978631496429443,
"learning_rate": 1.399226905223193e-06,
"loss": 1.0654164552688599,
"step": 3074
},
{
"epoch": 2.328538985616957,
"grad_norm": 1.7303528785705566,
"learning_rate": 1.3975664784480653e-06,
"loss": 0.9460194110870361,
"step": 3076
},
{
"epoch": 2.3300529901589706,
"grad_norm": 1.330672025680542,
"learning_rate": 1.3959064472875406e-06,
"loss": 0.11510738730430603,
"step": 3078
},
{
"epoch": 2.3315669947009843,
"grad_norm": 3.222262144088745,
"learning_rate": 1.3942468143432263e-06,
"loss": 0.6555764675140381,
"step": 3080
},
{
"epoch": 2.3330809992429975,
"grad_norm": 3.8371193408966064,
"learning_rate": 1.3925875822161034e-06,
"loss": 0.09196102619171143,
"step": 3082
},
{
"epoch": 2.334595003785011,
"grad_norm": 2.885496139526367,
"learning_rate": 1.3909287535065254e-06,
"loss": 0.4630357623100281,
"step": 3084
},
{
"epoch": 2.336109008327025,
"grad_norm": 1.3782941102981567,
"learning_rate": 1.3892703308142144e-06,
"loss": 0.39131051301956177,
"step": 3086
},
{
"epoch": 2.3376230128690385,
"grad_norm": 1.547868251800537,
"learning_rate": 1.3876123167382551e-06,
"loss": 0.9046714901924133,
"step": 3088
},
{
"epoch": 2.339137017411052,
"grad_norm": 1.5988070964813232,
"learning_rate": 1.3859547138770932e-06,
"loss": 0.5064923167228699,
"step": 3090
},
{
"epoch": 2.340651021953066,
"grad_norm": 2.8355846405029297,
"learning_rate": 1.3842975248285284e-06,
"loss": 0.6032267212867737,
"step": 3092
},
{
"epoch": 2.3421650264950795,
"grad_norm": 1.981899380683899,
"learning_rate": 1.382640752189712e-06,
"loss": 0.8943812847137451,
"step": 3094
},
{
"epoch": 2.343679031037093,
"grad_norm": 4.821799278259277,
"learning_rate": 1.380984398557145e-06,
"loss": 0.99314284324646,
"step": 3096
},
{
"epoch": 2.345193035579107,
"grad_norm": 2.2897233963012695,
"learning_rate": 1.3793284665266681e-06,
"loss": 0.3706459105014801,
"step": 3098
},
{
"epoch": 2.3467070401211205,
"grad_norm": 2.3508996963500977,
"learning_rate": 1.3776729586934643e-06,
"loss": 0.09634893387556076,
"step": 3100
},
{
"epoch": 2.348221044663134,
"grad_norm": 3.219851016998291,
"learning_rate": 1.3760178776520502e-06,
"loss": 0.0941806435585022,
"step": 3102
},
{
"epoch": 2.3497350492051474,
"grad_norm": 2.2882401943206787,
"learning_rate": 1.3743632259962745e-06,
"loss": 0.9920378923416138,
"step": 3104
},
{
"epoch": 2.351249053747161,
"grad_norm": 2.4805197715759277,
"learning_rate": 1.3727090063193114e-06,
"loss": 0.2563851475715637,
"step": 3106
},
{
"epoch": 2.3527630582891748,
"grad_norm": 1.203678846359253,
"learning_rate": 1.3710552212136604e-06,
"loss": 0.09712015837430954,
"step": 3108
},
{
"epoch": 2.3542770628311884,
"grad_norm": 1.679890751838684,
"learning_rate": 1.3694018732711379e-06,
"loss": 0.5343849658966064,
"step": 3110
},
{
"epoch": 2.355791067373202,
"grad_norm": 2.469313859939575,
"learning_rate": 1.367748965082876e-06,
"loss": 0.45109063386917114,
"step": 3112
},
{
"epoch": 2.3573050719152158,
"grad_norm": 2.0011284351348877,
"learning_rate": 1.3660964992393176e-06,
"loss": 0.9690456390380859,
"step": 3114
},
{
"epoch": 2.3588190764572294,
"grad_norm": 2.9229679107666016,
"learning_rate": 1.3644444783302122e-06,
"loss": 0.4914405941963196,
"step": 3116
},
{
"epoch": 2.360333080999243,
"grad_norm": 4.822331428527832,
"learning_rate": 1.3627929049446132e-06,
"loss": 0.9364929795265198,
"step": 3118
},
{
"epoch": 2.3618470855412568,
"grad_norm": 2.3485610485076904,
"learning_rate": 1.3611417816708704e-06,
"loss": 0.08451011031866074,
"step": 3120
},
{
"epoch": 2.3633610900832704,
"grad_norm": 3.3834290504455566,
"learning_rate": 1.3594911110966294e-06,
"loss": 0.7591638565063477,
"step": 3122
},
{
"epoch": 2.364875094625284,
"grad_norm": 3.9524335861206055,
"learning_rate": 1.357840895808827e-06,
"loss": 0.5419999957084656,
"step": 3124
},
{
"epoch": 2.3663890991672973,
"grad_norm": 2.177633047103882,
"learning_rate": 1.3561911383936855e-06,
"loss": 0.5028657913208008,
"step": 3126
},
{
"epoch": 2.367903103709311,
"grad_norm": 1.8020979166030884,
"learning_rate": 1.3545418414367094e-06,
"loss": 0.5325472354888916,
"step": 3128
},
{
"epoch": 2.3694171082513247,
"grad_norm": 2.580286741256714,
"learning_rate": 1.3528930075226817e-06,
"loss": 0.7464925646781921,
"step": 3130
},
{
"epoch": 2.3709311127933383,
"grad_norm": 1.5663801431655884,
"learning_rate": 1.3512446392356616e-06,
"loss": 0.10833124816417694,
"step": 3132
},
{
"epoch": 2.372445117335352,
"grad_norm": 2.8936641216278076,
"learning_rate": 1.3495967391589757e-06,
"loss": 0.6852737665176392,
"step": 3134
},
{
"epoch": 2.3739591218773657,
"grad_norm": 16.39162254333496,
"learning_rate": 1.347949309875219e-06,
"loss": 0.14308148622512817,
"step": 3136
},
{
"epoch": 2.3754731264193794,
"grad_norm": 2.1235458850860596,
"learning_rate": 1.3463023539662466e-06,
"loss": 0.5667533278465271,
"step": 3138
},
{
"epoch": 2.376987130961393,
"grad_norm": 4.373684406280518,
"learning_rate": 1.3446558740131748e-06,
"loss": 0.15360027551651,
"step": 3140
},
{
"epoch": 2.3785011355034067,
"grad_norm": 2.0062785148620605,
"learning_rate": 1.3430098725963704e-06,
"loss": 0.7532010078430176,
"step": 3142
},
{
"epoch": 2.38001514004542,
"grad_norm": 4.252340316772461,
"learning_rate": 1.3413643522954523e-06,
"loss": 0.11521820724010468,
"step": 3144
},
{
"epoch": 2.3815291445874336,
"grad_norm": 2.2112393379211426,
"learning_rate": 1.3397193156892861e-06,
"loss": 0.234744593501091,
"step": 3146
},
{
"epoch": 2.3830431491294473,
"grad_norm": 2.064846992492676,
"learning_rate": 1.3380747653559774e-06,
"loss": 0.3616696298122406,
"step": 3148
},
{
"epoch": 2.384557153671461,
"grad_norm": 3.4467649459838867,
"learning_rate": 1.3364307038728712e-06,
"loss": 0.5284278392791748,
"step": 3150
},
{
"epoch": 2.3860711582134746,
"grad_norm": 1.8391021490097046,
"learning_rate": 1.3347871338165446e-06,
"loss": 0.21906384825706482,
"step": 3152
},
{
"epoch": 2.3875851627554883,
"grad_norm": 2.8953263759613037,
"learning_rate": 1.3331440577628072e-06,
"loss": 0.5176718831062317,
"step": 3154
},
{
"epoch": 2.389099167297502,
"grad_norm": 1.4748371839523315,
"learning_rate": 1.3315014782866924e-06,
"loss": 0.9320688843727112,
"step": 3156
},
{
"epoch": 2.3906131718395156,
"grad_norm": 4.977100849151611,
"learning_rate": 1.3298593979624551e-06,
"loss": 0.07454890012741089,
"step": 3158
},
{
"epoch": 2.3921271763815293,
"grad_norm": 2.885956048965454,
"learning_rate": 1.3282178193635696e-06,
"loss": 0.20941953361034393,
"step": 3160
},
{
"epoch": 2.393641180923543,
"grad_norm": 1.426200032234192,
"learning_rate": 1.3265767450627227e-06,
"loss": 0.20474065840244293,
"step": 3162
},
{
"epoch": 2.3951551854655566,
"grad_norm": 17.105777740478516,
"learning_rate": 1.3249361776318117e-06,
"loss": 0.33967655897140503,
"step": 3164
},
{
"epoch": 2.39666919000757,
"grad_norm": 7.441493988037109,
"learning_rate": 1.3232961196419376e-06,
"loss": 0.5956662893295288,
"step": 3166
},
{
"epoch": 2.3981831945495835,
"grad_norm": 0.9323331713676453,
"learning_rate": 1.321656573663406e-06,
"loss": 0.12122826278209686,
"step": 3168
},
{
"epoch": 2.399697199091597,
"grad_norm": 1.7128007411956787,
"learning_rate": 1.3200175422657182e-06,
"loss": 0.5543103814125061,
"step": 3170
},
{
"epoch": 2.401211203633611,
"grad_norm": 1.926504135131836,
"learning_rate": 1.318379028017568e-06,
"loss": 0.09472674131393433,
"step": 3172
},
{
"epoch": 2.4027252081756245,
"grad_norm": 1.6713393926620483,
"learning_rate": 1.3167410334868418e-06,
"loss": 0.5352200269699097,
"step": 3174
},
{
"epoch": 2.404239212717638,
"grad_norm": 1.0860508680343628,
"learning_rate": 1.3151035612406088e-06,
"loss": 0.6046788096427917,
"step": 3176
},
{
"epoch": 2.405753217259652,
"grad_norm": 1.8982021808624268,
"learning_rate": 1.3134666138451209e-06,
"loss": 0.1095985621213913,
"step": 3178
},
{
"epoch": 2.4072672218016655,
"grad_norm": 4.064739227294922,
"learning_rate": 1.3118301938658064e-06,
"loss": 0.37438011169433594,
"step": 3180
},
{
"epoch": 2.408781226343679,
"grad_norm": 24.300867080688477,
"learning_rate": 1.3101943038672687e-06,
"loss": 0.14832612872123718,
"step": 3182
},
{
"epoch": 2.4102952308856924,
"grad_norm": 3.4894673824310303,
"learning_rate": 1.30855894641328e-06,
"loss": 0.051485493779182434,
"step": 3184
},
{
"epoch": 2.4118092354277065,
"grad_norm": 1.913360834121704,
"learning_rate": 1.3069241240667765e-06,
"loss": 0.8969357013702393,
"step": 3186
},
{
"epoch": 2.4133232399697198,
"grad_norm": 2.673508644104004,
"learning_rate": 1.3052898393898576e-06,
"loss": 1.0752180814743042,
"step": 3188
},
{
"epoch": 2.4148372445117334,
"grad_norm": 1.7779754400253296,
"learning_rate": 1.303656094943779e-06,
"loss": 0.637548565864563,
"step": 3190
},
{
"epoch": 2.416351249053747,
"grad_norm": 2.6721699237823486,
"learning_rate": 1.3020228932889508e-06,
"loss": 0.1810387670993805,
"step": 3192
},
{
"epoch": 2.4178652535957608,
"grad_norm": 3.1952381134033203,
"learning_rate": 1.3003902369849306e-06,
"loss": 0.5444831848144531,
"step": 3194
},
{
"epoch": 2.4193792581377744,
"grad_norm": 2.336055040359497,
"learning_rate": 1.2987581285904236e-06,
"loss": 0.5591921210289001,
"step": 3196
},
{
"epoch": 2.420893262679788,
"grad_norm": 5.412187576293945,
"learning_rate": 1.2971265706632747e-06,
"loss": 0.17424336075782776,
"step": 3198
},
{
"epoch": 2.4224072672218018,
"grad_norm": 1.522705078125,
"learning_rate": 1.2954955657604666e-06,
"loss": 0.08939556032419205,
"step": 3200
},
{
"epoch": 2.4239212717638154,
"grad_norm": 1.936441421508789,
"learning_rate": 1.293865116438115e-06,
"loss": 0.17253662645816803,
"step": 3202
},
{
"epoch": 2.425435276305829,
"grad_norm": 6.317035675048828,
"learning_rate": 1.2922352252514653e-06,
"loss": 0.17524859309196472,
"step": 3204
},
{
"epoch": 2.4269492808478423,
"grad_norm": 10.295120239257812,
"learning_rate": 1.2906058947548886e-06,
"loss": 0.510208249092102,
"step": 3206
},
{
"epoch": 2.428463285389856,
"grad_norm": 1.5548295974731445,
"learning_rate": 1.2889771275018757e-06,
"loss": 0.9413988590240479,
"step": 3208
},
{
"epoch": 2.4299772899318697,
"grad_norm": 2.107252597808838,
"learning_rate": 1.287348926045037e-06,
"loss": 0.44147780537605286,
"step": 3210
},
{
"epoch": 2.4314912944738833,
"grad_norm": 2.5720386505126953,
"learning_rate": 1.285721292936094e-06,
"loss": 0.5519513487815857,
"step": 3212
},
{
"epoch": 2.433005299015897,
"grad_norm": 1.7843011617660522,
"learning_rate": 1.2840942307258784e-06,
"loss": 0.5143168568611145,
"step": 3214
},
{
"epoch": 2.4345193035579107,
"grad_norm": 2.339644432067871,
"learning_rate": 1.2824677419643277e-06,
"loss": 0.5728105902671814,
"step": 3216
},
{
"epoch": 2.4360333080999244,
"grad_norm": 4.653833866119385,
"learning_rate": 1.2808418292004795e-06,
"loss": 0.48528730869293213,
"step": 3218
},
{
"epoch": 2.437547312641938,
"grad_norm": 9.157849311828613,
"learning_rate": 1.2792164949824702e-06,
"loss": 0.5953377485275269,
"step": 3220
},
{
"epoch": 2.4390613171839517,
"grad_norm": 1.5174750089645386,
"learning_rate": 1.2775917418575284e-06,
"loss": 0.5056532621383667,
"step": 3222
},
{
"epoch": 2.4405753217259654,
"grad_norm": 2.567638874053955,
"learning_rate": 1.275967572371971e-06,
"loss": 0.5474219918251038,
"step": 3224
},
{
"epoch": 2.442089326267979,
"grad_norm": 4.96596097946167,
"learning_rate": 1.2743439890712035e-06,
"loss": 0.12137976288795471,
"step": 3226
},
{
"epoch": 2.4436033308099923,
"grad_norm": 1.7557021379470825,
"learning_rate": 1.2727209944997099e-06,
"loss": 0.867196261882782,
"step": 3228
},
{
"epoch": 2.445117335352006,
"grad_norm": 3.688567876815796,
"learning_rate": 1.2710985912010514e-06,
"loss": 0.6646262407302856,
"step": 3230
},
{
"epoch": 2.4466313398940196,
"grad_norm": 1.933400273323059,
"learning_rate": 1.2694767817178651e-06,
"loss": 0.4808739423751831,
"step": 3232
},
{
"epoch": 2.4481453444360333,
"grad_norm": 1.9126027822494507,
"learning_rate": 1.2678555685918549e-06,
"loss": 0.7815792560577393,
"step": 3234
},
{
"epoch": 2.449659348978047,
"grad_norm": 5.408775329589844,
"learning_rate": 1.2662349543637915e-06,
"loss": 0.5138024091720581,
"step": 3236
},
{
"epoch": 2.4511733535200606,
"grad_norm": 6.258272647857666,
"learning_rate": 1.2646149415735061e-06,
"loss": 0.15966981649398804,
"step": 3238
},
{
"epoch": 2.4526873580620743,
"grad_norm": 6.899496555328369,
"learning_rate": 1.2629955327598884e-06,
"loss": 0.4678770899772644,
"step": 3240
},
{
"epoch": 2.454201362604088,
"grad_norm": 1.813686490058899,
"learning_rate": 1.2613767304608808e-06,
"loss": 0.9744421243667603,
"step": 3242
},
{
"epoch": 2.4557153671461016,
"grad_norm": 7.620000839233398,
"learning_rate": 1.2597585372134754e-06,
"loss": 0.43850672245025635,
"step": 3244
},
{
"epoch": 2.457229371688115,
"grad_norm": 1.3412652015686035,
"learning_rate": 1.2581409555537087e-06,
"loss": 0.06274896115064621,
"step": 3246
},
{
"epoch": 2.4587433762301285,
"grad_norm": 1.733108639717102,
"learning_rate": 1.2565239880166613e-06,
"loss": 0.020732510834932327,
"step": 3248
},
{
"epoch": 2.460257380772142,
"grad_norm": 1.8441457748413086,
"learning_rate": 1.2549076371364487e-06,
"loss": 0.09537363052368164,
"step": 3250
},
{
"epoch": 2.461771385314156,
"grad_norm": 3.559648036956787,
"learning_rate": 1.2532919054462209e-06,
"loss": 0.21012158691883087,
"step": 3252
},
{
"epoch": 2.4632853898561695,
"grad_norm": 1.165614128112793,
"learning_rate": 1.2516767954781588e-06,
"loss": 0.09470941871404648,
"step": 3254
},
{
"epoch": 2.464799394398183,
"grad_norm": 2.507286310195923,
"learning_rate": 1.250062309763467e-06,
"loss": 0.46259403228759766,
"step": 3256
},
{
"epoch": 2.466313398940197,
"grad_norm": 2.2950599193573,
"learning_rate": 1.248448450832373e-06,
"loss": 0.0759514644742012,
"step": 3258
},
{
"epoch": 2.4678274034822105,
"grad_norm": 0.58045494556427,
"learning_rate": 1.2468352212141202e-06,
"loss": 0.11328712105751038,
"step": 3260
},
{
"epoch": 2.469341408024224,
"grad_norm": 3.1575472354888916,
"learning_rate": 1.245222623436969e-06,
"loss": 0.7599681615829468,
"step": 3262
},
{
"epoch": 2.470855412566238,
"grad_norm": 1.8420294523239136,
"learning_rate": 1.243610660028186e-06,
"loss": 0.2730555236339569,
"step": 3264
},
{
"epoch": 2.4723694171082515,
"grad_norm": 4.679914951324463,
"learning_rate": 1.2419993335140467e-06,
"loss": 0.5341447591781616,
"step": 3266
},
{
"epoch": 2.4738834216502648,
"grad_norm": 1.712031364440918,
"learning_rate": 1.2403886464198259e-06,
"loss": 0.44335949420928955,
"step": 3268
},
{
"epoch": 2.4753974261922784,
"grad_norm": 2.764129638671875,
"learning_rate": 1.2387786012697987e-06,
"loss": 0.5020673274993896,
"step": 3270
},
{
"epoch": 2.476911430734292,
"grad_norm": 6.6505842208862305,
"learning_rate": 1.237169200587232e-06,
"loss": 0.021848905831575394,
"step": 3272
},
{
"epoch": 2.4784254352763058,
"grad_norm": 1.904347538948059,
"learning_rate": 1.235560446894383e-06,
"loss": 0.23235182464122772,
"step": 3274
},
{
"epoch": 2.4799394398183194,
"grad_norm": 3.316446542739868,
"learning_rate": 1.233952342712497e-06,
"loss": 0.16898798942565918,
"step": 3276
},
{
"epoch": 2.481453444360333,
"grad_norm": 3.6128406524658203,
"learning_rate": 1.232344890561799e-06,
"loss": 0.5249270796775818,
"step": 3278
},
{
"epoch": 2.4829674489023468,
"grad_norm": 2.04036021232605,
"learning_rate": 1.2307380929614932e-06,
"loss": 0.5710758566856384,
"step": 3280
},
{
"epoch": 2.4844814534443604,
"grad_norm": 1.612221121788025,
"learning_rate": 1.2291319524297573e-06,
"loss": 0.4911920726299286,
"step": 3282
},
{
"epoch": 2.485995457986374,
"grad_norm": 2.189225196838379,
"learning_rate": 1.2275264714837408e-06,
"loss": 0.12952546775341034,
"step": 3284
},
{
"epoch": 2.4875094625283873,
"grad_norm": 2.0172598361968994,
"learning_rate": 1.225921652639558e-06,
"loss": 0.3338145315647125,
"step": 3286
},
{
"epoch": 2.4890234670704015,
"grad_norm": 1.3855159282684326,
"learning_rate": 1.2243174984122853e-06,
"loss": 0.4880499541759491,
"step": 3288
},
{
"epoch": 2.4905374716124147,
"grad_norm": 1.6459860801696777,
"learning_rate": 1.2227140113159594e-06,
"loss": 1.032573938369751,
"step": 3290
},
{
"epoch": 2.4920514761544283,
"grad_norm": 3.479882001876831,
"learning_rate": 1.2211111938635695e-06,
"loss": 0.8720075488090515,
"step": 3292
},
{
"epoch": 2.493565480696442,
"grad_norm": 2.3142521381378174,
"learning_rate": 1.2195090485670563e-06,
"loss": 0.9480175971984863,
"step": 3294
},
{
"epoch": 2.4950794852384557,
"grad_norm": 2.713301181793213,
"learning_rate": 1.2179075779373064e-06,
"loss": 0.597554624080658,
"step": 3296
},
{
"epoch": 2.4965934897804694,
"grad_norm": 1.2404519319534302,
"learning_rate": 1.216306784484151e-06,
"loss": 0.5238319039344788,
"step": 3298
},
{
"epoch": 2.498107494322483,
"grad_norm": 2.1829569339752197,
"learning_rate": 1.2147066707163578e-06,
"loss": 0.2818773686885834,
"step": 3300
},
{
"epoch": 2.4996214988644967,
"grad_norm": 0.9500485062599182,
"learning_rate": 1.2131072391416298e-06,
"loss": 0.5698157548904419,
"step": 3302
},
{
"epoch": 2.5011355034065104,
"grad_norm": 42.315731048583984,
"learning_rate": 1.2115084922666007e-06,
"loss": 0.926612138748169,
"step": 3304
},
{
"epoch": 2.502649507948524,
"grad_norm": 2.067148447036743,
"learning_rate": 1.2099104325968327e-06,
"loss": 0.5611676573753357,
"step": 3306
},
{
"epoch": 2.5041635124905373,
"grad_norm": 1.7812166213989258,
"learning_rate": 1.20831306263681e-06,
"loss": 0.3297508656978607,
"step": 3308
},
{
"epoch": 2.5056775170325514,
"grad_norm": 2.64509916305542,
"learning_rate": 1.2067163848899345e-06,
"loss": 0.36095792055130005,
"step": 3310
},
{
"epoch": 2.5071915215745646,
"grad_norm": 1.3159887790679932,
"learning_rate": 1.2051204018585258e-06,
"loss": 0.45377278327941895,
"step": 3312
},
{
"epoch": 2.5087055261165783,
"grad_norm": 1.2811301946640015,
"learning_rate": 1.203525116043813e-06,
"loss": 0.5460622906684875,
"step": 3314
},
{
"epoch": 2.510219530658592,
"grad_norm": 1.6105045080184937,
"learning_rate": 1.201930529945933e-06,
"loss": 0.5222399830818176,
"step": 3316
},
{
"epoch": 2.5117335352006056,
"grad_norm": 1.8894621133804321,
"learning_rate": 1.2003366460639257e-06,
"loss": 0.47699031233787537,
"step": 3318
},
{
"epoch": 2.5132475397426193,
"grad_norm": 7.224452972412109,
"learning_rate": 1.1987434668957316e-06,
"loss": 0.18395495414733887,
"step": 3320
},
{
"epoch": 2.514761544284633,
"grad_norm": 2.792215585708618,
"learning_rate": 1.1971509949381862e-06,
"loss": 0.962587833404541,
"step": 3322
},
{
"epoch": 2.5162755488266466,
"grad_norm": 2.333487033843994,
"learning_rate": 1.1955592326870153e-06,
"loss": 0.9294099807739258,
"step": 3324
},
{
"epoch": 2.51778955336866,
"grad_norm": 2.340708017349243,
"learning_rate": 1.1939681826368353e-06,
"loss": 0.551225483417511,
"step": 3326
},
{
"epoch": 2.519303557910674,
"grad_norm": 4.234295845031738,
"learning_rate": 1.192377847281144e-06,
"loss": 0.5359384417533875,
"step": 3328
},
{
"epoch": 2.520817562452687,
"grad_norm": 9.935158729553223,
"learning_rate": 1.1907882291123196e-06,
"loss": 0.18878018856048584,
"step": 3330
},
{
"epoch": 2.522331566994701,
"grad_norm": 1.8245105743408203,
"learning_rate": 1.1891993306216168e-06,
"loss": 0.5585557222366333,
"step": 3332
},
{
"epoch": 2.5238455715367145,
"grad_norm": 3.751849889755249,
"learning_rate": 1.187611154299163e-06,
"loss": 0.695787250995636,
"step": 3334
},
{
"epoch": 2.525359576078728,
"grad_norm": 1.6436039209365845,
"learning_rate": 1.1860237026339524e-06,
"loss": 0.5512560606002808,
"step": 3336
},
{
"epoch": 2.526873580620742,
"grad_norm": 1.9282137155532837,
"learning_rate": 1.1844369781138445e-06,
"loss": 0.5277320146560669,
"step": 3338
},
{
"epoch": 2.5283875851627555,
"grad_norm": 1.7889491319656372,
"learning_rate": 1.1828509832255586e-06,
"loss": 0.790787935256958,
"step": 3340
},
{
"epoch": 2.529901589704769,
"grad_norm": 1.8602066040039062,
"learning_rate": 1.181265720454671e-06,
"loss": 0.5047529935836792,
"step": 3342
},
{
"epoch": 2.531415594246783,
"grad_norm": 1.6128684282302856,
"learning_rate": 1.1796811922856107e-06,
"loss": 1.01747727394104,
"step": 3344
},
{
"epoch": 2.5329295987887965,
"grad_norm": 2.1299707889556885,
"learning_rate": 1.1780974012016552e-06,
"loss": 0.1209339052438736,
"step": 3346
},
{
"epoch": 2.5344436033308098,
"grad_norm": 1.494289755821228,
"learning_rate": 1.1765143496849262e-06,
"loss": 0.11673936992883682,
"step": 3348
},
{
"epoch": 2.535957607872824,
"grad_norm": 10.133544921875,
"learning_rate": 1.1749320402163878e-06,
"loss": 0.04495559260249138,
