{
  "best_global_step": 5724,
  "best_metric": 0.4436005651950836,
  "best_model_checkpoint": "saves_multiple/lora/llama-3-8b-instruct/train_codealpacapy_42_1760705166/checkpoint-5724",
  "epoch": 20.0,
  "eval_steps": 1908,
  "global_step": 38160,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002620545073375262,
      "grad_norm": 0.8205227255821228,
      "learning_rate": 5.241090146750525e-08,
      "loss": 1.1832,
      "num_input_tokens_seen": 3264,
      "step": 5
    },
    {
      "epoch": 0.005241090146750524,
      "grad_norm": 0.8240294456481934,
      "learning_rate": 1.1792452830188679e-07,
      "loss": 1.2463,
      "num_input_tokens_seen": 5888,
      "step": 10
    },
    {
      "epoch": 0.007861635220125786,
      "grad_norm": 0.7573654651641846,
      "learning_rate": 1.8343815513626837e-07,
      "loss": 1.3478,
      "num_input_tokens_seen": 8480,
      "step": 15
    },
    {
      "epoch": 0.010482180293501049,
      "grad_norm": 2.1399152278900146,
      "learning_rate": 2.489517819706499e-07,
      "loss": 1.4246,
      "num_input_tokens_seen": 11488,
      "step": 20
    },
    {
      "epoch": 0.01310272536687631,
      "grad_norm": 1.2890939712524414,
      "learning_rate": 3.144654088050315e-07,
      "loss": 1.1789,
      "num_input_tokens_seen": 14880,
      "step": 25
    },
    {
      "epoch": 0.015723270440251572,
      "grad_norm": 0.8084590435028076,
      "learning_rate": 3.79979035639413e-07,
      "loss": 1.0941,
      "num_input_tokens_seen": 18592,
      "step": 30
    },
    {
      "epoch": 0.018343815513626835,
      "grad_norm": 1.2790471315383911,
      "learning_rate": 4.454926624737945e-07,
      "loss": 1.2534,
      "num_input_tokens_seen": 21568,
      "step": 35
    },
    {
      "epoch": 0.020964360587002098,
      "grad_norm": 1.1338529586791992,
      "learning_rate": 5.110062893081761e-07,
      "loss": 1.526,
      "num_input_tokens_seen": 24160,
      "step": 40
    },
    {
      "epoch": 0.02358490566037736,
      "grad_norm": 1.7503551244735718,
      "learning_rate": 5.765199161425577e-07,
      "loss": 1.4217,
      "num_input_tokens_seen": 27168,
      "step": 45
    },
    {
      "epoch": 0.02620545073375262,
      "grad_norm": 0.7311147451400757,
      "learning_rate": 6.420335429769392e-07,
      "loss": 1.1089,
      "num_input_tokens_seen": 30080,
      "step": 50
    },
    {
      "epoch": 0.028825995807127882,
      "grad_norm": 1.4762769937515259,
      "learning_rate": 7.075471698113208e-07,
      "loss": 1.4844,
      "num_input_tokens_seen": 32608,
      "step": 55
    },
    {
      "epoch": 0.031446540880503145,
      "grad_norm": 0.48823967576026917,
      "learning_rate": 7.730607966457024e-07,
      "loss": 1.167,
      "num_input_tokens_seen": 36448,
      "step": 60
    },
    {
      "epoch": 0.034067085953878404,
      "grad_norm": 0.8829717636108398,
      "learning_rate": 8.38574423480084e-07,
      "loss": 1.0816,
      "num_input_tokens_seen": 38944,
      "step": 65
    },
    {
      "epoch": 0.03668763102725367,
      "grad_norm": 2.8800909519195557,
      "learning_rate": 9.040880503144654e-07,
      "loss": 1.4655,
      "num_input_tokens_seen": 41824,
      "step": 70
    },
    {
      "epoch": 0.03930817610062893,
      "grad_norm": 0.8156294226646423,
      "learning_rate": 9.69601677148847e-07,
      "loss": 1.1356,
      "num_input_tokens_seen": 44896,
      "step": 75
    },
    {
      "epoch": 0.041928721174004195,
      "grad_norm": 1.2287204265594482,
      "learning_rate": 1.0351153039832286e-06,
      "loss": 1.1658,
      "num_input_tokens_seen": 47616,
      "step": 80
    },
    {
      "epoch": 0.044549266247379454,
      "grad_norm": 1.1691946983337402,
      "learning_rate": 1.10062893081761e-06,
      "loss": 1.3044,
      "num_input_tokens_seen": 50368,
      "step": 85
    },
    {
      "epoch": 0.04716981132075472,
      "grad_norm": 0.9649033546447754,
      "learning_rate": 1.1661425576519918e-06,
      "loss": 1.1184,
      "num_input_tokens_seen": 54016,
      "step": 90
    },
    {
      "epoch": 0.04979035639412998,
      "grad_norm": 0.5681169033050537,
      "learning_rate": 1.2316561844863732e-06,
      "loss": 0.8766,
      "num_input_tokens_seen": 57888,
      "step": 95
    },
    {
      "epoch": 0.05241090146750524,
      "grad_norm": 1.2145397663116455,
      "learning_rate": 1.2971698113207548e-06,
      "loss": 1.381,
      "num_input_tokens_seen": 61024,
      "step": 100
    },
    {
      "epoch": 0.055031446540880505,
      "grad_norm": 1.322643518447876,
      "learning_rate": 1.3626834381551364e-06,
      "loss": 1.1214,
      "num_input_tokens_seen": 64672,
      "step": 105
    },
    {
      "epoch": 0.057651991614255764,
      "grad_norm": 0.9954149723052979,
      "learning_rate": 1.4281970649895177e-06,
      "loss": 1.0507,
      "num_input_tokens_seen": 67616,
      "step": 110
    },
    {
      "epoch": 0.06027253668763103,
      "grad_norm": 0.508130669593811,
      "learning_rate": 1.4937106918238995e-06,
      "loss": 1.0299,
      "num_input_tokens_seen": 71232,
      "step": 115
    },
    {
      "epoch": 0.06289308176100629,
      "grad_norm": 1.5304887294769287,
      "learning_rate": 1.559224318658281e-06,
      "loss": 0.8744,
      "num_input_tokens_seen": 74880,
      "step": 120
    },
    {
      "epoch": 0.06551362683438156,
      "grad_norm": 3.4267847537994385,
      "learning_rate": 1.6247379454926625e-06,
      "loss": 1.1865,
      "num_input_tokens_seen": 78432,
      "step": 125
    },
    {
      "epoch": 0.06813417190775681,
      "grad_norm": 0.947415828704834,
      "learning_rate": 1.690251572327044e-06,
      "loss": 1.0666,
      "num_input_tokens_seen": 81088,
      "step": 130
    },
    {
      "epoch": 0.07075471698113207,
      "grad_norm": 0.8221093416213989,
      "learning_rate": 1.7557651991614257e-06,
      "loss": 0.9921,
      "num_input_tokens_seen": 84896,
      "step": 135
    },
    {
      "epoch": 0.07337526205450734,
      "grad_norm": 0.8060053586959839,
      "learning_rate": 1.821278825995807e-06,
      "loss": 1.2219,
      "num_input_tokens_seen": 88736,
      "step": 140
    },
    {
      "epoch": 0.0759958071278826,
      "grad_norm": 2.414795398712158,
      "learning_rate": 1.8867924528301887e-06,
      "loss": 1.5264,
      "num_input_tokens_seen": 92224,
      "step": 145
    },
    {
      "epoch": 0.07861635220125786,
      "grad_norm": 1.9082872867584229,
      "learning_rate": 1.9523060796645703e-06,
      "loss": 1.2471,
      "num_input_tokens_seen": 95168,
      "step": 150
    },
    {
      "epoch": 0.08123689727463312,
      "grad_norm": 1.640378475189209,
      "learning_rate": 2.017819706498952e-06,
      "loss": 0.9038,
      "num_input_tokens_seen": 98976,
      "step": 155
    },
    {
      "epoch": 0.08385744234800839,
      "grad_norm": 0.8147537708282471,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 0.8808,
      "num_input_tokens_seen": 102176,
      "step": 160
    },
    {
      "epoch": 0.08647798742138364,
      "grad_norm": 1.0699872970581055,
      "learning_rate": 2.148846960167715e-06,
      "loss": 0.851,
      "num_input_tokens_seen": 105824,
      "step": 165
    },
    {
      "epoch": 0.08909853249475891,
      "grad_norm": 1.042777419090271,
      "learning_rate": 2.2143605870020966e-06,
      "loss": 0.9619,
      "num_input_tokens_seen": 108512,
      "step": 170
    },
    {
      "epoch": 0.09171907756813417,
      "grad_norm": 1.1758551597595215,
      "learning_rate": 2.279874213836478e-06,
      "loss": 0.8462,
      "num_input_tokens_seen": 111872,
      "step": 175
    },
    {
      "epoch": 0.09433962264150944,
      "grad_norm": 2.27659010887146,
      "learning_rate": 2.34538784067086e-06,
      "loss": 0.8412,
      "num_input_tokens_seen": 115552,
      "step": 180
    },
    {
      "epoch": 0.09696016771488469,
      "grad_norm": 1.48320734500885,
      "learning_rate": 2.410901467505241e-06,
      "loss": 1.0189,
      "num_input_tokens_seen": 118784,
      "step": 185
    },
    {
      "epoch": 0.09958071278825996,
      "grad_norm": 2.00223970413208,
      "learning_rate": 2.476415094339623e-06,
      "loss": 0.7937,
      "num_input_tokens_seen": 122528,
      "step": 190
    },
    {
      "epoch": 0.10220125786163523,
      "grad_norm": 0.9536972045898438,
      "learning_rate": 2.5419287211740044e-06,
      "loss": 1.0417,
      "num_input_tokens_seen": 125024,
      "step": 195
    },
    {
      "epoch": 0.10482180293501048,
      "grad_norm": 1.1440117359161377,
      "learning_rate": 2.6074423480083857e-06,
      "loss": 0.9406,
      "num_input_tokens_seen": 128096,
      "step": 200
    },
    {
      "epoch": 0.10744234800838574,
      "grad_norm": 3.121914863586426,
      "learning_rate": 2.6729559748427675e-06,
      "loss": 1.07,
      "num_input_tokens_seen": 130496,
      "step": 205
    },
    {
      "epoch": 0.11006289308176101,
      "grad_norm": 1.0613490343093872,
      "learning_rate": 2.738469601677149e-06,
      "loss": 0.8269,
      "num_input_tokens_seen": 133152,
      "step": 210
    },
    {
      "epoch": 0.11268343815513626,
      "grad_norm": 2.5879688262939453,
      "learning_rate": 2.8039832285115303e-06,
      "loss": 1.0083,
      "num_input_tokens_seen": 136384,
      "step": 215
    },
    {
      "epoch": 0.11530398322851153,
      "grad_norm": 1.1826859712600708,
      "learning_rate": 2.869496855345912e-06,
      "loss": 0.8413,
      "num_input_tokens_seen": 139456,
      "step": 220
    },
    {
      "epoch": 0.1179245283018868,
      "grad_norm": 7.297121047973633,
      "learning_rate": 2.935010482180294e-06,
      "loss": 0.8547,
      "num_input_tokens_seen": 142816,
      "step": 225
    },
    {
      "epoch": 0.12054507337526206,
      "grad_norm": 1.19997239112854,
      "learning_rate": 3.0005241090146753e-06,
      "loss": 0.7547,
      "num_input_tokens_seen": 145696,
      "step": 230
    },
    {
      "epoch": 0.12316561844863731,
      "grad_norm": 0.9507931470870972,
      "learning_rate": 3.0660377358490567e-06,
      "loss": 0.8013,
      "num_input_tokens_seen": 149024,
      "step": 235
    },
    {
      "epoch": 0.12578616352201258,
      "grad_norm": 2.4328994750976562,
      "learning_rate": 3.131551362683438e-06,
      "loss": 0.7734,
      "num_input_tokens_seen": 151840,
      "step": 240
    },
    {
      "epoch": 0.12840670859538783,
      "grad_norm": 0.805752158164978,
      "learning_rate": 3.1970649895178203e-06,
      "loss": 0.8132,
      "num_input_tokens_seen": 155456,
      "step": 245
    },
    {
      "epoch": 0.1310272536687631,
      "grad_norm": 1.6080076694488525,
      "learning_rate": 3.2625786163522012e-06,
      "loss": 0.6914,
      "num_input_tokens_seen": 158528,
      "step": 250
    },
    {
      "epoch": 0.13364779874213836,
      "grad_norm": 1.2698324918746948,
      "learning_rate": 3.3280922431865826e-06,
      "loss": 0.6045,
      "num_input_tokens_seen": 166496,
      "step": 255
    },
    {
      "epoch": 0.13626834381551362,
      "grad_norm": 0.832471489906311,
      "learning_rate": 3.393605870020965e-06,
      "loss": 0.5689,
      "num_input_tokens_seen": 169568,
      "step": 260
    },
    {
      "epoch": 0.1388888888888889,
      "grad_norm": 1.864088773727417,
      "learning_rate": 3.459119496855346e-06,
      "loss": 0.7277,
      "num_input_tokens_seen": 172704,
      "step": 265
    },
    {
      "epoch": 0.14150943396226415,
      "grad_norm": 0.8005948662757874,
      "learning_rate": 3.524633123689727e-06,
      "loss": 0.7914,
      "num_input_tokens_seen": 176480,
      "step": 270
    },
    {
      "epoch": 0.1441299790356394,
      "grad_norm": 0.5403485298156738,
      "learning_rate": 3.5901467505241094e-06,
      "loss": 0.6047,
      "num_input_tokens_seen": 181408,
      "step": 275
    },
    {
      "epoch": 0.14675052410901468,
      "grad_norm": 0.7174608707427979,
      "learning_rate": 3.6556603773584908e-06,
      "loss": 0.702,
      "num_input_tokens_seen": 184736,
      "step": 280
    },
    {
      "epoch": 0.14937106918238993,
      "grad_norm": 0.9025600552558899,
      "learning_rate": 3.7211740041928726e-06,
      "loss": 0.5998,
      "num_input_tokens_seen": 188352,
      "step": 285
    },
    {
      "epoch": 0.1519916142557652,
      "grad_norm": 1.8641772270202637,
      "learning_rate": 3.786687631027254e-06,
      "loss": 0.5684,
      "num_input_tokens_seen": 193408,
      "step": 290
    },
    {
      "epoch": 0.15461215932914046,
      "grad_norm": 0.8043381571769714,
      "learning_rate": 3.852201257861635e-06,
      "loss": 0.4496,
      "num_input_tokens_seen": 196384,
      "step": 295
    },
    {
      "epoch": 0.15723270440251572,
      "grad_norm": 1.5230621099472046,
      "learning_rate": 3.917714884696017e-06,
      "loss": 0.6989,
      "num_input_tokens_seen": 199360,
      "step": 300
    },
    {
      "epoch": 0.159853249475891,
      "grad_norm": 0.6261242628097534,
      "learning_rate": 3.9832285115303985e-06,
      "loss": 0.593,
      "num_input_tokens_seen": 202848,
      "step": 305
    },
    {
      "epoch": 0.16247379454926625,
      "grad_norm": 0.9283719062805176,
      "learning_rate": 4.04874213836478e-06,
      "loss": 0.5252,
      "num_input_tokens_seen": 207104,
      "step": 310
    },
    {
      "epoch": 0.1650943396226415,
      "grad_norm": 1.4439867734909058,
      "learning_rate": 4.114255765199162e-06,
      "loss": 0.7438,
      "num_input_tokens_seen": 209600,
      "step": 315
    },
    {
      "epoch": 0.16771488469601678,
      "grad_norm": 0.6770917773246765,
      "learning_rate": 4.1797693920335435e-06,
      "loss": 0.654,
      "num_input_tokens_seen": 213152,
      "step": 320
    },
    {
      "epoch": 0.17033542976939203,
      "grad_norm": 2.0970702171325684,
      "learning_rate": 4.245283018867925e-06,
      "loss": 0.7399,
      "num_input_tokens_seen": 215616,
      "step": 325
    },
    {
      "epoch": 0.17295597484276728,
      "grad_norm": 0.7083891034126282,
      "learning_rate": 4.310796645702306e-06,
      "loss": 0.6385,
      "num_input_tokens_seen": 218528,
      "step": 330
    },
    {
      "epoch": 0.17557651991614256,
      "grad_norm": 2.4348249435424805,
      "learning_rate": 4.376310272536688e-06,
      "loss": 0.5043,
      "num_input_tokens_seen": 221824,
      "step": 335
    },
    {
      "epoch": 0.17819706498951782,
      "grad_norm": 0.5274974703788757,
      "learning_rate": 4.441823899371069e-06,
      "loss": 0.5674,
      "num_input_tokens_seen": 225280,
      "step": 340
    },
    {
      "epoch": 0.18081761006289307,
      "grad_norm": 0.9751011729240417,
      "learning_rate": 4.507337526205451e-06,
      "loss": 0.4653,
      "num_input_tokens_seen": 230912,
      "step": 345
    },
    {
      "epoch": 0.18343815513626835,
      "grad_norm": 1.0073169469833374,
      "learning_rate": 4.572851153039833e-06,
      "loss": 0.5789,
      "num_input_tokens_seen": 234592,
      "step": 350
    },
    {
      "epoch": 0.1860587002096436,
      "grad_norm": 0.6552371978759766,
      "learning_rate": 4.638364779874214e-06,
      "loss": 0.4994,
      "num_input_tokens_seen": 237600,
      "step": 355
    },
    {
      "epoch": 0.18867924528301888,
      "grad_norm": 0.8065017461776733,
      "learning_rate": 4.703878406708595e-06,
      "loss": 0.6418,
      "num_input_tokens_seen": 241088,
      "step": 360
    },
    {
      "epoch": 0.19129979035639413,
      "grad_norm": 1.2085505723953247,
      "learning_rate": 4.769392033542977e-06,
      "loss": 0.5864,
      "num_input_tokens_seen": 243520,
      "step": 365
    },
    {
      "epoch": 0.19392033542976939,
      "grad_norm": 1.486066222190857,
      "learning_rate": 4.834905660377359e-06,
      "loss": 0.7582,
      "num_input_tokens_seen": 247456,
      "step": 370
    },
    {
      "epoch": 0.19654088050314467,
      "grad_norm": 0.639738142490387,
      "learning_rate": 4.90041928721174e-06,
      "loss": 0.5221,
      "num_input_tokens_seen": 250784,
      "step": 375
    },
    {
      "epoch": 0.19916142557651992,
      "grad_norm": 1.2019051313400269,
      "learning_rate": 4.965932914046122e-06,
      "loss": 0.5538,
      "num_input_tokens_seen": 255040,
      "step": 380
    },
    {
      "epoch": 0.20178197064989517,
      "grad_norm": 0.9065296649932861,
      "learning_rate": 5.031446540880504e-06,
      "loss": 0.6454,
      "num_input_tokens_seen": 258240,
      "step": 385
    },
    {
      "epoch": 0.20440251572327045,
      "grad_norm": 0.4876384437084198,
      "learning_rate": 5.096960167714885e-06,
      "loss": 0.4812,
      "num_input_tokens_seen": 261152,
      "step": 390
    },
    {
      "epoch": 0.2070230607966457,
      "grad_norm": 0.8193243741989136,
      "learning_rate": 5.162473794549266e-06,
      "loss": 0.592,
      "num_input_tokens_seen": 264544,
      "step": 395
    },
    {
      "epoch": 0.20964360587002095,
      "grad_norm": 1.3125677108764648,
      "learning_rate": 5.227987421383648e-06,
      "loss": 0.6587,
      "num_input_tokens_seen": 267136,
      "step": 400
    },
    {
      "epoch": 0.21226415094339623,
      "grad_norm": 0.9630743861198425,
      "learning_rate": 5.2935010482180295e-06,
      "loss": 0.5139,
      "num_input_tokens_seen": 270464,
      "step": 405
    },
    {
      "epoch": 0.2148846960167715,
      "grad_norm": 0.9439541697502136,
      "learning_rate": 5.359014675052411e-06,
      "loss": 0.5254,
      "num_input_tokens_seen": 274528,
      "step": 410
    },
    {
      "epoch": 0.21750524109014674,
      "grad_norm": 1.0347297191619873,
      "learning_rate": 5.424528301886793e-06,
      "loss": 0.5191,
      "num_input_tokens_seen": 277376,
      "step": 415
    },
    {
      "epoch": 0.22012578616352202,
      "grad_norm": 1.2875181436538696,
      "learning_rate": 5.4900419287211744e-06,
      "loss": 0.4459,
      "num_input_tokens_seen": 279616,
      "step": 420
    },
    {
      "epoch": 0.22274633123689727,
      "grad_norm": 1.5931833982467651,
      "learning_rate": 5.555555555555556e-06,
      "loss": 0.4737,
      "num_input_tokens_seen": 282656,
      "step": 425
    },
    {
      "epoch": 0.22536687631027252,
      "grad_norm": 0.7922357320785522,
      "learning_rate": 5.621069182389937e-06,
      "loss": 0.5683,
      "num_input_tokens_seen": 285600,
      "step": 430
    },
    {
      "epoch": 0.2279874213836478,
      "grad_norm": 0.5078657269477844,
      "learning_rate": 5.686582809224319e-06,
      "loss": 0.4026,
      "num_input_tokens_seen": 289568,
      "step": 435
    },
    {
      "epoch": 0.23060796645702306,
      "grad_norm": 0.9720231890678406,
      "learning_rate": 5.752096436058701e-06,
      "loss": 0.5869,
      "num_input_tokens_seen": 292768,
      "step": 440
    },
    {
      "epoch": 0.23322851153039834,
      "grad_norm": 0.9289599657058716,
      "learning_rate": 5.817610062893082e-06,
      "loss": 0.6488,
      "num_input_tokens_seen": 295456,
      "step": 445
    },
    {
      "epoch": 0.2358490566037736,
      "grad_norm": 0.9577594995498657,
      "learning_rate": 5.8831236897274636e-06,
      "loss": 0.5954,
      "num_input_tokens_seen": 299488,
      "step": 450
    },
    {
      "epoch": 0.23846960167714884,
      "grad_norm": 1.2273191213607788,
      "learning_rate": 5.948637316561846e-06,
      "loss": 0.4674,
      "num_input_tokens_seen": 302976,
      "step": 455
    },
    {
      "epoch": 0.24109014675052412,
      "grad_norm": 0.961017906665802,
      "learning_rate": 6.014150943396227e-06,
      "loss": 0.5378,
      "num_input_tokens_seen": 306560,
      "step": 460
    },
    {
      "epoch": 0.24371069182389937,
      "grad_norm": 0.8706506490707397,
      "learning_rate": 6.079664570230608e-06,
      "loss": 0.4583,
      "num_input_tokens_seen": 309536,
      "step": 465
    },
    {
      "epoch": 0.24633123689727462,
      "grad_norm": 0.8577477931976318,
      "learning_rate": 6.14517819706499e-06,
      "loss": 0.5143,
      "num_input_tokens_seen": 312576,
      "step": 470
    },
    {
      "epoch": 0.2489517819706499,
      "grad_norm": 0.7478653192520142,
      "learning_rate": 6.210691823899371e-06,
      "loss": 0.5207,
      "num_input_tokens_seen": 315616,
      "step": 475
    },
    {
      "epoch": 0.25157232704402516,
      "grad_norm": 0.9047702550888062,
      "learning_rate": 6.2762054507337535e-06,
      "loss": 0.5284,
      "num_input_tokens_seen": 319136,
      "step": 480
    },
    {
      "epoch": 0.25419287211740044,
      "grad_norm": 1.1311345100402832,
      "learning_rate": 6.341719077568134e-06,
      "loss": 0.692,
      "num_input_tokens_seen": 321376,
      "step": 485
    },
    {
      "epoch": 0.25681341719077566,
      "grad_norm": 1.6585030555725098,
      "learning_rate": 6.407232704402516e-06,
      "loss": 0.586,
      "num_input_tokens_seen": 324928,
      "step": 490
    },
    {
      "epoch": 0.25943396226415094,
      "grad_norm": 1.1445555686950684,
      "learning_rate": 6.472746331236898e-06,
      "loss": 0.6713,
      "num_input_tokens_seen": 329248,
      "step": 495
    },
    {
      "epoch": 0.2620545073375262,
      "grad_norm": 0.8413041234016418,
      "learning_rate": 6.538259958071278e-06,
      "loss": 0.672,
      "num_input_tokens_seen": 332288,
      "step": 500
    },
    {
      "epoch": 0.26467505241090145,
      "grad_norm": 1.0694005489349365,
      "learning_rate": 6.60377358490566e-06,
      "loss": 0.5511,
      "num_input_tokens_seen": 334752,
      "step": 505
    },
    {
      "epoch": 0.2672955974842767,
      "grad_norm": 0.7332883477210999,
      "learning_rate": 6.669287211740043e-06,
      "loss": 0.5566,
      "num_input_tokens_seen": 337856,
      "step": 510
    },
    {
      "epoch": 0.269916142557652,
      "grad_norm": 0.6495680809020996,
      "learning_rate": 6.734800838574423e-06,
      "loss": 0.6574,
      "num_input_tokens_seen": 341792,
      "step": 515
    },
    {
      "epoch": 0.27253668763102723,
      "grad_norm": 0.8753320574760437,
      "learning_rate": 6.800314465408805e-06,
      "loss": 0.6471,
      "num_input_tokens_seen": 344768,
      "step": 520
    },
    {
      "epoch": 0.2751572327044025,
      "grad_norm": 1.0310944318771362,
      "learning_rate": 6.865828092243188e-06,
      "loss": 0.4742,
      "num_input_tokens_seen": 347360,
      "step": 525
    },
    {
      "epoch": 0.2777777777777778,
      "grad_norm": 0.8016573190689087,
      "learning_rate": 6.931341719077568e-06,
      "loss": 0.489,
      "num_input_tokens_seen": 350592,
      "step": 530
    },
    {
      "epoch": 0.280398322851153,
      "grad_norm": 1.0115299224853516,
      "learning_rate": 6.9968553459119495e-06,
      "loss": 0.4979,
      "num_input_tokens_seen": 353568,
      "step": 535
    },
    {
      "epoch": 0.2830188679245283,
      "grad_norm": 0.6745246052742004,
      "learning_rate": 7.062368972746332e-06,
      "loss": 0.6198,
      "num_input_tokens_seen": 356320,
      "step": 540
    },
    {
      "epoch": 0.2856394129979036,
      "grad_norm": 1.1727603673934937,
      "learning_rate": 7.127882599580712e-06,
      "loss": 0.5123,
      "num_input_tokens_seen": 359968,
      "step": 545
    },
    {
      "epoch": 0.2882599580712788,
      "grad_norm": 1.0280649662017822,
      "learning_rate": 7.1933962264150945e-06,
      "loss": 0.4506,
      "num_input_tokens_seen": 362752,
      "step": 550
    },
    {
      "epoch": 0.2908805031446541,
      "grad_norm": 1.2703386545181274,
      "learning_rate": 7.258909853249477e-06,
      "loss": 0.5807,
      "num_input_tokens_seen": 365312,
      "step": 555
    },
    {
      "epoch": 0.29350104821802936,
      "grad_norm": 0.6820155382156372,
      "learning_rate": 7.324423480083858e-06,
      "loss": 0.4796,
      "num_input_tokens_seen": 368896,
      "step": 560
    },
    {
      "epoch": 0.29612159329140464,
      "grad_norm": 0.894085168838501,
      "learning_rate": 7.3899371069182395e-06,
      "loss": 0.5394,
      "num_input_tokens_seen": 371648,
      "step": 565
    },
    {
      "epoch": 0.29874213836477986,
      "grad_norm": 0.8079003691673279,
      "learning_rate": 7.455450733752621e-06,
      "loss": 0.5959,
      "num_input_tokens_seen": 374848,
      "step": 570
    },
    {
      "epoch": 0.30136268343815514,
      "grad_norm": 1.1119354963302612,
      "learning_rate": 7.520964360587003e-06,
      "loss": 0.5357,
      "num_input_tokens_seen": 377184,
      "step": 575
    },
    {
      "epoch": 0.3039832285115304,
      "grad_norm": 0.6858542561531067,
      "learning_rate": 7.586477987421384e-06,
      "loss": 0.6301,
      "num_input_tokens_seen": 381504,
      "step": 580
    },
    {
      "epoch": 0.30660377358490565,
      "grad_norm": 1.1593818664550781,
      "learning_rate": 7.651991614255765e-06,
      "loss": 0.5376,
      "num_input_tokens_seen": 384672,
      "step": 585
    },
    {
      "epoch": 0.30922431865828093,
      "grad_norm": 0.9169436097145081,
      "learning_rate": 7.717505241090148e-06,
      "loss": 0.5333,
      "num_input_tokens_seen": 389376,
      "step": 590
    },
    {
      "epoch": 0.3118448637316562,
      "grad_norm": 0.748104989528656,
      "learning_rate": 7.783018867924528e-06,
      "loss": 0.5857,
      "num_input_tokens_seen": 392512,
      "step": 595
    },
    {
      "epoch": 0.31446540880503143,
      "grad_norm": 0.8801977634429932,
      "learning_rate": 7.848532494758911e-06,
      "loss": 0.5178,
      "num_input_tokens_seen": 396864,
      "step": 600
    },
    {
      "epoch": 0.3170859538784067,
      "grad_norm": 1.2647403478622437,
      "learning_rate": 7.914046121593292e-06,
      "loss": 0.4808,
      "num_input_tokens_seen": 399520,
      "step": 605
    },
    {
      "epoch": 0.319706498951782,
      "grad_norm": 5.327150344848633,
      "learning_rate": 7.979559748427674e-06,
      "loss": 0.5141,
      "num_input_tokens_seen": 402560,
      "step": 610
    },
    {
      "epoch": 0.3223270440251572,
      "grad_norm": 1.2796356678009033,
      "learning_rate": 8.045073375262055e-06,
      "loss": 0.6383,
      "num_input_tokens_seen": 406240,
      "step": 615
    },
    {
      "epoch": 0.3249475890985325,
      "grad_norm": 1.2944592237472534,
      "learning_rate": 8.110587002096436e-06,
      "loss": 0.5466,
      "num_input_tokens_seen": 408960,
      "step": 620
    },
    {
      "epoch": 0.3275681341719078,
      "grad_norm": 0.9365005493164062,
      "learning_rate": 8.176100628930818e-06,
      "loss": 0.6671,
      "num_input_tokens_seen": 412224,
      "step": 625
    },
    {
      "epoch": 0.330188679245283,
      "grad_norm": 0.8552364110946655,
      "learning_rate": 8.241614255765199e-06,
      "loss": 0.4833,
      "num_input_tokens_seen": 415296,
      "step": 630
    },
    {
      "epoch": 0.3328092243186583,
      "grad_norm": 1.4086257219314575,
      "learning_rate": 8.307127882599582e-06,
      "loss": 0.442,
      "num_input_tokens_seen": 417504,
      "step": 635
    },
    {
      "epoch": 0.33542976939203356,
      "grad_norm": 0.9798508286476135,
      "learning_rate": 8.372641509433962e-06,
      "loss": 0.5329,
      "num_input_tokens_seen": 420288,
      "step": 640
    },
    {
      "epoch": 0.3380503144654088,
      "grad_norm": 1.0245188474655151,
      "learning_rate": 8.438155136268345e-06,
      "loss": 0.6621,
      "num_input_tokens_seen": 422560,
      "step": 645
    },
    {
      "epoch": 0.34067085953878407,
      "grad_norm": 1.4904325008392334,
      "learning_rate": 8.503668763102726e-06,
      "loss": 0.5998,
      "num_input_tokens_seen": 425472,
      "step": 650
    },
    {
      "epoch": 0.34329140461215935,
      "grad_norm": 1.0816766023635864,
      "learning_rate": 8.569182389937106e-06,
      "loss": 0.5169,
      "num_input_tokens_seen": 429056,
      "step": 655
    },
    {
      "epoch": 0.34591194968553457,
      "grad_norm": 0.8332240581512451,
      "learning_rate": 8.634696016771489e-06,
      "loss": 0.4821,
      "num_input_tokens_seen": 433024,
      "step": 660
    },
    {
      "epoch": 0.34853249475890985,
      "grad_norm": 0.957460880279541,
      "learning_rate": 8.70020964360587e-06,
      "loss": 0.541,
      "num_input_tokens_seen": 435776,
      "step": 665
    },
    {
      "epoch": 0.35115303983228513,
      "grad_norm": 0.7531293630599976,
      "learning_rate": 8.765723270440252e-06,
      "loss": 0.5098,
      "num_input_tokens_seen": 438432,
      "step": 670
    },
    {
      "epoch": 0.35377358490566035,
      "grad_norm": 1.0103346109390259,
      "learning_rate": 8.831236897274633e-06,
      "loss": 0.5123,
      "num_input_tokens_seen": 442400,
      "step": 675
    },
    {
      "epoch": 0.35639412997903563,
      "grad_norm": 1.4068387746810913,
      "learning_rate": 8.896750524109015e-06,
      "loss": 0.6015,
      "num_input_tokens_seen": 445888,
      "step": 680
    },
    {
      "epoch": 0.3590146750524109,
      "grad_norm": 0.4443424344062805,
      "learning_rate": 8.962264150943396e-06,
      "loss": 0.4717,
      "num_input_tokens_seen": 448864,
      "step": 685
    },
    {
      "epoch": 0.36163522012578614,
      "grad_norm": 0.9575173258781433,
      "learning_rate": 9.027777777777777e-06,
      "loss": 0.6748,
      "num_input_tokens_seen": 452640,
      "step": 690
    },
    {
      "epoch": 0.3642557651991614,
      "grad_norm": 0.9844288229942322,
      "learning_rate": 9.09329140461216e-06,
      "loss": 0.6025,
      "num_input_tokens_seen": 455296,
      "step": 695
    },
    {
      "epoch": 0.3668763102725367,
      "grad_norm": 2.342177629470825,
      "learning_rate": 9.15880503144654e-06,
      "loss": 0.5342,
      "num_input_tokens_seen": 457888,
      "step": 700
    },
    {
      "epoch": 0.3694968553459119,
      "grad_norm": 0.9594718813896179,
      "learning_rate": 9.224318658280923e-06,
      "loss": 0.6051,
      "num_input_tokens_seen": 461664,
      "step": 705
    },
    {
      "epoch": 0.3721174004192872,
      "grad_norm": 0.8786901831626892,
      "learning_rate": 9.289832285115305e-06,
      "loss": 0.5981,
      "num_input_tokens_seen": 464352,
      "step": 710
    },
    {
      "epoch": 0.3747379454926625,
      "grad_norm": 0.9277939796447754,
      "learning_rate": 9.355345911949686e-06,
      "loss": 0.6076,
      "num_input_tokens_seen": 468256,
      "step": 715
    },
    {
      "epoch": 0.37735849056603776,
      "grad_norm": 2.679940700531006,
      "learning_rate": 9.420859538784067e-06,
      "loss": 0.4993,
      "num_input_tokens_seen": 471104,
      "step": 720
    },
    {
      "epoch": 0.379979035639413,
      "grad_norm": 2.5472517013549805,
      "learning_rate": 9.486373165618449e-06,
      "loss": 0.6031,
      "num_input_tokens_seen": 473696,
      "step": 725
    },
    {
      "epoch": 0.38259958071278827,
      "grad_norm": 0.9716710448265076,
      "learning_rate": 9.551886792452832e-06,
      "loss": 0.5003,
      "num_input_tokens_seen": 476512,
      "step": 730
    },
    {
      "epoch": 0.38522012578616355,
      "grad_norm": 0.7798869013786316,
      "learning_rate": 9.617400419287211e-06,
      "loss": 0.5751,
      "num_input_tokens_seen": 480128,
      "step": 735
    },
    {
      "epoch": 0.38784067085953877,
      "grad_norm": 1.7348568439483643,
      "learning_rate": 9.682914046121595e-06,
      "loss": 0.6421,
      "num_input_tokens_seen": 483808,
      "step": 740
    },
    {
      "epoch": 0.39046121593291405,
      "grad_norm": 1.4418957233428955,
      "learning_rate": 9.748427672955976e-06,
      "loss": 0.5895,
      "num_input_tokens_seen": 486240,
      "step": 745
    },
    {
      "epoch": 0.39308176100628933,
      "grad_norm": 0.8076645731925964,
      "learning_rate": 9.813941299790357e-06,
      "loss": 0.4405,
      "num_input_tokens_seen": 488448,
      "step": 750
    },
    {
      "epoch": 0.39570230607966456,
      "grad_norm": 1.7293415069580078,
      "learning_rate": 9.879454926624739e-06,
      "loss": 0.5547,
      "num_input_tokens_seen": 491392,
      "step": 755
    },
    {
      "epoch": 0.39832285115303984,
      "grad_norm": 1.342331051826477,
      "learning_rate": 9.94496855345912e-06,
      "loss": 0.5118,
      "num_input_tokens_seen": 495232,
      "step": 760
    },
    {
      "epoch": 0.4009433962264151,
      "grad_norm": 1.5226211547851562,
      "learning_rate": 1.0010482180293501e-05,
      "loss": 0.4765,
      "num_input_tokens_seen": 498016,
      "step": 765
    },
    {
      "epoch": 0.40356394129979034,
      "grad_norm": 1.05174720287323,
      "learning_rate": 1.0075995807127883e-05,
      "loss": 0.4893,
      "num_input_tokens_seen": 501152,
      "step": 770
    },
    {
      "epoch": 0.4061844863731656,
      "grad_norm": 0.835838258266449,
      "learning_rate": 1.0141509433962266e-05,
      "loss": 0.4828,
      "num_input_tokens_seen": 503872,
      "step": 775
    },
    {
      "epoch": 0.4088050314465409,
      "grad_norm": 1.1031776666641235,
      "learning_rate": 1.0207023060796646e-05,
      "loss": 0.507,
      "num_input_tokens_seen": 507328,
      "step": 780
    },
    {
      "epoch": 0.4114255765199161,
      "grad_norm": 0.7189106345176697,
      "learning_rate": 1.0272536687631027e-05,
      "loss": 0.3975,
      "num_input_tokens_seen": 510912,
      "step": 785
    },
    {
      "epoch": 0.4140461215932914,
      "grad_norm": 0.7598391771316528,
      "learning_rate": 1.033805031446541e-05,
      "loss": 0.5741,
      "num_input_tokens_seen": 514304,
      "step": 790
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 1.5290573835372925,
      "learning_rate": 1.040356394129979e-05,
      "loss": 0.429,
      "num_input_tokens_seen": 517440,
      "step": 795
    },
    {
      "epoch": 0.4192872117400419,
      "grad_norm": 1.433996558189392,
      "learning_rate": 1.0469077568134173e-05,
      "loss": 0.6307,
      "num_input_tokens_seen": 520416,
      "step": 800
    },
    {
      "epoch": 0.4219077568134172,
      "grad_norm": 1.1718108654022217,
      "learning_rate": 1.0534591194968554e-05,
      "loss": 0.6155,
      "num_input_tokens_seen": 523456,
      "step": 805
    },
    {
      "epoch": 0.42452830188679247,
      "grad_norm": 0.8563489317893982,
      "learning_rate": 1.0600104821802936e-05,
      "loss": 0.48,
      "num_input_tokens_seen": 526720,
      "step": 810
    },
    {
      "epoch": 0.4271488469601677,
      "grad_norm": 1.3665646314620972,
      "learning_rate": 1.0665618448637317e-05,
      "loss": 0.4073,
      "num_input_tokens_seen": 530272,
      "step": 815
    },
    {
      "epoch": 0.429769392033543,
      "grad_norm": 1.382773995399475,
      "learning_rate": 1.0731132075471698e-05,
      "loss": 0.6008,
      "num_input_tokens_seen": 533408,
      "step": 820
    },
    {
      "epoch": 0.43238993710691825,
      "grad_norm": 3.7248806953430176,
      "learning_rate": 1.079664570230608e-05,
      "loss": 0.6423,
      "num_input_tokens_seen": 536768,
      "step": 825
    },
    {
      "epoch": 0.4350104821802935,
      "grad_norm": 0.51448655128479,
      "learning_rate": 1.0862159329140461e-05,
      "loss": 0.5863,
      "num_input_tokens_seen": 541184,
      "step": 830
    },
    {
      "epoch": 0.43763102725366876,
      "grad_norm": 1.3559582233428955,
      "learning_rate": 1.0927672955974844e-05,
      "loss": 0.4385,
      "num_input_tokens_seen": 544576,
      "step": 835
    },
    {
      "epoch": 0.44025157232704404,
      "grad_norm": 0.9712456464767456,
      "learning_rate": 1.0993186582809224e-05,
      "loss": 0.5698,
      "num_input_tokens_seen": 549408,
      "step": 840
    },
    {
      "epoch": 0.44287211740041926,
      "grad_norm": 1.5917351245880127,
      "learning_rate": 1.1058700209643607e-05,
      "loss": 0.4578,
      "num_input_tokens_seen": 552032,
      "step": 845
    },
    {
      "epoch": 0.44549266247379454,
      "grad_norm": 1.2992106676101685,
      "learning_rate": 1.1124213836477988e-05,
      "loss": 0.4871,
      "num_input_tokens_seen": 555872,
      "step": 850
    },
    {
      "epoch": 0.4481132075471698,
      "grad_norm": 1.0848115682601929,
      "learning_rate": 1.118972746331237e-05,
      "loss": 0.6104,
      "num_input_tokens_seen": 559136,
      "step": 855
    },
    {
      "epoch": 0.45073375262054505,
      "grad_norm": 0.7571024894714355,
      "learning_rate": 1.1255241090146751e-05,
      "loss": 0.5126,
      "num_input_tokens_seen": 561984,
      "step": 860
    },
    {
      "epoch": 0.4533542976939203,
      "grad_norm": 0.7712518572807312,
      "learning_rate": 1.1320754716981132e-05,
      "loss": 0.4655,
      "num_input_tokens_seen": 564448,
      "step": 865
    },
    {
      "epoch": 0.4559748427672956,
      "grad_norm": 1.5538264513015747,
      "learning_rate": 1.1386268343815514e-05,
      "loss": 0.5675,
      "num_input_tokens_seen": 566752,
      "step": 870
    },
    {
      "epoch": 0.4585953878406709,
      "grad_norm": 0.9733453989028931,
      "learning_rate": 1.1451781970649895e-05,
      "loss": 0.5029,
      "num_input_tokens_seen": 569856,
      "step": 875
    },
    {
      "epoch": 0.4612159329140461,
      "grad_norm": 1.1016618013381958,
      "learning_rate": 1.1517295597484278e-05,
      "loss": 0.4912,
      "num_input_tokens_seen": 573056,
      "step": 880
    },
    {
      "epoch": 0.4638364779874214,
      "grad_norm": 1.1775985956192017,
      "learning_rate": 1.158280922431866e-05,
      "loss": 0.5648,
      "num_input_tokens_seen": 575712,
      "step": 885
    },
    {
      "epoch": 0.46645702306079667,
      "grad_norm": 0.8499876260757446,
      "learning_rate": 1.164832285115304e-05,
      "loss": 0.4117,
      "num_input_tokens_seen": 578880,
      "step": 890
    },
    {
      "epoch": 0.4690775681341719,
      "grad_norm": 0.7927697896957397,
      "learning_rate": 1.1713836477987422e-05,
      "loss": 0.4729,
      "num_input_tokens_seen": 581920,
      "step": 895
    },
    {
      "epoch": 0.4716981132075472,
      "grad_norm": 0.9637202024459839,
      "learning_rate": 1.1779350104821804e-05,
      "loss": 0.4929,
      "num_input_tokens_seen": 584672,
      "step": 900
    },
    {
      "epoch": 0.47431865828092246,
      "grad_norm": 1.097861647605896,
      "learning_rate": 1.1844863731656185e-05,
      "loss": 0.4797,
      "num_input_tokens_seen": 588704,
      "step": 905
    },
    {
      "epoch": 0.4769392033542977,
      "grad_norm": 1.5111232995986938,
      "learning_rate": 1.1910377358490566e-05,
      "loss": 0.4715,
      "num_input_tokens_seen": 591712,
      "step": 910
    },
    {
      "epoch": 0.47955974842767296,
      "grad_norm": 1.5032505989074707,
      "learning_rate": 1.197589098532495e-05,
      "loss": 0.5462,
      "num_input_tokens_seen": 595104,
      "step": 915
    },
    {
      "epoch": 0.48218029350104824,
      "grad_norm": 1.0917888879776,
      "learning_rate": 1.204140461215933e-05,
      "loss": 0.4997,
      "num_input_tokens_seen": 599808,
      "step": 920
    },
    {
      "epoch": 0.48480083857442346,
      "grad_norm": 1.573121428489685,
      "learning_rate": 1.210691823899371e-05,
      "loss": 0.4167,
      "num_input_tokens_seen": 602720,
      "step": 925
    },
    {
      "epoch": 0.48742138364779874,
      "grad_norm": 1.1952193975448608,
      "learning_rate": 1.2172431865828094e-05,
      "loss": 0.4892,
      "num_input_tokens_seen": 606176,
      "step": 930
    },
    {
      "epoch": 0.490041928721174,
      "grad_norm": 0.7996892333030701,
      "learning_rate": 1.2237945492662473e-05,
      "loss": 0.4515,
      "num_input_tokens_seen": 608864,
      "step": 935
    },
    {
      "epoch": 0.49266247379454925,
      "grad_norm": 0.873376190662384,
      "learning_rate": 1.2303459119496856e-05,
      "loss": 0.4897,
      "num_input_tokens_seen": 611968,
      "step": 940
    },
    {
      "epoch": 0.49528301886792453,
      "grad_norm": 1.355468988418579,
      "learning_rate": 1.2368972746331238e-05,
      "loss": 0.6032,
      "num_input_tokens_seen": 614848,
      "step": 945
    },
    {
      "epoch": 0.4979035639412998,
      "grad_norm": 0.8410788178443909,
      "learning_rate": 1.243448637316562e-05,
      "loss": 0.6433,
      "num_input_tokens_seen": 618080,
      "step": 950
    },
    {
      "epoch": 0.500524109014675,
      "grad_norm": 1.1951797008514404,
      "learning_rate": 1.25e-05,
      "loss": 0.3607,
      "num_input_tokens_seen": 620736,
      "step": 955
    },
    {
      "epoch": 0.5031446540880503,
      "grad_norm": 1.7047975063323975,
      "learning_rate": 1.2565513626834382e-05,
      "loss": 0.4556,
      "num_input_tokens_seen": 624480,
      "step": 960
    },
    {
      "epoch": 0.5057651991614256,
      "grad_norm": 1.4052996635437012,
      "learning_rate": 1.2631027253668765e-05,
      "loss": 0.4292,
      "num_input_tokens_seen": 628448,
      "step": 965
    },
    {
      "epoch": 0.5083857442348009,
      "grad_norm": 0.8621872663497925,
      "learning_rate": 1.2696540880503146e-05,
      "loss": 0.5608,
      "num_input_tokens_seen": 632000,
      "step": 970
    },
    {
      "epoch": 0.5110062893081762,
      "grad_norm": 1.0580344200134277,
      "learning_rate": 1.2762054507337526e-05,
      "loss": 0.5787,
      "num_input_tokens_seen": 635680,
      "step": 975
    },
    {
      "epoch": 0.5136268343815513,
      "grad_norm": 1.8231464624404907,
      "learning_rate": 1.2827568134171907e-05,
      "loss": 0.7183,
      "num_input_tokens_seen": 638496,
      "step": 980
    },
    {
      "epoch": 0.5162473794549266,
      "grad_norm": 1.5397305488586426,
      "learning_rate": 1.289308176100629e-05,
      "loss": 0.4619,
      "num_input_tokens_seen": 642272,
      "step": 985
    },
    {
      "epoch": 0.5188679245283019,
      "grad_norm": 1.1306794881820679,
      "learning_rate": 1.2958595387840672e-05,
      "loss": 0.5036,
      "num_input_tokens_seen": 645888,
      "step": 990
    },
    {
      "epoch": 0.5214884696016772,
      "grad_norm": 0.9472182393074036,
      "learning_rate": 1.3024109014675053e-05,
      "loss": 0.3552,
      "num_input_tokens_seen": 651808,
      "step": 995
    },
    {
      "epoch": 0.5241090146750524,
      "grad_norm": 1.3353122472763062,
      "learning_rate": 1.3089622641509436e-05,
      "loss": 0.5156,
      "num_input_tokens_seen": 655488,
      "step": 1000
    },
    {
      "epoch": 0.5267295597484277,
      "grad_norm": 0.7200545072555542,
      "learning_rate": 1.3155136268343814e-05,
      "loss": 0.4554,
      "num_input_tokens_seen": 658976,
      "step": 1005
    },
    {
      "epoch": 0.5293501048218029,
      "grad_norm": 1.271064281463623,
      "learning_rate": 1.3220649895178197e-05,
      "loss": 0.4545,
      "num_input_tokens_seen": 662048,
      "step": 1010
    },
    {
      "epoch": 0.5319706498951782,
      "grad_norm": 0.8694362640380859,
      "learning_rate": 1.3286163522012579e-05,
      "loss": 0.5622,
      "num_input_tokens_seen": 664640,
      "step": 1015
    },
    {
      "epoch": 0.5345911949685535,
      "grad_norm": 1.3837326765060425,
      "learning_rate": 1.3351677148846962e-05,
      "loss": 0.4746,
      "num_input_tokens_seen": 669056,
      "step": 1020
    },
    {
      "epoch": 0.5372117400419287,
      "grad_norm": 1.612887978553772,
      "learning_rate": 1.3417190775681343e-05,
      "loss": 0.4993,
      "num_input_tokens_seen": 672128,
      "step": 1025
    },
    {
      "epoch": 0.539832285115304,
      "grad_norm": 1.957095742225647,
      "learning_rate": 1.3482704402515725e-05,
      "loss": 0.4712,
      "num_input_tokens_seen": 675136,
      "step": 1030
    },
    {
      "epoch": 0.5424528301886793,
      "grad_norm": 1.3736881017684937,
      "learning_rate": 1.3548218029350104e-05,
      "loss": 0.5905,
      "num_input_tokens_seen": 678976,
      "step": 1035
    },
    {
      "epoch": 0.5450733752620545,
      "grad_norm": 1.8141653537750244,
      "learning_rate": 1.3613731656184486e-05,
      "loss": 0.6451,
      "num_input_tokens_seen": 682176,
      "step": 1040
    },
    {
      "epoch": 0.5476939203354297,
      "grad_norm": 1.547648549079895,
      "learning_rate": 1.3679245283018869e-05,
      "loss": 0.5455,
      "num_input_tokens_seen": 684768,
      "step": 1045
    },
    {
      "epoch": 0.550314465408805,
      "grad_norm": 1.0553088188171387,
      "learning_rate": 1.374475890985325e-05,
      "loss": 0.5193,
      "num_input_tokens_seen": 687392,
      "step": 1050
    },
    {
      "epoch": 0.5529350104821803,
      "grad_norm": 2.95029354095459,
      "learning_rate": 1.3810272536687632e-05,
      "loss": 0.4716,
      "num_input_tokens_seen": 690496,
      "step": 1055
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 0.7156791687011719,
      "learning_rate": 1.3875786163522015e-05,
      "loss": 0.648,
      "num_input_tokens_seen": 693376,
      "step": 1060
    },
    {
      "epoch": 0.5581761006289309,
      "grad_norm": 1.5009517669677734,
      "learning_rate": 1.3941299790356394e-05,
      "loss": 0.4584,
      "num_input_tokens_seen": 695808,
      "step": 1065
    },
    {
      "epoch": 0.560796645702306,
      "grad_norm": 0.8654805421829224,
      "learning_rate": 1.4006813417190776e-05,
      "loss": 0.4674,
      "num_input_tokens_seen": 699424,
      "step": 1070
    },
    {
      "epoch": 0.5634171907756813,
      "grad_norm": 0.9775951504707336,
      "learning_rate": 1.4072327044025157e-05,
      "loss": 0.5117,
      "num_input_tokens_seen": 703584,
      "step": 1075
    },
    {
      "epoch": 0.5660377358490566,
      "grad_norm": 1.385132908821106,
      "learning_rate": 1.413784067085954e-05,
      "loss": 0.5031,
      "num_input_tokens_seen": 706144,
      "step": 1080
    },
    {
      "epoch": 0.5686582809224319,
      "grad_norm": 1.0059760808944702,
      "learning_rate": 1.4203354297693922e-05,
      "loss": 0.5082,
      "num_input_tokens_seen": 708864,
      "step": 1085
    },
    {
      "epoch": 0.5712788259958071,
      "grad_norm": 0.9934499263763428,
      "learning_rate": 1.4268867924528303e-05,
      "loss": 0.531,
      "num_input_tokens_seen": 712352,
      "step": 1090
    },
    {
      "epoch": 0.5738993710691824,
      "grad_norm": 1.363346815109253,
      "learning_rate": 1.4334381551362683e-05,
      "loss": 0.7076,
      "num_input_tokens_seen": 715296,
      "step": 1095
    },
    {
      "epoch": 0.5765199161425576,
      "grad_norm": 0.9601722955703735,
      "learning_rate": 1.4399895178197064e-05,
      "loss": 0.4285,
      "num_input_tokens_seen": 718528,
      "step": 1100
    },
    {
      "epoch": 0.5791404612159329,
      "grad_norm": 0.8689362406730652,
      "learning_rate": 1.4465408805031447e-05,
      "loss": 0.4393,
      "num_input_tokens_seen": 722752,
      "step": 1105
    },
    {
      "epoch": 0.5817610062893082,
      "grad_norm": 0.9518976211547852,
      "learning_rate": 1.4530922431865828e-05,
      "loss": 0.4549,
      "num_input_tokens_seen": 725824,
      "step": 1110
    },
    {
      "epoch": 0.5843815513626834,
      "grad_norm": 1.0228487253189087,
      "learning_rate": 1.4596436058700211e-05,
      "loss": 0.5646,
      "num_input_tokens_seen": 730080,
      "step": 1115
    },
    {
      "epoch": 0.5870020964360587,
      "grad_norm": 1.1679801940917969,
      "learning_rate": 1.4661949685534593e-05,
      "loss": 0.4795,
      "num_input_tokens_seen": 733216,
      "step": 1120
    },
    {
      "epoch": 0.589622641509434,
      "grad_norm": 1.5282623767852783,
      "learning_rate": 1.4727463312368974e-05,
      "loss": 0.6458,
      "num_input_tokens_seen": 736416,
      "step": 1125
    },
    {
      "epoch": 0.5922431865828093,
      "grad_norm": 1.0802359580993652,
      "learning_rate": 1.4792976939203354e-05,
      "loss": 0.4424,
      "num_input_tokens_seen": 739808,
      "step": 1130
    },
    {
      "epoch": 0.5948637316561844,
      "grad_norm": 1.3909298181533813,
      "learning_rate": 1.4858490566037735e-05,
      "loss": 0.5327,
      "num_input_tokens_seen": 742400,
      "step": 1135
    },
    {
      "epoch": 0.5974842767295597,
      "grad_norm": 2.0607757568359375,
      "learning_rate": 1.4924004192872118e-05,
      "loss": 0.3865,
      "num_input_tokens_seen": 745760,
      "step": 1140
    },
    {
      "epoch": 0.600104821802935,
      "grad_norm": 1.5127876996994019,
      "learning_rate": 1.49895178197065e-05,
      "loss": 0.5602,
      "num_input_tokens_seen": 748736,
      "step": 1145
    },
    {
      "epoch": 0.6027253668763103,
      "grad_norm": 1.311554193496704,
      "learning_rate": 1.5055031446540883e-05,
      "loss": 0.5174,
      "num_input_tokens_seen": 752032,
      "step": 1150
    },
    {
      "epoch": 0.6053459119496856,
      "grad_norm": 0.8153391480445862,
      "learning_rate": 1.5120545073375264e-05,
      "loss": 0.4738,
      "num_input_tokens_seen": 756736,
      "step": 1155
    },
    {
      "epoch": 0.6079664570230608,
      "grad_norm": 0.9632670879364014,
      "learning_rate": 1.5186058700209644e-05,
      "loss": 0.4665,
      "num_input_tokens_seen": 759680,
      "step": 1160
    },
    {
      "epoch": 0.610587002096436,
      "grad_norm": 1.6845914125442505,
      "learning_rate": 1.5251572327044025e-05,
      "loss": 0.5003,
      "num_input_tokens_seen": 762560,
      "step": 1165
    },
    {
      "epoch": 0.6132075471698113,
      "grad_norm": 0.9617845416069031,
      "learning_rate": 1.5317085953878407e-05,
      "loss": 0.4183,
      "num_input_tokens_seen": 765600,
      "step": 1170
    },
    {
      "epoch": 0.6158280922431866,
      "grad_norm": 0.7735141515731812,
      "learning_rate": 1.538259958071279e-05,
      "loss": 0.3932,
      "num_input_tokens_seen": 768736,
      "step": 1175
    },
    {
      "epoch": 0.6184486373165619,
      "grad_norm": 1.1683390140533447,
      "learning_rate": 1.5448113207547173e-05,
      "loss": 0.6402,
      "num_input_tokens_seen": 771872,
      "step": 1180
    },
    {
      "epoch": 0.6210691823899371,
      "grad_norm": 1.049702763557434,
      "learning_rate": 1.5513626834381552e-05,
      "loss": 0.5965,
      "num_input_tokens_seen": 775008,
      "step": 1185
    },
    {
      "epoch": 0.6236897274633124,
      "grad_norm": 0.8270210027694702,
      "learning_rate": 1.5579140461215932e-05,
      "loss": 0.3828,
      "num_input_tokens_seen": 777920,
      "step": 1190
    },
    {
      "epoch": 0.6263102725366876,
      "grad_norm": 0.9411687850952148,
      "learning_rate": 1.5644654088050315e-05,
      "loss": 0.4934,
      "num_input_tokens_seen": 781056,
      "step": 1195
    },
    {
      "epoch": 0.6289308176100629,
      "grad_norm": 1.5087552070617676,
      "learning_rate": 1.5710167714884695e-05,
      "loss": 0.5518,
      "num_input_tokens_seen": 784128,
      "step": 1200
    },
    {
      "epoch": 0.6315513626834381,
      "grad_norm": 1.045322299003601,
      "learning_rate": 1.5775681341719078e-05,
      "loss": 0.513,
      "num_input_tokens_seen": 786624,
      "step": 1205
    },
    {
      "epoch": 0.6341719077568134,
      "grad_norm": 1.5057857036590576,
      "learning_rate": 1.584119496855346e-05,
      "loss": 0.5068,
      "num_input_tokens_seen": 789664,
      "step": 1210
    },
    {
      "epoch": 0.6367924528301887,
      "grad_norm": 1.0071712732315063,
      "learning_rate": 1.5906708595387844e-05,
      "loss": 0.4174,
      "num_input_tokens_seen": 793824,
      "step": 1215
    },
    {
      "epoch": 0.639412997903564,
      "grad_norm": 1.6219818592071533,
      "learning_rate": 1.597222222222222e-05,
      "loss": 0.5991,
      "num_input_tokens_seen": 796544,
      "step": 1220
    },
    {
      "epoch": 0.6420335429769392,
      "grad_norm": 0.8515250086784363,
      "learning_rate": 1.6037735849056604e-05,
      "loss": 0.3933,
      "num_input_tokens_seen": 799520,
      "step": 1225
    },
    {
      "epoch": 0.6446540880503144,
      "grad_norm": 1.4816514253616333,
      "learning_rate": 1.6103249475890987e-05,
      "loss": 0.3265,
      "num_input_tokens_seen": 802464,
      "step": 1230
    },
    {
      "epoch": 0.6472746331236897,
      "grad_norm": 1.5994969606399536,
      "learning_rate": 1.6168763102725366e-05,
      "loss": 0.5744,
      "num_input_tokens_seen": 805536,
      "step": 1235
    },
    {
      "epoch": 0.649895178197065,
      "grad_norm": 0.9286901354789734,
      "learning_rate": 1.623427672955975e-05,
      "loss": 0.3562,
      "num_input_tokens_seen": 808448,
      "step": 1240
    },
    {
      "epoch": 0.6525157232704403,
      "grad_norm": 1.1449795961380005,
      "learning_rate": 1.6299790356394132e-05,
      "loss": 0.4164,
      "num_input_tokens_seen": 812160,
      "step": 1245
    },
    {