"step": 3350
},
{
"epoch": 2.537471612414837,
"grad_norm": 1.7074990272521973,
"learning_rate": 1.1733504752758404e-06,
"loss": 0.6414316892623901,
"step": 3352
},
{
"epoch": 2.5389856169568508,
"grad_norm": 2.5590038299560547,
"learning_rate": 1.1717696573419162e-06,
"loss": 0.27318763732910156,
"step": 3354
},
{
"epoch": 2.5404996214988644,
"grad_norm": 2.422823667526245,
"learning_rate": 1.1701895888920792e-06,
"loss": 0.3937920331954956,
"step": 3356
},
{
"epoch": 2.542013626040878,
"grad_norm": 5.183249473571777,
"learning_rate": 1.1686102724026177e-06,
"loss": 0.6359275579452515,
"step": 3358
},
{
"epoch": 2.5435276305828918,
"grad_norm": 2.020559549331665,
"learning_rate": 1.1670317103486403e-06,
"loss": 0.5142892599105835,
"step": 3360
},
{
"epoch": 2.5450416351249054,
"grad_norm": 5.478321552276611,
"learning_rate": 1.165453905204076e-06,
"loss": 0.32605746388435364,
"step": 3362
},
{
"epoch": 2.546555639666919,
"grad_norm": 1.0260086059570312,
"learning_rate": 1.1638768594416648e-06,
"loss": 0.6114858388900757,
"step": 3364
},
{
"epoch": 2.548069644208933,
"grad_norm": 15.645800590515137,
"learning_rate": 1.162300575532958e-06,
"loss": 0.07219941914081573,
"step": 3366
},
{
"epoch": 2.5495836487509465,
"grad_norm": 2.069432497024536,
"learning_rate": 1.1607250559483121e-06,
"loss": 0.9569191336631775,
"step": 3368
},
{
"epoch": 2.5510976532929597,
"grad_norm": 1.3018686771392822,
"learning_rate": 1.1591503031568875e-06,
"loss": 0.5765461921691895,
"step": 3370
},
{
"epoch": 2.5526116578349733,
"grad_norm": 3.0679144859313965,
"learning_rate": 1.1575763196266412e-06,
"loss": 0.6863498687744141,
"step": 3372
},
{
"epoch": 2.554125662376987,
"grad_norm": 3.4148383140563965,
"learning_rate": 1.1560031078243248e-06,
"loss": 0.5839001536369324,
"step": 3374
},
{
"epoch": 2.5556396669190007,
"grad_norm": 7.998443603515625,
"learning_rate": 1.1544306702154807e-06,
"loss": 0.8170933723449707,
"step": 3376
},
{
"epoch": 2.5571536714610144,
"grad_norm": 2.7478621006011963,
"learning_rate": 1.1528590092644387e-06,
"loss": 0.3825238049030304,
"step": 3378
},
{
"epoch": 2.558667676003028,
"grad_norm": 2.4262313842773438,
"learning_rate": 1.1512881274343105e-06,
"loss": 0.9992386102676392,
"step": 3380
},
{
"epoch": 2.5601816805450417,
"grad_norm": 3.3657608032226562,
"learning_rate": 1.1497180271869862e-06,
"loss": 0.5366327166557312,
"step": 3382
},
{
"epoch": 2.5616956850870554,
"grad_norm": 3.6988351345062256,
"learning_rate": 1.1481487109831329e-06,
"loss": 0.9917261004447937,
"step": 3384
},
{
"epoch": 2.563209689629069,
"grad_norm": 2.595430374145508,
"learning_rate": 1.1465801812821875e-06,
"loss": 0.06481704860925674,
"step": 3386
},
{
"epoch": 2.5647236941710823,
"grad_norm": 2.457298994064331,
"learning_rate": 1.1450124405423544e-06,
"loss": 0.8893528580665588,
"step": 3388
},
{
"epoch": 2.5662376987130964,
"grad_norm": 2.416517972946167,
"learning_rate": 1.1434454912206018e-06,
"loss": 0.14643681049346924,
"step": 3390
},
{
"epoch": 2.5677517032551096,
"grad_norm": 2.499912977218628,
"learning_rate": 1.1418793357726579e-06,
"loss": 0.07367072999477386,
"step": 3392
},
{
"epoch": 2.5692657077971233,
"grad_norm": 1.3387746810913086,
"learning_rate": 1.1403139766530063e-06,
"loss": 0.48521560430526733,
"step": 3394
},
{
"epoch": 2.570779712339137,
"grad_norm": 1.4146757125854492,
"learning_rate": 1.1387494163148827e-06,
"loss": 0.07745841890573502,
"step": 3396
},
{
"epoch": 2.5722937168811506,
"grad_norm": 3.8005497455596924,
"learning_rate": 1.1371856572102705e-06,
"loss": 0.27372169494628906,
"step": 3398
},
{
"epoch": 2.5738077214231643,
"grad_norm": 5.660574913024902,
"learning_rate": 1.1356227017898985e-06,
"loss": 0.5278230905532837,
"step": 3400
},
{
"epoch": 2.575321725965178,
"grad_norm": 1.703931212425232,
"learning_rate": 1.1340605525032353e-06,
"loss": 0.5331737995147705,
"step": 3402
},
{
"epoch": 2.5768357305071916,
"grad_norm": 3.0242526531219482,
"learning_rate": 1.1324992117984852e-06,
"loss": 0.9489548206329346,
"step": 3404
},
{
"epoch": 2.5783497350492053,
"grad_norm": 18.639877319335938,
"learning_rate": 1.1309386821225879e-06,
"loss": 0.5219172239303589,
"step": 3406
},
{
"epoch": 2.579863739591219,
"grad_norm": 2.207932233810425,
"learning_rate": 1.1293789659212089e-06,
"loss": 0.5801731944084167,
"step": 3408
},
{
"epoch": 2.581377744133232,
"grad_norm": 1.7648814916610718,
"learning_rate": 1.127820065638741e-06,
"loss": 0.8760764598846436,
"step": 3410
},
{
"epoch": 2.5828917486752463,
"grad_norm": 5.7800517082214355,
"learning_rate": 1.1262619837182968e-06,
"loss": 0.18599385023117065,
"step": 3412
},
{
"epoch": 2.5844057532172595,
"grad_norm": 2.5123376846313477,
"learning_rate": 1.1247047226017085e-06,
"loss": 0.5670759677886963,
"step": 3414
},
{
"epoch": 2.585919757759273,
"grad_norm": 2.1895976066589355,
"learning_rate": 1.1231482847295195e-06,
"loss": 0.8945675492286682,
"step": 3416
},
{
"epoch": 2.587433762301287,
"grad_norm": 2.59700345993042,
"learning_rate": 1.1215926725409841e-06,
"loss": 0.17124387621879578,
"step": 3418
},
{
"epoch": 2.5889477668433005,
"grad_norm": 1.7561355829238892,
"learning_rate": 1.1200378884740637e-06,
"loss": 0.5565921068191528,
"step": 3420
},
{
"epoch": 2.590461771385314,
"grad_norm": 4.272517204284668,
"learning_rate": 1.1184839349654195e-06,
"loss": 0.17669063806533813,
"step": 3422
},
{
"epoch": 2.591975775927328,
"grad_norm": 2.457369327545166,
"learning_rate": 1.116930814450413e-06,
"loss": 0.9974216222763062,
"step": 3424
},
{
"epoch": 2.5934897804693415,
"grad_norm": 3.1485443115234375,
"learning_rate": 1.1153785293630988e-06,
"loss": 0.49294838309288025,
"step": 3426
},
{
"epoch": 2.5950037850113548,
"grad_norm": 1.056787133216858,
"learning_rate": 1.1138270821362239e-06,
"loss": 0.9152722358703613,
"step": 3428
},
{
"epoch": 2.596517789553369,
"grad_norm": 3.3435306549072266,
"learning_rate": 1.1122764752012208e-06,
"loss": 0.10038082301616669,
"step": 3430
},
{
"epoch": 2.598031794095382,
"grad_norm": 1.7850005626678467,
"learning_rate": 1.1107267109882053e-06,
"loss": 0.6021050810813904,
"step": 3432
},
{
"epoch": 2.5995457986373958,
"grad_norm": 4.395745277404785,
"learning_rate": 1.1091777919259723e-06,
"loss": 0.5003665089607239,
"step": 3434
},
{
"epoch": 2.6010598031794094,
"grad_norm": 1.6627795696258545,
"learning_rate": 1.107629720441994e-06,
"loss": 0.3916073143482208,
"step": 3436
},
{
"epoch": 2.602573807721423,
"grad_norm": 1.2196578979492188,
"learning_rate": 1.1060824989624123e-06,
"loss": 0.6092214584350586,
"step": 3438
},
{
"epoch": 2.6040878122634368,
"grad_norm": 1.57082200050354,
"learning_rate": 1.1045361299120364e-06,
"loss": 0.06277672201395035,
"step": 3440
},
{
"epoch": 2.6056018168054504,
"grad_norm": 3.2692267894744873,
"learning_rate": 1.1029906157143425e-06,
"loss": 0.7701391577720642,
"step": 3442
},
{
"epoch": 2.607115821347464,
"grad_norm": 4.937279224395752,
"learning_rate": 1.1014459587914638e-06,
"loss": 0.5263334512710571,
"step": 3444
},
{
"epoch": 2.608629825889478,
"grad_norm": 1.818088412284851,
"learning_rate": 1.0999021615641927e-06,
"loss": 0.3035053312778473,
"step": 3446
},
{
"epoch": 2.6101438304314915,
"grad_norm": 2.087446451187134,
"learning_rate": 1.0983592264519717e-06,
"loss": 0.537712812423706,
"step": 3448
},
{
"epoch": 2.6116578349735047,
"grad_norm": 1.2241779565811157,
"learning_rate": 1.0968171558728945e-06,
"loss": 0.5087027549743652,
"step": 3450
},
{
"epoch": 2.613171839515519,
"grad_norm": 18.9871826171875,
"learning_rate": 1.0952759522436987e-06,
"loss": 0.8322153687477112,
"step": 3452
},
{
"epoch": 2.614685844057532,
"grad_norm": 1.5005947351455688,
"learning_rate": 1.0937356179797627e-06,
"loss": 0.970147967338562,
"step": 3454
},
{
"epoch": 2.6161998485995457,
"grad_norm": 4.126323699951172,
"learning_rate": 1.092196155495105e-06,
"loss": 0.18110370635986328,
"step": 3456
},
{
"epoch": 2.6177138531415594,
"grad_norm": 1.706655502319336,
"learning_rate": 1.0906575672023743e-06,
"loss": 0.723210334777832,
"step": 3458
},
{
"epoch": 2.619227857683573,
"grad_norm": 1.5822142362594604,
"learning_rate": 1.0891198555128516e-06,
"loss": 0.9544146656990051,
"step": 3460
},
{
"epoch": 2.6207418622255867,
"grad_norm": 2.0501010417938232,
"learning_rate": 1.0875830228364431e-06,
"loss": 0.5467653870582581,
"step": 3462
},
{
"epoch": 2.6222558667676004,
"grad_norm": 1.315675139427185,
"learning_rate": 1.0860470715816785e-06,
"loss": 0.13321208953857422,
"step": 3464
},
{
"epoch": 2.623769871309614,
"grad_norm": 3.8847084045410156,
"learning_rate": 1.0845120041557049e-06,
"loss": 0.7330424785614014,
"step": 3466
},
{
"epoch": 2.6252838758516277,
"grad_norm": 8.901777267456055,
"learning_rate": 1.0829778229642848e-06,
"loss": 0.04311962053179741,
"step": 3468
},
{
"epoch": 2.6267978803936414,
"grad_norm": 2.727778434753418,
"learning_rate": 1.0814445304117917e-06,
"loss": 0.07144928723573685,
"step": 3470
},
{
"epoch": 2.6283118849356546,
"grad_norm": 2.459710121154785,
"learning_rate": 1.0799121289012068e-06,
"loss": 0.5683809518814087,
"step": 3472
},
{
"epoch": 2.6298258894776687,
"grad_norm": 4.345935344696045,
"learning_rate": 1.0783806208341141e-06,
"loss": 0.2112594097852707,
"step": 3474
},
{
"epoch": 2.631339894019682,
"grad_norm": 2.6355857849121094,
"learning_rate": 1.0768500086106978e-06,
"loss": 0.14096690714359283,
"step": 3476
},
{
"epoch": 2.6328538985616956,
"grad_norm": 0.5131763219833374,
"learning_rate": 1.075320294629739e-06,
"loss": 0.46342194080352783,
"step": 3478
},
{
"epoch": 2.6343679031037093,
"grad_norm": 1.176900029182434,
"learning_rate": 1.0737914812886094e-06,
"loss": 0.5241625308990479,
"step": 3480
},
{
"epoch": 2.635881907645723,
"grad_norm": 3.1469740867614746,
"learning_rate": 1.07226357098327e-06,
"loss": 0.14466527104377747,
"step": 3482
},
{
"epoch": 2.6373959121877366,
"grad_norm": 1.3433618545532227,
"learning_rate": 1.0707365661082674e-06,
"loss": 0.1578199863433838,
"step": 3484
},
{
"epoch": 2.6389099167297503,
"grad_norm": 2.3391337394714355,
"learning_rate": 1.069210469056727e-06,
"loss": 0.03919367492198944,
"step": 3486
},
{
"epoch": 2.640423921271764,
"grad_norm": 3.255871057510376,
"learning_rate": 1.0676852822203547e-06,
"loss": 0.3218518793582916,
"step": 3488
},
{
"epoch": 2.641937925813777,
"grad_norm": 3.4141852855682373,
"learning_rate": 1.0661610079894268e-06,
"loss": 1.0428060293197632,
"step": 3490
},
{
"epoch": 2.6434519303557913,
"grad_norm": 2.1213839054107666,
"learning_rate": 1.0646376487527907e-06,
"loss": 0.5885476469993591,
"step": 3492
},
{
"epoch": 2.6449659348978045,
"grad_norm": 4.027782440185547,
"learning_rate": 1.0631152068978604e-06,
"loss": 0.17618165910243988,
"step": 3494
},
{
"epoch": 2.646479939439818,
"grad_norm": 1.679085373878479,
"learning_rate": 1.0615936848106113e-06,
"loss": 0.9989380836486816,
"step": 3496
},
{
"epoch": 2.647993943981832,
"grad_norm": 1.7569453716278076,
"learning_rate": 1.0600730848755767e-06,
"loss": 0.47763514518737793,
"step": 3498
},
{
"epoch": 2.6495079485238455,
"grad_norm": 3.1439387798309326,
"learning_rate": 1.058553409475847e-06,
"loss": 0.4873977601528168,
"step": 3500
},
{
"epoch": 2.651021953065859,
"grad_norm": 1.0988174676895142,
"learning_rate": 1.0570346609930612e-06,
"loss": 0.39969107508659363,
"step": 3502
},
{
"epoch": 2.652535957607873,
"grad_norm": 1.5461552143096924,
"learning_rate": 1.0555168418074074e-06,
"loss": 0.21987293660640717,
"step": 3504
},
{
"epoch": 2.6540499621498865,
"grad_norm": 1.8398942947387695,
"learning_rate": 1.0539999542976152e-06,
"loss": 0.9838016629219055,
"step": 3506
},
{
"epoch": 2.6555639666919,
"grad_norm": 2.6948370933532715,
"learning_rate": 1.0524840008409575e-06,
"loss": 0.08101513981819153,
"step": 3508
},
{
"epoch": 2.657077971233914,
"grad_norm": 6.337255477905273,
"learning_rate": 1.0509689838132395e-06,
"loss": 0.010323734022676945,
"step": 3510
},
{
"epoch": 2.658591975775927,
"grad_norm": 2.0004448890686035,
"learning_rate": 1.0494549055888013e-06,
"loss": 0.5439300537109375,
"step": 3512
},
{
"epoch": 2.660105980317941,
"grad_norm": 1.2642723321914673,
"learning_rate": 1.0479417685405115e-06,
"loss": 0.8853827714920044,
"step": 3514
},
{
"epoch": 2.6616199848599544,
"grad_norm": 5.2497687339782715,
"learning_rate": 1.0464295750397626e-06,
"loss": 0.08229418843984604,
"step": 3516
},
{
"epoch": 2.663133989401968,
"grad_norm": 2.096442937850952,
"learning_rate": 1.044918327456469e-06,
"loss": 0.7499485015869141,
"step": 3518
},
{
"epoch": 2.6646479939439818,
"grad_norm": 1.2751201391220093,
"learning_rate": 1.0434080281590626e-06,
"loss": 0.6057901382446289,
"step": 3520
},
{
"epoch": 2.6661619984859954,
"grad_norm": 2.4840683937072754,
"learning_rate": 1.0418986795144896e-06,
"loss": 0.2760433852672577,
"step": 3522
},
{
"epoch": 2.667676003028009,
"grad_norm": 2.1772100925445557,
"learning_rate": 1.0403902838882056e-06,
"loss": 0.6145902276039124,
"step": 3524
},
{
"epoch": 2.669190007570023,
"grad_norm": 1.2478702068328857,
"learning_rate": 1.0388828436441733e-06,
"loss": 0.9742065072059631,
"step": 3526
},
{
"epoch": 2.6707040121120365,
"grad_norm": 0.7841724753379822,
"learning_rate": 1.0373763611448567e-06,
"loss": 0.47128647565841675,
"step": 3528
},
{
"epoch": 2.67221801665405,
"grad_norm": 1.6649569272994995,
"learning_rate": 1.035870838751221e-06,
"loss": 0.5873335599899292,
"step": 3530
},
{
"epoch": 2.673732021196064,
"grad_norm": 30.892988204956055,
"learning_rate": 1.0343662788227249e-06,
"loss": 0.08930249512195587,
"step": 3532
},
{
"epoch": 2.675246025738077,
"grad_norm": 2.34309458732605,
"learning_rate": 1.0328626837173202e-06,
"loss": 0.5236271619796753,
"step": 3534
},
{
"epoch": 2.6767600302800907,
"grad_norm": 1.6401013135910034,
"learning_rate": 1.0313600557914452e-06,
"loss": 0.28992849588394165,
"step": 3536
},
{
"epoch": 2.6782740348221044,
"grad_norm": 1.5496026277542114,
"learning_rate": 1.029858397400023e-06,
"loss": 0.11498652398586273,
"step": 3538
},
{
"epoch": 2.679788039364118,
"grad_norm": 2.9359993934631348,
"learning_rate": 1.028357710896458e-06,
"loss": 0.45273083448410034,
"step": 3540
},
{
"epoch": 2.6813020439061317,
"grad_norm": 3.2220873832702637,
"learning_rate": 1.0268579986326298e-06,
"loss": 0.5863102674484253,
"step": 3542
},
{
"epoch": 2.6828160484481454,
"grad_norm": 4.268006324768066,
"learning_rate": 1.0253592629588934e-06,
"loss": 0.9020994305610657,
"step": 3544
},
{
"epoch": 2.684330052990159,
"grad_norm": 2.0981273651123047,
"learning_rate": 1.0238615062240713e-06,
"loss": 0.4718743860721588,
"step": 3546
},
{
"epoch": 2.6858440575321727,
"grad_norm": 2.6294424533843994,
"learning_rate": 1.0223647307754524e-06,
"loss": 0.38430675864219666,
"step": 3548
},
{
"epoch": 2.6873580620741864,
"grad_norm": 0.6526815295219421,
"learning_rate": 1.0208689389587875e-06,
"loss": 0.10240335762500763,
"step": 3550
},
{
"epoch": 2.6888720666161996,
"grad_norm": 1.712877869606018,
"learning_rate": 1.0193741331182873e-06,
"loss": 0.18102604150772095,
"step": 3552
},
{
"epoch": 2.6903860711582137,
"grad_norm": 3.130396842956543,
"learning_rate": 1.0178803155966158e-06,
"loss": 0.34598445892333984,
"step": 3554
},
{
"epoch": 2.691900075700227,
"grad_norm": 0.3503313660621643,
"learning_rate": 1.0163874887348873e-06,
"loss": 0.09671512991189957,
"step": 3556
},
{
"epoch": 2.6934140802422406,
"grad_norm": 6.0429792404174805,
"learning_rate": 1.0148956548726668e-06,
"loss": 0.5806812644004822,
"step": 3558
},
{
"epoch": 2.6949280847842543,
"grad_norm": 1.3720959424972534,
"learning_rate": 1.0134048163479599e-06,
"loss": 1.0482804775238037,
"step": 3560
},
{
"epoch": 2.696442089326268,
"grad_norm": 1.4913266897201538,
"learning_rate": 1.0119149754972132e-06,
"loss": 0.5744074583053589,
"step": 3562
},
{
"epoch": 2.6979560938682816,
"grad_norm": 2.419517755508423,
"learning_rate": 1.0104261346553096e-06,
"loss": 0.5351262092590332,
"step": 3564
},
{
"epoch": 2.6994700984102953,
"grad_norm": 2.7999117374420166,
"learning_rate": 1.0089382961555663e-06,
"loss": 0.6164992451667786,
"step": 3566
},
{
"epoch": 2.700984102952309,
"grad_norm": 1.8278381824493408,
"learning_rate": 1.0074514623297277e-06,
"loss": 0.5365289449691772,
"step": 3568
},
{
"epoch": 2.7024981074943226,
"grad_norm": 1.0970771312713623,
"learning_rate": 1.005965635507964e-06,
"loss": 0.36719387769699097,
"step": 3570
},
{
"epoch": 2.7040121120363363,
"grad_norm": 2.4264323711395264,
"learning_rate": 1.0044808180188685e-06,
"loss": 0.24815315008163452,
"step": 3572
},
{
"epoch": 2.7055261165783495,
"grad_norm": 0.4873034656047821,
"learning_rate": 1.0029970121894516e-06,
"loss": 0.14031077921390533,
"step": 3574
},
{
"epoch": 2.7070401211203636,
"grad_norm": 0.6915985345840454,
"learning_rate": 1.0015142203451384e-06,
"loss": 0.2932851314544678,
"step": 3576
},
{
"epoch": 2.708554125662377,
"grad_norm": 2.271310806274414,
"learning_rate": 1.000032444809764e-06,
"loss": 0.4614425599575043,
"step": 3578
},
{
"epoch": 2.7100681302043905,
"grad_norm": 1.5097922086715698,
"learning_rate": 9.985516879055733e-07,
"loss": 0.9115333557128906,
"step": 3580
},
{
"epoch": 2.711582134746404,
"grad_norm": 1.59455406665802,
"learning_rate": 9.970719519532123e-07,
"loss": 0.5052527785301208,
"step": 3582
},
{
"epoch": 2.713096139288418,
"grad_norm": 2.1244618892669678,
"learning_rate": 9.955932392717273e-07,
"loss": 0.8183184266090393,
"step": 3584
},
{
"epoch": 2.7146101438304315,
"grad_norm": 5.1214399337768555,
"learning_rate": 9.941155521785622e-07,
"loss": 0.538994550704956,
"step": 3586
},
{
"epoch": 2.716124148372445,
"grad_norm": 4.378061294555664,
"learning_rate": 9.926388929895523e-07,
"loss": 0.2428453415632248,
"step": 3588
},
{
"epoch": 2.717638152914459,
"grad_norm": 7.363132476806641,
"learning_rate": 9.91163264018923e-07,
"loss": 0.06354464590549469,
"step": 3590
},
{
"epoch": 2.719152157456472,
"grad_norm": 3.2854292392730713,
"learning_rate": 9.89688667579284e-07,
"loss": 0.617462694644928,
"step": 3592
},
{
"epoch": 2.720666161998486,
"grad_norm": 1.5972332954406738,
"learning_rate": 9.882151059816286e-07,
"loss": 0.4859785735607147,
"step": 3594
},
{
"epoch": 2.7221801665404994,
"grad_norm": 2.845470905303955,
"learning_rate": 9.867425815353263e-07,
"loss": 0.5740007162094116,
"step": 3596
},
{
"epoch": 2.723694171082513,
"grad_norm": 2.0157742500305176,
"learning_rate": 9.852710965481219e-07,
"loss": 0.06600300222635269,
"step": 3598
},
{
"epoch": 2.7252081756245268,
"grad_norm": 1.7511472702026367,
"learning_rate": 9.83800653326131e-07,
"loss": 0.847884476184845,
"step": 3600
},
{
"epoch": 2.7267221801665404,
"grad_norm": 1.3995939493179321,
"learning_rate": 9.823312541738378e-07,
"loss": 0.8712521195411682,
"step": 3602
},
{
"epoch": 2.728236184708554,
"grad_norm": 1.488347053527832,
"learning_rate": 9.808629013940889e-07,
"loss": 0.5205467343330383,
"step": 3604
},
{
"epoch": 2.729750189250568,
"grad_norm": 1.5527247190475464,
"learning_rate": 9.793955972880904e-07,
"loss": 0.5302475094795227,
"step": 3606
},
{
"epoch": 2.7312641937925815,
"grad_norm": 1.9730198383331299,
"learning_rate": 9.779293441554072e-07,
"loss": 0.05066138133406639,
"step": 3608
},
{
"epoch": 2.732778198334595,
"grad_norm": 1.394031286239624,
"learning_rate": 9.764641442939552e-07,
"loss": 0.5910233855247498,
"step": 3610
},
{
"epoch": 2.734292202876609,
"grad_norm": 7.18306303024292,
"learning_rate": 9.750000000000004e-07,
"loss": 0.7828827500343323,
"step": 3612
},
{
"epoch": 2.735806207418622,
"grad_norm": 1.5465166568756104,
"learning_rate": 9.735369135681535e-07,
"loss": 0.4381512999534607,
"step": 3614
},
{
"epoch": 2.737320211960636,
"grad_norm": 1.4410380125045776,
"learning_rate": 9.720748872913692e-07,
"loss": 0.030536165460944176,
"step": 3616
},
{
"epoch": 2.7388342165026494,
"grad_norm": 5.242462158203125,
"learning_rate": 9.706139234609395e-07,
"loss": 0.5820929408073425,
"step": 3618
},
{
"epoch": 2.740348221044663,
"grad_norm": 2.1642913818359375,
"learning_rate": 9.691540243664913e-07,
"loss": 0.23760852217674255,
"step": 3620
},
{
"epoch": 2.7418622255866767,
"grad_norm": 2.6150195598602295,
"learning_rate": 9.67695192295982e-07,
"loss": 0.8972626328468323,
"step": 3622
},
{
"epoch": 2.7433762301286904,
"grad_norm": 0.4961766302585602,
"learning_rate": 9.662374295356995e-07,
"loss": 0.48937559127807617,
"step": 3624
},
{
"epoch": 2.744890234670704,
"grad_norm": 1.2891615629196167,
"learning_rate": 9.647807383702534e-07,
"loss": 0.08011382818222046,
"step": 3626
},
{
"epoch": 2.7464042392127177,