      "epoch": 0.6551362683438156,
      "grad_norm": 2.589778423309326,
      "learning_rate": 1.6365303983228512e-05,
      "loss": 0.6204,
      "num_input_tokens_seen": 814752,
      "step": 1250
    },
    {
      "epoch": 0.6577568134171907,
      "grad_norm": 2.101564884185791,
      "learning_rate": 1.6430817610062892e-05,
      "loss": 0.3871,
      "num_input_tokens_seen": 817568,
      "step": 1255
    },
    {
      "epoch": 0.660377358490566,
      "grad_norm": 1.564061164855957,
      "learning_rate": 1.6496331236897275e-05,
      "loss": 0.4705,
      "num_input_tokens_seen": 820384,
      "step": 1260
    },
    {
      "epoch": 0.6629979035639413,
      "grad_norm": 1.1729989051818848,
      "learning_rate": 1.6561844863731658e-05,
      "loss": 0.2982,
      "num_input_tokens_seen": 823040,
      "step": 1265
    },
    {
      "epoch": 0.6656184486373166,
      "grad_norm": 1.3853052854537964,
      "learning_rate": 1.6627358490566038e-05,
      "loss": 0.5674,
      "num_input_tokens_seen": 825600,
      "step": 1270
    },
    {
      "epoch": 0.6682389937106918,
      "grad_norm": 1.4008939266204834,
      "learning_rate": 1.669287211740042e-05,
      "loss": 0.5757,
      "num_input_tokens_seen": 828032,
      "step": 1275
    },
    {
      "epoch": 0.6708595387840671,
      "grad_norm": 1.811542272567749,
      "learning_rate": 1.6758385744234804e-05,
      "loss": 0.4297,
      "num_input_tokens_seen": 831296,
      "step": 1280
    },
    {
      "epoch": 0.6734800838574424,
      "grad_norm": 1.1963437795639038,
      "learning_rate": 1.6823899371069183e-05,
      "loss": 0.5952,
      "num_input_tokens_seen": 834208,
      "step": 1285
    },
    {
      "epoch": 0.6761006289308176,
      "grad_norm": 1.326469898223877,
      "learning_rate": 1.6889412997903563e-05,
      "loss": 0.4442,
      "num_input_tokens_seen": 838048,
      "step": 1290
    },
    {
      "epoch": 0.6787211740041929,
      "grad_norm": 1.9483048915863037,
      "learning_rate": 1.6954926624737946e-05,
      "loss": 0.5615,
      "num_input_tokens_seen": 841536,
      "step": 1295
    },
    {
      "epoch": 0.6813417190775681,
      "grad_norm": 0.9757769107818604,
      "learning_rate": 1.702044025157233e-05,
      "loss": 0.4259,
      "num_input_tokens_seen": 845184,
      "step": 1300
    },
    {
      "epoch": 0.6839622641509434,
      "grad_norm": 0.9183277487754822,
      "learning_rate": 1.708595387840671e-05,
      "loss": 0.4256,
      "num_input_tokens_seen": 848000,
      "step": 1305
    },
    {
      "epoch": 0.6865828092243187,
      "grad_norm": 2.2321949005126953,
      "learning_rate": 1.7151467505241092e-05,
      "loss": 0.5729,
      "num_input_tokens_seen": 851040,
      "step": 1310
    },
    {
      "epoch": 0.689203354297694,
      "grad_norm": 1.6911252737045288,
      "learning_rate": 1.7216981132075472e-05,
      "loss": 0.5813,
      "num_input_tokens_seen": 853920,
      "step": 1315
    },
    {
      "epoch": 0.6918238993710691,
      "grad_norm": 1.55675208568573,
      "learning_rate": 1.7282494758909855e-05,
      "loss": 0.5031,
      "num_input_tokens_seen": 857312,
      "step": 1320
    },
    {
      "epoch": 0.6944444444444444,
      "grad_norm": 1.2560739517211914,
      "learning_rate": 1.7348008385744234e-05,
      "loss": 0.4048,
      "num_input_tokens_seen": 859904,
      "step": 1325
    },
    {
      "epoch": 0.6970649895178197,
      "grad_norm": 1.1688133478164673,
      "learning_rate": 1.7413522012578618e-05,
      "loss": 0.5433,
      "num_input_tokens_seen": 863680,
      "step": 1330
    },
    {
      "epoch": 0.699685534591195,
      "grad_norm": 1.0300755500793457,
      "learning_rate": 1.7479035639413e-05,
      "loss": 0.5655,
      "num_input_tokens_seen": 866880,
      "step": 1335
    },
    {
      "epoch": 0.7023060796645703,
      "grad_norm": 1.703035593032837,
      "learning_rate": 1.754454926624738e-05,
      "loss": 0.3886,
      "num_input_tokens_seen": 869856,
      "step": 1340
    },
    {
      "epoch": 0.7049266247379455,
      "grad_norm": 1.2870471477508545,
      "learning_rate": 1.761006289308176e-05,
      "loss": 0.5134,
      "num_input_tokens_seen": 873472,
      "step": 1345
    },
    {
      "epoch": 0.7075471698113207,
      "grad_norm": 1.0636744499206543,
      "learning_rate": 1.7675576519916143e-05,
      "loss": 0.4377,
      "num_input_tokens_seen": 876704,
      "step": 1350
    },
    {
      "epoch": 0.710167714884696,
      "grad_norm": 0.8409854769706726,
      "learning_rate": 1.7741090146750526e-05,
      "loss": 0.4494,
      "num_input_tokens_seen": 880576,
      "step": 1355
    },
    {
      "epoch": 0.7127882599580713,
      "grad_norm": 1.2556918859481812,
      "learning_rate": 1.7806603773584906e-05,
      "loss": 0.4681,
      "num_input_tokens_seen": 883680,
      "step": 1360
    },
    {
      "epoch": 0.7154088050314465,
      "grad_norm": 1.3437458276748657,
      "learning_rate": 1.787211740041929e-05,
      "loss": 0.5344,
      "num_input_tokens_seen": 886464,
      "step": 1365
    },
    {
      "epoch": 0.7180293501048218,
      "grad_norm": 1.5086357593536377,
      "learning_rate": 1.7937631027253672e-05,
      "loss": 0.5148,
      "num_input_tokens_seen": 889536,
      "step": 1370
    },
    {
      "epoch": 0.7206498951781971,
      "grad_norm": 1.2264437675476074,
      "learning_rate": 1.8003144654088048e-05,
      "loss": 0.5034,
      "num_input_tokens_seen": 894272,
      "step": 1375
    },
    {
      "epoch": 0.7232704402515723,
      "grad_norm": 0.8101043701171875,
      "learning_rate": 1.806865828092243e-05,
      "loss": 0.4383,
      "num_input_tokens_seen": 897120,
      "step": 1380
    },
    {
      "epoch": 0.7258909853249476,
      "grad_norm": 1.69804847240448,
      "learning_rate": 1.8134171907756814e-05,
      "loss": 0.4895,
      "num_input_tokens_seen": 899808,
      "step": 1385
    },
    {
      "epoch": 0.7285115303983228,
      "grad_norm": 2.094693660736084,
      "learning_rate": 1.8199685534591197e-05,
      "loss": 0.4811,
      "num_input_tokens_seen": 903520,
      "step": 1390
    },
    {
      "epoch": 0.7311320754716981,
      "grad_norm": 1.258455753326416,
      "learning_rate": 1.8265199161425577e-05,
      "loss": 0.5223,
      "num_input_tokens_seen": 907200,
      "step": 1395
    },
    {
      "epoch": 0.7337526205450734,
      "grad_norm": 0.7063034176826477,
      "learning_rate": 1.833071278825996e-05,
      "loss": 0.4542,
      "num_input_tokens_seen": 911392,
      "step": 1400
    },
    {
      "epoch": 0.7363731656184487,
      "grad_norm": 1.5733360052108765,
      "learning_rate": 1.839622641509434e-05,
      "loss": 0.4827,
      "num_input_tokens_seen": 913440,
      "step": 1405
    },
    {
      "epoch": 0.7389937106918238,
      "grad_norm": 1.2907094955444336,
      "learning_rate": 1.846174004192872e-05,
      "loss": 0.4415,
      "num_input_tokens_seen": 917888,
      "step": 1410
    },
    {
      "epoch": 0.7416142557651991,
      "grad_norm": 0.8765714168548584,
      "learning_rate": 1.8527253668763103e-05,
      "loss": 0.5106,
      "num_input_tokens_seen": 921376,
      "step": 1415
    },
    {
      "epoch": 0.7442348008385744,
      "grad_norm": 0.9896162152290344,
      "learning_rate": 1.8592767295597486e-05,
      "loss": 0.5129,
      "num_input_tokens_seen": 924800,
      "step": 1420
    },
    {
      "epoch": 0.7468553459119497,
      "grad_norm": 3.794854164123535,
      "learning_rate": 1.865828092243187e-05,
      "loss": 0.6049,
      "num_input_tokens_seen": 927648,
      "step": 1425
    },
    {
      "epoch": 0.749475890985325,
      "grad_norm": 1.6597323417663574,
      "learning_rate": 1.872379454926625e-05,
      "loss": 0.3857,
      "num_input_tokens_seen": 930144,
      "step": 1430
    },
    {
      "epoch": 0.7520964360587002,
      "grad_norm": 1.176046371459961,
      "learning_rate": 1.878930817610063e-05,
      "loss": 0.4099,
      "num_input_tokens_seen": 933184,
      "step": 1435
    },
    {
      "epoch": 0.7547169811320755,
      "grad_norm": 1.334855079650879,
      "learning_rate": 1.885482180293501e-05,
      "loss": 0.527,
      "num_input_tokens_seen": 937376,
      "step": 1440
    },
    {
      "epoch": 0.7573375262054507,
      "grad_norm": 1.6718733310699463,
      "learning_rate": 1.892033542976939e-05,
      "loss": 0.4444,
      "num_input_tokens_seen": 940160,
      "step": 1445
    },
    {
      "epoch": 0.759958071278826,
      "grad_norm": 1.4209359884262085,
      "learning_rate": 1.8985849056603774e-05,
      "loss": 0.5044,
      "num_input_tokens_seen": 942816,
      "step": 1450
    },
    {
      "epoch": 0.7625786163522013,
      "grad_norm": 1.5979580879211426,
      "learning_rate": 1.9051362683438157e-05,
      "loss": 0.5484,
      "num_input_tokens_seen": 945952,
      "step": 1455
    },
    {
      "epoch": 0.7651991614255765,
      "grad_norm": 1.9523781538009644,
      "learning_rate": 1.9116876310272537e-05,
      "loss": 0.3872,
      "num_input_tokens_seen": 949472,
      "step": 1460
    },
    {
      "epoch": 0.7678197064989518,
      "grad_norm": 1.9014869928359985,
      "learning_rate": 1.918238993710692e-05,
      "loss": 0.494,
      "num_input_tokens_seen": 952672,
      "step": 1465
    },
    {
      "epoch": 0.7704402515723271,
      "grad_norm": 0.8762264251708984,
      "learning_rate": 1.92479035639413e-05,
      "loss": 0.5147,
      "num_input_tokens_seen": 955520,
      "step": 1470
    },
    {
      "epoch": 0.7730607966457023,
      "grad_norm": 1.3464457988739014,
      "learning_rate": 1.9313417190775683e-05,
      "loss": 0.5512,
      "num_input_tokens_seen": 958464,
      "step": 1475
    },
    {
      "epoch": 0.7756813417190775,
      "grad_norm": 2.2840254306793213,
      "learning_rate": 1.9378930817610062e-05,
      "loss": 0.4661,
      "num_input_tokens_seen": 960768,
      "step": 1480
    },
    {
      "epoch": 0.7783018867924528,
      "grad_norm": 1.7242305278778076,
      "learning_rate": 1.9444444444444445e-05,
      "loss": 0.6797,
      "num_input_tokens_seen": 963616,
      "step": 1485
    },
    {
      "epoch": 0.7809224318658281,
      "grad_norm": 0.9389321804046631,
      "learning_rate": 1.950995807127883e-05,
      "loss": 0.6789,
      "num_input_tokens_seen": 967936,
      "step": 1490
    },
    {
      "epoch": 0.7835429769392034,
      "grad_norm": 0.975024938583374,
      "learning_rate": 1.9575471698113208e-05,
      "loss": 0.4609,
      "num_input_tokens_seen": 971744,
      "step": 1495
    },
    {
      "epoch": 0.7861635220125787,
      "grad_norm": 1.4652554988861084,
      "learning_rate": 1.9640985324947588e-05,
      "loss": 0.5024,
      "num_input_tokens_seen": 974080,
      "step": 1500
    },
    {
      "epoch": 0.7887840670859538,
      "grad_norm": 1.0781551599502563,
      "learning_rate": 1.970649895178197e-05,
      "loss": 0.4585,
      "num_input_tokens_seen": 976992,
      "step": 1505
    },
    {
      "epoch": 0.7914046121593291,
      "grad_norm": 1.7573494911193848,
      "learning_rate": 1.9772012578616354e-05,
      "loss": 0.483,
      "num_input_tokens_seen": 979648,
      "step": 1510
    },
    {
      "epoch": 0.7940251572327044,
      "grad_norm": 1.0790897607803345,
      "learning_rate": 1.9837526205450734e-05,
      "loss": 0.5373,
      "num_input_tokens_seen": 983136,
      "step": 1515
    },
    {
      "epoch": 0.7966457023060797,
      "grad_norm": 0.7371461391448975,
      "learning_rate": 1.9903039832285117e-05,
      "loss": 0.4786,
      "num_input_tokens_seen": 986976,
      "step": 1520
    },
    {
      "epoch": 0.799266247379455,
      "grad_norm": 0.8000678420066833,
      "learning_rate": 1.99685534591195e-05,
      "loss": 0.4544,
      "num_input_tokens_seen": 990272,
      "step": 1525
    },
    {
      "epoch": 0.8018867924528302,
      "grad_norm": 0.985109269618988,
      "learning_rate": 2.003406708595388e-05,
      "loss": 0.4592,
      "num_input_tokens_seen": 994368,
      "step": 1530
    },
    {
      "epoch": 0.8045073375262054,
      "grad_norm": 1.2530500888824463,
      "learning_rate": 2.009958071278826e-05,
      "loss": 0.4177,
      "num_input_tokens_seen": 997504,
      "step": 1535
    },
    {
      "epoch": 0.8071278825995807,
      "grad_norm": 1.5178852081298828,
      "learning_rate": 2.0165094339622642e-05,
      "loss": 0.5462,
      "num_input_tokens_seen": 1000352,
      "step": 1540
    },
    {
      "epoch": 0.809748427672956,
      "grad_norm": 0.9786360263824463,
      "learning_rate": 2.0230607966457025e-05,
      "loss": 0.555,
      "num_input_tokens_seen": 1003616,
      "step": 1545
    },
    {
      "epoch": 0.8123689727463312,
      "grad_norm": 3.062628984451294,
      "learning_rate": 2.0296121593291405e-05,
      "loss": 0.5242,
      "num_input_tokens_seen": 1006240,
      "step": 1550
    },
    {
      "epoch": 0.8149895178197065,
      "grad_norm": 1.508867621421814,
      "learning_rate": 2.0361635220125788e-05,
      "loss": 0.5376,
      "num_input_tokens_seen": 1009440,
      "step": 1555
    },
    {
      "epoch": 0.8176100628930818,
      "grad_norm": 1.1389445066452026,
      "learning_rate": 2.0427148846960168e-05,
      "loss": 0.5241,
      "num_input_tokens_seen": 1011936,
      "step": 1560
    },
    {
      "epoch": 0.820230607966457,
      "grad_norm": 1.1279791593551636,
      "learning_rate": 2.049266247379455e-05,
      "loss": 0.4363,
      "num_input_tokens_seen": 1015040,
      "step": 1565
    },
    {
      "epoch": 0.8228511530398323,
      "grad_norm": 1.061981201171875,
      "learning_rate": 2.055817610062893e-05,
      "loss": 0.5781,
      "num_input_tokens_seen": 1018240,
      "step": 1570
    },
    {
      "epoch": 0.8254716981132075,
      "grad_norm": 0.6686882376670837,
      "learning_rate": 2.0623689727463314e-05,
      "loss": 0.4877,
      "num_input_tokens_seen": 1022336,
      "step": 1575
    },
    {
      "epoch": 0.8280922431865828,
      "grad_norm": 1.9870752096176147,
      "learning_rate": 2.0689203354297697e-05,
      "loss": 0.497,
      "num_input_tokens_seen": 1025184,
      "step": 1580
    },
    {
      "epoch": 0.8307127882599581,
      "grad_norm": 2.887237787246704,
      "learning_rate": 2.0754716981132076e-05,
      "loss": 0.5509,
      "num_input_tokens_seen": 1028448,
      "step": 1585
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 1.2696454524993896,
      "learning_rate": 2.0820230607966456e-05,
      "loss": 0.5238,
      "num_input_tokens_seen": 1031072,
      "step": 1590
    },
    {
      "epoch": 0.8359538784067087,
      "grad_norm": 1.3730674982070923,
      "learning_rate": 2.088574423480084e-05,
      "loss": 0.4635,
      "num_input_tokens_seen": 1034208,
      "step": 1595
    },
    {
      "epoch": 0.8385744234800838,
      "grad_norm": 1.3564382791519165,
      "learning_rate": 2.0951257861635222e-05,
      "loss": 0.4309,
      "num_input_tokens_seen": 1038624,
      "step": 1600
    },
    {
      "epoch": 0.8411949685534591,
      "grad_norm": 0.9022404551506042,
      "learning_rate": 2.1016771488469602e-05,
      "loss": 0.5429,
      "num_input_tokens_seen": 1041152,
      "step": 1605
    },
    {
      "epoch": 0.8438155136268344,
      "grad_norm": 1.0096526145935059,
      "learning_rate": 2.1082285115303985e-05,
      "loss": 0.3888,
      "num_input_tokens_seen": 1043904,
      "step": 1610
    },
    {
      "epoch": 0.8464360587002097,
      "grad_norm": 0.8734282851219177,
      "learning_rate": 2.1147798742138368e-05,
      "loss": 0.5035,
      "num_input_tokens_seen": 1047456,
      "step": 1615
    },
    {
      "epoch": 0.8490566037735849,
      "grad_norm": 1.1849571466445923,
      "learning_rate": 2.1213312368972748e-05,
      "loss": 0.5425,
      "num_input_tokens_seen": 1050304,
      "step": 1620
    },
    {
      "epoch": 0.8516771488469602,
      "grad_norm": 1.307526707649231,
      "learning_rate": 2.1278825995807127e-05,
      "loss": 0.5164,
      "num_input_tokens_seen": 1053600,
      "step": 1625
    },
    {
      "epoch": 0.8542976939203354,
      "grad_norm": 1.2662079334259033,
      "learning_rate": 2.134433962264151e-05,
      "loss": 0.3348,
      "num_input_tokens_seen": 1056672,
      "step": 1630
    },
    {
      "epoch": 0.8569182389937107,
      "grad_norm": 1.7193341255187988,
      "learning_rate": 2.1409853249475894e-05,
      "loss": 0.3944,
      "num_input_tokens_seen": 1059904,
      "step": 1635
    },
    {
      "epoch": 0.859538784067086,
      "grad_norm": 1.1774004697799683,
      "learning_rate": 2.1475366876310273e-05,
      "loss": 0.4881,
      "num_input_tokens_seen": 1063360,
      "step": 1640
    },
    {
      "epoch": 0.8621593291404612,
      "grad_norm": 1.8001474142074585,
      "learning_rate": 2.1540880503144656e-05,
      "loss": 0.4784,
      "num_input_tokens_seen": 1065728,
      "step": 1645
    },
    {
      "epoch": 0.8647798742138365,
      "grad_norm": 1.3021634817123413,
      "learning_rate": 2.160639412997904e-05,
      "loss": 0.4638,
      "num_input_tokens_seen": 1068448,
      "step": 1650
    },
    {
      "epoch": 0.8674004192872118,
      "grad_norm": 1.2274174690246582,
      "learning_rate": 2.1671907756813416e-05,
      "loss": 0.4839,
      "num_input_tokens_seen": 1071104,
      "step": 1655
    },
    {
      "epoch": 0.870020964360587,
      "grad_norm": 1.547707200050354,
      "learning_rate": 2.17374213836478e-05,
      "loss": 0.3998,
      "num_input_tokens_seen": 1075072,
      "step": 1660
    },
    {
      "epoch": 0.8726415094339622,
      "grad_norm": 1.1717965602874756,
      "learning_rate": 2.1802935010482182e-05,
      "loss": 0.4994,
      "num_input_tokens_seen": 1077824,
      "step": 1665
    },
    {
      "epoch": 0.8752620545073375,
      "grad_norm": 1.961364507675171,
      "learning_rate": 2.186844863731656e-05,
      "loss": 0.4661,
      "num_input_tokens_seen": 1080832,
      "step": 1670
    },
    {
      "epoch": 0.8778825995807128,
      "grad_norm": 1.834125280380249,
      "learning_rate": 2.1933962264150945e-05,
      "loss": 0.516,
      "num_input_tokens_seen": 1084096,
      "step": 1675
    },
    {
      "epoch": 0.8805031446540881,
      "grad_norm": 1.0227700471878052,
      "learning_rate": 2.1999475890985328e-05,
      "loss": 0.4845,
      "num_input_tokens_seen": 1087104,
      "step": 1680
    },
    {
      "epoch": 0.8831236897274634,
      "grad_norm": 1.0569632053375244,
      "learning_rate": 2.2064989517819707e-05,
      "loss": 0.5591,
      "num_input_tokens_seen": 1090304,
      "step": 1685
    },
    {
      "epoch": 0.8857442348008385,
      "grad_norm": 1.7256577014923096,
      "learning_rate": 2.2130503144654087e-05,
      "loss": 0.7647,
      "num_input_tokens_seen": 1094112,
      "step": 1690
    },
    {
      "epoch": 0.8883647798742138,
      "grad_norm": 1.108201503753662,
      "learning_rate": 2.219601677148847e-05,
      "loss": 0.5043,
      "num_input_tokens_seen": 1098496,
      "step": 1695
    },
    {
      "epoch": 0.8909853249475891,
      "grad_norm": 0.9137705564498901,
      "learning_rate": 2.2261530398322853e-05,
      "loss": 0.4166,
      "num_input_tokens_seen": 1100992,
      "step": 1700
    },
    {
      "epoch": 0.8936058700209644,
      "grad_norm": 1.479645013809204,
      "learning_rate": 2.2327044025157233e-05,
      "loss": 0.5268,
      "num_input_tokens_seen": 1103776,
      "step": 1705
    },
    {
      "epoch": 0.8962264150943396,
      "grad_norm": 1.4655557870864868,
      "learning_rate": 2.2392557651991616e-05,
      "loss": 0.4441,
      "num_input_tokens_seen": 1106368,
      "step": 1710
    },
    {
      "epoch": 0.8988469601677149,
      "grad_norm": 1.4439606666564941,
      "learning_rate": 2.2458071278825996e-05,
      "loss": 0.428,
      "num_input_tokens_seen": 1109376,
      "step": 1715
    },
    {
      "epoch": 0.9014675052410901,
      "grad_norm": 0.7830042243003845,
      "learning_rate": 2.252358490566038e-05,
      "loss": 0.4502,
      "num_input_tokens_seen": 1112448,
      "step": 1720
    },
    {
      "epoch": 0.9040880503144654,
      "grad_norm": 0.7430910468101501,
      "learning_rate": 2.258909853249476e-05,
      "loss": 0.4634,
      "num_input_tokens_seen": 1115520,
      "step": 1725
    },
    {
      "epoch": 0.9067085953878407,
      "grad_norm": 1.687449336051941,
      "learning_rate": 2.265461215932914e-05,
      "loss": 0.6346,
      "num_input_tokens_seen": 1118784,
      "step": 1730
    },
    {
      "epoch": 0.9093291404612159,
      "grad_norm": 3.640740156173706,
      "learning_rate": 2.2720125786163524e-05,
      "loss": 0.6952,
      "num_input_tokens_seen": 1122112,
      "step": 1735
    },
    {
      "epoch": 0.9119496855345912,
      "grad_norm": 1.3885586261749268,
      "learning_rate": 2.2785639412997904e-05,
      "loss": 0.4455,
      "num_input_tokens_seen": 1125952,
      "step": 1740
    },
    {
      "epoch": 0.9145702306079665,
      "grad_norm": 1.779643177986145,
      "learning_rate": 2.2851153039832284e-05,
      "loss": 0.628,
      "num_input_tokens_seen": 1129024,
      "step": 1745
    },
    {
      "epoch": 0.9171907756813418,
      "grad_norm": 1.5251678228378296,
      "learning_rate": 2.2916666666666667e-05,
      "loss": 0.4316,
      "num_input_tokens_seen": 1131488,
      "step": 1750
    },
    {
      "epoch": 0.9198113207547169,
      "grad_norm": 1.1568266153335571,
      "learning_rate": 2.298218029350105e-05,
      "loss": 0.4429,
      "num_input_tokens_seen": 1134176,
      "step": 1755
    },
    {
      "epoch": 0.9224318658280922,
      "grad_norm": 0.640579342842102,
      "learning_rate": 2.304769392033543e-05,
      "loss": 0.5033,
      "num_input_tokens_seen": 1137888,
      "step": 1760
    },
    {
      "epoch": 0.9250524109014675,
      "grad_norm": 2.3611345291137695,
      "learning_rate": 2.3113207547169813e-05,
      "loss": 0.5781,
      "num_input_tokens_seen": 1141184,
      "step": 1765
    },
    {
      "epoch": 0.9276729559748428,
      "grad_norm": 2.310351848602295,
      "learning_rate": 2.3178721174004196e-05,
      "loss": 0.5285,
      "num_input_tokens_seen": 1144064,
      "step": 1770
    },
    {
      "epoch": 0.9302935010482181,
      "grad_norm": 1.7239028215408325,
      "learning_rate": 2.3244234800838576e-05,
      "loss": 0.4625,
      "num_input_tokens_seen": 1147584,
      "step": 1775
    },
    {
      "epoch": 0.9329140461215933,
      "grad_norm": 0.65884929895401,
      "learning_rate": 2.3309748427672955e-05,
      "loss": 0.4607,
      "num_input_tokens_seen": 1154784,
      "step": 1780
    },
    {
      "epoch": 0.9355345911949685,
      "grad_norm": 2.241380453109741,
      "learning_rate": 2.3375262054507338e-05,
      "loss": 0.467,
      "num_input_tokens_seen": 1157440,
      "step": 1785
    },
    {
      "epoch": 0.9381551362683438,
      "grad_norm": 0.7356058955192566,
      "learning_rate": 2.344077568134172e-05,
      "loss": 0.6042,
      "num_input_tokens_seen": 1160224,
      "step": 1790
    },
    {
      "epoch": 0.9407756813417191,
      "grad_norm": 1.7074998617172241,
      "learning_rate": 2.35062893081761e-05,
      "loss": 0.4937,
      "num_input_tokens_seen": 1164480,
      "step": 1795
    },
    {
      "epoch": 0.9433962264150944,
      "grad_norm": 1.9943757057189941,
      "learning_rate": 2.3571802935010484e-05,
      "loss": 0.3796,
      "num_input_tokens_seen": 1168704,
      "step": 1800
    },
    {
      "epoch": 0.9460167714884696,
      "grad_norm": 1.36757493019104,
      "learning_rate": 2.3637316561844867e-05,
      "loss": 0.3974,
      "num_input_tokens_seen": 1172384,
      "step": 1805
    },
    {
      "epoch": 0.9486373165618449,
      "grad_norm": 1.2833868265151978,
      "learning_rate": 2.3702830188679247e-05,
      "loss": 0.4716,
      "num_input_tokens_seen": 1175392,
      "step": 1810
    },
    {
      "epoch": 0.9512578616352201,
      "grad_norm": 1.2130563259124756,
      "learning_rate": 2.3768343815513627e-05,
      "loss": 0.4404,
      "num_input_tokens_seen": 1178368,
      "step": 1815
    },
    {
      "epoch": 0.9538784067085954,
      "grad_norm": 1.3863474130630493,
      "learning_rate": 2.383385744234801e-05,
      "loss": 0.3871,
      "num_input_tokens_seen": 1182240,
      "step": 1820
    },
    {
      "epoch": 0.9564989517819706,
      "grad_norm": 0.9348484873771667,
      "learning_rate": 2.3899371069182393e-05,
      "loss": 0.4496,
      "num_input_tokens_seen": 1185504,
      "step": 1825
    },
    {
      "epoch": 0.9591194968553459,
      "grad_norm": 2.223841428756714,
      "learning_rate": 2.3964884696016772e-05,
      "loss": 0.4575,
      "num_input_tokens_seen": 1188288,
      "step": 1830
    },
    {
      "epoch": 0.9617400419287212,
      "grad_norm": 1.4784348011016846,
      "learning_rate": 2.4030398322851155e-05,
      "loss": 0.4372,
      "num_input_tokens_seen": 1192064,
      "step": 1835
    },
    {
      "epoch": 0.9643605870020965,
      "grad_norm": 0.9507945775985718,
      "learning_rate": 2.4095911949685535e-05,
      "loss": 0.4375,
      "num_input_tokens_seen": 1195104,
      "step": 1840
    },
    {
      "epoch": 0.9669811320754716,
      "grad_norm": 1.2066198587417603,
      "learning_rate": 2.4161425576519918e-05,
      "loss": 0.5851,
      "num_input_tokens_seen": 1197920,
      "step": 1845
    },
    {
      "epoch": 0.9696016771488469,
      "grad_norm": 0.917510986328125,
      "learning_rate": 2.4226939203354298e-05,
      "loss": 0.432,
      "num_input_tokens_seen": 1200896,
      "step": 1850
    },
    {
      "epoch": 0.9722222222222222,
      "grad_norm": 1.6063488721847534,
      "learning_rate": 2.429245283018868e-05,
      "loss": 0.452,
      "num_input_tokens_seen": 1204032,
      "step": 1855
    },
    {
      "epoch": 0.9748427672955975,
      "grad_norm": 1.1325654983520508,
      "learning_rate": 2.4357966457023064e-05,
      "loss": 0.3894,
      "num_input_tokens_seen": 1207264,
      "step": 1860
    },
    {
      "epoch": 0.9774633123689728,
      "grad_norm": 1.860935091972351,
      "learning_rate": 2.4423480083857444e-05,
      "loss": 0.5438,
      "num_input_tokens_seen": 1209984,
      "step": 1865
    },
    {
      "epoch": 0.980083857442348,
      "grad_norm": 1.5019116401672363,
      "learning_rate": 2.4488993710691823e-05,
      "loss": 0.4144,
      "num_input_tokens_seen": 1213856,
      "step": 1870
    },
    {
      "epoch": 0.9827044025157232,
      "grad_norm": 0.9762663245201111,
      "learning_rate": 2.4554507337526206e-05,
      "loss": 0.4386,
      "num_input_tokens_seen": 1217408,
      "step": 1875
    },
    {
      "epoch": 0.9853249475890985,
      "grad_norm": 0.8043728470802307,
      "learning_rate": 2.4620020964360586e-05,
      "loss": 0.4918,
      "num_input_tokens_seen": 1221024,
      "step": 1880
    },
    {
      "epoch": 0.9879454926624738,
      "grad_norm": 1.25347101688385,
      "learning_rate": 2.468553459119497e-05,
      "loss": 0.3908,
      "num_input_tokens_seen": 1224320,
      "step": 1885
    },
    {
      "epoch": 0.9905660377358491,
      "grad_norm": 0.9051137566566467,
      "learning_rate": 2.4751048218029352e-05,
      "loss": 0.3765,
      "num_input_tokens_seen": 1228000,
      "step": 1890
    },
    {
      "epoch": 0.9931865828092243,
      "grad_norm": 1.5234224796295166,
      "learning_rate": 2.4816561844863735e-05,
      "loss": 0.4787,
      "num_input_tokens_seen": 1232320,
      "step": 1895
    },
    {
      "epoch": 0.9958071278825996,
      "grad_norm": 1.5348492860794067,
      "learning_rate": 2.488207547169811e-05,
      "loss": 0.5755,
      "num_input_tokens_seen": 1238272,
      "step": 1900
    },
    {
      "epoch": 0.9984276729559748,
      "grad_norm": 2.71412992477417,
      "learning_rate": 2.4947589098532495e-05,
      "loss": 0.4698,
      "num_input_tokens_seen": 1241344,
      "step": 1905
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.45764854550361633,
      "eval_runtime": 11.6051,
      "eval_samples_per_second": 73.071,
      "eval_steps_per_second": 18.268,
      "num_input_tokens_seen": 1243048,
      "step": 1908
    },
    {
      "epoch": 1.00104821802935,
      "grad_norm": 1.3951404094696045,
      "learning_rate": 2.5013102725366878e-05,
      "loss": 0.4145,
      "num_input_tokens_seen": 1244520,
      "step": 1910
    },
    {
      "epoch": 1.0036687631027255,
      "grad_norm": 1.242878794670105,
      "learning_rate": 2.507861635220126e-05,
      "loss": 0.4683,
      "num_input_tokens_seen": 1247528,
      "step": 1915
    },
    {
      "epoch": 1.0062893081761006,
      "grad_norm": 1.1943484544754028,
      "learning_rate": 2.514412997903564e-05,
      "loss": 0.5786,
      "num_input_tokens_seen": 1250824,
      "step": 1920
    },
    {
      "epoch": 1.0089098532494758,
      "grad_norm": 1.0758882761001587,
      "learning_rate": 2.520964360587002e-05,
      "loss": 0.4937,
      "num_input_tokens_seen": 1254024,
      "step": 1925
    },
    {
      "epoch": 1.0115303983228512,
      "grad_norm": 1.3744534254074097,
      "learning_rate": 2.5275157232704407e-05,
      "loss": 0.4288,
      "num_input_tokens_seen": 1260328,
      "step": 1930
    },
    {
      "epoch": 1.0141509433962264,
      "grad_norm": 1.4037724733352661,
      "learning_rate": 2.5340670859538783e-05,
      "loss": 0.6649,
      "num_input_tokens_seen": 1263656,
      "step": 1935
    },
    {
      "epoch": 1.0167714884696017,
      "grad_norm": 1.1114431619644165,
      "learning_rate": 2.540618448637317e-05,
      "loss": 0.5047,
      "num_input_tokens_seen": 1267528,
      "step": 1940
    },
    {
      "epoch": 1.019392033542977,
      "grad_norm": 1.9627832174301147,
      "learning_rate": 2.547169811320755e-05,
      "loss": 0.546,
      "num_input_tokens_seen": 1270536,
      "step": 1945
    },
    {
      "epoch": 1.0220125786163523,
      "grad_norm": 1.3622410297393799,
      "learning_rate": 2.553721174004193e-05,
      "loss": 0.3788,
      "num_input_tokens_seen": 1273416,
      "step": 1950
    },
    {
      "epoch": 1.0246331236897275,
      "grad_norm": 1.731316089630127,
      "learning_rate": 2.5602725366876312e-05,
      "loss": 0.3575,
      "num_input_tokens_seen": 1276232,
      "step": 1955
    },
    {
      "epoch": 1.0272536687631026,
      "grad_norm": 0.8621290326118469,
      "learning_rate": 2.566823899371069e-05,
      "loss": 0.3769,
      "num_input_tokens_seen": 1279400,
      "step": 1960
    },
    {
      "epoch": 1.029874213836478,
      "grad_norm": 0.8138803243637085,
      "learning_rate": 2.5733752620545075e-05,
      "loss": 0.4193,
      "num_input_tokens_seen": 1283752,
      "step": 1965
    },
    {
      "epoch": 1.0324947589098532,
      "grad_norm": 1.057543158531189,
      "learning_rate": 2.5799266247379454e-05,
      "loss": 0.4056,
      "num_input_tokens_seen": 1287048,
      "step": 1970
    },
    {
      "epoch": 1.0351153039832286,
      "grad_norm": 0.9177941679954529,
      "learning_rate": 2.586477987421384e-05,
      "loss": 0.4921,
      "num_input_tokens_seen": 1289544,
      "step": 1975
    },
    {
      "epoch": 1.0377358490566038,
      "grad_norm": 1.1500276327133179,
      "learning_rate": 2.593029350104822e-05,
      "loss": 0.4499,
      "num_input_tokens_seen": 1292392,
      "step": 1980
    },
    {
      "epoch": 1.040356394129979,
      "grad_norm": 1.6190882921218872,
      "learning_rate": 2.59958071278826e-05,
      "loss": 0.5525,
      "num_input_tokens_seen": 1295400,
      "step": 1985
    },
    {
      "epoch": 1.0429769392033543,
      "grad_norm": 0.9400023221969604,
      "learning_rate": 2.6061320754716983e-05,
      "loss": 0.4702,
      "num_input_tokens_seen": 1298152,
      "step": 1990
    },
    {
      "epoch": 1.0455974842767295,
      "grad_norm": 1.6061880588531494,
      "learning_rate": 2.6126834381551363e-05,
      "loss": 0.565,
      "num_input_tokens_seen": 1301384,
      "step": 1995
    },
    {
      "epoch": 1.0482180293501049,
      "grad_norm": 1.8873757123947144,
      "learning_rate": 2.6192348008385746e-05,
      "loss": 0.5155,
      "num_input_tokens_seen": 1304296,
      "step": 2000
    },
    {
      "epoch": 1.05083857442348,
      "grad_norm": 1.1825017929077148,
      "learning_rate": 2.6257861635220126e-05,
      "loss": 0.5595,
      "num_input_tokens_seen": 1307048,
      "step": 2005
    },
    {
      "epoch": 1.0534591194968554,
      "grad_norm": 0.879233717918396,
      "learning_rate": 2.6323375262054505e-05,
      "loss": 0.4398,
      "num_input_tokens_seen": 1310792,
      "step": 2010
    },
    {
      "epoch": 1.0560796645702306,
      "grad_norm": 1.3454135656356812,
      "learning_rate": 2.6388888888888892e-05,
      "loss": 0.573,
      "num_input_tokens_seen": 1313448,
      "step": 2015
    },
    {
      "epoch": 1.0587002096436058,
      "grad_norm": 1.9946339130401611,
      "learning_rate": 2.645440251572327e-05,
      "loss": 0.3893,
      "num_input_tokens_seen": 1316200,
      "step": 2020
    },
    {
      "epoch": 1.0613207547169812,
      "grad_norm": 1.100577473640442,
      "learning_rate": 2.6519916142557655e-05,
      "loss": 0.7711,
      "num_input_tokens_seen": 1319720,
      "step": 2025
    },
    {
      "epoch": 1.0639412997903563,
      "grad_norm": 1.5511314868927002,
      "learning_rate": 2.6585429769392034e-05,
      "loss": 0.5039,
      "num_input_tokens_seen": 1323432,
      "step": 2030
    },
    {
      "epoch": 1.0665618448637317,
      "grad_norm": 1.1565511226654053,
      "learning_rate": 2.6650943396226417e-05,
      "loss": 0.4633,
      "num_input_tokens_seen": 1326760,
      "step": 2035
    },
    {
      "epoch": 1.069182389937107,
      "grad_norm": 1.0776543617248535,
      "learning_rate": 2.6716457023060797e-05,
      "loss": 0.4381,
      "num_input_tokens_seen": 1330120,
      "step": 2040
    },
    {
      "epoch": 1.0718029350104823,
      "grad_norm": 1.1901791095733643,
      "learning_rate": 2.6781970649895177e-05,
      "loss": 0.5081,
      "num_input_tokens_seen": 1333032,
      "step": 2045
    },
    {
      "epoch": 1.0744234800838575,
      "grad_norm": 1.964064359664917,
      "learning_rate": 2.6847484276729563e-05,
      "loss": 0.5564,
      "num_input_tokens_seen": 1336296,
      "step": 2050
    },
    {
      "epoch": 1.0770440251572326,
      "grad_norm": 0.8275517821311951,
      "learning_rate": 2.6912997903563943e-05,
      "loss": 0.3459,
      "num_input_tokens_seen": 1339592,
      "step": 2055
    },
    {
      "epoch": 1.079664570230608,
      "grad_norm": 1.4435774087905884,
      "learning_rate": 2.6978511530398326e-05,
      "loss": 0.4609,
      "num_input_tokens_seen": 1342664,
      "step": 2060
    },
    {
      "epoch": 1.0822851153039832,
      "grad_norm": 1.0435208082199097,
      "learning_rate": 2.7044025157232706e-05,
      "loss": 0.5558,
      "num_input_tokens_seen": 1345288,
      "step": 2065
    },
    {
      "epoch": 1.0849056603773586,
      "grad_norm": 1.0102232694625854,
      "learning_rate": 2.710953878406709e-05,
      "loss": 0.4467,
      "num_input_tokens_seen": 1348488,
      "step": 2070
    },
    {
      "epoch": 1.0875262054507338,
      "grad_norm": 1.0095633268356323,
      "learning_rate": 2.717505241090147e-05,
      "loss": 0.5915,
      "num_input_tokens_seen": 1352712,
      "step": 2075
    },
    {
      "epoch": 1.090146750524109,
      "grad_norm": 0.8426591157913208,
      "learning_rate": 2.7240566037735848e-05,
      "loss": 0.4181,
      "num_input_tokens_seen": 1355496,
      "step": 2080
    },
    {
      "epoch": 1.0927672955974843,
      "grad_norm": 0.6997132301330566,
      "learning_rate": 2.7306079664570235e-05,
      "loss": 0.4979,
      "num_input_tokens_seen": 1360552,
      "step": 2085
    },
    {
      "epoch": 1.0953878406708595,
      "grad_norm": 1.202507495880127,
      "learning_rate": 2.737159329140461e-05,
      "loss": 0.3942,
      "num_input_tokens_seen": 1363432,
      "step": 2090
    },
    {
      "epoch": 1.0980083857442349,
      "grad_norm": 1.3303406238555908,
      "learning_rate": 2.7437106918238997e-05,
      "loss": 0.3669,
      "num_input_tokens_seen": 1366824,
      "step": 2095
    },
    {
      "epoch": 1.10062893081761,
      "grad_norm": 1.8759270906448364,
      "learning_rate": 2.7502620545073377e-05,
      "loss": 0.532,
      "num_input_tokens_seen": 1370312,
      "step": 2100
    },
    {
      "epoch": 1.1032494758909852,
      "grad_norm": 1.6064817905426025,
      "learning_rate": 2.7568134171907757e-05,
      "loss": 0.6896,
      "num_input_tokens_seen": 1373160,
      "step": 2105
    },
    {
      "epoch": 1.1058700209643606,
      "grad_norm": 1.4358925819396973,
      "learning_rate": 2.763364779874214e-05,
      "loss": 0.5286,
      "num_input_tokens_seen": 1376008,
      "step": 2110
    },
    {
      "epoch": 1.1084905660377358,
      "grad_norm": 1.3522140979766846,
      "learning_rate": 2.769916142557652e-05,
      "loss": 0.5117,
      "num_input_tokens_seen": 1378440,
      "step": 2115
    },
    {
      "epoch": 1.1111111111111112,
      "grad_norm": 0.9641623497009277,
      "learning_rate": 2.7764675052410906e-05,
      "loss": 0.5102,
      "num_input_tokens_seen": 1382152,
      "step": 2120
    },
    {
      "epoch": 1.1137316561844863,
      "grad_norm": 1.187131643295288,
      "learning_rate": 2.7830188679245282e-05,
      "loss": 0.6187,
      "num_input_tokens_seen": 1385416,
      "step": 2125
    },
    {
      "epoch": 1.1163522012578617,
      "grad_norm": 1.2743580341339111,
      "learning_rate": 2.789570230607967e-05,
      "loss": 0.4246,
      "num_input_tokens_seen": 1389288,
      "step": 2130
    },
    {
      "epoch": 1.118972746331237,
      "grad_norm": 1.557121753692627,
      "learning_rate": 2.796121593291405e-05,
      "loss": 0.4851,
      "num_input_tokens_seen": 1392136,
      "step": 2135
    },
    {
      "epoch": 1.121593291404612,
      "grad_norm": 1.169112205505371,
      "learning_rate": 2.8026729559748428e-05,
      "loss": 0.4842,
      "num_input_tokens_seen": 1395208,
      "step": 2140
    },
    {
      "epoch": 1.1242138364779874,
      "grad_norm": 1.1552951335906982,
      "learning_rate": 2.809224318658281e-05,
      "loss": 0.3817,
      "num_input_tokens_seen": 1398024,
      "step": 2145
    },
    {
      "epoch": 1.1268343815513626,
      "grad_norm": 1.0491361618041992,
      "learning_rate": 2.815775681341719e-05,
      "loss": 0.4111,
      "num_input_tokens_seen": 1401256,
      "step": 2150
    },
    {
      "epoch": 1.129454926624738,
      "grad_norm": 1.4127061367034912,
      "learning_rate": 2.8223270440251577e-05,
      "loss": 0.3341,
      "num_input_tokens_seen": 1403848,
      "step": 2155
    },
    {
      "epoch": 1.1320754716981132,
      "grad_norm": 1.320347547531128,
      "learning_rate": 2.8288784067085954e-05,
      "loss": 0.5555,
      "num_input_tokens_seen": 1408040,
      "step": 2160
    },
    {
      "epoch": 1.1346960167714886,
      "grad_norm": 2.3959596157073975,
      "learning_rate": 2.8354297693920333e-05,
      "loss": 0.5699,
      "num_input_tokens_seen": 1411688,
      "step": 2165
    },
    {
      "epoch": 1.1373165618448637,
      "grad_norm": 1.3890527486801147,
      "learning_rate": 2.841981132075472e-05,
      "loss": 0.5595,
      "num_input_tokens_seen": 1415464,
      "step": 2170
    },
    {
      "epoch": 1.139937106918239,
      "grad_norm": 2.1928627490997314,
      "learning_rate": 2.84853249475891e-05,
      "loss": 0.382,
      "num_input_tokens_seen": 1418664,
      "step": 2175
    },
    {
      "epoch": 1.1425576519916143,
      "grad_norm": 1.394763469696045,
      "learning_rate": 2.8550838574423482e-05,
      "loss": 0.405,
      "num_input_tokens_seen": 1422088,
      "step": 2180
    },
    {
      "epoch": 1.1451781970649895,
      "grad_norm": 1.6144269704818726,
      "learning_rate": 2.8616352201257862e-05,
      "loss": 0.4127,
      "num_input_tokens_seen": 1424872,
      "step": 2185
    },
    {
      "epoch": 1.1477987421383649,
      "grad_norm": 0.9721892476081848,
      "learning_rate": 2.868186582809225e-05,
      "loss": 0.5824,
      "num_input_tokens_seen": 1429320,
      "step": 2190
    },
    {
      "epoch": 1.15041928721174,
      "grad_norm": 1.0294116735458374,
      "learning_rate": 2.8747379454926625e-05,
      "loss": 0.4546,
      "num_input_tokens_seen": 1433704,
      "step": 2195
    },
    {
      "epoch": 1.1530398322851152,
      "grad_norm": 1.2692652940750122,
      "learning_rate": 2.8812893081761005e-05,
      "loss": 0.4911,
      "num_input_tokens_seen": 1436872,
      "step": 2200
    },
    {
      "epoch": 1.1556603773584906,
      "grad_norm": 1.910491704940796,
      "learning_rate": 2.887840670859539e-05,
      "loss": 0.4726,
      "num_input_tokens_seen": 1440104,
      "step": 2205
    },
    {
      "epoch": 1.1582809224318658,
      "grad_norm": 1.0858498811721802,
      "learning_rate": 2.894392033542977e-05,
      "loss": 0.6264,
      "num_input_tokens_seen": 1443176,
      "step": 2210
    },
    {
      "epoch": 1.1609014675052411,
      "grad_norm": 0.8792463541030884,
      "learning_rate": 2.9009433962264154e-05,
      "loss": 0.4375,
      "num_input_tokens_seen": 1446440,
      "step": 2215
    },
    {
      "epoch": 1.1635220125786163,
      "grad_norm": 1.7898666858673096,
      "learning_rate": 2.9074947589098533e-05,
      "loss": 0.4171,
      "num_input_tokens_seen": 1450120,
      "step": 2220
    },
    {
      "epoch": 1.1661425576519917,
      "grad_norm": 1.560465931892395,
      "learning_rate": 2.9140461215932913e-05,
      "loss": 0.442,
      "num_input_tokens_seen": 1453608,
      "step": 2225
    },
    {
      "epoch": 1.1687631027253669,
      "grad_norm": 1.0627460479736328,
      "learning_rate": 2.9205974842767296e-05,
      "loss": 0.5879,
      "num_input_tokens_seen": 1457832,
      "step": 2230
    },
    {
      "epoch": 1.171383647798742,
      "grad_norm": 0.8470693230628967,
      "learning_rate": 2.9271488469601676e-05,
      "loss": 0.5157,
      "num_input_tokens_seen": 1462280,
      "step": 2235
    },
    {
      "epoch": 1.1740041928721174,
      "grad_norm": 0.9859356880187988,
      "learning_rate": 2.9337002096436062e-05,
      "loss": 0.4479,
      "num_input_tokens_seen": 1466024,
      "step": 2240
    },
    {
      "epoch": 1.1766247379454926,
      "grad_norm": 1.0907349586486816,
      "learning_rate": 2.9402515723270442e-05,
      "loss": 0.4378,
      "num_input_tokens_seen": 1468872,
      "step": 2245
    },
    {
      "epoch": 1.179245283018868,
      "grad_norm": 1.8835879564285278,
      "learning_rate": 2.9468029350104825e-05,
      "loss": 0.4721,
      "num_input_tokens_seen": 1471336,
      "step": 2250
    },
    {
      "epoch": 1.1818658280922432,
      "grad_norm": 1.5184288024902344,
      "learning_rate": 2.9533542976939205e-05,
      "loss": 0.5936,
      "num_input_tokens_seen": 1473928,
      "step": 2255
    },
    {
      "epoch": 1.1844863731656186,
      "grad_norm": 1.1125786304473877,
      "learning_rate": 2.9599056603773584e-05,
      "loss": 0.407,
      "num_input_tokens_seen": 1478728,
      "step": 2260
    },
    {
      "epoch": 1.1871069182389937,
      "grad_norm": 1.1606321334838867,
      "learning_rate": 2.9664570230607968e-05,
      "loss": 0.4638,
      "num_input_tokens_seen": 1481576,
      "step": 2265
    },
    {
      "epoch": 1.189727463312369,
      "grad_norm": 1.6021349430084229,
      "learning_rate": 2.9730083857442347e-05,
      "loss": 0.5146,
      "num_input_tokens_seen": 1484104,
      "step": 2270
    },
    {
      "epoch": 1.1923480083857443,
      "grad_norm": 1.3095132112503052,
      "learning_rate": 2.9795597484276734e-05,
      "loss": 0.5386,
      "num_input_tokens_seen": 1487048,
      "step": 2275
    },
    {
      "epoch": 1.1949685534591195,
      "grad_norm": 1.0416784286499023,
      "learning_rate": 2.9861111111111113e-05,
      "loss": 0.4406,
      "num_input_tokens_seen": 1490152,
      "step": 2280
    },
    {
      "epoch": 1.1975890985324948,
      "grad_norm": 1.239342451095581,
      "learning_rate": 2.9926624737945496e-05,
      "loss": 0.3945,
      "num_input_tokens_seen": 1492648,
      "step": 2285
    },
    {
      "epoch": 1.20020964360587,
      "grad_norm": 1.1947742700576782,
      "learning_rate": 2.9992138364779876e-05,
      "loss": 0.5533,
      "num_input_tokens_seen": 1495272,
      "step": 2290
    },
    {
      "epoch": 1.2028301886792452,
      "grad_norm": 1.2276586294174194,
      "learning_rate": 3.0057651991614256e-05,
      "loss": 0.4806,
      "num_input_tokens_seen": 1497832,
      "step": 2295
    },
    {
      "epoch": 1.2054507337526206,
      "grad_norm": 1.2546507120132446,
      "learning_rate": 3.012316561844864e-05,
      "loss": 0.3666,
      "num_input_tokens_seen": 1501896,
      "step": 2300
    },
    {
      "epoch": 1.2080712788259957,
      "grad_norm": 1.5474010705947876,
      "learning_rate": 3.018867924528302e-05,
      "loss": 0.4347,
      "num_input_tokens_seen": 1504328,
      "step": 2305
    },
    {
      "epoch": 1.2106918238993711,
      "grad_norm": 1.100217580795288,
      "learning_rate": 3.0254192872117405e-05,
      "loss": 0.4981,
      "num_input_tokens_seen": 1508264,
      "step": 2310
    },
    {
      "epoch": 1.2133123689727463,
      "grad_norm": 1.3960362672805786,
      "learning_rate": 3.0319706498951785e-05,
      "loss": 0.5756,
      "num_input_tokens_seen": 1511048,
      "step": 2315
    },
    {
      "epoch": 1.2159329140461215,
      "grad_norm": 2.0230188369750977,
      "learning_rate": 3.038522012578616e-05,
      "loss": 0.6449,
      "num_input_tokens_seen": 1513928,
      "step": 2320
    },
    {
      "epoch": 1.2185534591194969,
      "grad_norm": 1.0618895292282104,
      "learning_rate": 3.0450733752620547e-05,
      "loss": 0.4031,
      "num_input_tokens_seen": 1517512,
      "step": 2325
    },
    {
      "epoch": 1.221174004192872,
      "grad_norm": 2.308286666870117,
      "learning_rate": 3.0516247379454927e-05,
      "loss": 0.365,
      "num_input_tokens_seen": 1521256,
      "step": 2330
    },
    {
      "epoch": 1.2237945492662474,
      "grad_norm": 1.8189404010772705,
      "learning_rate": 3.0581761006289314e-05,
      "loss": 0.5966,
      "num_input_tokens_seen": 1523144,
      "step": 2335
    },
    {
      "epoch": 1.2264150943396226,
      "grad_norm": 1.240368127822876,
      "learning_rate": 3.064727463312369e-05,
      "loss": 0.4564,
      "num_input_tokens_seen": 1527624,
      "step": 2340
    },
    {
      "epoch": 1.229035639412998,
      "grad_norm": 1.203876256942749,
      "learning_rate": 3.071278825995807e-05,
      "loss": 0.4483,
      "num_input_tokens_seen": 1531560,
      "step": 2345
    },
    {
      "epoch": 1.2316561844863732,
      "grad_norm": 1.4041351079940796,
      "learning_rate": 3.0778301886792456e-05,
      "loss": 0.4513,
      "num_input_tokens_seen": 1533832,
      "step": 2350
    },
    {
      "epoch": 1.2342767295597485,
      "grad_norm": 1.0257201194763184,
      "learning_rate": 3.084381551362683e-05,
      "loss": 0.4732,
      "num_input_tokens_seen": 1536584,
      "step": 2355
    },
    {
      "epoch": 1.2368972746331237,
      "grad_norm": 0.8661887645721436,
      "learning_rate": 3.0909329140461215e-05,
      "loss": 0.452,
      "num_input_tokens_seen": 1539720,
      "step": 2360
    },
    {
      "epoch": 1.2395178197064989,
      "grad_norm": 1.0366275310516357,
      "learning_rate": 3.09748427672956e-05,
      "loss": 0.3658,
      "num_input_tokens_seen": 1543112,
      "step": 2365
    },
    {
      "epoch": 1.2421383647798743,
      "grad_norm": 1.532339334487915,
      "learning_rate": 3.104035639412998e-05,
      "loss": 0.4978,
      "num_input_tokens_seen": 1546568,
      "step": 2370
    },
    {
      "epoch": 1.2447589098532494,
      "grad_norm": 1.6136833429336548,
      "learning_rate": 3.1105870020964365e-05,
      "loss": 0.4951,
      "num_input_tokens_seen": 1549864,
      "step": 2375
    },
    {
      "epoch": 1.2473794549266248,
      "grad_norm": 1.0414328575134277,
      "learning_rate": 3.117138364779874e-05,
      "loss": 0.4202,
      "num_input_tokens_seen": 1552456,
      "step": 2380
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.688106656074524,
      "learning_rate": 3.1236897274633124e-05,
      "loss": 0.5531,
      "num_input_tokens_seen": 1555240,
      "step": 2385
    },
    {
      "epoch": 1.2526205450733752,
      "grad_norm": 1.3146761655807495,
      "learning_rate": 3.130241090146751e-05,
      "loss": 0.3974,
      "num_input_tokens_seen": 1558568,
      "step": 2390
    },
    {
      "epoch": 1.2552410901467506,
      "grad_norm": 1.9374762773513794,
      "learning_rate": 3.136792452830189e-05,
      "loss": 0.5166,
      "num_input_tokens_seen": 1561288,
      "step": 2395
    },
    {
      "epoch": 1.2578616352201257,
      "grad_norm": 2.064882516860962,
      "learning_rate": 3.1433438155136266e-05,
      "loss": 0.5356,
      "num_input_tokens_seen": 1564200,
      "step": 2400
    },
    {
      "epoch": 1.2604821802935011,
      "grad_norm": 0.984368622303009,
      "learning_rate": 3.1498951781970656e-05,
      "loss": 0.4814,
      "num_input_tokens_seen": 1567208,
      "step": 2405
    },
    {
      "epoch": 1.2631027253668763,
      "grad_norm": 1.0797394514083862,
      "learning_rate": 3.156446540880503e-05,
      "loss": 0.4207,
      "num_input_tokens_seen": 1571240,
      "step": 2410
    },
    {
      "epoch": 1.2657232704402515,
      "grad_norm": 2.8989994525909424,
      "learning_rate": 3.162997903563941e-05,
      "loss": 0.473,
      "num_input_tokens_seen": 1574376,
      "step": 2415
    },
    {
      "epoch": 1.2683438155136268,
      "grad_norm": 1.2793159484863281,
      "learning_rate": 3.16954926624738e-05,
      "loss": 0.5321,
      "num_input_tokens_seen": 1577672,
      "step": 2420
    },
    {
      "epoch": 1.270964360587002,
      "grad_norm": 0.8851102590560913,
      "learning_rate": 3.1761006289308175e-05,
      "loss": 0.4264,
      "num_input_tokens_seen": 1581192,
      "step": 2425
    },
    {
      "epoch": 1.2735849056603774,
      "grad_norm": 0.8006277084350586,
      "learning_rate": 3.182651991614256e-05,
      "loss": 0.4068,
      "num_input_tokens_seen": 1584424,
      "step": 2430
    },
    {
      "epoch": 1.2762054507337526,
      "grad_norm": 1.6265790462493896,
      "learning_rate": 3.189203354297694e-05,
      "loss": 0.4625,
      "num_input_tokens_seen": 1587912,
      "step": 2435
    },
    {
      "epoch": 1.2788259958071277,
      "grad_norm": 0.8737910389900208,
      "learning_rate": 3.1957547169811324e-05,
      "loss": 0.3474,
      "num_input_tokens_seen": 1591624,
      "step": 2440
    },
    {
      "epoch": 1.2814465408805031,
      "grad_norm": 1.0248068571090698,
      "learning_rate": 3.202306079664571e-05,
      "loss": 0.5013,
      "num_input_tokens_seen": 1594376,
      "step": 2445
    },
    {
      "epoch": 1.2840670859538785,
      "grad_norm": 0.8238078355789185,
      "learning_rate": 3.2088574423480084e-05,
      "loss": 0.4359,
      "num_input_tokens_seen": 1597608,
      "step": 2450
    },
    {
      "epoch": 1.2866876310272537,
      "grad_norm": 1.5882819890975952,
      "learning_rate": 3.215408805031447e-05,
      "loss": 0.4301,
      "num_input_tokens_seen": 1600136,
      "step": 2455
    },
    {
      "epoch": 1.2893081761006289,
      "grad_norm": 1.2747819423675537,
      "learning_rate": 3.221960167714885e-05,
      "loss": 0.47,
      "num_input_tokens_seen": 1603592,
      "step": 2460
    },
    {
      "epoch": 1.2919287211740043,
      "grad_norm": 1.469145655632019,
      "learning_rate": 3.228511530398323e-05,
      "loss": 0.573,
      "num_input_tokens_seen": 1606504,
      "step": 2465
    },
    {
      "epoch": 1.2945492662473794,
      "grad_norm": 1.1032861471176147,
      "learning_rate": 3.235062893081761e-05,
      "loss": 0.4185,
      "num_input_tokens_seen": 1613064,
      "step": 2470
    },
    {
      "epoch": 1.2971698113207548,
      "grad_norm": 0.8421196341514587,
      "learning_rate": 3.241614255765199e-05,
      "loss": 0.4498,
      "num_input_tokens_seen": 1616136,
      "step": 2475
    },
    {
      "epoch": 1.29979035639413,
      "grad_norm": 1.3260595798492432,
      "learning_rate": 3.2481656184486375e-05,
      "loss": 0.4345,
      "num_input_tokens_seen": 1618792,
      "step": 2480
    },
    {
      "epoch": 1.3024109014675052,
      "grad_norm": 2.0527920722961426,
      "learning_rate": 3.254716981132075e-05,
      "loss": 0.3777,
      "num_input_tokens_seen": 1621096,
      "step": 2485
    },
    {
      "epoch": 1.3050314465408805,
      "grad_norm": 1.6384433507919312,
      "learning_rate": 3.261268343815514e-05,
      "loss": 0.4064,
      "num_input_tokens_seen": 1623720,
      "step": 2490
    },
    {
      "epoch": 1.3076519916142557,
      "grad_norm": 1.2367465496063232,
      "learning_rate": 3.267819706498952e-05,
      "loss": 0.4822,
      "num_input_tokens_seen": 1626888,
      "step": 2495
    },
    {
      "epoch": 1.310272536687631,
      "grad_norm": 1.2715122699737549,
      "learning_rate": 3.27437106918239e-05,
      "loss": 0.6021,
      "num_input_tokens_seen": 1630472,
      "step": 2500
    },
    {
      "epoch": 1.3128930817610063,
      "grad_norm": 1.1390084028244019,
      "learning_rate": 3.2809224318658284e-05,
      "loss": 0.4244,
      "num_input_tokens_seen": 1633192,
      "step": 2505
    },
    {
      "epoch": 1.3155136268343814,
      "grad_norm": 0.6730809211730957,
      "learning_rate": 3.287473794549266e-05,
      "loss": 0.5112,
      "num_input_tokens_seen": 1637128,
      "step": 2510
    },
    {
      "epoch": 1.3181341719077568,
      "grad_norm": 0.7717965841293335,
      "learning_rate": 3.294025157232704e-05,
      "loss": 0.4847,
      "num_input_tokens_seen": 1640456,
      "step": 2515
    },
    {
      "epoch": 1.320754716981132,
      "grad_norm": 1.540792465209961,
      "learning_rate": 3.3005765199161426e-05,
      "loss": 0.5761,
      "num_input_tokens_seen": 1644008,
      "step": 2520
    },
    {
      "epoch": 1.3233752620545074,
      "grad_norm": 1.50412118434906,
      "learning_rate": 3.307127882599581e-05,
      "loss": 0.4404,
      "num_input_tokens_seen": 1647720,
      "step": 2525
    },
    {
      "epoch": 1.3259958071278826,
      "grad_norm": 2.297187328338623,
      "learning_rate": 3.313679245283019e-05,
      "loss": 0.5949,
      "num_input_tokens_seen": 1650440,
      "step": 2530
    },
    {
      "epoch": 1.3286163522012577,
      "grad_norm": 1.4714189767837524,
      "learning_rate": 3.320230607966457e-05,
      "loss": 0.5759,
      "num_input_tokens_seen": 1653352,
      "step": 2535
    },
    {
      "epoch": 1.3312368972746331,
      "grad_norm": 0.9426354169845581,
      "learning_rate": 3.326781970649895e-05,
      "loss": 0.4521,
      "num_input_tokens_seen": 1656360,
      "step": 2540
    },
    {
      "epoch": 1.3338574423480085,
      "grad_norm": 0.8318505883216858,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.5844,
      "num_input_tokens_seen": 1659336,
      "step": 2545
    },
    {
      "epoch": 1.3364779874213837,
      "grad_norm": 3.3244550228118896,
      "learning_rate": 3.339884696016772e-05,
      "loss": 0.6533,
      "num_input_tokens_seen": 1661896,
      "step": 2550
    },
    {
      "epoch": 1.3390985324947589,
      "grad_norm": 0.673663318157196,
      "learning_rate": 3.3464360587002094e-05,
      "loss": 0.3581,
      "num_input_tokens_seen": 1665832,
      "step": 2555
    },
    {
      "epoch": 1.3417190775681342,
      "grad_norm": 1.7450103759765625,
      "learning_rate": 3.3529874213836484e-05,
      "loss": 0.5172,
      "num_input_tokens_seen": 1668520,
      "step": 2560
    },
    {
      "epoch": 1.3443396226415094,
      "grad_norm": 0.9766102433204651,
      "learning_rate": 3.359538784067086e-05,
      "loss": 0.4554,
      "num_input_tokens_seen": 1671912,
      "step": 2565
    },
    {
      "epoch": 1.3469601677148848,
      "grad_norm": 1.018040657043457,
      "learning_rate": 3.3660901467505244e-05,
      "loss": 0.4609,
      "num_input_tokens_seen": 1675080,
      "step": 2570
    },
    {
      "epoch": 1.34958071278826,
      "grad_norm": 1.5062663555145264,
      "learning_rate": 3.3726415094339627e-05,
      "loss": 0.3312,
      "num_input_tokens_seen": 1681512,
      "step": 2575
    },
    {
      "epoch": 1.3522012578616351,
      "grad_norm": 1.0331921577453613,
      "learning_rate": 3.3791928721174e-05,
      "loss": 0.4662,
      "num_input_tokens_seen": 1685000,
      "step": 2580
    },
    {
      "epoch": 1.3548218029350105,
      "grad_norm": 1.478691577911377,
      "learning_rate": 3.3857442348008386e-05,
      "loss": 0.4279,
      "num_input_tokens_seen": 1687528,
      "step": 2585
    },
    {
      "epoch": 1.3574423480083857,
      "grad_norm": 1.1968681812286377,
      "learning_rate": 3.392295597484277e-05,
      "loss": 0.5724,
      "num_input_tokens_seen": 1690280,
      "step": 2590
    },
    {
      "epoch": 1.360062893081761,
      "grad_norm": 1.1447725296020508,
      "learning_rate": 3.398846960167715e-05,
      "loss": 0.3845,
      "num_input_tokens_seen": 1693352,
      "step": 2595
    },
    {
      "epoch": 1.3626834381551363,
      "grad_norm": 1.540444254875183,
      "learning_rate": 3.4053983228511535e-05,
      "loss": 0.4538,
      "num_input_tokens_seen": 1696456,
      "step": 2600
    },
    {
      "epoch": 1.3653039832285114,
      "grad_norm": 0.8005126118659973,
      "learning_rate": 3.411949685534591e-05,
      "loss": 0.4132,
      "num_input_tokens_seen": 1700488,
      "step": 2605
    },
    {
      "epoch": 1.3679245283018868,
      "grad_norm": 1.765428066253662,
      "learning_rate": 3.4185010482180295e-05,
      "loss": 0.4135,
      "num_input_tokens_seen": 1703304,
      "step": 2610
    },
    {
      "epoch": 1.370545073375262,
      "grad_norm": 1.8026145696640015,
      "learning_rate": 3.425052410901468e-05,
      "loss": 0.5016,
      "num_input_tokens_seen": 1705864,
      "step": 2615
    },
    {
      "epoch": 1.3731656184486374,
      "grad_norm": 0.9400510787963867,
      "learning_rate": 3.431603773584906e-05,
      "loss": 0.5819,
      "num_input_tokens_seen": 1708616,
      "step": 2620
    },
    {
      "epoch": 1.3757861635220126,
      "grad_norm": 0.6010936498641968,
      "learning_rate": 3.438155136268344e-05,
      "loss": 0.4647,
      "num_input_tokens_seen": 1712232,
      "step": 2625
    },
    {
      "epoch": 1.3784067085953877,
      "grad_norm": 2.118695020675659,
      "learning_rate": 3.444706498951782e-05,
      "loss": 0.3136,
      "num_input_tokens_seen": 1715144,
      "step": 2630
    },
    {
      "epoch": 1.381027253668763,
      "grad_norm": 2.0431413650512695,
      "learning_rate": 3.45125786163522e-05,
      "loss": 0.4222,
      "num_input_tokens_seen": 1717800,
      "step": 2635
    },
    {
      "epoch": 1.3836477987421385,
      "grad_norm": 1.4738153219223022,
      "learning_rate": 3.457809224318658e-05,
      "loss": 0.4642,
      "num_input_tokens_seen": 1721128,
      "step": 2640
    },
    {
      "epoch": 1.3862683438155137,
      "grad_norm": 0.7047557234764099,
      "learning_rate": 3.464360587002097e-05,
      "loss": 0.4404,
      "num_input_tokens_seen": 1724328,
      "step": 2645
    },
    {
      "epoch": 1.3888888888888888,
      "grad_norm": 1.3799458742141724,
      "learning_rate": 3.4709119496855346e-05,
      "loss": 0.527,
      "num_input_tokens_seen": 1728328,
      "step": 2650
    },
    {
      "epoch": 1.3915094339622642,
      "grad_norm": 1.121680736541748,
      "learning_rate": 3.477463312368973e-05,
      "loss": 0.4407,
      "num_input_tokens_seen": 1731080,
      "step": 2655
    },
    {
      "epoch": 1.3941299790356394,
      "grad_norm": 1.1330811977386475,
      "learning_rate": 3.484014675052411e-05,
      "loss": 0.5066,
      "num_input_tokens_seen": 1736008,
      "step": 2660
    },
    {
      "epoch": 1.3967505241090148,
      "grad_norm": 1.5804840326309204,
      "learning_rate": 3.490566037735849e-05,
      "loss": 0.3868,
      "num_input_tokens_seen": 1739368,
      "step": 2665
    },
    {
      "epoch": 1.39937106918239,
      "grad_norm": 1.5836706161499023,
      "learning_rate": 3.497117400419288e-05,
      "loss": 0.6773,
      "num_input_tokens_seen": 1742408,
      "step": 2670
    },
    {
      "epoch": 1.4019916142557651,
      "grad_norm": 1.6122781038284302,
      "learning_rate": 3.5036687631027254e-05,
      "loss": 0.5109,
      "num_input_tokens_seen": 1746120,
      "step": 2675
    },
    {
      "epoch": 1.4046121593291405,
      "grad_norm": 1.2775896787643433,
      "learning_rate": 3.510220125786164e-05,
      "loss": 0.5083,
      "num_input_tokens_seen": 1749096,
      "step": 2680
    },
    {
      "epoch": 1.4072327044025157,
      "grad_norm": 3.4531774520874023,
      "learning_rate": 3.516771488469602e-05,
      "loss": 0.5244,
      "num_input_tokens_seen": 1751624,
      "step": 2685
    },
    {
      "epoch": 1.409853249475891,
      "grad_norm": 1.7423375844955444,
      "learning_rate": 3.5233228511530397e-05,
      "loss": 0.4094,
      "num_input_tokens_seen": 1754216,
      "step": 2690
    },
    {
      "epoch": 1.4124737945492662,
      "grad_norm": 0.7744272351264954,
      "learning_rate": 3.529874213836478e-05,
      "loss": 0.5469,
      "num_input_tokens_seen": 1758280,
      "step": 2695
    },
    {
      "epoch": 1.4150943396226414,
      "grad_norm": 1.584863305091858,
      "learning_rate": 3.536425576519916e-05,
      "loss": 0.4638,
      "num_input_tokens_seen": 1761352,
      "step": 2700
    },
    {
      "epoch": 1.4177148846960168,
      "grad_norm": 0.897788941860199,
      "learning_rate": 3.5429769392033546e-05,
      "loss": 0.4154,
      "num_input_tokens_seen": 1764648,
      "step": 2705
    },
    {
      "epoch": 1.420335429769392,
      "grad_norm": 1.2170075178146362,
      "learning_rate": 3.549528301886792e-05,
      "loss": 0.5887,
      "num_input_tokens_seen": 1767592,
      "step": 2710
    },
    {
      "epoch": 1.4229559748427674,
      "grad_norm": 0.8976203799247742,
      "learning_rate": 3.556079664570231e-05,
      "loss": 0.4266,
      "num_input_tokens_seen": 1774600,
      "step": 2715
    },
    {
      "epoch": 1.4255765199161425,
      "grad_norm": 2.0601093769073486,
      "learning_rate": 3.562631027253669e-05,
      "loss": 0.5598,
      "num_input_tokens_seen": 1778248,
      "step": 2720
    },
    {
      "epoch": 1.4281970649895177,
      "grad_norm": 0.9280978441238403,
      "learning_rate": 3.569182389937107e-05,
      "loss": 0.449,
      "num_input_tokens_seen": 1781608,
      "step": 2725
    },
    {
      "epoch": 1.430817610062893,
      "grad_norm": 1.0596946477890015,
      "learning_rate": 3.5757337526205454e-05,
      "loss": 0.6144,
      "num_input_tokens_seen": 1784776,
      "step": 2730
    },
    {
      "epoch": 1.4334381551362683,
      "grad_norm": 1.4671211242675781,
      "learning_rate": 3.582285115303983e-05,
      "loss": 0.5452,
      "num_input_tokens_seen": 1787688,
      "step": 2735
    },
    {
      "epoch": 1.4360587002096437,
      "grad_norm": 0.8749681711196899,
      "learning_rate": 3.588836477987422e-05,
      "loss": 0.5631,
      "num_input_tokens_seen": 1791016,
      "step": 2740
    },
    {
      "epoch": 1.4386792452830188,
      "grad_norm": 2.2055110931396484,
      "learning_rate": 3.59538784067086e-05,
      "loss": 0.4865,
      "num_input_tokens_seen": 1794280,
      "step": 2745
    },
    {
      "epoch": 1.441299790356394,
      "grad_norm": 1.793230414390564,
      "learning_rate": 3.601939203354298e-05,
      "loss": 0.3811,
      "num_input_tokens_seen": 1797480,
      "step": 2750
    },
    {
      "epoch": 1.4439203354297694,
      "grad_norm": 1.085011601448059,
      "learning_rate": 3.608490566037736e-05,
      "loss": 0.6,
      "num_input_tokens_seen": 1801160,
      "step": 2755
    },
    {
      "epoch": 1.4465408805031448,
      "grad_norm": 1.4905444383621216,
      "learning_rate": 3.615041928721174e-05,
      "loss": 0.5601,
      "num_input_tokens_seen": 1803464,
      "step": 2760
    },
    {
      "epoch": 1.44916142557652,
      "grad_norm": 1.3433620929718018,
      "learning_rate": 3.621593291404612e-05,
      "loss": 0.4327,
      "num_input_tokens_seen": 1806728,
      "step": 2765
    },
    {
      "epoch": 1.4517819706498951,
      "grad_norm": 0.8563610315322876,
      "learning_rate": 3.6281446540880505e-05,
      "loss": 0.4127,
      "num_input_tokens_seen": 1810088,
      "step": 2770
    },
    {
      "epoch": 1.4544025157232705,
      "grad_norm": 0.6807305812835693,
      "learning_rate": 3.634696016771489e-05,
      "loss": 0.407,
      "num_input_tokens_seen": 1815144,
      "step": 2775
    },
    {
      "epoch": 1.4570230607966457,
      "grad_norm": 0.6236798167228699,
      "learning_rate": 3.6412473794549265e-05,
      "loss": 0.3922,
      "num_input_tokens_seen": 1818792,
      "step": 2780
    },
    {
      "epoch": 1.459643605870021,
      "grad_norm": 1.0571566820144653,
      "learning_rate": 3.647798742138365e-05,
      "loss": 0.5781,
      "num_input_tokens_seen": 1821768,
      "step": 2785
    },
    {
      "epoch": 1.4622641509433962,
      "grad_norm": 1.401486873626709,
      "learning_rate": 3.654350104821803e-05,
      "loss": 0.5927,
      "num_input_tokens_seen": 1825512,
      "step": 2790
    },
    {
      "epoch": 1.4648846960167714,
      "grad_norm": 1.1671866178512573,
      "learning_rate": 3.6609014675052414e-05,
      "loss": 0.3364,
      "num_input_tokens_seen": 1828968,
      "step": 2795
    },
    {
      "epoch": 1.4675052410901468,
      "grad_norm": 1.1266391277313232,
      "learning_rate": 3.66745283018868e-05,
      "loss": 0.5168,
      "num_input_tokens_seen": 1831336,
      "step": 2800
    },
    {
      "epoch": 1.470125786163522,
      "grad_norm": 0.760977566242218,
      "learning_rate": 3.6740041928721173e-05,
      "loss": 0.5038,
      "num_input_tokens_seen": 1834536,
      "step": 2805
    },
    {
      "epoch": 1.4727463312368974,
      "grad_norm": 0.9826387166976929,
      "learning_rate": 3.6805555555555556e-05,
      "loss": 0.4532,
      "num_input_tokens_seen": 1837480,
      "step": 2810
    },
    {
      "epoch": 1.4753668763102725,
      "grad_norm": 2.2463929653167725,
      "learning_rate": 3.687106918238994e-05,
      "loss": 0.5398,
      "num_input_tokens_seen": 1840808,
      "step": 2815
    },
    {
      "epoch": 1.4779874213836477,
      "grad_norm": 1.0031195878982544,
      "learning_rate": 3.6936582809224316e-05,
      "loss": 0.5923,
      "num_input_tokens_seen": 1844392,
      "step": 2820
    },
    {
      "epoch": 1.480607966457023,
      "grad_norm": 1.0289939641952515,
      "learning_rate": 3.7002096436058706e-05,
      "loss": 0.3768,
      "num_input_tokens_seen": 1848008,
      "step": 2825
    },
    {
      "epoch": 1.4832285115303983,
      "grad_norm": 1.1709580421447754,
      "learning_rate": 3.706761006289308e-05,
      "loss": 0.4032,
      "num_input_tokens_seen": 1850888,
      "step": 2830
    },
    {
      "epoch": 1.4858490566037736,
      "grad_norm": 2.656733274459839,
      "learning_rate": 3.7133123689727465e-05,
      "loss": 0.6512,
      "num_input_tokens_seen": 1854312,
      "step": 2835
    },
    {
      "epoch": 1.4884696016771488,
      "grad_norm": 0.9019132256507874,
      "learning_rate": 3.719863731656185e-05,
      "loss": 0.4555,
      "num_input_tokens_seen": 1858792,
      "step": 2840
    },
    {
      "epoch": 1.491090146750524,
      "grad_norm": 0.951124370098114,
      "learning_rate": 3.7264150943396224e-05,
      "loss": 0.446,
      "num_input_tokens_seen": 1861672,
      "step": 2845
    },
    {
      "epoch": 1.4937106918238994,
      "grad_norm": 1.3014642000198364,
      "learning_rate": 3.732966457023061e-05,
      "loss": 0.4186,
      "num_input_tokens_seen": 1864360,
      "step": 2850
    },
    {
      "epoch": 1.4963312368972748,
      "grad_norm": 2.0450518131256104,
      "learning_rate": 3.739517819706499e-05,
      "loss": 0.4589,
      "num_input_tokens_seen": 1867048,
      "step": 2855
    },
    {
      "epoch": 1.49895178197065,
      "grad_norm": 0.7959545850753784,
      "learning_rate": 3.7460691823899374e-05,
      "loss": 0.4346,
      "num_input_tokens_seen": 1870280,
      "step": 2860
    },
    {
      "epoch": 1.501572327044025,
      "grad_norm": 1.3625580072402954,
      "learning_rate": 3.752620545073376e-05,
      "loss": 0.636,
      "num_input_tokens_seen": 1873896,
      "step": 2865
    },
    {
      "epoch": 1.5041928721174003,
      "grad_norm": 0.9380871653556824,
      "learning_rate": 3.759171907756814e-05,
      "loss": 0.4886,
      "num_input_tokens_seen": 1876808,
      "step": 2870
    },
    {
      "epoch": 1.5068134171907757,
      "grad_norm": 1.1384090185165405,
      "learning_rate": 3.7657232704402516e-05,
      "loss": 0.4951,
      "num_input_tokens_seen": 1880168,
      "step": 2875
    },
    {
      "epoch": 1.509433962264151,
      "grad_norm": 1.8057973384857178,
      "learning_rate": 3.77227463312369e-05,
      "loss": 0.3986,
      "num_input_tokens_seen": 1882824,
      "step": 2880
    },
    {
      "epoch": 1.5120545073375262,
      "grad_norm": 0.8760766386985779,
      "learning_rate": 3.778825995807128e-05,
      "loss": 0.44,
      "num_input_tokens_seen": 1886184,
      "step": 2885
    },
    {
      "epoch": 1.5146750524109014,
      "grad_norm": 0.8551605343818665,
      "learning_rate": 3.785377358490566e-05,
      "loss": 0.5052,
      "num_input_tokens_seen": 1888968,
      "step": 2890
    },
    {
      "epoch": 1.5172955974842768,
      "grad_norm": 1.2571018934249878,
      "learning_rate": 3.791928721174005e-05,
      "loss": 0.5608,
      "num_input_tokens_seen": 1892232,
      "step": 2895
    },
    {
      "epoch": 1.519916142557652,
      "grad_norm": 1.0707275867462158,
      "learning_rate": 3.7984800838574425e-05,
      "loss": 0.4287,
      "num_input_tokens_seen": 1895240,
      "step": 2900
    },
    {
      "epoch": 1.5225366876310273,
      "grad_norm": 0.4566957354545593,
      "learning_rate": 3.805031446540881e-05,
      "loss": 0.3324,
      "num_input_tokens_seen": 1899144,
      "step": 2905
    },
    {
      "epoch": 1.5251572327044025,
      "grad_norm": 1.0467137098312378,
      "learning_rate": 3.811582809224319e-05,
      "loss": 0.4383,
      "num_input_tokens_seen": 1901896,
      "step": 2910
    },
    {
      "epoch": 1.5277777777777777,
      "grad_norm": 0.8400961756706238,
      "learning_rate": 3.818134171907757e-05,
      "loss": 0.4515,
      "num_input_tokens_seen": 1904936,
      "step": 2915
    },
    {
      "epoch": 1.530398322851153,
      "grad_norm": 0.9379619359970093,
      "learning_rate": 3.824685534591195e-05,
      "loss": 0.4959,
      "num_input_tokens_seen": 1908328,
      "step": 2920
    },
    {
      "epoch": 1.5330188679245285,
      "grad_norm": 1.0034526586532593,
      "learning_rate": 3.831236897274633e-05,
      "loss": 0.5499,
      "num_input_tokens_seen": 1911496,
      "step": 2925
    },
    {
      "epoch": 1.5356394129979036,
      "grad_norm": 1.072290062904358,
      "learning_rate": 3.8377882599580716e-05,
      "loss": 0.3959,
      "num_input_tokens_seen": 1913992,
      "step": 2930
    },
    {
      "epoch": 1.5382599580712788,
      "grad_norm": 1.6351321935653687,
      "learning_rate": 3.844339622641509e-05,
      "loss": 0.5283,
      "num_input_tokens_seen": 1917640,
      "step": 2935
    },
    {
      "epoch": 1.540880503144654,
      "grad_norm": 1.1044200658798218,
      "learning_rate": 3.8508909853249476e-05,
      "loss": 0.3869,
      "num_input_tokens_seen": 1920616,
      "step": 2940
    },
    {
      "epoch": 1.5435010482180294,
      "grad_norm": 1.3494473695755005,
      "learning_rate": 3.857442348008386e-05,
      "loss": 0.4372,
      "num_input_tokens_seen": 1923208,
      "step": 2945
    },
    {
      "epoch": 1.5461215932914047,
      "grad_norm": 0.8983946442604065,
      "learning_rate": 3.863993710691824e-05,
      "loss": 0.3871,
      "num_input_tokens_seen": 1926952,
      "step": 2950
    },
    {
      "epoch": 1.54874213836478,
      "grad_norm": 1.4385360479354858,
      "learning_rate": 3.8705450733752625e-05,
      "loss": 0.4544,
      "num_input_tokens_seen": 1930696,
      "step": 2955
    },
    {
      "epoch": 1.551362683438155,
      "grad_norm": 0.8728206753730774,
      "learning_rate": 3.8770964360587e-05,
      "loss": 0.4583,
      "num_input_tokens_seen": 1933864,
      "step": 2960
    },
    {
      "epoch": 1.5539832285115303,
      "grad_norm": 1.3538697957992554,
      "learning_rate": 3.883647798742139e-05,
      "loss": 0.4866,
      "num_input_tokens_seen": 1938440,
      "step": 2965
    },
    {
      "epoch": 1.5566037735849056,
      "grad_norm": 0.7315838932991028,
      "learning_rate": 3.890199161425577e-05,
      "loss": 0.5511,
      "num_input_tokens_seen": 1942184,
      "step": 2970
    },
    {
      "epoch": 1.559224318658281,
      "grad_norm": 1.027138590812683,
      "learning_rate": 3.8967505241090144e-05,
      "loss": 0.4188,
      "num_input_tokens_seen": 1945928,
      "step": 2975
    },
    {
      "epoch": 1.5618448637316562,
      "grad_norm": 1.4428842067718506,
      "learning_rate": 3.9033018867924534e-05,
      "loss": 0.5282,
      "num_input_tokens_seen": 1949416,
      "step": 2980
    },
    {
      "epoch": 1.5644654088050314,
      "grad_norm": 0.7228180766105652,
      "learning_rate": 3.909853249475891e-05,
      "loss": 0.4164,
      "num_input_tokens_seen": 1952008,
      "step": 2985
    },
    {
      "epoch": 1.5670859538784065,
      "grad_norm": 1.349337100982666,
      "learning_rate": 3.916404612159329e-05,
      "loss": 0.4134,
      "num_input_tokens_seen": 1955656,
      "step": 2990
    },
    {
      "epoch": 1.569706498951782,
      "grad_norm": 1.4314677715301514,
      "learning_rate": 3.9229559748427676e-05,
      "loss": 0.6458,
      "num_input_tokens_seen": 1958632,
      "step": 2995
    },
    {
      "epoch": 1.5723270440251573,
      "grad_norm": 0.6153419017791748,
      "learning_rate": 3.929507337526205e-05,
      "loss": 0.4756,
      "num_input_tokens_seen": 1961256,
      "step": 3000
    },
    {
      "epoch": 1.5749475890985325,
      "grad_norm": 0.7827326655387878,
      "learning_rate": 3.9360587002096435e-05,
      "loss": 0.3964,
      "num_input_tokens_seen": 1965480,
      "step": 3005
    },
    {
      "epoch": 1.5775681341719077,
      "grad_norm": 1.2421772480010986,
      "learning_rate": 3.942610062893082e-05,
      "loss": 0.4718,
      "num_input_tokens_seen": 1968552,
      "step": 3010
    },
    {
      "epoch": 1.580188679245283,
      "grad_norm": 0.7349922060966492,
      "learning_rate": 3.94916142557652e-05,
      "loss": 0.4229,
      "num_input_tokens_seen": 1971400,
      "step": 3015
    },
    {
      "epoch": 1.5828092243186582,
      "grad_norm": 0.8203959465026855,
      "learning_rate": 3.9557127882599585e-05,
      "loss": 0.3692,
      "num_input_tokens_seen": 1974312,
      "step": 3020
    },
    {
      "epoch": 1.5854297693920336,
      "grad_norm": 0.8275688886642456,
      "learning_rate": 3.962264150943397e-05,
      "loss": 0.3586,
      "num_input_tokens_seen": 1976904,
      "step": 3025
    },
    {
      "epoch": 1.5880503144654088,
      "grad_norm": 1.6995158195495605,
      "learning_rate": 3.9688155136268344e-05,
      "loss": 0.5364,
      "num_input_tokens_seen": 1979432,
      "step": 3030
    },
    {
      "epoch": 1.590670859538784,
      "grad_norm": 1.2726598978042603,
      "learning_rate": 3.975366876310273e-05,
      "loss": 0.4837,
      "num_input_tokens_seen": 1982952,
      "step": 3035
    },
    {
      "epoch": 1.5932914046121593,
      "grad_norm": 1.8415179252624512,
      "learning_rate": 3.981918238993711e-05,
      "loss": 0.4492,
      "num_input_tokens_seen": 1986408,
      "step": 3040
    },
    {
      "epoch": 1.5959119496855347,
      "grad_norm": 1.2181613445281982,
      "learning_rate": 3.9884696016771486e-05,
      "loss": 0.421,
      "num_input_tokens_seen": 1988936,
      "step": 3045
    },
    {
      "epoch": 1.59853249475891,
      "grad_norm": 1.2834832668304443,
      "learning_rate": 3.9950209643605876e-05,
      "loss": 0.5978,
      "num_input_tokens_seen": 1992264,
      "step": 3050
    },
    {
      "epoch": 1.601153039832285,
      "grad_norm": 1.4289264678955078,
      "learning_rate": 4.001572327044025e-05,
      "loss": 0.5623,
      "num_input_tokens_seen": 1995240,
      "step": 3055
    },
    {
      "epoch": 1.6037735849056602,
      "grad_norm": 1.1506186723709106,
      "learning_rate": 4.0081236897274636e-05,
      "loss": 0.5217,
      "num_input_tokens_seen": 1998408,
      "step": 3060
    },
    {
      "epoch": 1.6063941299790356,
      "grad_norm": 1.1092169284820557,
      "learning_rate": 4.014675052410902e-05,
      "loss": 0.449,
      "num_input_tokens_seen": 2001416,
      "step": 3065
    },
    {
      "epoch": 1.609014675052411,
      "grad_norm": 0.8440787196159363,
      "learning_rate": 4.0212264150943395e-05,
      "loss": 0.4394,
      "num_input_tokens_seen": 2003752,
      "step": 3070
    },
    {
      "epoch": 1.6116352201257862,
      "grad_norm": 0.542572021484375,
      "learning_rate": 4.027777777777778e-05,
      "loss": 0.4401,
      "num_input_tokens_seen": 2006728,
      "step": 3075
    },
    {
      "epoch": 1.6142557651991614,
      "grad_norm": 1.136190414428711,
      "learning_rate": 4.034329140461216e-05,
      "loss": 0.4383,
      "num_input_tokens_seen": 2009928,
      "step": 3080
    },
    {
      "epoch": 1.6168763102725365,
      "grad_norm": 0.573619544506073,
      "learning_rate": 4.0408805031446544e-05,
      "loss": 0.3229,
      "num_input_tokens_seen": 2012648,
      "step": 3085
    },
    {
      "epoch": 1.619496855345912,
      "grad_norm": 0.5764094591140747,
      "learning_rate": 4.047431865828093e-05,
      "loss": 0.3858,
      "num_input_tokens_seen": 2016936,
      "step": 3090
    },
    {
      "epoch": 1.6221174004192873,
      "grad_norm": 1.3180034160614014,
      "learning_rate": 4.0539832285115304e-05,
      "loss": 0.3302,
      "num_input_tokens_seen": 2019720,
      "step": 3095
    },
    {
      "epoch": 1.6247379454926625,
      "grad_norm": 1.45188307762146,
      "learning_rate": 4.0605345911949687e-05,
      "loss": 0.4303,
      "num_input_tokens_seen": 2022760,
      "step": 3100
    },
    {
      "epoch": 1.6273584905660377,
      "grad_norm": 1.0400422811508179,
      "learning_rate": 4.067085953878407e-05,
      "loss": 0.4059,
      "num_input_tokens_seen": 2025544,
      "step": 3105
    },
    {
      "epoch": 1.629979035639413,
      "grad_norm": 1.0507490634918213,
      "learning_rate": 4.073637316561845e-05,
      "loss": 0.411,
      "num_input_tokens_seen": 2028872,
      "step": 3110
    },
    {
      "epoch": 1.6325995807127882,
      "grad_norm": 0.826316773891449,
      "learning_rate": 4.080188679245283e-05,
      "loss": 0.3832,
      "num_input_tokens_seen": 2031816,
      "step": 3115
    },
    {
      "epoch": 1.6352201257861636,
      "grad_norm": 1.207930088043213,
      "learning_rate": 4.086740041928722e-05,
      "loss": 0.5543,
      "num_input_tokens_seen": 2034280,
      "step": 3120
    },
    {
      "epoch": 1.6378406708595388,
      "grad_norm": 0.8593827486038208,
      "learning_rate": 4.0932914046121595e-05,
      "loss": 0.4611,
      "num_input_tokens_seen": 2038056,
      "step": 3125
    },
    {
      "epoch": 1.640461215932914,
      "grad_norm": 1.3300840854644775,
      "learning_rate": 4.099842767295597e-05,
      "loss": 0.4982,
      "num_input_tokens_seen": 2041192,
      "step": 3130
    },
    {
      "epoch": 1.6430817610062893,
      "grad_norm": 0.9433104395866394,
      "learning_rate": 4.106394129979036e-05,
      "loss": 0.4315,
      "num_input_tokens_seen": 2043720,
      "step": 3135
    },
    {
      "epoch": 1.6457023060796647,
      "grad_norm": 1.0004593133926392,
      "learning_rate": 4.112945492662474e-05,
      "loss": 0.5487,
      "num_input_tokens_seen": 2046792,
      "step": 3140
    },
    {
      "epoch": 1.64832285115304,
      "grad_norm": 0.9116191267967224,
      "learning_rate": 4.119496855345912e-05,
      "loss": 0.6025,
      "num_input_tokens_seen": 2049768,
      "step": 3145
    },
    {
      "epoch": 1.650943396226415,
      "grad_norm": 1.5254124402999878,
      "learning_rate": 4.1260482180293504e-05,
      "loss": 0.4899,
      "num_input_tokens_seen": 2052840,
      "step": 3150
    },
    {
      "epoch": 1.6535639412997902,
      "grad_norm": 0.9666997790336609,
      "learning_rate": 4.132599580712788e-05,
      "loss": 0.5088,
      "num_input_tokens_seen": 2055432,
      "step": 3155
    },
    {
      "epoch": 1.6561844863731656,
      "grad_norm": 1.2607451677322388,
      "learning_rate": 4.139150943396227e-05,
      "loss": 0.5882,
      "num_input_tokens_seen": 2058792,
      "step": 3160
    },
    {
      "epoch": 1.658805031446541,
      "grad_norm": 0.9152466654777527,
      "learning_rate": 4.1457023060796646e-05,
      "loss": 0.4532,
      "num_input_tokens_seen": 2061960,
      "step": 3165
    },
    {
      "epoch": 1.6614255765199162,
      "grad_norm": 0.9514005184173584,
      "learning_rate": 4.152253668763103e-05,
      "loss": 0.5864,
      "num_input_tokens_seen": 2064712,
      "step": 3170
    },
    {
      "epoch": 1.6640461215932913,
      "grad_norm": 0.9590349793434143,
      "learning_rate": 4.158805031446541e-05,
      "loss": 0.4027,
      "num_input_tokens_seen": 2068360,
      "step": 3175
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.9214494228363037,
      "learning_rate": 4.1653563941299795e-05,
      "loss": 0.4353,
      "num_input_tokens_seen": 2071304,
      "step": 3180
    },
    {
      "epoch": 1.669287211740042,
      "grad_norm": 1.0045995712280273,
      "learning_rate": 4.171907756813417e-05,
      "loss": 0.6798,
      "num_input_tokens_seen": 2075016,
      "step": 3185
    },
    {
      "epoch": 1.6719077568134173,
      "grad_norm": 0.684696614742279,
      "learning_rate": 4.1784591194968555e-05,
      "loss": 0.4369,
      "num_input_tokens_seen": 2079016,
      "step": 3190
    },
    {
      "epoch": 1.6745283018867925,
      "grad_norm": 0.8319454193115234,
      "learning_rate": 4.185010482180294e-05,
      "loss": 0.3474,
      "num_input_tokens_seen": 2081960,
      "step": 3195
    },
    {
      "epoch": 1.6771488469601676,
      "grad_norm": 1.1087456941604614,
      "learning_rate": 4.1915618448637314e-05,
      "loss": 0.4176,
      "num_input_tokens_seen": 2084648,
      "step": 3200
    },
    {
      "epoch": 1.679769392033543,
      "grad_norm": 1.0380762815475464,
      "learning_rate": 4.1981132075471704e-05,
      "loss": 0.4341,
      "num_input_tokens_seen": 2088840,
      "step": 3205
    },
    {
      "epoch": 1.6823899371069182,
      "grad_norm": 1.769669532775879,
      "learning_rate": 4.204664570230608e-05,
      "loss": 0.4815,
      "num_input_tokens_seen": 2091944,
      "step": 3210
    },
    {
      "epoch": 1.6850104821802936,
      "grad_norm": 0.9789177179336548,
      "learning_rate": 4.2112159329140463e-05,
      "loss": 0.51,
      "num_input_tokens_seen": 2095144,
      "step": 3215
    },
    {
      "epoch": 1.6876310272536688,
      "grad_norm": 0.9462720155715942,
      "learning_rate": 4.2177672955974846e-05,
      "loss": 0.5195,
      "num_input_tokens_seen": 2098984,
      "step": 3220
    },
    {
      "epoch": 1.690251572327044,
      "grad_norm": 0.5802416801452637,
      "learning_rate": 4.224318658280922e-05,
      "loss": 0.3393,
      "num_input_tokens_seen": 2101736,
      "step": 3225
    },
    {
      "epoch": 1.6928721174004193,
      "grad_norm": 0.8598908185958862,
      "learning_rate": 4.2308700209643606e-05,
      "loss": 0.5148,
      "num_input_tokens_seen": 2104904,
      "step": 3230
    },
    {
      "epoch": 1.6954926624737947,
      "grad_norm": 1.1730434894561768,
      "learning_rate": 4.237421383647799e-05,
      "loss": 0.5083,
      "num_input_tokens_seen": 2108392,
      "step": 3235
    },
    {
      "epoch": 1.6981132075471699,
      "grad_norm": 0.8606053590774536,
      "learning_rate": 4.243972746331237e-05,
      "loss": 0.5869,
      "num_input_tokens_seen": 2112360,
      "step": 3240
    },
    {
      "epoch": 1.700733752620545,
      "grad_norm": 1.674991250038147,
      "learning_rate": 4.2505241090146755e-05,
      "loss": 0.4928,
      "num_input_tokens_seen": 2117992,
      "step": 3245
    },
    {
      "epoch": 1.7033542976939202,
      "grad_norm": 1.1607316732406616,
      "learning_rate": 4.257075471698113e-05,
      "loss": 0.399,
      "num_input_tokens_seen": 2121096,
      "step": 3250
    },
    {
      "epoch": 1.7059748427672956,
      "grad_norm": 1.2073736190795898,
      "learning_rate": 4.2636268343815514e-05,
      "loss": 0.4604,
      "num_input_tokens_seen": 2125064,
      "step": 3255
    },
    {
      "epoch": 1.708595387840671,
      "grad_norm": 0.9735557436943054,
      "learning_rate": 4.27017819706499e-05,
      "loss": 0.3427,
      "num_input_tokens_seen": 2129032,
      "step": 3260
    },
    {
      "epoch": 1.7112159329140462,
      "grad_norm": 2.0052895545959473,
      "learning_rate": 4.276729559748428e-05,
      "loss": 0.5489,
      "num_input_tokens_seen": 2131656,
      "step": 3265
    },
    {
      "epoch": 1.7138364779874213,
      "grad_norm": 1.854122519493103,
      "learning_rate": 4.283280922431866e-05,
      "loss": 0.5561,
      "num_input_tokens_seen": 2134760,
      "step": 3270
    },
    {
      "epoch": 1.7164570230607965,
      "grad_norm": 1.4985655546188354,
      "learning_rate": 4.289832285115305e-05,
      "loss": 0.5612,
      "num_input_tokens_seen": 2137960,
      "step": 3275
    },
    {
      "epoch": 1.719077568134172,
      "grad_norm": 1.2340143918991089,
      "learning_rate": 4.296383647798742e-05,
      "loss": 0.4454,
      "num_input_tokens_seen": 2140552,
      "step": 3280
    },
    {
      "epoch": 1.7216981132075473,
      "grad_norm": 1.7480430603027344,
      "learning_rate": 4.3029350104821806e-05,
      "loss": 0.3548,
      "num_input_tokens_seen": 2143272,
      "step": 3285
    },
    {
      "epoch": 1.7243186582809225,
      "grad_norm": 0.6575835347175598,
      "learning_rate": 4.309486373165619e-05,
      "loss": 0.4926,
      "num_input_tokens_seen": 2147304,
      "step": 3290
    },
    {
      "epoch": 1.7269392033542976,
      "grad_norm": 0.868812084197998,
      "learning_rate": 4.3160377358490565e-05,
      "loss": 0.6274,
      "num_input_tokens_seen": 2151624,
      "step": 3295
    },
    {
      "epoch": 1.7295597484276728,
      "grad_norm": 1.2645814418792725,
      "learning_rate": 4.322589098532495e-05,
      "loss": 0.5252,
      "num_input_tokens_seen": 2154248,
      "step": 3300
    },
    {
      "epoch": 1.7321802935010482,
      "grad_norm": 1.267414927482605,
      "learning_rate": 4.329140461215933e-05,
      "loss": 0.5252,
      "num_input_tokens_seen": 2157000,
      "step": 3305
    },
    {
      "epoch": 1.7348008385744236,
      "grad_norm": 1.3883538246154785,
      "learning_rate": 4.335691823899371e-05,
      "loss": 0.4176,
      "num_input_tokens_seen": 2160904,
      "step": 3310
    },
    {
      "epoch": 1.7374213836477987,
      "grad_norm": 0.5792728662490845,
      "learning_rate": 4.34224318658281e-05,
      "loss": 0.5104,
      "num_input_tokens_seen": 2163720,
      "step": 3315
    },
    {
      "epoch": 1.740041928721174,
      "grad_norm": 0.8338179588317871,
      "learning_rate": 4.3487945492662474e-05,
      "loss": 0.4063,
      "num_input_tokens_seen": 2167912,
      "step": 3320
    },
    {
      "epoch": 1.7426624737945493,
      "grad_norm": 0.6946846842765808,
      "learning_rate": 4.355345911949686e-05,
      "loss": 0.4343,
      "num_input_tokens_seen": 2171592,
      "step": 3325
    },
    {
      "epoch": 1.7452830188679245,
      "grad_norm": 1.11611807346344,
      "learning_rate": 4.361897274633124e-05,
      "loss": 0.4914,
      "num_input_tokens_seen": 2173864,
      "step": 3330
    },
    {
      "epoch": 1.7479035639412999,
      "grad_norm": 1.0559475421905518,
      "learning_rate": 4.368448637316562e-05,
      "loss": 0.4031,
      "num_input_tokens_seen": 2176744,
      "step": 3335
    },
    {
      "epoch": 1.750524109014675,
      "grad_norm": 1.0013973712921143,
      "learning_rate": 4.375e-05,
      "loss": 0.6316,
      "num_input_tokens_seen": 2179528,
      "step": 3340
    },
    {
      "epoch": 1.7531446540880502,
      "grad_norm": 1.887018084526062,
      "learning_rate": 4.381551362683438e-05,
      "loss": 0.414,
      "num_input_tokens_seen": 2182856,
      "step": 3345
    },
    {
      "epoch": 1.7557651991614256,
      "grad_norm": 0.9272993206977844,
      "learning_rate": 4.3881027253668766e-05,
      "loss": 0.4201,
      "num_input_tokens_seen": 2185832,
      "step": 3350
    },
    {
      "epoch": 1.758385744234801,
      "grad_norm": 0.7918432950973511,
      "learning_rate": 4.394654088050314e-05,
      "loss": 0.4705,
      "num_input_tokens_seen": 2189352,
      "step": 3355
    },
    {
      "epoch": 1.7610062893081762,
      "grad_norm": 1.6135493516921997,
      "learning_rate": 4.401205450733753e-05,
      "loss": 0.4361,
      "num_input_tokens_seen": 2192264,
      "step": 3360
    },
    {
      "epoch": 1.7636268343815513,
      "grad_norm": 0.7338452339172363,
      "learning_rate": 4.407756813417191e-05,
      "loss": 0.4424,
      "num_input_tokens_seen": 2195560,
      "step": 3365
    },
    {
      "epoch": 1.7662473794549265,
      "grad_norm": 0.6852299571037292,
      "learning_rate": 4.414308176100629e-05,
      "loss": 0.3939,
      "num_input_tokens_seen": 2199080,
      "step": 3370
    },
    {
      "epoch": 1.7688679245283019,
      "grad_norm": 0.8697541356086731,
      "learning_rate": 4.4208595387840674e-05,
      "loss": 0.3718,
      "num_input_tokens_seen": 2202408,
      "step": 3375
    },
    {
      "epoch": 1.7714884696016773,
      "grad_norm": 0.9632282257080078,
      "learning_rate": 4.427410901467505e-05,
      "loss": 0.4827,
      "num_input_tokens_seen": 2206056,
      "step": 3380
    },
    {
      "epoch": 1.7741090146750524,
      "grad_norm": 1.4939157962799072,
      "learning_rate": 4.433962264150944e-05,
      "loss": 0.3999,
      "num_input_tokens_seen": 2208456,
      "step": 3385
    },
    {
      "epoch": 1.7767295597484276,
      "grad_norm": 0.6387580037117004,
      "learning_rate": 4.440513626834382e-05,
      "loss": 0.3563,
      "num_input_tokens_seen": 2211848,
      "step": 3390
    },
    {
      "epoch": 1.7793501048218028,
      "grad_norm": 0.802864670753479,
      "learning_rate": 4.44706498951782e-05,
      "loss": 0.385,
      "num_input_tokens_seen": 2215112,
      "step": 3395
    },
    {
      "epoch": 1.7819706498951782,
      "grad_norm": 0.8572812080383301,
      "learning_rate": 4.453616352201258e-05,
      "loss": 0.5019,
      "num_input_tokens_seen": 2219528,
      "step": 3400
    },
    {
      "epoch": 1.7845911949685536,
      "grad_norm": 0.8951612114906311,
      "learning_rate": 4.460167714884696e-05,
      "loss": 0.5198,
      "num_input_tokens_seen": 2223624,
      "step": 3405
    },
    {
      "epoch": 1.7872117400419287,
      "grad_norm": 0.841438353061676,
      "learning_rate": 4.466719077568134e-05,
      "loss": 0.47,
      "num_input_tokens_seen": 2227464,
      "step": 3410
    },
    {
      "epoch": 1.789832285115304,
      "grad_norm": 0.463076651096344,
      "learning_rate": 4.4732704402515725e-05,
      "loss": 0.4104,
      "num_input_tokens_seen": 2231240,
      "step": 3415
    },
    {
      "epoch": 1.7924528301886793,
      "grad_norm": 1.1127301454544067,
      "learning_rate": 4.479821802935011e-05,
      "loss": 0.5675,
      "num_input_tokens_seen": 2234600,
      "step": 3420
    },
    {
      "epoch": 1.7950733752620545,
      "grad_norm": 0.7443017363548279,
      "learning_rate": 4.4863731656184485e-05,
      "loss": 0.4788,
      "num_input_tokens_seen": 2238056,
      "step": 3425
    },
    {
      "epoch": 1.7976939203354299,
      "grad_norm": 0.8162218332290649,
      "learning_rate": 4.4929245283018875e-05,
      "loss": 0.4026,
      "num_input_tokens_seen": 2241640,
      "step": 3430
    },
    {
      "epoch": 1.800314465408805,
      "grad_norm": 0.8135399222373962,
      "learning_rate": 4.499475890985325e-05,
      "loss": 0.5026,
      "num_input_tokens_seen": 2245832,
      "step": 3435
    },
    {
      "epoch": 1.8029350104821802,
      "grad_norm": 1.1893302202224731,
      "learning_rate": 4.5060272536687634e-05,
      "loss": 0.6857,
      "num_input_tokens_seen": 2248552,
      "step": 3440
    },
    {
      "epoch": 1.8055555555555556,
      "grad_norm": 1.3282021284103394,
      "learning_rate": 4.512578616352202e-05,
      "loss": 0.528,
      "num_input_tokens_seen": 2251432,
      "step": 3445
    },
    {
      "epoch": 1.808176100628931,
      "grad_norm": 1.6716575622558594,
      "learning_rate": 4.519129979035639e-05,
      "loss": 0.5477,
      "num_input_tokens_seen": 2254408,
      "step": 3450
    },
    {
      "epoch": 1.8107966457023061,
      "grad_norm": 0.6996035575866699,
      "learning_rate": 4.525681341719078e-05,
      "loss": 0.6428,
      "num_input_tokens_seen": 2257032,
      "step": 3455
    },
    {
      "epoch": 1.8134171907756813,
      "grad_norm": 1.7313289642333984,
      "learning_rate": 4.532232704402516e-05,
      "loss": 0.4038,
      "num_input_tokens_seen": 2259560,
      "step": 3460
    },
    {
      "epoch": 1.8160377358490565,
      "grad_norm": 0.8738412857055664,
      "learning_rate": 4.5387840670859536e-05,
      "loss": 0.5037,
      "num_input_tokens_seen": 2263240,
      "step": 3465
    },
    {
      "epoch": 1.8186582809224319,
      "grad_norm": 1.1142024993896484,
      "learning_rate": 4.5453354297693926e-05,
      "loss": 0.4423,
      "num_input_tokens_seen": 2267144,
      "step": 3470
    },
    {
      "epoch": 1.8212788259958073,
      "grad_norm": 1.5690100193023682,
      "learning_rate": 4.55188679245283e-05,
      "loss": 0.5345,
      "num_input_tokens_seen": 2270856,
      "step": 3475
    },
    {
      "epoch": 1.8238993710691824,
      "grad_norm": 1.0953623056411743,
      "learning_rate": 4.5584381551362685e-05,
      "loss": 0.5461,
      "num_input_tokens_seen": 2273672,
      "step": 3480
    },
    {
      "epoch": 1.8265199161425576,
      "grad_norm": 0.896149218082428,
      "learning_rate": 4.564989517819707e-05,
      "loss": 0.4563,
      "num_input_tokens_seen": 2276520,
      "step": 3485
    },
    {
      "epoch": 1.8291404612159328,
      "grad_norm": 0.8463414907455444,
      "learning_rate": 4.571540880503145e-05,
      "loss": 0.3901,
      "num_input_tokens_seen": 2279368,
      "step": 3490
    },
    {
      "epoch": 1.8317610062893082,
      "grad_norm": 0.7656681537628174,
      "learning_rate": 4.578092243186583e-05,
      "loss": 0.6794,
      "num_input_tokens_seen": 2281992,
      "step": 3495
    },
    {
      "epoch": 1.8343815513626835,
      "grad_norm": 0.6303684711456299,
      "learning_rate": 4.584643605870021e-05,
      "loss": 0.4243,
      "num_input_tokens_seen": 2285576,
      "step": 3500
    },
    {
      "epoch": 1.8370020964360587,
      "grad_norm": 1.5597957372665405,
      "learning_rate": 4.5911949685534594e-05,
      "loss": 0.4141,
      "num_input_tokens_seen": 2288072,
      "step": 3505
    },
    {
      "epoch": 1.8396226415094339,
      "grad_norm": 0.8286660313606262,
      "learning_rate": 4.5977463312368977e-05,
      "loss": 0.3789,
      "num_input_tokens_seen": 2291336,
      "step": 3510
    },
    {
      "epoch": 1.8422431865828093,
      "grad_norm": 1.0928454399108887,
      "learning_rate": 4.604297693920336e-05,
      "loss": 0.5142,
      "num_input_tokens_seen": 2294152,
      "step": 3515
    },
    {
      "epoch": 1.8448637316561844,
      "grad_norm": 0.7828997373580933,
      "learning_rate": 4.6108490566037736e-05,
      "loss": 0.3872,
      "num_input_tokens_seen": 2296872,
      "step": 3520
    },
    {
      "epoch": 1.8474842767295598,
      "grad_norm": 0.8442279100418091,
      "learning_rate": 4.617400419287212e-05,
      "loss": 0.455,
      "num_input_tokens_seen": 2299528,
      "step": 3525
    },
    {
      "epoch": 1.850104821802935,
      "grad_norm": 1.184253454208374,
      "learning_rate": 4.62395178197065e-05,
      "loss": 0.6129,
      "num_input_tokens_seen": 2301864,
      "step": 3530
    },
    {
      "epoch": 1.8527253668763102,
      "grad_norm": 0.5400843620300293,
      "learning_rate": 4.630503144654088e-05,
      "loss": 0.4182,
      "num_input_tokens_seen": 2306088,
      "step": 3535
    },
    {
      "epoch": 1.8553459119496856,
      "grad_norm": 0.9346745610237122,
      "learning_rate": 4.637054507337527e-05,
      "loss": 0.5193,
      "num_input_tokens_seen": 2309416,
      "step": 3540
    },
    {
      "epoch": 1.857966457023061,
      "grad_norm": 1.0127010345458984,
      "learning_rate": 4.6436058700209645e-05,
      "loss": 0.4959,
      "num_input_tokens_seen": 2312328,
      "step": 3545
    },
    {
      "epoch": 1.8605870020964361,
      "grad_norm": 1.0209459066390991,
      "learning_rate": 4.650157232704403e-05,
      "loss": 0.5157,
      "num_input_tokens_seen": 2314472,
      "step": 3550
    },
    {
      "epoch": 1.8632075471698113,
      "grad_norm": 0.9008644223213196,
      "learning_rate": 4.656708595387841e-05,
      "loss": 0.3558,
      "num_input_tokens_seen": 2318664,
      "step": 3555
    },
    {
      "epoch": 1.8658280922431865,
      "grad_norm": 1.2753163576126099,
      "learning_rate": 4.663259958071279e-05,
      "loss": 0.5393,
      "num_input_tokens_seen": 2323784,
      "step": 3560
    },
    {
      "epoch": 1.8684486373165619,
      "grad_norm": 1.1637455224990845,
      "learning_rate": 4.669811320754717e-05,
      "loss": 0.5012,
      "num_input_tokens_seen": 2326536,
      "step": 3565
    },
    {
      "epoch": 1.8710691823899372,
      "grad_norm": 0.6091561317443848,
      "learning_rate": 4.676362683438155e-05,
      "loss": 0.4613,
      "num_input_tokens_seen": 2330440,
      "step": 3570
    },
    {
      "epoch": 1.8736897274633124,
      "grad_norm": 0.20964308083057404,
      "learning_rate": 4.6829140461215936e-05,
      "loss": 0.544,
      "num_input_tokens_seen": 2335688,
      "step": 3575
    },
    {
      "epoch": 1.8763102725366876,
      "grad_norm": 0.81111741065979,
      "learning_rate": 4.689465408805032e-05,
      "loss": 0.4221,
      "num_input_tokens_seen": 2338568,
      "step": 3580
    },
    {
      "epoch": 1.8789308176100628,
      "grad_norm": 1.0544487237930298,
      "learning_rate": 4.69601677148847e-05,
      "loss": 0.4788,
      "num_input_tokens_seen": 2342120,
      "step": 3585
    },
    {
      "epoch": 1.8815513626834381,
      "grad_norm": 0.6985263228416443,
      "learning_rate": 4.702568134171908e-05,
      "loss": 0.3939,
      "num_input_tokens_seen": 2346184,
      "step": 3590
    },
    {
      "epoch": 1.8841719077568135,
      "grad_norm": 0.743550717830658,
      "learning_rate": 4.709119496855346e-05,
      "loss": 0.57,
      "num_input_tokens_seen": 2348936,
      "step": 3595
    },
    {
      "epoch": 1.8867924528301887,
      "grad_norm": 0.9323901534080505,
      "learning_rate": 4.7156708595387845e-05,
      "loss": 0.4603,
      "num_input_tokens_seen": 2351976,
      "step": 3600
    },
    {
      "epoch": 1.8894129979035639,
      "grad_norm": 0.6325273513793945,
      "learning_rate": 4.722222222222222e-05,
      "loss": 0.45,
      "num_input_tokens_seen": 2357192,
      "step": 3605
    },
    {
      "epoch": 1.892033542976939,
      "grad_norm": 1.2783602476119995,
      "learning_rate": 4.728773584905661e-05,
      "loss": 0.5986,
      "num_input_tokens_seen": 2360072,
      "step": 3610
    },
    {
      "epoch": 1.8946540880503144,
      "grad_norm": 0.9632793664932251,
      "learning_rate": 4.735324947589099e-05,
      "loss": 0.4661,
      "num_input_tokens_seen": 2363656,
      "step": 3615
    },
    {
      "epoch": 1.8972746331236898,
      "grad_norm": 1.4721882343292236,
      "learning_rate": 4.7418763102725364e-05,
      "loss": 0.5155,
      "num_input_tokens_seen": 2366216,
      "step": 3620
    },
    {
      "epoch": 1.899895178197065,
      "grad_norm": 1.364186406135559,
      "learning_rate": 4.7484276729559753e-05,
      "loss": 0.4809,
      "num_input_tokens_seen": 2369448,
      "step": 3625
    },
    {
      "epoch": 1.9025157232704402,
      "grad_norm": 0.87712562084198,
      "learning_rate": 4.754979035639413e-05,
      "loss": 0.6187,
      "num_input_tokens_seen": 2372744,
      "step": 3630
    },
    {
      "epoch": 1.9051362683438156,
      "grad_norm": 1.1222022771835327,
      "learning_rate": 4.761530398322851e-05,
      "loss": 0.4092,
      "num_input_tokens_seen": 2375080,
      "step": 3635
    },
    {
      "epoch": 1.9077568134171907,
      "grad_norm": 0.7448204159736633,
      "learning_rate": 4.7680817610062896e-05,
      "loss": 0.493,
      "num_input_tokens_seen": 2377864,
      "step": 3640
    },
    {
      "epoch": 1.9103773584905661,
      "grad_norm": 0.8202388882637024,
      "learning_rate": 4.774633123689728e-05,
      "loss": 0.4656,
      "num_input_tokens_seen": 2381224,
      "step": 3645
    },
    {
      "epoch": 1.9129979035639413,
      "grad_norm": 0.8536428809165955,
      "learning_rate": 4.7811844863731655e-05,
      "loss": 0.4934,
      "num_input_tokens_seen": 2384424,
      "step": 3650
    },
    {
      "epoch": 1.9156184486373165,
      "grad_norm": 1.019440770149231,
      "learning_rate": 4.787735849056604e-05,
      "loss": 0.428,
      "num_input_tokens_seen": 2388936,
      "step": 3655
    },
    {
      "epoch": 1.9182389937106918,
      "grad_norm": 0.9200148582458496,
      "learning_rate": 4.794287211740042e-05,
      "loss": 0.4931,
      "num_input_tokens_seen": 2392744,
      "step": 3660
    },
    {
      "epoch": 1.9208595387840672,
      "grad_norm": 1.0341315269470215,
      "learning_rate": 4.8008385744234804e-05,
      "loss": 0.5716,
      "num_input_tokens_seen": 2395112,
      "step": 3665
    },
    {
      "epoch": 1.9234800838574424,
      "grad_norm": 0.9090929627418518,
      "learning_rate": 4.807389937106919e-05,
      "loss": 0.3997,
      "num_input_tokens_seen": 2398344,
      "step": 3670
    },
    {
      "epoch": 1.9261006289308176,
      "grad_norm": 0.552129864692688,
      "learning_rate": 4.8139412997903564e-05,
      "loss": 0.4779,
      "num_input_tokens_seen": 2401704,
      "step": 3675
    },
    {
      "epoch": 1.9287211740041927,
      "grad_norm": 1.2622177600860596,
      "learning_rate": 4.8204926624737954e-05,
      "loss": 0.3633,
      "num_input_tokens_seen": 2404168,
      "step": 3680
    },
    {
      "epoch": 1.9313417190775681,
      "grad_norm": 0.813545286655426,
      "learning_rate": 4.827044025157233e-05,
      "loss": 0.3784,
      "num_input_tokens_seen": 2407688,
      "step": 3685
    },
    {
      "epoch": 1.9339622641509435,
      "grad_norm": 0.7179453372955322,
      "learning_rate": 4.8335953878406706e-05,
      "loss": 0.3858,
      "num_input_tokens_seen": 2411528,
      "step": 3690
    },
    {
      "epoch": 1.9365828092243187,
      "grad_norm": 1.247065782546997,
      "learning_rate": 4.8401467505241096e-05,
      "loss": 0.5172,
      "num_input_tokens_seen": 2414440,
      "step": 3695
    },
    {
      "epoch": 1.9392033542976939,
      "grad_norm": 0.8892459273338318,
      "learning_rate": 4.846698113207547e-05,
      "loss": 0.4015,
      "num_input_tokens_seen": 2417448,
      "step": 3700
    },
    {
      "epoch": 1.941823899371069,
      "grad_norm": 1.0732240676879883,
      "learning_rate": 4.8532494758909855e-05,
      "loss": 0.4427,
      "num_input_tokens_seen": 2420680,
      "step": 3705
    },
    {
      "epoch": 1.9444444444444444,
      "grad_norm": 1.1266803741455078,
      "learning_rate": 4.859800838574424e-05,
      "loss": 0.4584,
      "num_input_tokens_seen": 2423752,
      "step": 3710
    },
    {
      "epoch": 1.9470649895178198,
      "grad_norm": 0.9860086441040039,
      "learning_rate": 4.8663522012578615e-05,
      "loss": 0.468,
      "num_input_tokens_seen": 2426216,
      "step": 3715
    },
    {
      "epoch": 1.949685534591195,
      "grad_norm": 0.8163867592811584,
      "learning_rate": 4.8729035639413e-05,
      "loss": 0.5276,
      "num_input_tokens_seen": 2429640,
      "step": 3720
    },
    {
      "epoch": 1.9523060796645701,
      "grad_norm": 1.437093734741211,
      "learning_rate": 4.879454926624738e-05,
      "loss": 0.413,
      "num_input_tokens_seen": 2433224,
      "step": 3725
    },
    {
      "epoch": 1.9549266247379455,
      "grad_norm": 0.8240796327590942,
      "learning_rate": 4.8860062893081764e-05,
      "loss": 0.5152,