"grad_norm": 1.4327223300933838,
"learning_rate": 9.63325121082574e-07,
"loss": 0.9761390686035156,
"step": 3628
},
{
"epoch": 2.7479182437547314,
"grad_norm": 0.7498922348022461,
"learning_rate": 9.618705799539105e-07,
"loss": 0.11114507913589478,
"step": 3630
},
{
"epoch": 2.749432248296745,
"grad_norm": 2.269970417022705,
"learning_rate": 9.604171172638233e-07,
"loss": 0.546760082244873,
"step": 3632
},
{
"epoch": 2.7509462528387587,
"grad_norm": 2.625152111053467,
"learning_rate": 9.589647352901837e-07,
"loss": 0.15670128166675568,
"step": 3634
},
{
"epoch": 2.752460257380772,
"grad_norm": 2.0255126953125,
"learning_rate": 9.575134363091702e-07,
"loss": 0.6736434698104858,
"step": 3636
},
{
"epoch": 2.7539742619227856,
"grad_norm": 4.662033557891846,
"learning_rate": 9.560632225952626e-07,
"loss": 0.15046308934688568,
"step": 3638
},
{
"epoch": 2.7554882664647993,
"grad_norm": 1.6699665784835815,
"learning_rate": 9.546140964212397e-07,
"loss": 1.004642128944397,
"step": 3640
},
{
"epoch": 2.757002271006813,
"grad_norm": 5.824310779571533,
"learning_rate": 9.531660600581774e-07,
"loss": 0.22400666773319244,
"step": 3642
},
{
"epoch": 2.7585162755488266,
"grad_norm": 1.661800742149353,
"learning_rate": 9.517191157754421e-07,
"loss": 0.04683089628815651,
"step": 3644
},
{
"epoch": 2.7600302800908403,
"grad_norm": 2.4442532062530518,
"learning_rate": 9.502732658406903e-07,
"loss": 0.670370876789093,
"step": 3646
},
{
"epoch": 2.761544284632854,
"grad_norm": 3.1336352825164795,
"learning_rate": 9.488285125198622e-07,
"loss": 0.1903330534696579,
"step": 3648
},
{
"epoch": 2.7630582891748676,
"grad_norm": 2.6961746215820312,
"learning_rate": 9.473848580771794e-07,
"loss": 0.9148120880126953,
"step": 3650
},
{
"epoch": 2.7645722937168813,
"grad_norm": 1.6039396524429321,
"learning_rate": 9.459423047751417e-07,
"loss": 0.9865699410438538,
"step": 3652
},
{
"epoch": 2.7660862982588945,
"grad_norm": 2.1121411323547363,
"learning_rate": 9.445008548745238e-07,
"loss": 0.245061993598938,
"step": 3654
},
{
"epoch": 2.7676003028009086,
"grad_norm": 1.7591969966888428,
"learning_rate": 9.430605106343696e-07,
"loss": 0.0845777839422226,
"step": 3656
},
{
"epoch": 2.769114307342922,
"grad_norm": 4.009654521942139,
"learning_rate": 9.416212743119911e-07,
"loss": 0.519393265247345,
"step": 3658
},
{
"epoch": 2.7706283118849355,
"grad_norm": 1.430734395980835,
"learning_rate": 9.401831481629649e-07,
"loss": 0.48087310791015625,
"step": 3660
},
{
"epoch": 2.772142316426949,
"grad_norm": 1.65854811668396,
"learning_rate": 9.387461344411263e-07,
"loss": 0.9938499331474304,
"step": 3662
},
{
"epoch": 2.773656320968963,
"grad_norm": 0.5344567894935608,
"learning_rate": 9.373102353985668e-07,
"loss": 0.48683303594589233,
"step": 3664
},
{
"epoch": 2.7751703255109765,
"grad_norm": 1.240643858909607,
"learning_rate": 9.358754532856334e-07,
"loss": 0.4459698796272278,
"step": 3666
},
{
"epoch": 2.77668433005299,
"grad_norm": 1.141143798828125,
"learning_rate": 9.344417903509201e-07,
"loss": 0.5696487426757812,
"step": 3668
},
{
"epoch": 2.778198334595004,
"grad_norm": 2.9075422286987305,
"learning_rate": 9.330092488412687e-07,
"loss": 0.4729193150997162,
"step": 3670
},
{
"epoch": 2.7797123391370175,
"grad_norm": 0.8197333216667175,
"learning_rate": 9.315778310017616e-07,
"loss": 0.5160009860992432,
"step": 3672
},
{
"epoch": 2.781226343679031,
"grad_norm": 1.4015450477600098,
"learning_rate": 9.301475390757222e-07,
"loss": 0.5251963138580322,
"step": 3674
},
{
"epoch": 2.7827403482210444,
"grad_norm": 2.8256664276123047,
"learning_rate": 9.287183753047082e-07,
"loss": 0.1814594566822052,
"step": 3676
},
{
"epoch": 2.7842543527630585,
"grad_norm": 4.373061180114746,
"learning_rate": 9.272903419285096e-07,
"loss": 0.11841978132724762,
"step": 3678
},
{
"epoch": 2.7857683573050718,
"grad_norm": 2.711310386657715,
"learning_rate": 9.258634411851445e-07,
"loss": 0.9179194569587708,
"step": 3680
},
{
"epoch": 2.7872823618470854,
"grad_norm": 3.6902360916137695,
"learning_rate": 9.244376753108567e-07,
"loss": 0.14111346006393433,
"step": 3682
},
{
"epoch": 2.788796366389099,
"grad_norm": 5.682900905609131,
"learning_rate": 9.230130465401107e-07,
"loss": 0.0874456986784935,
"step": 3684
},
{
"epoch": 2.790310370931113,
"grad_norm": 3.762624740600586,
"learning_rate": 9.215895571055886e-07,
"loss": 0.22180086374282837,
"step": 3686
},
{
"epoch": 2.7918243754731265,
"grad_norm": 1.6312108039855957,
"learning_rate": 9.201672092381885e-07,
"loss": 0.6041345000267029,
"step": 3688
},
{
"epoch": 2.79333838001514,
"grad_norm": 3.2012195587158203,
"learning_rate": 9.187460051670173e-07,
"loss": 0.5570380091667175,
"step": 3690
},
{
"epoch": 2.794852384557154,
"grad_norm": 1.1236485242843628,
"learning_rate": 9.173259471193918e-07,
"loss": 0.5052518844604492,
"step": 3692
},
{
"epoch": 2.7963663890991675,
"grad_norm": 2.8305928707122803,
"learning_rate": 9.159070373208301e-07,
"loss": 0.19987015426158905,
"step": 3694
},
{
"epoch": 2.797880393641181,
"grad_norm": 7.895492076873779,
"learning_rate": 9.144892779950532e-07,
"loss": 0.737027645111084,
"step": 3696
},
{
"epoch": 2.7993943981831944,
"grad_norm": 3.232949733734131,
"learning_rate": 9.130726713639774e-07,
"loss": 0.09915422648191452,
"step": 3698
},
{
"epoch": 2.800908402725208,
"grad_norm": 1.2412161827087402,
"learning_rate": 9.116572196477129e-07,
"loss": 0.5296503305435181,
"step": 3700
},
{
"epoch": 2.8024224072672217,
"grad_norm": 2.9720041751861572,
"learning_rate": 9.102429250645598e-07,
"loss": 0.9459198713302612,
"step": 3702
},
{
"epoch": 2.8039364118092354,
"grad_norm": 1.1881636381149292,
"learning_rate": 9.088297898310059e-07,
"loss": 0.7172422409057617,
"step": 3704
},
{
"epoch": 2.805450416351249,
"grad_norm": 4.256730079650879,
"learning_rate": 9.074178161617206e-07,
"loss": 0.339231014251709,
"step": 3706
},
{
"epoch": 2.8069644208932627,
"grad_norm": 7.478480339050293,
"learning_rate": 9.06007006269553e-07,
"loss": 0.5643231272697449,
"step": 3708
},
{
"epoch": 2.8084784254352764,
"grad_norm": 3.0805113315582275,
"learning_rate": 9.045973623655298e-07,
"loss": 0.45885780453681946,
"step": 3710
},
{
"epoch": 2.80999242997729,
"grad_norm": 4.949220657348633,
"learning_rate": 9.031888866588486e-07,
"loss": 0.5311449766159058,
"step": 3712
},
{
"epoch": 2.8115064345193037,
"grad_norm": 1.7975176572799683,
"learning_rate": 9.017815813568773e-07,
"loss": 0.5856797099113464,
"step": 3714
},
{
"epoch": 2.813020439061317,
"grad_norm": 2.2618489265441895,
"learning_rate": 9.003754486651483e-07,
"loss": 0.15607936680316925,
"step": 3716
},
{
"epoch": 2.814534443603331,
"grad_norm": 2.57824969291687,
"learning_rate": 8.989704907873585e-07,
"loss": 0.5110048055648804,
"step": 3718
},
{
"epoch": 2.8160484481453443,
"grad_norm": 1.3291974067687988,
"learning_rate": 8.975667099253615e-07,
"loss": 0.0944562554359436,
"step": 3720
},
{
"epoch": 2.817562452687358,
"grad_norm": 2.122201442718506,
"learning_rate": 8.961641082791665e-07,
"loss": 0.029286984354257584,
"step": 3722
},
{
"epoch": 2.8190764572293716,
"grad_norm": 1.4434117078781128,
"learning_rate": 8.947626880469365e-07,
"loss": 0.5472773909568787,
"step": 3724
},
{
"epoch": 2.8205904617713853,
"grad_norm": 1.7789900302886963,
"learning_rate": 8.933624514249809e-07,
"loss": 0.6741805076599121,
"step": 3726
},
{
"epoch": 2.822104466313399,
"grad_norm": 1.073285698890686,
"learning_rate": 8.919634006077551e-07,
"loss": 0.43643903732299805,
"step": 3728
},
{
"epoch": 2.8236184708554126,
"grad_norm": 7.309019565582275,
"learning_rate": 8.905655377878552e-07,
"loss": 0.7717113494873047,
"step": 3730
},
{
"epoch": 2.8251324753974263,
"grad_norm": 1.233903169631958,
"learning_rate": 8.891688651560177e-07,
"loss": 0.032879047095775604,
"step": 3732
},
{
"epoch": 2.82664647993944,
"grad_norm": 1.6785112619400024,
"learning_rate": 8.87773384901111e-07,
"loss": 0.9319685101509094,
"step": 3734
},
{
"epoch": 2.8281604844814536,
"grad_norm": 30.342561721801758,
"learning_rate": 8.863790992101359e-07,
"loss": 0.5449196100234985,
"step": 3736
},
{
"epoch": 2.829674489023467,
"grad_norm": 3.7048192024230957,
"learning_rate": 8.849860102682226e-07,
"loss": 0.7348836064338684,
"step": 3738
},
{
"epoch": 2.831188493565481,
"grad_norm": 2.610330104827881,
"learning_rate": 8.835941202586237e-07,
"loss": 0.06106247752904892,
"step": 3740
},
{
"epoch": 2.832702498107494,
"grad_norm": 3.5520286560058594,
"learning_rate": 8.822034313627131e-07,
"loss": 1.0147613286972046,
"step": 3742
},
{
"epoch": 2.834216502649508,
"grad_norm": 5.207727909088135,
"learning_rate": 8.808139457599839e-07,
"loss": 0.5845840573310852,
"step": 3744
},
{
"epoch": 2.8357305071915215,
"grad_norm": 2.8798627853393555,
"learning_rate": 8.794256656280411e-07,
"loss": 0.520442008972168,
"step": 3746
},
{
"epoch": 2.837244511733535,
"grad_norm": 2.0684001445770264,
"learning_rate": 8.780385931426028e-07,
"loss": 0.1020556092262268,
"step": 3748
},
{
"epoch": 2.838758516275549,
"grad_norm": 0.7989282608032227,
"learning_rate": 8.766527304774929e-07,
"loss": 0.18852849304676056,
"step": 3750
},
{
"epoch": 2.8402725208175625,
"grad_norm": 2.2457497119903564,
"learning_rate": 8.752680798046388e-07,
"loss": 0.13739685714244843,
"step": 3752
},
{
"epoch": 2.841786525359576,
"grad_norm": 1.6464593410491943,
"learning_rate": 8.738846432940708e-07,
"loss": 0.4965594410896301,
"step": 3754
},
{
"epoch": 2.8433005299015894,
"grad_norm": 3.5233702659606934,
"learning_rate": 8.725024231139141e-07,
"loss": 0.5871769785881042,
"step": 3756
},
{
"epoch": 2.8448145344436035,
"grad_norm": 0.7635524272918701,
"learning_rate": 8.711214214303882e-07,
"loss": 0.5196180939674377,
"step": 3758
},
{
"epoch": 2.8463285389856168,
"grad_norm": 1.6084498167037964,
"learning_rate": 8.69741640407804e-07,
"loss": 0.18574142456054688,
"step": 3760
},
{
"epoch": 2.8478425435276304,
"grad_norm": 0.49115055799484253,
"learning_rate": 8.683630822085586e-07,
"loss": 0.19083711504936218,
"step": 3762
},
{
"epoch": 2.849356548069644,
"grad_norm": 1.3124921321868896,
"learning_rate": 8.669857489931323e-07,
"loss": 0.43203872442245483,
"step": 3764
},
{
"epoch": 2.850870552611658,
"grad_norm": 1.9938055276870728,
"learning_rate": 8.656096429200857e-07,
"loss": 0.05717456340789795,
"step": 3766
},
{
"epoch": 2.8523845571536715,
"grad_norm": 15.529302597045898,
"learning_rate": 8.642347661460574e-07,
"loss": 0.0767684355378151,
"step": 3768
},
{
"epoch": 2.853898561695685,
"grad_norm": 0.2853293716907501,
"learning_rate": 8.628611208257582e-07,
"loss": 0.4549844264984131,
"step": 3770
},
{
"epoch": 2.855412566237699,
"grad_norm": 1.8048611879348755,
"learning_rate": 8.614887091119692e-07,
"loss": 0.49801960587501526,
"step": 3772
},
{
"epoch": 2.8569265707797125,
"grad_norm": 3.1209964752197266,
"learning_rate": 8.60117533155538e-07,
"loss": 0.07470957934856415,
"step": 3774
},
{
"epoch": 2.858440575321726,
"grad_norm": 3.946650266647339,
"learning_rate": 8.587475951053769e-07,
"loss": 0.4709521234035492,
"step": 3776
},
{
"epoch": 2.8599545798637394,
"grad_norm": 1.5060186386108398,
"learning_rate": 8.573788971084563e-07,
"loss": 0.2602544128894806,
"step": 3778
},
{
"epoch": 2.8614685844057535,
"grad_norm": 2.0307724475860596,
"learning_rate": 8.560114413098036e-07,
"loss": 0.7113937139511108,
"step": 3780
},
{
"epoch": 2.8629825889477667,
"grad_norm": 2.940063714981079,
"learning_rate": 8.54645229852501e-07,
"loss": 0.10309739410877228,
"step": 3782
},
{
"epoch": 2.8644965934897804,
"grad_norm": 1.7410873174667358,
"learning_rate": 8.532802648776786e-07,
"loss": 0.5519030690193176,
"step": 3784
},
{
"epoch": 2.866010598031794,
"grad_norm": 2.5878376960754395,
"learning_rate": 8.519165485245139e-07,
"loss": 0.1054067388176918,
"step": 3786
},
{
"epoch": 2.8675246025738077,
"grad_norm": 1.256569266319275,
"learning_rate": 8.505540829302267e-07,
"loss": 0.5424407124519348,
"step": 3788
},
{
"epoch": 2.8690386071158214,
"grad_norm": 0.8407960534095764,
"learning_rate": 8.491928702300788e-07,
"loss": 0.4280465245246887,
"step": 3790
},
{
"epoch": 2.870552611657835,
"grad_norm": 0.8045430183410645,
"learning_rate": 8.478329125573654e-07,
"loss": 0.03474471718072891,
"step": 3792
},
{
"epoch": 2.8720666161998487,
"grad_norm": 2.285611629486084,
"learning_rate": 8.464742120434181e-07,
"loss": 0.07297685742378235,
"step": 3794
},
{
"epoch": 2.8735806207418624,
"grad_norm": 1.3411763906478882,
"learning_rate": 8.451167708175949e-07,
"loss": 0.18974605202674866,
"step": 3796
},
{
"epoch": 2.875094625283876,
"grad_norm": 1.9717655181884766,
"learning_rate": 8.437605910072835e-07,
"loss": 0.49169692397117615,
"step": 3798
},
{
"epoch": 2.8766086298258893,
"grad_norm": 1.0608391761779785,
"learning_rate": 8.424056747378924e-07,
"loss": 0.46762970089912415,
"step": 3800
},
{
"epoch": 2.878122634367903,
"grad_norm": 2.073852777481079,
"learning_rate": 8.410520241328499e-07,
"loss": 0.5141778588294983,
"step": 3802
},
{
"epoch": 2.8796366389099166,
"grad_norm": 2.4379963874816895,
"learning_rate": 8.396996413136029e-07,
"loss": 0.06461524963378906,
"step": 3804
},
{
"epoch": 2.8811506434519303,
"grad_norm": 1.331201195716858,
"learning_rate": 8.383485283996091e-07,
"loss": 0.5797151327133179,
"step": 3806
},
{
"epoch": 2.882664647993944,
"grad_norm": 2.5884780883789062,
"learning_rate": 8.369986875083369e-07,
"loss": 0.48049142956733704,
"step": 3808
},
{
"epoch": 2.8841786525359576,
"grad_norm": 1.6268047094345093,
"learning_rate": 8.356501207552611e-07,
"loss": 0.5873973965644836,
"step": 3810
},
{
"epoch": 2.8856926570779713,
"grad_norm": 2.7758703231811523,
"learning_rate": 8.3430283025386e-07,
"loss": 0.5925127863883972,
"step": 3812
},
{
"epoch": 2.887206661619985,
"grad_norm": 21.12485694885254,
"learning_rate": 8.329568181156116e-07,
"loss": 0.981511652469635,
"step": 3814
},
{
"epoch": 2.8887206661619986,
"grad_norm": 1.62020742893219,
"learning_rate": 8.316120864499896e-07,
"loss": 0.2684553861618042,
"step": 3816
},
{
"epoch": 2.890234670704012,
"grad_norm": 0.5304837226867676,
"learning_rate": 8.302686373644626e-07,
"loss": 0.4767705798149109,
"step": 3818
},
{
"epoch": 2.891748675246026,
"grad_norm": 1.8308970928192139,
"learning_rate": 8.289264729644878e-07,
"loss": 0.5119712352752686,
"step": 3820
},
{
"epoch": 2.893262679788039,
"grad_norm": 4.053905487060547,
"learning_rate": 8.275855953535097e-07,
"loss": 0.5398560762405396,
"step": 3822
},
{
"epoch": 2.894776684330053,
"grad_norm": 1.6479356288909912,
"learning_rate": 8.262460066329549e-07,
"loss": 0.46708399057388306,
"step": 3824
},
{
"epoch": 2.8962906888720665,
"grad_norm": 0.3225676417350769,
"learning_rate": 8.249077089022323e-07,
"loss": 0.5151532888412476,
"step": 3826
},
{
"epoch": 2.89780469341408,
"grad_norm": 1.7271003723144531,
"learning_rate": 8.235707042587258e-07,
"loss": 1.0721513032913208,
"step": 3828
},
{
"epoch": 2.899318697956094,
"grad_norm": 3.939084768295288,
"learning_rate": 8.222349947977929e-07,
"loss": 0.13510605692863464,
"step": 3830
},
{
"epoch": 2.9008327024981075,
"grad_norm": 1.4582927227020264,
"learning_rate": 8.209005826127616e-07,
"loss": 0.5551086664199829,
"step": 3832
},
{
"epoch": 2.902346707040121,
"grad_norm": 1.3123462200164795,
"learning_rate": 8.195674697949277e-07,
"loss": 0.7564160227775574,
"step": 3834
},
{
"epoch": 2.903860711582135,
"grad_norm": 3.6382358074188232,
"learning_rate": 8.182356584335491e-07,
"loss": 0.11954665184020996,
"step": 3836
},
{
"epoch": 2.9053747161241485,
"grad_norm": 1.0889722108840942,
"learning_rate": 8.169051506158443e-07,
"loss": 1.0122618675231934,
"step": 3838
},
{
"epoch": 2.9068887206661618,
"grad_norm": 1.8109678030014038,
"learning_rate": 8.155759484269905e-07,
"loss": 0.770567774772644,
"step": 3840
},
{
"epoch": 2.908402725208176,
"grad_norm": 1.7070538997650146,
"learning_rate": 8.142480539501167e-07,
"loss": 0.5919016599655151,
"step": 3842
},
{
"epoch": 2.909916729750189,
"grad_norm": 2.2287955284118652,
"learning_rate": 8.129214692663032e-07,
"loss": 0.6235976219177246,
"step": 3844
},
{
"epoch": 2.911430734292203,
"grad_norm": 1.6173175573349,
"learning_rate": 8.115961964545783e-07,
"loss": 0.08541975915431976,
"step": 3846
},
{
"epoch": 2.9129447388342165,
"grad_norm": 1.2753394842147827,
"learning_rate": 8.10272237591913e-07,
"loss": 0.02214045450091362,
"step": 3848
},
{
"epoch": 2.91445874337623,
"grad_norm": 1.2947241067886353,
"learning_rate": 8.089495947532204e-07,
"loss": 0.6888188719749451,
"step": 3850
},
{
"epoch": 2.915972747918244,
"grad_norm": 0.6611459255218506,
"learning_rate": 8.0762827001135e-07,
"loss": 0.14420601725578308,
"step": 3852
},
{
"epoch": 2.9174867524602575,
"grad_norm": 1.7915887832641602,
"learning_rate": 8.063082654370859e-07,
"loss": 0.04857180640101433,
"step": 3854
},
{
"epoch": 2.919000757002271,
"grad_norm": 1.6762808561325073,
"learning_rate": 8.049895830991442e-07,
"loss": 0.03882667049765587,
"step": 3856
},
{
"epoch": 2.920514761544285,
"grad_norm": 1.8337093591690063,
"learning_rate": 8.036722250641675e-07,
"loss": 0.5603622794151306,
"step": 3858
},
{
"epoch": 2.9220287660862985,
"grad_norm": 2.0373966693878174,
"learning_rate": 8.023561933967231e-07,
"loss": 0.021611543372273445,
"step": 3860
},
{
"epoch": 2.9235427706283117,
"grad_norm": 1.4119657278060913,
"learning_rate": 8.010414901593006e-07,
"loss": 0.028445757925510406,
"step": 3862
},
{
"epoch": 2.9250567751703254,
"grad_norm": 1.7821069955825806,
"learning_rate": 7.997281174123065e-07,
"loss": 0.5556588172912598,
"step": 3864
},
{
"epoch": 2.926570779712339,
"grad_norm": 3.5669658184051514,
"learning_rate": 7.98416077214063e-07,
"loss": 0.05701752379536629,
"step": 3866
},
{
"epoch": 2.9280847842543527,
"grad_norm": 4.570811748504639,
"learning_rate": 7.971053716208031e-07,
"loss": 0.033300720155239105,
"step": 3868
},
{
"epoch": 2.9295987887963664,
"grad_norm": 2.6463215351104736,
"learning_rate": 7.957960026866695e-07,
"loss": 0.5267956256866455,
"step": 3870
},
{
"epoch": 2.93111279333838,
"grad_norm": 1.602906584739685,
"learning_rate": 7.944879724637089e-07,
"loss": 1.021190881729126,
"step": 3872
},
{
"epoch": 2.9326267978803937,
"grad_norm": 1.8349531888961792,
"learning_rate": 7.931812830018696e-07,
"loss": 0.5174588561058044,
"step": 3874
},
{
"epoch": 2.9341408024224074,
"grad_norm": 1.067012906074524,
"learning_rate": 7.918759363490007e-07,
"loss": 0.4694302976131439,
"step": 3876
},
{
"epoch": 2.935654806964421,
"grad_norm": 7.647733688354492,
"learning_rate": 7.905719345508448e-07,
"loss": 0.5283836126327515,
"step": 3878
},
{
"epoch": 2.9371688115064343,
"grad_norm": 2.6443963050842285,
"learning_rate": 7.89269279651038e-07,
"loss": 0.6997630000114441,
"step": 3880
},
{
"epoch": 2.9386828160484484,
"grad_norm": 2.3616750240325928,
"learning_rate": 7.879679736911043e-07,
"loss": 0.4031399190425873,
"step": 3882
},
{
"epoch": 2.9401968205904616,
"grad_norm": 1.977614164352417,
"learning_rate": 7.866680187104554e-07,
"loss": 0.6264676451683044,
"step": 3884
},
{
"epoch": 2.9417108251324753,
"grad_norm": 1.331671953201294,
"learning_rate": 7.853694167463845e-07,
"loss": 0.5368511080741882,
"step": 3886
},
{
"epoch": 2.943224829674489,
"grad_norm": 1.7582005262374878,
"learning_rate": 7.840721698340645e-07,
"loss": 0.4475153088569641,
"step": 3888
},
{
"epoch": 2.9447388342165026,
"grad_norm": 2.403273344039917,
"learning_rate": 7.827762800065447e-07,
"loss": 0.05570833012461662,
"step": 3890
},
{
"epoch": 2.9462528387585163,
"grad_norm": 0.9218311905860901,
"learning_rate": 7.814817492947481e-07,
"loss": 0.10007733106613159,
"step": 3892
},
{
"epoch": 2.94776684330053,
"grad_norm": 2.469407081604004,
"learning_rate": 7.801885797274668e-07,
"loss": 0.5435525178909302,
"step": 3894
},
{
"epoch": 2.9492808478425436,
"grad_norm": 2.3963382244110107,
"learning_rate": 7.788967733313607e-07,
"loss": 0.5850597620010376,
"step": 3896
},
{
"epoch": 2.9507948523845573,
"grad_norm": 1.2060469388961792,
"learning_rate": 7.776063321309522e-07,
"loss": 0.004651893395930529,
"step": 3898
},
{
"epoch": 2.952308856926571,
"grad_norm": 3.9787027835845947,
"learning_rate": 7.763172581486256e-07,
"loss": 0.04053030163049698,
"step": 3900
},
{
"epoch": 2.953822861468584,
"grad_norm": 1.6185312271118164,
"learning_rate": 7.750295534046214e-07,
"loss": 1.1225244998931885,
"step": 3902
},
{
"epoch": 2.9553368660105983,
"grad_norm": 1.8362376689910889,
"learning_rate": 7.737432199170336e-07,
"loss": 0.07839284092187881,
"step": 3904
},
{
"epoch": 2.9568508705526115,
"grad_norm": 2.360868215560913,