      "num_input_tokens_seen": 2437576,
      "step": 3730
    },
    {
      "epoch": 1.9575471698113207,
      "grad_norm": 1.262935996055603,
      "learning_rate": 4.892557651991615e-05,
      "loss": 0.444,
      "num_input_tokens_seen": 2439976,
      "step": 3735
    },
    {
      "epoch": 1.960167714884696,
      "grad_norm": 0.9454325437545776,
      "learning_rate": 4.899109014675053e-05,
      "loss": 0.4359,
      "num_input_tokens_seen": 2443272,
      "step": 3740
    },
    {
      "epoch": 1.9627882599580713,
      "grad_norm": 0.941551923751831,
      "learning_rate": 4.9056603773584906e-05,
      "loss": 0.5304,
      "num_input_tokens_seen": 2445832,
      "step": 3745
    },
    {
      "epoch": 1.9654088050314464,
      "grad_norm": 0.8618049025535583,
      "learning_rate": 4.912211740041929e-05,
      "loss": 0.3571,
      "num_input_tokens_seen": 2448456,
      "step": 3750
    },
    {
      "epoch": 1.9680293501048218,
      "grad_norm": 1.313106894493103,
      "learning_rate": 4.918763102725367e-05,
      "loss": 0.5296,
      "num_input_tokens_seen": 2451208,
      "step": 3755
    },
    {
      "epoch": 1.9706498951781972,
      "grad_norm": 1.0273698568344116,
      "learning_rate": 4.925314465408805e-05,
      "loss": 0.4411,
      "num_input_tokens_seen": 2454152,
      "step": 3760
    },
    {
      "epoch": 1.9732704402515724,
      "grad_norm": 2.446551561355591,
      "learning_rate": 4.931865828092244e-05,
      "loss": 0.3742,
      "num_input_tokens_seen": 2457736,
      "step": 3765
    },
    {
      "epoch": 1.9758909853249476,
      "grad_norm": 0.9038395881652832,
      "learning_rate": 4.9384171907756815e-05,
      "loss": 0.5857,
      "num_input_tokens_seen": 2461384,
      "step": 3770
    },
    {
      "epoch": 1.9785115303983227,
      "grad_norm": 1.141337275505066,
      "learning_rate": 4.944968553459119e-05,
      "loss": 0.3685,
      "num_input_tokens_seen": 2464776,
      "step": 3775
    },
    {
      "epoch": 1.9811320754716981,
      "grad_norm": 1.860746145248413,
      "learning_rate": 4.951519916142558e-05,
      "loss": 0.4416,
      "num_input_tokens_seen": 2467144,
      "step": 3780
    },
    {
      "epoch": 1.9837526205450735,
      "grad_norm": 0.7765856385231018,
      "learning_rate": 4.958071278825996e-05,
      "loss": 0.5668,
      "num_input_tokens_seen": 2469736,
      "step": 3785
    },
    {
      "epoch": 1.9863731656184487,
      "grad_norm": 0.8486813306808472,
      "learning_rate": 4.964622641509434e-05,
      "loss": 0.356,
      "num_input_tokens_seen": 2474536,
      "step": 3790
    },
    {
      "epoch": 1.9889937106918238,
      "grad_norm": 0.7607100009918213,
      "learning_rate": 4.9711740041928724e-05,
      "loss": 0.4793,
      "num_input_tokens_seen": 2477704,
      "step": 3795
    },
    {
      "epoch": 1.991614255765199,
      "grad_norm": 0.7863984704017639,
      "learning_rate": 4.977725366876311e-05,
      "loss": 0.4346,
      "num_input_tokens_seen": 2480296,
      "step": 3800
    },
    {
      "epoch": 1.9942348008385744,
      "grad_norm": 1.0903955698013306,
      "learning_rate": 4.984276729559749e-05,
      "loss": 0.5318,
      "num_input_tokens_seen": 2483368,
      "step": 3805
    },
    {
      "epoch": 1.9968553459119498,
      "grad_norm": 0.6669129133224487,
      "learning_rate": 4.9908280922431866e-05,
      "loss": 0.363,
      "num_input_tokens_seen": 2486440,
      "step": 3810
    },
    {
      "epoch": 1.999475890985325,
      "grad_norm": 0.8713302612304688,
      "learning_rate": 4.997379454926625e-05,
      "loss": 0.3335,
      "num_input_tokens_seen": 2489352,
      "step": 3815
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.4474548399448395,
      "eval_runtime": 11.931,
      "eval_samples_per_second": 71.075,
      "eval_steps_per_second": 17.769,
      "num_input_tokens_seen": 2489456,
      "step": 3816
    },
    {
      "epoch": 2.0020964360587,
      "grad_norm": 1.4656645059585571,
      "learning_rate": 4.999999905865133e-05,
      "loss": 0.5107,
      "num_input_tokens_seen": 2492112,
      "step": 3820
    },
    {
      "epoch": 2.0047169811320753,
      "grad_norm": 1.0772984027862549,
      "learning_rate": 4.999999330596526e-05,
      "loss": 0.3508,
      "num_input_tokens_seen": 2494800,
      "step": 3825
    },
    {
      "epoch": 2.007337526205451,
      "grad_norm": 1.0950380563735962,
      "learning_rate": 4.9999982323565806e-05,
      "loss": 0.4487,
      "num_input_tokens_seen": 2498896,
      "step": 3830
    },
    {
      "epoch": 2.009958071278826,
      "grad_norm": 0.8371256589889526,
      "learning_rate": 4.999996611145527e-05,
      "loss": 0.4981,
      "num_input_tokens_seen": 2502160,
      "step": 3835
    },
    {
      "epoch": 2.0125786163522013,
      "grad_norm": 0.736445963382721,
      "learning_rate": 4.9999944669637024e-05,
      "loss": 0.4523,
      "num_input_tokens_seen": 2504752,
      "step": 3840
    },
    {
      "epoch": 2.0151991614255764,
      "grad_norm": 0.9549121856689453,
      "learning_rate": 4.999991799811557e-05,
      "loss": 0.4437,
      "num_input_tokens_seen": 2507536,
      "step": 3845
    },
    {
      "epoch": 2.0178197064989516,
      "grad_norm": 0.8869922757148743,
      "learning_rate": 4.99998860968965e-05,
      "loss": 0.4477,
      "num_input_tokens_seen": 2510352,
      "step": 3850
    },
    {
      "epoch": 2.020440251572327,
      "grad_norm": 0.863684356212616,
      "learning_rate": 4.999984896598646e-05,
      "loss": 0.4021,
      "num_input_tokens_seen": 2512784,
      "step": 3855
    },
    {
      "epoch": 2.0230607966457024,
      "grad_norm": 0.8242151141166687,
      "learning_rate": 4.999980660539323e-05,
      "loss": 0.3771,
      "num_input_tokens_seen": 2516304,
      "step": 3860
    },
    {
      "epoch": 2.0256813417190775,
      "grad_norm": 1.1989774703979492,
      "learning_rate": 4.999975901512567e-05,
      "loss": 0.4822,
      "num_input_tokens_seen": 2519184,
      "step": 3865
    },
    {
      "epoch": 2.0283018867924527,
      "grad_norm": 0.7164724469184875,
      "learning_rate": 4.999970619519374e-05,
      "loss": 0.5089,
      "num_input_tokens_seen": 2522448,
      "step": 3870
    },
    {
      "epoch": 2.030922431865828,
      "grad_norm": 0.982212245464325,
      "learning_rate": 4.999964814560849e-05,
      "loss": 0.4055,
      "num_input_tokens_seen": 2525744,
      "step": 3875
    },
    {
      "epoch": 2.0335429769392035,
      "grad_norm": 0.9044710397720337,
      "learning_rate": 4.999958486638205e-05,
      "loss": 0.3917,
      "num_input_tokens_seen": 2528240,
      "step": 3880
    },
    {
      "epoch": 2.0361635220125787,
      "grad_norm": 1.0388551950454712,
      "learning_rate": 4.999951635752767e-05,
      "loss": 0.417,
      "num_input_tokens_seen": 2530928,
      "step": 3885
    },
    {
      "epoch": 2.038784067085954,
      "grad_norm": 0.9117083549499512,
      "learning_rate": 4.999944261905968e-05,
      "loss": 0.4165,
      "num_input_tokens_seen": 2533936,
      "step": 3890
    },
    {
      "epoch": 2.041404612159329,
      "grad_norm": 1.119097113609314,
      "learning_rate": 4.99993636509935e-05,
      "loss": 0.5408,
      "num_input_tokens_seen": 2536720,
      "step": 3895
    },
    {
      "epoch": 2.0440251572327046,
      "grad_norm": 0.6053121089935303,
      "learning_rate": 4.999927945334565e-05,
      "loss": 0.434,
      "num_input_tokens_seen": 2540656,
      "step": 3900
    },
    {
      "epoch": 2.04664570230608,
      "grad_norm": 0.8253674507141113,
      "learning_rate": 4.9999190026133745e-05,
      "loss": 0.4038,
      "num_input_tokens_seen": 2545040,
      "step": 3905
    },
    {
      "epoch": 2.049266247379455,
      "grad_norm": 1.7863781452178955,
      "learning_rate": 4.9999095369376506e-05,
      "loss": 0.6488,
      "num_input_tokens_seen": 2547888,
      "step": 3910
    },
    {
      "epoch": 2.05188679245283,
      "grad_norm": 1.247628927230835,
      "learning_rate": 4.99989954830937e-05,
      "loss": 0.3205,
      "num_input_tokens_seen": 2550256,
      "step": 3915
    },
    {
      "epoch": 2.0545073375262053,
      "grad_norm": 0.6571066379547119,
      "learning_rate": 4.9998890367306255e-05,
      "loss": 0.3404,
      "num_input_tokens_seen": 2553712,
      "step": 3920
    },
    {
      "epoch": 2.057127882599581,
      "grad_norm": 0.6834980845451355,
      "learning_rate": 4.999878002203614e-05,
      "loss": 0.4717,
      "num_input_tokens_seen": 2557392,
      "step": 3925
    },
    {
      "epoch": 2.059748427672956,
      "grad_norm": 0.8796876072883606,
      "learning_rate": 4.999866444730645e-05,
      "loss": 0.5957,
      "num_input_tokens_seen": 2560656,
      "step": 3930
    },
    {
      "epoch": 2.0623689727463312,
      "grad_norm": 0.6990813612937927,
      "learning_rate": 4.999854364314136e-05,
      "loss": 0.3857,
      "num_input_tokens_seen": 2563984,
      "step": 3935
    },
    {
      "epoch": 2.0649895178197064,
      "grad_norm": 1.1092183589935303,
      "learning_rate": 4.999841760956614e-05,
      "loss": 0.555,
      "num_input_tokens_seen": 2567088,
      "step": 3940
    },
    {
      "epoch": 2.0676100628930816,
      "grad_norm": 0.9268121719360352,
      "learning_rate": 4.999828634660715e-05,
      "loss": 0.393,
      "num_input_tokens_seen": 2570288,
      "step": 3945
    },
    {
      "epoch": 2.070230607966457,
      "grad_norm": 1.6172271966934204,
      "learning_rate": 4.9998149854291856e-05,
      "loss": 0.4718,
      "num_input_tokens_seen": 2573136,
      "step": 3950
    },
    {
      "epoch": 2.0728511530398324,
      "grad_norm": 0.961003303527832,
      "learning_rate": 4.99980081326488e-05,
      "loss": 0.4452,
      "num_input_tokens_seen": 2576912,
      "step": 3955
    },
    {
      "epoch": 2.0754716981132075,
      "grad_norm": 1.940852165222168,
      "learning_rate": 4.999786118170763e-05,
      "loss": 0.4432,
      "num_input_tokens_seen": 2579824,
      "step": 3960
    },
    {
      "epoch": 2.0780922431865827,
      "grad_norm": 1.0727357864379883,
      "learning_rate": 4.99977090014991e-05,
      "loss": 0.4543,
      "num_input_tokens_seen": 2583664,
      "step": 3965
    },
    {
      "epoch": 2.080712788259958,
      "grad_norm": 0.7790111899375916,
      "learning_rate": 4.9997551592055034e-05,
      "loss": 0.4481,
      "num_input_tokens_seen": 2587344,
      "step": 3970
    },
    {
      "epoch": 2.0833333333333335,
      "grad_norm": 0.849469006061554,
      "learning_rate": 4.999738895340836e-05,
      "loss": 0.4263,
      "num_input_tokens_seen": 2590800,
      "step": 3975
    },
    {
      "epoch": 2.0859538784067087,
      "grad_norm": 1.3668849468231201,
      "learning_rate": 4.999722108559311e-05,
      "loss": 0.5641,
      "num_input_tokens_seen": 2593872,
      "step": 3980
    },
    {
      "epoch": 2.088574423480084,
      "grad_norm": 0.839858889579773,
      "learning_rate": 4.9997047988644385e-05,
      "loss": 0.5347,
      "num_input_tokens_seen": 2596976,
      "step": 3985
    },
    {
      "epoch": 2.091194968553459,
      "grad_norm": 1.3720848560333252,
      "learning_rate": 4.99968696625984e-05,
      "loss": 0.4499,
      "num_input_tokens_seen": 2599824,
      "step": 3990
    },
    {
      "epoch": 2.0938155136268346,
      "grad_norm": 1.2311910390853882,
      "learning_rate": 4.999668610749246e-05,
      "loss": 0.4991,
      "num_input_tokens_seen": 2602992,
      "step": 3995
    },
    {
      "epoch": 2.0964360587002098,
      "grad_norm": 1.2093485593795776,
      "learning_rate": 4.999649732336498e-05,
      "loss": 0.3371,
      "num_input_tokens_seen": 2606448,
      "step": 4000
    },
    {
      "epoch": 2.099056603773585,
      "grad_norm": 0.8151189684867859,
      "learning_rate": 4.999630331025542e-05,
      "loss": 0.3184,
      "num_input_tokens_seen": 2610864,
      "step": 4005
    },
    {
      "epoch": 2.10167714884696,
      "grad_norm": 0.6393641829490662,
      "learning_rate": 4.9996104068204385e-05,
      "loss": 0.3774,
      "num_input_tokens_seen": 2614256,
      "step": 4010
    },
    {
      "epoch": 2.1042976939203353,
      "grad_norm": 0.9324632287025452,
      "learning_rate": 4.999589959725355e-05,
      "loss": 0.6124,
      "num_input_tokens_seen": 2617072,
      "step": 4015
    },
    {
      "epoch": 2.106918238993711,
      "grad_norm": 1.19242525100708,
      "learning_rate": 4.999568989744568e-05,
      "loss": 0.415,
      "num_input_tokens_seen": 2619792,
      "step": 4020
    },
    {
      "epoch": 2.109538784067086,
      "grad_norm": 0.8424113988876343,
      "learning_rate": 4.999547496882466e-05,
      "loss": 0.3508,
      "num_input_tokens_seen": 2622448,
      "step": 4025
    },
    {
      "epoch": 2.1121593291404612,
      "grad_norm": 1.10183584690094,
      "learning_rate": 4.999525481143543e-05,
      "loss": 0.4774,
      "num_input_tokens_seen": 2625328,
      "step": 4030
    },
    {
      "epoch": 2.1147798742138364,
      "grad_norm": 1.0388648509979248,
      "learning_rate": 4.999502942532406e-05,
      "loss": 0.4601,
      "num_input_tokens_seen": 2628336,
      "step": 4035
    },
    {
      "epoch": 2.1174004192872116,
      "grad_norm": 1.1229667663574219,
      "learning_rate": 4.999479881053769e-05,
      "loss": 0.4549,
      "num_input_tokens_seen": 2634128,
      "step": 4040
    },
    {
      "epoch": 2.120020964360587,
      "grad_norm": 0.8021851181983948,
      "learning_rate": 4.999456296712457e-05,
      "loss": 0.3678,
      "num_input_tokens_seen": 2638672,
      "step": 4045
    },
    {
      "epoch": 2.1226415094339623,
      "grad_norm": 0.512188196182251,
      "learning_rate": 4.9994321895134025e-05,
      "loss": 0.4933,
      "num_input_tokens_seen": 2642768,
      "step": 4050
    },
    {
      "epoch": 2.1252620545073375,
      "grad_norm": 1.5770092010498047,
      "learning_rate": 4.999407559461649e-05,
      "loss": 0.4182,
      "num_input_tokens_seen": 2645552,
      "step": 4055
    },
    {
      "epoch": 2.1278825995807127,
      "grad_norm": 0.8530145883560181,
      "learning_rate": 4.999382406562349e-05,
      "loss": 0.4014,
      "num_input_tokens_seen": 2647728,
      "step": 4060
    },
    {
      "epoch": 2.130503144654088,
      "grad_norm": 0.7648189067840576,
      "learning_rate": 4.999356730820765e-05,
      "loss": 0.5229,
      "num_input_tokens_seen": 2650992,
      "step": 4065
    },
    {
      "epoch": 2.1331236897274635,
      "grad_norm": 0.6658935546875,
      "learning_rate": 4.999330532242266e-05,
      "loss": 0.4591,
      "num_input_tokens_seen": 2653744,
      "step": 4070
    },
    {
      "epoch": 2.1357442348008386,
      "grad_norm": 1.165187120437622,
      "learning_rate": 4.999303810832333e-05,
      "loss": 0.2745,
      "num_input_tokens_seen": 2656336,
      "step": 4075
    },
    {
      "epoch": 2.138364779874214,
      "grad_norm": 0.9380889534950256,
      "learning_rate": 4.999276566596557e-05,
      "loss": 0.6862,
      "num_input_tokens_seen": 2659216,
      "step": 4080
    },
    {
      "epoch": 2.140985324947589,
      "grad_norm": 1.173069715499878,
      "learning_rate": 4.9992487995406364e-05,
      "loss": 0.4968,
      "num_input_tokens_seen": 2662224,
      "step": 4085
    },
    {
      "epoch": 2.1436058700209646,
      "grad_norm": 0.8560549020767212,
      "learning_rate": 4.9992205096703806e-05,
      "loss": 0.3321,
      "num_input_tokens_seen": 2665936,
      "step": 4090
    },
    {
      "epoch": 2.1462264150943398,
      "grad_norm": 1.0017282962799072,
      "learning_rate": 4.9991916969917054e-05,
      "loss": 0.3169,
      "num_input_tokens_seen": 2669648,
      "step": 4095
    },
    {
      "epoch": 2.148846960167715,
      "grad_norm": 1.4129838943481445,
      "learning_rate": 4.99916236151064e-05,
      "loss": 0.6616,
      "num_input_tokens_seen": 2673552,
      "step": 4100
    },
    {
      "epoch": 2.15146750524109,
      "grad_norm": 1.061894416809082,
      "learning_rate": 4.999132503233321e-05,
      "loss": 0.5661,
      "num_input_tokens_seen": 2675824,
      "step": 4105
    },
    {
      "epoch": 2.1540880503144653,
      "grad_norm": 2.1526591777801514,
      "learning_rate": 4.999102122165994e-05,
      "loss": 0.4874,
      "num_input_tokens_seen": 2678256,
      "step": 4110
    },
    {
      "epoch": 2.156708595387841,
      "grad_norm": 0.8565110564231873,
      "learning_rate": 4.9990712183150135e-05,
      "loss": 0.3798,
      "num_input_tokens_seen": 2682224,
      "step": 4115
    },
    {
      "epoch": 2.159329140461216,
      "grad_norm": 1.1748943328857422,
      "learning_rate": 4.9990397916868456e-05,
      "loss": 0.3121,
      "num_input_tokens_seen": 2684848,
      "step": 4120
    },
    {
      "epoch": 2.161949685534591,
      "grad_norm": 0.7252142429351807,
      "learning_rate": 4.999007842288064e-05,
      "loss": 0.4874,
      "num_input_tokens_seen": 2688816,
      "step": 4125
    },
    {
      "epoch": 2.1645702306079664,
      "grad_norm": 0.7251995205879211,
      "learning_rate": 4.9989753701253514e-05,
      "loss": 0.8701,
      "num_input_tokens_seen": 2691728,
      "step": 4130
    },
    {
      "epoch": 2.1671907756813416,
      "grad_norm": 2.7537245750427246,
      "learning_rate": 4.998942375205502e-05,
      "loss": 0.5363,
      "num_input_tokens_seen": 2694608,
      "step": 4135
    },
    {
      "epoch": 2.169811320754717,
      "grad_norm": 0.7329400777816772,
      "learning_rate": 4.998908857535416e-05,
      "loss": 0.504,
      "num_input_tokens_seen": 2697872,
      "step": 4140
    },
    {
      "epoch": 2.1724318658280923,
      "grad_norm": 0.7157714366912842,
      "learning_rate": 4.9988748171221065e-05,
      "loss": 0.3899,
      "num_input_tokens_seen": 2701040,
      "step": 4145
    },
    {
      "epoch": 2.1750524109014675,
      "grad_norm": 1.4288867712020874,
      "learning_rate": 4.998840253972695e-05,
      "loss": 0.597,
      "num_input_tokens_seen": 2703760,
      "step": 4150
    },
    {
      "epoch": 2.1776729559748427,
      "grad_norm": 0.6005299687385559,
      "learning_rate": 4.9988051680944094e-05,
      "loss": 0.4522,
      "num_input_tokens_seen": 2706480,
      "step": 4155
    },
    {
      "epoch": 2.180293501048218,
      "grad_norm": 0.8318734765052795,
      "learning_rate": 4.998769559494591e-05,
      "loss": 0.3014,
      "num_input_tokens_seen": 2709488,
      "step": 4160
    },
    {
      "epoch": 2.1829140461215935,
      "grad_norm": 0.6237682700157166,
      "learning_rate": 4.998733428180689e-05,
      "loss": 0.3957,
      "num_input_tokens_seen": 2712880,
      "step": 4165
    },
    {
      "epoch": 2.1855345911949686,
      "grad_norm": 0.9714605212211609,
      "learning_rate": 4.998696774160259e-05,
      "loss": 0.4353,
      "num_input_tokens_seen": 2716112,
      "step": 4170
    },
    {
      "epoch": 2.188155136268344,
      "grad_norm": 0.9631900787353516,
      "learning_rate": 4.9986595974409714e-05,
      "loss": 0.481,
      "num_input_tokens_seen": 2719088,
      "step": 4175
    },
    {
      "epoch": 2.190775681341719,
      "grad_norm": 1.5418260097503662,
      "learning_rate": 4.998621898030603e-05,
      "loss": 0.4316,
      "num_input_tokens_seen": 2722064,
      "step": 4180
    },
    {
      "epoch": 2.1933962264150946,
      "grad_norm": 1.2410324811935425,
      "learning_rate": 4.998583675937039e-05,
      "loss": 0.2999,
      "num_input_tokens_seen": 2725072,
      "step": 4185
    },
    {
      "epoch": 2.1960167714884697,
      "grad_norm": 0.8358880281448364,
      "learning_rate": 4.998544931168274e-05,
      "loss": 0.4002,
      "num_input_tokens_seen": 2728368,
      "step": 4190
    },
    {
      "epoch": 2.198637316561845,
      "grad_norm": 0.8042575716972351,
      "learning_rate": 4.9985056637324154e-05,
      "loss": 0.5005,
      "num_input_tokens_seen": 2731984,
      "step": 4195
    },
    {
      "epoch": 2.20125786163522,
      "grad_norm": 0.9775447249412537,
      "learning_rate": 4.998465873637677e-05,
      "loss": 0.4111,
      "num_input_tokens_seen": 2735632,
      "step": 4200
    },
    {
      "epoch": 2.2038784067085953,
      "grad_norm": 0.8370229601860046,
      "learning_rate": 4.9984255608923806e-05,
      "loss": 0.4845,
      "num_input_tokens_seen": 2738928,
      "step": 4205
    },
    {
      "epoch": 2.2064989517819704,
      "grad_norm": 0.9184961915016174,
      "learning_rate": 4.998384725504961e-05,
      "loss": 0.4446,
      "num_input_tokens_seen": 2743248,
      "step": 4210
    },
    {
      "epoch": 2.209119496855346,
      "grad_norm": 1.1140669584274292,
      "learning_rate": 4.9983433674839596e-05,
      "loss": 0.5275,
      "num_input_tokens_seen": 2745744,
      "step": 4215
    },
    {
      "epoch": 2.211740041928721,
      "grad_norm": 1.1439636945724487,
      "learning_rate": 4.998301486838028e-05,
      "loss": 0.4552,
      "num_input_tokens_seen": 2748432,
      "step": 4220
    },
    {
      "epoch": 2.2143605870020964,
      "grad_norm": 1.2783759832382202,
      "learning_rate": 4.998259083575929e-05,
      "loss": 0.5251,
      "num_input_tokens_seen": 2751280,
      "step": 4225
    },
    {
      "epoch": 2.2169811320754715,
      "grad_norm": 1.0821152925491333,
      "learning_rate": 4.99821615770653e-05,
      "loss": 0.4966,
      "num_input_tokens_seen": 2754096,
      "step": 4230
    },
    {
      "epoch": 2.219601677148847,
      "grad_norm": 1.0124520063400269,
      "learning_rate": 4.998172709238813e-05,
      "loss": 0.5646,
      "num_input_tokens_seen": 2757200,
      "step": 4235
    },
    {
      "epoch": 2.2222222222222223,
      "grad_norm": 1.404278039932251,
      "learning_rate": 4.9981287381818654e-05,
      "loss": 0.4729,
      "num_input_tokens_seen": 2760240,
      "step": 4240
    },
    {
      "epoch": 2.2248427672955975,
      "grad_norm": 1.9620577096939087,
      "learning_rate": 4.9980842445448856e-05,
      "loss": 0.4168,
      "num_input_tokens_seen": 2763248,
      "step": 4245
    },
    {
      "epoch": 2.2274633123689727,
      "grad_norm": 0.737175464630127,
      "learning_rate": 4.9980392283371815e-05,
      "loss": 0.4303,
      "num_input_tokens_seen": 2766192,
      "step": 4250
    },
    {
      "epoch": 2.230083857442348,
      "grad_norm": 0.6156315803527832,
      "learning_rate": 4.99799368956817e-05,
      "loss": 0.4411,
      "num_input_tokens_seen": 2770160,
      "step": 4255
    },
    {
      "epoch": 2.2327044025157234,
      "grad_norm": 0.7731545567512512,
      "learning_rate": 4.997947628247378e-05,
      "loss": 0.3699,
      "num_input_tokens_seen": 2773424,
      "step": 4260
    },
    {
      "epoch": 2.2353249475890986,
      "grad_norm": 1.0454668998718262,
      "learning_rate": 4.9979010443844395e-05,
      "loss": 0.4459,
      "num_input_tokens_seen": 2776240,
      "step": 4265
    },
    {
      "epoch": 2.237945492662474,
      "grad_norm": 1.0968807935714722,
      "learning_rate": 4.9978539379891e-05,
      "loss": 0.4421,
      "num_input_tokens_seen": 2779088,
      "step": 4270
    },
    {
      "epoch": 2.240566037735849,
      "grad_norm": 0.8163143396377563,
      "learning_rate": 4.997806309071214e-05,
      "loss": 0.3835,
      "num_input_tokens_seen": 2782032,
      "step": 4275
    },
    {
      "epoch": 2.243186582809224,
      "grad_norm": 0.9425626397132874,
      "learning_rate": 4.997758157640745e-05,
      "loss": 0.5835,
      "num_input_tokens_seen": 2786000,
      "step": 4280
    },
    {
      "epoch": 2.2458071278825997,
      "grad_norm": 1.1718863248825073,
      "learning_rate": 4.997709483707765e-05,
      "loss": 0.3568,
      "num_input_tokens_seen": 2789584,
      "step": 4285
    },
    {
      "epoch": 2.248427672955975,
      "grad_norm": 1.2670750617980957,
      "learning_rate": 4.997660287282456e-05,
      "loss": 0.4902,
      "num_input_tokens_seen": 2792976,
      "step": 4290
    },
    {
      "epoch": 2.25104821802935,
      "grad_norm": 1.1793638467788696,
      "learning_rate": 4.99761056837511e-05,
      "loss": 0.4697,
      "num_input_tokens_seen": 2795856,
      "step": 4295
    },
    {
      "epoch": 2.2536687631027252,
      "grad_norm": 1.2128055095672607,
      "learning_rate": 4.997560326996128e-05,
      "loss": 0.4594,
      "num_input_tokens_seen": 2798384,
      "step": 4300
    },
    {
      "epoch": 2.2562893081761004,
      "grad_norm": 0.5516351461410522,
      "learning_rate": 4.997509563156018e-05,
      "loss": 0.3397,
      "num_input_tokens_seen": 2801552,
      "step": 4305
    },
    {
      "epoch": 2.258909853249476,
      "grad_norm": 1.5651358366012573,
      "learning_rate": 4.997458276865401e-05,
      "loss": 0.5156,
      "num_input_tokens_seen": 2804816,
      "step": 4310
    },
    {
      "epoch": 2.261530398322851,
      "grad_norm": 2.2358498573303223,
      "learning_rate": 4.997406468135005e-05,
      "loss": 0.7232,
      "num_input_tokens_seen": 2807536,
      "step": 4315
    },
    {
      "epoch": 2.2641509433962264,
      "grad_norm": 0.7245827913284302,
      "learning_rate": 4.9973541369756686e-05,
      "loss": 0.4373,
      "num_input_tokens_seen": 2812080,
      "step": 4320
    },
    {
      "epoch": 2.2667714884696015,
      "grad_norm": 2.2785959243774414,
      "learning_rate": 4.9973012833983365e-05,
      "loss": 0.4585,
      "num_input_tokens_seen": 2814768,
      "step": 4325
    },
    {
      "epoch": 2.269392033542977,
      "grad_norm": 1.7039425373077393,
      "learning_rate": 4.9972479074140684e-05,
      "loss": 0.4723,
      "num_input_tokens_seen": 2817328,
      "step": 4330
    },
    {
      "epoch": 2.2720125786163523,
      "grad_norm": 0.8373345732688904,
      "learning_rate": 4.9971940090340276e-05,
      "loss": 0.5673,
      "num_input_tokens_seen": 2820624,
      "step": 4335
    },
    {
      "epoch": 2.2746331236897275,
      "grad_norm": 0.9481750726699829,
      "learning_rate": 4.997139588269489e-05,
      "loss": 0.3582,
      "num_input_tokens_seen": 2824720,
      "step": 4340
    },
    {
      "epoch": 2.2772536687631026,
      "grad_norm": 0.7933593392372131,
      "learning_rate": 4.997084645131838e-05,
      "loss": 0.5414,
      "num_input_tokens_seen": 2828560,
      "step": 4345
    },
    {
      "epoch": 2.279874213836478,
      "grad_norm": 0.9161211848258972,
      "learning_rate": 4.9970291796325666e-05,
      "loss": 0.497,
      "num_input_tokens_seen": 2832080,
      "step": 4350
    },
    {
      "epoch": 2.2824947589098534,
      "grad_norm": 1.9454519748687744,
      "learning_rate": 4.996973191783279e-05,
      "loss": 0.5238,
      "num_input_tokens_seen": 2834640,
      "step": 4355
    },
    {
      "epoch": 2.2851153039832286,
      "grad_norm": 0.9391509294509888,
      "learning_rate": 4.996916681595687e-05,
      "loss": 0.4669,
      "num_input_tokens_seen": 2837136,
      "step": 4360
    },
    {
      "epoch": 2.2877358490566038,
      "grad_norm": 1.015981912612915,
      "learning_rate": 4.996859649081611e-05,
      "loss": 0.4279,
      "num_input_tokens_seen": 2841040,
      "step": 4365
    },
    {
      "epoch": 2.290356394129979,
      "grad_norm": 1.3806508779525757,
      "learning_rate": 4.996802094252983e-05,
      "loss": 0.4924,
      "num_input_tokens_seen": 2843984,
      "step": 4370
    },
    {
      "epoch": 2.2929769392033545,
      "grad_norm": 1.1927618980407715,
      "learning_rate": 4.996744017121841e-05,
      "loss": 0.5769,
      "num_input_tokens_seen": 2846832,
      "step": 4375
    },
    {
      "epoch": 2.2955974842767297,
      "grad_norm": 0.5859869718551636,
      "learning_rate": 4.9966854177003355e-05,
      "loss": 0.3767,
      "num_input_tokens_seen": 2850512,
      "step": 4380
    },
    {
      "epoch": 2.298218029350105,
      "grad_norm": 0.7933828830718994,
      "learning_rate": 4.996626296000724e-05,
      "loss": 0.3258,
      "num_input_tokens_seen": 2854800,
      "step": 4385
    },
    {
      "epoch": 2.30083857442348,
      "grad_norm": 0.9330230951309204,
      "learning_rate": 4.996566652035374e-05,
      "loss": 0.3747,
      "num_input_tokens_seen": 2857552,
      "step": 4390
    },
    {
      "epoch": 2.3034591194968552,
      "grad_norm": 1.044500708580017,
      "learning_rate": 4.996506485816763e-05,
      "loss": 0.4383,
      "num_input_tokens_seen": 2860432,
      "step": 4395
    },
    {
      "epoch": 2.3060796645702304,
      "grad_norm": 0.7829755544662476,
      "learning_rate": 4.996445797357477e-05,
      "loss": 0.5179,
      "num_input_tokens_seen": 2863568,
      "step": 4400
    },
    {
      "epoch": 2.308700209643606,
      "grad_norm": 0.9705772995948792,
      "learning_rate": 4.996384586670211e-05,
      "loss": 0.3751,
      "num_input_tokens_seen": 2865968,
      "step": 4405
    },
    {
      "epoch": 2.311320754716981,
      "grad_norm": 0.9790675640106201,
      "learning_rate": 4.99632285376777e-05,
      "loss": 0.5877,
      "num_input_tokens_seen": 2868848,
      "step": 4410
    },
    {
      "epoch": 2.3139412997903563,
      "grad_norm": 0.9194088578224182,
      "learning_rate": 4.9962605986630674e-05,
      "loss": 0.3939,
      "num_input_tokens_seen": 2871792,
      "step": 4415
    },
    {
      "epoch": 2.3165618448637315,
      "grad_norm": 0.5604981184005737,
      "learning_rate": 4.996197821369126e-05,
      "loss": 0.4491,
      "num_input_tokens_seen": 2874832,
      "step": 4420
    },
    {
      "epoch": 2.319182389937107,
      "grad_norm": 1.489734172821045,
      "learning_rate": 4.9961345218990796e-05,
      "loss": 0.4429,
      "num_input_tokens_seen": 2877648,
      "step": 4425
    },
    {
      "epoch": 2.3218029350104823,
      "grad_norm": 0.979749858379364,
      "learning_rate": 4.996070700266168e-05,
      "loss": 0.4481,
      "num_input_tokens_seen": 2880016,
      "step": 4430
    },
    {
      "epoch": 2.3244234800838575,
      "grad_norm": 1.138347864151001,
      "learning_rate": 4.9960063564837424e-05,
      "loss": 0.566,
      "num_input_tokens_seen": 2883184,
      "step": 4435
    },
    {
      "epoch": 2.3270440251572326,
      "grad_norm": 0.9958848357200623,
      "learning_rate": 4.995941490565262e-05,
      "loss": 0.3714,
      "num_input_tokens_seen": 2887408,
      "step": 4440
    },
    {
      "epoch": 2.329664570230608,
      "grad_norm": 0.6549350023269653,
      "learning_rate": 4.995876102524299e-05,
      "loss": 0.274,
      "num_input_tokens_seen": 2892912,
      "step": 4445
    },
    {
      "epoch": 2.3322851153039834,
      "grad_norm": 0.9244667291641235,
      "learning_rate": 4.995810192374529e-05,
      "loss": 0.4812,
      "num_input_tokens_seen": 2895504,
      "step": 4450
    },
    {
      "epoch": 2.3349056603773586,
      "grad_norm": 1.0004820823669434,
      "learning_rate": 4.99574376012974e-05,
      "loss": 0.537,
      "num_input_tokens_seen": 2900208,
      "step": 4455
    },
    {
      "epoch": 2.3375262054507338,
      "grad_norm": 0.9930649399757385,
      "learning_rate": 4.995676805803829e-05,
      "loss": 0.3442,
      "num_input_tokens_seen": 2903568,
      "step": 4460
    },
    {
      "epoch": 2.340146750524109,
      "grad_norm": 0.9735274314880371,
      "learning_rate": 4.995609329410804e-05,
      "loss": 0.5498,
      "num_input_tokens_seen": 2906864,
      "step": 4465
    },
    {
      "epoch": 2.342767295597484,
      "grad_norm": 1.1804614067077637,
      "learning_rate": 4.9955413309647776e-05,
      "loss": 0.3877,
      "num_input_tokens_seen": 2910480,
      "step": 4470
    },
    {
      "epoch": 2.3453878406708597,
      "grad_norm": 1.055669903755188,
      "learning_rate": 4.9954728104799756e-05,
      "loss": 0.4083,
      "num_input_tokens_seen": 2913552,
      "step": 4475
    },
    {
      "epoch": 2.348008385744235,
      "grad_norm": 0.9249239563941956,
      "learning_rate": 4.995403767970731e-05,
      "loss": 0.5075,
      "num_input_tokens_seen": 2916368,
      "step": 4480
    },
    {
      "epoch": 2.35062893081761,
      "grad_norm": 0.5661365389823914,
      "learning_rate": 4.995334203451488e-05,
      "loss": 0.4631,
      "num_input_tokens_seen": 2918960,
      "step": 4485
    },
    {
      "epoch": 2.353249475890985,
      "grad_norm": 0.7261754870414734,
      "learning_rate": 4.995264116936798e-05,
      "loss": 0.4633,
      "num_input_tokens_seen": 2922768,
      "step": 4490
    },
    {
      "epoch": 2.3558700209643604,
      "grad_norm": 0.8291407227516174,
      "learning_rate": 4.995193508441322e-05,
      "loss": 0.4728,
      "num_input_tokens_seen": 2925840,
      "step": 4495
    },
    {
      "epoch": 2.358490566037736,
      "grad_norm": 0.7861765623092651,
      "learning_rate": 4.99512237797983e-05,
      "loss": 0.3731,
      "num_input_tokens_seen": 2929008,
      "step": 4500
    },
    {
      "epoch": 2.361111111111111,
      "grad_norm": 1.200788974761963,
      "learning_rate": 4.995050725567203e-05,
      "loss": 0.4488,
      "num_input_tokens_seen": 2931632,
      "step": 4505
    },
    {
      "epoch": 2.3637316561844863,
      "grad_norm": 1.063108205795288,
      "learning_rate": 4.994978551218429e-05,
      "loss": 0.4958,
      "num_input_tokens_seen": 2935824,
      "step": 4510
    },
    {
      "epoch": 2.3663522012578615,
      "grad_norm": 1.0391241312026978,
      "learning_rate": 4.9949058549486066e-05,
      "loss": 0.3303,
      "num_input_tokens_seen": 2938832,
      "step": 4515
    },
    {
      "epoch": 2.368972746331237,
      "grad_norm": 0.855276346206665,
      "learning_rate": 4.994832636772943e-05,
      "loss": 0.4098,
      "num_input_tokens_seen": 2942096,
      "step": 4520
    },
    {
      "epoch": 2.3715932914046123,
      "grad_norm": 1.1944711208343506,
      "learning_rate": 4.994758896706754e-05,
      "loss": 0.4392,
      "num_input_tokens_seen": 2945968,
      "step": 4525
    },
    {
      "epoch": 2.3742138364779874,
      "grad_norm": 0.6256291270256042,
      "learning_rate": 4.994684634765465e-05,
      "loss": 0.4971,
      "num_input_tokens_seen": 2949456,
      "step": 4530
    },
    {
      "epoch": 2.3768343815513626,
      "grad_norm": 1.0037697553634644,
      "learning_rate": 4.9946098509646125e-05,
      "loss": 0.4504,
      "num_input_tokens_seen": 2952592,
      "step": 4535
    },
    {
      "epoch": 2.379454926624738,
      "grad_norm": 0.5848961472511292,
      "learning_rate": 4.994534545319838e-05,
      "loss": 0.3113,
      "num_input_tokens_seen": 2955760,
      "step": 4540
    },
    {
      "epoch": 2.3820754716981134,
      "grad_norm": 0.865924596786499,
      "learning_rate": 4.994458717846896e-05,
      "loss": 0.2765,
      "num_input_tokens_seen": 2958768,
      "step": 4545
    },
    {
      "epoch": 2.3846960167714886,
      "grad_norm": 1.6815069913864136,
      "learning_rate": 4.9943823685616485e-05,
      "loss": 0.3596,
      "num_input_tokens_seen": 2961552,
      "step": 4550
    },
    {
      "epoch": 2.3873165618448637,
      "grad_norm": 0.7139689922332764,
      "learning_rate": 4.994305497480067e-05,
      "loss": 0.3855,
      "num_input_tokens_seen": 2964464,
      "step": 4555
    },
    {
      "epoch": 2.389937106918239,
      "grad_norm": 0.9641523957252502,
      "learning_rate": 4.994228104618233e-05,
      "loss": 0.4767,
      "num_input_tokens_seen": 2966832,
      "step": 4560
    },
    {
      "epoch": 2.392557651991614,
      "grad_norm": 1.0758674144744873,
      "learning_rate": 4.994150189992334e-05,
      "loss": 0.4166,
      "num_input_tokens_seen": 2969968,
      "step": 4565
    },
    {
      "epoch": 2.3951781970649897,
      "grad_norm": 1.0770525932312012,
      "learning_rate": 4.99407175361867e-05,
      "loss": 0.3973,
      "num_input_tokens_seen": 2973040,
      "step": 4570
    },
    {
      "epoch": 2.397798742138365,
      "grad_norm": 0.8648043870925903,
      "learning_rate": 4.99399279551365e-05,
      "loss": 0.4734,
      "num_input_tokens_seen": 2977520,
      "step": 4575
    },
    {
      "epoch": 2.40041928721174,
      "grad_norm": 0.8953893780708313,
      "learning_rate": 4.9939133156937886e-05,
      "loss": 0.3482,
      "num_input_tokens_seen": 2980880,
      "step": 4580
    },
    {
      "epoch": 2.403039832285115,
      "grad_norm": 0.9735058546066284,
      "learning_rate": 4.993833314175716e-05,
      "loss": 0.5588,
      "num_input_tokens_seen": 2983632,
      "step": 4585
    },
    {
      "epoch": 2.4056603773584904,
      "grad_norm": 1.1186206340789795,
      "learning_rate": 4.993752790976163e-05,
      "loss": 0.429,
      "num_input_tokens_seen": 2986256,
      "step": 4590
    },
    {
      "epoch": 2.408280922431866,
      "grad_norm": 0.8397535681724548,
      "learning_rate": 4.993671746111977e-05,
      "loss": 0.3987,
      "num_input_tokens_seen": 2991056,
      "step": 4595
    },
    {
      "epoch": 2.410901467505241,
      "grad_norm": 0.9885100722312927,
      "learning_rate": 4.993590179600111e-05,
      "loss": 0.3373,
      "num_input_tokens_seen": 2994320,
      "step": 4600
    },
    {
      "epoch": 2.4135220125786163,
      "grad_norm": 1.729630708694458,
      "learning_rate": 4.9935080914576284e-05,
      "loss": 0.3817,
      "num_input_tokens_seen": 2997104,
      "step": 4605
    },
    {
      "epoch": 2.4161425576519915,
      "grad_norm": 0.9556085467338562,
      "learning_rate": 4.9934254817016995e-05,
      "loss": 0.4524,
      "num_input_tokens_seen": 3000112,
      "step": 4610
    },
    {
      "epoch": 2.418763102725367,
      "grad_norm": 0.6744325757026672,
      "learning_rate": 4.993342350349607e-05,
      "loss": 0.4074,
      "num_input_tokens_seen": 3003568,
      "step": 4615
    },
    {
      "epoch": 2.4213836477987423,
      "grad_norm": 0.7257193326950073,
      "learning_rate": 4.9932586974187404e-05,
      "loss": 0.3944,
      "num_input_tokens_seen": 3007696,
      "step": 4620
    },
    {
      "epoch": 2.4240041928721174,
      "grad_norm": 1.002777099609375,
      "learning_rate": 4.9931745229265986e-05,
      "loss": 0.4933,
      "num_input_tokens_seen": 3011472,
      "step": 4625
    },
    {
      "epoch": 2.4266247379454926,
      "grad_norm": 1.2076959609985352,
      "learning_rate": 4.99308982689079e-05,
      "loss": 0.4064,
      "num_input_tokens_seen": 3014320,
      "step": 4630
    },
    {
      "epoch": 2.4292452830188678,
      "grad_norm": 1.2529457807540894,
      "learning_rate": 4.993004609329033e-05,
      "loss": 0.5509,
      "num_input_tokens_seen": 3018224,
      "step": 4635
    },
    {
      "epoch": 2.431865828092243,
      "grad_norm": 1.005791425704956,
      "learning_rate": 4.992918870259153e-05,
      "loss": 0.4754,
      "num_input_tokens_seen": 3021072,
      "step": 4640
    },
    {
      "epoch": 2.4344863731656186,
      "grad_norm": 0.6750101447105408,
      "learning_rate": 4.992832609699086e-05,
      "loss": 0.3303,
      "num_input_tokens_seen": 3023600,
      "step": 4645
    },
    {
      "epoch": 2.4371069182389937,
      "grad_norm": 1.0362941026687622,
      "learning_rate": 4.992745827666877e-05,
      "loss": 0.4479,
      "num_input_tokens_seen": 3026768,
      "step": 4650
    },
    {
      "epoch": 2.439727463312369,
      "grad_norm": 0.902818500995636,
      "learning_rate": 4.99265852418068e-05,
      "loss": 0.4577,
      "num_input_tokens_seen": 3033424,
      "step": 4655
    },
    {
      "epoch": 2.442348008385744,
      "grad_norm": 1.021113634109497,
      "learning_rate": 4.992570699258757e-05,
      "loss": 0.4324,
      "num_input_tokens_seen": 3035952,
      "step": 4660
    },
    {
      "epoch": 2.4449685534591197,
      "grad_norm": 1.3096625804901123,
      "learning_rate": 4.99248235291948e-05,
      "loss": 0.4367,
      "num_input_tokens_seen": 3039216,
      "step": 4665
    },
    {
      "epoch": 2.447589098532495,
      "grad_norm": 1.1413288116455078,
      "learning_rate": 4.9923934851813314e-05,
      "loss": 0.6547,
      "num_input_tokens_seen": 3042224,
      "step": 4670
    },
    {
      "epoch": 2.45020964360587,
      "grad_norm": 0.7873940467834473,
      "learning_rate": 4.9923040960628996e-05,
      "loss": 0.3808,
      "num_input_tokens_seen": 3044752,
      "step": 4675
    },
    {
      "epoch": 2.452830188679245,
      "grad_norm": 1.0250298976898193,
      "learning_rate": 4.9922141855828856e-05,
      "loss": 0.5268,
      "num_input_tokens_seen": 3048336,
      "step": 4680
    },
    {
      "epoch": 2.4554507337526204,
      "grad_norm": 1.2443618774414062,
      "learning_rate": 4.992123753760096e-05,
      "loss": 0.4071,
      "num_input_tokens_seen": 3050864,
      "step": 4685
    },
    {
      "epoch": 2.458071278825996,
      "grad_norm": 1.6683346033096313,
      "learning_rate": 4.9920328006134484e-05,
      "loss": 0.4296,
      "num_input_tokens_seen": 3053712,
      "step": 4690
    },
    {
      "epoch": 2.460691823899371,
      "grad_norm": 0.9617761373519897,
      "learning_rate": 4.991941326161971e-05,
      "loss": 0.6334,
      "num_input_tokens_seen": 3056944,
      "step": 4695
    },
    {
      "epoch": 2.4633123689727463,
      "grad_norm": 0.7137762308120728,
      "learning_rate": 4.991849330424796e-05,
      "loss": 0.4892,
      "num_input_tokens_seen": 3060176,
      "step": 4700
    },
    {
      "epoch": 2.4659329140461215,
      "grad_norm": 1.0713907480239868,
      "learning_rate": 4.9917568134211703e-05,
      "loss": 0.6407,
      "num_input_tokens_seen": 3062352,
      "step": 4705
    },
    {
      "epoch": 2.468553459119497,
      "grad_norm": 0.9384681582450867,
      "learning_rate": 4.991663775170447e-05,
      "loss": 0.4386,
      "num_input_tokens_seen": 3064848,
      "step": 4710
    },
    {
      "epoch": 2.4711740041928723,
      "grad_norm": 0.7501437664031982,
      "learning_rate": 4.991570215692088e-05,
      "loss": 0.505,
      "num_input_tokens_seen": 3068368,
      "step": 4715
    },
    {
      "epoch": 2.4737945492662474,
      "grad_norm": 1.283486008644104,
      "learning_rate": 4.991476135005665e-05,
      "loss": 0.5222,
      "num_input_tokens_seen": 3071536,
      "step": 4720
    },
    {
      "epoch": 2.4764150943396226,
      "grad_norm": 1.2074484825134277,
      "learning_rate": 4.991381533130859e-05,
      "loss": 0.3407,
      "num_input_tokens_seen": 3074224,
      "step": 4725
    },
    {
      "epoch": 2.4790356394129978,
      "grad_norm": 1.0529160499572754,
      "learning_rate": 4.99128641008746e-05,
      "loss": 0.4823,
      "num_input_tokens_seen": 3077744,
      "step": 4730
    },
    {
      "epoch": 2.481656184486373,
      "grad_norm": 1.2097177505493164,
      "learning_rate": 4.9911907658953664e-05,
      "loss": 0.4396,
      "num_input_tokens_seen": 3080656,
      "step": 4735
    },
    {
      "epoch": 2.4842767295597485,
      "grad_norm": 0.4353210926055908,
      "learning_rate": 4.991094600574584e-05,
      "loss": 0.4499,
      "num_input_tokens_seen": 3084624,
      "step": 4740
    },
    {
      "epoch": 2.4868972746331237,
      "grad_norm": 2.2328853607177734,
      "learning_rate": 4.9909979141452324e-05,
      "loss": 0.4092,
      "num_input_tokens_seen": 3087760,
      "step": 4745
    },
    {
      "epoch": 2.489517819706499,
      "grad_norm": 0.7471996545791626,
      "learning_rate": 4.9909007066275346e-05,
      "loss": 0.4767,
      "num_input_tokens_seen": 3091440,
      "step": 4750
    },
    {
      "epoch": 2.492138364779874,
      "grad_norm": 1.1769191026687622,
      "learning_rate": 4.990802978041828e-05,
      "loss": 0.4273,
      "num_input_tokens_seen": 3094192,
      "step": 4755
    },
    {
      "epoch": 2.4947589098532497,
      "grad_norm": 1.20206880569458,
      "learning_rate": 4.990704728408555e-05,
      "loss": 0.3734,
      "num_input_tokens_seen": 3097584,
      "step": 4760
    },
    {
      "epoch": 2.497379454926625,
      "grad_norm": 0.8029869198799133,
      "learning_rate": 4.990605957748268e-05,
      "loss": 0.4343,
      "num_input_tokens_seen": 3101232,
      "step": 4765
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9252493381500244,
      "learning_rate": 4.990506666081628e-05,
      "loss": 0.2973,
      "num_input_tokens_seen": 3103888,
      "step": 4770
    },
    {
      "epoch": 2.502620545073375,
      "grad_norm": 1.2343990802764893,
      "learning_rate": 4.990406853429408e-05,
      "loss": 0.3422,
      "num_input_tokens_seen": 3111408,
      "step": 4775
    },
    {
      "epoch": 2.5052410901467503,
      "grad_norm": 0.7937312722206116,
      "learning_rate": 4.990306519812485e-05,
      "loss": 0.3683,
      "num_input_tokens_seen": 3114288,
      "step": 4780
    },
    {
      "epoch": 2.507861635220126,
      "grad_norm": 1.288478136062622,
      "learning_rate": 4.99020566525185e-05,
      "loss": 0.5025,
      "num_input_tokens_seen": 3117296,
      "step": 4785
    },
    {
      "epoch": 2.510482180293501,
      "grad_norm": 1.3445608615875244,
      "learning_rate": 4.9901042897685996e-05,
      "loss": 0.3321,
      "num_input_tokens_seen": 3119792,
      "step": 4790
    },
    {
      "epoch": 2.5131027253668763,
      "grad_norm": 0.9254457950592041,
      "learning_rate": 4.99000239338394e-05,
      "loss": 0.5088,
      "num_input_tokens_seen": 3122544,
      "step": 4795
    },
    {
      "epoch": 2.5157232704402515,
      "grad_norm": 1.0418550968170166,
      "learning_rate": 4.989899976119187e-05,
      "loss": 0.3961,
      "num_input_tokens_seen": 3125648,
      "step": 4800
    },
    {
      "epoch": 2.518343815513627,
      "grad_norm": 1.1586980819702148,
      "learning_rate": 4.989797037995766e-05,
      "loss": 0.3972,
      "num_input_tokens_seen": 3128400,
      "step": 4805
    },
    {
      "epoch": 2.5209643605870022,
      "grad_norm": 1.365269422531128,
      "learning_rate": 4.989693579035209e-05,
      "loss": 0.4077,
      "num_input_tokens_seen": 3131280,
      "step": 4810
    },
    {
      "epoch": 2.5235849056603774,
      "grad_norm": 0.7335320115089417,
      "learning_rate": 4.989589599259159e-05,
      "loss": 0.4652,
      "num_input_tokens_seen": 3134288,
      "step": 4815
    },
    {
      "epoch": 2.5262054507337526,
      "grad_norm": 1.2541950941085815,
      "learning_rate": 4.9894850986893685e-05,
      "loss": 0.4439,
      "num_input_tokens_seen": 3136784,
      "step": 4820
    },
    {
      "epoch": 2.5288259958071277,
      "grad_norm": 1.2145841121673584,
      "learning_rate": 4.9893800773476965e-05,
      "loss": 0.4684,
      "num_input_tokens_seen": 3139664,
      "step": 4825
    },
    {
      "epoch": 2.531446540880503,
      "grad_norm": 0.9356163144111633,
      "learning_rate": 4.9892745352561134e-05,
      "loss": 0.6552,
      "num_input_tokens_seen": 3142896,
      "step": 4830
    },
    {
      "epoch": 2.5340670859538785,
      "grad_norm": 0.7196177244186401,
      "learning_rate": 4.9891684724366955e-05,
      "loss": 0.4881,
      "num_input_tokens_seen": 3146224,
      "step": 4835
    },
    {
      "epoch": 2.5366876310272537,
      "grad_norm": 0.9789611101150513,
      "learning_rate": 4.989061888911631e-05,
      "loss": 0.4852,
      "num_input_tokens_seen": 3149136,
      "step": 4840
    },
    {
      "epoch": 2.539308176100629,
      "grad_norm": 1.10703444480896,
      "learning_rate": 4.988954784703217e-05,
      "loss": 0.4753,
      "num_input_tokens_seen": 3151984,
      "step": 4845
    },
    {
      "epoch": 2.541928721174004,
      "grad_norm": 0.9073973298072815,
      "learning_rate": 4.9888471598338574e-05,
      "loss": 0.4851,
      "num_input_tokens_seen": 3155440,
      "step": 4850
    },
    {
      "epoch": 2.5445492662473796,
      "grad_norm": 0.4508844017982483,
      "learning_rate": 4.988739014326066e-05,
      "loss": 0.3348,
      "num_input_tokens_seen": 3159216,
      "step": 4855
    },
    {
      "epoch": 2.547169811320755,
      "grad_norm": 0.8315840363502502,
      "learning_rate": 4.9886303482024656e-05,
      "loss": 0.5455,
      "num_input_tokens_seen": 3162640,
      "step": 4860
    },
    {
      "epoch": 2.54979035639413,
      "grad_norm": 1.0924392938613892,
      "learning_rate": 4.988521161485789e-05,
      "loss": 0.392,
      "num_input_tokens_seen": 3165968,
      "step": 4865
    },
    {
      "epoch": 2.552410901467505,
      "grad_norm": 0.7823301553726196,
      "learning_rate": 4.9884114541988747e-05,
      "loss": 0.3932,
      "num_input_tokens_seen": 3168688,
      "step": 4870
    },
    {
      "epoch": 2.5550314465408803,
      "grad_norm": 0.9025758504867554,
      "learning_rate": 4.988301226364674e-05,
      "loss": 0.5136,
      "num_input_tokens_seen": 3172400,
      "step": 4875
    },
    {
      "epoch": 2.5576519916142555,
      "grad_norm": 1.1708762645721436,
      "learning_rate": 4.988190478006245e-05,
      "loss": 0.5356,
      "num_input_tokens_seen": 3175472,
      "step": 4880
    },
    {
      "epoch": 2.560272536687631,
      "grad_norm": 1.2822697162628174,
      "learning_rate": 4.988079209146754e-05,
      "loss": 0.6595,
      "num_input_tokens_seen": 3178032,
      "step": 4885
    },
    {
      "epoch": 2.5628930817610063,
      "grad_norm": 0.57208251953125,
      "learning_rate": 4.9879674198094794e-05,
      "loss": 0.3299,
      "num_input_tokens_seen": 3181744,
      "step": 4890
    },
    {
      "epoch": 2.5655136268343814,
      "grad_norm": 0.9787645936012268,
      "learning_rate": 4.987855110017804e-05,
      "loss": 0.4228,
      "num_input_tokens_seen": 3184656,
      "step": 4895
    },
    {
      "epoch": 2.568134171907757,
      "grad_norm": 0.739425539970398,
      "learning_rate": 4.9877422797952224e-05,
      "loss": 0.4141,
      "num_input_tokens_seen": 3187504,
      "step": 4900
    },
    {
      "epoch": 2.5707547169811322,
      "grad_norm": 1.3433233499526978,
      "learning_rate": 4.987628929165338e-05,
      "loss": 0.4792,
      "num_input_tokens_seen": 3190608,
      "step": 4905
    },
    {
      "epoch": 2.5733752620545074,
      "grad_norm": 0.8213831782341003,
      "learning_rate": 4.9875150581518615e-05,
      "loss": 0.4729,
      "num_input_tokens_seen": 3193584,
      "step": 4910
    },
    {
      "epoch": 2.5759958071278826,
      "grad_norm": 0.9422950744628906,
      "learning_rate": 4.987400666778614e-05,
      "loss": 0.5824,
      "num_input_tokens_seen": 3197488,
      "step": 4915
    },
    {
      "epoch": 2.5786163522012577,
      "grad_norm": 1.1146032810211182,
      "learning_rate": 4.987285755069524e-05,
      "loss": 0.5322,
      "num_input_tokens_seen": 3200592,
      "step": 4920
    },
    {
      "epoch": 2.581236897274633,
      "grad_norm": 1.086198329925537,
      "learning_rate": 4.987170323048632e-05,
      "loss": 0.4148,
      "num_input_tokens_seen": 3204208,
      "step": 4925
    },
    {
      "epoch": 2.5838574423480085,
      "grad_norm": 1.3187867403030396,
      "learning_rate": 4.987054370740083e-05,
      "loss": 0.5161,
      "num_input_tokens_seen": 3208048,
      "step": 4930
    },
    {
      "epoch": 2.5864779874213837,
      "grad_norm": 0.6446576714515686,
      "learning_rate": 4.986937898168134e-05,
      "loss": 0.3898,
      "num_input_tokens_seen": 3210576,
      "step": 4935
    },
    {
      "epoch": 2.589098532494759,
      "grad_norm": 0.5804134607315063,
      "learning_rate": 4.986820905357148e-05,
      "loss": 0.3515,
      "num_input_tokens_seen": 3213648,
      "step": 4940
    },
    {
      "epoch": 2.591719077568134,
      "grad_norm": 0.6212309002876282,
      "learning_rate": 4.9867033923315995e-05,
      "loss": 0.6213,
      "num_input_tokens_seen": 3216592,
      "step": 4945
    },
    {
      "epoch": 2.5943396226415096,
      "grad_norm": 1.1735824346542358,
      "learning_rate": 4.9865853591160724e-05,
      "loss": 0.4112,
      "num_input_tokens_seen": 3219760,
      "step": 4950
    },
    {
      "epoch": 2.596960167714885,
      "grad_norm": 1.316888689994812,
      "learning_rate": 4.9864668057352556e-05,
      "loss": 0.6612,
      "num_input_tokens_seen": 3223600,
      "step": 4955
    },
    {
      "epoch": 2.59958071278826,
      "grad_norm": 1.4506900310516357,
      "learning_rate": 4.9863477322139517e-05,
      "loss": 0.5574,
      "num_input_tokens_seen": 3226416,
      "step": 4960
    },
    {
      "epoch": 2.602201257861635,
      "grad_norm": 0.7096754312515259,
      "learning_rate": 4.986228138577067e-05,
      "loss": 0.3585,
      "num_input_tokens_seen": 3230288,
      "step": 4965
    },
    {
      "epoch": 2.6048218029350103,
      "grad_norm": 1.5886013507843018,
      "learning_rate": 4.986108024849619e-05,
      "loss": 0.4779,
      "num_input_tokens_seen": 3233104,