"learning_rate": 7.724582597018097e-07,
"loss": 0.5056535005569458,
"step": 3906
},
{
"epoch": 2.958364875094625,
"grad_norm": 1.8739367723464966,
"learning_rate": 7.711746747727421e-07,
"loss": 0.5829885601997375,
"step": 3908
},
{
"epoch": 2.959878879636639,
"grad_norm": 1.7606217861175537,
"learning_rate": 7.698924671414689e-07,
"loss": 0.8447286486625671,
"step": 3910
},
{
"epoch": 2.9613928841786525,
"grad_norm": 18.142555236816406,
"learning_rate": 7.686116388174711e-07,
"loss": 0.3059283196926117,
"step": 3912
},
{
"epoch": 2.962906888720666,
"grad_norm": 0.9034169316291809,
"learning_rate": 7.67332191808066e-07,
"loss": 0.41916462779045105,
"step": 3914
},
{
"epoch": 2.96442089326268,
"grad_norm": 1.391530156135559,
"learning_rate": 7.660541281184074e-07,
"loss": 0.8247898817062378,
"step": 3916
},
{
"epoch": 2.9659348978046935,
"grad_norm": 2.9683992862701416,
"learning_rate": 7.6477744975148e-07,
"loss": 0.470153272151947,
"step": 3918
},
{
"epoch": 2.9674489023467068,
"grad_norm": 1.7733650207519531,
"learning_rate": 7.63502158708099e-07,
"loss": 0.5974211096763611,
"step": 3920
},
{
"epoch": 2.968962906888721,
"grad_norm": 1.377882480621338,
"learning_rate": 7.62228256986904e-07,
"loss": 0.9456145167350769,
"step": 3922
},
{
"epoch": 2.970476911430734,
"grad_norm": 0.9141403436660767,
"learning_rate": 7.609557465843581e-07,
"loss": 0.440866619348526,
"step": 3924
},
{
"epoch": 2.971990915972748,
"grad_norm": 1.7118134498596191,
"learning_rate": 7.596846294947427e-07,
"loss": 0.21067911386489868,
"step": 3926
},
{
"epoch": 2.9735049205147615,
"grad_norm": 1.9212812185287476,
"learning_rate": 7.584149077101576e-07,
"loss": 0.47009432315826416,
"step": 3928
},
{
"epoch": 2.975018925056775,
"grad_norm": 0.7594667673110962,
"learning_rate": 7.571465832205142e-07,
"loss": 0.1051906868815422,
"step": 3930
},
{
"epoch": 2.976532929598789,
"grad_norm": 3.166438341140747,
"learning_rate": 7.558796580135345e-07,
"loss": 1.0215100049972534,
"step": 3932
},
{
"epoch": 2.9780469341408025,
"grad_norm": 4.067107677459717,
"learning_rate": 7.546141340747478e-07,
"loss": 0.4328348934650421,
"step": 3934
},
{
"epoch": 2.979560938682816,
"grad_norm": 7.947187900543213,
"learning_rate": 7.533500133874874e-07,
"loss": 0.37044578790664673,
"step": 3936
},
{
"epoch": 2.98107494322483,
"grad_norm": 1.890057921409607,
"learning_rate": 7.52087297932887e-07,
"loss": 0.6495625376701355,
"step": 3938
},
{
"epoch": 2.9825889477668435,
"grad_norm": 1.2888870239257812,
"learning_rate": 7.508259896898774e-07,
"loss": 0.9499898552894592,
"step": 3940
},
{
"epoch": 2.9841029523088567,
"grad_norm": 1.0993458032608032,
"learning_rate": 7.495660906351864e-07,
"loss": 0.5022351741790771,
"step": 3942
},
{
"epoch": 2.985616956850871,
"grad_norm": 1.2433116436004639,
"learning_rate": 7.483076027433309e-07,
"loss": 0.5429167151451111,
"step": 3944
},
{
"epoch": 2.987130961392884,
"grad_norm": 1.759813666343689,
"learning_rate": 7.470505279866165e-07,
"loss": 0.5135501623153687,
"step": 3946
},
{
"epoch": 2.9886449659348977,
"grad_norm": 1.7812200784683228,
"learning_rate": 7.457948683351357e-07,
"loss": 0.5178404450416565,
"step": 3948
},
{
"epoch": 2.9901589704769114,
"grad_norm": 1.8572566509246826,
"learning_rate": 7.445406257567613e-07,
"loss": 0.5528455376625061,
"step": 3950
},
{
"epoch": 2.991672975018925,
"grad_norm": 1.6367700099945068,
"learning_rate": 7.432878022171473e-07,
"loss": 0.5564888715744019,
"step": 3952
},
{
"epoch": 2.9931869795609387,
"grad_norm": 1.6826878786087036,
"learning_rate": 7.420363996797214e-07,
"loss": 0.08277818560600281,
"step": 3954
},
{
"epoch": 2.9947009841029524,
"grad_norm": 2.2437868118286133,
"learning_rate": 7.407864201056869e-07,
"loss": 0.4363061785697937,
"step": 3956
},
{
"epoch": 2.996214988644966,
"grad_norm": 8.740562438964844,
"learning_rate": 7.395378654540147e-07,
"loss": 0.7006982564926147,
"step": 3958
},
{
"epoch": 2.9977289931869797,
"grad_norm": 1.6342284679412842,
"learning_rate": 7.382907376814441e-07,
"loss": 0.5745993256568909,
"step": 3960
},
{
"epoch": 2.9992429977289934,
"grad_norm": 1.8306623697280884,
"learning_rate": 7.370450387424767e-07,
"loss": 0.9066737294197083,
"step": 3962
},
{
"epoch": 3.0007570022710066,
"grad_norm": 1.6099592447280884,
"learning_rate": 7.358007705893771e-07,
"loss": 0.56649249792099,
"step": 3964
},
{
"epoch": 3.0022710068130203,
"grad_norm": 0.6058228015899658,
"learning_rate": 7.345579351721656e-07,
"loss": 0.4400520920753479,
"step": 3966
},
{
"epoch": 3.003785011355034,
"grad_norm": 2.340282440185547,
"learning_rate": 7.333165344386171e-07,
"loss": 0.2406800538301468,
"step": 3968
},
{
"epoch": 3.0052990158970476,
"grad_norm": 7.830657005310059,
"learning_rate": 7.3207657033426e-07,
"loss": 0.19384807348251343,
"step": 3970
},
{
"epoch": 3.0068130204390613,
"grad_norm": 0.14254476130008698,
"learning_rate": 7.308380448023691e-07,
"loss": 0.47479867935180664,
"step": 3972
},
{
"epoch": 3.008327024981075,
"grad_norm": 1.5983010530471802,
"learning_rate": 7.296009597839658e-07,
"loss": 0.41168585419654846,
"step": 3974
},
{
"epoch": 3.0098410295230886,
"grad_norm": 1.721829891204834,
"learning_rate": 7.283653172178129e-07,
"loss": 0.4500812888145447,
"step": 3976
},
{
"epoch": 3.0113550340651023,
"grad_norm": 1.6810276508331299,
"learning_rate": 7.271311190404144e-07,
"loss": 0.4039463996887207,
"step": 3978
},
{
"epoch": 3.012869038607116,
"grad_norm": 1.7916243076324463,
"learning_rate": 7.25898367186009e-07,
"loss": 0.9354234337806702,
"step": 3980
},
{
"epoch": 3.0143830431491296,
"grad_norm": 1.3198893070220947,
"learning_rate": 7.246670635865692e-07,
"loss": 0.13221746683120728,
"step": 3982
},
{
"epoch": 3.015897047691143,
"grad_norm": 1.7246809005737305,
"learning_rate": 7.234372101717974e-07,
"loss": 0.8873751759529114,
"step": 3984
},
{
"epoch": 3.0174110522331565,
"grad_norm": 1.7376207113265991,
"learning_rate": 7.222088088691246e-07,
"loss": 0.451565146446228,
"step": 3986
},
{
"epoch": 3.01892505677517,
"grad_norm": 2.4644076824188232,
"learning_rate": 7.209818616037046e-07,
"loss": 0.014454965479671955,
"step": 3988
},
{
"epoch": 3.020439061317184,
"grad_norm": 2.7672579288482666,
"learning_rate": 7.197563702984131e-07,
"loss": 0.4274209439754486,
"step": 3990
},
{
"epoch": 3.0219530658591975,
"grad_norm": 1.235212802886963,
"learning_rate": 7.185323368738442e-07,
"loss": 0.4580550193786621,
"step": 3992
},
{
"epoch": 3.023467070401211,
"grad_norm": 1.6338485479354858,
"learning_rate": 7.173097632483067e-07,
"loss": 0.47135257720947266,
"step": 3994
},
{
"epoch": 3.024981074943225,
"grad_norm": 0.5066173672676086,
"learning_rate": 7.160886513378211e-07,
"loss": 0.47466373443603516,
"step": 3996
},
{
"epoch": 3.0264950794852385,
"grad_norm": 1.0611977577209473,
"learning_rate": 7.148690030561192e-07,
"loss": 0.42254048585891724,
"step": 3998
},
{
"epoch": 3.028009084027252,
"grad_norm": 1.4310609102249146,
"learning_rate": 7.136508203146364e-07,
"loss": 0.42798805236816406,
"step": 4000
},
{
"epoch": 3.029523088569266,
"grad_norm": 0.9165547490119934,
"learning_rate": 7.124341050225133e-07,
"loss": 0.0023502728436142206,
"step": 4002
},
{
"epoch": 3.031037093111279,
"grad_norm": 2.101602077484131,
"learning_rate": 7.112188590865894e-07,
"loss": 0.5950231552124023,
"step": 4004
},
{
"epoch": 3.032551097653293,
"grad_norm": 1.025341272354126,
"learning_rate": 7.10005084411402e-07,
"loss": 0.03122119978070259,
"step": 4006
},
{
"epoch": 3.0340651021953065,
"grad_norm": 1.6090428829193115,
"learning_rate": 7.087927828991828e-07,
"loss": 0.32853564620018005,
"step": 4008
},
{
"epoch": 3.03557910673732,
"grad_norm": 1.611303687095642,
"learning_rate": 7.075819564498545e-07,
"loss": 0.3890955448150635,
"step": 4010
},
{
"epoch": 3.037093111279334,
"grad_norm": 3.2842607498168945,
"learning_rate": 7.063726069610276e-07,
"loss": 0.42601776123046875,
"step": 4012
},
{
"epoch": 3.0386071158213475,
"grad_norm": 4.464943885803223,
"learning_rate": 7.05164736327999e-07,
"loss": 0.1855679452419281,
"step": 4014
},
{
"epoch": 3.040121120363361,
"grad_norm": 2.2660470008850098,
"learning_rate": 7.039583464437473e-07,
"loss": 0.535140335559845,
"step": 4016
},
{
"epoch": 3.041635124905375,
"grad_norm": 1.3778276443481445,
"learning_rate": 7.027534391989301e-07,
"loss": 0.3070753216743469,
"step": 4018
},
{
"epoch": 3.0431491294473885,
"grad_norm": 1.3708959817886353,
"learning_rate": 7.015500164818816e-07,
"loss": 0.011551040224730968,
"step": 4020
},
{
"epoch": 3.044663133989402,
"grad_norm": 1.6248399019241333,
"learning_rate": 7.003480801786104e-07,
"loss": 0.7897078394889832,
"step": 4022
},
{
"epoch": 3.046177138531416,
"grad_norm": 1.192365050315857,
"learning_rate": 6.991476321727945e-07,
"loss": 0.015628105029463768,
"step": 4024
},
{
"epoch": 3.047691143073429,
"grad_norm": 1.2526236772537231,
"learning_rate": 6.979486743457794e-07,
"loss": 0.09048985689878464,
"step": 4026
},
{
"epoch": 3.0492051476154427,
"grad_norm": 3.9266533851623535,
"learning_rate": 6.967512085765763e-07,
"loss": 0.42560815811157227,
"step": 4028
},
{
"epoch": 3.0507191521574564,
"grad_norm": 2.915327548980713,
"learning_rate": 6.955552367418566e-07,
"loss": 0.21876971423625946,
"step": 4030
},
{
"epoch": 3.05223315669947,
"grad_norm": 0.17085859179496765,
"learning_rate": 6.943607607159516e-07,
"loss": 0.4021713435649872,
"step": 4032
},
{
"epoch": 3.0537471612414837,
"grad_norm": 1.7475470304489136,
"learning_rate": 6.93167782370847e-07,
"loss": 0.5051688551902771,
"step": 4034
},
{
"epoch": 3.0552611657834974,
"grad_norm": 2.716975688934326,
"learning_rate": 6.919763035761835e-07,
"loss": 0.4326034188270569,
"step": 4036
},
{
"epoch": 3.056775170325511,
"grad_norm": 1.3036128282546997,
"learning_rate": 6.907863261992494e-07,
"loss": 0.5222240686416626,
"step": 4038
},
{
"epoch": 3.0582891748675247,
"grad_norm": 2.3033268451690674,
"learning_rate": 6.895978521049816e-07,
"loss": 0.0866556465625763,
"step": 4040
},
{
"epoch": 3.0598031794095384,
"grad_norm": 1.632857084274292,
"learning_rate": 6.884108831559594e-07,
"loss": 0.44751986861228943,
"step": 4042
},
{
"epoch": 3.061317183951552,
"grad_norm": 0.7052401900291443,
"learning_rate": 6.872254212124053e-07,
"loss": 0.43911439180374146,
"step": 4044
},
{
"epoch": 3.0628311884935653,
"grad_norm": 2.376359701156616,
"learning_rate": 6.860414681321787e-07,
"loss": 0.20868328213691711,
"step": 4046
},
{
"epoch": 3.064345193035579,
"grad_norm": 1.0090512037277222,
"learning_rate": 6.848590257707741e-07,
"loss": 0.45377230644226074,
"step": 4048
},
{
"epoch": 3.0658591975775926,
"grad_norm": 1.676525354385376,
"learning_rate": 6.836780959813194e-07,
"loss": 0.37644314765930176,
"step": 4050
},
{
"epoch": 3.0673732021196063,
"grad_norm": 42.375518798828125,
"learning_rate": 6.82498680614571e-07,
"loss": 0.474353164434433,
"step": 4052
},
{
"epoch": 3.06888720666162,
"grad_norm": 1.7034921646118164,
"learning_rate": 6.81320781518913e-07,
"loss": 0.5057543516159058,
"step": 4054
},
{
"epoch": 3.0704012112036336,
"grad_norm": 1.43311607837677,
"learning_rate": 6.801444005403517e-07,
"loss": 0.005710270255804062,
"step": 4056
},
{
"epoch": 3.0719152157456473,
"grad_norm": 1.3534364700317383,
"learning_rate": 6.789695395225158e-07,
"loss": 0.3858188986778259,
"step": 4058
},
{
"epoch": 3.073429220287661,
"grad_norm": 2.685340404510498,
"learning_rate": 6.777962003066511e-07,
"loss": 0.45474767684936523,
"step": 4060
},
{
"epoch": 3.0749432248296746,
"grad_norm": 1.9002461433410645,
"learning_rate": 6.766243847316176e-07,
"loss": 0.45055925846099854,
"step": 4062
},
{
"epoch": 3.0764572293716883,
"grad_norm": 1.7495943307876587,
"learning_rate": 6.754540946338894e-07,
"loss": 0.28715768456459045,
"step": 4064
},
{
"epoch": 3.0779712339137015,
"grad_norm": 3.2639667987823486,
"learning_rate": 6.742853318475486e-07,
"loss": 0.6930996775627136,
"step": 4066
},
{
"epoch": 3.079485238455715,
"grad_norm": 1.3847533464431763,
"learning_rate": 6.731180982042835e-07,
"loss": 0.4592818319797516,
"step": 4068
},
{
"epoch": 3.080999242997729,
"grad_norm": 6.530165672302246,
"learning_rate": 6.719523955333861e-07,
"loss": 0.031308241188526154,
"step": 4070
},
{
"epoch": 3.0825132475397425,
"grad_norm": 2.292935609817505,
"learning_rate": 6.707882256617498e-07,
"loss": 0.3984447419643402,
"step": 4072
},
{
"epoch": 3.084027252081756,
"grad_norm": 1.8727905750274658,
"learning_rate": 6.696255904138654e-07,
"loss": 0.5818535089492798,
"step": 4074
},
{
"epoch": 3.08554125662377,
"grad_norm": 1.011590838432312,
"learning_rate": 6.684644916118179e-07,
"loss": 0.3981076776981354,
"step": 4076
},
{
"epoch": 3.0870552611657835,
"grad_norm": 2.1105289459228516,
"learning_rate": 6.673049310752851e-07,
"loss": 0.46739429235458374,
"step": 4078
},
{
"epoch": 3.088569265707797,
"grad_norm": 7.887820243835449,
"learning_rate": 6.661469106215343e-07,
"loss": 0.11569129675626755,
"step": 4080
},
{
"epoch": 3.090083270249811,
"grad_norm": 1.2003979682922363,
"learning_rate": 6.649904320654185e-07,
"loss": 0.3800753057003021,
"step": 4082
},
{
"epoch": 3.0915972747918246,
"grad_norm": 0.3290994465351105,
"learning_rate": 6.638354972193742e-07,
"loss": 0.011019574478268623,
"step": 4084
},
{
"epoch": 3.0931112793338382,
"grad_norm": 1.8022518157958984,
"learning_rate": 6.626821078934197e-07,
"loss": 0.5054448246955872,
"step": 4086
},
{
"epoch": 3.0946252838758515,
"grad_norm": 1.4422074556350708,
"learning_rate": 6.615302658951497e-07,
"loss": 0.8614604473114014,
"step": 4088
},
{
"epoch": 3.096139288417865,
"grad_norm": 0.8293083906173706,
"learning_rate": 6.603799730297347e-07,
"loss": 0.0018426375463604927,
"step": 4090
},
{
"epoch": 3.097653292959879,
"grad_norm": 1.566907525062561,
"learning_rate": 6.592312310999173e-07,
"loss": 0.4373411536216736,
"step": 4092
},
{
"epoch": 3.0991672975018925,
"grad_norm": 2.7237191200256348,
"learning_rate": 6.580840419060095e-07,
"loss": 0.02366696111857891,
"step": 4094
},
{
"epoch": 3.100681302043906,
"grad_norm": 1.0543371438980103,
"learning_rate": 6.5693840724589e-07,
"loss": 0.6218542456626892,
"step": 4096
},
{
"epoch": 3.10219530658592,
"grad_norm": 0.6855198740959167,
"learning_rate": 6.557943289150002e-07,
"loss": 0.0873081237077713,
"step": 4098
},
{
"epoch": 3.1037093111279335,
"grad_norm": 0.9665345549583435,
"learning_rate": 6.546518087063444e-07,
"loss": 0.0009544052300043404,
"step": 4100
},
{
"epoch": 3.105223315669947,
"grad_norm": 0.9792335629463196,
"learning_rate": 6.535108484104827e-07,
"loss": 0.08898274600505829,
"step": 4102
},
{
"epoch": 3.106737320211961,
"grad_norm": 2.054558753967285,
"learning_rate": 6.523714498155326e-07,
"loss": 0.04483972489833832,
"step": 4104
},
{
"epoch": 3.1082513247539745,
"grad_norm": 2.7286934852600098,
"learning_rate": 6.512336147071624e-07,
"loss": 0.613974392414093,
"step": 4106
},
{
"epoch": 3.1097653292959877,
"grad_norm": 1.400834560394287,
"learning_rate": 6.500973448685914e-07,
"loss": 0.24737434089183807,
"step": 4108
},
{
"epoch": 3.1112793338380014,
"grad_norm": 0.4912169277667999,
"learning_rate": 6.489626420805851e-07,
"loss": 0.37733766436576843,
"step": 4110
},
{
"epoch": 3.112793338380015,
"grad_norm": 6.200310707092285,
"learning_rate": 6.47829508121453e-07,
"loss": 0.05362119525671005,
"step": 4112
},
{
"epoch": 3.1143073429220287,
"grad_norm": 3.9851672649383545,
"learning_rate": 6.466979447670463e-07,
"loss": 0.1483810842037201,
"step": 4114
},
{
"epoch": 3.1158213474640424,
"grad_norm": 0.546320915222168,
"learning_rate": 6.455679537907548e-07,
"loss": 0.007687057368457317,
"step": 4116
},
{
"epoch": 3.117335352006056,
"grad_norm": 1.32984459400177,
"learning_rate": 6.44439536963504e-07,
"loss": 0.7310623526573181,
"step": 4118
},
{
"epoch": 3.1188493565480697,
"grad_norm": 3.537933826446533,
"learning_rate": 6.433126960537513e-07,
"loss": 0.6782516837120056,
"step": 4120
},
{
"epoch": 3.1203633610900834,
"grad_norm": 1.9288111925125122,
"learning_rate": 6.421874328274865e-07,
"loss": 0.8860838413238525,
"step": 4122
},
{
"epoch": 3.121877365632097,
"grad_norm": 2.1687636375427246,
"learning_rate": 6.410637490482252e-07,
"loss": 0.8524699211120605,
"step": 4124
},
{
"epoch": 3.1233913701741107,
"grad_norm": 2.177737236022949,
"learning_rate": 6.399416464770082e-07,
"loss": 0.19032512605190277,
"step": 4126
},
{
"epoch": 3.124905374716124,
"grad_norm": 2.1528878211975098,
"learning_rate": 6.388211268723975e-07,
"loss": 0.5917698740959167,
"step": 4128
},
{
"epoch": 3.1264193792581376,
"grad_norm": 3.231980085372925,
"learning_rate": 6.377021919904758e-07,
"loss": 0.05690597742795944,
"step": 4130
},
{
"epoch": 3.1279333838001513,
"grad_norm": 1.488774061203003,
"learning_rate": 6.365848435848412e-07,
"loss": 0.4813673794269562,
"step": 4132
},
{
"epoch": 3.129447388342165,
"grad_norm": 2.938225269317627,
"learning_rate": 6.354690834066054e-07,
"loss": 0.9174196124076843,
"step": 4134
},
{
"epoch": 3.1309613928841786,
"grad_norm": 7.670570373535156,
"learning_rate": 6.343549132043909e-07,
"loss": 0.20087343454360962,
"step": 4136
},
{
"epoch": 3.1324753974261923,
"grad_norm": 1.8709648847579956,
"learning_rate": 6.332423347243294e-07,
"loss": 0.9196281433105469,
"step": 4138
},
{
"epoch": 3.133989401968206,
"grad_norm": 1.3427828550338745,
"learning_rate": 6.321313497100571e-07,
"loss": 0.3599630296230316,
"step": 4140
},
{
"epoch": 3.1355034065102196,
"grad_norm": 3.089582920074463,
"learning_rate": 6.310219599027128e-07,
"loss": 0.6179099678993225,
"step": 4142
},
{
"epoch": 3.1370174110522333,
"grad_norm": 2.12060809135437,
"learning_rate": 6.299141670409361e-07,
"loss": 0.5258981585502625,
"step": 4144
},
{
"epoch": 3.138531415594247,
"grad_norm": 1.4477745294570923,
"learning_rate": 6.288079728608635e-07,
"loss": 0.9324302673339844,
"step": 4146
},
{
"epoch": 3.14004542013626,
"grad_norm": 1.8090686798095703,
"learning_rate": 6.277033790961259e-07,
"loss": 0.80943763256073,
"step": 4148
},
{
"epoch": 3.141559424678274,
"grad_norm": 2.9717986583709717,
"learning_rate": 6.266003874778456e-07,
"loss": 0.49069246649742126,
"step": 4150
},
{
"epoch": 3.1430734292202875,
"grad_norm": 4.860805034637451,
"learning_rate": 6.254989997346353e-07,
"loss": 0.03697580099105835,
"step": 4152
},
{
"epoch": 3.144587433762301,
"grad_norm": 1.3705027103424072,
"learning_rate": 6.243992175925925e-07,
"loss": 0.28662988543510437,
"step": 4154
},
{
"epoch": 3.146101438304315,
"grad_norm": 0.6647135019302368,
"learning_rate": 6.233010427753001e-07,
"loss": 0.46875080466270447,
"step": 4156
},
{
"epoch": 3.1476154428463285,
"grad_norm": 0.3081839382648468,
"learning_rate": 6.222044770038203e-07,
"loss": 0.3812503516674042,
"step": 4158
},
{
"epoch": 3.149129447388342,
"grad_norm": 5.093878746032715,
"learning_rate": 6.21109521996695e-07,
"loss": 0.872576892375946,
"step": 4160
},
{
"epoch": 3.150643451930356,
"grad_norm": 1.9643348455429077,
"learning_rate": 6.20016179469941e-07,
"loss": 0.7427912950515747,
"step": 4162
},
{
"epoch": 3.1521574564723696,
"grad_norm": 1.43099045753479,
"learning_rate": 6.189244511370476e-07,
"loss": 0.5047937035560608,
"step": 4164
},
{
"epoch": 3.1536714610143832,
"grad_norm": 2.2820796966552734,
"learning_rate": 6.178343387089756e-07,
"loss": 0.7190396189689636,
"step": 4166
},
{
"epoch": 3.1551854655563965,
"grad_norm": 1.949791669845581,
"learning_rate": 6.16745843894152e-07,
"loss": 0.4036288559436798,
"step": 4168
},
{
"epoch": 3.15669947009841,
"grad_norm": 3.984266757965088,
"learning_rate": 6.156589683984697e-07,
"loss": 0.04631911590695381,
"step": 4170
},
{
"epoch": 3.158213474640424,
"grad_norm": 4.274811744689941,
"learning_rate": 6.145737139252829e-07,
"loss": 0.0733456015586853,
"step": 4172
},
{
"epoch": 3.1597274791824375,
"grad_norm": 5.016620635986328,
"learning_rate": 6.134900821754063e-07,
"loss": 0.06877418607473373,
"step": 4174
},
{
"epoch": 3.161241483724451,
"grad_norm": 2.872525453567505,
"learning_rate": 6.124080748471109e-07,
"loss": 0.08794441819190979,
"step": 4176
},
{
"epoch": 3.162755488266465,
"grad_norm": 1.5351786613464355,
"learning_rate": 6.113276936361215e-07,
"loss": 0.38341233134269714,
"step": 4178
},
{
"epoch": 3.1642694928084785,
"grad_norm": 2.5572965145111084,
"learning_rate": 6.102489402356153e-07,
"loss": 0.9050129652023315,
"step": 4180
},
{
"epoch": 3.165783497350492,
"grad_norm": 0.9266190528869629,
"learning_rate": 6.091718163362182e-07,
"loss": 0.500056803226471,
"step": 4182
},
{
"epoch": 3.167297501892506,
"grad_norm": 2.0204591751098633,
"learning_rate": 6.080963236260016e-07,