      "step": 4970
    },
    {
      "epoch": 2.6074423480083855,
      "grad_norm": 0.8577755689620972,
      "learning_rate": 4.985987391056737e-05,
      "loss": 0.4827,
      "num_input_tokens_seen": 3237872,
      "step": 4975
    },
    {
      "epoch": 2.610062893081761,
      "grad_norm": 0.6386369466781616,
      "learning_rate": 4.9858662372236544e-05,
      "loss": 0.4078,
      "num_input_tokens_seen": 3242064,
      "step": 4980
    },
    {
      "epoch": 2.6126834381551363,
      "grad_norm": 1.2718181610107422,
      "learning_rate": 4.985744563375714e-05,
      "loss": 0.4355,
      "num_input_tokens_seen": 3245712,
      "step": 4985
    },
    {
      "epoch": 2.6153039832285114,
      "grad_norm": 1.1082735061645508,
      "learning_rate": 4.985622369538371e-05,
      "loss": 0.4385,
      "num_input_tokens_seen": 3248560,
      "step": 4990
    },
    {
      "epoch": 2.617924528301887,
      "grad_norm": 0.8504509329795837,
      "learning_rate": 4.985499655737185e-05,
      "loss": 0.4655,
      "num_input_tokens_seen": 3251440,
      "step": 4995
    },
    {
      "epoch": 2.620545073375262,
      "grad_norm": 1.3407795429229736,
      "learning_rate": 4.985376421997827e-05,
      "loss": 0.4461,
      "num_input_tokens_seen": 3254640,
      "step": 5000
    },
    {
      "epoch": 2.6231656184486374,
      "grad_norm": 0.7997627854347229,
      "learning_rate": 4.985252668346076e-05,
      "loss": 0.3712,
      "num_input_tokens_seen": 3258320,
      "step": 5005
    },
    {
      "epoch": 2.6257861635220126,
      "grad_norm": 0.8602678179740906,
      "learning_rate": 4.9851283948078206e-05,
      "loss": 0.412,
      "num_input_tokens_seen": 3261168,
      "step": 5010
    },
    {
      "epoch": 2.6284067085953877,
      "grad_norm": 1.0512553453445435,
      "learning_rate": 4.985003601409057e-05,
      "loss": 0.5116,
      "num_input_tokens_seen": 3264816,
      "step": 5015
    },
    {
      "epoch": 2.631027253668763,
      "grad_norm": 1.0839368104934692,
      "learning_rate": 4.9848782881758906e-05,
      "loss": 0.3333,
      "num_input_tokens_seen": 3268080,
      "step": 5020
    },
    {
      "epoch": 2.6336477987421385,
      "grad_norm": 0.6600587964057922,
      "learning_rate": 4.9847524551345345e-05,
      "loss": 0.491,
      "num_input_tokens_seen": 3272528,
      "step": 5025
    },
    {
      "epoch": 2.6362683438155137,
      "grad_norm": 1.0102354288101196,
      "learning_rate": 4.984626102311313e-05,
      "loss": 0.4814,
      "num_input_tokens_seen": 3275888,
      "step": 5030
    },
    {
      "epoch": 2.638888888888889,
      "grad_norm": 1.7791262865066528,
      "learning_rate": 4.984499229732656e-05,
      "loss": 0.3763,
      "num_input_tokens_seen": 3281072,
      "step": 5035
    },
    {
      "epoch": 2.641509433962264,
      "grad_norm": 0.6830974221229553,
      "learning_rate": 4.984371837425106e-05,
      "loss": 0.445,
      "num_input_tokens_seen": 3285808,
      "step": 5040
    },
    {
      "epoch": 2.6441299790356396,
      "grad_norm": 0.6661378145217896,
      "learning_rate": 4.9842439254153106e-05,
      "loss": 0.3733,
      "num_input_tokens_seen": 3289168,
      "step": 5045
    },
    {
      "epoch": 2.646750524109015,
      "grad_norm": 1.0633200407028198,
      "learning_rate": 4.984115493730027e-05,
      "loss": 0.6042,
      "num_input_tokens_seen": 3292336,
      "step": 5050
    },
    {
      "epoch": 2.64937106918239,
      "grad_norm": 0.8059085011482239,
      "learning_rate": 4.983986542396123e-05,
      "loss": 0.4125,
      "num_input_tokens_seen": 3295024,
      "step": 5055
    },
    {
      "epoch": 2.651991614255765,
      "grad_norm": 1.8177906274795532,
      "learning_rate": 4.983857071440573e-05,
      "loss": 0.6008,
      "num_input_tokens_seen": 3297776,
      "step": 5060
    },
    {
      "epoch": 2.6546121593291403,
      "grad_norm": 0.7970789670944214,
      "learning_rate": 4.9837270808904616e-05,
      "loss": 0.3576,
      "num_input_tokens_seen": 3300560,
      "step": 5065
    },
    {
      "epoch": 2.6572327044025155,
      "grad_norm": 0.985478401184082,
      "learning_rate": 4.9835965707729807e-05,
      "loss": 0.461,
      "num_input_tokens_seen": 3304432,
      "step": 5070
    },
    {
      "epoch": 2.659853249475891,
      "grad_norm": 0.8625039458274841,
      "learning_rate": 4.983465541115431e-05,
      "loss": 0.4179,
      "num_input_tokens_seen": 3307344,
      "step": 5075
    },
    {
      "epoch": 2.6624737945492662,
      "grad_norm": 0.7454243302345276,
      "learning_rate": 4.983333991945223e-05,
      "loss": 0.453,
      "num_input_tokens_seen": 3311056,
      "step": 5080
    },
    {
      "epoch": 2.6650943396226414,
      "grad_norm": 0.7812857031822205,
      "learning_rate": 4.983201923289875e-05,
      "loss": 0.422,
      "num_input_tokens_seen": 3313712,
      "step": 5085
    },
    {
      "epoch": 2.667714884696017,
      "grad_norm": 0.5986628532409668,
      "learning_rate": 4.983069335177014e-05,
      "loss": 0.4173,
      "num_input_tokens_seen": 3317232,
      "step": 5090
    },
    {
      "epoch": 2.670335429769392,
      "grad_norm": 0.9181172251701355,
      "learning_rate": 4.982936227634378e-05,
      "loss": 0.5827,
      "num_input_tokens_seen": 3320048,
      "step": 5095
    },
    {
      "epoch": 2.6729559748427674,
      "grad_norm": 0.7135785818099976,
      "learning_rate": 4.982802600689809e-05,
      "loss": 0.3902,
      "num_input_tokens_seen": 3324720,
      "step": 5100
    },
    {
      "epoch": 2.6755765199161425,
      "grad_norm": 0.968137264251709,
      "learning_rate": 4.982668454371261e-05,
      "loss": 0.4107,
      "num_input_tokens_seen": 3327696,
      "step": 5105
    },
    {
      "epoch": 2.6781970649895177,
      "grad_norm": 0.7652660012245178,
      "learning_rate": 4.9825337887067966e-05,
      "loss": 0.3357,
      "num_input_tokens_seen": 3330608,
      "step": 5110
    },
    {
      "epoch": 2.680817610062893,
      "grad_norm": 1.4063889980316162,
      "learning_rate": 4.982398603724586e-05,
      "loss": 0.4457,
      "num_input_tokens_seen": 3333296,
      "step": 5115
    },
    {
      "epoch": 2.6834381551362685,
      "grad_norm": 1.2540115118026733,
      "learning_rate": 4.982262899452909e-05,
      "loss": 0.4086,
      "num_input_tokens_seen": 3337552,
      "step": 5120
    },
    {
      "epoch": 2.6860587002096437,
      "grad_norm": 0.6868038773536682,
      "learning_rate": 4.982126675920151e-05,
      "loss": 0.4373,
      "num_input_tokens_seen": 3341296,
      "step": 5125
    },
    {
      "epoch": 2.688679245283019,
      "grad_norm": 1.5026353597640991,
      "learning_rate": 4.98198993315481e-05,
      "loss": 0.4737,
      "num_input_tokens_seen": 3343920,
      "step": 5130
    },
    {
      "epoch": 2.691299790356394,
      "grad_norm": 1.0666499137878418,
      "learning_rate": 4.981852671185492e-05,
      "loss": 0.4174,
      "num_input_tokens_seen": 3346736,
      "step": 5135
    },
    {
      "epoch": 2.6939203354297696,
      "grad_norm": 0.8690690398216248,
      "learning_rate": 4.9817148900409093e-05,
      "loss": 0.4503,
      "num_input_tokens_seen": 3350320,
      "step": 5140
    },
    {
      "epoch": 2.6965408805031448,
      "grad_norm": 0.9242812395095825,
      "learning_rate": 4.981576589749885e-05,
      "loss": 0.3747,
      "num_input_tokens_seen": 3353552,
      "step": 5145
    },
    {
      "epoch": 2.69916142557652,
      "grad_norm": 1.2391897439956665,
      "learning_rate": 4.981437770341348e-05,
      "loss": 0.4814,
      "num_input_tokens_seen": 3357136,
      "step": 5150
    },
    {
      "epoch": 2.701781970649895,
      "grad_norm": 2.3874592781066895,
      "learning_rate": 4.9812984318443414e-05,
      "loss": 0.4274,
      "num_input_tokens_seen": 3360048,
      "step": 5155
    },
    {
      "epoch": 2.7044025157232703,
      "grad_norm": 1.023868441581726,
      "learning_rate": 4.98115857428801e-05,
      "loss": 0.3894,
      "num_input_tokens_seen": 3363280,
      "step": 5160
    },
    {
      "epoch": 2.7070230607966455,
      "grad_norm": 1.1285157203674316,
      "learning_rate": 4.9810181977016104e-05,
      "loss": 0.5784,
      "num_input_tokens_seen": 3366000,
      "step": 5165
    },
    {
      "epoch": 2.709643605870021,
      "grad_norm": 0.6571842432022095,
      "learning_rate": 4.9808773021145096e-05,
      "loss": 0.4855,
      "num_input_tokens_seen": 3369648,
      "step": 5170
    },
    {
      "epoch": 2.7122641509433962,
      "grad_norm": 1.4293038845062256,
      "learning_rate": 4.98073588755618e-05,
      "loss": 0.4188,
      "num_input_tokens_seen": 3372528,
      "step": 5175
    },
    {
      "epoch": 2.7148846960167714,
      "grad_norm": 1.1438287496566772,
      "learning_rate": 4.980593954056205e-05,
      "loss": 0.47,
      "num_input_tokens_seen": 3375728,
      "step": 5180
    },
    {
      "epoch": 2.717505241090147,
      "grad_norm": 1.494735598564148,
      "learning_rate": 4.9804515016442754e-05,
      "loss": 0.3931,
      "num_input_tokens_seen": 3378864,
      "step": 5185
    },
    {
      "epoch": 2.720125786163522,
      "grad_norm": 0.9476129412651062,
      "learning_rate": 4.9803085303501896e-05,
      "loss": 0.3976,
      "num_input_tokens_seen": 3381936,
      "step": 5190
    },
    {
      "epoch": 2.7227463312368974,
      "grad_norm": 0.8415091037750244,
      "learning_rate": 4.9801650402038555e-05,
      "loss": 0.4896,
      "num_input_tokens_seen": 3384464,
      "step": 5195
    },
    {
      "epoch": 2.7253668763102725,
      "grad_norm": 0.9627417922019958,
      "learning_rate": 4.980021031235291e-05,
      "loss": 0.5356,
      "num_input_tokens_seen": 3387312,
      "step": 5200
    },
    {
      "epoch": 2.7279874213836477,
      "grad_norm": 0.9052270650863647,
      "learning_rate": 4.9798765034746206e-05,
      "loss": 0.4052,
      "num_input_tokens_seen": 3391376,
      "step": 5205
    },
    {
      "epoch": 2.730607966457023,
      "grad_norm": 0.8063744306564331,
      "learning_rate": 4.979731456952077e-05,
      "loss": 0.4543,
      "num_input_tokens_seen": 3394352,
      "step": 5210
    },
    {
      "epoch": 2.7332285115303985,
      "grad_norm": 0.6582739353179932,
      "learning_rate": 4.9795858916980034e-05,
      "loss": 0.4886,
      "num_input_tokens_seen": 3397456,
      "step": 5215
    },
    {
      "epoch": 2.7358490566037736,
      "grad_norm": 1.0979852676391602,
      "learning_rate": 4.9794398077428495e-05,
      "loss": 0.507,
      "num_input_tokens_seen": 3403792,
      "step": 5220
    },
    {
      "epoch": 2.738469601677149,
      "grad_norm": 1.5845332145690918,
      "learning_rate": 4.979293205117175e-05,
      "loss": 0.4387,
      "num_input_tokens_seen": 3407184,
      "step": 5225
    },
    {
      "epoch": 2.741090146750524,
      "grad_norm": 0.9269558787345886,
      "learning_rate": 4.979146083851647e-05,
      "loss": 0.4762,
      "num_input_tokens_seen": 3410256,
      "step": 5230
    },
    {
      "epoch": 2.7437106918238996,
      "grad_norm": 1.412714958190918,
      "learning_rate": 4.978998443977042e-05,
      "loss": 0.5183,
      "num_input_tokens_seen": 3413168,
      "step": 5235
    },
    {
      "epoch": 2.7463312368972748,
      "grad_norm": 1.1447649002075195,
      "learning_rate": 4.9788502855242436e-05,
      "loss": 0.5985,
      "num_input_tokens_seen": 3415728,
      "step": 5240
    },
    {
      "epoch": 2.74895178197065,
      "grad_norm": 1.7008122205734253,
      "learning_rate": 4.978701608524247e-05,
      "loss": 0.3673,
      "num_input_tokens_seen": 3419568,
      "step": 5245
    },
    {
      "epoch": 2.751572327044025,
      "grad_norm": 0.7436463832855225,
      "learning_rate": 4.978552413008152e-05,
      "loss": 0.3355,
      "num_input_tokens_seen": 3422736,
      "step": 5250
    },
    {
      "epoch": 2.7541928721174003,
      "grad_norm": 0.9495092630386353,
      "learning_rate": 4.9784026990071687e-05,
      "loss": 0.3927,
      "num_input_tokens_seen": 3427472,
      "step": 5255
    },
    {
      "epoch": 2.7568134171907754,
      "grad_norm": 0.7953334450721741,
      "learning_rate": 4.978252466552615e-05,
      "loss": 0.3525,
      "num_input_tokens_seen": 3430256,
      "step": 5260
    },
    {
      "epoch": 2.759433962264151,
      "grad_norm": 0.9564311504364014,
      "learning_rate": 4.9781017156759205e-05,
      "loss": 0.5528,
      "num_input_tokens_seen": 3433328,
      "step": 5265
    },
    {
      "epoch": 2.762054507337526,
      "grad_norm": 0.9244142770767212,
      "learning_rate": 4.977950446408618e-05,
      "loss": 0.4015,
      "num_input_tokens_seen": 3437456,
      "step": 5270
    },
    {
      "epoch": 2.7646750524109014,
      "grad_norm": 0.8607593774795532,
      "learning_rate": 4.9777986587823523e-05,
      "loss": 0.3408,
      "num_input_tokens_seen": 3440240,
      "step": 5275
    },
    {
      "epoch": 2.767295597484277,
      "grad_norm": 1.0431565046310425,
      "learning_rate": 4.977646352828875e-05,
      "loss": 0.415,
      "num_input_tokens_seen": 3443824,
      "step": 5280
    },
    {
      "epoch": 2.769916142557652,
      "grad_norm": 0.6645194888114929,
      "learning_rate": 4.9774935285800476e-05,
      "loss": 0.4875,
      "num_input_tokens_seen": 3447184,
      "step": 5285
    },
    {
      "epoch": 2.7725366876310273,
      "grad_norm": 1.36495041847229,
      "learning_rate": 4.977340186067838e-05,
      "loss": 0.3967,
      "num_input_tokens_seen": 3450320,
      "step": 5290
    },
    {
      "epoch": 2.7751572327044025,
      "grad_norm": 0.5820550322532654,
      "learning_rate": 4.977186325324325e-05,
      "loss": 0.5463,
      "num_input_tokens_seen": 3454640,
      "step": 5295
    },
    {
      "epoch": 2.7777777777777777,
      "grad_norm": 1.1585031747817993,
      "learning_rate": 4.977031946381694e-05,
      "loss": 0.4289,
      "num_input_tokens_seen": 3458000,
      "step": 5300
    },
    {
      "epoch": 2.780398322851153,
      "grad_norm": 0.9700713753700256,
      "learning_rate": 4.9768770492722394e-05,
      "loss": 0.5385,
      "num_input_tokens_seen": 3461296,
      "step": 5305
    },
    {
      "epoch": 2.7830188679245285,
      "grad_norm": 0.9362121224403381,
      "learning_rate": 4.976721634028363e-05,
      "loss": 0.3874,
      "num_input_tokens_seen": 3464336,
      "step": 5310
    },
    {
      "epoch": 2.7856394129979036,
      "grad_norm": 0.9527879953384399,
      "learning_rate": 4.976565700682578e-05,
      "loss": 0.539,
      "num_input_tokens_seen": 3467184,
      "step": 5315
    },
    {
      "epoch": 2.788259958071279,
      "grad_norm": 1.1295264959335327,
      "learning_rate": 4.976409249267502e-05,
      "loss": 0.5055,
      "num_input_tokens_seen": 3470320,
      "step": 5320
    },
    {
      "epoch": 2.790880503144654,
      "grad_norm": 1.2875906229019165,
      "learning_rate": 4.976252279815863e-05,
      "loss": 0.36,
      "num_input_tokens_seen": 3474128,
      "step": 5325
    },
    {
      "epoch": 2.7935010482180296,
      "grad_norm": 1.4348409175872803,
      "learning_rate": 4.976094792360498e-05,
      "loss": 0.577,
      "num_input_tokens_seen": 3476336,
      "step": 5330
    },
    {
      "epoch": 2.7961215932914047,
      "grad_norm": 0.9171625375747681,
      "learning_rate": 4.975936786934351e-05,
      "loss": 0.482,
      "num_input_tokens_seen": 3480016,
      "step": 5335
    },
    {
      "epoch": 2.79874213836478,
      "grad_norm": 2.2007579803466797,
      "learning_rate": 4.975778263570475e-05,
      "loss": 0.4047,
      "num_input_tokens_seen": 3482608,
      "step": 5340
    },
    {
      "epoch": 2.801362683438155,
      "grad_norm": 0.9061279296875,
      "learning_rate": 4.975619222302032e-05,
      "loss": 0.6517,
      "num_input_tokens_seen": 3485200,
      "step": 5345
    },
    {
      "epoch": 2.8039832285115303,
      "grad_norm": 1.0758183002471924,
      "learning_rate": 4.975459663162291e-05,
      "loss": 0.3309,
      "num_input_tokens_seen": 3488336,
      "step": 5350
    },
    {
      "epoch": 2.8066037735849054,
      "grad_norm": 0.8546314239501953,
      "learning_rate": 4.9752995861846296e-05,
      "loss": 0.5006,
      "num_input_tokens_seen": 3492080,
      "step": 5355
    },
    {
      "epoch": 2.809224318658281,
      "grad_norm": 0.7514743804931641,
      "learning_rate": 4.975138991402535e-05,
      "loss": 0.3933,
      "num_input_tokens_seen": 3495952,
      "step": 5360
    },
    {
      "epoch": 2.811844863731656,
      "grad_norm": 0.7470415234565735,
      "learning_rate": 4.974977878849601e-05,
      "loss": 0.3931,
      "num_input_tokens_seen": 3498224,
      "step": 5365
    },
    {
      "epoch": 2.8144654088050314,
      "grad_norm": 1.650909185409546,
      "learning_rate": 4.9748162485595303e-05,
      "loss": 0.515,
      "num_input_tokens_seen": 3501872,
      "step": 5370
    },
    {
      "epoch": 2.8170859538784065,
      "grad_norm": 0.7705399394035339,
      "learning_rate": 4.974654100566135e-05,
      "loss": 0.3861,
      "num_input_tokens_seen": 3505008,
      "step": 5375
    },
    {
      "epoch": 2.819706498951782,
      "grad_norm": 0.7937591671943665,
      "learning_rate": 4.974491434903335e-05,
      "loss": 0.3451,
      "num_input_tokens_seen": 3508016,
      "step": 5380
    },
    {
      "epoch": 2.8223270440251573,
      "grad_norm": 0.9606471061706543,
      "learning_rate": 4.9743282516051556e-05,
      "loss": 0.4796,
      "num_input_tokens_seen": 3512912,
      "step": 5385
    },
    {
      "epoch": 2.8249475890985325,
      "grad_norm": 0.9241462349891663,
      "learning_rate": 4.974164550705736e-05,
      "loss": 0.4252,
      "num_input_tokens_seen": 3516112,
      "step": 5390
    },
    {
      "epoch": 2.8275681341719077,
      "grad_norm": 1.0494232177734375,
      "learning_rate": 4.974000332239319e-05,
      "loss": 0.5489,
      "num_input_tokens_seen": 3518800,
      "step": 5395
    },
    {
      "epoch": 2.830188679245283,
      "grad_norm": 1.0086110830307007,
      "learning_rate": 4.973835596240257e-05,
      "loss": 0.3501,
      "num_input_tokens_seen": 3522608,
      "step": 5400
    },
    {
      "epoch": 2.832809224318658,
      "grad_norm": 1.4169795513153076,
      "learning_rate": 4.9736703427430114e-05,
      "loss": 0.463,
      "num_input_tokens_seen": 3526512,
      "step": 5405
    },
    {
      "epoch": 2.8354297693920336,
      "grad_norm": 0.9603434205055237,
      "learning_rate": 4.9735045717821514e-05,
      "loss": 0.4033,
      "num_input_tokens_seen": 3529456,
      "step": 5410
    },
    {
      "epoch": 2.838050314465409,
      "grad_norm": 0.9611970782279968,
      "learning_rate": 4.973338283392354e-05,
      "loss": 0.4206,
      "num_input_tokens_seen": 3532464,
      "step": 5415
    },
    {
      "epoch": 2.840670859538784,
      "grad_norm": 1.395412802696228,
      "learning_rate": 4.9731714776084046e-05,
      "loss": 0.4314,
      "num_input_tokens_seen": 3535824,
      "step": 5420
    },
    {
      "epoch": 2.8432914046121596,
      "grad_norm": 0.6288367509841919,
      "learning_rate": 4.973004154465198e-05,
      "loss": 0.4669,
      "num_input_tokens_seen": 3539696,
      "step": 5425
    },
    {
      "epoch": 2.8459119496855347,
      "grad_norm": 0.8665655255317688,
      "learning_rate": 4.972836313997736e-05,
      "loss": 0.4684,
      "num_input_tokens_seen": 3543408,
      "step": 5430
    },
    {
      "epoch": 2.84853249475891,
      "grad_norm": 2.017082452774048,
      "learning_rate": 4.9726679562411285e-05,
      "loss": 0.3697,
      "num_input_tokens_seen": 3546864,
      "step": 5435
    },
    {
      "epoch": 2.851153039832285,
      "grad_norm": 0.6396265029907227,
      "learning_rate": 4.972499081230595e-05,
      "loss": 0.4213,
      "num_input_tokens_seen": 3549904,
      "step": 5440
    },
    {
      "epoch": 2.8537735849056602,
      "grad_norm": 0.8320884704589844,
      "learning_rate": 4.9723296890014605e-05,
      "loss": 0.4101,
      "num_input_tokens_seen": 3553072,
      "step": 5445
    },
    {
      "epoch": 2.8563941299790354,
      "grad_norm": 0.6239486932754517,
      "learning_rate": 4.9721597795891605e-05,
      "loss": 0.3192,
      "num_input_tokens_seen": 3556784,
      "step": 5450
    },
    {
      "epoch": 2.859014675052411,
      "grad_norm": 0.4893045425415039,
      "learning_rate": 4.9719893530292396e-05,
      "loss": 0.4702,
      "num_input_tokens_seen": 3561232,
      "step": 5455
    },
    {
      "epoch": 2.861635220125786,
      "grad_norm": 0.6547728180885315,
      "learning_rate": 4.9718184093573475e-05,
      "loss": 0.4839,
      "num_input_tokens_seen": 3564048,
      "step": 5460
    },
    {
      "epoch": 2.8642557651991614,
      "grad_norm": 0.8729127645492554,
      "learning_rate": 4.9716469486092435e-05,
      "loss": 0.3598,
      "num_input_tokens_seen": 3567664,
      "step": 5465
    },
    {
      "epoch": 2.8668763102725365,
      "grad_norm": 0.7432447075843811,
      "learning_rate": 4.971474970820797e-05,
      "loss": 0.4443,
      "num_input_tokens_seen": 3571344,
      "step": 5470
    },
    {
      "epoch": 2.869496855345912,
      "grad_norm": 0.6880806684494019,
      "learning_rate": 4.971302476027982e-05,
      "loss": 0.4081,
      "num_input_tokens_seen": 3574128,
      "step": 5475
    },
    {
      "epoch": 2.8721174004192873,
      "grad_norm": 0.5838350057601929,
      "learning_rate": 4.971129464266884e-05,
      "loss": 0.4189,
      "num_input_tokens_seen": 3577552,
      "step": 5480
    },
    {
      "epoch": 2.8747379454926625,
      "grad_norm": 0.9052523374557495,
      "learning_rate": 4.970955935573693e-05,
      "loss": 0.6149,
      "num_input_tokens_seen": 3581040,
      "step": 5485
    },
    {
      "epoch": 2.8773584905660377,
      "grad_norm": 1.1471489667892456,
      "learning_rate": 4.9707818899847106e-05,
      "loss": 0.4032,
      "num_input_tokens_seen": 3583760,
      "step": 5490
    },
    {
      "epoch": 2.879979035639413,
      "grad_norm": 0.9404237866401672,
      "learning_rate": 4.970607327536346e-05,
      "loss": 0.4575,
      "num_input_tokens_seen": 3587632,
      "step": 5495
    },
    {
      "epoch": 2.882599580712788,
      "grad_norm": 1.260815143585205,
      "learning_rate": 4.970432248265114e-05,
      "loss": 0.4628,
      "num_input_tokens_seen": 3590960,
      "step": 5500
    },
    {
      "epoch": 2.8852201257861636,
      "grad_norm": 0.7424999475479126,
      "learning_rate": 4.97025665220764e-05,
      "loss": 0.3573,
      "num_input_tokens_seen": 3594672,
      "step": 5505
    },
    {
      "epoch": 2.8878406708595388,
      "grad_norm": 0.7291228771209717,
      "learning_rate": 4.970080539400656e-05,
      "loss": 0.4054,
      "num_input_tokens_seen": 3597872,
      "step": 5510
    },
    {
      "epoch": 2.890461215932914,
      "grad_norm": 0.7429918646812439,
      "learning_rate": 4.969903909881004e-05,
      "loss": 0.3349,
      "num_input_tokens_seen": 3600528,
      "step": 5515
    },
    {
      "epoch": 2.8930817610062896,
      "grad_norm": 0.8882690668106079,
      "learning_rate": 4.9697267636856317e-05,
      "loss": 0.4779,
      "num_input_tokens_seen": 3603344,
      "step": 5520
    },
    {
      "epoch": 2.8957023060796647,
      "grad_norm": 0.9663309454917908,
      "learning_rate": 4.969549100851597e-05,
      "loss": 0.4231,
      "num_input_tokens_seen": 3606448,
      "step": 5525
    },
    {
      "epoch": 2.89832285115304,
      "grad_norm": 0.9424374103546143,
      "learning_rate": 4.9693709214160645e-05,
      "loss": 0.6146,
      "num_input_tokens_seen": 3609776,
      "step": 5530
    },
    {
      "epoch": 2.900943396226415,
      "grad_norm": 1.1766417026519775,
      "learning_rate": 4.969192225416307e-05,
      "loss": 0.4306,
      "num_input_tokens_seen": 3612080,
      "step": 5535
    },
    {
      "epoch": 2.9035639412997902,
      "grad_norm": 0.958577036857605,
      "learning_rate": 4.969013012889706e-05,
      "loss": 0.4844,
      "num_input_tokens_seen": 3615504,
      "step": 5540
    },
    {
      "epoch": 2.9061844863731654,
      "grad_norm": 1.734978199005127,
      "learning_rate": 4.9688332838737504e-05,
      "loss": 0.6429,
      "num_input_tokens_seen": 3618096,
      "step": 5545
    },
    {
      "epoch": 2.908805031446541,
      "grad_norm": 1.069920301437378,
      "learning_rate": 4.9686530384060386e-05,
      "loss": 0.4,
      "num_input_tokens_seen": 3621296,
      "step": 5550
    },
    {
      "epoch": 2.911425576519916,
      "grad_norm": 0.9809463024139404,
      "learning_rate": 4.968472276524275e-05,
      "loss": 0.4287,
      "num_input_tokens_seen": 3624784,
      "step": 5555
    },
    {
      "epoch": 2.9140461215932913,
      "grad_norm": 0.7285333275794983,
      "learning_rate": 4.9682909982662725e-05,
      "loss": 0.5197,
      "num_input_tokens_seen": 3628336,
      "step": 5560
    },
    {
      "epoch": 2.9166666666666665,
      "grad_norm": 1.1332446336746216,
      "learning_rate": 4.968109203669953e-05,
      "loss": 0.5159,
      "num_input_tokens_seen": 3631280,
      "step": 5565
    },
    {
      "epoch": 2.919287211740042,
      "grad_norm": 0.5295753479003906,
      "learning_rate": 4.967926892773346e-05,
      "loss": 0.6069,
      "num_input_tokens_seen": 3634448,
      "step": 5570
    },
    {
      "epoch": 2.9219077568134173,
      "grad_norm": 0.8691968321800232,
      "learning_rate": 4.967744065614588e-05,
      "loss": 0.4296,
      "num_input_tokens_seen": 3637072,
      "step": 5575
    },
    {
      "epoch": 2.9245283018867925,
      "grad_norm": 0.707309365272522,
      "learning_rate": 4.9675607222319255e-05,
      "loss": 0.5988,
      "num_input_tokens_seen": 3640880,
      "step": 5580
    },
    {
      "epoch": 2.9271488469601676,
      "grad_norm": 1.185329556465149,
      "learning_rate": 4.967376862663711e-05,
      "loss": 0.4646,
      "num_input_tokens_seen": 3643312,
      "step": 5585
    },
    {
      "epoch": 2.929769392033543,
      "grad_norm": 0.8162965774536133,
      "learning_rate": 4.9671924869484064e-05,
      "loss": 0.4615,
      "num_input_tokens_seen": 3647152,
      "step": 5590
    },
    {
      "epoch": 2.932389937106918,
      "grad_norm": 1.4033641815185547,
      "learning_rate": 4.967007595124581e-05,
      "loss": 0.4353,
      "num_input_tokens_seen": 3649904,
      "step": 5595
    },
    {
      "epoch": 2.9350104821802936,
      "grad_norm": 0.7443110942840576,
      "learning_rate": 4.966822187230911e-05,
      "loss": 0.3546,
      "num_input_tokens_seen": 3653680,
      "step": 5600
    },
    {
      "epoch": 2.9376310272536688,
      "grad_norm": 0.9073502421379089,
      "learning_rate": 4.966636263306183e-05,
      "loss": 0.518,
      "num_input_tokens_seen": 3657328,
      "step": 5605
    },
    {
      "epoch": 2.940251572327044,
      "grad_norm": 0.8228192925453186,
      "learning_rate": 4.9664498233892895e-05,
      "loss": 0.4539,
      "num_input_tokens_seen": 3661008,
      "step": 5610
    },
    {
      "epoch": 2.9428721174004195,
      "grad_norm": 1.0688167810440063,
      "learning_rate": 4.966262867519232e-05,
      "loss": 0.4255,
      "num_input_tokens_seen": 3664144,
      "step": 5615
    },
    {
      "epoch": 2.9454926624737947,
      "grad_norm": 1.3177976608276367,
      "learning_rate": 4.966075395735119e-05,
      "loss": 0.3712,
      "num_input_tokens_seen": 3666800,
      "step": 5620
    },
    {
      "epoch": 2.94811320754717,
      "grad_norm": 0.9915875792503357,
      "learning_rate": 4.965887408076167e-05,
      "loss": 0.4087,
      "num_input_tokens_seen": 3669936,
      "step": 5625
    },
    {
      "epoch": 2.950733752620545,
      "grad_norm": 0.6767895817756653,
      "learning_rate": 4.965698904581703e-05,
      "loss": 0.3773,
      "num_input_tokens_seen": 3673584,
      "step": 5630
    },
    {
      "epoch": 2.95335429769392,
      "grad_norm": 1.028439998626709,
      "learning_rate": 4.965509885291157e-05,
      "loss": 0.4033,
      "num_input_tokens_seen": 3676816,
      "step": 5635
    },
    {
      "epoch": 2.9559748427672954,
      "grad_norm": 1.339932918548584,
      "learning_rate": 4.9653203502440716e-05,
      "loss": 0.5415,
      "num_input_tokens_seen": 3679792,
      "step": 5640
    },
    {
      "epoch": 2.958595387840671,
      "grad_norm": 0.712233304977417,
      "learning_rate": 4.965130299480095e-05,
      "loss": 0.4415,
      "num_input_tokens_seen": 3682704,
      "step": 5645
    },
    {
      "epoch": 2.961215932914046,
      "grad_norm": 1.4298301935195923,
      "learning_rate": 4.9649397330389834e-05,
      "loss": 0.4074,
      "num_input_tokens_seen": 3685968,
      "step": 5650
    },
    {
      "epoch": 2.9638364779874213,
      "grad_norm": 0.8330389261245728,
      "learning_rate": 4.964748650960601e-05,
      "loss": 0.4525,
      "num_input_tokens_seen": 3688528,
      "step": 5655
    },
    {
      "epoch": 2.9664570230607965,
      "grad_norm": 1.0927540063858032,
      "learning_rate": 4.96455705328492e-05,
      "loss": 0.4549,
      "num_input_tokens_seen": 3692464,
      "step": 5660
    },
    {
      "epoch": 2.969077568134172,
      "grad_norm": 0.6584956645965576,
      "learning_rate": 4.964364940052021e-05,
      "loss": 0.3615,
      "num_input_tokens_seen": 3695920,
      "step": 5665
    },
    {
      "epoch": 2.9716981132075473,
      "grad_norm": 0.9563530683517456,
      "learning_rate": 4.964172311302091e-05,
      "loss": 0.5778,
      "num_input_tokens_seen": 3699600,
      "step": 5670
    },
    {
      "epoch": 2.9743186582809225,
      "grad_norm": 1.130722165107727,
      "learning_rate": 4.9639791670754265e-05,
      "loss": 0.4848,
      "num_input_tokens_seen": 3703024,
      "step": 5675
    },
    {
      "epoch": 2.9769392033542976,
      "grad_norm": 0.5082243084907532,
      "learning_rate": 4.9637855074124305e-05,
      "loss": 0.3386,
      "num_input_tokens_seen": 3706544,
      "step": 5680
    },
    {
      "epoch": 2.979559748427673,
      "grad_norm": 0.8272328972816467,
      "learning_rate": 4.963591332353615e-05,
      "loss": 0.5491,
      "num_input_tokens_seen": 3709584,
      "step": 5685
    },
    {
      "epoch": 2.982180293501048,
      "grad_norm": 0.8553819060325623,
      "learning_rate": 4.9633966419396e-05,
      "loss": 0.4739,
      "num_input_tokens_seen": 3713616,
      "step": 5690
    },
    {
      "epoch": 2.9848008385744236,
      "grad_norm": 0.6822053790092468,
      "learning_rate": 4.96320143621111e-05,
      "loss": 0.4633,
      "num_input_tokens_seen": 3716656,
      "step": 5695
    },
    {
      "epoch": 2.9874213836477987,
      "grad_norm": 0.7984822988510132,
      "learning_rate": 4.963005715208981e-05,
      "loss": 0.3566,
      "num_input_tokens_seen": 3719504,
      "step": 5700
    },
    {
      "epoch": 2.990041928721174,
      "grad_norm": 1.1876837015151978,
      "learning_rate": 4.962809478974157e-05,
      "loss": 0.4098,
      "num_input_tokens_seen": 3722672,
      "step": 5705
    },
    {
      "epoch": 2.9926624737945495,
      "grad_norm": 0.614684522151947,
      "learning_rate": 4.962612727547686e-05,
      "loss": 0.4436,
      "num_input_tokens_seen": 3725104,
      "step": 5710
    },
    {
      "epoch": 2.9952830188679247,
      "grad_norm": 0.7445694804191589,
      "learning_rate": 4.962415460970728e-05,
      "loss": 0.5716,
      "num_input_tokens_seen": 3728272,
      "step": 5715
    },
    {
      "epoch": 2.9979035639413,
      "grad_norm": 1.19176185131073,
      "learning_rate": 4.962217679284548e-05,
      "loss": 0.3801,
      "num_input_tokens_seen": 3731344,
      "step": 5720
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.4436005651950836,
      "eval_runtime": 11.9666,
      "eval_samples_per_second": 70.864,
      "eval_steps_per_second": 17.716,
      "num_input_tokens_seen": 3733736,
      "step": 5724
    },
    {
      "epoch": 3.000524109014675,
      "grad_norm": 1.1276592016220093,
      "learning_rate": 4.962019382530521e-05,
      "loss": 0.4118,
      "num_input_tokens_seen": 3734280,
      "step": 5725
    },
    {
      "epoch": 3.00314465408805,
      "grad_norm": 0.8087549209594727,
      "learning_rate": 4.9618205707501256e-05,
      "loss": 0.3595,
      "num_input_tokens_seen": 3736776,
      "step": 5730
    },
    {
      "epoch": 3.0057651991614254,
      "grad_norm": 0.8381653428077698,
      "learning_rate": 4.961621243984954e-05,
      "loss": 0.4056,
      "num_input_tokens_seen": 3739496,
      "step": 5735
    },
    {
      "epoch": 3.008385744234801,
      "grad_norm": 0.9387534260749817,
      "learning_rate": 4.9614214022767014e-05,
      "loss": 0.331,
      "num_input_tokens_seen": 3741992,
      "step": 5740
    },
    {
      "epoch": 3.011006289308176,
      "grad_norm": 1.0709123611450195,
      "learning_rate": 4.9612210456671735e-05,
      "loss": 0.4764,
      "num_input_tokens_seen": 3744488,
      "step": 5745
    },
    {
      "epoch": 3.0136268343815513,
      "grad_norm": 0.812597393989563,
      "learning_rate": 4.961020174198281e-05,
      "loss": 0.4542,
      "num_input_tokens_seen": 3747688,
      "step": 5750
    },
    {
      "epoch": 3.0162473794549265,
      "grad_norm": 0.5824412107467651,
      "learning_rate": 4.9608187879120454e-05,
      "loss": 0.3836,
      "num_input_tokens_seen": 3753064,
      "step": 5755
    },
    {
      "epoch": 3.018867924528302,
      "grad_norm": 0.7600882053375244,
      "learning_rate": 4.9606168868505936e-05,
      "loss": 0.4727,
      "num_input_tokens_seen": 3756136,
      "step": 5760
    },
    {
      "epoch": 3.0214884696016773,
      "grad_norm": 0.7037802338600159,
      "learning_rate": 4.9604144710561616e-05,
      "loss": 0.4733,
      "num_input_tokens_seen": 3758888,
      "step": 5765
    },
    {
      "epoch": 3.0241090146750524,
      "grad_norm": 0.9471564292907715,
      "learning_rate": 4.960211540571092e-05,
      "loss": 0.3228,
      "num_input_tokens_seen": 3761832,
      "step": 5770
    },
    {
      "epoch": 3.0267295597484276,
      "grad_norm": 0.8098539710044861,
      "learning_rate": 4.960008095437836e-05,
      "loss": 0.3951,
      "num_input_tokens_seen": 3765032,
      "step": 5775
    },
    {
      "epoch": 3.029350104821803,
      "grad_norm": 1.0982370376586914,
      "learning_rate": 4.959804135698951e-05,
      "loss": 0.4978,
      "num_input_tokens_seen": 3767944,
      "step": 5780
    },
    {
      "epoch": 3.0319706498951784,
      "grad_norm": 1.1651369333267212,
      "learning_rate": 4.959599661397104e-05,
      "loss": 0.395,
      "num_input_tokens_seen": 3771112,
      "step": 5785
    },
    {
      "epoch": 3.0345911949685536,
      "grad_norm": 1.1010791063308716,
      "learning_rate": 4.959394672575068e-05,
      "loss": 0.4989,
      "num_input_tokens_seen": 3773960,
      "step": 5790
    },
    {
      "epoch": 3.0372117400419287,
      "grad_norm": 0.8245370388031006,
      "learning_rate": 4.959189169275725e-05,
      "loss": 0.4659,
      "num_input_tokens_seen": 3777192,
      "step": 5795
    },
    {
      "epoch": 3.039832285115304,
      "grad_norm": 1.167127251625061,
      "learning_rate": 4.958983151542065e-05,
      "loss": 0.4532,
      "num_input_tokens_seen": 3780488,
      "step": 5800
    },
    {
      "epoch": 3.042452830188679,
      "grad_norm": 0.691291868686676,
      "learning_rate": 4.958776619417182e-05,
      "loss": 0.4124,
      "num_input_tokens_seen": 3784328,
      "step": 5805
    },
    {
      "epoch": 3.0450733752620547,
      "grad_norm": 1.2179802656173706,
      "learning_rate": 4.9585695729442814e-05,
      "loss": 0.406,
      "num_input_tokens_seen": 3786696,
      "step": 5810
    },
    {
      "epoch": 3.04769392033543,
      "grad_norm": 0.7308257222175598,
      "learning_rate": 4.9583620121666754e-05,
      "loss": 0.4748,
      "num_input_tokens_seen": 3789768,
      "step": 5815
    },
    {
      "epoch": 3.050314465408805,
      "grad_norm": 0.8851591944694519,
      "learning_rate": 4.958153937127783e-05,
      "loss": 0.4276,
      "num_input_tokens_seen": 3792936,
      "step": 5820
    },
    {
      "epoch": 3.05293501048218,
      "grad_norm": 0.9433837532997131,
      "learning_rate": 4.95794534787113e-05,
      "loss": 0.4774,
      "num_input_tokens_seen": 3796168,
      "step": 5825
    },
    {
      "epoch": 3.0555555555555554,
      "grad_norm": 1.4860334396362305,
      "learning_rate": 4.957736244440353e-05,
      "loss": 0.3982,
      "num_input_tokens_seen": 3798696,
      "step": 5830
    },
    {
      "epoch": 3.058176100628931,
      "grad_norm": 1.4935872554779053,
      "learning_rate": 4.957526626879193e-05,
      "loss": 0.3567,
      "num_input_tokens_seen": 3801992,
      "step": 5835
    },
    {
      "epoch": 3.060796645702306,
      "grad_norm": 0.7341769933700562,
      "learning_rate": 4.957316495231499e-05,
      "loss": 0.3186,
      "num_input_tokens_seen": 3804648,
      "step": 5840
    },
    {
      "epoch": 3.0634171907756813,
      "grad_norm": 0.7455694675445557,
      "learning_rate": 4.9571058495412294e-05,
      "loss": 0.5449,
      "num_input_tokens_seen": 3808296,
      "step": 5845
    },
    {
      "epoch": 3.0660377358490565,
      "grad_norm": 0.8649548888206482,
      "learning_rate": 4.9568946898524485e-05,
      "loss": 0.434,
      "num_input_tokens_seen": 3811688,
      "step": 5850
    },
    {
      "epoch": 3.068658280922432,
      "grad_norm": 0.7672677636146545,
      "learning_rate": 4.956683016209327e-05,
      "loss": 0.3668,
      "num_input_tokens_seen": 3817416,
      "step": 5855
    },
    {
      "epoch": 3.0712788259958073,
      "grad_norm": 1.1332206726074219,
      "learning_rate": 4.956470828656147e-05,
      "loss": 0.4149,
      "num_input_tokens_seen": 3820168,
      "step": 5860
    },
    {
      "epoch": 3.0738993710691824,
      "grad_norm": 0.8085625767707825,
      "learning_rate": 4.956258127237294e-05,
      "loss": 0.3658,
      "num_input_tokens_seen": 3823720,
      "step": 5865
    },
    {
      "epoch": 3.0765199161425576,
      "grad_norm": 1.4324132204055786,
      "learning_rate": 4.956044911997263e-05,
      "loss": 0.4006,
      "num_input_tokens_seen": 3827432,
      "step": 5870
    },
    {
      "epoch": 3.0791404612159328,
      "grad_norm": 0.9290270805358887,
      "learning_rate": 4.955831182980657e-05,
      "loss": 0.4354,
      "num_input_tokens_seen": 3829832,
      "step": 5875
    },
    {
      "epoch": 3.0817610062893084,
      "grad_norm": 0.8877552151679993,
      "learning_rate": 4.955616940232185e-05,
      "loss": 0.3875,
      "num_input_tokens_seen": 3832040,
      "step": 5880
    },
    {
      "epoch": 3.0843815513626835,
      "grad_norm": 1.0286595821380615,
      "learning_rate": 4.9554021837966636e-05,
      "loss": 0.3911,
      "num_input_tokens_seen": 3834984,
      "step": 5885
    },
    {
      "epoch": 3.0870020964360587,
      "grad_norm": 0.8961387276649475,
      "learning_rate": 4.955186913719019e-05,
      "loss": 0.4218,
      "num_input_tokens_seen": 3841800,
      "step": 5890
    },
    {
      "epoch": 3.089622641509434,
      "grad_norm": 1.0418752431869507,
      "learning_rate": 4.954971130044281e-05,
      "loss": 0.3863,
      "num_input_tokens_seen": 3844456,
      "step": 5895
    },
    {
      "epoch": 3.092243186582809,
      "grad_norm": 0.9727331399917603,
      "learning_rate": 4.954754832817592e-05,
      "loss": 0.3133,
      "num_input_tokens_seen": 3847912,
      "step": 5900
    },
    {
      "epoch": 3.0948637316561847,
      "grad_norm": 0.6993407011032104,
      "learning_rate": 4.954538022084196e-05,
      "loss": 0.3056,
      "num_input_tokens_seen": 3851208,
      "step": 5905
    },
    {
      "epoch": 3.09748427672956,
      "grad_norm": 0.4802064895629883,
      "learning_rate": 4.9543206978894496e-05,
      "loss": 0.3126,
      "num_input_tokens_seen": 3854920,
      "step": 5910
    },
    {
      "epoch": 3.100104821802935,
      "grad_norm": 0.7421488761901855,
      "learning_rate": 4.954102860278813e-05,
      "loss": 0.421,
      "num_input_tokens_seen": 3858408,
      "step": 5915
    },
    {
      "epoch": 3.10272536687631,
      "grad_norm": 1.40814208984375,
      "learning_rate": 4.9538845092978567e-05,
      "loss": 0.4308,
      "num_input_tokens_seen": 3861352,
      "step": 5920
    },
    {
      "epoch": 3.1053459119496853,
      "grad_norm": 1.0323113203048706,
      "learning_rate": 4.953665644992256e-05,
      "loss": 0.3318,
      "num_input_tokens_seen": 3864680,
      "step": 5925
    },
    {
      "epoch": 3.107966457023061,
      "grad_norm": 0.9712859392166138,
      "learning_rate": 4.953446267407795e-05,
      "loss": 0.3854,
      "num_input_tokens_seen": 3867176,
      "step": 5930
    },
    {
      "epoch": 3.110587002096436,
      "grad_norm": 1.1362375020980835,
      "learning_rate": 4.953226376590365e-05,
      "loss": 0.3963,
      "num_input_tokens_seen": 3869864,
      "step": 5935
    },
    {
      "epoch": 3.1132075471698113,
      "grad_norm": 0.7223990559577942,
      "learning_rate": 4.953005972585965e-05,
      "loss": 0.3791,
      "num_input_tokens_seen": 3872744,
      "step": 5940
    },
    {
      "epoch": 3.1158280922431865,
      "grad_norm": 1.2037715911865234,
      "learning_rate": 4.952785055440701e-05,
      "loss": 0.42,
      "num_input_tokens_seen": 3875880,
      "step": 5945
    },
    {
      "epoch": 3.1184486373165616,
      "grad_norm": 1.261928677558899,
      "learning_rate": 4.9525636252007864e-05,
      "loss": 0.342,
      "num_input_tokens_seen": 3879048,
      "step": 5950
    },
    {
      "epoch": 3.1210691823899372,
      "grad_norm": 1.0340309143066406,
      "learning_rate": 4.9523416819125425e-05,
      "loss": 0.2877,
      "num_input_tokens_seen": 3882376,
      "step": 5955
    },
    {
      "epoch": 3.1236897274633124,
      "grad_norm": 0.698131263256073,
      "learning_rate": 4.952119225622396e-05,
      "loss": 0.3403,
      "num_input_tokens_seen": 3885576,
      "step": 5960
    },
    {
      "epoch": 3.1263102725366876,
      "grad_norm": 1.5765507221221924,
      "learning_rate": 4.951896256376882e-05,
      "loss": 0.3799,
      "num_input_tokens_seen": 3888808,
      "step": 5965
    },
    {
      "epoch": 3.1289308176100628,
      "grad_norm": 1.1219828128814697,
      "learning_rate": 4.9516727742226444e-05,
      "loss": 0.3869,
      "num_input_tokens_seen": 3891976,
      "step": 5970
    },
    {
      "epoch": 3.131551362683438,
      "grad_norm": 0.834359884262085,
      "learning_rate": 4.9514487792064323e-05,
      "loss": 0.3886,
      "num_input_tokens_seen": 3895592,
      "step": 5975
    },
    {
      "epoch": 3.1341719077568135,
      "grad_norm": 0.9926354885101318,
      "learning_rate": 4.951224271375104e-05,
      "loss": 0.5383,
      "num_input_tokens_seen": 3899336,
      "step": 5980
    },
    {
      "epoch": 3.1367924528301887,
      "grad_norm": 1.2236173152923584,
      "learning_rate": 4.950999250775623e-05,
      "loss": 0.3282,
      "num_input_tokens_seen": 3902280,
      "step": 5985
    },
    {
      "epoch": 3.139412997903564,
      "grad_norm": 0.8497328162193298,
      "learning_rate": 4.950773717455061e-05,
      "loss": 0.3936,
      "num_input_tokens_seen": 3905672,
      "step": 5990
    },
    {
      "epoch": 3.142033542976939,
      "grad_norm": 0.8646019697189331,
      "learning_rate": 4.950547671460596e-05,
      "loss": 0.3407,
      "num_input_tokens_seen": 3908392,
      "step": 5995
    },
    {
      "epoch": 3.1446540880503147,
      "grad_norm": 0.9519004225730896,
      "learning_rate": 4.950321112839517e-05,
      "loss": 0.4646,
      "num_input_tokens_seen": 3912200,
      "step": 6000
    },
    {
      "epoch": 3.14727463312369,
      "grad_norm": 0.9908936023712158,
      "learning_rate": 4.9500940416392156e-05,
      "loss": 0.4663,
      "num_input_tokens_seen": 3915304,
      "step": 6005
    },
    {
      "epoch": 3.149895178197065,
      "grad_norm": 1.347917079925537,
      "learning_rate": 4.949866457907193e-05,
      "loss": 0.4484,
      "num_input_tokens_seen": 3918952,
      "step": 6010
    },
    {
      "epoch": 3.15251572327044,
      "grad_norm": 0.9105600118637085,
      "learning_rate": 4.9496383616910565e-05,
      "loss": 0.5275,
      "num_input_tokens_seen": 3921864,
      "step": 6015
    },
    {
      "epoch": 3.1551362683438153,
      "grad_norm": 0.8782954216003418,
      "learning_rate": 4.949409753038522e-05,
      "loss": 0.3472,
      "num_input_tokens_seen": 3925256,
      "step": 6020
    },
    {
      "epoch": 3.157756813417191,
      "grad_norm": 1.079141616821289,
      "learning_rate": 4.949180631997411e-05,
      "loss": 0.3559,
      "num_input_tokens_seen": 3927432,
      "step": 6025
    },
    {
      "epoch": 3.160377358490566,
      "grad_norm": 1.4222123622894287,
      "learning_rate": 4.948950998615654e-05,
      "loss": 0.3798,
      "num_input_tokens_seen": 3931368,
      "step": 6030
    },
    {
      "epoch": 3.1629979035639413,
      "grad_norm": 1.452899694442749,
      "learning_rate": 4.9487208529412876e-05,
      "loss": 0.3492,
      "num_input_tokens_seen": 3934504,
      "step": 6035
    },
    {
      "epoch": 3.1656184486373165,
      "grad_norm": 0.9443694949150085,
      "learning_rate": 4.9484901950224546e-05,
      "loss": 0.371,
      "num_input_tokens_seen": 3937384,
      "step": 6040
    },
    {
      "epoch": 3.1682389937106916,
      "grad_norm": 1.4499586820602417,
      "learning_rate": 4.948259024907407e-05,
      "loss": 0.4939,
      "num_input_tokens_seen": 3940168,
      "step": 6045
    },
    {
      "epoch": 3.1708595387840672,
      "grad_norm": 0.9191929697990417,
      "learning_rate": 4.948027342644502e-05,
      "loss": 0.3664,
      "num_input_tokens_seen": 3942504,
      "step": 6050
    },
    {
      "epoch": 3.1734800838574424,
      "grad_norm": 0.6803352236747742,
      "learning_rate": 4.947795148282206e-05,
      "loss": 0.442,
      "num_input_tokens_seen": 3945480,
      "step": 6055
    },
    {
      "epoch": 3.1761006289308176,
      "grad_norm": 1.3428441286087036,
      "learning_rate": 4.947562441869091e-05,
      "loss": 0.3956,
      "num_input_tokens_seen": 3948520,
      "step": 6060
    },
    {
      "epoch": 3.1787211740041927,
      "grad_norm": 0.60222327709198,
      "learning_rate": 4.947329223453836e-05,
      "loss": 0.4821,
      "num_input_tokens_seen": 3952904,
      "step": 6065
    },
    {
      "epoch": 3.181341719077568,
      "grad_norm": 0.9909277558326721,
      "learning_rate": 4.947095493085229e-05,
      "loss": 0.4253,
      "num_input_tokens_seen": 3955816,
      "step": 6070
    },
    {
      "epoch": 3.1839622641509435,
      "grad_norm": 3.8593947887420654,
      "learning_rate": 4.9468612508121616e-05,
      "loss": 0.4036,
      "num_input_tokens_seen": 3958120,
      "step": 6075
    },
    {
      "epoch": 3.1865828092243187,
      "grad_norm": 1.2177132368087769,
      "learning_rate": 4.9466264966836365e-05,
      "loss": 0.4205,
      "num_input_tokens_seen": 3960520,
      "step": 6080
    },
    {
      "epoch": 3.189203354297694,
      "grad_norm": 1.4875718355178833,
      "learning_rate": 4.94639123074876e-05,
      "loss": 0.4086,
      "num_input_tokens_seen": 3963208,
      "step": 6085
    },
    {
      "epoch": 3.191823899371069,
      "grad_norm": 1.7334163188934326,
      "learning_rate": 4.946155453056749e-05,
      "loss": 0.4606,
      "num_input_tokens_seen": 3966568,
      "step": 6090
    },
    {
      "epoch": 3.1944444444444446,
      "grad_norm": 1.3273634910583496,
      "learning_rate": 4.945919163656923e-05,
      "loss": 0.485,
      "num_input_tokens_seen": 3970184,
      "step": 6095
    },
    {
      "epoch": 3.19706498951782,
      "grad_norm": 0.9589629769325256,
      "learning_rate": 4.945682362598715e-05,
      "loss": 0.4413,
      "num_input_tokens_seen": 3973192,
      "step": 6100
    },
    {
      "epoch": 3.199685534591195,
      "grad_norm": 1.2858089208602905,
      "learning_rate": 4.945445049931656e-05,
      "loss": 0.3642,
      "num_input_tokens_seen": 3975656,
      "step": 6105
    },
    {
      "epoch": 3.20230607966457,
      "grad_norm": 0.7108312845230103,
      "learning_rate": 4.945207225705393e-05,
      "loss": 0.3625,
      "num_input_tokens_seen": 3979240,
      "step": 6110
    },
    {
      "epoch": 3.2049266247379453,
      "grad_norm": 0.6286655068397522,
      "learning_rate": 4.944968889969674e-05,
      "loss": 0.38,
      "num_input_tokens_seen": 3983464,
      "step": 6115
    },
    {
      "epoch": 3.207547169811321,
      "grad_norm": 1.1475647687911987,
      "learning_rate": 4.944730042774358e-05,
      "loss": 0.4326,
      "num_input_tokens_seen": 3986568,
      "step": 6120
    },
    {
      "epoch": 3.210167714884696,
      "grad_norm": 1.199719786643982,
      "learning_rate": 4.944490684169408e-05,
      "loss": 0.3973,
      "num_input_tokens_seen": 3989224,
      "step": 6125
    },
    {
      "epoch": 3.2127882599580713,
      "grad_norm": 0.8042672276496887,
      "learning_rate": 4.9442508142048945e-05,
      "loss": 0.5504,
      "num_input_tokens_seen": 3992648,
      "step": 6130
    },
    {
      "epoch": 3.2154088050314464,
      "grad_norm": 0.6426253914833069,
      "learning_rate": 4.9440104329309956e-05,
      "loss": 0.3884,
      "num_input_tokens_seen": 3995528,
      "step": 6135
    },
    {
      "epoch": 3.2180293501048216,
      "grad_norm": 1.084206223487854,
      "learning_rate": 4.943769540397998e-05,
      "loss": 0.4018,
      "num_input_tokens_seen": 3998536,
      "step": 6140
    },
    {
      "epoch": 3.220649895178197,
      "grad_norm": 1.2554007768630981,
      "learning_rate": 4.943528136656292e-05,
      "loss": 0.4113,
      "num_input_tokens_seen": 4001224,
      "step": 6145
    },
    {
      "epoch": 3.2232704402515724,
      "grad_norm": 0.8149362206459045,
      "learning_rate": 4.943286221756377e-05,
      "loss": 0.3497,
      "num_input_tokens_seen": 4004328,
      "step": 6150
    },
    {
      "epoch": 3.2258909853249476,
      "grad_norm": 0.9029017090797424,
      "learning_rate": 4.9430437957488586e-05,
      "loss": 0.3934,
      "num_input_tokens_seen": 4007944,
      "step": 6155
    },
    {
      "epoch": 3.2285115303983227,
      "grad_norm": 1.931251883506775,
      "learning_rate": 4.94280085868445e-05,
      "loss": 0.406,
      "num_input_tokens_seen": 4011144,
      "step": 6160
    },
    {
      "epoch": 3.231132075471698,
      "grad_norm": 1.0669505596160889,
      "learning_rate": 4.942557410613972e-05,
      "loss": 0.3428,
      "num_input_tokens_seen": 4014376,
      "step": 6165
    },
    {
      "epoch": 3.2337526205450735,
      "grad_norm": 0.6336100697517395,
      "learning_rate": 4.942313451588348e-05,
      "loss": 0.4156,
      "num_input_tokens_seen": 4018216,
      "step": 6170
    },
    {
      "epoch": 3.2363731656184487,
      "grad_norm": 1.2897422313690186,
      "learning_rate": 4.942068981658614e-05,
      "loss": 0.4443,
      "num_input_tokens_seen": 4021032,
      "step": 6175
    },
    {
      "epoch": 3.238993710691824,
      "grad_norm": 0.970782995223999,
      "learning_rate": 4.9418240008759097e-05,
      "loss": 0.4841,
      "num_input_tokens_seen": 4024040,
      "step": 6180
    },
    {
      "epoch": 3.241614255765199,
      "grad_norm": 1.2762718200683594,
      "learning_rate": 4.9415785092914826e-05,
      "loss": 0.2822,
      "num_input_tokens_seen": 4027336,
      "step": 6185
    },
    {
      "epoch": 3.2442348008385746,
      "grad_norm": 1.1479854583740234,
      "learning_rate": 4.941332506956686e-05,
      "loss": 0.4607,
      "num_input_tokens_seen": 4030472,
      "step": 6190
    },
    {
      "epoch": 3.24685534591195,
      "grad_norm": 0.6795832514762878,
      "learning_rate": 4.941085993922981e-05,
      "loss": 0.4021,
      "num_input_tokens_seen": 4034312,
      "step": 6195
    },
    {
      "epoch": 3.249475890985325,
      "grad_norm": 0.6156799793243408,
      "learning_rate": 4.9408389702419354e-05,
      "loss": 0.3975,
      "num_input_tokens_seen": 4038632,
      "step": 6200
    },
    {
      "epoch": 3.2520964360587,
      "grad_norm": 0.9672621488571167,
      "learning_rate": 4.940591435965224e-05,
      "loss": 0.4599,
      "num_input_tokens_seen": 4042824,
      "step": 6205
    },
    {
      "epoch": 3.2547169811320753,
      "grad_norm": 1.8667571544647217,
      "learning_rate": 4.9403433911446285e-05,
      "loss": 0.3996,
      "num_input_tokens_seen": 4045896,