"loss": 0.8148636221885681,
"step": 4184
},
{
"epoch": 3.1688115064345195,
"grad_norm": 0.7167491316795349,
"learning_rate": 6.070224637904811e-07,
"loss": 0.4477299451828003,
"step": 4186
},
{
"epoch": 3.170325510976533,
"grad_norm": 2.1239423751831055,
"learning_rate": 6.059502385126138e-07,
"loss": 0.8294087052345276,
"step": 4188
},
{
"epoch": 3.1718395155185464,
"grad_norm": 1.3196355104446411,
"learning_rate": 6.04879649472794e-07,
"loss": 0.5490575432777405,
"step": 4190
},
{
"epoch": 3.17335352006056,
"grad_norm": 1.4261651039123535,
"learning_rate": 6.038106983488523e-07,
"loss": 0.06415217369794846,
"step": 4192
},
{
"epoch": 3.1748675246025737,
"grad_norm": 1.8780325651168823,
"learning_rate": 6.027433868160518e-07,
"loss": 0.05316773056983948,
"step": 4194
},
{
"epoch": 3.1763815291445874,
"grad_norm": 0.09676197916269302,
"learning_rate": 6.016777165470872e-07,
"loss": 0.0006479769363068044,
"step": 4196
},
{
"epoch": 3.177895533686601,
"grad_norm": 4.501723289489746,
"learning_rate": 6.0061368921208e-07,
"loss": 0.030987899750471115,
"step": 4198
},
{
"epoch": 3.1794095382286147,
"grad_norm": 3.0383286476135254,
"learning_rate": 5.99551306478577e-07,
"loss": 0.10911338776350021,
"step": 4200
},
{
"epoch": 3.1809235427706284,
"grad_norm": 1.3916982412338257,
"learning_rate": 5.98490570011548e-07,
"loss": 0.442693829536438,
"step": 4202
},
{
"epoch": 3.182437547312642,
"grad_norm": 1.404009461402893,
"learning_rate": 5.97431481473382e-07,
"loss": 0.3937862515449524,
"step": 4204
},
{
"epoch": 3.1839515518546557,
"grad_norm": 2.481732130050659,
"learning_rate": 5.963740425238867e-07,
"loss": 0.8848980665206909,
"step": 4206
},
{
"epoch": 3.1854655563966694,
"grad_norm": 2.028106927871704,
"learning_rate": 5.953182548202828e-07,
"loss": 0.1351696103811264,
"step": 4208
},
{
"epoch": 3.1869795609386826,
"grad_norm": 2.2008683681488037,
"learning_rate": 5.94264120017205e-07,
"loss": 0.03230151906609535,
"step": 4210
},
{
"epoch": 3.1884935654806963,
"grad_norm": 1.489862322807312,
"learning_rate": 5.932116397666961e-07,
"loss": 0.23553022742271423,
"step": 4212
},
{
"epoch": 3.19000757002271,
"grad_norm": 1.8328267335891724,
"learning_rate": 5.921608157182062e-07,
"loss": 0.8542577028274536,
"step": 4214
},
{
"epoch": 3.1915215745647236,
"grad_norm": 1.3404337167739868,
"learning_rate": 5.911116495185907e-07,
"loss": 0.5208070278167725,
"step": 4216
},
{
"epoch": 3.1930355791067373,
"grad_norm": 3.571223020553589,
"learning_rate": 5.900641428121059e-07,
"loss": 0.48536592721939087,
"step": 4218
},
{
"epoch": 3.194549583648751,
"grad_norm": 0.9041838049888611,
"learning_rate": 5.890182972404074e-07,
"loss": 0.035800751298666,
"step": 4220
},
{
"epoch": 3.1960635881907646,
"grad_norm": 1.311282992362976,
"learning_rate": 5.879741144425475e-07,
"loss": 0.479990690946579,
"step": 4222
},
{
"epoch": 3.1975775927327783,
"grad_norm": 9.55543327331543,
"learning_rate": 5.869315960549734e-07,
"loss": 0.46675267815589905,
"step": 4224
},
{
"epoch": 3.199091597274792,
"grad_norm": 0.6630218029022217,
"learning_rate": 5.858907437115225e-07,
"loss": 0.003185458015650511,
"step": 4226
},
{
"epoch": 3.2006056018168056,
"grad_norm": 3.2958920001983643,
"learning_rate": 5.848515590434222e-07,
"loss": 0.9570237994194031,
"step": 4228
},
{
"epoch": 3.202119606358819,
"grad_norm": 1.514442801475525,
"learning_rate": 5.838140436792856e-07,
"loss": 0.8392640948295593,
"step": 4230
},
{
"epoch": 3.2036336109008325,
"grad_norm": 0.65553218126297,
"learning_rate": 5.827781992451105e-07,
"loss": 0.15988874435424805,
"step": 4232
},
{
"epoch": 3.205147615442846,
"grad_norm": 0.5351942777633667,
"learning_rate": 5.817440273642755e-07,
"loss": 0.016632379963994026,
"step": 4234
},
{
"epoch": 3.20666161998486,
"grad_norm": 5.285929203033447,
"learning_rate": 5.807115296575374e-07,
"loss": 0.8525801301002502,
"step": 4236
},
{
"epoch": 3.2081756245268735,
"grad_norm": 2.787135601043701,
"learning_rate": 5.796807077430305e-07,
"loss": 0.4894711375236511,
"step": 4238
},
{
"epoch": 3.209689629068887,
"grad_norm": 1.6540796756744385,
"learning_rate": 5.78651563236262e-07,
"loss": 0.03871167451143265,
"step": 4240
},
{
"epoch": 3.211203633610901,
"grad_norm": 1.149295687675476,
"learning_rate": 5.776240977501102e-07,
"loss": 0.5460255742073059,
"step": 4242
},
{
"epoch": 3.2127176381529146,
"grad_norm": 1.5972412824630737,
"learning_rate": 5.765983128948217e-07,
"loss": 0.024195147678256035,
"step": 4244
},
{
"epoch": 3.2142316426949282,
"grad_norm": 3.290018320083618,
"learning_rate": 5.75574210278011e-07,
"loss": 0.030811673030257225,
"step": 4246
},
{
"epoch": 3.215745647236942,
"grad_norm": 0.38647592067718506,
"learning_rate": 5.745517915046542e-07,
"loss": 0.005239240825176239,
"step": 4248
},
{
"epoch": 3.2172596517789556,
"grad_norm": 1.172089695930481,
"learning_rate": 5.735310581770891e-07,
"loss": 0.19742485880851746,
"step": 4250
},
{
"epoch": 3.218773656320969,
"grad_norm": 2.1019949913024902,
"learning_rate": 5.725120118950119e-07,
"loss": 0.7881650924682617,
"step": 4252
},
{
"epoch": 3.2202876608629825,
"grad_norm": 3.427130937576294,
"learning_rate": 5.71494654255476e-07,
"loss": 0.8992434740066528,
"step": 4254
},
{
"epoch": 3.221801665404996,
"grad_norm": 1.4113669395446777,
"learning_rate": 5.704789868528865e-07,
"loss": 0.4951695203781128,
"step": 4256
},
{
"epoch": 3.22331566994701,
"grad_norm": 8.905265808105469,
"learning_rate": 5.694650112790013e-07,
"loss": 0.011985964141786098,
"step": 4258
},
{
"epoch": 3.2248296744890235,
"grad_norm": 0.7892521619796753,
"learning_rate": 5.684527291229256e-07,
"loss": 0.01021597534418106,
"step": 4260
},
{
"epoch": 3.226343679031037,
"grad_norm": 0.35012581944465637,
"learning_rate": 5.674421419711116e-07,
"loss": 0.042361292988061905,
"step": 4262
},
{
"epoch": 3.227857683573051,
"grad_norm": 2.7546911239624023,
"learning_rate": 5.664332514073544e-07,
"loss": 0.3553725779056549,
"step": 4264
},
{
"epoch": 3.2293716881150645,
"grad_norm": 4.400173664093018,
"learning_rate": 5.6542605901279e-07,
"loss": 0.503367006778717,
"step": 4266
},
{
"epoch": 3.230885692657078,
"grad_norm": 1.0059622526168823,
"learning_rate": 5.644205663658943e-07,
"loss": 0.03507407009601593,
"step": 4268
},
{
"epoch": 3.2323996971990914,
"grad_norm": 2.090927839279175,
"learning_rate": 5.634167750424781e-07,
"loss": 0.4575502276420593,
"step": 4270
},
{
"epoch": 3.233913701741105,
"grad_norm": 4.920496463775635,
"learning_rate": 5.624146866156859e-07,
"loss": 0.10926653444766998,
"step": 4272
},
{
"epoch": 3.2354277062831187,
"grad_norm": 5.562958717346191,
"learning_rate": 5.614143026559947e-07,
"loss": 0.33701470494270325,
"step": 4274
},
{
"epoch": 3.2369417108251324,
"grad_norm": 1.8385343551635742,
"learning_rate": 5.604156247312088e-07,
"loss": 0.405283659696579,
"step": 4276
},
{
"epoch": 3.238455715367146,
"grad_norm": 1.3062556982040405,
"learning_rate": 5.594186544064595e-07,
"loss": 0.1007096916437149,
"step": 4278
},
{
"epoch": 3.2399697199091597,
"grad_norm": 3.5049281120300293,
"learning_rate": 5.584233932442013e-07,
"loss": 0.14424999058246613,
"step": 4280
},
{
"epoch": 3.2414837244511734,
"grad_norm": 0.5313048958778381,
"learning_rate": 5.574298428042114e-07,
"loss": 0.15579357743263245,
"step": 4282
},
{
"epoch": 3.242997728993187,
"grad_norm": 0.5158863067626953,
"learning_rate": 5.564380046435849e-07,
"loss": 0.48773688077926636,
"step": 4284
},
{
"epoch": 3.2445117335352007,
"grad_norm": 2.8779616355895996,
"learning_rate": 5.554478803167336e-07,
"loss": 0.05835890769958496,
"step": 4286
},
{
"epoch": 3.2460257380772144,
"grad_norm": 1.8299925327301025,
"learning_rate": 5.544594713753831e-07,
"loss": 0.09974260628223419,
"step": 4288
},
{
"epoch": 3.247539742619228,
"grad_norm": 4.498845100402832,
"learning_rate": 5.534727793685715e-07,
"loss": 0.1410599797964096,
"step": 4290
},
{
"epoch": 3.2490537471612413,
"grad_norm": 0.45668092370033264,
"learning_rate": 5.524878058426454e-07,
"loss": 0.0029913324397057295,
"step": 4292
},
{
"epoch": 3.250567751703255,
"grad_norm": 0.4835710823535919,
"learning_rate": 5.515045523412582e-07,
"loss": 0.4473477602005005,
"step": 4294
},
{
"epoch": 3.2520817562452686,
"grad_norm": 1.6173064708709717,
"learning_rate": 5.505230204053681e-07,
"loss": 0.8970786929130554,
"step": 4296
},
{
"epoch": 3.2535957607872823,
"grad_norm": 1.275166630744934,
"learning_rate": 5.495432115732351e-07,
"loss": 0.5767251253128052,
"step": 4298
},
{
"epoch": 3.255109765329296,
"grad_norm": 2.301729202270508,
"learning_rate": 5.485651273804185e-07,
"loss": 0.5156683325767517,
"step": 4300
},
{
"epoch": 3.2566237698713096,
"grad_norm": 0.3116729259490967,
"learning_rate": 5.475887693597747e-07,
"loss": 0.4290734827518463,
"step": 4302
},
{
"epoch": 3.2581377744133233,
"grad_norm": 2.6406760215759277,
"learning_rate": 5.466141390414554e-07,
"loss": 0.8496693968772888,
"step": 4304
},
{
"epoch": 3.259651778955337,
"grad_norm": 3.133697032928467,
"learning_rate": 5.456412379529038e-07,
"loss": 0.45309460163116455,
"step": 4306
},
{
"epoch": 3.2611657834973506,
"grad_norm": 1.8580689430236816,
"learning_rate": 5.446700676188544e-07,
"loss": 0.4132268726825714,
"step": 4308
},
{
"epoch": 3.262679788039364,
"grad_norm": 1.7730848789215088,
"learning_rate": 5.437006295613273e-07,
"loss": 0.4618237018585205,
"step": 4310
},
{
"epoch": 3.264193792581378,
"grad_norm": 1.9740703105926514,
"learning_rate": 5.427329252996299e-07,
"loss": 0.4546656310558319,
"step": 4312
},
{
"epoch": 3.265707797123391,
"grad_norm": 1.3898946046829224,
"learning_rate": 5.417669563503508e-07,
"loss": 0.5297324657440186,
"step": 4314
},
{
"epoch": 3.267221801665405,
"grad_norm": 3.3110592365264893,
"learning_rate": 5.408027242273592e-07,
"loss": 0.5353357195854187,
"step": 4316
},
{
"epoch": 3.2687358062074185,
"grad_norm": 1.8744398355484009,
"learning_rate": 5.398402304418033e-07,
"loss": 0.48005539178848267,
"step": 4318
},
{
"epoch": 3.270249810749432,
"grad_norm": 1.2090200185775757,
"learning_rate": 5.388794765021063e-07,
"loss": 0.411783367395401,
"step": 4320
},
{
"epoch": 3.271763815291446,
"grad_norm": 1.5509945154190063,
"learning_rate": 5.379204639139646e-07,
"loss": 0.7165201902389526,
"step": 4322
},
{
"epoch": 3.2732778198334596,
"grad_norm": 1.963220477104187,
"learning_rate": 5.369631941803455e-07,
"loss": 0.8382374048233032,
"step": 4324
},
{
"epoch": 3.2747918243754732,
"grad_norm": 0.9289786219596863,
"learning_rate": 5.360076688014856e-07,
"loss": 0.05874892696738243,
"step": 4326
},
{
"epoch": 3.276305828917487,
"grad_norm": 1.2153873443603516,
"learning_rate": 5.350538892748871e-07,
"loss": 0.2230982780456543,
"step": 4328
},
{
"epoch": 3.2778198334595006,
"grad_norm": 1.8008383512496948,
"learning_rate": 5.341018570953158e-07,
"loss": 0.5720658898353577,
"step": 4330
},
{
"epoch": 3.279333838001514,
"grad_norm": 0.8865185976028442,
"learning_rate": 5.331515737548002e-07,
"loss": 0.42918962240219116,
"step": 4332
},
{
"epoch": 3.2808478425435275,
"grad_norm": 1.399993658065796,
"learning_rate": 5.32203040742627e-07,
"loss": 0.34146907925605774,
"step": 4334
},
{
"epoch": 3.282361847085541,
"grad_norm": 0.1398351937532425,
"learning_rate": 5.3125625954534e-07,
"loss": 0.02566056326031685,
"step": 4336
},
{
"epoch": 3.283875851627555,
"grad_norm": 1.386157512664795,
"learning_rate": 5.303112316467375e-07,
"loss": 0.5041114091873169,
"step": 4338
},
{
"epoch": 3.2853898561695685,
"grad_norm": 2.1241602897644043,
"learning_rate": 5.293679585278709e-07,
"loss": 0.5188767313957214,
"step": 4340
},
{
"epoch": 3.286903860711582,
"grad_norm": 2.159628391265869,
"learning_rate": 5.284264416670402e-07,
"loss": 0.3858943283557892,
"step": 4342
},
{
"epoch": 3.288417865253596,
"grad_norm": 0.35991570353507996,
"learning_rate": 5.274866825397937e-07,
"loss": 0.033612895756959915,
"step": 4344
},
{
"epoch": 3.2899318697956095,
"grad_norm": 8.98794174194336,
"learning_rate": 5.265486826189246e-07,
"loss": 0.065807044506073,
"step": 4346
},
{
"epoch": 3.291445874337623,
"grad_norm": 0.7042241096496582,
"learning_rate": 5.256124433744697e-07,
"loss": 0.4922759532928467,
"step": 4348
},
{
"epoch": 3.292959878879637,
"grad_norm": 1.7255072593688965,
"learning_rate": 5.246779662737058e-07,
"loss": 0.35371652245521545,
"step": 4350
},
{
"epoch": 3.2944738834216505,
"grad_norm": 1.7350187301635742,
"learning_rate": 5.237452527811482e-07,
"loss": 0.5240334272384644,
"step": 4352
},
{
"epoch": 3.2959878879636637,
"grad_norm": 1.2877637147903442,
"learning_rate": 5.228143043585488e-07,
"loss": 0.0797341838479042,
"step": 4354
},
{
"epoch": 3.2975018925056774,
"grad_norm": 0.647106945514679,
"learning_rate": 5.21885122464893e-07,
"loss": 0.001529219443909824,
"step": 4356
},
{
"epoch": 3.299015897047691,
"grad_norm": 0.8982298374176025,
"learning_rate": 5.209577085563968e-07,
"loss": 0.4877578914165497,
"step": 4358
},
{
"epoch": 3.3005299015897047,
"grad_norm": 2.524348020553589,
"learning_rate": 5.200320640865069e-07,
"loss": 0.45231226086616516,
"step": 4360
},
{
"epoch": 3.3020439061317184,
"grad_norm": 0.022618478164076805,
"learning_rate": 5.191081905058956e-07,
"loss": 0.3647257089614868,
"step": 4362
},
{
"epoch": 3.303557910673732,
"grad_norm": 1.8879731893539429,
"learning_rate": 5.181860892624612e-07,
"loss": 0.4972490072250366,
"step": 4364
},
{
"epoch": 3.3050719152157457,
"grad_norm": 2.4274215698242188,
"learning_rate": 5.17265761801323e-07,
"loss": 0.28176480531692505,
"step": 4366
},
{
"epoch": 3.3065859197577594,
"grad_norm": 6.416411876678467,
"learning_rate": 5.163472095648216e-07,
"loss": 0.40716928243637085,
"step": 4368
},
{
"epoch": 3.308099924299773,
"grad_norm": 1.85149085521698,
"learning_rate": 5.154304339925146e-07,
"loss": 0.9484966993331909,
"step": 4370
},
{
"epoch": 3.3096139288417863,
"grad_norm": 4.972357273101807,
"learning_rate": 5.145154365211757e-07,
"loss": 0.09232676029205322,
"step": 4372
},
{
"epoch": 3.3111279333838,
"grad_norm": 1.1895396709442139,
"learning_rate": 5.136022185847912e-07,
"loss": 0.03847669064998627,
"step": 4374
},
{
"epoch": 3.3126419379258136,
"grad_norm": 1.5939233303070068,
"learning_rate": 5.126907816145599e-07,
"loss": 0.500072181224823,
"step": 4376
},
{
"epoch": 3.3141559424678273,
"grad_norm": 0.835131049156189,
"learning_rate": 5.117811270388885e-07,
"loss": 0.008409892208874226,
"step": 4378
},
{
"epoch": 3.315669947009841,
"grad_norm": 29.27231216430664,
"learning_rate": 5.108732562833903e-07,
"loss": 0.5254101157188416,
"step": 4380
},
{
"epoch": 3.3171839515518546,
"grad_norm": 1.65263032913208,
"learning_rate": 5.09967170770883e-07,
"loss": 0.09114937484264374,
"step": 4382
},
{
"epoch": 3.3186979560938683,
"grad_norm": 2.0233285427093506,
"learning_rate": 5.090628719213871e-07,
"loss": 0.6537255644798279,
"step": 4384
},
{
"epoch": 3.320211960635882,
"grad_norm": 2.8925328254699707,
"learning_rate": 5.081603611521223e-07,
"loss": 0.027660584077239037,
"step": 4386
},
{
"epoch": 3.3217259651778956,
"grad_norm": 2.887153387069702,
"learning_rate": 5.072596398775062e-07,
"loss": 0.001339872251264751,
"step": 4388
},
{
"epoch": 3.3232399697199093,
"grad_norm": 1.8406734466552734,
"learning_rate": 5.063607095091524e-07,
"loss": 0.0365619957447052,
"step": 4390
},
{
"epoch": 3.324753974261923,
"grad_norm": 1.0864671468734741,
"learning_rate": 5.054635714558673e-07,
"loss": 0.8768489360809326,
"step": 4392
},
{
"epoch": 3.326267978803936,
"grad_norm": 17.93305206298828,
"learning_rate": 5.045682271236481e-07,
"loss": 0.40835052728652954,
"step": 4394
},
{
"epoch": 3.32778198334595,
"grad_norm": 4.551692962646484,
"learning_rate": 5.036746779156812e-07,
"loss": 0.06771939992904663,
"step": 4396
},
{
"epoch": 3.3292959878879635,
"grad_norm": 1.5928632020950317,
"learning_rate": 5.027829252323402e-07,
"loss": 0.40175971388816833,
"step": 4398
},
{
"epoch": 3.330809992429977,
"grad_norm": 1.8709660768508911,
"learning_rate": 5.018929704711824e-07,
"loss": 0.5133115649223328,
"step": 4400
},
{
"epoch": 3.332323996971991,
"grad_norm": 1.6498949527740479,
"learning_rate": 5.010048150269478e-07,
"loss": 0.03898582234978676,
"step": 4402
},
{
"epoch": 3.3338380015140046,
"grad_norm": 0.4957684278488159,
"learning_rate": 5.001184602915561e-07,
"loss": 0.007185522932559252,
"step": 4404
},
{
"epoch": 3.3353520060560182,
"grad_norm": 1.3830325603485107,
"learning_rate": 4.992339076541056e-07,
"loss": 0.4123181998729706,
"step": 4406
},
{
"epoch": 3.336866010598032,
"grad_norm": 1.9279829263687134,
"learning_rate": 4.983511585008695e-07,
"loss": 0.8564941883087158,
"step": 4408
},
{
"epoch": 3.3383800151400456,
"grad_norm": 1.7395809888839722,
"learning_rate": 4.974702142152955e-07,
"loss": 0.3417756259441376,
"step": 4410
},
{
"epoch": 3.3398940196820592,
"grad_norm": 1.7301632165908813,
"learning_rate": 4.965910761780018e-07,
"loss": 0.4873124063014984,
"step": 4412
},
{
"epoch": 3.341408024224073,
"grad_norm": 0.8227750658988953,
"learning_rate": 4.957137457667769e-07,
"loss": 0.009282869286835194,
"step": 4414
},
{
"epoch": 3.342922028766086,
"grad_norm": 0.2841896116733551,
"learning_rate": 4.948382243565753e-07,
"loss": 0.01866152510046959,
"step": 4416
},
{
"epoch": 3.3444360333081,
"grad_norm": 1.9856455326080322,
"learning_rate": 4.939645133195168e-07,
"loss": 0.40884366631507874,
"step": 4418
},
{
"epoch": 3.3459500378501135,
"grad_norm": 2.2969679832458496,
"learning_rate": 4.930926140248843e-07,
"loss": 0.7881647944450378,
"step": 4420
},
{
"epoch": 3.347464042392127,
"grad_norm": 0.34639841318130493,
"learning_rate": 4.922225278391212e-07,
"loss": 0.0036944563034921885,
"step": 4422
},
{
"epoch": 3.348978046934141,
"grad_norm": 2.087341070175171,
"learning_rate": 4.913542561258286e-07,
"loss": 0.6937224864959717,
"step": 4424
},
{
"epoch": 3.3504920514761545,
"grad_norm": 0.256864070892334,
"learning_rate": 4.904878002457658e-07,
"loss": 0.41196000576019287,
"step": 4426
},
{
"epoch": 3.352006056018168,
"grad_norm": 0.23681002855300903,
"learning_rate": 4.896231615568442e-07,
"loss": 0.06520024687051773,
"step": 4428
},
{
"epoch": 3.353520060560182,
"grad_norm": 2.564225673675537,
"learning_rate": 4.887603414141288e-07,
"loss": 0.8333681225776672,
"step": 4430
},
{
"epoch": 3.3550340651021955,
"grad_norm": 4.068796157836914,
"learning_rate": 4.878993411698335e-07,
"loss": 0.3338027596473694,
"step": 4432
},
{
"epoch": 3.3565480696442087,
"grad_norm": 0.886523425579071,
"learning_rate": 4.87040162173321e-07,
"loss": 0.03563188016414642,
"step": 4434
},
{
"epoch": 3.3580620741862224,
"grad_norm": 1.7797421216964722,
"learning_rate": 4.861828057710993e-07,
"loss": 0.7589153051376343,
"step": 4436
},
{
"epoch": 3.359576078728236,
"grad_norm": 1.5448027849197388,
"learning_rate": 4.853272733068198e-07,
"loss": 0.8362768888473511,
"step": 4438
},
{
"epoch": 3.3610900832702497,
"grad_norm": 2.307023763656616,
"learning_rate": 4.844735661212755e-07,
"loss": 0.48151662945747375,
"step": 4440
},
{
"epoch": 3.3626040878122634,
"grad_norm": 2.37272572517395,
"learning_rate": 4.836216855523995e-07,
"loss": 0.8961389660835266,
"step": 4442
},
{
"epoch": 3.364118092354277,
"grad_norm": 2.659275531768799,
"learning_rate": 4.827716329352615e-07,
"loss": 0.7384055256843567,
"step": 4444
},
{
"epoch": 3.3656320968962907,
"grad_norm": 1.6435630321502686,
"learning_rate": 4.819234096020662e-07,
"loss": 0.13354086875915527,
"step": 4446
},
{
"epoch": 3.3671461014383044,
"grad_norm": 8.236377716064453,
"learning_rate": 4.810770168821524e-07,
"loss": 0.10777951031923294,
"step": 4448
},
{
"epoch": 3.368660105980318,
"grad_norm": 3.1149511337280273,
"learning_rate": 4.802324561019895e-07,
"loss": 0.11652589589357376,
"step": 4450
},
{
"epoch": 3.3701741105223317,
"grad_norm": 2.0627260208129883,
"learning_rate": 4.793897285851753e-07,
"loss": 0.3498725891113281,
"step": 4452
},
{
"epoch": 3.3716881150643454,
"grad_norm": 2.8674232959747314,
"learning_rate": 4.785488356524347e-07,
"loss": 0.6957799196243286,
"step": 4454
},
{
"epoch": 3.3732021196063586,
"grad_norm": 22.25187873840332,
"learning_rate": 4.777097786216188e-07,
"loss": 0.8306207656860352,
"step": 4456
},
{
"epoch": 3.3747161241483723,
"grad_norm": 1.1734298467636108,
"learning_rate": 4.768725588076997e-07,
"loss": 0.48902827501296997,
"step": 4458
},
{
"epoch": 3.376230128690386,
"grad_norm": 0.1804582178592682,
"learning_rate": 4.7603717752277094e-07,
"loss": 0.2803400754928589,
"step": 4460
},
{
"epoch": 3.3777441332323996,
"grad_norm": 2.76470947265625,
"learning_rate": 4.752036360760449e-07,
"loss": 0.5458463430404663,
"step": 4462
},
{
"epoch": 3.3792581377744133,
"grad_norm": 2.0508320331573486,
"learning_rate": 4.743719357738499e-07,
"loss": 0.5317753553390503,
"step": 4464
},
{
"epoch": 3.380772142316427,
"grad_norm": 2.33542537689209,
"learning_rate": 4.735420779196299e-07,
"loss": 0.6131026744842529,
"step": 4466
},
{
"epoch": 3.3822861468584406,
"grad_norm": 1.4460207223892212,
"learning_rate": 4.727140638139402e-07,
"loss": 0.098182313144207,
"step": 4468
},
{
"epoch": 3.3838001514004543,
"grad_norm": 1.3774964809417725,
"learning_rate": 4.718878947544473e-07,
"loss": 0.4701768457889557,
"step": 4470
},
{
"epoch": 3.385314155942468,
"grad_norm": 0.5767019987106323,
"learning_rate": 4.7106357203592613e-07,
"loss": 0.09576580673456192,
"step": 4472
},
{
"epoch": 3.386828160484481,
"grad_norm": 0.33554038405418396,
"learning_rate": 4.702410969502575e-07,
"loss": 0.3836270272731781,
"step": 4474
},
{
"epoch": 3.3883421650264953,
"grad_norm": 2.1231112480163574,
"learning_rate": 4.694204707864267e-07,
"loss": 0.4471987783908844,
"step": 4476
},
{
"epoch": 3.3898561695685085,
"grad_norm": 0.8759449124336243,
"learning_rate": 4.6860169483052253e-07,
"loss": 0.4150061309337616,
"step": 4478
},
{
"epoch": 3.391370174110522,
"grad_norm": 2.5322000980377197,
"learning_rate": 4.677847703657323e-07,
"loss": 0.9512776732444763,
"step": 4480
},
{
"epoch": 3.392884178652536,
"grad_norm": 1.4753117561340332,
"learning_rate": 4.669696986723429e-07,
"loss": 0.5149673819541931,
"step": 4482
},
{
"epoch": 3.3943981831945496,
"grad_norm": 3.7654807567596436,
"learning_rate": 4.6615648102773745e-07,
"loss": 0.48217836022377014,
"step": 4484
},
{
"epoch": 3.3959121877365632,
"grad_norm": 3.2758965492248535,
"learning_rate": 4.653451187063932e-07,
"loss": 0.4106427729129791,
"step": 4486
},
{
"epoch": 3.397426192278577,
"grad_norm": 1.7070201635360718,
"learning_rate": 4.6453561297987923e-07,
"loss": 0.4504314661026001,
"step": 4488
},
{
"epoch": 3.3989401968205906,
"grad_norm": 1.3219845294952393,
"learning_rate": 4.637279651168556e-07,
"loss": 0.0958370640873909,
"step": 4490
},
{
"epoch": 3.4004542013626042,
"grad_norm": 1.774338960647583,
"learning_rate": 4.6292217638307104e-07,
"loss": 0.8819460272789001,
"step": 4492
},
{
"epoch": 3.401968205904618,
"grad_norm": 1.8408117294311523,
"learning_rate": 4.6211824804135973e-07,
"loss": 0.6723110675811768,
"step": 4494
},
{
"epoch": 3.403482210446631,
"grad_norm": 0.3831016719341278,
"learning_rate": 4.6131618135164077e-07,
"loss": 0.0017858326900750399,
"step": 4496
},
{
"epoch": 3.404996214988645,
"grad_norm": 3.7257351875305176,
"learning_rate": 4.605159775709153e-07,
"loss": 0.845191478729248,
"step": 4498
},
{
"epoch": 3.4065102195306585,
"grad_norm": 1.4804571866989136,
"learning_rate": 4.5971763795326586e-07,
"loss": 0.5123839974403381,
"step": 4500
},
{
"epoch": 3.408024224072672,
"grad_norm": 1.1878763437271118,
"learning_rate": 4.589211637498522e-07,
"loss": 0.047980278730392456,
"step": 4502
},
{
"epoch": 3.409538228614686,
"grad_norm": 2.475844621658325,
"learning_rate": 4.5812655620891124e-07,
"loss": 0.5012723803520203,
"step": 4504
},
{
"epoch": 3.4110522331566995,
"grad_norm": 1.7084704637527466,
"learning_rate": 4.5733381657575493e-07,
"loss": 0.5272079706192017,
"step": 4506
},
{
"epoch": 3.412566237698713,
"grad_norm": 2.307476282119751,
"learning_rate": 4.5654294609276676e-07,
"loss": 0.9087414145469666,
"step": 4508
},
{
"epoch": 3.414080242240727,
"grad_norm": 1.5252008438110352,
"learning_rate": 4.5575394599940126e-07,
"loss": 0.4586748480796814,
"step": 4510
},
{
"epoch": 3.4155942467827405,
"grad_norm": 0.6368719339370728,
"learning_rate": 4.5496681753218254e-07,
"loss": 0.030393214896321297,
"step": 4512
},
{
"epoch": 3.417108251324754,
"grad_norm": 2.486400604248047,
"learning_rate": 4.541815619247002e-07,
"loss": 0.4503438174724579,
"step": 4514
},
{
"epoch": 3.418622255866768,
"grad_norm": 1.8390527963638306,
"learning_rate": 4.533981804076097e-07,
"loss": 0.41847169399261475,
"step": 4516
},
{
"epoch": 3.420136260408781,
"grad_norm": 3.992145299911499,
"learning_rate": 4.526166742086288e-07,
"loss": 0.6022506952285767,
"step": 4518
},
{
"epoch": 3.4216502649507947,
"grad_norm": 3.6967203617095947,
"learning_rate": 4.5183704455253603e-07,
"loss": 0.03936862200498581,
"step": 4520
},
{
"epoch": 3.4231642694928084,
"grad_norm": 1.9748190641403198,
"learning_rate": 4.5105929266117035e-07,
"loss": 0.8123414516448975,
"step": 4522
},
{
"epoch": 3.424678274034822,
"grad_norm": 1.5579864978790283,
"learning_rate": 4.5028341975342617e-07,
"loss": 0.46440210938453674,
"step": 4524
},
{
"epoch": 3.4261922785768357,
"grad_norm": 1.6632148027420044,
"learning_rate": 4.495094270452542e-07,
"loss": 0.22790756821632385,
"step": 4526
},
{
"epoch": 3.4277062831188494,
"grad_norm": 2.2529702186584473,
"learning_rate": 4.487373157496584e-07,
"loss": 0.07386959344148636,
"step": 4528
},
{
"epoch": 3.429220287660863,
"grad_norm": 3.140373945236206,
"learning_rate": 4.479670870766938e-07,
"loss": 0.02634722925722599,
"step": 4530
},
{
"epoch": 3.4307342922028767,
"grad_norm": 1.9195737838745117,
"learning_rate": 4.4719874223346523e-07,
"loss": 0.8603999614715576,
"step": 4532
},
{
"epoch": 3.4322482967448904,
"grad_norm": 1.6826436519622803,
"learning_rate": 4.4643228242412494e-07,
"loss": 0.3530040979385376,
"step": 4534
},
{
"epoch": 3.4337623012869036,
"grad_norm": 2.034973382949829,
"learning_rate": 4.456677088498715e-07,
"loss": 0.47168779373168945,
"step": 4536
},
{
"epoch": 3.4352763058289173,
"grad_norm": 1.598543643951416,
"learning_rate": 4.449050227089469e-07,
"loss": 0.7359284162521362,
"step": 4538
},
{
"epoch": 3.436790310370931,
"grad_norm": 1.941410779953003,
"learning_rate": 4.4414422519663487e-07,
"loss": 0.475564181804657,
"step": 4540
},
{
"epoch": 3.4383043149129446,
"grad_norm": 1.9490824937820435,
"learning_rate": 4.4338531750526014e-07,
"loss": 0.7518396973609924,
"step": 4542
},
{
"epoch": 3.4398183194549583,
"grad_norm": 3.9313268661499023,
"learning_rate": 4.426283008241852e-07,
"loss": 0.5432718992233276,
"step": 4544
},
{
"epoch": 3.441332323996972,
"grad_norm": 0.7866805195808411,
"learning_rate": 4.418731763398087e-07,
"loss": 0.4719991087913513,
"step": 4546
},
{
"epoch": 3.4428463285389856,
"grad_norm": 1.8675153255462646,
"learning_rate": 4.4111994523556387e-07,
"loss": 0.42668935656547546,
"step": 4548
},
{
"epoch": 3.4443603330809993,
"grad_norm": 2.0621190071105957,
"learning_rate": 4.4036860869191734e-07,
"loss": 0.4930598735809326,
"step": 4550
},
{
"epoch": 3.445874337623013,
"grad_norm": 1.0746771097183228,
"learning_rate": 4.3961916788636594e-07,
"loss": 0.10933978855609894,
"step": 4552
},
{
"epoch": 3.4473883421650267,
"grad_norm": 1.4768182039260864,
"learning_rate": 4.388716239934357e-07,
"loss": 0.43619269132614136,
"step": 4554
},
{
"epoch": 3.4489023467070403,
"grad_norm": 2.2217190265655518,
"learning_rate": 4.381259781846793e-07,
"loss": 0.0945279598236084,
"step": 4556
},
{
"epoch": 3.4504163512490535,
"grad_norm": 0.46245077252388,
"learning_rate": 4.3738223162867586e-07,
"loss": 0.013268718495965004,
"step": 4558
},
{
"epoch": 3.451930355791067,
"grad_norm": 10.982232093811035,
"learning_rate": 4.3664038549102674e-07,
"loss": 0.7407252192497253,
"step": 4560
},
{
"epoch": 3.453444360333081,
"grad_norm": 1.541535496711731,
"learning_rate": 4.3590044093435584e-07,
"loss": 0.40889009833335876,
"step": 4562
},
{
"epoch": 3.4549583648750946,
"grad_norm": 1.5232385396957397,
"learning_rate": 4.351623991183065e-07,
"loss": 0.038662832230329514,
"step": 4564
},
{
"epoch": 3.4564723694171082,
"grad_norm": 0.6275953650474548,
"learning_rate": 4.3442626119953994e-07,
"loss": 0.4955153167247772,
"step": 4566
},
{
"epoch": 3.457986373959122,
"grad_norm": 1.9511977434158325,
"learning_rate": 4.336920283317344e-07,
"loss": 0.3769458532333374,
"step": 4568
},
{
"epoch": 3.4595003785011356,
"grad_norm": 0.9865078330039978,
"learning_rate": 4.329597016655811e-07,
"loss": 0.06349517405033112,
"step": 4570
},
{
"epoch": 3.4610143830431492,
"grad_norm": 1.6756504774093628,
"learning_rate": 4.3222928234878545e-07,
"loss": 0.4160110652446747,
"step": 4572
},
{
"epoch": 3.462528387585163,
"grad_norm": 1.8068053722381592,
"learning_rate": 4.315007715260625e-07,
"loss": 0.9335572719573975,
"step": 4574
},
{
"epoch": 3.4640423921271766,
"grad_norm": 1.5421242713928223,
"learning_rate": 4.307741703391366e-07,
"loss": 0.0467277355492115,
"step": 4576
},
{
"epoch": 3.4655563966691902,
"grad_norm": 1.0545746088027954,
"learning_rate": 4.3004947992673973e-07,
"loss": 0.4467891752719879,
"step": 4578
},
{
"epoch": 3.4670704012112035,
"grad_norm": 0.03795475885272026,
"learning_rate": 4.2932670142460887e-07,
"loss": 0.005698340944945812,
"step": 4580
},
{
"epoch": 3.468584405753217,
"grad_norm": 1.9916728734970093,
"learning_rate": 4.286058359654846e-07,
"loss": 0.5108667612075806,
"step": 4582
},
{
"epoch": 3.470098410295231,
"grad_norm": 6.060962677001953,
"learning_rate": 4.278868846791094e-07,
"loss": 0.33193695545196533,
"step": 4584
},
{
"epoch": 3.4716124148372445,
"grad_norm": 4.610118389129639,
"learning_rate": 4.2716984869222625e-07,
"loss": 0.49625271558761597,
"step": 4586
},
{
"epoch": 3.473126419379258,
"grad_norm": 3.719632148742676,
"learning_rate": 4.2645472912857617e-07,
"loss": 0.41371193528175354,
"step": 4588
},
{
"epoch": 3.474640423921272,
"grad_norm": 0.6784191727638245,
"learning_rate": 4.257415271088966e-07,
"loss": 0.05130447819828987,
"step": 4590
},
{
"epoch": 3.4761544284632855,
"grad_norm": 1.290065884590149,
"learning_rate": 4.2503024375092e-07,
"loss": 0.4512989819049835,
"step": 4592
},
{
"epoch": 3.477668433005299,
"grad_norm": 1.6803733110427856,
"learning_rate": 4.24320880169372e-07,
"loss": 0.4859740436077118,
"step": 4594
},
{
"epoch": 3.479182437547313,
"grad_norm": 1.7524383068084717,
"learning_rate": 4.236134374759694e-07,
"loss": 0.7698912620544434,
"step": 4596
},
{
"epoch": 3.480696442089326,
"grad_norm": 10.697996139526367,
"learning_rate": 4.229079167794184e-07,
"loss": 0.0986369401216507,
"step": 4598
},
{
"epoch": 3.4822104466313397,
"grad_norm": 1.8657169342041016,
"learning_rate": 4.2220431918541345e-07,
"loss": 0.8880587816238403,
"step": 4600
},
{
"epoch": 3.4837244511733534,
"grad_norm": 0.8085569739341736,
"learning_rate": 4.2150264579663514e-07,
"loss": 0.4447839856147766,
"step": 4602
},
{
"epoch": 3.485238455715367,
"grad_norm": 1.6194043159484863,
"learning_rate": 4.208028977127476e-07,
"loss": 0.025584876537322998,
"step": 4604
},
{
"epoch": 3.4867524602573807,
"grad_norm": 1.456982970237732,
"learning_rate": 4.2010507603039853e-07,
"loss": 0.7217623591423035,
"step": 4606
},
{
"epoch": 3.4882664647993944,
"grad_norm": 2.8561789989471436,
"learning_rate": 4.194091818432163e-07,
"loss": 0.43590599298477173,
"step": 4608
},
{
"epoch": 3.489780469341408,
"grad_norm": 1.112181544303894,
"learning_rate": 4.187152162418084e-07,
"loss": 0.47518157958984375,
"step": 4610
},
{
"epoch": 3.4912944738834217,
"grad_norm": 2.6314733028411865,
"learning_rate": 4.1802318031375967e-07,
"loss": 0.8224350810050964,
"step": 4612
},
{
"epoch": 3.4928084784254354,
"grad_norm": 1.5108767747879028,
"learning_rate": 4.1733307514363146e-07,
"loss": 0.3456285893917084,
"step": 4614
},
{
"epoch": 3.494322482967449,
"grad_norm": 6.438747406005859,
"learning_rate": 4.1664490181295813e-07,
"loss": 0.03416212648153305,
"step": 4616
},
{
"epoch": 3.4958364875094627,
"grad_norm": 2.583399534225464,
"learning_rate": 4.1595866140024767e-07,
"loss": 0.48854929208755493,
"step": 4618
},
{
"epoch": 3.497350492051476,
"grad_norm": 2.204908847808838,
"learning_rate": 4.1527435498097803e-07,
"loss": 0.4382208287715912,
"step": 4620
},
{
"epoch": 3.4988644965934896,
"grad_norm": 0.6002346873283386,
"learning_rate": 4.145919836275961e-07,
"loss": 0.5514084696769714,
"step": 4622
},
{
"epoch": 3.5003785011355033,
"grad_norm": 0.23682530224323273,
"learning_rate": 4.1391154840951664e-07,
"loss": 0.13242147862911224,
"step": 4624
},
{
"epoch": 3.501892505677517,
"grad_norm": 2.97751784324646,
"learning_rate": 4.1323305039311985e-07,
"loss": 0.5178206562995911,
"step": 4626
},
{
"epoch": 3.5034065102195306,
"grad_norm": 0.030940603464841843,
"learning_rate": 4.125564906417497e-07,
"loss": 0.46128955483436584,
"step": 4628
},
{
"epoch": 3.5049205147615443,
"grad_norm": 2.5979979038238525,
"learning_rate": 4.118818702157131e-07,
"loss": 0.08022566884756088,
"step": 4630
},
{
"epoch": 3.506434519303558,
"grad_norm": 1.7580132484436035,
"learning_rate": 4.1120919017227715e-07,
"loss": 0.43764054775238037,
"step": 4632
},
{
"epoch": 3.5079485238455717,
"grad_norm": 5.585171699523926,
"learning_rate": 4.105384515656678e-07,
"loss": 0.06207115575671196,
"step": 4634
},
{
"epoch": 3.5094625283875853,
"grad_norm": 5.287844657897949,
"learning_rate": 4.098696554470691e-07,
"loss": 0.0812431275844574,
"step": 4636
},
{
"epoch": 3.5109765329295985,
"grad_norm": 2.570967674255371,
"learning_rate": 4.0920280286462046e-07,
"loss": 0.9307460188865662,
"step": 4638
},
{
"epoch": 3.5124905374716127,
"grad_norm": 0.485571950674057,
"learning_rate": 4.0853789486341506e-07,
"loss": 0.006304420065134764,
"step": 4640
},
{
"epoch": 3.514004542013626,
"grad_norm": 2.8949780464172363,
"learning_rate": 4.078749324854988e-07,
"loss": 0.023514462634921074,
"step": 4642
},
{
"epoch": 3.5155185465556396,
"grad_norm": 1.7327907085418701,
"learning_rate": 4.0721391676986864e-07,
"loss": 0.8459428548812866,
"step": 4644
},
{
"epoch": 3.5170325510976532,
"grad_norm": 3.3375747203826904,
"learning_rate": 4.0655484875247025e-07,
"loss": 0.8779842853546143,
"step": 4646
},
{
"epoch": 3.518546555639667,
"grad_norm": 0.9328950643539429,
"learning_rate": 4.058977294661972e-07,
"loss": 0.4344834089279175,
"step": 4648
},
{
"epoch": 3.5200605601816806,
"grad_norm": 7.156875133514404,
"learning_rate": 4.0524255994088855e-07,
"loss": 0.14196467399597168,
"step": 4650
},
{
"epoch": 3.5215745647236942,
"grad_norm": 2.4609146118164062,
"learning_rate": 4.0458934120332856e-07,
"loss": 0.3472437560558319,
"step": 4652
},
{
"epoch": 3.523088569265708,
"grad_norm": 1.547291874885559,
"learning_rate": 4.039380742772435e-07,
"loss": 0.5102356672286987,
"step": 4654
},
{
"epoch": 3.5246025738077216,
"grad_norm": 0.6036778688430786,
"learning_rate": 4.032887601833006e-07,
"loss": 0.5507434010505676,
"step": 4656
},
{
"epoch": 3.5261165783497352,
"grad_norm": 0.9084627032279968,
"learning_rate": 4.026413999391074e-07,
"loss": 0.001643902505747974,
"step": 4658
},
{
"epoch": 3.5276305828917485,
"grad_norm": 2.3534629344940186,
"learning_rate": 4.0199599455920866e-07,
"loss": 0.19594229757785797,
"step": 4660
},
{
"epoch": 3.529144587433762,
"grad_norm": 3.39477276802063,
"learning_rate": 4.013525450550858e-07,
"loss": 0.447501003742218,
"step": 4662
},
{
"epoch": 3.530658591975776,
"grad_norm": 3.5711495876312256,
"learning_rate": 4.007110524351546e-07,
"loss": 0.490939736366272,
"step": 4664
},
{
"epoch": 3.5321725965177895,
"grad_norm": 0.9418548941612244,
"learning_rate": 4.000715177047648e-07,
"loss": 0.45484209060668945,
"step": 4666
},
{
"epoch": 3.533686601059803,
"grad_norm": 0.17408324778079987,
"learning_rate": 3.994339418661967e-07,
"loss": 0.43646398186683655,
"step": 4668
},
{
"epoch": 3.535200605601817,
"grad_norm": 0.4622572064399719,
"learning_rate": 3.9879832591866166e-07,
"loss": 0.008858655579388142,
"step": 4670
},
{
"epoch": 3.5367146101438305,
"grad_norm": 0.3369908034801483,
"learning_rate": 3.981646708582985e-07,
"loss": 0.39352118968963623,
"step": 4672
},
{
"epoch": 3.538228614685844,
"grad_norm": 1.5281133651733398,
"learning_rate": 3.9753297767817374e-07,
"loss": 0.012237350456416607,
"step": 4674
},
{
"epoch": 3.539742619227858,
"grad_norm": 2.077275037765503,
"learning_rate": 3.969032473682789e-07,
"loss": 0.024612778797745705,
"step": 4676
},
{
"epoch": 3.541256623769871,
"grad_norm": 1.828776240348816,
"learning_rate": 3.962754809155289e-07,
"loss": 0.9028818011283875,
"step": 4678
},
{
"epoch": 3.542770628311885,
"grad_norm": 1.3136188983917236,
"learning_rate": 3.956496793037618e-07,
"loss": 0.44292280077934265,
"step": 4680
},
{
"epoch": 3.5442846328538984,
"grad_norm": 2.2508389949798584,
"learning_rate": 3.950258435137358e-07,
"loss": 0.5394406318664551,
"step": 4682
},
{
"epoch": 3.545798637395912,
"grad_norm": 1.111860752105713,
"learning_rate": 3.944039745231279e-07,
"loss": 0.41304436326026917,
"step": 4684
},
{
"epoch": 3.5473126419379257,
"grad_norm": 0.2608751356601715,
"learning_rate": 3.937840733065333e-07,
"loss": 0.09430210292339325,
"step": 4686
},
{
"epoch": 3.5488266464799394,
"grad_norm": 0.8345021605491638,
"learning_rate": 3.9316614083546326e-07,
"loss": 0.4794743061065674,
"step": 4688
},
{
"epoch": 3.550340651021953,
"grad_norm": 1.2579665184020996,
"learning_rate": 3.925501780783433e-07,
"loss": 0.025660770013928413,
"step": 4690
},
{
"epoch": 3.5518546555639667,
"grad_norm": 2.4215896129608154,
"learning_rate": 3.9193618600051217e-07,
"loss": 0.45629942417144775,
"step": 4692
},
{
"epoch": 3.5533686601059804,
"grad_norm": 1.93221116065979,
"learning_rate": 3.913241655642205e-07,
"loss": 0.4120534658432007,
"step": 4694
},
{
"epoch": 3.554882664647994,
"grad_norm": 1.3998444080352783,
"learning_rate": 3.907141177286283e-07,
"loss": 0.5035297870635986,
"step": 4696
},
{
"epoch": 3.5563966691900077,
"grad_norm": 2.830394744873047,
"learning_rate": 3.901060434498047e-07,
"loss": 0.5286551713943481,
"step": 4698
},
{
"epoch": 3.557910673732021,
"grad_norm": 1.3005797863006592,
"learning_rate": 3.89499943680725e-07,
"loss": 0.48577815294265747,
"step": 4700
},
{
"epoch": 3.559424678274035,
"grad_norm": 8.160534858703613,
"learning_rate": 3.8889581937127145e-07,
"loss": 0.7468402981758118,
"step": 4702
},
{
"epoch": 3.5609386828160483,
"grad_norm": 1.0731797218322754,
"learning_rate": 3.882936714682292e-07,
"loss": 0.035015299916267395,
"step": 4704
},
{
"epoch": 3.562452687358062,
"grad_norm": 0.23713716864585876,
"learning_rate": 3.876935009152862e-07,
"loss": 0.0008841986418701708,
"step": 4706
},
{
"epoch": 3.5639666919000756,
"grad_norm": 2.5999464988708496,
"learning_rate": 3.870953086530317e-07,
"loss": 0.19958657026290894,
"step": 4708
},
{
"epoch": 3.5654806964420893,
"grad_norm": 1.8141475915908813,
"learning_rate": 3.8649909561895453e-07,
"loss": 0.8769348859786987,
"step": 4710
},
{
"epoch": 3.566994700984103,
"grad_norm": 3.1186091899871826,
"learning_rate": 3.8590486274744177e-07,
"loss": 0.142172709107399,
"step": 4712
},
{
"epoch": 3.5685087055261167,
"grad_norm": 4.5143046379089355,
"learning_rate": 3.853126109697766e-07,
"loss": 0.07558415085077286,
"step": 4714
},
{
"epoch": 3.5700227100681303,
"grad_norm": 0.4608166217803955,
"learning_rate": 3.8472234121413844e-07,
"loss": 0.01279566902667284,
"step": 4716
},
{
"epoch": 3.571536714610144,
"grad_norm": 1.0723479986190796,
"learning_rate": 3.841340544055992e-07,
"loss": 0.02807643637061119,
"step": 4718
},
{
"epoch": 3.5730507191521577,
"grad_norm": 2.0866928100585938,
"learning_rate": 3.8354775146612457e-07,
"loss": 0.48381567001342773,
"step": 4720
},
{
"epoch": 3.574564723694171,
"grad_norm": 1.0565446615219116,
"learning_rate": 3.8296343331456966e-07,
"loss": 0.4933178424835205,
"step": 4722
},
{
"epoch": 3.5760787282361846,
"grad_norm": 1.5202834606170654,
"learning_rate": 3.823811008666803e-07,
"loss": 0.7812383770942688,
"step": 4724
},
{
"epoch": 3.5775927327781982,
"grad_norm": 1.1805784702301025,
"learning_rate": 3.818007550350892e-07,
"loss": 0.009394466876983643,
"step": 4726
},
{
"epoch": 3.579106737320212,
"grad_norm": 0.8975813984870911,
"learning_rate": 3.8122239672931647e-07,
"loss": 0.10344241559505463,
"step": 4728
},
{
"epoch": 3.5806207418622256,
"grad_norm": 1.8160611391067505,
"learning_rate": 3.8064602685576696e-07,
"loss": 0.8045218586921692,
"step": 4730
},
{
"epoch": 3.5821347464042392,
"grad_norm": 0.764239490032196,
"learning_rate": 3.800716463177295e-07,
"loss": 0.02110106125473976,
"step": 4732
},
{
"epoch": 3.583648750946253,
"grad_norm": 1.9680922031402588,
"learning_rate": 3.7949925601537495e-07,
"loss": 0.22424939274787903,
"step": 4734
},
{
"epoch": 3.5851627554882666,
"grad_norm": 2.4243404865264893,
"learning_rate": 3.789288568457548e-07,
"loss": 0.47614777088165283,
"step": 4736
},
{
"epoch": 3.5866767600302802,
"grad_norm": 1.1433113813400269,
"learning_rate": 3.78360449702801e-07,
"loss": 0.45122095942497253,