      "step": 6210
    },
    {
      "epoch": 3.257337526205451,
      "grad_norm": 1.263399600982666,
      "learning_rate": 4.9400948358320356e-05,
      "loss": 0.4463,
      "num_input_tokens_seen": 4048712,
      "step": 6215
    },
    {
      "epoch": 3.259958071278826,
      "grad_norm": 0.8394221663475037,
      "learning_rate": 4.939845770079442e-05,
      "loss": 0.4342,
      "num_input_tokens_seen": 4051176,
      "step": 6220
    },
    {
      "epoch": 3.2625786163522013,
      "grad_norm": 0.9216780662536621,
      "learning_rate": 4.9395961939389475e-05,
      "loss": 0.3123,
      "num_input_tokens_seen": 4054056,
      "step": 6225
    },
    {
      "epoch": 3.2651991614255764,
      "grad_norm": 0.982231080532074,
      "learning_rate": 4.939346107462762e-05,
      "loss": 0.3504,
      "num_input_tokens_seen": 4058120,
      "step": 6230
    },
    {
      "epoch": 3.2678197064989516,
      "grad_norm": 1.5753626823425293,
      "learning_rate": 4.9390955107032e-05,
      "loss": 0.4411,
      "num_input_tokens_seen": 4060840,
      "step": 6235
    },
    {
      "epoch": 3.270440251572327,
      "grad_norm": 0.5701562762260437,
      "learning_rate": 4.938844403712685e-05,
      "loss": 0.4549,
      "num_input_tokens_seen": 4064136,
      "step": 6240
    },
    {
      "epoch": 3.2730607966457024,
      "grad_norm": 0.8589027523994446,
      "learning_rate": 4.938592786543743e-05,
      "loss": 0.408,
      "num_input_tokens_seen": 4066824,
      "step": 6245
    },
    {
      "epoch": 3.2756813417190775,
      "grad_norm": 1.009003758430481,
      "learning_rate": 4.938340659249012e-05,
      "loss": 0.5398,
      "num_input_tokens_seen": 4069864,
      "step": 6250
    },
    {
      "epoch": 3.2783018867924527,
      "grad_norm": 0.48534920811653137,
      "learning_rate": 4.938088021881233e-05,
      "loss": 0.3084,
      "num_input_tokens_seen": 4072936,
      "step": 6255
    },
    {
      "epoch": 3.280922431865828,
      "grad_norm": 0.7315918803215027,
      "learning_rate": 4.937834874493255e-05,
      "loss": 0.5458,
      "num_input_tokens_seen": 4075400,
      "step": 6260
    },
    {
      "epoch": 3.2835429769392035,
      "grad_norm": 1.61141037940979,
      "learning_rate": 4.9375812171380334e-05,
      "loss": 0.4462,
      "num_input_tokens_seen": 4078696,
      "step": 6265
    },
    {
      "epoch": 3.2861635220125787,
      "grad_norm": 1.226337194442749,
      "learning_rate": 4.9373270498686306e-05,
      "loss": 0.4093,
      "num_input_tokens_seen": 4082824,
      "step": 6270
    },
    {
      "epoch": 3.288784067085954,
      "grad_norm": 1.378833293914795,
      "learning_rate": 4.937072372738216e-05,
      "loss": 0.5266,
      "num_input_tokens_seen": 4085448,
      "step": 6275
    },
    {
      "epoch": 3.291404612159329,
      "grad_norm": 1.2038683891296387,
      "learning_rate": 4.9368171858000635e-05,
      "loss": 0.4563,
      "num_input_tokens_seen": 4088424,
      "step": 6280
    },
    {
      "epoch": 3.2940251572327046,
      "grad_norm": 0.7184142470359802,
      "learning_rate": 4.9365614891075576e-05,
      "loss": 0.4712,
      "num_input_tokens_seen": 4091752,
      "step": 6285
    },
    {
      "epoch": 3.29664570230608,
      "grad_norm": 1.278807520866394,
      "learning_rate": 4.936305282714185e-05,
      "loss": 0.4039,
      "num_input_tokens_seen": 4094568,
      "step": 6290
    },
    {
      "epoch": 3.299266247379455,
      "grad_norm": 0.8060164451599121,
      "learning_rate": 4.936048566673542e-05,
      "loss": 0.4301,
      "num_input_tokens_seen": 4097352,
      "step": 6295
    },
    {
      "epoch": 3.30188679245283,
      "grad_norm": 0.8819817900657654,
      "learning_rate": 4.935791341039332e-05,
      "loss": 0.3483,
      "num_input_tokens_seen": 4100328,
      "step": 6300
    },
    {
      "epoch": 3.3045073375262053,
      "grad_norm": 1.3920414447784424,
      "learning_rate": 4.935533605865361e-05,
      "loss": 0.3686,
      "num_input_tokens_seen": 4102728,
      "step": 6305
    },
    {
      "epoch": 3.307127882599581,
      "grad_norm": 0.7212669253349304,
      "learning_rate": 4.935275361205547e-05,
      "loss": 0.3171,
      "num_input_tokens_seen": 4105352,
      "step": 6310
    },
    {
      "epoch": 3.309748427672956,
      "grad_norm": 1.1000702381134033,
      "learning_rate": 4.93501660711391e-05,
      "loss": 0.3669,
      "num_input_tokens_seen": 4108744,
      "step": 6315
    },
    {
      "epoch": 3.3123689727463312,
      "grad_norm": 0.9713027477264404,
      "learning_rate": 4.934757343644578e-05,
      "loss": 0.304,
      "num_input_tokens_seen": 4112168,
      "step": 6320
    },
    {
      "epoch": 3.3149895178197064,
      "grad_norm": 0.763192355632782,
      "learning_rate": 4.934497570851788e-05,
      "loss": 0.5749,
      "num_input_tokens_seen": 4115240,
      "step": 6325
    },
    {
      "epoch": 3.3176100628930816,
      "grad_norm": 0.7287151217460632,
      "learning_rate": 4.9342372887898804e-05,
      "loss": 0.3962,
      "num_input_tokens_seen": 4117864,
      "step": 6330
    },
    {
      "epoch": 3.320230607966457,
      "grad_norm": 1.8309669494628906,
      "learning_rate": 4.9339764975133026e-05,
      "loss": 0.3901,
      "num_input_tokens_seen": 4121960,
      "step": 6335
    },
    {
      "epoch": 3.3228511530398324,
      "grad_norm": 0.993034839630127,
      "learning_rate": 4.93371519707661e-05,
      "loss": 0.4823,
      "num_input_tokens_seen": 4125160,
      "step": 6340
    },
    {
      "epoch": 3.3254716981132075,
      "grad_norm": 0.9403859972953796,
      "learning_rate": 4.933453387534464e-05,
      "loss": 0.4624,
      "num_input_tokens_seen": 4128168,
      "step": 6345
    },
    {
      "epoch": 3.3280922431865827,
      "grad_norm": 1.2773547172546387,
      "learning_rate": 4.933191068941631e-05,
      "loss": 0.4812,
      "num_input_tokens_seen": 4134632,
      "step": 6350
    },
    {
      "epoch": 3.330712788259958,
      "grad_norm": 0.6516627669334412,
      "learning_rate": 4.932928241352987e-05,
      "loss": 0.46,
      "num_input_tokens_seen": 4138088,
      "step": 6355
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 1.1404597759246826,
      "learning_rate": 4.9326649048235095e-05,
      "loss": 0.3699,
      "num_input_tokens_seen": 4141608,
      "step": 6360
    },
    {
      "epoch": 3.3359538784067087,
      "grad_norm": 1.0205936431884766,
      "learning_rate": 4.932401059408287e-05,
      "loss": 0.3363,
      "num_input_tokens_seen": 4144712,
      "step": 6365
    },
    {
      "epoch": 3.338574423480084,
      "grad_norm": 1.651198148727417,
      "learning_rate": 4.932136705162514e-05,
      "loss": 0.3393,
      "num_input_tokens_seen": 4147624,
      "step": 6370
    },
    {
      "epoch": 3.341194968553459,
      "grad_norm": 0.9476110339164734,
      "learning_rate": 4.93187184214149e-05,
      "loss": 0.3207,
      "num_input_tokens_seen": 4150728,
      "step": 6375
    },
    {
      "epoch": 3.3438155136268346,
      "grad_norm": 1.0697945356369019,
      "learning_rate": 4.9316064704006205e-05,
      "loss": 0.3731,
      "num_input_tokens_seen": 4153928,
      "step": 6380
    },
    {
      "epoch": 3.3464360587002098,
      "grad_norm": 0.9830542206764221,
      "learning_rate": 4.931340589995419e-05,
      "loss": 0.35,
      "num_input_tokens_seen": 4157608,
      "step": 6385
    },
    {
      "epoch": 3.349056603773585,
      "grad_norm": 0.9970285892486572,
      "learning_rate": 4.9310742009815035e-05,
      "loss": 0.3788,
      "num_input_tokens_seen": 4160648,
      "step": 6390
    },
    {
      "epoch": 3.35167714884696,
      "grad_norm": 0.9040840864181519,
      "learning_rate": 4.930807303414601e-05,
      "loss": 0.4356,
      "num_input_tokens_seen": 4163720,
      "step": 6395
    },
    {
      "epoch": 3.3542976939203353,
      "grad_norm": 1.1255191564559937,
      "learning_rate": 4.930539897350542e-05,
      "loss": 0.3991,
      "num_input_tokens_seen": 4166920,
      "step": 6400
    },
    {
      "epoch": 3.3569182389937104,
      "grad_norm": 0.9268117547035217,
      "learning_rate": 4.930271982845266e-05,
      "loss": 0.2989,
      "num_input_tokens_seen": 4170024,
      "step": 6405
    },
    {
      "epoch": 3.359538784067086,
      "grad_norm": 0.9821680784225464,
      "learning_rate": 4.930003559954817e-05,
      "loss": 0.3232,
      "num_input_tokens_seen": 4173288,
      "step": 6410
    },
    {
      "epoch": 3.3621593291404612,
      "grad_norm": 1.6421144008636475,
      "learning_rate": 4.929734628735346e-05,
      "loss": 0.3416,
      "num_input_tokens_seen": 4175688,
      "step": 6415
    },
    {
      "epoch": 3.3647798742138364,
      "grad_norm": 1.5972604751586914,
      "learning_rate": 4.9294651892431115e-05,
      "loss": 0.3222,
      "num_input_tokens_seen": 4178504,
      "step": 6420
    },
    {
      "epoch": 3.3674004192872116,
      "grad_norm": 1.1488289833068848,
      "learning_rate": 4.929195241534476e-05,
      "loss": 0.3318,
      "num_input_tokens_seen": 4180840,
      "step": 6425
    },
    {
      "epoch": 3.370020964360587,
      "grad_norm": 0.8587620258331299,
      "learning_rate": 4.9289247856659095e-05,
      "loss": 0.3987,
      "num_input_tokens_seen": 4184328,
      "step": 6430
    },
    {
      "epoch": 3.3726415094339623,
      "grad_norm": 0.7146474123001099,
      "learning_rate": 4.9286538216939883e-05,
      "loss": 0.5072,
      "num_input_tokens_seen": 4188072,
      "step": 6435
    },
    {
      "epoch": 3.3752620545073375,
      "grad_norm": 0.9223299026489258,
      "learning_rate": 4.9283823496753944e-05,
      "loss": 0.377,
      "num_input_tokens_seen": 4191848,
      "step": 6440
    },
    {
      "epoch": 3.3778825995807127,
      "grad_norm": 0.8883240222930908,
      "learning_rate": 4.928110369666918e-05,
      "loss": 0.3647,
      "num_input_tokens_seen": 4194440,
      "step": 6445
    },
    {
      "epoch": 3.380503144654088,
      "grad_norm": 1.210593581199646,
      "learning_rate": 4.927837881725455e-05,
      "loss": 0.4449,
      "num_input_tokens_seen": 4197256,
      "step": 6450
    },
    {
      "epoch": 3.3831236897274635,
      "grad_norm": 0.9623485207557678,
      "learning_rate": 4.927564885908003e-05,
      "loss": 0.3953,
      "num_input_tokens_seen": 4200872,
      "step": 6455
    },
    {
      "epoch": 3.3857442348008386,
      "grad_norm": 0.907528281211853,
      "learning_rate": 4.9272913822716735e-05,
      "loss": 0.3399,
      "num_input_tokens_seen": 4204232,
      "step": 6460
    },
    {
      "epoch": 3.388364779874214,
      "grad_norm": 1.333681344985962,
      "learning_rate": 4.927017370873678e-05,
      "loss": 0.3857,
      "num_input_tokens_seen": 4207304,
      "step": 6465
    },
    {
      "epoch": 3.390985324947589,
      "grad_norm": 1.2970720529556274,
      "learning_rate": 4.926742851771338e-05,
      "loss": 0.371,
      "num_input_tokens_seen": 4209832,
      "step": 6470
    },
    {
      "epoch": 3.3936058700209646,
      "grad_norm": 0.83357173204422,
      "learning_rate": 4.926467825022079e-05,
      "loss": 0.3747,
      "num_input_tokens_seen": 4213128,
      "step": 6475
    },
    {
      "epoch": 3.3962264150943398,
      "grad_norm": 1.2785167694091797,
      "learning_rate": 4.926192290683434e-05,
      "loss": 0.33,
      "num_input_tokens_seen": 4215976,
      "step": 6480
    },
    {
      "epoch": 3.398846960167715,
      "grad_norm": 1.3169469833374023,
      "learning_rate": 4.9259162488130404e-05,
      "loss": 0.5633,
      "num_input_tokens_seen": 4218632,
      "step": 6485
    },
    {
      "epoch": 3.40146750524109,
      "grad_norm": 1.7832547426223755,
      "learning_rate": 4.925639699468645e-05,
      "loss": 0.4088,
      "num_input_tokens_seen": 4221416,
      "step": 6490
    },
    {
      "epoch": 3.4040880503144653,
      "grad_norm": 1.2289718389511108,
      "learning_rate": 4.925362642708097e-05,
      "loss": 0.3082,
      "num_input_tokens_seen": 4224392,
      "step": 6495
    },
    {
      "epoch": 3.4067085953878404,
      "grad_norm": 1.7332994937896729,
      "learning_rate": 4.9250850785893536e-05,
      "loss": 0.377,
      "num_input_tokens_seen": 4227176,
      "step": 6500
    },
    {
      "epoch": 3.409329140461216,
      "grad_norm": 1.1506484746932983,
      "learning_rate": 4.924807007170479e-05,
      "loss": 0.4426,
      "num_input_tokens_seen": 4229960,
      "step": 6505
    },
    {
      "epoch": 3.411949685534591,
      "grad_norm": 1.3971794843673706,
      "learning_rate": 4.924528428509643e-05,
      "loss": 0.3391,
      "num_input_tokens_seen": 4232136,
      "step": 6510
    },
    {
      "epoch": 3.4145702306079664,
      "grad_norm": 1.136109709739685,
      "learning_rate": 4.9242493426651196e-05,
      "loss": 0.4878,
      "num_input_tokens_seen": 4235624,
      "step": 6515
    },
    {
      "epoch": 3.4171907756813416,
      "grad_norm": 0.5613622069358826,
      "learning_rate": 4.92396974969529e-05,
      "loss": 0.4769,
      "num_input_tokens_seen": 4240616,
      "step": 6520
    },
    {
      "epoch": 3.419811320754717,
      "grad_norm": 1.0391418933868408,
      "learning_rate": 4.923689649658644e-05,
      "loss": 0.3938,
      "num_input_tokens_seen": 4243528,
      "step": 6525
    },
    {
      "epoch": 3.4224318658280923,
      "grad_norm": 0.6256993412971497,
      "learning_rate": 4.923409042613774e-05,
      "loss": 0.4468,
      "num_input_tokens_seen": 4247208,
      "step": 6530
    },
    {
      "epoch": 3.4250524109014675,
      "grad_norm": 1.1770777702331543,
      "learning_rate": 4.92312792861938e-05,
      "loss": 0.2934,
      "num_input_tokens_seen": 4252424,
      "step": 6535
    },
    {
      "epoch": 3.4276729559748427,
      "grad_norm": 1.0962988138198853,
      "learning_rate": 4.922846307734268e-05,
      "loss": 0.441,
      "num_input_tokens_seen": 4255752,
      "step": 6540
    },
    {
      "epoch": 3.430293501048218,
      "grad_norm": 0.8378953337669373,
      "learning_rate": 4.9225641800173496e-05,
      "loss": 0.4253,
      "num_input_tokens_seen": 4258472,
      "step": 6545
    },
    {
      "epoch": 3.4329140461215935,
      "grad_norm": 0.9646049737930298,
      "learning_rate": 4.922281545527643e-05,
      "loss": 0.3165,
      "num_input_tokens_seen": 4260904,
      "step": 6550
    },
    {
      "epoch": 3.4355345911949686,
      "grad_norm": 1.005703330039978,
      "learning_rate": 4.9219984043242715e-05,
      "loss": 0.497,
      "num_input_tokens_seen": 4264104,
      "step": 6555
    },
    {
      "epoch": 3.438155136268344,
      "grad_norm": 0.7121990323066711,
      "learning_rate": 4.921714756466466e-05,
      "loss": 0.4634,
      "num_input_tokens_seen": 4268648,
      "step": 6560
    },
    {
      "epoch": 3.440775681341719,
      "grad_norm": 1.2076674699783325,
      "learning_rate": 4.921430602013561e-05,
      "loss": 0.4589,
      "num_input_tokens_seen": 4272776,
      "step": 6565
    },
    {
      "epoch": 3.4433962264150946,
      "grad_norm": 1.0302152633666992,
      "learning_rate": 4.9211459410249995e-05,
      "loss": 0.4033,
      "num_input_tokens_seen": 4275784,
      "step": 6570
    },
    {
      "epoch": 3.4460167714884697,
      "grad_norm": 0.8913229703903198,
      "learning_rate": 4.92086077356033e-05,
      "loss": 0.5345,
      "num_input_tokens_seen": 4279400,
      "step": 6575
    },
    {
      "epoch": 3.448637316561845,
      "grad_norm": 0.9824510812759399,
      "learning_rate": 4.920575099679204e-05,
      "loss": 0.3449,
      "num_input_tokens_seen": 4282920,
      "step": 6580
    },
    {
      "epoch": 3.45125786163522,
      "grad_norm": 0.9357719421386719,
      "learning_rate": 4.920288919441383e-05,
      "loss": 0.4258,
      "num_input_tokens_seen": 4286856,
      "step": 6585
    },
    {
      "epoch": 3.4538784067085953,
      "grad_norm": 0.6542942523956299,
      "learning_rate": 4.9200022329067316e-05,
      "loss": 0.451,
      "num_input_tokens_seen": 4290216,
      "step": 6590
    },
    {
      "epoch": 3.4564989517819704,
      "grad_norm": 2.106917142868042,
      "learning_rate": 4.9197150401352224e-05,
      "loss": 0.56,
      "num_input_tokens_seen": 4293768,
      "step": 6595
    },
    {
      "epoch": 3.459119496855346,
      "grad_norm": 0.5384864807128906,
      "learning_rate": 4.919427341186933e-05,
      "loss": 0.3869,
      "num_input_tokens_seen": 4297288,
      "step": 6600
    },
    {
      "epoch": 3.461740041928721,
      "grad_norm": 1.280884027481079,
      "learning_rate": 4.919139136122045e-05,
      "loss": 0.3756,
      "num_input_tokens_seen": 4299592,
      "step": 6605
    },
    {
      "epoch": 3.4643605870020964,
      "grad_norm": 1.4849138259887695,
      "learning_rate": 4.918850425000849e-05,
      "loss": 0.4151,
      "num_input_tokens_seen": 4302600,
      "step": 6610
    },
    {
      "epoch": 3.4669811320754715,
      "grad_norm": 1.8397822380065918,
      "learning_rate": 4.918561207883741e-05,
      "loss": 0.5114,
      "num_input_tokens_seen": 4305640,
      "step": 6615
    },
    {
      "epoch": 3.469601677148847,
      "grad_norm": 1.4299182891845703,
      "learning_rate": 4.9182714848312195e-05,
      "loss": 0.3597,
      "num_input_tokens_seen": 4308168,
      "step": 6620
    },
    {
      "epoch": 3.4722222222222223,
      "grad_norm": 1.157762050628662,
      "learning_rate": 4.917981255903893e-05,
      "loss": 0.3516,
      "num_input_tokens_seen": 4311208,
      "step": 6625
    },
    {
      "epoch": 3.4748427672955975,
      "grad_norm": 1.2243340015411377,
      "learning_rate": 4.917690521162474e-05,
      "loss": 0.467,
      "num_input_tokens_seen": 4314312,
      "step": 6630
    },
    {
      "epoch": 3.4774633123689727,
      "grad_norm": 1.0844427347183228,
      "learning_rate": 4.9173992806677796e-05,
      "loss": 0.4617,
      "num_input_tokens_seen": 4316936,
      "step": 6635
    },
    {
      "epoch": 3.480083857442348,
      "grad_norm": 1.5236471891403198,
      "learning_rate": 4.917107534480735e-05,
      "loss": 0.354,
      "num_input_tokens_seen": 4320232,
      "step": 6640
    },
    {
      "epoch": 3.4827044025157234,
      "grad_norm": 1.1897436380386353,
      "learning_rate": 4.916815282662369e-05,
      "loss": 0.5025,
      "num_input_tokens_seen": 4324104,
      "step": 6645
    },
    {
      "epoch": 3.4853249475890986,
      "grad_norm": 0.7816571593284607,
      "learning_rate": 4.91652252527382e-05,
      "loss": 0.5879,
      "num_input_tokens_seen": 4327304,
      "step": 6650
    },
    {
      "epoch": 3.487945492662474,
      "grad_norm": 1.1815670728683472,
      "learning_rate": 4.9162292623763275e-05,
      "loss": 0.4306,
      "num_input_tokens_seen": 4332008,
      "step": 6655
    },
    {
      "epoch": 3.490566037735849,
      "grad_norm": 0.984565794467926,
      "learning_rate": 4.9159354940312386e-05,
      "loss": 0.4133,
      "num_input_tokens_seen": 4334728,
      "step": 6660
    },
    {
      "epoch": 3.4931865828092246,
      "grad_norm": 0.9274908304214478,
      "learning_rate": 4.915641220300007e-05,
      "loss": 0.3889,
      "num_input_tokens_seen": 4337416,
      "step": 6665
    },
    {
      "epoch": 3.4958071278825997,
      "grad_norm": 0.8563914895057678,
      "learning_rate": 4.9153464412441915e-05,
      "loss": 0.3109,
      "num_input_tokens_seen": 4340328,
      "step": 6670
    },
    {
      "epoch": 3.498427672955975,
      "grad_norm": 0.9087803363800049,
      "learning_rate": 4.915051156925456e-05,
      "loss": 0.3919,
      "num_input_tokens_seen": 4344456,
      "step": 6675
    },
    {
      "epoch": 3.50104821802935,
      "grad_norm": 0.9049600958824158,
      "learning_rate": 4.91475536740557e-05,
      "loss": 0.3791,
      "num_input_tokens_seen": 4347944,
      "step": 6680
    },
    {
      "epoch": 3.5036687631027252,
      "grad_norm": 0.5957373976707458,
      "learning_rate": 4.914459072746411e-05,
      "loss": 0.4286,
      "num_input_tokens_seen": 4350760,
      "step": 6685
    },
    {
      "epoch": 3.5062893081761004,
      "grad_norm": 0.7596591711044312,
      "learning_rate": 4.91416227300996e-05,
      "loss": 0.4403,
      "num_input_tokens_seen": 4354760,
      "step": 6690
    },
    {
      "epoch": 3.508909853249476,
      "grad_norm": 1.2277746200561523,
      "learning_rate": 4.913864968258303e-05,
      "loss": 0.3484,
      "num_input_tokens_seen": 4358120,
      "step": 6695
    },
    {
      "epoch": 3.511530398322851,
      "grad_norm": 0.6656280755996704,
      "learning_rate": 4.913567158553634e-05,
      "loss": 0.3614,
      "num_input_tokens_seen": 4361352,
      "step": 6700
    },
    {
      "epoch": 3.5141509433962264,
      "grad_norm": 1.5374025106430054,
      "learning_rate": 4.9132688439582505e-05,
      "loss": 0.4954,
      "num_input_tokens_seen": 4363784,
      "step": 6705
    },
    {
      "epoch": 3.5167714884696015,
      "grad_norm": 0.7973252534866333,
      "learning_rate": 4.9129700245345565e-05,
      "loss": 0.367,
      "num_input_tokens_seen": 4367400,
      "step": 6710
    },
    {
      "epoch": 3.519392033542977,
      "grad_norm": 0.8128684163093567,
      "learning_rate": 4.912670700345063e-05,
      "loss": 0.4562,
      "num_input_tokens_seen": 4372232,
      "step": 6715
    },
    {
      "epoch": 3.5220125786163523,
      "grad_norm": 0.7820790410041809,
      "learning_rate": 4.912370871452383e-05,
      "loss": 0.3785,
      "num_input_tokens_seen": 4375560,
      "step": 6720
    },
    {
      "epoch": 3.5246331236897275,
      "grad_norm": 1.9953089952468872,
      "learning_rate": 4.91207053791924e-05,
      "loss": 0.4077,
      "num_input_tokens_seen": 4378216,
      "step": 6725
    },
    {
      "epoch": 3.5272536687631026,
      "grad_norm": 1.762518048286438,
      "learning_rate": 4.911769699808457e-05,
      "loss": 0.5375,
      "num_input_tokens_seen": 4380744,
      "step": 6730
    },
    {
      "epoch": 3.529874213836478,
      "grad_norm": 0.9607658982276917,
      "learning_rate": 4.911468357182969e-05,
      "loss": 0.3535,
      "num_input_tokens_seen": 4384264,
      "step": 6735
    },
    {
      "epoch": 3.532494758909853,
      "grad_norm": 0.864285945892334,
      "learning_rate": 4.9111665101058116e-05,
      "loss": 0.5431,
      "num_input_tokens_seen": 4388200,
      "step": 6740
    },
    {
      "epoch": 3.5351153039832286,
      "grad_norm": 1.6000064611434937,
      "learning_rate": 4.910864158640128e-05,
      "loss": 0.4126,
      "num_input_tokens_seen": 4390760,
      "step": 6745
    },
    {
      "epoch": 3.5377358490566038,
      "grad_norm": 2.3871049880981445,
      "learning_rate": 4.910561302849168e-05,
      "loss": 0.4769,
      "num_input_tokens_seen": 4393416,
      "step": 6750
    },
    {
      "epoch": 3.540356394129979,
      "grad_norm": 0.6293793320655823,
      "learning_rate": 4.910257942796284e-05,
      "loss": 0.4138,
      "num_input_tokens_seen": 4396392,
      "step": 6755
    },
    {
      "epoch": 3.5429769392033545,
      "grad_norm": 0.5463760495185852,
      "learning_rate": 4.909954078544936e-05,
      "loss": 0.361,
      "num_input_tokens_seen": 4399496,
      "step": 6760
    },
    {
      "epoch": 3.5455974842767297,
      "grad_norm": 0.6085829138755798,
      "learning_rate": 4.909649710158689e-05,
      "loss": 0.4195,
      "num_input_tokens_seen": 4402440,
      "step": 6765
    },
    {
      "epoch": 3.548218029350105,
      "grad_norm": 1.1748483180999756,
      "learning_rate": 4.909344837701214e-05,
      "loss": 0.5096,
      "num_input_tokens_seen": 4405672,
      "step": 6770
    },
    {
      "epoch": 3.55083857442348,
      "grad_norm": 1.8194676637649536,
      "learning_rate": 4.909039461236285e-05,
      "loss": 0.4912,
      "num_input_tokens_seen": 4408872,
      "step": 6775
    },
    {
      "epoch": 3.5534591194968552,
      "grad_norm": 1.167304515838623,
      "learning_rate": 4.908733580827786e-05,
      "loss": 0.3798,
      "num_input_tokens_seen": 4411112,
      "step": 6780
    },
    {
      "epoch": 3.5560796645702304,
      "grad_norm": 1.0087831020355225,
      "learning_rate": 4.9084271965397014e-05,
      "loss": 0.3478,
      "num_input_tokens_seen": 4414248,
      "step": 6785
    },
    {
      "epoch": 3.558700209643606,
      "grad_norm": 1.297440528869629,
      "learning_rate": 4.908120308436124e-05,
      "loss": 0.4939,
      "num_input_tokens_seen": 4417160,
      "step": 6790
    },
    {
      "epoch": 3.561320754716981,
      "grad_norm": 1.3149268627166748,
      "learning_rate": 4.907812916581251e-05,
      "loss": 0.3657,
      "num_input_tokens_seen": 4421032,
      "step": 6795
    },
    {
      "epoch": 3.5639412997903563,
      "grad_norm": 1.0576889514923096,
      "learning_rate": 4.907505021039386e-05,
      "loss": 0.5331,
      "num_input_tokens_seen": 4423848,
      "step": 6800
    },
    {
      "epoch": 3.5665618448637315,
      "grad_norm": 1.0414042472839355,
      "learning_rate": 4.907196621874937e-05,
      "loss": 0.3711,
      "num_input_tokens_seen": 4427144,
      "step": 6805
    },
    {
      "epoch": 3.569182389937107,
      "grad_norm": 0.9795339703559875,
      "learning_rate": 4.9068877191524176e-05,
      "loss": 0.4426,
      "num_input_tokens_seen": 4429896,
      "step": 6810
    },
    {
      "epoch": 3.5718029350104823,
      "grad_norm": 1.255523920059204,
      "learning_rate": 4.906578312936446e-05,
      "loss": 0.4417,
      "num_input_tokens_seen": 4435720,
      "step": 6815
    },
    {
      "epoch": 3.5744234800838575,
      "grad_norm": 0.8333434462547302,
      "learning_rate": 4.906268403291747e-05,
      "loss": 0.3887,
      "num_input_tokens_seen": 4439048,
      "step": 6820
    },
    {
      "epoch": 3.5770440251572326,
      "grad_norm": 0.9159239530563354,
      "learning_rate": 4.905957990283151e-05,
      "loss": 0.4433,
      "num_input_tokens_seen": 4441928,
      "step": 6825
    },
    {
      "epoch": 3.579664570230608,
      "grad_norm": 0.7120321393013,
      "learning_rate": 4.905647073975592e-05,
      "loss": 0.3595,
      "num_input_tokens_seen": 4445160,
      "step": 6830
    },
    {
      "epoch": 3.582285115303983,
      "grad_norm": 1.4039952754974365,
      "learning_rate": 4.9053356544341095e-05,
      "loss": 0.4288,
      "num_input_tokens_seen": 4448328,
      "step": 6835
    },
    {
      "epoch": 3.5849056603773586,
      "grad_norm": 0.9985044002532959,
      "learning_rate": 4.90502373172385e-05,
      "loss": 0.3353,
      "num_input_tokens_seen": 4451816,
      "step": 6840
    },
    {
      "epoch": 3.5875262054507338,
      "grad_norm": 1.2913278341293335,
      "learning_rate": 4.9047113059100635e-05,
      "loss": 0.4467,
      "num_input_tokens_seen": 4454856,
      "step": 6845
    },
    {
      "epoch": 3.590146750524109,
      "grad_norm": 1.9000893831253052,
      "learning_rate": 4.904398377058106e-05,
      "loss": 0.4262,
      "num_input_tokens_seen": 4457544,
      "step": 6850
    },
    {
      "epoch": 3.5927672955974845,
      "grad_norm": 0.9648944139480591,
      "learning_rate": 4.90408494523344e-05,
      "loss": 0.4769,
      "num_input_tokens_seen": 4461160,
      "step": 6855
    },
    {
      "epoch": 3.5953878406708597,
      "grad_norm": 1.0440672636032104,
      "learning_rate": 4.90377101050163e-05,
      "loss": 0.364,
      "num_input_tokens_seen": 4465000,
      "step": 6860
    },
    {
      "epoch": 3.598008385744235,
      "grad_norm": 0.9734400510787964,
      "learning_rate": 4.903456572928348e-05,
      "loss": 0.4255,
      "num_input_tokens_seen": 4467944,
      "step": 6865
    },
    {
      "epoch": 3.60062893081761,
      "grad_norm": 1.0626991987228394,
      "learning_rate": 4.903141632579371e-05,
      "loss": 0.4761,
      "num_input_tokens_seen": 4471880,
      "step": 6870
    },
    {
      "epoch": 3.603249475890985,
      "grad_norm": 1.18192458152771,
      "learning_rate": 4.902826189520581e-05,
      "loss": 0.3656,
      "num_input_tokens_seen": 4474184,
      "step": 6875
    },
    {
      "epoch": 3.6058700209643604,
      "grad_norm": 0.8430137038230896,
      "learning_rate": 4.9025102438179656e-05,
      "loss": 0.3548,
      "num_input_tokens_seen": 4476584,
      "step": 6880
    },
    {
      "epoch": 3.608490566037736,
      "grad_norm": 1.6094813346862793,
      "learning_rate": 4.902193795537616e-05,
      "loss": 0.351,
      "num_input_tokens_seen": 4479592,
      "step": 6885
    },
    {
      "epoch": 3.611111111111111,
      "grad_norm": 1.5240206718444824,
      "learning_rate": 4.90187684474573e-05,
      "loss": 0.3937,
      "num_input_tokens_seen": 4482280,
      "step": 6890
    },
    {
      "epoch": 3.6137316561844863,
      "grad_norm": 1.4217917919158936,
      "learning_rate": 4.901559391508611e-05,
      "loss": 0.3319,
      "num_input_tokens_seen": 4485224,
      "step": 6895
    },
    {
      "epoch": 3.6163522012578615,
      "grad_norm": 1.0608932971954346,
      "learning_rate": 4.9012414358926644e-05,
      "loss": 0.4962,
      "num_input_tokens_seen": 4488680,
      "step": 6900
    },
    {
      "epoch": 3.618972746331237,
      "grad_norm": 1.4690831899642944,
      "learning_rate": 4.900922977964405e-05,
      "loss": 0.3992,
      "num_input_tokens_seen": 4492104,
      "step": 6905
    },
    {
      "epoch": 3.6215932914046123,
      "grad_norm": 0.8474972248077393,
      "learning_rate": 4.90060401779045e-05,
      "loss": 0.3954,
      "num_input_tokens_seen": 4495240,
      "step": 6910
    },
    {
      "epoch": 3.6242138364779874,
      "grad_norm": 1.2352638244628906,
      "learning_rate": 4.900284555437521e-05,
      "loss": 0.2925,
      "num_input_tokens_seen": 4498088,
      "step": 6915
    },
    {
      "epoch": 3.6268343815513626,
      "grad_norm": 1.176666498184204,
      "learning_rate": 4.899964590972448e-05,
      "loss": 0.3877,
      "num_input_tokens_seen": 4501224,
      "step": 6920
    },
    {
      "epoch": 3.629454926624738,
      "grad_norm": 1.1437629461288452,
      "learning_rate": 4.899644124462163e-05,
      "loss": 0.4713,
      "num_input_tokens_seen": 4503784,
      "step": 6925
    },
    {
      "epoch": 3.632075471698113,
      "grad_norm": 1.349230170249939,
      "learning_rate": 4.8993231559737026e-05,
      "loss": 0.4187,
      "num_input_tokens_seen": 4506568,
      "step": 6930
    },
    {
      "epoch": 3.6346960167714886,
      "grad_norm": 1.1238436698913574,
      "learning_rate": 4.8990016855742114e-05,
      "loss": 0.3329,
      "num_input_tokens_seen": 4509096,
      "step": 6935
    },
    {
      "epoch": 3.6373165618448637,
      "grad_norm": 0.8705005645751953,
      "learning_rate": 4.898679713330937e-05,
      "loss": 0.2966,
      "num_input_tokens_seen": 4512200,
      "step": 6940
    },
    {
      "epoch": 3.639937106918239,
      "grad_norm": 0.8434275388717651,
      "learning_rate": 4.898357239311232e-05,
      "loss": 0.3631,
      "num_input_tokens_seen": 4514984,
      "step": 6945
    },
    {
      "epoch": 3.6425576519916145,
      "grad_norm": 0.7391166687011719,
      "learning_rate": 4.898034263582554e-05,
      "loss": 0.3888,
      "num_input_tokens_seen": 4518568,
      "step": 6950
    },
    {
      "epoch": 3.6451781970649897,
      "grad_norm": 1.4811266660690308,
      "learning_rate": 4.897710786212467e-05,
      "loss": 0.7161,
      "num_input_tokens_seen": 4521832,
      "step": 6955
    },
    {
      "epoch": 3.647798742138365,
      "grad_norm": 1.760496973991394,
      "learning_rate": 4.897386807268638e-05,
      "loss": 0.4817,
      "num_input_tokens_seen": 4525640,
      "step": 6960
    },
    {
      "epoch": 3.65041928721174,
      "grad_norm": 1.3298124074935913,
      "learning_rate": 4.897062326818839e-05,
      "loss": 0.3832,
      "num_input_tokens_seen": 4529608,
      "step": 6965
    },
    {
      "epoch": 3.653039832285115,
      "grad_norm": 1.5391603708267212,
      "learning_rate": 4.8967373449309495e-05,
      "loss": 0.4933,
      "num_input_tokens_seen": 4532328,
      "step": 6970
    },
    {
      "epoch": 3.6556603773584904,
      "grad_norm": 0.8147874474525452,
      "learning_rate": 4.89641186167295e-05,
      "loss": 0.4055,
      "num_input_tokens_seen": 4535336,
      "step": 6975
    },
    {
      "epoch": 3.658280922431866,
      "grad_norm": 0.8389527797698975,
      "learning_rate": 4.8960858771129294e-05,
      "loss": 0.3161,
      "num_input_tokens_seen": 4538760,
      "step": 6980
    },
    {
      "epoch": 3.660901467505241,
      "grad_norm": 1.058800458908081,
      "learning_rate": 4.895759391319079e-05,
      "loss": 0.3564,
      "num_input_tokens_seen": 4546376,
      "step": 6985
    },
    {
      "epoch": 3.6635220125786163,
      "grad_norm": 1.2404736280441284,
      "learning_rate": 4.8954324043596954e-05,
      "loss": 0.4287,
      "num_input_tokens_seen": 4550376,
      "step": 6990
    },
    {
      "epoch": 3.6661425576519915,
      "grad_norm": 0.9133602380752563,
      "learning_rate": 4.8951049163031826e-05,
      "loss": 0.5272,
      "num_input_tokens_seen": 4553736,
      "step": 6995
    },
    {
      "epoch": 3.668763102725367,
      "grad_norm": 0.6850408315658569,
      "learning_rate": 4.894776927218045e-05,
      "loss": 0.3568,
      "num_input_tokens_seen": 4556936,
      "step": 7000
    },
    {
      "epoch": 3.6713836477987423,
      "grad_norm": 1.2048203945159912,
      "learning_rate": 4.894448437172896e-05,
      "loss": 0.392,
      "num_input_tokens_seen": 4559816,
      "step": 7005
    },
    {
      "epoch": 3.6740041928721174,
      "grad_norm": 1.5040264129638672,
      "learning_rate": 4.894119446236452e-05,
      "loss": 0.3839,
      "num_input_tokens_seen": 4564520,
      "step": 7010
    },
    {
      "epoch": 3.6766247379454926,
      "grad_norm": 1.1475716829299927,
      "learning_rate": 4.893789954477532e-05,
      "loss": 0.4314,
      "num_input_tokens_seen": 4567592,
      "step": 7015
    },
    {
      "epoch": 3.6792452830188678,
      "grad_norm": 0.8891334533691406,
      "learning_rate": 4.893459961965063e-05,
      "loss": 0.414,
      "num_input_tokens_seen": 4570408,
      "step": 7020
    },
    {
      "epoch": 3.681865828092243,
      "grad_norm": 1.290022850036621,
      "learning_rate": 4.8931294687680775e-05,
      "loss": 0.3899,
      "num_input_tokens_seen": 4573320,
      "step": 7025
    },
    {
      "epoch": 3.6844863731656186,
      "grad_norm": 0.7076283693313599,
      "learning_rate": 4.8927984749557086e-05,
      "loss": 0.4816,
      "num_input_tokens_seen": 4576680,
      "step": 7030
    },
    {
      "epoch": 3.6871069182389937,
      "grad_norm": 1.1218749284744263,
      "learning_rate": 4.892466980597198e-05,
      "loss": 0.398,
      "num_input_tokens_seen": 4579368,
      "step": 7035
    },
    {
      "epoch": 3.689727463312369,
      "grad_norm": 0.8436912298202515,
      "learning_rate": 4.892134985761889e-05,
      "loss": 0.5091,
      "num_input_tokens_seen": 4582440,
      "step": 7040
    },
    {
      "epoch": 3.6923480083857445,
      "grad_norm": 0.8576446175575256,
      "learning_rate": 4.891802490519232e-05,
      "loss": 0.3445,
      "num_input_tokens_seen": 4585128,
      "step": 7045
    },
    {
      "epoch": 3.6949685534591197,
      "grad_norm": 1.5902254581451416,
      "learning_rate": 4.891469494938781e-05,
      "loss": 0.3954,
      "num_input_tokens_seen": 4588168,
      "step": 7050
    },
    {
      "epoch": 3.697589098532495,
      "grad_norm": 1.0492736101150513,
      "learning_rate": 4.891135999090195e-05,
      "loss": 0.3826,
      "num_input_tokens_seen": 4591400,
      "step": 7055
    },
    {
      "epoch": 3.70020964360587,
      "grad_norm": 1.6428471803665161,
      "learning_rate": 4.890802003043238e-05,
      "loss": 0.4033,
      "num_input_tokens_seen": 4594280,
      "step": 7060
    },
    {
      "epoch": 3.702830188679245,
      "grad_norm": 0.7974018454551697,
      "learning_rate": 4.890467506867777e-05,
      "loss": 0.3075,
      "num_input_tokens_seen": 4598536,
      "step": 7065
    },
    {
      "epoch": 3.7054507337526204,
      "grad_norm": 1.0932921171188354,
      "learning_rate": 4.890132510633786e-05,
      "loss": 0.4461,
      "num_input_tokens_seen": 4602440,
      "step": 7070
    },
    {
      "epoch": 3.708071278825996,
      "grad_norm": 0.6856849193572998,
      "learning_rate": 4.8897970144113416e-05,
      "loss": 0.4231,
      "num_input_tokens_seen": 4605352,
      "step": 7075
    },
    {
      "epoch": 3.710691823899371,
      "grad_norm": 0.7217128872871399,
      "learning_rate": 4.889461018270626e-05,
      "loss": 0.3129,
      "num_input_tokens_seen": 4610152,
      "step": 7080
    },
    {
      "epoch": 3.7133123689727463,
      "grad_norm": 0.8175809383392334,
      "learning_rate": 4.889124522281926e-05,
      "loss": 0.3586,
      "num_input_tokens_seen": 4612872,
      "step": 7085
    },
    {
      "epoch": 3.7159329140461215,
      "grad_norm": 0.941222071647644,
      "learning_rate": 4.888787526515632e-05,
      "loss": 0.4715,
      "num_input_tokens_seen": 4617064,
      "step": 7090
    },
    {
      "epoch": 3.718553459119497,
      "grad_norm": 0.9366174936294556,
      "learning_rate": 4.8884500310422406e-05,
      "loss": 0.4661,
      "num_input_tokens_seen": 4621608,
      "step": 7095
    },
    {
      "epoch": 3.7211740041928723,
      "grad_norm": 2.5156311988830566,
      "learning_rate": 4.8881120359323516e-05,
      "loss": 0.4161,
      "num_input_tokens_seen": 4623880,
      "step": 7100
    },
    {
      "epoch": 3.7237945492662474,
      "grad_norm": 1.0754098892211914,
      "learning_rate": 4.887773541256669e-05,
      "loss": 0.3733,
      "num_input_tokens_seen": 4627080,
      "step": 7105
    },
    {
      "epoch": 3.7264150943396226,
      "grad_norm": 1.0330195426940918,
      "learning_rate": 4.887434547086004e-05,
      "loss": 0.3604,
      "num_input_tokens_seen": 4630536,
      "step": 7110
    },
    {
      "epoch": 3.7290356394129978,
      "grad_norm": 1.4790118932724,
      "learning_rate": 4.887095053491269e-05,
      "loss": 0.4398,
      "num_input_tokens_seen": 4633256,
      "step": 7115
    },
    {
      "epoch": 3.731656184486373,
      "grad_norm": 0.8518035411834717,
      "learning_rate": 4.8867550605434806e-05,
      "loss": 0.4221,
      "num_input_tokens_seen": 4636104,
      "step": 7120
    },
    {
      "epoch": 3.7342767295597485,
      "grad_norm": 0.9881477952003479,
      "learning_rate": 4.8864145683137644e-05,
      "loss": 0.3672,
      "num_input_tokens_seen": 4639272,
      "step": 7125
    },
    {
      "epoch": 3.7368972746331237,
      "grad_norm": 1.0333510637283325,
      "learning_rate": 4.886073576873346e-05,
      "loss": 0.3055,
      "num_input_tokens_seen": 4641800,
      "step": 7130
    },
    {
      "epoch": 3.739517819706499,
      "grad_norm": 0.7686599493026733,
      "learning_rate": 4.885732086293556e-05,
      "loss": 0.5432,
      "num_input_tokens_seen": 4645160,
      "step": 7135
    },
    {
      "epoch": 3.742138364779874,
      "grad_norm": 1.119585633277893,
      "learning_rate": 4.8853900966458335e-05,
      "loss": 0.3152,
      "num_input_tokens_seen": 4648488,
      "step": 7140
    },
    {
      "epoch": 3.7447589098532497,
      "grad_norm": 1.1383711099624634,
      "learning_rate": 4.8850476080017146e-05,
      "loss": 0.3847,
      "num_input_tokens_seen": 4651944,
      "step": 7145
    },
    {
      "epoch": 3.747379454926625,
      "grad_norm": 0.8685924410820007,
      "learning_rate": 4.884704620432848e-05,
      "loss": 0.4368,
      "num_input_tokens_seen": 4655816,
      "step": 7150
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.1029866933822632,
      "learning_rate": 4.8843611340109794e-05,
      "loss": 0.3765,
      "num_input_tokens_seen": 4658824,
      "step": 7155
    },
    {
      "epoch": 3.752620545073375,
      "grad_norm": 0.8463260531425476,
      "learning_rate": 4.884017148807965e-05,
      "loss": 0.4366,
      "num_input_tokens_seen": 4662440,
      "step": 7160
    },
    {
      "epoch": 3.7552410901467503,
      "grad_norm": 1.2845873832702637,
      "learning_rate": 4.883672664895761e-05,
      "loss": 0.4077,
      "num_input_tokens_seen": 4666280,
      "step": 7165
    },
    {
      "epoch": 3.757861635220126,
      "grad_norm": 0.7143834233283997,
      "learning_rate": 4.8833276823464296e-05,
      "loss": 0.3468,
      "num_input_tokens_seen": 4669544,
      "step": 7170
    },
    {
      "epoch": 3.760482180293501,
      "grad_norm": 0.7375748753547668,
      "learning_rate": 4.882982201232138e-05,
      "loss": 0.3639,
      "num_input_tokens_seen": 4672392,
      "step": 7175
    },
    {
      "epoch": 3.7631027253668763,
      "grad_norm": 0.9987705945968628,
      "learning_rate": 4.882636221625156e-05,
      "loss": 0.4725,
      "num_input_tokens_seen": 4676872,
      "step": 7180
    },
    {
      "epoch": 3.7657232704402515,
      "grad_norm": 0.588759183883667,
      "learning_rate": 4.8822897435978586e-05,
      "loss": 0.5295,
      "num_input_tokens_seen": 4680520,
      "step": 7185
    },
    {
      "epoch": 3.768343815513627,
      "grad_norm": 0.5891568660736084,
      "learning_rate": 4.8819427672227264e-05,
      "loss": 0.4158,
      "num_input_tokens_seen": 4683464,
      "step": 7190
    },
    {
      "epoch": 3.7709643605870022,
      "grad_norm": 0.6737049221992493,
      "learning_rate": 4.881595292572341e-05,
      "loss": 0.3889,
      "num_input_tokens_seen": 4686728,
      "step": 7195
    },
    {
      "epoch": 3.7735849056603774,
      "grad_norm": 1.1879229545593262,
      "learning_rate": 4.881247319719392e-05,
      "loss": 0.4084,
      "num_input_tokens_seen": 4690376,
      "step": 7200
    },
    {
      "epoch": 3.7762054507337526,
      "grad_norm": 1.1835381984710693,
      "learning_rate": 4.88089884873667e-05,
      "loss": 0.3668,
      "num_input_tokens_seen": 4695304,
      "step": 7205
    },
    {
      "epoch": 3.7788259958071277,
      "grad_norm": 1.5781872272491455,
      "learning_rate": 4.880549879697071e-05,
      "loss": 0.4661,
      "num_input_tokens_seen": 4698216,
      "step": 7210
    },
    {
      "epoch": 3.781446540880503,
      "grad_norm": 1.0947580337524414,
      "learning_rate": 4.8802004126735965e-05,
      "loss": 0.4899,
      "num_input_tokens_seen": 4701512,
      "step": 7215
    },
    {
      "epoch": 3.7840670859538785,
      "grad_norm": 1.1945852041244507,
      "learning_rate": 4.87985044773935e-05,
      "loss": 0.4764,
      "num_input_tokens_seen": 4704744,
      "step": 7220
    },
    {
      "epoch": 3.7866876310272537,
      "grad_norm": 0.9997036457061768,
      "learning_rate": 4.879499984967541e-05,
      "loss": 0.5354,
      "num_input_tokens_seen": 4707880,
      "step": 7225
    },
    {
      "epoch": 3.789308176100629,
      "grad_norm": 0.6964150071144104,
      "learning_rate": 4.8791490244314824e-05,
      "loss": 0.3703,
      "num_input_tokens_seen": 4711080,
      "step": 7230
    },
    {
      "epoch": 3.791928721174004,
      "grad_norm": 0.9721601009368896,
      "learning_rate": 4.8787975662045904e-05,
      "loss": 0.3478,
      "num_input_tokens_seen": 4714760,
      "step": 7235
    },
    {
      "epoch": 3.7945492662473796,
      "grad_norm": 0.6385480165481567,
      "learning_rate": 4.878445610360386e-05,
      "loss": 0.342,
      "num_input_tokens_seen": 4718248,
      "step": 7240
    },
    {
      "epoch": 3.797169811320755,
      "grad_norm": 1.346512794494629,
      "learning_rate": 4.878093156972494e-05,
      "loss": 0.6998,
      "num_input_tokens_seen": 4721480,
      "step": 7245
    },
    {
      "epoch": 3.79979035639413,
      "grad_norm": 0.807059109210968,
      "learning_rate": 4.877740206114645e-05,
      "loss": 0.3935,
      "num_input_tokens_seen": 4724328,
      "step": 7250
    },
    {
      "epoch": 3.802410901467505,
      "grad_norm": 1.0380648374557495,
      "learning_rate": 4.8773867578606716e-05,
      "loss": 0.3463,
      "num_input_tokens_seen": 4727432,
      "step": 7255
    },
    {
      "epoch": 3.8050314465408803,
      "grad_norm": 0.5576488375663757,
      "learning_rate": 4.877032812284511e-05,
      "loss": 0.4861,
      "num_input_tokens_seen": 4730952,
      "step": 7260
    },
    {
      "epoch": 3.8076519916142555,
      "grad_norm": 1.6327413320541382,
      "learning_rate": 4.876678369460205e-05,
      "loss": 0.4487,
      "num_input_tokens_seen": 4733704,
      "step": 7265
    },
    {
      "epoch": 3.810272536687631,
      "grad_norm": 0.8489397764205933,
      "learning_rate": 4.8763234294618985e-05,
      "loss": 0.4217,
      "num_input_tokens_seen": 4736424,
      "step": 7270
    },
    {
      "epoch": 3.8128930817610063,
      "grad_norm": 0.702168881893158,
      "learning_rate": 4.875967992363841e-05,
      "loss": 0.389,
      "num_input_tokens_seen": 4739944,
      "step": 7275
    },
    {
      "epoch": 3.8155136268343814,
      "grad_norm": 0.8440321087837219,
      "learning_rate": 4.875612058240386e-05,
      "loss": 0.3352,
      "num_input_tokens_seen": 4743656,
      "step": 7280
    },
    {
      "epoch": 3.818134171907757,
      "grad_norm": 1.145267367362976,
      "learning_rate": 4.875255627165991e-05,
      "loss": 0.3845,
      "num_input_tokens_seen": 4746440,
      "step": 7285
    },
    {
      "epoch": 3.8207547169811322,
      "grad_norm": 1.0708993673324585,
      "learning_rate": 4.874898699215217e-05,
      "loss": 0.4107,
      "num_input_tokens_seen": 4749672,
      "step": 7290
    },
    {
      "epoch": 3.8233752620545074,
      "grad_norm": 0.8365983366966248,
      "learning_rate": 4.8745412744627296e-05,
      "loss": 0.4923,
      "num_input_tokens_seen": 4753224,
      "step": 7295
    },
    {
      "epoch": 3.8259958071278826,
      "grad_norm": 1.2231765985488892,
      "learning_rate": 4.874183352983297e-05,
      "loss": 0.3478,
      "num_input_tokens_seen": 4756072,
      "step": 7300
    },
    {
      "epoch": 3.8286163522012577,
      "grad_norm": 0.6578301191329956,
      "learning_rate": 4.873824934851794e-05,
      "loss": 0.3661,
      "num_input_tokens_seen": 4759368,
      "step": 7305
    },
    {
      "epoch": 3.831236897274633,
      "grad_norm": 0.6878703236579895,
      "learning_rate": 4.873466020143196e-05,
      "loss": 0.4834,
      "num_input_tokens_seen": 4762280,
      "step": 7310
    },
    {
      "epoch": 3.8338574423480085,
      "grad_norm": 0.8841665983200073,
      "learning_rate": 4.873106608932585e-05,
      "loss": 0.3535,
      "num_input_tokens_seen": 4765544,
      "step": 7315
    },
    {
      "epoch": 3.8364779874213837,
      "grad_norm": 2.0477614402770996,
      "learning_rate": 4.872746701295145e-05,
      "loss": 0.4008,
      "num_input_tokens_seen": 4768424,
      "step": 7320
    },
    {
      "epoch": 3.839098532494759,
      "grad_norm": 1.4802749156951904,
      "learning_rate": 4.872386297306165e-05,
      "loss": 0.4015,
      "num_input_tokens_seen": 4771464,
      "step": 7325
    },
    {
      "epoch": 3.841719077568134,
      "grad_norm": 0.783979594707489,
      "learning_rate": 4.872025397041037e-05,
      "loss": 0.3689,
      "num_input_tokens_seen": 4775208,
      "step": 7330
    },
    {
      "epoch": 3.8443396226415096,
      "grad_norm": 1.0347563028335571,
      "learning_rate": 4.871664000575258e-05,
      "loss": 0.317,
      "num_input_tokens_seen": 4777768,
      "step": 7335
    },
    {
      "epoch": 3.846960167714885,
      "grad_norm": 0.8686429858207703,
      "learning_rate": 4.871302107984427e-05,
      "loss": 0.3611,
      "num_input_tokens_seen": 4781864,
      "step": 7340
    },
    {
      "epoch": 3.84958071278826,
      "grad_norm": 1.0228917598724365,
      "learning_rate": 4.8709397193442474e-05,
      "loss": 0.3442,
      "num_input_tokens_seen": 4785352,
      "step": 7345
    },
    {
      "epoch": 3.852201257861635,
      "grad_norm": 3.0762743949890137,
      "learning_rate": 4.870576834730529e-05,
      "loss": 0.3924,
      "num_input_tokens_seen": 4787848,
      "step": 7350
    },
    {
      "epoch": 3.8548218029350103,
      "grad_norm": 0.9841413497924805,
      "learning_rate": 4.8702134542191816e-05,
      "loss": 0.2888,
      "num_input_tokens_seen": 4790952,
      "step": 7355
    },
    {
      "epoch": 3.8574423480083855,
      "grad_norm": 0.39285123348236084,
      "learning_rate": 4.86984957788622e-05,
      "loss": 0.2566,
      "num_input_tokens_seen": 4794824,
      "step": 7360
    },
    {
      "epoch": 3.860062893081761,
      "grad_norm": 0.8710809350013733,
      "learning_rate": 4.869485205807763e-05,
      "loss": 0.3679,
      "num_input_tokens_seen": 4798120,
      "step": 7365
    },
    {
      "epoch": 3.8626834381551363,
      "grad_norm": 0.9596309065818787,
      "learning_rate": 4.869120338060034e-05,
      "loss": 0.3664,
      "num_input_tokens_seen": 4801704,
      "step": 7370
    },
    {
      "epoch": 3.8653039832285114,
      "grad_norm": 1.1519687175750732,
      "learning_rate": 4.8687549747193585e-05,
      "loss": 0.5577,
      "num_input_tokens_seen": 4804776,
      "step": 7375
    },
    {
      "epoch": 3.867924528301887,
      "grad_norm": 0.7259008884429932,
      "learning_rate": 4.868389115862166e-05,
      "loss": 0.5522,
      "num_input_tokens_seen": 4807496,
      "step": 7380
    },
    {
      "epoch": 3.870545073375262,
      "grad_norm": 0.884678840637207,
      "learning_rate": 4.868022761564991e-05,
      "loss": 0.4244,
      "num_input_tokens_seen": 4810152,
      "step": 7385
    },
    {
      "epoch": 3.8731656184486374,
      "grad_norm": 0.7229189276695251,
      "learning_rate": 4.8676559119044696e-05,
      "loss": 0.2965,
      "num_input_tokens_seen": 4813736,
      "step": 7390
    },
    {
      "epoch": 3.8757861635220126,
      "grad_norm": 1.4034565687179565,
      "learning_rate": 4.867288566957344e-05,
      "loss": 0.4559,
      "num_input_tokens_seen": 4816904,
      "step": 7395
    },
    {
      "epoch": 3.8784067085953877,
      "grad_norm": 0.873242974281311,
      "learning_rate": 4.866920726800457e-05,
      "loss": 0.357,
      "num_input_tokens_seen": 4820168,
      "step": 7400
    },
    {
      "epoch": 3.881027253668763,
      "grad_norm": 1.2845332622528076,
      "learning_rate": 4.8665523915107576e-05,
      "loss": 0.5646,
      "num_input_tokens_seen": 4823176,
      "step": 7405
    },
    {
      "epoch": 3.8836477987421385,
      "grad_norm": 0.9738649725914001,
      "learning_rate": 4.866183561165296e-05,
      "loss": 0.362,
      "num_input_tokens_seen": 4826344,
      "step": 7410
    },
    {
      "epoch": 3.8862683438155137,
      "grad_norm": 0.9298363327980042,
      "learning_rate": 4.86581423584123e-05,
      "loss": 0.3922,
      "num_input_tokens_seen": 4830120,
      "step": 7415
    },
    {
      "epoch": 3.888888888888889,
      "grad_norm": 0.7659473419189453,
      "learning_rate": 4.8654444156158154e-05,
      "loss": 0.3735,
      "num_input_tokens_seen": 4833576,
      "step": 7420
    },
    {
      "epoch": 3.891509433962264,
      "grad_norm": 0.9985786080360413,
      "learning_rate": 4.865074100566415e-05,
      "loss": 0.3427,
      "num_input_tokens_seen": 4836776,
      "step": 7425
    },
    {
      "epoch": 3.8941299790356396,
      "grad_norm": 0.7982422113418579,
      "learning_rate": 4.864703290770495e-05,
      "loss": 0.5216,
      "num_input_tokens_seen": 4840808,
      "step": 7430
    },
    {
      "epoch": 3.896750524109015,
      "grad_norm": 0.8334930539131165,
      "learning_rate": 4.864331986305626e-05,
      "loss": 0.4944,
      "num_input_tokens_seen": 4847624,
      "step": 7435
    },
    {
      "epoch": 3.89937106918239,
      "grad_norm": 0.8653190732002258,
      "learning_rate": 4.8639601872494775e-05,
      "loss": 0.4826,
      "num_input_tokens_seen": 4850504,
      "step": 7440
    },
    {
      "epoch": 3.901991614255765,
      "grad_norm": 0.817204475402832,
      "learning_rate": 4.8635878936798287e-05,
      "loss": 0.3235,
      "num_input_tokens_seen": 4853448,
      "step": 7445
    },
    {
      "epoch": 3.9046121593291403,
      "grad_norm": 0.8675603270530701,
      "learning_rate": 4.863215105674557e-05,
      "loss": 0.4077,
      "num_input_tokens_seen": 4856040,
      "step": 7450
    },
    {
      "epoch": 3.9072327044025155,
      "grad_norm": 0.6766879558563232,
      "learning_rate": 4.8628418233116457e-05,
      "loss": 0.4361,
      "num_input_tokens_seen": 4859976,
      "step": 7455
    },
    {
      "epoch": 3.909853249475891,
      "grad_norm": 0.8153372406959534,
      "learning_rate": 4.8624680466691824e-05,
      "loss": 0.3918,
      "num_input_tokens_seen": 4863304,
      "step": 7460
    },
    {
      "epoch": 3.9124737945492662,
      "grad_norm": 1.339172601699829,
      "learning_rate": 4.862093775825356e-05,
      "loss": 0.505,
      "num_input_tokens_seen": 4866568,
      "step": 7465
    },
    {
      "epoch": 3.9150943396226414,
      "grad_norm": 1.0416719913482666,
      "learning_rate": 4.86171901085846e-05,
      "loss": 0.4544,
      "num_input_tokens_seen": 4870312,
      "step": 7470
    },
    {
      "epoch": 3.917714884696017,
      "grad_norm": 0.9289392828941345,
      "learning_rate": 4.861343751846891e-05,
      "loss": 0.5531,
      "num_input_tokens_seen": 4873864,
      "step": 7475
    },
    {
      "epoch": 3.920335429769392,
      "grad_norm": 0.6406198740005493,
      "learning_rate": 4.860967998869148e-05,
      "loss": 0.479,
      "num_input_tokens_seen": 4877032,
      "step": 7480
    },
    {
      "epoch": 3.9229559748427674,
      "grad_norm": 0.759374737739563,
      "learning_rate": 4.860591752003836e-05,
      "loss": 0.4252,
      "num_input_tokens_seen": 4880136,
      "step": 7485
    },
    {
      "epoch": 3.9255765199161425,
      "grad_norm": 0.9786805510520935,
      "learning_rate": 4.8602150113296596e-05,
      "loss": 0.3979,
      "num_input_tokens_seen": 4883720,
      "step": 7490
    },
    {
      "epoch": 3.9281970649895177,
      "grad_norm": 1.5820579528808594,
      "learning_rate": 4.8598377769254305e-05,
      "loss": 0.507,
      "num_input_tokens_seen": 4886440,
      "step": 7495
    },
    {
      "epoch": 3.930817610062893,
      "grad_norm": 1.0557595491409302,
      "learning_rate": 4.85946004887006e-05,
      "loss": 0.4179,
      "num_input_tokens_seen": 4888936,
      "step": 7500
    },
    {
      "epoch": 3.9334381551362685,
      "grad_norm": 4.490665435791016,
      "learning_rate": 4.859081827242566e-05,
      "loss": 0.4317,
      "num_input_tokens_seen": 4890952,
      "step": 7505
    },
    {
      "epoch": 3.9360587002096437,
      "grad_norm": 1.444970965385437,
      "learning_rate": 4.858703112122068e-05,
      "loss": 0.4762,
      "num_input_tokens_seen": 4893736,
      "step": 7510
    },
    {
      "epoch": 3.938679245283019,
      "grad_norm": 0.8661037683486938,
      "learning_rate": 4.8583239035877876e-05,
      "loss": 0.4715,
      "num_input_tokens_seen": 4897384,
      "step": 7515
    },
    {
      "epoch": 3.941299790356394,
      "grad_norm": 1.313609004020691,
      "learning_rate": 4.8579442017190526e-05,
      "loss": 0.4165,
      "num_input_tokens_seen": 4900808,
      "step": 7520
    },
    {
      "epoch": 3.9439203354297696,
      "grad_norm": 1.521091103553772,
      "learning_rate": 4.8575640065952905e-05,
      "loss": 0.7028,
      "num_input_tokens_seen": 4903592,
      "step": 7525
    },
    {
      "epoch": 3.9465408805031448,
      "grad_norm": 1.4048899412155151,
      "learning_rate": 4.857183318296035e-05,
      "loss": 0.4185,
      "num_input_tokens_seen": 4907048,
      "step": 7530
    },
    {
      "epoch": 3.94916142557652,
      "grad_norm": 0.9241065979003906,
      "learning_rate": 4.856802136900922e-05,
      "loss": 0.3937,
      "num_input_tokens_seen": 4909960,
      "step": 7535
    },
    {
      "epoch": 3.951781970649895,
      "grad_norm": 0.9400728344917297,
      "learning_rate": 4.856420462489689e-05,
      "loss": 0.3057,
      "num_input_tokens_seen": 4913768,
      "step": 7540
    },
    {
      "epoch": 3.9544025157232703,
      "grad_norm": 1.3109662532806396,
      "learning_rate": 4.856038295142179e-05,
      "loss": 0.4654,
      "num_input_tokens_seen": 4917416,
      "step": 7545
    },
    {
      "epoch": 3.9570230607966455,
      "grad_norm": 0.7647598385810852,
      "learning_rate": 4.855655634938337e-05,
      "loss": 0.3188,
      "num_input_tokens_seen": 4922056,
      "step": 7550
    },
    {
      "epoch": 3.959643605870021,
      "grad_norm": 1.5252599716186523,
      "learning_rate": 4.8552724819582106e-05,
      "loss": 0.411,
      "num_input_tokens_seen": 4925320,
      "step": 7555
    },
    {
      "epoch": 3.9622641509433962,
      "grad_norm": 2.0356266498565674,
      "learning_rate": 4.8548888362819515e-05,
      "loss": 0.5983,
      "num_input_tokens_seen": 4928168,
      "step": 7560
    },
    {
      "epoch": 3.9648846960167714,
      "grad_norm": 0.9542834758758545,
      "learning_rate": 4.854504697989814e-05,
      "loss": 0.3844,
      "num_input_tokens_seen": 4931176,
      "step": 7565
    },
    {
      "epoch": 3.967505241090147,
      "grad_norm": 0.7942293286323547,
      "learning_rate": 4.8541200671621554e-05,
      "loss": 0.3499,
      "num_input_tokens_seen": 4935304,
      "step": 7570
    },
    {
      "epoch": 3.970125786163522,
      "grad_norm": 0.9961225986480713,
      "learning_rate": 4.8537349438794354e-05,
      "loss": 0.4541,
      "num_input_tokens_seen": 4938600,
      "step": 7575
    },
    {
      "epoch": 3.9727463312368974,
      "grad_norm": 1.1141470670700073,
      "learning_rate": 4.853349328222219e-05,
      "loss": 0.346,
      "num_input_tokens_seen": 4941512,
      "step": 7580
    },
    {
      "epoch": 3.9753668763102725,
      "grad_norm": 0.6628134846687317,
      "learning_rate": 4.8529632202711706e-05,
      "loss": 0.4036,
      "num_input_tokens_seen": 4944744,
      "step": 7585
    },
    {
      "epoch": 3.9779874213836477,
      "grad_norm": 2.3975608348846436,
      "learning_rate": 4.852576620107061e-05,
      "loss": 0.3904,
      "num_input_tokens_seen": 4948072,
      "step": 7590
    },
    {
      "epoch": 3.980607966457023,
      "grad_norm": 1.0637272596359253,
      "learning_rate": 4.8521895278107615e-05,
      "loss": 0.4869,
      "num_input_tokens_seen": 4951656,
      "step": 7595
    },
    {
      "epoch": 3.9832285115303985,
      "grad_norm": 0.328180730342865,
      "learning_rate": 4.8518019434632484e-05,
      "loss": 0.2537,
      "num_input_tokens_seen": 4955400,
      "step": 7600
    },
    {
      "epoch": 3.9858490566037736,
      "grad_norm": 0.611456036567688,
      "learning_rate": 4.8514138671455995e-05,
      "loss": 0.3579,
      "num_input_tokens_seen": 4958568,
      "step": 7605
    },
    {
      "epoch": 3.988469601677149,
      "grad_norm": 0.9889593720436096,
      "learning_rate": 4.851025298938996e-05,
      "loss": 0.4127,
      "num_input_tokens_seen": 4961608,
      "step": 7610
    },
    {
      "epoch": 3.991090146750524,
      "grad_norm": 1.7451515197753906,
      "learning_rate": 4.850636238924722e-05,
      "loss": 0.4076,
      "num_input_tokens_seen": 4965288,
      "step": 7615
    },
    {
      "epoch": 3.9937106918238996,
      "grad_norm": 0.7453892230987549,
      "learning_rate": 4.850246687184163e-05,
      "loss": 0.352,
      "num_input_tokens_seen": 4968424,
      "step": 7620
    },
    {
      "epoch": 3.9963312368972748,
      "grad_norm": 0.8471987247467041,
      "learning_rate": 4.849856643798812e-05,
      "loss": 0.4357,
      "num_input_tokens_seen": 4972200,
      "step": 7625
    },
    {
      "epoch": 3.99895178197065,
      "grad_norm": 0.9382302761077881,
      "learning_rate": 4.849466108850259e-05,
      "loss": 0.4428,
      "num_input_tokens_seen": 4975560,
      "step": 7630
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.449357271194458,
      "eval_runtime": 11.7742,
      "eval_samples_per_second": 72.022,
      "eval_steps_per_second": 18.005,
      "num_input_tokens_seen": 4976128,
      "step": 7632
    },
    {
      "epoch": 4.001572327044025,
      "grad_norm": 0.5730118155479431,
      "learning_rate": 4.8490750824202e-05,
      "loss": 0.4016,
      "num_input_tokens_seen": 4977984,
      "step": 7635
    },
    {
      "epoch": 4.0041928721174,
      "grad_norm": 0.8346168398857117,
      "learning_rate": 4.8486835645904324e-05,
      "loss": 0.2949,
      "num_input_tokens_seen": 4980896,
      "step": 7640
    },
    {
      "epoch": 4.006813417190775,
      "grad_norm": 0.9339603185653687,
      "learning_rate": 4.84829155544286e-05,
      "loss": 0.3009,
      "num_input_tokens_seen": 4984512,
      "step": 7645
    },
    {
      "epoch": 4.009433962264151,
      "grad_norm": 0.9111030101776123,
      "learning_rate": 4.847899055059484e-05,
      "loss": 0.3271,
      "num_input_tokens_seen": 4988320,
      "step": 7650
    },
    {
      "epoch": 4.012054507337526,
      "grad_norm": 0.7804256081581116,
      "learning_rate": 4.847506063522412e-05,
      "loss": 0.4973,
      "num_input_tokens_seen": 4991392,
      "step": 7655
    },
    {
      "epoch": 4.014675052410902,
      "grad_norm": 0.858859658241272,
      "learning_rate": 4.847112580913854e-05,
      "loss": 0.3902,
      "num_input_tokens_seen": 4995680,
      "step": 7660
    },
    {
      "epoch": 4.017295597484277,
      "grad_norm": 0.9101559519767761,
      "learning_rate": 4.846718607316121e-05,
      "loss": 0.465,
      "num_input_tokens_seen": 4999424,
      "step": 7665
    },
    {
      "epoch": 4.019916142557652,
      "grad_norm": 1.5234929323196411,
      "learning_rate": 4.846324142811628e-05,
      "loss": 0.3631,
      "num_input_tokens_seen": 5002272,
      "step": 7670
    },
    {
      "epoch": 4.022536687631027,
      "grad_norm": 0.788788378238678,
      "learning_rate": 4.8459291874828924e-05,
      "loss": 0.3091,
      "num_input_tokens_seen": 5005888,
      "step": 7675
    },
    {
      "epoch": 4.0251572327044025,
      "grad_norm": 0.5986142158508301,
      "learning_rate": 4.845533741412534e-05,
      "loss": 0.262,
      "num_input_tokens_seen": 5009600,
      "step": 7680
    },
    {
      "epoch": 4.027777777777778,
      "grad_norm": 1.0792289972305298,
      "learning_rate": 4.845137804683277e-05,
      "loss": 0.3075,
      "num_input_tokens_seen": 5013568,
      "step": 7685
    },
    {
      "epoch": 4.030398322851153,
      "grad_norm": 1.0693577527999878,
      "learning_rate": 4.844741377377946e-05,
      "loss": 0.3937,
      "num_input_tokens_seen": 5016800,
      "step": 7690
    },
    {
      "epoch": 4.033018867924528,
      "grad_norm": 1.2411524057388306,
      "learning_rate": 4.844344459579469e-05,
      "loss": 0.3707,
      "num_input_tokens_seen": 5020032,
      "step": 7695
    },
    {
      "epoch": 4.035639412997903,
      "grad_norm": 0.9627851843833923,
      "learning_rate": 4.8439470513708766e-05,
      "loss": 0.3564,
      "num_input_tokens_seen": 5023168,
      "step": 7700
    },
    {
      "epoch": 4.038259958071279,
      "grad_norm": 0.9308719635009766,
      "learning_rate": 4.8435491528353026e-05,
      "loss": 0.294,
      "num_input_tokens_seen": 5027040,
      "step": 7705
    },
    {
      "epoch": 4.040880503144654,
      "grad_norm": 1.6037845611572266,
      "learning_rate": 4.843150764055982e-05,
      "loss": 0.4697,
      "num_input_tokens_seen": 5030112,
      "step": 7710
    },
    {
      "epoch": 4.04350104821803,
      "grad_norm": 1.295727252960205,
      "learning_rate": 4.842751885116254e-05,
      "loss": 0.3042,
      "num_input_tokens_seen": 5033600,
      "step": 7715
    },
    {
      "epoch": 4.046121593291405,
      "grad_norm": 0.9542921185493469,
      "learning_rate": 4.842352516099559e-05,
      "loss": 0.2028,
      "num_input_tokens_seen": 5036416,
      "step": 7720
    },
    {
      "epoch": 4.04874213836478,
      "grad_norm": 1.105029821395874,
      "learning_rate": 4.84195265708944e-05,
      "loss": 0.4199,
      "num_input_tokens_seen": 5039552,
      "step": 7725
    },
    {
      "epoch": 4.051362683438155,
      "grad_norm": 0.9341323971748352,
      "learning_rate": 4.841552308169545e-05,
      "loss": 0.3892,
      "num_input_tokens_seen": 5044032,
      "step": 7730
    },
    {
      "epoch": 4.05398322851153,
      "grad_norm": 1.0147525072097778,
      "learning_rate": 4.841151469423619e-05,
      "loss": 0.284,
      "num_input_tokens_seen": 5047232,
      "step": 7735
    },
    {
      "epoch": 4.056603773584905,
      "grad_norm": 1.011113166809082,
      "learning_rate": 4.840750140935516e-05,
      "loss": 0.3951,
      "num_input_tokens_seen": 5050624,
      "step": 7740
    },
    {
      "epoch": 4.059224318658281,
      "grad_norm": 2.0226891040802,
      "learning_rate": 4.840348322789189e-05,
      "loss": 0.3549,
      "num_input_tokens_seen": 5055008,
      "step": 7745
    },
    {
      "epoch": 4.061844863731656,
      "grad_norm": 1.1756411790847778,
      "learning_rate": 4.839946015068692e-05,
      "loss": 0.3397,
      "num_input_tokens_seen": 5058240,
      "step": 7750
    },
    {
      "epoch": 4.064465408805032,
      "grad_norm": 1.0956069231033325,
      "learning_rate": 4.839543217858184e-05,
      "loss": 0.2394,
      "num_input_tokens_seen": 5060864,
      "step": 7755
    },
    {
      "epoch": 4.067085953878407,
      "grad_norm": 1.245397686958313,
      "learning_rate": 4.839139931241926e-05,
      "loss": 0.4289,
      "num_input_tokens_seen": 5064384,
      "step": 7760
    },
    {
      "epoch": 4.069706498951782,
      "grad_norm": 0.9443243145942688,
      "learning_rate": 4.838736155304281e-05,
      "loss": 0.3299,
      "num_input_tokens_seen": 5068096,
      "step": 7765
    },
    {
      "epoch": 4.072327044025157,
      "grad_norm": 1.193755030632019,
      "learning_rate": 4.8383318901297135e-05,
      "loss": 0.4132,
      "num_input_tokens_seen": 5070944,
      "step": 7770
    },
    {
      "epoch": 4.0749475890985325,
      "grad_norm": 1.012054443359375,
      "learning_rate": 4.837927135802792e-05,
      "loss": 0.3587,
      "num_input_tokens_seen": 5074624,
      "step": 7775
    },
    {
      "epoch": 4.077568134171908,
      "grad_norm": 1.82693612575531,
      "learning_rate": 4.8375218924081856e-05,
      "loss": 0.4493,
      "num_input_tokens_seen": 5077216,
      "step": 7780
    },
    {
      "epoch": 4.080188679245283,
      "grad_norm": 1.4093406200408936,
      "learning_rate": 4.837116160030668e-05,
      "loss": 0.3992,
      "num_input_tokens_seen": 5079872,
      "step": 7785
    },
    {
      "epoch": 4.082809224318658,
      "grad_norm": 1.9681963920593262,
      "learning_rate": 4.8367099387551126e-05,
      "loss": 0.4144,
      "num_input_tokens_seen": 5082592,
      "step": 7790
    },
    {
      "epoch": 4.085429769392033,
      "grad_norm": 0.7948949337005615,
      "learning_rate": 4.8363032286664964e-05,
      "loss": 0.3245,
      "num_input_tokens_seen": 5086144,
      "step": 7795
    },
    {
      "epoch": 4.088050314465409,
      "grad_norm": 0.890657365322113,
      "learning_rate": 4.835896029849899e-05,
      "loss": 0.2877,
      "num_input_tokens_seen": 5089696,
      "step": 7800
    },
    {
      "epoch": 4.090670859538784,
      "grad_norm": 0.9735006093978882,
      "learning_rate": 4.835488342390502e-05,
      "loss": 0.3757,
      "num_input_tokens_seen": 5092672,
      "step": 7805
    },
    {
      "epoch": 4.09329140461216,
      "grad_norm": 0.8100546002388,
      "learning_rate": 4.835080166373587e-05,
      "loss": 0.3696,
      "num_input_tokens_seen": 5094848,
      "step": 7810
    },
    {
      "epoch": 4.095911949685535,
      "grad_norm": 0.9957624077796936,
      "learning_rate": 4.834671501884542e-05,
      "loss": 0.2687,
      "num_input_tokens_seen": 5097408,
      "step": 7815
    },
    {
      "epoch": 4.09853249475891,
      "grad_norm": 1.2325752973556519,
      "learning_rate": 4.8342623490088543e-05,
      "loss": 0.349,
      "num_input_tokens_seen": 5099872,
      "step": 7820
    },
    {
      "epoch": 4.101153039832285,
      "grad_norm": 1.3452959060668945,
      "learning_rate": 4.833852707832114e-05,
      "loss": 0.4247,
      "num_input_tokens_seen": 5103392,
      "step": 7825
    },
    {
      "epoch": 4.10377358490566,
      "grad_norm": 1.7404333353042603,
      "learning_rate": 4.833442578440013e-05,
      "loss": 0.3411,
      "num_input_tokens_seen": 5105696,
      "step": 7830
    },
    {
      "epoch": 4.106394129979035,
      "grad_norm": 2.0780794620513916,
      "learning_rate": 4.833031960918346e-05,
      "loss": 0.4329,
      "num_input_tokens_seen": 5109312,
      "step": 7835
    },
    {
      "epoch": 4.109014675052411,
      "grad_norm": 2.4417691230773926,
      "learning_rate": 4.8326208553530095e-05,
      "loss": 0.4568,
      "num_input_tokens_seen": 5111840,
      "step": 7840
    },
    {
      "epoch": 4.111635220125786,
      "grad_norm": 1.0295507907867432,
      "learning_rate": 4.832209261830002e-05,
      "loss": 0.4483,
      "num_input_tokens_seen": 5114752,
      "step": 7845
    },
    {
      "epoch": 4.114255765199162,
      "grad_norm": 2.1418850421905518,
      "learning_rate": 4.831797180435423e-05,
      "loss": 0.4323,
      "num_input_tokens_seen": 5117760,
      "step": 7850
    },
    {
      "epoch": 4.116876310272537,
      "grad_norm": 0.7287174463272095,
      "learning_rate": 4.831384611255478e-05,
      "loss": 0.4172,
      "num_input_tokens_seen": 5122176,
      "step": 7855
    },
    {
      "epoch": 4.119496855345912,
      "grad_norm": 1.0003873109817505,
      "learning_rate": 4.8309715543764697e-05,
      "loss": 0.6657,
      "num_input_tokens_seen": 5124256,
      "step": 7860
    },
    {
      "epoch": 4.122117400419287,
      "grad_norm": 2.0253818035125732,
      "learning_rate": 4.8305580098848046e-05,
      "loss": 0.4133,
      "num_input_tokens_seen": 5126912,
      "step": 7865
    },
    {
      "epoch": 4.1247379454926625,
      "grad_norm": 0.9522197842597961,
      "learning_rate": 4.830143977866993e-05,
      "loss": 0.4173,
      "num_input_tokens_seen": 5130144,
      "step": 7870
    },
    {
      "epoch": 4.127358490566038,
      "grad_norm": 1.2196166515350342,
      "learning_rate": 4.829729458409645e-05,
      "loss": 0.3525,
      "num_input_tokens_seen": 5133440,
      "step": 7875
    },
    {
      "epoch": 4.129979035639413,
      "grad_norm": 1.3418511152267456,
      "learning_rate": 4.8293144515994725e-05,
      "loss": 0.429,
      "num_input_tokens_seen": 5140064,
      "step": 7880
    },
    {
      "epoch": 4.132599580712788,
      "grad_norm": 0.8837196826934814,
      "learning_rate": 4.828898957523292e-05,
      "loss": 0.3736,
      "num_input_tokens_seen": 5142720,
      "step": 7885
    },
    {
      "epoch": 4.135220125786163,
      "grad_norm": 0.9924672842025757,
      "learning_rate": 4.8284829762680183e-05,
      "loss": 0.3534,
      "num_input_tokens_seen": 5145632,
      "step": 7890
    },
    {
      "epoch": 4.137840670859539,
      "grad_norm": 1.2397042512893677,
      "learning_rate": 4.828066507920671e-05,
      "loss": 0.3259,
      "num_input_tokens_seen": 5148224,
      "step": 7895
    },
    {
      "epoch": 4.140461215932914,
      "grad_norm": 1.3328347206115723,
      "learning_rate": 4.82764955256837e-05,
      "loss": 0.4386,
      "num_input_tokens_seen": 5152096,
      "step": 7900
    },
    {
      "epoch": 4.1430817610062896,
      "grad_norm": 0.655504047870636,
      "learning_rate": 4.827232110298338e-05,
      "loss": 0.4328,
      "num_input_tokens_seen": 5155968,
      "step": 7905
    },
    {
      "epoch": 4.145702306079665,
      "grad_norm": 1.1567009687423706,
      "learning_rate": 4.8268141811978995e-05,
      "loss": 0.3201,
      "num_input_tokens_seen": 5159168,
      "step": 7910
    },
    {
      "epoch": 4.14832285115304,
      "grad_norm": 0.9898234009742737,
      "learning_rate": 4.826395765354479e-05,
      "loss": 0.2831,
      "num_input_tokens_seen": 5162816,
      "step": 7915
    },
    {
      "epoch": 4.150943396226415,
      "grad_norm": 1.6828056573867798,
      "learning_rate": 4.825976862855606e-05,
      "loss": 0.397,
      "num_input_tokens_seen": 5167232,
      "step": 7920
    },
    {
      "epoch": 4.15356394129979,
      "grad_norm": 1.389029622077942,
      "learning_rate": 4.8255574737889094e-05,
      "loss": 0.3876,
      "num_input_tokens_seen": 5170560,
      "step": 7925
    },
    {
      "epoch": 4.156184486373165,
      "grad_norm": 1.968853235244751,
      "learning_rate": 4.82513759824212e-05,
      "loss": 0.3079,
      "num_input_tokens_seen": 5173184,
      "step": 7930
    },
    {
      "epoch": 4.158805031446541,
      "grad_norm": 0.8298845291137695,
      "learning_rate": 4.824717236303072e-05,
      "loss": 0.3825,
      "num_input_tokens_seen": 5176288,
      "step": 7935
    },
    {
      "epoch": 4.161425576519916,
      "grad_norm": 1.1438369750976562,
      "learning_rate": 4.8242963880597005e-05,
      "loss": 0.3356,
      "num_input_tokens_seen": 5181216,
      "step": 7940
    },
    {
      "epoch": 4.164046121593292,
      "grad_norm": 1.5181914567947388,
      "learning_rate": 4.823875053600041e-05,
      "loss": 0.2918,
      "num_input_tokens_seen": 5184256,
      "step": 7945
    },
    {
      "epoch": 4.166666666666667,
      "grad_norm": 0.8578251600265503,
      "learning_rate": 4.823453233012233e-05,
      "loss": 0.2805,
      "num_input_tokens_seen": 5187616,
      "step": 7950
    },
    {
      "epoch": 4.169287211740042,
      "grad_norm": 1.2354592084884644,
      "learning_rate": 4.8230309263845155e-05,
      "loss": 0.5091,
      "num_input_tokens_seen": 5189952,
      "step": 7955
    },
    {
      "epoch": 4.171907756813417,
      "grad_norm": 2.193784713745117,
      "learning_rate": 4.822608133805231e-05,
      "loss": 0.4007,
      "num_input_tokens_seen": 5192608,
      "step": 7960
    },
    {
      "epoch": 4.1745283018867925,
      "grad_norm": 1.0779697895050049,
      "learning_rate": 4.8221848553628226e-05,
      "loss": 0.3193,
      "num_input_tokens_seen": 5196256,
      "step": 7965
    },
    {
      "epoch": 4.177148846960168,
      "grad_norm": 0.9781007766723633,
      "learning_rate": 4.8217610911458346e-05,
      "loss": 0.3374,
      "num_input_tokens_seen": 5199840,
      "step": 7970
    },
    {
      "epoch": 4.179769392033543,
      "grad_norm": 0.8168816566467285,
      "learning_rate": 4.821336841242915e-05,
      "loss": 0.2478,
      "num_input_tokens_seen": 5202688,
      "step": 7975
    },
    {
      "epoch": 4.182389937106918,
      "grad_norm": 1.0874241590499878,
      "learning_rate": 4.820912105742811e-05,
      "loss": 0.4929,
      "num_input_tokens_seen": 5205248,
      "step": 7980
    },
    {
      "epoch": 4.185010482180293,
      "grad_norm": 1.357988715171814,
      "learning_rate": 4.8204868847343736e-05,
      "loss": 0.3016,
      "num_input_tokens_seen": 5207648,
      "step": 7985
    },
    {
      "epoch": 4.187631027253669,
      "grad_norm": 1.24289870262146,
      "learning_rate": 4.820061178306553e-05,
      "loss": 0.3293,
      "num_input_tokens_seen": 5210272,
      "step": 7990
    },
    {
      "epoch": 4.190251572327044,
      "grad_norm": 0.8396263122558594,
      "learning_rate": 4.819634986548403e-05,
      "loss": 0.4488,
      "num_input_tokens_seen": 5214848,
      "step": 7995
    },
    {
      "epoch": 4.1928721174004195,
      "grad_norm": 0.9044924378395081,
      "learning_rate": 4.819208309549077e-05,
      "loss": 0.3471,
      "num_input_tokens_seen": 5217888,
      "step": 8000
    },
    {
      "epoch": 4.195492662473795,
      "grad_norm": 1.192331075668335,
      "learning_rate": 4.8187811473978316e-05,
      "loss": 0.4189,
      "num_input_tokens_seen": 5221952,
      "step": 8005
    },
    {
      "epoch": 4.19811320754717,
      "grad_norm": 1.7018119096755981,
      "learning_rate": 4.818353500184025e-05,
      "loss": 0.4566,
      "num_input_tokens_seen": 5225120,
      "step": 8010
    },
    {
      "epoch": 4.200733752620545,
      "grad_norm": 1.8907630443572998,
      "learning_rate": 4.817925367997115e-05,
      "loss": 0.4073,
      "num_input_tokens_seen": 5228448,
      "step": 8015
    },
    {
      "epoch": 4.20335429769392,
      "grad_norm": 1.6287211179733276,
      "learning_rate": 4.8174967509266615e-05,
      "loss": 0.4213,
      "num_input_tokens_seen": 5230976,
      "step": 8020
    },
    {
      "epoch": 4.205974842767295,
      "grad_norm": 1.0492326021194458,
      "learning_rate": 4.817067649062328e-05,
      "loss": 0.4323,
      "num_input_tokens_seen": 5234112,
      "step": 8025
    },
    {
      "epoch": 4.2085953878406706,
      "grad_norm": 0.9897317290306091,
      "learning_rate": 4.816638062493876e-05,
      "loss": 0.4693,
      "num_input_tokens_seen": 5238592,
      "step": 8030
    },
    {
      "epoch": 4.211215932914046,
      "grad_norm": 0.7404037714004517,
      "learning_rate": 4.8162079913111716e-05,
      "loss": 0.3547,
      "num_input_tokens_seen": 5241568,
      "step": 8035
    },
    {
      "epoch": 4.213836477987422,
      "grad_norm": 1.2093251943588257,
      "learning_rate": 4.81577743560418e-05,
      "loss": 0.3145,
      "num_input_tokens_seen": 5244704,
      "step": 8040
    },
    {
      "epoch": 4.216457023060797,
      "grad_norm": 0.8480321168899536,
      "learning_rate": 4.8153463954629684e-05,
      "loss": 0.3072,
      "num_input_tokens_seen": 5248000,
      "step": 8045
    },
    {
      "epoch": 4.219077568134172,
      "grad_norm": 2.245274782180786,
      "learning_rate": 4.814914870977706e-05,
      "loss": 0.3076,
      "num_input_tokens_seen": 5252096,
      "step": 8050
    },
    {
      "epoch": 4.221698113207547,
      "grad_norm": 1.367749810218811,
      "learning_rate": 4.8144828622386627e-05,
      "loss": 0.4247,
      "num_input_tokens_seen": 5255360,
      "step": 8055
    },
    {
      "epoch": 4.2243186582809225,
      "grad_norm": 0.6556640267372131,
      "learning_rate": 4.8140503693362095e-05,
      "loss": 0.4836,
      "num_input_tokens_seen": 5259712,
      "step": 8060
    },
    {
      "epoch": 4.226939203354298,
      "grad_norm": 1.8777360916137695,
      "learning_rate": 4.813617392360818e-05,
      "loss": 0.3457,
      "num_input_tokens_seen": 5262912,
      "step": 8065
    },
    {
      "epoch": 4.229559748427673,
      "grad_norm": 1.252976894378662,
      "learning_rate": 4.813183931403065e-05,
      "loss": 0.3399,
      "num_input_tokens_seen": 5265664,
      "step": 8070
    },
    {
      "epoch": 4.232180293501048,
      "grad_norm": 1.0824438333511353,
      "learning_rate": 4.8127499865536226e-05,
      "loss": 0.2909,
      "num_input_tokens_seen": 5268928,
      "step": 8075
    },
    {
      "epoch": 4.234800838574423,
      "grad_norm": 1.0833334922790527,
      "learning_rate": 4.812315557903269e-05,
      "loss": 0.3685,
      "num_input_tokens_seen": 5272032,
      "step": 8080
    },
    {
      "epoch": 4.237421383647799,
      "grad_norm": 1.1651922464370728,
      "learning_rate": 4.8118806455428806e-05,
      "loss": 0.3288,
      "num_input_tokens_seen": 5275520,
      "step": 8085
    },
    {
      "epoch": 4.240041928721174,
      "grad_norm": 1.627028465270996,
      "learning_rate": 4.811445249563437e-05,
      "loss": 0.389,
      "num_input_tokens_seen": 5278528,
      "step": 8090
    },
    {
      "epoch": 4.2426624737945495,
      "grad_norm": 1.6361976861953735,
      "learning_rate": 4.811009370056017e-05,
      "loss": 0.4953,
      "num_input_tokens_seen": 5281536,
      "step": 8095
    },
    {
      "epoch": 4.245283018867925,
      "grad_norm": 0.9488562941551208,
      "learning_rate": 4.8105730071118016e-05,
      "loss": 0.357,
      "num_input_tokens_seen": 5287168,
      "step": 8100
    },
    {
      "epoch": 4.2479035639413,
      "grad_norm": 1.0645561218261719,
      "learning_rate": 4.810136160822074e-05,
      "loss": 0.3182,
      "num_input_tokens_seen": 5291008,
      "step": 8105
    },
    {
      "epoch": 4.250524109014675,
      "grad_norm": 1.2151612043380737,
      "learning_rate": 4.8096988312782174e-05,
      "loss": 0.2833,
      "num_input_tokens_seen": 5294144,
      "step": 8110
    },
    {
      "epoch": 4.25314465408805,
      "grad_norm": 2.6613221168518066,
      "learning_rate": 4.809261018571715e-05,
      "loss": 0.4831,
      "num_input_tokens_seen": 5297184,
      "step": 8115
    },
    {
      "epoch": 4.255765199161425,
      "grad_norm": 0.9963695406913757,
      "learning_rate": 4.808822722794153e-05,
      "loss": 0.3393,
      "num_input_tokens_seen": 5300480,
      "step": 8120
    },
    {
      "epoch": 4.2583857442348005,
      "grad_norm": 0.8165751695632935,
      "learning_rate": 4.8083839440372194e-05,
      "loss": 0.354,
      "num_input_tokens_seen": 5303936,
      "step": 8125
    },
    {
      "epoch": 4.261006289308176,
      "grad_norm": 0.4941924810409546,
      "learning_rate": 4.8079446823926984e-05,
      "loss": 0.3205,
      "num_input_tokens_seen": 5306816,
      "step": 8130
    },
    {
      "epoch": 4.263626834381552,
      "grad_norm": 1.3726789951324463,
      "learning_rate": 4.8075049379524805e-05,
      "loss": 0.3084,
      "num_input_tokens_seen": 5309792,
      "step": 8135
    },
    {
      "epoch": 4.266247379454927,
      "grad_norm": 1.0622731447219849,
      "learning_rate": 4.807064710808555e-05,
      "loss": 0.4866,
      "num_input_tokens_seen": 5313056,
      "step": 8140
    },
    {
      "epoch": 4.268867924528302,
      "grad_norm": 0.8768707513809204,
      "learning_rate": 4.806624001053013e-05,
      "loss": 0.2785,
      "num_input_tokens_seen": 5315488,
      "step": 8145
    },
    {
      "epoch": 4.271488469601677,
      "grad_norm": 1.1485509872436523,
      "learning_rate": 4.8061828087780455e-05,
      "loss": 0.3047,
      "num_input_tokens_seen": 5318272,
      "step": 8150
    },
    {
      "epoch": 4.274109014675052,
      "grad_norm": 1.1550036668777466,
      "learning_rate": 4.805741134075944e-05,
      "loss": 0.3913,
      "num_input_tokens_seen": 5321408,
      "step": 8155
    },
    {
      "epoch": 4.276729559748428,
      "grad_norm": 0.8378115892410278,
      "learning_rate": 4.8052989770391026e-05,
      "loss": 0.3289,
      "num_input_tokens_seen": 5325696,
      "step": 8160
    },
    {
      "epoch": 4.279350104821803,
      "grad_norm": 2.257681369781494,
      "learning_rate": 4.804856337760016e-05,
      "loss": 0.3817,
      "num_input_tokens_seen": 5328832,
      "step": 8165
    },
    {
      "epoch": 4.281970649895178,
      "grad_norm": 1.23528254032135,
      "learning_rate": 4.804413216331278e-05,
      "loss": 0.3932,
      "num_input_tokens_seen": 5332448,
      "step": 8170
    },
    {
      "epoch": 4.284591194968553,
      "grad_norm": 0.9620116353034973,
      "learning_rate": 4.8039696128455856e-05,
      "loss": 0.4121,
      "num_input_tokens_seen": 5335520,
      "step": 8175
    },
    {
      "epoch": 4.287211740041929,
      "grad_norm": 1.3486989736557007,
      "learning_rate": 4.803525527395736e-05,
      "loss": 0.2828,
      "num_input_tokens_seen": 5339328,
      "step": 8180
    },
    {
      "epoch": 4.289832285115304,
      "grad_norm": 1.1271761655807495,
      "learning_rate": 4.803080960074625e-05,
      "loss": 0.3838,
      "num_input_tokens_seen": 5342368,
      "step": 8185
    },
    {
      "epoch": 4.2924528301886795,
      "grad_norm": 1.1484956741333008,
      "learning_rate": 4.8026359109752525e-05,
      "loss": 0.4601,
      "num_input_tokens_seen": 5345632,
      "step": 8190
    },
    {
      "epoch": 4.295073375262055,
      "grad_norm": 1.3742469549179077,
      "learning_rate": 4.8021903801907175e-05,
      "loss": 0.484,
      "num_input_tokens_seen": 5348608,
      "step": 8195
    },
    {
      "epoch": 4.29769392033543,
      "grad_norm": 2.438331127166748,
      "learning_rate": 4.801744367814219e-05,
      "loss": 0.2852,
      "num_input_tokens_seen": 5351776,
      "step": 8200
    },
    {
      "epoch": 4.300314465408805,
      "grad_norm": 0.9135802388191223,
      "learning_rate": 4.801297873939059e-05,
      "loss": 0.4895,
      "num_input_tokens_seen": 5354880,
      "step": 8205
    },
    {
      "epoch": 4.30293501048218,
      "grad_norm": 1.704871416091919,
      "learning_rate": 4.8008508986586384e-05,
      "loss": 0.322,
      "num_input_tokens_seen": 5357824,
      "step": 8210
    },
    {
      "epoch": 4.305555555555555,
      "grad_norm": 1.5114399194717407,
      "learning_rate": 4.800403442066459e-05,
      "loss": 0.3579,
      "num_input_tokens_seen": 5360512,
      "step": 8215
    },
    {
      "epoch": 4.3081761006289305,
      "grad_norm": 1.3549792766571045,
      "learning_rate": 4.799955504256125e-05,
      "loss": 0.3114,
      "num_input_tokens_seen": 5364064,
      "step": 8220
    },
    {
      "epoch": 4.310796645702306,
      "grad_norm": 1.1498757600784302,
      "learning_rate": 4.7995070853213376e-05,
      "loss": 0.4156,
      "num_input_tokens_seen": 5367200,
      "step": 8225
    },
    {
      "epoch": 4.313417190775682,
      "grad_norm": 0.8061573505401611,
      "learning_rate": 4.799058185355903e-05,
      "loss": 0.3914,
      "num_input_tokens_seen": 5370272,
      "step": 8230
    },
    {
      "epoch": 4.316037735849057,
      "grad_norm": 1.3487993478775024,
      "learning_rate": 4.7986088044537235e-05,
      "loss": 0.3632,
      "num_input_tokens_seen": 5372896,
      "step": 8235
    },
    {
      "epoch": 4.318658280922432,
      "grad_norm": 1.6483031511306763,
      "learning_rate": 4.798158942708807e-05,
      "loss": 0.4471,
      "num_input_tokens_seen": 5376800,
      "step": 8240
    },
    {
      "epoch": 4.321278825995807,
      "grad_norm": 1.1559029817581177,
      "learning_rate": 4.797708600215258e-05,
      "loss": 0.5084,
      "num_input_tokens_seen": 5380288,
      "step": 8245
    },
    {
      "epoch": 4.323899371069182,
      "grad_norm": 1.8854478597640991,
      "learning_rate": 4.7972577770672845e-05,
      "loss": 0.3107,
      "num_input_tokens_seen": 5383072,
      "step": 8250
    },
    {
      "epoch": 4.326519916142558,
      "grad_norm": 1.3483561277389526,
      "learning_rate": 4.796806473359192e-05,
      "loss": 0.4745,
      "num_input_tokens_seen": 5386400,
      "step": 8255
    },
    {
      "epoch": 4.329140461215933,
      "grad_norm": 0.9021615386009216,
      "learning_rate": 4.796354689185389e-05,
      "loss": 0.4677,
      "num_input_tokens_seen": 5388832,
      "step": 8260
    },
    {
      "epoch": 4.331761006289308,
      "grad_norm": 1.7556591033935547,
      "learning_rate": 4.795902424640383e-05,
      "loss": 0.3666,
      "num_input_tokens_seen": 5391968,
      "step": 8265
    },
    {
      "epoch": 4.334381551362683,
      "grad_norm": 1.2085058689117432,
      "learning_rate": 4.795449679818782e-05,
      "loss": 0.3265,
      "num_input_tokens_seen": 5398624,
      "step": 8270
    },
    {
      "epoch": 4.337002096436059,
      "grad_norm": 0.9423232078552246,
      "learning_rate": 4.794996454815297e-05,
      "loss": 0.3871,
      "num_input_tokens_seen": 5401824,
      "step": 8275
    },
    {
      "epoch": 4.339622641509434,
      "grad_norm": 1.4261900186538696,
      "learning_rate": 4.7945427497247355e-05,
      "loss": 0.3078,
      "num_input_tokens_seen": 5404928,
      "step": 8280
    },
    {
      "epoch": 4.3422431865828095,
      "grad_norm": 0.8871475458145142,
      "learning_rate": 4.7940885646420084e-05,
      "loss": 0.3448,
      "num_input_tokens_seen": 5408320,
      "step": 8285
    },
    {
      "epoch": 4.344863731656185,
      "grad_norm": 1.3243836164474487,
      "learning_rate": 4.7936338996621264e-05,
      "loss": 0.4546,
      "num_input_tokens_seen": 5410752,
      "step": 8290
    },
    {
      "epoch": 4.34748427672956,
      "grad_norm": 1.0530179738998413,
      "learning_rate": 4.7931787548801996e-05,
      "loss": 0.3049,
      "num_input_tokens_seen": 5413312,
      "step": 8295
    },
    {
      "epoch": 4.350104821802935,
      "grad_norm": 1.1338564157485962,
      "learning_rate": 4.7927231303914394e-05,
      "loss": 0.4086,
      "num_input_tokens_seen": 5416992,
      "step": 8300
    },
    {
      "epoch": 4.35272536687631,
      "grad_norm": 0.8710578680038452,
      "learning_rate": 4.7922670262911565e-05,
      "loss": 0.2721,
      "num_input_tokens_seen": 5420672,
      "step": 8305
    },
    {
      "epoch": 4.355345911949685,
      "grad_norm": 1.5496972799301147,
      "learning_rate": 4.791810442674764e-05,
      "loss": 0.3064,
      "num_input_tokens_seen": 5423360,
      "step": 8310
    },
    {
      "epoch": 4.3579664570230605,
      "grad_norm": 1.129509687423706,
      "learning_rate": 4.791353379637773e-05,
      "loss": 0.3436,
      "num_input_tokens_seen": 5426336,
      "step": 8315
    },
    {
      "epoch": 4.360587002096436,
      "grad_norm": 1.8055766820907593,
      "learning_rate": 4.790895837275796e-05,
      "loss": 0.405,
      "num_input_tokens_seen": 5429632,
      "step": 8320
    },
    {
      "epoch": 4.363207547169811,
      "grad_norm": 1.0227378606796265,
      "learning_rate": 4.790437815684545e-05,
      "loss": 0.3448,
      "num_input_tokens_seen": 5433312,
      "step": 8325
    },
    {
      "epoch": 4.365828092243187,
      "grad_norm": 0.8805586695671082,
      "learning_rate": 4.789979314959835e-05,
      "loss": 0.4135,
      "num_input_tokens_seen": 5437248,
      "step": 8330
    },
    {
      "epoch": 4.368448637316562,
      "grad_norm": 1.3085105419158936,
      "learning_rate": 4.789520335197577e-05,
      "loss": 0.2836,
      "num_input_tokens_seen": 5440384,
      "step": 8335
    },
    {
      "epoch": 4.371069182389937,
      "grad_norm": 1.1774959564208984,
      "learning_rate": 4.789060876493785e-05,
      "loss": 0.4371,
      "num_input_tokens_seen": 5444000,
      "step": 8340
    },
    {
      "epoch": 4.373689727463312,
      "grad_norm": 1.4571043252944946,
      "learning_rate": 4.788600938944573e-05,
      "loss": 0.3826,
      "num_input_tokens_seen": 5447936,
      "step": 8345
    },
    {
      "epoch": 4.376310272536688,
      "grad_norm": 1.5362166166305542,
      "learning_rate": 4.788140522646155e-05,
      "loss": 0.4088,
      "num_input_tokens_seen": 5451488,
      "step": 8350
    },
    {
      "epoch": 4.378930817610063,
      "grad_norm": 1.1019164323806763,
      "learning_rate": 4.787679627694843e-05,
      "loss": 0.3765,
      "num_input_tokens_seen": 5455520,
      "step": 8355
    },
    {
      "epoch": 4.381551362683438,
      "grad_norm": 1.974021315574646,
      "learning_rate": 4.787218254187053e-05,
      "loss": 0.4728,
      "num_input_tokens_seen": 5458560,
      "step": 8360
    },
    {
      "epoch": 4.384171907756813,
      "grad_norm": 1.2423348426818848,
      "learning_rate": 4.7867564022192965e-05,
      "loss": 0.3663,
      "num_input_tokens_seen": 5461600,
      "step": 8365
    },
    {
      "epoch": 4.386792452830189,
      "grad_norm": 1.378522515296936,
      "learning_rate": 4.786294071888191e-05,
      "loss": 0.3045,
      "num_input_tokens_seen": 5465536,
      "step": 8370
    },
    {
      "epoch": 4.389412997903564,
      "grad_norm": 1.2814860343933105,
      "learning_rate": 4.785831263290449e-05,
      "loss": 0.3387,
      "num_input_tokens_seen": 5468416,
      "step": 8375
    },
    {
      "epoch": 4.3920335429769395,
      "grad_norm": 1.9284932613372803,
      "learning_rate": 4.7853679765228845e-05,
      "loss": 0.4088,
      "num_input_tokens_seen": 5472000,
      "step": 8380
    },
    {
      "epoch": 4.394654088050315,
      "grad_norm": 0.8994912505149841,
      "learning_rate": 4.7849042116824116e-05,
      "loss": 0.3028,
      "num_input_tokens_seen": 5475168,
      "step": 8385
    },
    {
      "epoch": 4.39727463312369,
      "grad_norm": 1.6328858137130737,
      "learning_rate": 4.7844399688660456e-05,
      "loss": 0.3168,
      "num_input_tokens_seen": 5477760,
      "step": 8390
    },
    {
      "epoch": 4.399895178197065,
      "grad_norm": 0.9904060959815979,
      "learning_rate": 4.783975248170901e-05,
      "loss": 0.308,
      "num_input_tokens_seen": 5480544,
      "step": 8395
    },
    {
      "epoch": 4.40251572327044,
      "grad_norm": 0.8949857354164124,
      "learning_rate": 4.783510049694191e-05,
      "loss": 0.3042,
      "num_input_tokens_seen": 5483776,
      "step": 8400
    },
    {
      "epoch": 4.405136268343815,
      "grad_norm": 2.204758405685425,
      "learning_rate": 4.78304437353323e-05,
      "loss": 0.3227,
      "num_input_tokens_seen": 5485888,
      "step": 8405
    },
    {
      "epoch": 4.4077568134171905,
      "grad_norm": 1.9388161897659302,
      "learning_rate": 4.782578219785433e-05,
      "loss": 0.4395,
      "num_input_tokens_seen": 5489056,
      "step": 8410
    },
    {
      "epoch": 4.410377358490566,
      "grad_norm": 1.0342813730239868,
      "learning_rate": 4.782111588548312e-05,
      "loss": 0.4599,
      "num_input_tokens_seen": 5492224,
      "step": 8415
    },
    {
      "epoch": 4.412997903563941,
      "grad_norm": 1.4935588836669922,
      "learning_rate": 4.781644479919484e-05,
      "loss": 0.3216,
      "num_input_tokens_seen": 5495872,
      "step": 8420
    },
    {
      "epoch": 4.415618448637317,
      "grad_norm": 0.7307144999504089,
      "learning_rate": 4.7811768939966605e-05,
      "loss": 0.2783,
      "num_input_tokens_seen": 5499488,
      "step": 8425
    },
    {
      "epoch": 4.418238993710692,
      "grad_norm": 1.1642063856124878,
      "learning_rate": 4.780708830877656e-05,
      "loss": 0.3863,
      "num_input_tokens_seen": 5503104,
      "step": 8430
    },
    {
      "epoch": 4.420859538784067,
      "grad_norm": 1.5032044649124146,
      "learning_rate": 4.780240290660384e-05,
      "loss": 0.34,
      "num_input_tokens_seen": 5506912,
      "step": 8435
    },
    {
      "epoch": 4.423480083857442,
      "grad_norm": 1.479324460029602,
      "learning_rate": 4.7797712734428575e-05,
      "loss": 0.4167,
      "num_input_tokens_seen": 5509312,
      "step": 8440
    },
    {
      "epoch": 4.426100628930818,
      "grad_norm": 1.6684149503707886,
      "learning_rate": 4.779301779323189e-05,
      "loss": 0.3027,
      "num_input_tokens_seen": 5512096,
      "step": 8445
    },
    {
      "epoch": 4.428721174004193,
      "grad_norm": 2.4914391040802,
      "learning_rate": 4.778831808399593e-05,
      "loss": 0.3265,
      "num_input_tokens_seen": 5515168,
      "step": 8450
    },
    {
      "epoch": 4.431341719077568,
      "grad_norm": 1.1879963874816895,
      "learning_rate": 4.77836136077038e-05,
      "loss": 0.4056,
      "num_input_tokens_seen": 5518464,
      "step": 8455
    },
    {
      "epoch": 4.433962264150943,
      "grad_norm": 0.8119494915008545,
      "learning_rate": 4.777890436533963e-05,
      "loss": 0.3068,
      "num_input_tokens_seen": 5521344,
      "step": 8460
    },
    {
      "epoch": 4.436582809224318,
      "grad_norm": 1.353156566619873,
      "learning_rate": 4.777419035788855e-05,
      "loss": 0.3684,
      "num_input_tokens_seen": 5524416,
      "step": 8465
    },
    {
      "epoch": 4.439203354297694,
      "grad_norm": 1.5599197149276733,
      "learning_rate": 4.776947158633667e-05,
      "loss": 0.441,
      "num_input_tokens_seen": 5527488,
      "step": 8470
    },
    {
      "epoch": 4.4418238993710695,
      "grad_norm": 2.224966049194336,
      "learning_rate": 4.7764748051671095e-05,
      "loss": 0.379,
      "num_input_tokens_seen": 5530144,
      "step": 8475
    },
    {
      "epoch": 4.444444444444445,
      "grad_norm": 1.1039947271347046,
      "learning_rate": 4.7760019754879946e-05,
      "loss": 0.5005,
      "num_input_tokens_seen": 5533536,
      "step": 8480
    },
    {
      "epoch": 4.44706498951782,
      "grad_norm": 1.0149861574172974,
      "learning_rate": 4.7755286696952324e-05,
      "loss": 0.3638,
      "num_input_tokens_seen": 5536032,
      "step": 8485
    },
    {
      "epoch": 4.449685534591195,
      "grad_norm": 0.8605139255523682,
      "learning_rate": 4.775054887887833e-05,
      "loss": 0.3032,
      "num_input_tokens_seen": 5539552,
      "step": 8490
    },
    {
      "epoch": 4.45230607966457,
      "grad_norm": 1.3635141849517822,
      "learning_rate": 4.774580630164906e-05,
      "loss": 0.3994,
      "num_input_tokens_seen": 5542528,
      "step": 8495
    },
    {
      "epoch": 4.454926624737945,
      "grad_norm": 1.6956642866134644,
      "learning_rate": 4.774105896625661e-05,
      "loss": 0.4197,
      "num_input_tokens_seen": 5545408,
      "step": 8500
    },
    {
      "epoch": 4.4575471698113205,
      "grad_norm": 1.1076093912124634,
      "learning_rate": 4.773630687369407e-05,
      "loss": 0.2953,
      "num_input_tokens_seen": 5549120,
      "step": 8505
    },
    {
      "epoch": 4.460167714884696,
      "grad_norm": 1.4194642305374146,
      "learning_rate": 4.773155002495552e-05,
      "loss": 0.4348,
      "num_input_tokens_seen": 5552352,
      "step": 8510
    },
    {
      "epoch": 4.462788259958071,
      "grad_norm": 0.6324037313461304,
      "learning_rate": 4.772678842103605e-05,
      "loss": 0.2995,
      "num_input_tokens_seen": 5555296,
      "step": 8515
    },
    {
      "epoch": 4.465408805031447,
      "grad_norm": 1.1924577951431274,
      "learning_rate": 4.772202206293172e-05,
      "loss": 0.2838,
      "num_input_tokens_seen": 5558720,
      "step": 8520
    },
    {
      "epoch": 4.468029350104822,
      "grad_norm": 1.4415512084960938,
      "learning_rate": 4.7717250951639586e-05,
      "loss": 0.4113,
      "num_input_tokens_seen": 5562176,
      "step": 8525
    },
    {
      "epoch": 4.470649895178197,
      "grad_norm": 0.993500828742981,
      "learning_rate": 4.771247508815774e-05,
      "loss": 0.4967,
      "num_input_tokens_seen": 5565408,
      "step": 8530
    },
    {
      "epoch": 4.473270440251572,
      "grad_norm": 0.7938562035560608,
      "learning_rate": 4.770769447348522e-05,
      "loss": 0.3576,
      "num_input_tokens_seen": 5568288,
      "step": 8535
    },
    {
      "epoch": 4.475890985324948,
      "grad_norm": 0.9447990655899048,
      "learning_rate": 4.7702909108622074e-05,
      "loss": 0.552,
      "num_input_tokens_seen": 5571744,
      "step": 8540
    },
    {
      "epoch": 4.478511530398323,
      "grad_norm": 0.8676931858062744,
      "learning_rate": 4.7698118994569355e-05,
      "loss": 0.4496,
      "num_input_tokens_seen": 5575008,
      "step": 8545
    },
    {
      "epoch": 4.481132075471698,
      "grad_norm": 0.9615928530693054,
      "learning_rate": 4.769332413232909e-05,
      "loss": 0.4814,
      "num_input_tokens_seen": 5577824,
      "step": 8550
    },
    {
      "epoch": 4.483752620545073,
      "grad_norm": 1.1629587411880493,
      "learning_rate": 4.768852452290432e-05,
      "loss": 0.3566,
      "num_input_tokens_seen": 5581216,
      "step": 8555
    },
    {
      "epoch": 4.486373165618448,
      "grad_norm": 1.2566441297531128,
      "learning_rate": 4.768372016729907e-05,
      "loss": 0.3703,
      "num_input_tokens_seen": 5586272,
      "step": 8560
    },
    {
      "epoch": 4.488993710691824,
      "grad_norm": 1.2847070693969727,
      "learning_rate": 4.767891106651834e-05,
      "loss": 0.3599,
      "num_input_tokens_seen": 5588608,
      "step": 8565
    },
    {
      "epoch": 4.4916142557651995,
      "grad_norm": 1.187262773513794,
      "learning_rate": 4.767409722156816e-05,
      "loss": 0.256,
      "num_input_tokens_seen": 5591296,
      "step": 8570
    },
    {
      "epoch": 4.494234800838575,
      "grad_norm": 1.4774585962295532,
      "learning_rate": 4.7669278633455515e-05,
      "loss": 0.3249,
      "num_input_tokens_seen": 5593568,
      "step": 8575
    },
    {
      "epoch": 4.49685534591195,
      "grad_norm": 1.9825531244277954,
      "learning_rate": 4.7664455303188404e-05,
      "loss": 0.3859,
      "num_input_tokens_seen": 5596480,
      "step": 8580
    },
    {
      "epoch": 4.499475890985325,
      "grad_norm": 0.7801294326782227,
      "learning_rate": 4.7659627231775815e-05,
      "loss": 0.4413,
      "num_input_tokens_seen": 5599968,
      "step": 8585
    },
    {
      "epoch": 4.5020964360587,
      "grad_norm": 0.8576105237007141,
      "learning_rate": 4.765479442022773e-05,
      "loss": 0.3811,
      "num_input_tokens_seen": 5603168,
      "step": 8590
    },
    {
      "epoch": 4.504716981132075,
      "grad_norm": 0.9140750765800476,
      "learning_rate": 4.76499568695551e-05,
      "loss": 0.4776,
      "num_input_tokens_seen": 5606464,
      "step": 8595
    },
    {
      "epoch": 4.5073375262054505,
      "grad_norm": 0.9336465001106262,
      "learning_rate": 4.764511458076991e-05,
      "loss": 0.4224,
      "num_input_tokens_seen": 5609216,
      "step": 8600
    },
    {
      "epoch": 4.509958071278826,
      "grad_norm": 1.0709844827651978,
      "learning_rate": 4.7640267554885085e-05,
      "loss": 0.3381,
      "num_input_tokens_seen": 5612800,
      "step": 8605
    },
    {
      "epoch": 4.512578616352201,
      "grad_norm": 1.2258676290512085,
      "learning_rate": 4.763541579291459e-05,
      "loss": 0.3093,
      "num_input_tokens_seen": 5615488,
      "step": 8610
    },
    {
      "epoch": 4.515199161425577,
      "grad_norm": 0.5194418430328369,
      "learning_rate": 4.763055929587336e-05,
      "loss": 0.3172,
      "num_input_tokens_seen": 5619264,
      "step": 8615
    },
    {
      "epoch": 4.517819706498952,
      "grad_norm": 1.7326011657714844,
      "learning_rate": 4.7625698064777294e-05,
      "loss": 0.3112,
      "num_input_tokens_seen": 5623104,
      "step": 8620
    },
    {
      "epoch": 4.520440251572327,
      "grad_norm": 1.5039441585540771,
      "learning_rate": 4.762083210064332e-05,
      "loss": 0.3589,
      "num_input_tokens_seen": 5626144,
      "step": 8625
    },
    {
      "epoch": 4.523060796645702,
      "grad_norm": 0.9488025903701782,
      "learning_rate": 4.761596140448934e-05,
      "loss": 0.383,
      "num_input_tokens_seen": 5629216,
      "step": 8630
    },
    {
      "epoch": 4.5256813417190775,
      "grad_norm": 1.892764925956726,
      "learning_rate": 4.7611085977334256e-05,
      "loss": 0.3936,
      "num_input_tokens_seen": 5632448,
      "step": 8635
    },
    {
      "epoch": 4.528301886792453,
      "grad_norm": 1.2072501182556152,
      "learning_rate": 4.7606205820197946e-05,
      "loss": 0.4205,
      "num_input_tokens_seen": 5636704,
      "step": 8640
    },
    {
      "epoch": 4.530922431865828,
      "grad_norm": 0.8080047965049744,
      "learning_rate": 4.760132093410129e-05,
      "loss": 0.3447,
      "num_input_tokens_seen": 5641408,
      "step": 8645
    },
    {
      "epoch": 4.533542976939203,
      "grad_norm": 1.045549750328064,
      "learning_rate": 4.759643132006614e-05,
      "loss": 0.3252,
      "num_input_tokens_seen": 5644096,
      "step": 8650
    },
    {
      "epoch": 4.536163522012579,
      "grad_norm": 0.9514559507369995,
      "learning_rate": 4.759153697911535e-05,
      "loss": 0.3154,
      "num_input_tokens_seen": 5646720,
      "step": 8655
    },
    {
      "epoch": 4.538784067085954,
      "grad_norm": 1.0145515203475952,
      "learning_rate": 4.758663791227276e-05,
      "loss": 0.4307,
      "num_input_tokens_seen": 5649696,
      "step": 8660
    },
    {
      "epoch": 4.5414046121593294,
      "grad_norm": 1.1130290031433105,
      "learning_rate": 4.7581734120563206e-05,
      "loss": 0.2793,
      "num_input_tokens_seen": 5653504,
      "step": 8665
    },
    {
      "epoch": 4.544025157232705,
      "grad_norm": 1.10697340965271,
      "learning_rate": 4.757682560501249e-05,
      "loss": 0.4215,
      "num_input_tokens_seen": 5656896,
      "step": 8670
    },
    {
      "epoch": 4.54664570230608,
      "grad_norm": 0.9463984966278076,
      "learning_rate": 4.757191236664744e-05,
      "loss": 0.3732,
      "num_input_tokens_seen": 5660928,
      "step": 8675
    },
    {
      "epoch": 4.549266247379455,
      "grad_norm": 1.090866208076477,
      "learning_rate": 4.7566994406495825e-05,
      "loss": 0.3901,
      "num_input_tokens_seen": 5664064,
      "step": 8680
    },
    {
      "epoch": 4.55188679245283,
      "grad_norm": 4.096395969390869,
      "learning_rate": 4.756207172558645e-05,
      "loss": 0.4559,
      "num_input_tokens_seen": 5667232,
      "step": 8685
    },
    {
      "epoch": 4.554507337526205,
      "grad_norm": 1.612423300743103,
      "learning_rate": 4.7557144324949065e-05,
      "loss": 0.3931,
      "num_input_tokens_seen": 5670176,
      "step": 8690
    },
    {
      "epoch": 4.5571278825995805,
      "grad_norm": 1.5405023097991943,
      "learning_rate": 4.755221220561443e-05,
      "loss": 0.2895,
      "num_input_tokens_seen": 5673120,
      "step": 8695
    },
    {
      "epoch": 4.559748427672956,
      "grad_norm": 0.9560888409614563,
      "learning_rate": 4.7547275368614295e-05,
      "loss": 0.3565,