"step": 4738
},
{
"epoch": 3.5881907645722935,
"grad_norm": 0.37726327776908875,
"learning_rate": 3.777940354773227e-07,
"loss": 0.008424730040133,
"step": 4740
},
{
"epoch": 3.5897047691143076,
"grad_norm": 2.8628084659576416,
"learning_rate": 3.7722961505700625e-07,
"loss": 0.4138200879096985,
"step": 4742
},
{
"epoch": 3.591218773656321,
"grad_norm": 1.5430349111557007,
"learning_rate": 3.766671893264126e-07,
"loss": 0.4892534911632538,
"step": 4744
},
{
"epoch": 3.5927327781983345,
"grad_norm": 0.7445770502090454,
"learning_rate": 3.7610675916697766e-07,
"loss": 0.038960278034210205,
"step": 4746
},
{
"epoch": 3.594246782740348,
"grad_norm": 4.10042142868042,
"learning_rate": 3.7554832545700917e-07,
"loss": 0.3631005883216858,
"step": 4748
},
{
"epoch": 3.595760787282362,
"grad_norm": 2.208441972732544,
"learning_rate": 3.7499188907168595e-07,
"loss": 0.4665554463863373,
"step": 4750
},
{
"epoch": 3.5972747918243755,
"grad_norm": 1.664963722229004,
"learning_rate": 3.7443745088305723e-07,
"loss": 0.05097402259707451,
"step": 4752
},
{
"epoch": 3.598788796366389,
"grad_norm": 2.0958125591278076,
"learning_rate": 3.7388501176004005e-07,
"loss": 0.0315166674554348,
"step": 4754
},
{
"epoch": 3.600302800908403,
"grad_norm": 0.5980046391487122,
"learning_rate": 3.733345725684187e-07,
"loss": 0.10055415332317352,
"step": 4756
},
{
"epoch": 3.6018168054504165,
"grad_norm": 1.2863233089447021,
"learning_rate": 3.727861341708432e-07,
"loss": 0.024909744039177895,
"step": 4758
},
{
"epoch": 3.60333080999243,
"grad_norm": 4.173029899597168,
"learning_rate": 3.7223969742682807e-07,
"loss": 0.541300356388092,
"step": 4760
},
{
"epoch": 3.6048448145344434,
"grad_norm": 4.239688396453857,
"learning_rate": 3.716952631927505e-07,
"loss": 0.09993429481983185,
"step": 4762
},
{
"epoch": 3.6063588190764575,
"grad_norm": 2.102480411529541,
"learning_rate": 3.711528323218495e-07,
"loss": 0.03141137957572937,
"step": 4764
},
{
"epoch": 3.6078728236184707,
"grad_norm": 0.5398826599121094,
"learning_rate": 3.7061240566422476e-07,
"loss": 0.5443601608276367,
"step": 4766
},
{
"epoch": 3.6093868281604844,
"grad_norm": 2.651846408843994,
"learning_rate": 3.700739840668343e-07,
"loss": 0.48372316360473633,
"step": 4768
},
{
"epoch": 3.610900832702498,
"grad_norm": 3.62178635597229,
"learning_rate": 3.695375683734941e-07,
"loss": 0.5068149566650391,
"step": 4770
},
{
"epoch": 3.6124148372445117,
"grad_norm": 0.22578562796115875,
"learning_rate": 3.6900315942487687e-07,
"loss": 0.4274672567844391,
"step": 4772
},
{
"epoch": 3.6139288417865254,
"grad_norm": 0.422225683927536,
"learning_rate": 3.6847075805850934e-07,
"loss": 0.003254303941503167,
"step": 4774
},
{
"epoch": 3.615442846328539,
"grad_norm": 2.6041109561920166,
"learning_rate": 3.6794036510877343e-07,
"loss": 0.4800873100757599,
"step": 4776
},
{
"epoch": 3.6169568508705527,
"grad_norm": 1.3889739513397217,
"learning_rate": 3.6741198140690205e-07,
"loss": 0.019312866032123566,
"step": 4778
},
{
"epoch": 3.618470855412566,
"grad_norm": 0.17547756433486938,
"learning_rate": 3.668856077809798e-07,
"loss": 0.48028045892715454,
"step": 4780
},
{
"epoch": 3.61998485995458,
"grad_norm": 1.1334937810897827,
"learning_rate": 3.663612450559414e-07,
"loss": 0.04994925111532211,
"step": 4782
},
{
"epoch": 3.6214988644965933,
"grad_norm": 0.4655733108520508,
"learning_rate": 3.658388940535696e-07,
"loss": 0.5019496083259583,
"step": 4784
},
{
"epoch": 3.623012869038607,
"grad_norm": 0.9031540751457214,
"learning_rate": 3.653185555924943e-07,
"loss": 0.45209193229675293,
"step": 4786
},
{
"epoch": 3.6245268735806206,
"grad_norm": 1.9063076972961426,
"learning_rate": 3.6480023048819196e-07,
"loss": 0.9164039492607117,
"step": 4788
},
{
"epoch": 3.6260408781226343,
"grad_norm": 2.208582639694214,
"learning_rate": 3.64283919552983e-07,
"loss": 0.05035603046417236,
"step": 4790
},
{
"epoch": 3.627554882664648,
"grad_norm": 1.7097535133361816,
"learning_rate": 3.637696235960316e-07,
"loss": 0.23326392471790314,
"step": 4792
},
{
"epoch": 3.6290688872066617,
"grad_norm": 1.653237223625183,
"learning_rate": 3.632573434233438e-07,
"loss": 0.4799347221851349,
"step": 4794
},
{
"epoch": 3.6305828917486753,
"grad_norm": 2.5724575519561768,
"learning_rate": 3.6274707983776723e-07,
"loss": 0.8532195091247559,
"step": 4796
},
{
"epoch": 3.632096896290689,
"grad_norm": 0.6447435021400452,
"learning_rate": 3.622388336389881e-07,
"loss": 0.4974871277809143,
"step": 4798
},
{
"epoch": 3.6336109008327027,
"grad_norm": 2.2832088470458984,
"learning_rate": 3.6173260562353163e-07,
"loss": 0.19416573643684387,
"step": 4800
},
{
"epoch": 3.635124905374716,
"grad_norm": 3.592654228210449,
"learning_rate": 3.6122839658475964e-07,
"loss": 0.6662434935569763,
"step": 4802
},
{
"epoch": 3.63663890991673,
"grad_norm": 1.6334006786346436,
"learning_rate": 3.6072620731287066e-07,
"loss": 0.4261821210384369,
"step": 4804
},
{
"epoch": 3.6381529144587432,
"grad_norm": 0.5838547945022583,
"learning_rate": 3.6022603859489704e-07,
"loss": 0.09783003479242325,
"step": 4806
},
{
"epoch": 3.639666919000757,
"grad_norm": 1.4045666456222534,
"learning_rate": 3.597278912147044e-07,
"loss": 0.07156740128993988,
"step": 4808
},
{
"epoch": 3.6411809235427706,
"grad_norm": 0.7741947174072266,
"learning_rate": 3.592317659529913e-07,
"loss": 0.5537245869636536,
"step": 4810
},
{
"epoch": 3.6426949280847842,
"grad_norm": 0.0055837552063167095,
"learning_rate": 3.5873766358728684e-07,
"loss": 0.024619167670607567,
"step": 4812
},
{
"epoch": 3.644208932626798,
"grad_norm": 1.4337111711502075,
"learning_rate": 3.582455848919496e-07,
"loss": 0.4184442460536957,
"step": 4814
},
{
"epoch": 3.6457229371688116,
"grad_norm": 1.548632264137268,
"learning_rate": 3.577555306381666e-07,
"loss": 0.7215121388435364,
"step": 4816
},
{
"epoch": 3.6472369417108252,
"grad_norm": 22.545286178588867,
"learning_rate": 3.5726750159395296e-07,
"loss": 0.8514040112495422,
"step": 4818
},
{
"epoch": 3.648750946252839,
"grad_norm": 1.5197862386703491,
"learning_rate": 3.5678149852414884e-07,
"loss": 0.39174315333366394,
"step": 4820
},
{
"epoch": 3.6502649507948526,
"grad_norm": 2.523045778274536,
"learning_rate": 3.5629752219042015e-07,
"loss": 0.37386053800582886,
"step": 4822
},
{
"epoch": 3.651778955336866,
"grad_norm": 1.6517491340637207,
"learning_rate": 3.558155733512557e-07,
"loss": 0.46329256892204285,
"step": 4824
},
{
"epoch": 3.6532929598788795,
"grad_norm": 1.114324927330017,
"learning_rate": 3.5533565276196766e-07,
"loss": 0.12746909260749817,
"step": 4826
},
{
"epoch": 3.654806964420893,
"grad_norm": 1.4792526960372925,
"learning_rate": 3.5485776117468887e-07,
"loss": 0.053770918399095535,
"step": 4828
},
{
"epoch": 3.656320968962907,
"grad_norm": 1.9576071500778198,
"learning_rate": 3.543818993383724e-07,
"loss": 0.857465922832489,
"step": 4830
},
{
"epoch": 3.6578349735049205,
"grad_norm": 0.609967827796936,
"learning_rate": 3.5390806799879065e-07,
"loss": 0.4741651117801666,
"step": 4832
},
{
"epoch": 3.659348978046934,
"grad_norm": 1.04575514793396,
"learning_rate": 3.534362678985334e-07,
"loss": 0.38848766684532166,
"step": 4834
},
{
"epoch": 3.660862982588948,
"grad_norm": 0.827594518661499,
"learning_rate": 3.529664997770074e-07,
"loss": 0.16450990736484528,
"step": 4836
},
{
"epoch": 3.6623769871309615,
"grad_norm": 4.762159824371338,
"learning_rate": 3.524987643704343e-07,
"loss": 0.8133305311203003,
"step": 4838
},
{
"epoch": 3.663890991672975,
"grad_norm": 4.865293979644775,
"learning_rate": 3.5203306241185103e-07,
"loss": 0.36955681443214417,
"step": 4840
},
{
"epoch": 3.6654049962149884,
"grad_norm": 0.7137389183044434,
"learning_rate": 3.515693946311069e-07,
"loss": 0.0033145768102258444,
"step": 4842
},
{
"epoch": 3.6669190007570025,
"grad_norm": 0.4564444124698639,
"learning_rate": 3.511077617548632e-07,
"loss": 0.0010724674211815,
"step": 4844
},
{
"epoch": 3.6684330052990157,
"grad_norm": 0.32473912835121155,
"learning_rate": 3.506481645065932e-07,
"loss": 0.4716408848762512,
"step": 4846
},
{
"epoch": 3.6699470098410294,
"grad_norm": 3.3602616786956787,
"learning_rate": 3.501906036065784e-07,
"loss": 0.5193576216697693,
"step": 4848
},
{
"epoch": 3.671461014383043,
"grad_norm": 2.077227830886841,
"learning_rate": 3.497350797719101e-07,
"loss": 0.18189232051372528,
"step": 4850
},
{
"epoch": 3.6729750189250567,
"grad_norm": 2.433666229248047,
"learning_rate": 3.4928159371648635e-07,
"loss": 0.46935853362083435,
"step": 4852
},
{
"epoch": 3.6744890234670704,
"grad_norm": 1.6422847509384155,
"learning_rate": 3.488301461510123e-07,
"loss": 0.4132682979106903,
"step": 4854
},
{
"epoch": 3.676003028009084,
"grad_norm": 1.8942492008209229,
"learning_rate": 3.4838073778299776e-07,
"loss": 0.7164862751960754,
"step": 4856
},
{
"epoch": 3.6775170325510977,
"grad_norm": 0.6586605906486511,
"learning_rate": 3.4793336931675693e-07,
"loss": 0.28098931908607483,
"step": 4858
},
{
"epoch": 3.6790310370931114,
"grad_norm": 1.4436407089233398,
"learning_rate": 3.4748804145340693e-07,
"loss": 0.6255571842193604,
"step": 4860
},
{
"epoch": 3.680545041635125,
"grad_norm": 6.418086051940918,
"learning_rate": 3.470447548908672e-07,
"loss": 0.5840961933135986,
"step": 4862
},
{
"epoch": 3.6820590461771383,
"grad_norm": 0.7285082340240479,
"learning_rate": 3.466035103238579e-07,
"loss": 0.12661457061767578,
"step": 4864
},
{
"epoch": 3.6835730507191524,
"grad_norm": 3.845855712890625,
"learning_rate": 3.461643084438984e-07,
"loss": 0.38669610023498535,
"step": 4866
},
{
"epoch": 3.6850870552611656,
"grad_norm": 2.364147663116455,
"learning_rate": 3.4572714993930797e-07,
"loss": 0.9998100996017456,
"step": 4868
},
{
"epoch": 3.6866010598031793,
"grad_norm": 3.6311380863189697,
"learning_rate": 3.4529203549520226e-07,
"loss": 0.14734582602977753,
"step": 4870
},
{
"epoch": 3.688115064345193,
"grad_norm": 0.20212996006011963,
"learning_rate": 3.4485896579349415e-07,
"loss": 0.004680413752794266,
"step": 4872
},
{
"epoch": 3.6896290688872067,
"grad_norm": 2.0408899784088135,
"learning_rate": 3.444279415128919e-07,
"loss": 0.06556116044521332,
"step": 4874
},
{
"epoch": 3.6911430734292203,
"grad_norm": 2.7118277549743652,
"learning_rate": 3.43998963328898e-07,
"loss": 0.8545077443122864,
"step": 4876
},
{
"epoch": 3.692657077971234,
"grad_norm": 1.9300854206085205,
"learning_rate": 3.435720319138084e-07,
"loss": 0.9223077297210693,
"step": 4878
},
{
"epoch": 3.6941710825132477,
"grad_norm": 0.09388744086027145,
"learning_rate": 3.431471479367113e-07,
"loss": 0.41516542434692383,
"step": 4880
},
{
"epoch": 3.6956850870552613,
"grad_norm": 0.01692943461239338,
"learning_rate": 3.4272431206348635e-07,
"loss": 0.37090763449668884,
"step": 4882
},
{
"epoch": 3.697199091597275,
"grad_norm": 2.315216302871704,
"learning_rate": 3.4230352495680335e-07,
"loss": 0.9441631436347961,
"step": 4884
},
{
"epoch": 3.6987130961392882,
"grad_norm": 2.9745566844940186,
"learning_rate": 3.4188478727612087e-07,
"loss": 0.9142878651618958,
"step": 4886
},
{
"epoch": 3.700227100681302,
"grad_norm": 1.6987249851226807,
"learning_rate": 3.4146809967768595e-07,
"loss": 0.8680942058563232,
"step": 4888
},
{
"epoch": 3.7017411052233156,
"grad_norm": 1.8989615440368652,
"learning_rate": 3.41053462814533e-07,
"loss": 0.022511256858706474,
"step": 4890
},
{
"epoch": 3.7032551097653292,
"grad_norm": 1.9308021068572998,
"learning_rate": 3.4064087733648184e-07,
"loss": 0.8625115752220154,
"step": 4892
},
{
"epoch": 3.704769114307343,
"grad_norm": 1.6914867162704468,
"learning_rate": 3.40230343890138e-07,
"loss": 0.8048926591873169,
"step": 4894
},
{
"epoch": 3.7062831188493566,
"grad_norm": 0.6650161147117615,
"learning_rate": 3.3982186311889046e-07,
"loss": 0.46356675028800964,
"step": 4896
},
{
"epoch": 3.7077971233913702,
"grad_norm": 2.505150556564331,
"learning_rate": 3.3941543566291193e-07,
"loss": 0.043180715292692184,
"step": 4898
},
{
"epoch": 3.709311127933384,
"grad_norm": 2.578604221343994,
"learning_rate": 3.390110621591566e-07,
"loss": 0.39062124490737915,
"step": 4900
},
{
"epoch": 3.7108251324753976,
"grad_norm": 1.5174658298492432,
"learning_rate": 3.3860874324135974e-07,
"loss": 0.46687743067741394,
"step": 4902
},
{
"epoch": 3.712339137017411,
"grad_norm": 1.998696208000183,
"learning_rate": 3.3820847954003713e-07,
"loss": 0.010921820998191833,
"step": 4904
},
{
"epoch": 3.713853141559425,
"grad_norm": 1.7215547561645508,
"learning_rate": 3.378102716824832e-07,
"loss": 0.0716339498758316,
"step": 4906
},
{
"epoch": 3.715367146101438,
"grad_norm": 0.12513110041618347,
"learning_rate": 3.3741412029277034e-07,
"loss": 0.4399803876876831,
"step": 4908
},
{
"epoch": 3.716881150643452,
"grad_norm": 0.37457987666130066,
"learning_rate": 3.370200259917483e-07,
"loss": 0.3699595034122467,
"step": 4910
},
{
"epoch": 3.7183951551854655,
"grad_norm": 1.7787532806396484,
"learning_rate": 3.3662798939704323e-07,
"loss": 0.021962502971291542,
"step": 4912
},
{
"epoch": 3.719909159727479,
"grad_norm": 1.0377203226089478,
"learning_rate": 3.3623801112305577e-07,
"loss": 0.09830473363399506,
"step": 4914
},
{
"epoch": 3.721423164269493,
"grad_norm": 0.755774736404419,
"learning_rate": 3.3585009178096114e-07,
"loss": 0.01071044523268938,
"step": 4916
},
{
"epoch": 3.7229371688115065,
"grad_norm": 0.1359897404909134,
"learning_rate": 3.3546423197870793e-07,
"loss": 0.46218374371528625,
"step": 4918
},
{
"epoch": 3.72445117335352,
"grad_norm": 4.3549065589904785,
"learning_rate": 3.350804323210165e-07,
"loss": 0.32898473739624023,
"step": 4920
},
{
"epoch": 3.725965177895534,
"grad_norm": 0.4687940180301666,
"learning_rate": 3.3469869340937915e-07,
"loss": 0.05365509167313576,
"step": 4922
},
{
"epoch": 3.7274791824375475,
"grad_norm": 0.1344975084066391,
"learning_rate": 3.3431901584205834e-07,
"loss": 0.4858251214027405,
"step": 4924
},
{
"epoch": 3.7289931869795607,
"grad_norm": 2.259783983230591,
"learning_rate": 3.3394140021408555e-07,
"loss": 0.47054964303970337,
"step": 4926
},
{
"epoch": 3.730507191521575,
"grad_norm": 5.918397903442383,
"learning_rate": 3.335658471172615e-07,
"loss": 0.3773191571235657,
"step": 4928
},
{
"epoch": 3.732021196063588,
"grad_norm": 1.230940580368042,
"learning_rate": 3.3319235714015426e-07,
"loss": 0.11731115728616714,
"step": 4930
},
{
"epoch": 3.7335352006056017,
"grad_norm": 0.3544609546661377,
"learning_rate": 3.328209308680981e-07,
"loss": 0.5121480822563171,
"step": 4932
},
{
"epoch": 3.7350492051476154,
"grad_norm": 0.2798723876476288,
"learning_rate": 3.324515688831939e-07,
"loss": 0.015740007162094116,
"step": 4934
},
{
"epoch": 3.736563209689629,
"grad_norm": 2.402623414993286,
"learning_rate": 3.320842717643066e-07,
"loss": 0.4829557538032532,
"step": 4936
},
{
"epoch": 3.7380772142316427,
"grad_norm": 3.5150060653686523,
"learning_rate": 3.317190400870653e-07,
"loss": 1.0066595077514648,
"step": 4938
},
{
"epoch": 3.7395912187736564,
"grad_norm": 2.071301221847534,
"learning_rate": 3.3135587442386263e-07,
"loss": 0.5190615653991699,
"step": 4940
},
{
"epoch": 3.74110522331567,
"grad_norm": 2.0592801570892334,
"learning_rate": 3.309947753438528e-07,
"loss": 0.4132930636405945,
"step": 4942
},
{
"epoch": 3.7426192278576833,
"grad_norm": 0.05614176392555237,
"learning_rate": 3.3063574341295124e-07,
"loss": 0.18080037832260132,
"step": 4944
},
{
"epoch": 3.7441332323996974,
"grad_norm": 1.8372803926467896,
"learning_rate": 3.302787791938339e-07,
"loss": 0.4844646155834198,
"step": 4946
},
{
"epoch": 3.7456472369417106,
"grad_norm": 0.5025854706764221,
"learning_rate": 3.299238832459364e-07,
"loss": 0.0169985331594944,
"step": 4948
},
{
"epoch": 3.7471612414837243,
"grad_norm": 2.4502007961273193,
"learning_rate": 3.2957105612545275e-07,
"loss": 0.3472217321395874,
"step": 4950
},
{
"epoch": 3.748675246025738,
"grad_norm": 0.560302197933197,
"learning_rate": 3.292202983853347e-07,
"loss": 0.46183523535728455,
"step": 4952
},
{
"epoch": 3.7501892505677517,
"grad_norm": 1.8078094720840454,
"learning_rate": 3.2887161057529067e-07,
"loss": 0.40099021792411804,
"step": 4954
},
{
"epoch": 3.7517032551097653,
"grad_norm": 1.7329490184783936,
"learning_rate": 3.2852499324178554e-07,
"loss": 0.761401355266571,
"step": 4956
},
{
"epoch": 3.753217259651779,
"grad_norm": 1.790453314781189,
"learning_rate": 3.281804469280392e-07,
"loss": 0.8640339970588684,
"step": 4958
},
{
"epoch": 3.7547312641937927,
"grad_norm": 0.2882123291492462,
"learning_rate": 3.2783797217402525e-07,
"loss": 0.4108726680278778,
"step": 4960
},
{
"epoch": 3.7562452687358063,
"grad_norm": 1.0697760581970215,
"learning_rate": 3.274975695164716e-07,
"loss": 0.27952757477760315,
"step": 4962
},
{
"epoch": 3.75775927327782,
"grad_norm": 0.7273476123809814,
"learning_rate": 3.271592394888583e-07,
"loss": 0.041864413768053055,
"step": 4964
},
{
"epoch": 3.7592732778198332,
"grad_norm": 1.7666912078857422,
"learning_rate": 3.2682298262141696e-07,
"loss": 0.9729198217391968,
"step": 4966
},
{
"epoch": 3.7607872823618473,
"grad_norm": 1.9599757194519043,
"learning_rate": 3.264887994411306e-07,
"loss": 0.10504017025232315,
"step": 4968
},
{
"epoch": 3.7623012869038606,
"grad_norm": 0.019330648705363274,
"learning_rate": 3.2615669047173225e-07,
"loss": 0.41043803095817566,
"step": 4970
},
{
"epoch": 3.7638152914458742,
"grad_norm": 0.7178271412849426,
"learning_rate": 3.2582665623370385e-07,
"loss": 0.47693270444869995,
"step": 4972
},
{
"epoch": 3.765329295987888,
"grad_norm": 0.08044314384460449,
"learning_rate": 3.2549869724427634e-07,
"loss": 0.04021545127034187,
"step": 4974
},
{
"epoch": 3.7668433005299016,
"grad_norm": 3.817652702331543,
"learning_rate": 3.251728140174279e-07,
"loss": 0.9784973859786987,
"step": 4976
},
{
"epoch": 3.7683573050719152,
"grad_norm": 3.3584280014038086,
"learning_rate": 3.248490070638837e-07,
"loss": 0.5802483558654785,
"step": 4978
},
{
"epoch": 3.769871309613929,
"grad_norm": 1.8819648027420044,
"learning_rate": 3.2452727689111543e-07,
"loss": 0.4363057017326355,
"step": 4980
},
{
"epoch": 3.7713853141559426,
"grad_norm": 0.22932952642440796,
"learning_rate": 3.2420762400333903e-07,
"loss": 0.43567347526550293,
"step": 4982
},
{
"epoch": 3.7728993186979563,
"grad_norm": 2.5808701515197754,
"learning_rate": 3.238900489015158e-07,
"loss": 0.48371621966362,
"step": 4984
},
{
"epoch": 3.77441332323997,
"grad_norm": 2.555543899536133,
"learning_rate": 3.235745520833506e-07,
"loss": 0.5207771062850952,
"step": 4986
},
{
"epoch": 3.775927327781983,
"grad_norm": 0.1696174144744873,
"learning_rate": 3.232611340432908e-07,
"loss": 0.4082969129085541,
"step": 4988
},
{
"epoch": 3.777441332323997,
"grad_norm": 3.984182357788086,
"learning_rate": 3.22949795272526e-07,
"loss": 0.1221313551068306,
"step": 4990
},
{
"epoch": 3.7789553368660105,
"grad_norm": 2.366755723953247,
"learning_rate": 3.226405362589877e-07,
"loss": 0.36068195104599,
"step": 4992
},
{
"epoch": 3.780469341408024,
"grad_norm": 0.05323546752333641,
"learning_rate": 3.2233335748734724e-07,
"loss": 0.5102896690368652,
"step": 4994
},
{
"epoch": 3.781983345950038,
"grad_norm": 0.11353172361850739,
"learning_rate": 3.220282594390161e-07,
"loss": 0.0019964941311627626,
"step": 4996
},
{
"epoch": 3.7834973504920515,
"grad_norm": 5.21306037902832,
"learning_rate": 3.2172524259214534e-07,
"loss": 0.42315593361854553,
"step": 4998
},
{
"epoch": 3.785011355034065,
"grad_norm": 0.41837674379348755,
"learning_rate": 3.2142430742162355e-07,
"loss": 0.00036322709638625383,
"step": 5000
},
{
"epoch": 3.786525359576079,
"grad_norm": 2.8493845462799072,
"learning_rate": 3.211254543990774e-07,
"loss": 0.009760625660419464,
"step": 5002
},
{
"epoch": 3.7880393641180925,
"grad_norm": 2.051926374435425,
"learning_rate": 3.208286839928701e-07,
"loss": 0.8073641657829285,
"step": 5004
},
{
"epoch": 3.7895533686601057,
"grad_norm": 8.931031227111816,
"learning_rate": 3.2053399666810133e-07,
"loss": 0.47890886664390564,
"step": 5006
},
{
"epoch": 3.79106737320212,
"grad_norm": 0.6862629652023315,
"learning_rate": 3.2024139288660594e-07,
"loss": 0.10860828310251236,
"step": 5008
},
{
"epoch": 3.792581377744133,
"grad_norm": 0.049004968255758286,
"learning_rate": 3.1995087310695357e-07,
"loss": 0.008655873127281666,
"step": 5010
},
{
"epoch": 3.7940953822861467,
"grad_norm": 0.8857827186584473,
"learning_rate": 3.196624377844476e-07,
"loss": 0.008771966211497784,
"step": 5012
},
{
"epoch": 3.7956093868281604,
"grad_norm": 1.8794935941696167,
"learning_rate": 3.1937608737112493e-07,
"loss": 0.4379273056983948,
"step": 5014
},
{
"epoch": 3.797123391370174,
"grad_norm": 0.3324967324733734,
"learning_rate": 3.190918223157546e-07,
"loss": 0.41587817668914795,
"step": 5016
},
{
"epoch": 3.7986373959121877,
"grad_norm": 0.26607972383499146,
"learning_rate": 3.188096430638377e-07,
"loss": 0.022580014541745186,
"step": 5018
},
{
"epoch": 3.8001514004542014,
"grad_norm": 2.0681216716766357,
"learning_rate": 3.1852955005760665e-07,
"loss": 0.3928464949131012,
"step": 5020
},
{
"epoch": 3.801665404996215,
"grad_norm": 0.1730499416589737,
"learning_rate": 3.1825154373602416e-07,
"loss": 0.48923325538635254,
"step": 5022
},
{
"epoch": 3.8031794095382288,
"grad_norm": 4.178548812866211,
"learning_rate": 3.179756245347822e-07,
"loss": 0.4484570324420929,
"step": 5024
},
{
"epoch": 3.8046934140802424,
"grad_norm": 0.5775771141052246,
"learning_rate": 3.1770179288630244e-07,
"loss": 0.20080740749835968,
"step": 5026
},
{
"epoch": 3.8062074186222556,
"grad_norm": 0.3481805920600891,
"learning_rate": 3.174300492197346e-07,
"loss": 0.01089209783822298,
"step": 5028
},
{
"epoch": 3.8077214231642698,
"grad_norm": 1.8364008665084839,
"learning_rate": 3.1716039396095646e-07,
"loss": 0.45432248711586,
"step": 5030
},
{
"epoch": 3.809235427706283,
"grad_norm": 0.7630714178085327,
"learning_rate": 3.168928275325724e-07,
"loss": 0.45632779598236084,
"step": 5032
},
{
"epoch": 3.8107494322482967,
"grad_norm": 5.66952657699585,
"learning_rate": 3.1662735035391325e-07,
"loss": 0.5279141664505005,
"step": 5034
},
{
"epoch": 3.8122634367903103,
"grad_norm": 1.6522698402404785,
"learning_rate": 3.1636396284103594e-07,
"loss": 0.8378298878669739,
"step": 5036
},
{
"epoch": 3.813777441332324,
"grad_norm": 2.261073589324951,
"learning_rate": 3.16102665406722e-07,
"loss": 0.8691202998161316,
"step": 5038
},
{
"epoch": 3.8152914458743377,
"grad_norm": 2.561520576477051,
"learning_rate": 3.1584345846047765e-07,
"loss": 0.47621315717697144,
"step": 5040
},
{
"epoch": 3.8168054504163513,
"grad_norm": 3.8017935752868652,
"learning_rate": 3.155863424085327e-07,
"loss": 0.40808743238449097,
"step": 5042
},
{
"epoch": 3.818319454958365,
"grad_norm": 2.399139165878296,
"learning_rate": 3.1533131765384063e-07,
"loss": 0.45948272943496704,
"step": 5044
},
{
"epoch": 3.8198334595003782,
"grad_norm": 1.641716480255127,
"learning_rate": 3.150783845960765e-07,
"loss": 0.5885670781135559,
"step": 5046
},
{
"epoch": 3.8213474640423923,
"grad_norm": 1.085816502571106,
"learning_rate": 3.148275436316381e-07,
"loss": 0.281469464302063,
"step": 5048
},
{
"epoch": 3.8228614685844056,
"grad_norm": 1.2711929082870483,
"learning_rate": 3.145787951536441e-07,
"loss": 0.32660719752311707,
"step": 5050
},
{
"epoch": 3.8243754731264192,
"grad_norm": 1.4348828792572021,
"learning_rate": 3.1433213955193404e-07,
"loss": 0.4572903513908386,
"step": 5052
},
{
"epoch": 3.825889477668433,
"grad_norm": 1.9743393659591675,
"learning_rate": 3.1408757721306693e-07,
"loss": 0.8791977763175964,
"step": 5054
},
{
"epoch": 3.8274034822104466,
"grad_norm": 2.0506911277770996,
"learning_rate": 3.1384510852032213e-07,
"loss": 0.005691791884601116,
"step": 5056
},
{
"epoch": 3.8289174867524602,
"grad_norm": 0.03359084948897362,
"learning_rate": 3.1360473385369705e-07,
"loss": 0.039005476981401443,
"step": 5058
},
{
"epoch": 3.830431491294474,
"grad_norm": 3.871607780456543,
"learning_rate": 3.133664535899078e-07,
"loss": 0.748072624206543,
"step": 5060
},
{
"epoch": 3.8319454958364876,
"grad_norm": 2.2511465549468994,
"learning_rate": 3.131302681023877e-07,
"loss": 0.862351655960083,
"step": 5062
},
{
"epoch": 3.8334595003785013,
"grad_norm": 1.9889755249023438,
"learning_rate": 3.128961777612876e-07,
"loss": 0.08380108326673508,
"step": 5064
},
{
"epoch": 3.834973504920515,
"grad_norm": 2.928863048553467,
"learning_rate": 3.1266418293347464e-07,
"loss": 0.4806826412677765,
"step": 5066
},
{
"epoch": 3.836487509462528,
"grad_norm": 5.594569683074951,
"learning_rate": 3.124342839825314e-07,
"loss": 0.020531795918941498,
"step": 5068
},
{
"epoch": 3.8380015140045423,
"grad_norm": 3.1761984825134277,
"learning_rate": 3.122064812687568e-07,
"loss": 0.6929623484611511,
"step": 5070
},
{
"epoch": 3.8395155185465555,
"grad_norm": 0.08937308937311172,
"learning_rate": 3.1198077514916374e-07,
"loss": 0.2494657337665558,
"step": 5072
},
{
"epoch": 3.841029523088569,
"grad_norm": 2.0265660285949707,
"learning_rate": 3.1175716597747953e-07,
"loss": 0.4214555621147156,
"step": 5074
},
{
"epoch": 3.842543527630583,
"grad_norm": 1.8746016025543213,
"learning_rate": 3.1153565410414514e-07,
"loss": 0.43491077423095703,
"step": 5076
},
{
"epoch": 3.8440575321725965,
"grad_norm": 0.8957091569900513,
"learning_rate": 3.113162398763148e-07,
"loss": 0.40135228633880615,
"step": 5078
},
{
"epoch": 3.84557153671461,
"grad_norm": 2.91975998878479,
"learning_rate": 3.110989236378549e-07,
"loss": 0.7896611094474792,
"step": 5080
},
{
"epoch": 3.847085541256624,
"grad_norm": 0.6040961742401123,
"learning_rate": 3.108837057293445e-07,
"loss": 0.012859215028584003,
"step": 5082
},
{
"epoch": 3.8485995457986375,
"grad_norm": 0.19318491220474243,
"learning_rate": 3.106705864880735e-07,
"loss": 0.47871896624565125,
"step": 5084
},
{
"epoch": 3.850113550340651,
"grad_norm": 2.807324171066284,
"learning_rate": 3.1045956624804317e-07,
"loss": 0.4998835325241089,
"step": 5086
},
{
"epoch": 3.851627554882665,
"grad_norm": 4.192824363708496,
"learning_rate": 3.1025064533996517e-07,
"loss": 0.39344674348831177,
"step": 5088
},
{
"epoch": 3.853141559424678,
"grad_norm": 6.574662208557129,
"learning_rate": 3.1004382409126064e-07,
"loss": 0.4768523871898651,
"step": 5090
},
{
"epoch": 3.8546555639666917,
"grad_norm": 1.621928334236145,
"learning_rate": 3.0983910282606083e-07,
"loss": 0.42897361516952515,
"step": 5092
},
{
"epoch": 3.8561695685087054,
"grad_norm": 2.074524402618408,
"learning_rate": 3.096364818652055e-07,
"loss": 0.445551335811615,
"step": 5094
},
{
"epoch": 3.857683573050719,
"grad_norm": 1.5912147760391235,
"learning_rate": 3.0943596152624284e-07,
"loss": 0.003714049933478236,
"step": 5096
},
{
"epoch": 3.8591975775927327,
"grad_norm": 1.4102367162704468,
"learning_rate": 3.0923754212342886e-07,
"loss": 0.698522686958313,
"step": 5098
},
{
"epoch": 3.8607115821347464,
"grad_norm": 1.337382435798645,
"learning_rate": 3.0904122396772705e-07,
"loss": 0.007130913436412811,
"step": 5100
},
{
"epoch": 3.86222558667676,
"grad_norm": 3.0540449619293213,
"learning_rate": 3.08847007366808e-07,
"loss": 0.4882618188858032,
"step": 5102
},
{
"epoch": 3.8637395912187738,
"grad_norm": 1.6839513778686523,
"learning_rate": 3.0865489262504844e-07,
"loss": 0.41174182295799255,
"step": 5104
},
{
"epoch": 3.8652535957607874,
"grad_norm": 0.445525199174881,
"learning_rate": 3.0846488004353116e-07,
"loss": 0.018767863512039185,
"step": 5106
},
{
"epoch": 3.8667676003028006,
"grad_norm": 3.5258700847625732,
"learning_rate": 3.0827696992004456e-07,
"loss": 0.8859530091285706,
"step": 5108
},
{
"epoch": 3.8682816048448148,
"grad_norm": 1.3775527477264404,
"learning_rate": 3.0809116254908205e-07,
"loss": 0.9377045631408691,
"step": 5110
},
{
"epoch": 3.869795609386828,
"grad_norm": 3.718674659729004,
"learning_rate": 3.079074582218412e-07,
"loss": 0.8108839392662048,
"step": 5112
},
{
"epoch": 3.8713096139288417,
"grad_norm": 1.806331992149353,
"learning_rate": 3.077258572262245e-07,
"loss": 0.5829481482505798,
"step": 5114
},
{
"epoch": 3.8728236184708553,
"grad_norm": 1.367382287979126,
"learning_rate": 3.0754635984683733e-07,
"loss": 0.3863900303840637,
"step": 5116
},
{
"epoch": 3.874337623012869,
"grad_norm": 0.5311092138290405,
"learning_rate": 3.0736896636498844e-07,
"loss": 0.24167108535766602,
"step": 5118
},
{
"epoch": 3.8758516275548827,
"grad_norm": 0.8665949106216431,
"learning_rate": 3.0719367705868947e-07,
"loss": 0.11795948445796967,
"step": 5120
},
{
"epoch": 3.8773656320968963,
"grad_norm": 1.5483936071395874,
"learning_rate": 3.0702049220265463e-07,
"loss": 0.4619280695915222,
"step": 5122
},
{
"epoch": 3.87887963663891,
"grad_norm": 1.4960511922836304,
"learning_rate": 3.0684941206829945e-07,
"loss": 0.1171555370092392,
"step": 5124
},
{
"epoch": 3.8803936411809237,
"grad_norm": 0.2610466778278351,
"learning_rate": 3.0668043692374135e-07,
"loss": 0.0014266979414969683,
"step": 5126
},
{
"epoch": 3.8819076457229373,
"grad_norm": 0.6856014132499695,
"learning_rate": 3.06513567033799e-07,
"loss": 0.4256713390350342,
"step": 5128
},
{
"epoch": 3.8834216502649506,
"grad_norm": 1.56218421459198,
"learning_rate": 3.063488026599908e-07,
"loss": 0.5585575699806213,
"step": 5130
},
{
"epoch": 3.8849356548069647,
"grad_norm": 1.8463737964630127,
"learning_rate": 3.061861440605366e-07,
"loss": 0.461412250995636,
"step": 5132
},
{
"epoch": 3.886449659348978,
"grad_norm": 0.11369356513023376,
"learning_rate": 3.060255914903552e-07,
"loss": 0.4456888735294342,
"step": 5134
},
{
"epoch": 3.8879636638909916,
"grad_norm": 0.9624853134155273,
"learning_rate": 3.05867145201065e-07,
"loss": 0.1871456354856491,
"step": 5136
},
{
"epoch": 3.8894776684330052,
"grad_norm": 1.1454511880874634,
"learning_rate": 3.057108054409837e-07,
"loss": 0.45751628279685974,
"step": 5138
},
{
"epoch": 3.890991672975019,
"grad_norm": 1.262891411781311,
"learning_rate": 3.0555657245512737e-07,
"loss": 0.04787885770201683,
"step": 5140
},
{
"epoch": 3.8925056775170326,
"grad_norm": 2.1630172729492188,
"learning_rate": 3.0540444648521044e-07,
"loss": 0.47380274534225464,
"step": 5142
},
{
"epoch": 3.8940196820590463,
"grad_norm": 0.3003329634666443,
"learning_rate": 3.052544277696452e-07,
"loss": 0.024536890909075737,
"step": 5144
},
{
"epoch": 3.89553368660106,
"grad_norm": 1.0984807014465332,
"learning_rate": 3.0510651654354144e-07,
"loss": 0.08106108009815216,
"step": 5146
},
{
"epoch": 3.8970476911430736,
"grad_norm": 1.8982774019241333,
"learning_rate": 3.0496071303870617e-07,
"loss": 0.011285691522061825,
"step": 5148
},
{
"epoch": 3.8985616956850873,
"grad_norm": 1.2132909297943115,
"learning_rate": 3.0481701748364293e-07,
"loss": 0.7924096584320068,
"step": 5150
},
{
"epoch": 3.9000757002271005,
"grad_norm": 1.6980708837509155,
"learning_rate": 3.04675430103552e-07,
"loss": 0.8110805749893188,
"step": 5152
},
{
"epoch": 3.901589704769114,
"grad_norm": 0.3759213984012604,
"learning_rate": 3.045359511203297e-07,
"loss": 0.4529341161251068,
"step": 5154
},
{
"epoch": 3.903103709311128,
"grad_norm": 1.5288448333740234,
"learning_rate": 3.043985807525675e-07,
"loss": 0.555069625377655,
"step": 5156
},
{
"epoch": 3.9046177138531415,
"grad_norm": 2.1005005836486816,
"learning_rate": 3.04263319215553e-07,
"loss": 0.1001816838979721,
"step": 5158
},
{
"epoch": 3.906131718395155,
"grad_norm": 2.5000596046447754,
"learning_rate": 3.041301667212684e-07,
"loss": 0.5128546357154846,
"step": 5160
},
{
"epoch": 3.907645722937169,
"grad_norm": 2.431713581085205,
"learning_rate": 3.039991234783906e-07,
"loss": 0.02077617682516575,
"step": 5162
},
{
"epoch": 3.9091597274791825,
"grad_norm": 1.96620512008667,
"learning_rate": 3.0387018969229133e-07,
"loss": 0.02237214706838131,
"step": 5164
},
{
"epoch": 3.910673732021196,
"grad_norm": 2.4163711071014404,
"learning_rate": 3.0374336556503574e-07,
"loss": 0.8278625011444092,
"step": 5166
},
{
"epoch": 3.91218773656321,
"grad_norm": 2.0597403049468994,
"learning_rate": 3.0361865129538293e-07,
"loss": 0.655526876449585,
"step": 5168
},
{
"epoch": 3.913701741105223,
"grad_norm": 3.301499843597412,
"learning_rate": 3.0349604707878565e-07,
"loss": 0.23730042576789856,
"step": 5170
},
{
"epoch": 3.915215745647237,
"grad_norm": 2.0083491802215576,
"learning_rate": 3.033755531073897e-07,
"loss": 0.27088233828544617,
"step": 5172
},
{
"epoch": 3.9167297501892504,
"grad_norm": 3.793509006500244,
"learning_rate": 3.0325716957003346e-07,
"loss": 0.4128110408782959,
"step": 5174
},
{
"epoch": 3.918243754731264,
"grad_norm": 2.2577743530273438,
"learning_rate": 3.03140896652248e-07,
"loss": 0.9377378821372986,
"step": 5176
},
{
"epoch": 3.9197577592732777,
"grad_norm": 2.377145528793335,
"learning_rate": 3.0302673453625645e-07,
"loss": 0.4497866928577423,
"step": 5178
},
{
"epoch": 3.9212717638152914,
"grad_norm": 0.19746972620487213,
"learning_rate": 3.0291468340097433e-07,
"loss": 0.0009020717116072774,
"step": 5180
},
{
"epoch": 3.922785768357305,
"grad_norm": 0.45454391837120056,
"learning_rate": 3.0280474342200826e-07,
"loss": 0.1682163029909134,
"step": 5182
},
{
"epoch": 3.9242997728993188,
"grad_norm": 1.051390528678894,
"learning_rate": 3.0269691477165676e-07,
"loss": 0.07067491859197617,
"step": 5184
},
{
"epoch": 3.9258137774413324,
"grad_norm": 1.027337908744812,
"learning_rate": 3.025911976189091e-07,
"loss": 0.08587085455656052,
"step": 5186
},
{
"epoch": 3.927327781983346,
"grad_norm": 1.2078237533569336,
"learning_rate": 3.024875921294456e-07,
"loss": 0.30312326550483704,
"step": 5188
},
{
"epoch": 3.9288417865253598,
"grad_norm": 1.9159677028656006,
"learning_rate": 3.0238609846563696e-07,
"loss": 0.4570953845977783,
"step": 5190
},
{
"epoch": 3.930355791067373,
"grad_norm": 2.077888250350952,
"learning_rate": 3.0228671678654446e-07,
"loss": 0.34210488200187683,
"step": 5192
},
{
"epoch": 3.931869795609387,
"grad_norm": 1.5590956211090088,
"learning_rate": 3.0218944724791933e-07,
"loss": 0.641905665397644,
"step": 5194
},
{
"epoch": 3.9333838001514003,
"grad_norm": 0.030448101460933685,
"learning_rate": 3.020942900022029e-07,
"loss": 0.0016837774310261011,
"step": 5196
},
{
"epoch": 3.934897804693414,
"grad_norm": 0.7177050113677979,
"learning_rate": 3.020012451985256e-07,
"loss": 0.03333354741334915,
"step": 5198
},
{
"epoch": 3.9364118092354277,
"grad_norm": 4.787274360656738,
"learning_rate": 3.019103129827078e-07,
"loss": 0.09767407178878784,
"step": 5200
},
{
"epoch": 3.9379258137774413,
"grad_norm": 1.6870243549346924,
"learning_rate": 3.018214934972586e-07,
"loss": 0.4257047474384308,
"step": 5202
},
{
"epoch": 3.939439818319455,
"grad_norm": 1.786167025566101,
"learning_rate": 3.0173478688137626e-07,
"loss": 0.07438768446445465,
"step": 5204
},
{
"epoch": 3.9409538228614687,
"grad_norm": 0.9023681879043579,
"learning_rate": 3.0165019327094754e-07,
"loss": 0.43461889028549194,
"step": 5206
},
{
"epoch": 3.9424678274034823,
"grad_norm": 0.7739644646644592,
"learning_rate": 3.0156771279854786e-07,
"loss": 0.44720742106437683,
"step": 5208
},
{
"epoch": 3.9439818319454956,
"grad_norm": 0.15927131474018097,
"learning_rate": 3.0148734559344096e-07,
"loss": 0.0011613935930654407,
"step": 5210
},
{
"epoch": 3.9454958364875097,
"grad_norm": 2.753117799758911,
"learning_rate": 3.0140909178157846e-07,
"loss": 0.49144524335861206,
"step": 5212
},
{
"epoch": 3.947009841029523,
"grad_norm": 0.03960929438471794,
"learning_rate": 3.0133295148559994e-07,
"loss": 0.26274147629737854,
"step": 5214
},
{
"epoch": 3.9485238455715366,
"grad_norm": 4.2824177742004395,
"learning_rate": 3.0125892482483296e-07,
"loss": 0.5472410321235657,
"step": 5216
},
{
"epoch": 3.9500378501135502,
"grad_norm": 0.7858144044876099,
"learning_rate": 3.0118701191529207e-07,
"loss": 0.03546662628650665,
"step": 5218
},
{
"epoch": 3.951551854655564,
"grad_norm": 4.754327297210693,
"learning_rate": 3.011172128696795e-07,
"loss": 0.7001851797103882,
"step": 5220
},
{
"epoch": 3.9530658591975776,
"grad_norm": 5.084624767303467,
"learning_rate": 3.0104952779738463e-07,
"loss": 0.19021165370941162,
"step": 5222
},
{
"epoch": 3.9545798637395913,
"grad_norm": 2.424482583999634,
"learning_rate": 3.0098395680448356e-07,
"loss": 0.8408768773078918,
"step": 5224
},
{
"epoch": 3.956093868281605,
"grad_norm": 2.059044361114502,
"learning_rate": 3.0092049999373956e-07,
"loss": 0.4773581922054291,
"step": 5226
},
{
"epoch": 3.9576078728236186,
"grad_norm": 0.9264152646064758,
"learning_rate": 3.00859157464602e-07,
"loss": 0.039199769496917725,
"step": 5228
},
{
"epoch": 3.9591218773656323,
"grad_norm": 2.8337342739105225,
"learning_rate": 3.0079992931320743e-07,
"loss": 0.35346996784210205,
"step": 5230
},
{
"epoch": 3.9606358819076455,
"grad_norm": 0.22176103293895721,
"learning_rate": 3.0074281563237815e-07,
"loss": 0.042797479778528214,
"step": 5232
},
{
"epoch": 3.9621498864496596,
"grad_norm": 1.4219660758972168,
"learning_rate": 3.0068781651162303e-07,
"loss": 0.44434934854507446,
"step": 5234
},
{
"epoch": 3.963663890991673,
"grad_norm": 2.8318819999694824,
"learning_rate": 3.006349320371369e-07,
"loss": 0.5036948323249817,
"step": 5236
},
{
"epoch": 3.9651778955336865,
"grad_norm": 2.2482142448425293,
"learning_rate": 3.005841622918003e-07,
"loss": 0.7446083426475525,
"step": 5238
},
{
"epoch": 3.9666919000757,
"grad_norm": 2.282947063446045,
"learning_rate": 3.0053550735517994e-07,
"loss": 0.9360232949256897,
"step": 5240
},
{
"epoch": 3.968205904617714,
"grad_norm": 38.85729217529297,
"learning_rate": 3.004889673035278e-07,
"loss": 0.4850436747074127,
"step": 5242
},
{
"epoch": 3.9697199091597275,
"grad_norm": 5.286346912384033,
"learning_rate": 3.0044454220978176e-07,
"loss": 0.13758641481399536,
"step": 5244
},
{
"epoch": 3.971233913701741,
"grad_norm": 3.7393798828125,
"learning_rate": 3.004022321435649e-07,
"loss": 0.3747367262840271,
"step": 5246
},
{
"epoch": 3.972747918243755,
"grad_norm": 1.6003137826919556,
"learning_rate": 3.003620371711856e-07,
"loss": 0.4701666831970215,
"step": 5248
},
{
"epoch": 3.9742619227857685,
"grad_norm": 1.439442753791809,
"learning_rate": 3.0032395735563773e-07,
"loss": 0.3731074333190918,
"step": 5250
},
{
"epoch": 3.975775927327782,
"grad_norm": 1.7399569749832153,
"learning_rate": 3.0028799275659976e-07,
"loss": 0.663485050201416,
"step": 5252
},
{
"epoch": 3.9772899318697954,
"grad_norm": 1.0444694757461548,
"learning_rate": 3.002541434304361e-07,
"loss": 0.46440961956977844,
"step": 5254
},
{
"epoch": 3.978803936411809,
"grad_norm": 1.0770765542984009,
"learning_rate": 3.0022240943019483e-07,
"loss": 0.04875878617167473,
"step": 5256
},
{
"epoch": 3.9803179409538227,
"grad_norm": 1.9652454853057861,
"learning_rate": 3.001927908056102e-07,
"loss": 0.44029855728149414,
"step": 5258
},
{
"epoch": 3.9818319454958364,
"grad_norm": 6.682809352874756,
"learning_rate": 3.0016528760310013e-07,
"loss": 0.09948641061782837,
"step": 5260
},
{
"epoch": 3.98334595003785,
"grad_norm": 2.8983285427093506,
"learning_rate": 3.00139899865768e-07,
"loss": 0.6057555079460144,
"step": 5262
},
{
"epoch": 3.9848599545798638,
"grad_norm": 2.302161931991577,
"learning_rate": 3.001166276334015e-07,
"loss": 0.07627967745065689,
"step": 5264
},
{
"epoch": 3.9863739591218774,
"grad_norm": 1.456366777420044,
"learning_rate": 3.0009547094247273e-07,
"loss": 0.007791428826749325,
"step": 5266
},
{
"epoch": 3.987887963663891,
"grad_norm": 1.7385027408599854,
"learning_rate": 3.000764298261389e-07,
"loss": 0.5243682861328125,
"step": 5268
},
{
"epoch": 3.9894019682059048,
"grad_norm": 0.8515236377716064,
"learning_rate": 3.000595043142412e-07,
"loss": 0.04642985761165619,
"step": 5270
},
{
"epoch": 3.990915972747918,
"grad_norm": 0.29582443833351135,
"learning_rate": 3.0004469443330506e-07,
"loss": 0.05615620315074921,
"step": 5272
},
{
"epoch": 3.992429977289932,
"grad_norm": 3.1598730087280273,
"learning_rate": 3.0003200020654085e-07,
"loss": 0.7824563980102539,
"step": 5274
},
{
"epoch": 3.9939439818319453,
"grad_norm": 1.9997477531433105,
"learning_rate": 3.0002142165384287e-07,
"loss": 0.4484213590621948,
"step": 5276
},
{
"epoch": 3.995457986373959,
"grad_norm": 0.953460156917572,
"learning_rate": 3.0001295879179e-07,
"loss": 0.013839129358530045,
"step": 5278
},
{
"epoch": 3.9969719909159727,
"grad_norm": 2.7543108463287354,
"learning_rate": 3.0000661163364527e-07,
"loss": 0.4201052784919739,
"step": 5280
},
{
"epoch": 3.9984859954579863,
"grad_norm": 5.094695568084717,
"learning_rate": 3.000023801893557e-07,
"loss": 0.08943390846252441,
"step": 5282
},
{
"epoch": 4.0,
"grad_norm": 2.777651071548462,
"learning_rate": 3.0000026446555307e-07,
"loss": 0.7438609004020691,
"step": 5284
},
{
"epoch": 4.0,
"step": 5284,
"total_flos": 6.453258552541708e+18,
"train_loss": 0.5996934499091338,
"train_runtime": 17171.6272,
"train_samples_per_second": 4.923,
"train_steps_per_second": 0.308
}
],
"logging_steps": 2,
"max_steps": 5284,
"num_input_tokens_seen": 0,
"num_train_epochs": 4,
"save_steps": 99999,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": false,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 6.453258552541708e+18,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}