      "num_input_tokens_seen": 5675936,
      "step": 8700
    },
    {
      "epoch": 4.562368972746331,
      "grad_norm": 0.7194001078605652,
      "learning_rate": 4.754233381498139e-05,
      "loss": 0.2813,
      "num_input_tokens_seen": 5679072,
      "step": 8705
    },
    {
      "epoch": 4.564989517819707,
      "grad_norm": 1.3700649738311768,
      "learning_rate": 4.753738754574942e-05,
      "loss": 0.3616,
      "num_input_tokens_seen": 5682112,
      "step": 8710
    },
    {
      "epoch": 4.567610062893082,
      "grad_norm": 2.3798539638519287,
      "learning_rate": 4.75324365619531e-05,
      "loss": 0.4266,
      "num_input_tokens_seen": 5685696,
      "step": 8715
    },
    {
      "epoch": 4.570230607966457,
      "grad_norm": 1.4507968425750732,
      "learning_rate": 4.752748086462811e-05,
      "loss": 0.3419,
      "num_input_tokens_seen": 5688096,
      "step": 8720
    },
    {
      "epoch": 4.572851153039832,
      "grad_norm": 1.1252162456512451,
      "learning_rate": 4.752252045481113e-05,
      "loss": 0.379,
      "num_input_tokens_seen": 5691072,
      "step": 8725
    },
    {
      "epoch": 4.5754716981132075,
      "grad_norm": 1.340590238571167,
      "learning_rate": 4.7517555333539834e-05,
      "loss": 0.3816,
      "num_input_tokens_seen": 5694592,
      "step": 8730
    },
    {
      "epoch": 4.578092243186583,
      "grad_norm": 0.7425200939178467,
      "learning_rate": 4.751258550185284e-05,
      "loss": 0.3122,
      "num_input_tokens_seen": 5697312,
      "step": 8735
    },
    {
      "epoch": 4.580712788259958,
      "grad_norm": 1.0181618928909302,
      "learning_rate": 4.7507610960789804e-05,
      "loss": 0.2528,
      "num_input_tokens_seen": 5701088,
      "step": 8740
    },
    {
      "epoch": 4.583333333333333,
      "grad_norm": 1.2320626974105835,
      "learning_rate": 4.7502631711391334e-05,
      "loss": 0.4213,
      "num_input_tokens_seen": 5703904,
      "step": 8745
    },
    {
      "epoch": 4.585953878406709,
      "grad_norm": 1.1615296602249146,
      "learning_rate": 4.749764775469904e-05,
      "loss": 0.4047,
      "num_input_tokens_seen": 5706816,
      "step": 8750
    },
    {
      "epoch": 4.588574423480084,
      "grad_norm": 1.5880504846572876,
      "learning_rate": 4.749265909175549e-05,
      "loss": 0.3818,
      "num_input_tokens_seen": 5709632,
      "step": 8755
    },
    {
      "epoch": 4.591194968553459,
      "grad_norm": 1.3872160911560059,
      "learning_rate": 4.7487665723604275e-05,
      "loss": 0.376,
      "num_input_tokens_seen": 5713472,
      "step": 8760
    },
    {
      "epoch": 4.593815513626835,
      "grad_norm": 1.531185269355774,
      "learning_rate": 4.7482667651289944e-05,
      "loss": 0.5348,
      "num_input_tokens_seen": 5716736,
      "step": 8765
    },
    {
      "epoch": 4.59643605870021,
      "grad_norm": 1.6779319047927856,
      "learning_rate": 4.747766487585803e-05,
      "loss": 0.3274,
      "num_input_tokens_seen": 5720672,
      "step": 8770
    },
    {
      "epoch": 4.599056603773585,
      "grad_norm": 0.9381993412971497,
      "learning_rate": 4.747265739835507e-05,
      "loss": 0.5268,
      "num_input_tokens_seen": 5724000,
      "step": 8775
    },
    {
      "epoch": 4.60167714884696,
      "grad_norm": 2.0642430782318115,
      "learning_rate": 4.746764521982856e-05,
      "loss": 0.4598,
      "num_input_tokens_seen": 5726784,
      "step": 8780
    },
    {
      "epoch": 4.604297693920335,
      "grad_norm": 1.0041966438293457,
      "learning_rate": 4.7462628341326995e-05,
      "loss": 0.3669,
      "num_input_tokens_seen": 5730272,
      "step": 8785
    },
    {
      "epoch": 4.6069182389937104,
      "grad_norm": 2.5606672763824463,
      "learning_rate": 4.7457606763899856e-05,
      "loss": 0.377,
      "num_input_tokens_seen": 5733120,
      "step": 8790
    },
    {
      "epoch": 4.609538784067086,
      "grad_norm": 1.300341248512268,
      "learning_rate": 4.745258048859758e-05,
      "loss": 0.2424,
      "num_input_tokens_seen": 5737824,
      "step": 8795
    },
    {
      "epoch": 4.612159329140461,
      "grad_norm": 1.0986857414245605,
      "learning_rate": 4.744754951647162e-05,
      "loss": 0.3125,
      "num_input_tokens_seen": 5740384,
      "step": 8800
    },
    {
      "epoch": 4.614779874213837,
      "grad_norm": 1.2719744443893433,
      "learning_rate": 4.74425138485744e-05,
      "loss": 0.3367,
      "num_input_tokens_seen": 5743232,
      "step": 8805
    },
    {
      "epoch": 4.617400419287212,
      "grad_norm": 1.0454394817352295,
      "learning_rate": 4.743747348595932e-05,
      "loss": 0.4027,
      "num_input_tokens_seen": 5746432,
      "step": 8810
    },
    {
      "epoch": 4.620020964360587,
      "grad_norm": 1.372251033782959,
      "learning_rate": 4.743242842968077e-05,
      "loss": 0.4135,
      "num_input_tokens_seen": 5749696,
      "step": 8815
    },
    {
      "epoch": 4.622641509433962,
      "grad_norm": 1.0988731384277344,
      "learning_rate": 4.742737868079411e-05,
      "loss": 0.3668,
      "num_input_tokens_seen": 5753152,
      "step": 8820
    },
    {
      "epoch": 4.6252620545073375,
      "grad_norm": 0.8244682550430298,
      "learning_rate": 4.7422324240355696e-05,
      "loss": 0.2973,
      "num_input_tokens_seen": 5755712,
      "step": 8825
    },
    {
      "epoch": 4.627882599580713,
      "grad_norm": 1.045543909072876,
      "learning_rate": 4.741726510942286e-05,
      "loss": 0.4224,
      "num_input_tokens_seen": 5759040,
      "step": 8830
    },
    {
      "epoch": 4.630503144654088,
      "grad_norm": 1.982983112335205,
      "learning_rate": 4.741220128905391e-05,
      "loss": 0.4452,
      "num_input_tokens_seen": 5761408,
      "step": 8835
    },
    {
      "epoch": 4.633123689727463,
      "grad_norm": 1.1812453269958496,
      "learning_rate": 4.740713278030815e-05,
      "loss": 0.393,
      "num_input_tokens_seen": 5764672,
      "step": 8840
    },
    {
      "epoch": 4.635744234800838,
      "grad_norm": 0.8320803642272949,
      "learning_rate": 4.740205958424584e-05,
      "loss": 0.3268,
      "num_input_tokens_seen": 5768160,
      "step": 8845
    },
    {
      "epoch": 4.638364779874214,
      "grad_norm": 1.7807332277297974,
      "learning_rate": 4.739698170192824e-05,
      "loss": 0.3238,
      "num_input_tokens_seen": 5770400,
      "step": 8850
    },
    {
      "epoch": 4.640985324947589,
      "grad_norm": 0.9505714178085327,
      "learning_rate": 4.7391899134417596e-05,
      "loss": 0.2936,
      "num_input_tokens_seen": 5774080,
      "step": 8855
    },
    {
      "epoch": 4.643605870020965,
      "grad_norm": 1.6893354654312134,
      "learning_rate": 4.73868118827771e-05,
      "loss": 0.3564,
      "num_input_tokens_seen": 5777184,
      "step": 8860
    },
    {
      "epoch": 4.64622641509434,
      "grad_norm": 1.4148666858673096,
      "learning_rate": 4.738171994807097e-05,
      "loss": 0.429,
      "num_input_tokens_seen": 5780736,
      "step": 8865
    },
    {
      "epoch": 4.648846960167715,
      "grad_norm": 1.2207605838775635,
      "learning_rate": 4.7376623331364364e-05,
      "loss": 0.4092,
      "num_input_tokens_seen": 5784320,
      "step": 8870
    },
    {
      "epoch": 4.65146750524109,
      "grad_norm": 0.8379622101783752,
      "learning_rate": 4.737152203372345e-05,
      "loss": 0.3416,
      "num_input_tokens_seen": 5787840,
      "step": 8875
    },
    {
      "epoch": 4.654088050314465,
      "grad_norm": 1.3303731679916382,
      "learning_rate": 4.7366416056215345e-05,
      "loss": 0.4468,
      "num_input_tokens_seen": 5791040,
      "step": 8880
    },
    {
      "epoch": 4.65670859538784,
      "grad_norm": 6.069678783416748,
      "learning_rate": 4.7361305399908185e-05,
      "loss": 0.5771,
      "num_input_tokens_seen": 5794560,
      "step": 8885
    },
    {
      "epoch": 4.659329140461216,
      "grad_norm": 1.7689399719238281,
      "learning_rate": 4.735619006587103e-05,
      "loss": 0.3722,
      "num_input_tokens_seen": 5797504,
      "step": 8890
    },
    {
      "epoch": 4.661949685534591,
      "grad_norm": 0.9148828983306885,
      "learning_rate": 4.735107005517398e-05,
      "loss": 0.3466,
      "num_input_tokens_seen": 5799808,
      "step": 8895
    },
    {
      "epoch": 4.664570230607967,
      "grad_norm": 1.188522458076477,
      "learning_rate": 4.734594536888806e-05,
      "loss": 0.2074,
      "num_input_tokens_seen": 5802752,
      "step": 8900
    },
    {
      "epoch": 4.667190775681342,
      "grad_norm": 1.027174711227417,
      "learning_rate": 4.734081600808531e-05,
      "loss": 0.3871,
      "num_input_tokens_seen": 5806144,
      "step": 8905
    },
    {
      "epoch": 4.669811320754717,
      "grad_norm": 1.1668784618377686,
      "learning_rate": 4.733568197383872e-05,
      "loss": 0.4709,
      "num_input_tokens_seen": 5809536,
      "step": 8910
    },
    {
      "epoch": 4.672431865828092,
      "grad_norm": 1.6509077548980713,
      "learning_rate": 4.7330543267222294e-05,
      "loss": 0.3462,
      "num_input_tokens_seen": 5813088,
      "step": 8915
    },
    {
      "epoch": 4.6750524109014675,
      "grad_norm": 1.2639977931976318,
      "learning_rate": 4.732539988931097e-05,
      "loss": 0.2659,
      "num_input_tokens_seen": 5815488,
      "step": 8920
    },
    {
      "epoch": 4.677672955974843,
      "grad_norm": 0.9580846428871155,
      "learning_rate": 4.732025184118068e-05,
      "loss": 0.471,
      "num_input_tokens_seen": 5818528,
      "step": 8925
    },
    {
      "epoch": 4.680293501048218,
      "grad_norm": 0.8668920397758484,
      "learning_rate": 4.731509912390837e-05,
      "loss": 0.4028,
      "num_input_tokens_seen": 5822016,
      "step": 8930
    },
    {
      "epoch": 4.682914046121593,
      "grad_norm": 0.6535959243774414,
      "learning_rate": 4.730994173857189e-05,
      "loss": 0.5018,
      "num_input_tokens_seen": 5825952,
      "step": 8935
    },
    {
      "epoch": 4.685534591194968,
      "grad_norm": 1.1090223789215088,
      "learning_rate": 4.7304779686250126e-05,
      "loss": 0.2878,
      "num_input_tokens_seen": 5828416,
      "step": 8940
    },
    {
      "epoch": 4.688155136268344,
      "grad_norm": 1.1584241390228271,
      "learning_rate": 4.729961296802292e-05,
      "loss": 0.3635,
      "num_input_tokens_seen": 5832352,
      "step": 8945
    },
    {
      "epoch": 4.690775681341719,
      "grad_norm": 1.3913192749023438,
      "learning_rate": 4.7294441584971095e-05,
      "loss": 0.4177,
      "num_input_tokens_seen": 5835488,
      "step": 8950
    },
    {
      "epoch": 4.693396226415095,
      "grad_norm": 1.2266885042190552,
      "learning_rate": 4.728926553817643e-05,
      "loss": 0.41,
      "num_input_tokens_seen": 5838624,
      "step": 8955
    },
    {
      "epoch": 4.69601677148847,
      "grad_norm": 0.7807854413986206,
      "learning_rate": 4.728408482872171e-05,
      "loss": 0.3558,
      "num_input_tokens_seen": 5842592,
      "step": 8960
    },
    {
      "epoch": 4.698637316561845,
      "grad_norm": 1.3110705614089966,
      "learning_rate": 4.727889945769067e-05,
      "loss": 0.3829,
      "num_input_tokens_seen": 5846080,
      "step": 8965
    },
    {
      "epoch": 4.70125786163522,
      "grad_norm": 1.3809294700622559,
      "learning_rate": 4.727370942616803e-05,
      "loss": 0.4018,
      "num_input_tokens_seen": 5849536,
      "step": 8970
    },
    {
      "epoch": 4.703878406708595,
      "grad_norm": 1.1403979063034058,
      "learning_rate": 4.72685147352395e-05,
      "loss": 0.3317,
      "num_input_tokens_seen": 5852032,
      "step": 8975
    },
    {
      "epoch": 4.70649895178197,
      "grad_norm": 1.6081867218017578,
      "learning_rate": 4.7263315385991735e-05,
      "loss": 0.4344,
      "num_input_tokens_seen": 5855072,
      "step": 8980
    },
    {
      "epoch": 4.709119496855346,
      "grad_norm": 1.1805025339126587,
      "learning_rate": 4.7258111379512384e-05,
      "loss": 0.4172,
      "num_input_tokens_seen": 5858272,
      "step": 8985
    },
    {
      "epoch": 4.711740041928721,
      "grad_norm": 0.8157265186309814,
      "learning_rate": 4.725290271689007e-05,
      "loss": 0.381,
      "num_input_tokens_seen": 5861856,
      "step": 8990
    },
    {
      "epoch": 4.714360587002097,
      "grad_norm": 1.02849280834198,
      "learning_rate": 4.724768939921438e-05,
      "loss": 0.3549,
      "num_input_tokens_seen": 5864448,
      "step": 8995
    },
    {
      "epoch": 4.716981132075472,
      "grad_norm": 0.8645599484443665,
      "learning_rate": 4.7242471427575884e-05,
      "loss": 0.3338,
      "num_input_tokens_seen": 5867744,
      "step": 9000
    },
    {
      "epoch": 4.719601677148847,
      "grad_norm": 1.4349156618118286,
      "learning_rate": 4.723724880306612e-05,
      "loss": 0.3142,
      "num_input_tokens_seen": 5870336,
      "step": 9005
    },
    {
      "epoch": 4.722222222222222,
      "grad_norm": 0.7482133507728577,
      "learning_rate": 4.72320215267776e-05,
      "loss": 0.4069,
      "num_input_tokens_seen": 5873728,
      "step": 9010
    },
    {
      "epoch": 4.7248427672955975,
      "grad_norm": 0.9016861319541931,
      "learning_rate": 4.722678959980381e-05,
      "loss": 0.3139,
      "num_input_tokens_seen": 5876320,
      "step": 9015
    },
    {
      "epoch": 4.727463312368973,
      "grad_norm": 1.9159198999404907,
      "learning_rate": 4.7221553023239216e-05,
      "loss": 0.3376,
      "num_input_tokens_seen": 5879552,
      "step": 9020
    },
    {
      "epoch": 4.730083857442348,
      "grad_norm": 2.4575114250183105,
      "learning_rate": 4.721631179817925e-05,
      "loss": 0.3308,
      "num_input_tokens_seen": 5882432,
      "step": 9025
    },
    {
      "epoch": 4.732704402515723,
      "grad_norm": 12.69784927368164,
      "learning_rate": 4.721106592572031e-05,
      "loss": 0.3203,
      "num_input_tokens_seen": 5885248,
      "step": 9030
    },
    {
      "epoch": 4.735324947589098,
      "grad_norm": 0.9872632026672363,
      "learning_rate": 4.7205815406959774e-05,
      "loss": 0.3185,
      "num_input_tokens_seen": 5887680,
      "step": 9035
    },
    {
      "epoch": 4.737945492662474,
      "grad_norm": 1.3234702348709106,
      "learning_rate": 4.7200560242996e-05,
      "loss": 0.3396,
      "num_input_tokens_seen": 5890848,
      "step": 9040
    },
    {
      "epoch": 4.740566037735849,
      "grad_norm": 0.8651976585388184,
      "learning_rate": 4.719530043492829e-05,
      "loss": 0.3635,
      "num_input_tokens_seen": 5893664,
      "step": 9045
    },
    {
      "epoch": 4.743186582809225,
      "grad_norm": 0.9768457412719727,
      "learning_rate": 4.7190035983856954e-05,
      "loss": 0.4235,
      "num_input_tokens_seen": 5897216,
      "step": 9050
    },
    {
      "epoch": 4.7458071278826,
      "grad_norm": 1.2452683448791504,
      "learning_rate": 4.718476689088325e-05,
      "loss": 0.3591,
      "num_input_tokens_seen": 5900992,
      "step": 9055
    },
    {
      "epoch": 4.748427672955975,
      "grad_norm": 0.7924380302429199,
      "learning_rate": 4.71794931571094e-05,
      "loss": 0.4886,
      "num_input_tokens_seen": 5903904,
      "step": 9060
    },
    {
      "epoch": 4.75104821802935,
      "grad_norm": 1.1683883666992188,
      "learning_rate": 4.717421478363862e-05,
      "loss": 0.4266,
      "num_input_tokens_seen": 5907008,
      "step": 9065
    },
    {
      "epoch": 4.753668763102725,
      "grad_norm": 1.81244695186615,
      "learning_rate": 4.7168931771575085e-05,
      "loss": 0.3103,
      "num_input_tokens_seen": 5910112,
      "step": 9070
    },
    {
      "epoch": 4.7562893081761,
      "grad_norm": 1.0366106033325195,
      "learning_rate": 4.716364412202394e-05,
      "loss": 0.3583,
      "num_input_tokens_seen": 5913216,
      "step": 9075
    },
    {
      "epoch": 4.758909853249476,
      "grad_norm": 1.0672365427017212,
      "learning_rate": 4.7158351836091306e-05,
      "loss": 0.3684,
      "num_input_tokens_seen": 5916256,
      "step": 9080
    },
    {
      "epoch": 4.761530398322851,
      "grad_norm": 1.0043846368789673,
      "learning_rate": 4.7153054914884266e-05,
      "loss": 0.3487,
      "num_input_tokens_seen": 5919040,
      "step": 9085
    },
    {
      "epoch": 4.764150943396227,
      "grad_norm": 0.9156254529953003,
      "learning_rate": 4.714775335951086e-05,
      "loss": 0.3554,
      "num_input_tokens_seen": 5922944,
      "step": 9090
    },
    {
      "epoch": 4.766771488469602,
      "grad_norm": 1.003435730934143,
      "learning_rate": 4.7142447171080137e-05,
      "loss": 0.291,
      "num_input_tokens_seen": 5926368,
      "step": 9095
    },
    {
      "epoch": 4.769392033542977,
      "grad_norm": 0.7068285346031189,
      "learning_rate": 4.713713635070208e-05,
      "loss": 0.3283,
      "num_input_tokens_seen": 5929632,
      "step": 9100
    },
    {
      "epoch": 4.772012578616352,
      "grad_norm": 0.6856381297111511,
      "learning_rate": 4.713182089948765e-05,
      "loss": 0.3891,
      "num_input_tokens_seen": 5932832,
      "step": 9105
    },
    {
      "epoch": 4.7746331236897275,
      "grad_norm": 1.2329001426696777,
      "learning_rate": 4.712650081854878e-05,
      "loss": 0.308,
      "num_input_tokens_seen": 5936608,
      "step": 9110
    },
    {
      "epoch": 4.777253668763103,
      "grad_norm": 1.8724792003631592,
      "learning_rate": 4.712117610899837e-05,
      "loss": 0.4416,
      "num_input_tokens_seen": 5940640,
      "step": 9115
    },
    {
      "epoch": 4.779874213836478,
      "grad_norm": 1.2197211980819702,
      "learning_rate": 4.7115846771950286e-05,
      "loss": 0.3526,
      "num_input_tokens_seen": 5944416,
      "step": 9120
    },
    {
      "epoch": 4.782494758909853,
      "grad_norm": 2.9818804264068604,
      "learning_rate": 4.711051280851938e-05,
      "loss": 0.3531,
      "num_input_tokens_seen": 5948352,
      "step": 9125
    },
    {
      "epoch": 4.785115303983228,
      "grad_norm": 0.973572313785553,
      "learning_rate": 4.7105174219821435e-05,
      "loss": 0.3485,
      "num_input_tokens_seen": 5952000,
      "step": 9130
    },
    {
      "epoch": 4.787735849056604,
      "grad_norm": 1.8251278400421143,
      "learning_rate": 4.709983100697323e-05,
      "loss": 0.3581,
      "num_input_tokens_seen": 5954528,
      "step": 9135
    },
    {
      "epoch": 4.790356394129979,
      "grad_norm": 1.1659032106399536,
      "learning_rate": 4.709448317109252e-05,
      "loss": 0.342,
      "num_input_tokens_seen": 5957664,
      "step": 9140
    },
    {
      "epoch": 4.7929769392033545,
      "grad_norm": 0.8124675154685974,
      "learning_rate": 4.708913071329798e-05,
      "loss": 0.2988,
      "num_input_tokens_seen": 5960448,
      "step": 9145
    },
    {
      "epoch": 4.79559748427673,
      "grad_norm": 0.7323048710823059,
      "learning_rate": 4.708377363470932e-05,
      "loss": 0.3339,
      "num_input_tokens_seen": 5963072,
      "step": 9150
    },
    {
      "epoch": 4.798218029350105,
      "grad_norm": 1.0206263065338135,
      "learning_rate": 4.7078411936447144e-05,
      "loss": 0.3062,
      "num_input_tokens_seen": 5966240,
      "step": 9155
    },
    {
      "epoch": 4.80083857442348,
      "grad_norm": 1.0868844985961914,
      "learning_rate": 4.707304561963308e-05,
      "loss": 0.2921,
      "num_input_tokens_seen": 5969664,
      "step": 9160
    },
    {
      "epoch": 4.803459119496855,
      "grad_norm": 1.3147882223129272,
      "learning_rate": 4.70676746853897e-05,
      "loss": 0.3439,
      "num_input_tokens_seen": 5972320,
      "step": 9165
    },
    {
      "epoch": 4.80607966457023,
      "grad_norm": 1.141404628753662,
      "learning_rate": 4.706229913484054e-05,
      "loss": 0.4023,
      "num_input_tokens_seen": 5975264,
      "step": 9170
    },
    {
      "epoch": 4.808700209643606,
      "grad_norm": 1.4498692750930786,
      "learning_rate": 4.705691896911009e-05,
      "loss": 0.3997,
      "num_input_tokens_seen": 5978336,
      "step": 9175
    },
    {
      "epoch": 4.811320754716981,
      "grad_norm": 1.4735835790634155,
      "learning_rate": 4.705153418932383e-05,
      "loss": 0.3836,
      "num_input_tokens_seen": 5981536,
      "step": 9180
    },
    {
      "epoch": 4.813941299790356,
      "grad_norm": 2.208831548690796,
      "learning_rate": 4.7046144796608206e-05,
      "loss": 0.4799,
      "num_input_tokens_seen": 5984640,
      "step": 9185
    },
    {
      "epoch": 4.816561844863732,
      "grad_norm": 1.3863133192062378,
      "learning_rate": 4.70407507920906e-05,
      "loss": 0.3904,
      "num_input_tokens_seen": 5987168,
      "step": 9190
    },
    {
      "epoch": 4.819182389937107,
      "grad_norm": 0.548150897026062,
      "learning_rate": 4.7035352176899384e-05,
      "loss": 0.3934,
      "num_input_tokens_seen": 5990880,
      "step": 9195
    },
    {
      "epoch": 4.821802935010482,
      "grad_norm": 1.0999853610992432,
      "learning_rate": 4.7029948952163885e-05,
      "loss": 0.3733,
      "num_input_tokens_seen": 5993664,
      "step": 9200
    },
    {
      "epoch": 4.8244234800838575,
      "grad_norm": 0.8130893707275391,
      "learning_rate": 4.7024541119014386e-05,
      "loss": 0.4489,
      "num_input_tokens_seen": 5997216,
      "step": 9205
    },
    {
      "epoch": 4.827044025157233,
      "grad_norm": 0.7600782513618469,
      "learning_rate": 4.701912867858217e-05,
      "loss": 0.2437,
      "num_input_tokens_seen": 5999840,
      "step": 9210
    },
    {
      "epoch": 4.829664570230608,
      "grad_norm": 2.1617965698242188,
      "learning_rate": 4.7013711631999434e-05,
      "loss": 0.3555,
      "num_input_tokens_seen": 6003008,
      "step": 9215
    },
    {
      "epoch": 4.832285115303983,
      "grad_norm": 2.2876923084259033,
      "learning_rate": 4.700828998039937e-05,
      "loss": 0.4084,
      "num_input_tokens_seen": 6007552,
      "step": 9220
    },
    {
      "epoch": 4.834905660377358,
      "grad_norm": 1.4835971593856812,
      "learning_rate": 4.700286372491612e-05,
      "loss": 0.4087,
      "num_input_tokens_seen": 6011040,
      "step": 9225
    },
    {
      "epoch": 4.837526205450734,
      "grad_norm": 1.0057721138000488,
      "learning_rate": 4.6997432866684816e-05,
      "loss": 0.3093,
      "num_input_tokens_seen": 6013536,
      "step": 9230
    },
    {
      "epoch": 4.840146750524109,
      "grad_norm": 1.6017415523529053,
      "learning_rate": 4.6991997406841504e-05,
      "loss": 0.3329,
      "num_input_tokens_seen": 6015808,
      "step": 9235
    },
    {
      "epoch": 4.8427672955974845,
      "grad_norm": 0.963726818561554,
      "learning_rate": 4.698655734652324e-05,
      "loss": 0.2893,
      "num_input_tokens_seen": 6019328,
      "step": 9240
    },
    {
      "epoch": 4.84538784067086,
      "grad_norm": 0.7825763821601868,
      "learning_rate": 4.698111268686801e-05,
      "loss": 0.3311,
      "num_input_tokens_seen": 6022496,
      "step": 9245
    },
    {
      "epoch": 4.848008385744235,
      "grad_norm": 1.5761195421218872,
      "learning_rate": 4.6975663429014785e-05,
      "loss": 0.35,
      "num_input_tokens_seen": 6025408,
      "step": 9250
    },
    {
      "epoch": 4.85062893081761,
      "grad_norm": 1.1397826671600342,
      "learning_rate": 4.697020957410347e-05,
      "loss": 0.4331,
      "num_input_tokens_seen": 6029056,
      "step": 9255
    },
    {
      "epoch": 4.853249475890985,
      "grad_norm": 1.1707979440689087,
      "learning_rate": 4.6964751123274974e-05,
      "loss": 0.2625,
      "num_input_tokens_seen": 6031584,
      "step": 9260
    },
    {
      "epoch": 4.85587002096436,
      "grad_norm": 1.4557384252548218,
      "learning_rate": 4.695928807767113e-05,
      "loss": 0.3804,
      "num_input_tokens_seen": 6034432,
      "step": 9265
    },
    {
      "epoch": 4.8584905660377355,
      "grad_norm": 1.277340054512024,
      "learning_rate": 4.695382043843474e-05,
      "loss": 0.3025,
      "num_input_tokens_seen": 6037792,
      "step": 9270
    },
    {
      "epoch": 4.861111111111111,
      "grad_norm": 1.271830439567566,
      "learning_rate": 4.694834820670958e-05,
      "loss": 0.3725,
      "num_input_tokens_seen": 6040832,
      "step": 9275
    },
    {
      "epoch": 4.863731656184486,
      "grad_norm": 0.9890521168708801,
      "learning_rate": 4.694287138364037e-05,
      "loss": 0.4973,
      "num_input_tokens_seen": 6044480,
      "step": 9280
    },
    {
      "epoch": 4.866352201257862,
      "grad_norm": 0.7881940007209778,
      "learning_rate": 4.6937389970372815e-05,
      "loss": 0.3345,
      "num_input_tokens_seen": 6047520,
      "step": 9285
    },
    {
      "epoch": 4.868972746331237,
      "grad_norm": 0.9062352776527405,
      "learning_rate": 4.693190396805354e-05,
      "loss": 0.3421,
      "num_input_tokens_seen": 6051008,
      "step": 9290
    },
    {
      "epoch": 4.871593291404612,
      "grad_norm": 0.7872847318649292,
      "learning_rate": 4.692641337783017e-05,
      "loss": 0.3915,
      "num_input_tokens_seen": 6053216,
      "step": 9295
    },
    {
      "epoch": 4.8742138364779874,
      "grad_norm": 1.1503041982650757,
      "learning_rate": 4.6920918200851274e-05,
      "loss": 0.4389,
      "num_input_tokens_seen": 6056864,
      "step": 9300
    },
    {
      "epoch": 4.876834381551363,
      "grad_norm": 1.0523735284805298,
      "learning_rate": 4.691541843826638e-05,
      "loss": 0.3055,
      "num_input_tokens_seen": 6059744,
      "step": 9305
    },
    {
      "epoch": 4.879454926624738,
      "grad_norm": 1.698865294456482,
      "learning_rate": 4.690991409122597e-05,
      "loss": 0.3036,
      "num_input_tokens_seen": 6062624,
      "step": 9310
    },
    {
      "epoch": 4.882075471698113,
      "grad_norm": 1.2991031408309937,
      "learning_rate": 4.690440516088149e-05,
      "loss": 0.3299,
      "num_input_tokens_seen": 6066080,
      "step": 9315
    },
    {
      "epoch": 4.884696016771488,
      "grad_norm": 0.9744247794151306,
      "learning_rate": 4.689889164838536e-05,
      "loss": 0.4093,
      "num_input_tokens_seen": 6069696,
      "step": 9320
    },
    {
      "epoch": 4.887316561844864,
      "grad_norm": 1.0086876153945923,
      "learning_rate": 4.6893373554890925e-05,
      "loss": 0.483,
      "num_input_tokens_seen": 6073408,
      "step": 9325
    },
    {
      "epoch": 4.889937106918239,
      "grad_norm": 0.9020155668258667,
      "learning_rate": 4.688785088155251e-05,
      "loss": 0.2234,
      "num_input_tokens_seen": 6078592,
      "step": 9330
    },
    {
      "epoch": 4.8925576519916145,
      "grad_norm": 1.021240472793579,
      "learning_rate": 4.688232362952541e-05,
      "loss": 0.2498,
      "num_input_tokens_seen": 6088128,
      "step": 9335
    },
    {
      "epoch": 4.89517819706499,
      "grad_norm": 1.74807870388031,
      "learning_rate": 4.687679179996585e-05,
      "loss": 0.3923,
      "num_input_tokens_seen": 6090816,
      "step": 9340
    },
    {
      "epoch": 4.897798742138365,
      "grad_norm": 1.0996885299682617,
      "learning_rate": 4.687125539403103e-05,
      "loss": 0.3716,
      "num_input_tokens_seen": 6093856,
      "step": 9345
    },
    {
      "epoch": 4.90041928721174,
      "grad_norm": 0.9066886901855469,
      "learning_rate": 4.68657144128791e-05,
      "loss": 0.3442,
      "num_input_tokens_seen": 6097792,
      "step": 9350
    },
    {
      "epoch": 4.903039832285115,
      "grad_norm": 0.8180421590805054,
      "learning_rate": 4.6860168857669175e-05,
      "loss": 0.2548,
      "num_input_tokens_seen": 6100160,
      "step": 9355
    },
    {
      "epoch": 4.90566037735849,
      "grad_norm": 1.499916672706604,
      "learning_rate": 4.6854618729561315e-05,
      "loss": 0.4514,
      "num_input_tokens_seen": 6103264,
      "step": 9360
    },
    {
      "epoch": 4.9082809224318655,
      "grad_norm": 1.0060608386993408,
      "learning_rate": 4.684906402971655e-05,
      "loss": 0.4603,
      "num_input_tokens_seen": 6105856,
      "step": 9365
    },
    {
      "epoch": 4.910901467505241,
      "grad_norm": 1.332993984222412,
      "learning_rate": 4.684350475929685e-05,
      "loss": 0.4297,
      "num_input_tokens_seen": 6108800,
      "step": 9370
    },
    {
      "epoch": 4.913522012578616,
      "grad_norm": 0.9837652444839478,
      "learning_rate": 4.683794091946517e-05,
      "loss": 0.3383,
      "num_input_tokens_seen": 6111552,
      "step": 9375
    },
    {
      "epoch": 4.916142557651992,
      "grad_norm": 0.9852614402770996,
      "learning_rate": 4.683237251138538e-05,
      "loss": 0.4413,
      "num_input_tokens_seen": 6115488,
      "step": 9380
    },
    {
      "epoch": 4.918763102725367,
      "grad_norm": 0.9798521399497986,
      "learning_rate": 4.6826799536222346e-05,
      "loss": 0.2944,
      "num_input_tokens_seen": 6119136,
      "step": 9385
    },
    {
      "epoch": 4.921383647798742,
      "grad_norm": 1.1147470474243164,
      "learning_rate": 4.6821221995141865e-05,
      "loss": 0.4902,
      "num_input_tokens_seen": 6121696,
      "step": 9390
    },
    {
      "epoch": 4.924004192872117,
      "grad_norm": 1.703153133392334,
      "learning_rate": 4.681563988931068e-05,
      "loss": 0.3278,
      "num_input_tokens_seen": 6124480,
      "step": 9395
    },
    {
      "epoch": 4.926624737945493,
      "grad_norm": 1.6408406496047974,
      "learning_rate": 4.681005321989652e-05,
      "loss": 0.3409,
      "num_input_tokens_seen": 6127008,
      "step": 9400
    },
    {
      "epoch": 4.929245283018868,
      "grad_norm": 1.5927685499191284,
      "learning_rate": 4.6804461988068055e-05,
      "loss": 0.2332,
      "num_input_tokens_seen": 6130080,
      "step": 9405
    },
    {
      "epoch": 4.931865828092243,
      "grad_norm": 0.9559873342514038,
      "learning_rate": 4.6798866194994896e-05,
      "loss": 0.3748,
      "num_input_tokens_seen": 6133632,
      "step": 9410
    },
    {
      "epoch": 4.934486373165618,
      "grad_norm": 0.8996161818504333,
      "learning_rate": 4.6793265841847624e-05,
      "loss": 0.3529,
      "num_input_tokens_seen": 6136992,
      "step": 9415
    },
    {
      "epoch": 4.937106918238994,
      "grad_norm": 1.2535182237625122,
      "learning_rate": 4.6787660929797766e-05,
      "loss": 0.2892,
      "num_input_tokens_seen": 6139776,
      "step": 9420
    },
    {
      "epoch": 4.939727463312369,
      "grad_norm": 0.9702702164649963,
      "learning_rate": 4.678205146001781e-05,
      "loss": 0.3974,
      "num_input_tokens_seen": 6142880,
      "step": 9425
    },
    {
      "epoch": 4.9423480083857445,
      "grad_norm": 1.3327330350875854,
      "learning_rate": 4.67764374336812e-05,
      "loss": 0.3239,
      "num_input_tokens_seen": 6147200,
      "step": 9430
    },
    {
      "epoch": 4.94496855345912,
      "grad_norm": 1.210758090019226,
      "learning_rate": 4.6770818851962305e-05,
      "loss": 0.3548,
      "num_input_tokens_seen": 6150016,
      "step": 9435
    },
    {
      "epoch": 4.947589098532495,
      "grad_norm": 0.7262565493583679,
      "learning_rate": 4.676519571603648e-05,
      "loss": 0.3082,
      "num_input_tokens_seen": 6152960,
      "step": 9440
    },
    {
      "epoch": 4.95020964360587,
      "grad_norm": 1.2515974044799805,
      "learning_rate": 4.6759568027080034e-05,
      "loss": 0.2848,
      "num_input_tokens_seen": 6156544,
      "step": 9445
    },
    {
      "epoch": 4.952830188679245,
      "grad_norm": 1.5096583366394043,
      "learning_rate": 4.67539357862702e-05,
      "loss": 0.365,
      "num_input_tokens_seen": 6159584,
      "step": 9450
    },
    {
      "epoch": 4.95545073375262,
      "grad_norm": 1.7924439907073975,
      "learning_rate": 4.6748298994785175e-05,
      "loss": 0.3809,
      "num_input_tokens_seen": 6162752,
      "step": 9455
    },
    {
      "epoch": 4.9580712788259955,
      "grad_norm": 1.0286904573440552,
      "learning_rate": 4.6742657653804116e-05,
      "loss": 0.242,
      "num_input_tokens_seen": 6166016,
      "step": 9460
    },
    {
      "epoch": 4.960691823899371,
      "grad_norm": 1.2308590412139893,
      "learning_rate": 4.673701176450713e-05,
      "loss": 0.3556,
      "num_input_tokens_seen": 6169184,
      "step": 9465
    },
    {
      "epoch": 4.963312368972746,
      "grad_norm": 1.7278921604156494,
      "learning_rate": 4.6731361328075286e-05,
      "loss": 0.3395,
      "num_input_tokens_seen": 6172960,
      "step": 9470
    },
    {
      "epoch": 4.965932914046122,
      "grad_norm": 0.8604191541671753,
      "learning_rate": 4.672570634569056e-05,
      "loss": 0.3533,
      "num_input_tokens_seen": 6176832,
      "step": 9475
    },
    {
      "epoch": 4.968553459119497,
      "grad_norm": 0.6770857572555542,
      "learning_rate": 4.672004681853593e-05,
      "loss": 0.3643,
      "num_input_tokens_seen": 6179392,
      "step": 9480
    },
    {
      "epoch": 4.971174004192872,
      "grad_norm": 0.6733262538909912,
      "learning_rate": 4.6714382747795304e-05,
      "loss": 0.3278,
      "num_input_tokens_seen": 6182208,
      "step": 9485
    },
    {
      "epoch": 4.973794549266247,
      "grad_norm": 0.9659993648529053,
      "learning_rate": 4.670871413465354e-05,
      "loss": 0.4258,
      "num_input_tokens_seen": 6185376,
      "step": 9490
    },
    {
      "epoch": 4.976415094339623,
      "grad_norm": 0.7661151885986328,
      "learning_rate": 4.670304098029645e-05,
      "loss": 0.3125,
      "num_input_tokens_seen": 6188896,
      "step": 9495
    },
    {
      "epoch": 4.979035639412998,
      "grad_norm": 1.4669857025146484,
      "learning_rate": 4.669736328591078e-05,
      "loss": 0.3697,
      "num_input_tokens_seen": 6191840,
      "step": 9500
    },
    {
      "epoch": 4.981656184486373,
      "grad_norm": 1.0389164686203003,
      "learning_rate": 4.669168105268424e-05,
      "loss": 0.3244,
      "num_input_tokens_seen": 6194784,
      "step": 9505
    },
    {
      "epoch": 4.984276729559748,
      "grad_norm": 1.767783522605896,
      "learning_rate": 4.6685994281805514e-05,
      "loss": 0.2687,
      "num_input_tokens_seen": 6198048,
      "step": 9510
    },
    {
      "epoch": 4.986897274633124,
      "grad_norm": 2.3912317752838135,
      "learning_rate": 4.6680302974464175e-05,
      "loss": 0.3458,
      "num_input_tokens_seen": 6202688,
      "step": 9515
    },
    {
      "epoch": 4.989517819706499,
      "grad_norm": 0.9871980547904968,
      "learning_rate": 4.66746071318508e-05,
      "loss": 0.4976,
      "num_input_tokens_seen": 6205952,
      "step": 9520
    },
    {
      "epoch": 4.9921383647798745,
      "grad_norm": 1.0321738719940186,
      "learning_rate": 4.6668906755156896e-05,
      "loss": 0.2755,
      "num_input_tokens_seen": 6209504,
      "step": 9525
    },
    {
      "epoch": 4.99475890985325,
      "grad_norm": 0.8485603332519531,
      "learning_rate": 4.6663201845574906e-05,
      "loss": 0.4095,
      "num_input_tokens_seen": 6212928,
      "step": 9530
    },
    {
      "epoch": 4.997379454926625,
      "grad_norm": 1.158661127090454,
      "learning_rate": 4.6657492404298234e-05,
      "loss": 0.3247,
      "num_input_tokens_seen": 6216672,
      "step": 9535
    },
    {
      "epoch": 5.0,
      "grad_norm": 4.78749418258667,
      "learning_rate": 4.665177843252124e-05,
      "loss": 0.4908,
      "num_input_tokens_seen": 6219592,
      "step": 9540
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.47002243995666504,
      "eval_runtime": 11.8241,
      "eval_samples_per_second": 71.718,
      "eval_steps_per_second": 17.929,
      "num_input_tokens_seen": 6219592,
      "step": 9540
    },
    {
      "epoch": 5.002620545073375,
      "grad_norm": 1.9964544773101807,
      "learning_rate": 4.664605993143921e-05,
      "loss": 0.3233,
      "num_input_tokens_seen": 6222216,
      "step": 9545
    },
    {
      "epoch": 5.00524109014675,
      "grad_norm": 0.9543737173080444,
      "learning_rate": 4.6640336902248396e-05,
      "loss": 0.3182,
      "num_input_tokens_seen": 6224872,
      "step": 9550
    },
    {
      "epoch": 5.0078616352201255,
      "grad_norm": 0.7479839324951172,
      "learning_rate": 4.6634609346145984e-05,
      "loss": 0.318,
      "num_input_tokens_seen": 6228072,
      "step": 9555
    },
    {
      "epoch": 5.010482180293501,
      "grad_norm": 1.2307790517807007,
      "learning_rate": 4.6628877264330116e-05,
      "loss": 0.3605,
      "num_input_tokens_seen": 6230984,
      "step": 9560
    },
    {
      "epoch": 5.013102725366877,
      "grad_norm": 1.4300060272216797,
      "learning_rate": 4.6623140657999886e-05,
      "loss": 0.3634,
      "num_input_tokens_seen": 6234056,
      "step": 9565
    },
    {
      "epoch": 5.015723270440252,
      "grad_norm": 0.8717216849327087,
      "learning_rate": 4.661739952835532e-05,
      "loss": 0.2107,
      "num_input_tokens_seen": 6237800,
      "step": 9570
    },
    {
      "epoch": 5.018343815513627,
      "grad_norm": 1.2599480152130127,
      "learning_rate": 4.66116538765974e-05,
      "loss": 0.4547,
      "num_input_tokens_seen": 6240808,
      "step": 9575
    },
    {
      "epoch": 5.020964360587002,
      "grad_norm": 1.324265480041504,
      "learning_rate": 4.6605903703928034e-05,
      "loss": 0.3098,
      "num_input_tokens_seen": 6244040,
      "step": 9580
    },
    {
      "epoch": 5.023584905660377,
      "grad_norm": 1.5132921934127808,
      "learning_rate": 4.6600149011550116e-05,
      "loss": 0.3682,
      "num_input_tokens_seen": 6246952,
      "step": 9585
    },
    {
      "epoch": 5.026205450733753,
      "grad_norm": 0.8215728998184204,
      "learning_rate": 4.659438980066745e-05,
      "loss": 0.3418,
      "num_input_tokens_seen": 6250472,
      "step": 9590
    },
    {
      "epoch": 5.028825995807128,
      "grad_norm": 0.9383131861686707,
      "learning_rate": 4.65886260724848e-05,
      "loss": 0.2926,
      "num_input_tokens_seen": 6253640,
      "step": 9595
    },
    {
      "epoch": 5.031446540880503,
      "grad_norm": 1.634689450263977,
      "learning_rate": 4.658285782820787e-05,
      "loss": 0.3047,
      "num_input_tokens_seen": 6257160,
      "step": 9600
    },
    {
      "epoch": 5.034067085953878,
      "grad_norm": 1.11027991771698,
      "learning_rate": 4.657708506904331e-05,
      "loss": 0.299,
      "num_input_tokens_seen": 6259688,
      "step": 9605
    },
    {
      "epoch": 5.036687631027253,
      "grad_norm": 1.4213694334030151,
      "learning_rate": 4.6571307796198724e-05,
      "loss": 0.3138,
      "num_input_tokens_seen": 6263080,
      "step": 9610
    },
    {
      "epoch": 5.039308176100629,
      "grad_norm": 0.6509796977043152,
      "learning_rate": 4.656552601088265e-05,
      "loss": 0.1984,
      "num_input_tokens_seen": 6269320,
      "step": 9615
    },
    {
      "epoch": 5.0419287211740045,
      "grad_norm": 1.0874027013778687,
      "learning_rate": 4.655973971430456e-05,
      "loss": 0.3192,
      "num_input_tokens_seen": 6272392,
      "step": 9620
    },
    {
      "epoch": 5.04454926624738,
      "grad_norm": 2.5805563926696777,
      "learning_rate": 4.6553948907674895e-05,
      "loss": 0.3853,
      "num_input_tokens_seen": 6275496,
      "step": 9625
    },
    {
      "epoch": 5.047169811320755,
      "grad_norm": 1.1560885906219482,
      "learning_rate": 4.654815359220501e-05,
      "loss": 0.2555,
      "num_input_tokens_seen": 6277960,
      "step": 9630
    },
    {
      "epoch": 5.04979035639413,
      "grad_norm": 0.9034286737442017,
      "learning_rate": 4.6542353769107226e-05,
      "loss": 0.6377,
      "num_input_tokens_seen": 6280520,
      "step": 9635
    },
    {
      "epoch": 5.052410901467505,
      "grad_norm": 1.451255440711975,
      "learning_rate": 4.653654943959481e-05,
      "loss": 0.2956,
      "num_input_tokens_seen": 6283304,
      "step": 9640
    },
    {
      "epoch": 5.05503144654088,
      "grad_norm": 1.6333487033843994,
      "learning_rate": 4.653074060488195e-05,
      "loss": 0.3868,
      "num_input_tokens_seen": 6286536,
      "step": 9645
    },
    {
      "epoch": 5.0576519916142555,
      "grad_norm": 1.3437552452087402,
      "learning_rate": 4.6524927266183796e-05,
      "loss": 0.315,
      "num_input_tokens_seen": 6289928,
      "step": 9650
    },
    {
      "epoch": 5.060272536687631,
      "grad_norm": 1.1394485235214233,
      "learning_rate": 4.6519109424716426e-05,
      "loss": 0.3601,
      "num_input_tokens_seen": 6293416,
      "step": 9655
    },
    {
      "epoch": 5.062893081761007,
      "grad_norm": 1.1523576974868774,
      "learning_rate": 4.6513287081696855e-05,
      "loss": 0.353,
      "num_input_tokens_seen": 6297320,
      "step": 9660
    },
    {
      "epoch": 5.065513626834382,
      "grad_norm": 1.0094197988510132,
      "learning_rate": 4.6507460238343076e-05,
      "loss": 0.2709,
      "num_input_tokens_seen": 6300008,
      "step": 9665
    },
    {
      "epoch": 5.068134171907757,
      "grad_norm": 0.6025351881980896,
      "learning_rate": 4.6501628895873984e-05,
      "loss": 0.3014,
      "num_input_tokens_seen": 6304136,
      "step": 9670
    },
    {
      "epoch": 5.070754716981132,
      "grad_norm": 2.0820729732513428,
      "learning_rate": 4.649579305550942e-05,
      "loss": 0.393,
      "num_input_tokens_seen": 6306792,
      "step": 9675
    },
    {
      "epoch": 5.073375262054507,
      "grad_norm": 1.583358883857727,
      "learning_rate": 4.64899527184702e-05,
      "loss": 0.3864,
      "num_input_tokens_seen": 6309800,
      "step": 9680
    },
    {
      "epoch": 5.075995807127883,
      "grad_norm": 2.282093048095703,
      "learning_rate": 4.6484107885978035e-05,
      "loss": 0.3512,
      "num_input_tokens_seen": 6312360,
      "step": 9685
    },
    {
      "epoch": 5.078616352201258,
      "grad_norm": 0.939406156539917,
      "learning_rate": 4.64782585592556e-05,
      "loss": 0.3235,
      "num_input_tokens_seen": 6316008,
      "step": 9690
    },
    {
      "epoch": 5.081236897274633,
      "grad_norm": 1.4070496559143066,
      "learning_rate": 4.647240473952652e-05,
      "loss": 0.3007,
      "num_input_tokens_seen": 6318952,
      "step": 9695
    },
    {
      "epoch": 5.083857442348008,
      "grad_norm": 1.5130107402801514,
      "learning_rate": 4.6466546428015336e-05,
      "loss": 0.3224,
      "num_input_tokens_seen": 6322056,
      "step": 9700
    },
    {
      "epoch": 5.086477987421383,
      "grad_norm": 1.2390795946121216,
      "learning_rate": 4.646068362594753e-05,
      "loss": 0.2989,
      "num_input_tokens_seen": 6325000,
      "step": 9705
    },
    {
      "epoch": 5.089098532494759,
      "grad_norm": 1.3182716369628906,
      "learning_rate": 4.645481633454956e-05,
      "loss": 0.2594,
      "num_input_tokens_seen": 6329960,
      "step": 9710
    },
    {
      "epoch": 5.0917190775681345,
      "grad_norm": 1.8424886465072632,
      "learning_rate": 4.644894455504878e-05,
      "loss": 0.2463,
      "num_input_tokens_seen": 6333640,
      "step": 9715
    },
    {
      "epoch": 5.09433962264151,
      "grad_norm": 1.9235484600067139,
      "learning_rate": 4.64430682886735e-05,
      "loss": 0.237,
      "num_input_tokens_seen": 6336488,
      "step": 9720
    },
    {
      "epoch": 5.096960167714885,
      "grad_norm": 1.6840115785598755,
      "learning_rate": 4.643718753665297e-05,
      "loss": 0.3363,
      "num_input_tokens_seen": 6340328,
      "step": 9725
    },
    {
      "epoch": 5.09958071278826,
      "grad_norm": 1.895187258720398,
      "learning_rate": 4.643130230021737e-05,
      "loss": 0.3549,
      "num_input_tokens_seen": 6343464,
      "step": 9730
    },
    {
      "epoch": 5.102201257861635,
      "grad_norm": 1.3950921297073364,
      "learning_rate": 4.6425412580597836e-05,
      "loss": 0.4864,
      "num_input_tokens_seen": 6349352,
      "step": 9735
    },
    {
      "epoch": 5.10482180293501,
      "grad_norm": 1.007415533065796,
      "learning_rate": 4.641951837902643e-05,
      "loss": 0.2565,
      "num_input_tokens_seen": 6351784,
      "step": 9740
    },
    {
      "epoch": 5.1074423480083855,
      "grad_norm": 1.6735148429870605,
      "learning_rate": 4.641361969673614e-05,
      "loss": 0.4213,
      "num_input_tokens_seen": 6355368,
      "step": 9745
    },
    {
      "epoch": 5.110062893081761,
      "grad_norm": 1.1172842979431152,
      "learning_rate": 4.640771653496091e-05,
      "loss": 0.3447,
      "num_input_tokens_seen": 6358312,
      "step": 9750
    },
    {
      "epoch": 5.112683438155136,
      "grad_norm": 1.5682086944580078,
      "learning_rate": 4.64018088949356e-05,
      "loss": 0.3278,
      "num_input_tokens_seen": 6360776,
      "step": 9755
    },
    {
      "epoch": 5.115303983228512,
      "grad_norm": 1.1639374494552612,
      "learning_rate": 4.6395896777896055e-05,
      "loss": 0.3673,
      "num_input_tokens_seen": 6364424,
      "step": 9760
    },
    {
      "epoch": 5.117924528301887,
      "grad_norm": 1.1247674226760864,
      "learning_rate": 4.638998018507899e-05,
      "loss": 0.3246,
      "num_input_tokens_seen": 6367496,
      "step": 9765
    },
    {
      "epoch": 5.120545073375262,
      "grad_norm": 2.1037991046905518,
      "learning_rate": 4.63840591177221e-05,
      "loss": 0.2713,
      "num_input_tokens_seen": 6370472,
      "step": 9770
    },
    {
      "epoch": 5.123165618448637,
      "grad_norm": 0.7721221446990967,
      "learning_rate": 4.637813357706401e-05,
      "loss": 0.2598,
      "num_input_tokens_seen": 6373832,
      "step": 9775
    },
    {
      "epoch": 5.1257861635220126,
      "grad_norm": 2.010204315185547,
      "learning_rate": 4.6372203564344264e-05,
      "loss": 0.3038,
      "num_input_tokens_seen": 6377672,
      "step": 9780
    },
    {
      "epoch": 5.128406708595388,
      "grad_norm": 1.7183772325515747,
      "learning_rate": 4.6366269080803364e-05,
      "loss": 0.308,
      "num_input_tokens_seen": 6380712,
      "step": 9785
    },
    {
      "epoch": 5.131027253668763,
      "grad_norm": 1.7281111478805542,
      "learning_rate": 4.636033012768273e-05,
      "loss": 0.3171,
      "num_input_tokens_seen": 6384776,
      "step": 9790
    },
    {
      "epoch": 5.133647798742138,
      "grad_norm": 1.9799747467041016,
      "learning_rate": 4.635438670622473e-05,
      "loss": 0.3212,
      "num_input_tokens_seen": 6387560,
      "step": 9795
    },
    {
      "epoch": 5.136268343815513,
      "grad_norm": 1.3516687154769897,
      "learning_rate": 4.6348438817672646e-05,
      "loss": 0.4025,
      "num_input_tokens_seen": 6390952,
      "step": 9800
    },
    {
      "epoch": 5.138888888888889,
      "grad_norm": 0.9197382926940918,
      "learning_rate": 4.634248646327072e-05,
      "loss": 0.3607,
      "num_input_tokens_seen": 6394088,
      "step": 9805
    },
    {
      "epoch": 5.1415094339622645,
      "grad_norm": 0.7118512392044067,
      "learning_rate": 4.6336529644264114e-05,
      "loss": 0.2604,
      "num_input_tokens_seen": 6397064,
      "step": 9810
    },
    {
      "epoch": 5.14412997903564,
      "grad_norm": 3.09429669380188,
      "learning_rate": 4.633056836189893e-05,
      "loss": 0.4306,
      "num_input_tokens_seen": 6400584,
      "step": 9815
    },
    {
      "epoch": 5.146750524109015,
      "grad_norm": 1.2641615867614746,
      "learning_rate": 4.632460261742219e-05,
      "loss": 0.2548,
      "num_input_tokens_seen": 6403976,
      "step": 9820
    },
    {
      "epoch": 5.14937106918239,
      "grad_norm": 1.5663002729415894,
      "learning_rate": 4.631863241208187e-05,
      "loss": 0.3362,
      "num_input_tokens_seen": 6406536,
      "step": 9825
    },
    {
      "epoch": 5.151991614255765,
      "grad_norm": 0.8604857325553894,
      "learning_rate": 4.6312657747126865e-05,
      "loss": 0.337,
      "num_input_tokens_seen": 6410312,
      "step": 9830
    },
    {
      "epoch": 5.15461215932914,
      "grad_norm": 1.637934923171997,
      "learning_rate": 4.630667862380701e-05,
      "loss": 0.2614,
      "num_input_tokens_seen": 6414184,
      "step": 9835
    },
    {
      "epoch": 5.1572327044025155,
      "grad_norm": 1.9471328258514404,
      "learning_rate": 4.630069504337306e-05,
      "loss": 0.5716,
      "num_input_tokens_seen": 6417224,
      "step": 9840
    },
    {
      "epoch": 5.159853249475891,
      "grad_norm": 1.1075913906097412,
      "learning_rate": 4.629470700707672e-05,
      "loss": 0.3987,
      "num_input_tokens_seen": 6421128,
      "step": 9845
    },
    {
      "epoch": 5.162473794549266,
      "grad_norm": 1.0821176767349243,
      "learning_rate": 4.628871451617062e-05,
      "loss": 0.277,
      "num_input_tokens_seen": 6424232,
      "step": 9850
    },
    {
      "epoch": 5.165094339622642,
      "grad_norm": 1.600584864616394,
      "learning_rate": 4.628271757190832e-05,
      "loss": 0.3418,
      "num_input_tokens_seen": 6427464,
      "step": 9855
    },
    {
      "epoch": 5.167714884696017,
      "grad_norm": 1.1336495876312256,
      "learning_rate": 4.62767161755443e-05,
      "loss": 0.2648,
      "num_input_tokens_seen": 6430504,
      "step": 9860
    },
    {
      "epoch": 5.170335429769392,
      "grad_norm": 1.3027011156082153,
      "learning_rate": 4.6270710328334004e-05,
      "loss": 0.3387,
      "num_input_tokens_seen": 6434056,
      "step": 9865
    },
    {
      "epoch": 5.172955974842767,
      "grad_norm": 1.2232093811035156,
      "learning_rate": 4.626470003153377e-05,
      "loss": 0.4623,
      "num_input_tokens_seen": 6437256,
      "step": 9870
    },
    {
      "epoch": 5.1755765199161425,
      "grad_norm": 2.6275484561920166,
      "learning_rate": 4.625868528640089e-05,
      "loss": 0.3147,
      "num_input_tokens_seen": 6441064,
      "step": 9875
    },
    {
      "epoch": 5.178197064989518,
      "grad_norm": 0.7722792625427246,
      "learning_rate": 4.6252666094193586e-05,
      "loss": 0.2942,
      "num_input_tokens_seen": 6444648,
      "step": 9880
    },
    {
      "epoch": 5.180817610062893,
      "grad_norm": 1.3471988439559937,
      "learning_rate": 4.6246642456171e-05,
      "loss": 0.351,
      "num_input_tokens_seen": 6447880,
      "step": 9885
    },
    {
      "epoch": 5.183438155136268,
      "grad_norm": 0.837742269039154,
      "learning_rate": 4.6240614373593196e-05,
      "loss": 0.3324,
      "num_input_tokens_seen": 6450728,
      "step": 9890
    },
    {
      "epoch": 5.186058700209643,
      "grad_norm": 1.0411025285720825,
      "learning_rate": 4.623458184772119e-05,
      "loss": 0.231,
      "num_input_tokens_seen": 6453640,
      "step": 9895
    },
    {
      "epoch": 5.188679245283019,
      "grad_norm": 1.1204572916030884,
      "learning_rate": 4.622854487981693e-05,
      "loss": 0.2594,
      "num_input_tokens_seen": 6456840,
      "step": 9900
    },
    {
      "epoch": 5.191299790356394,
      "grad_norm": 1.2837477922439575,
      "learning_rate": 4.622250347114326e-05,
      "loss": 0.2859,
      "num_input_tokens_seen": 6459848,
      "step": 9905
    },
    {
      "epoch": 5.19392033542977,
      "grad_norm": 2.0384533405303955,
      "learning_rate": 4.621645762296399e-05,
      "loss": 0.3763,
      "num_input_tokens_seen": 6463624,
      "step": 9910
    },
    {
      "epoch": 5.196540880503145,
      "grad_norm": 1.4995008707046509,
      "learning_rate": 4.6210407336543835e-05,
      "loss": 0.2676,
      "num_input_tokens_seen": 6467176,
      "step": 9915
    },
    {
      "epoch": 5.19916142557652,
      "grad_norm": 1.5130863189697266,
      "learning_rate": 4.620435261314844e-05,
      "loss": 0.3576,
      "num_input_tokens_seen": 6469896,
      "step": 9920
    },
    {
      "epoch": 5.201781970649895,
      "grad_norm": 1.1639512777328491,
      "learning_rate": 4.61982934540444e-05,
      "loss": 0.3694,
      "num_input_tokens_seen": 6472904,
      "step": 9925
    },
    {
      "epoch": 5.20440251572327,
      "grad_norm": 0.7639169096946716,
      "learning_rate": 4.6192229860499206e-05,
      "loss": 0.3274,
      "num_input_tokens_seen": 6477288,
      "step": 9930
    },
    {
      "epoch": 5.2070230607966455,
      "grad_norm": 0.9793940782546997,
      "learning_rate": 4.6186161833781295e-05,
      "loss": 0.2986,
      "num_input_tokens_seen": 6480488,
      "step": 9935
    },
    {
      "epoch": 5.209643605870021,
      "grad_norm": 1.167747139930725,
      "learning_rate": 4.618008937516003e-05,
      "loss": 0.2357,
      "num_input_tokens_seen": 6483368,
      "step": 9940
    },
    {
      "epoch": 5.212264150943396,
      "grad_norm": 1.8169198036193848,
      "learning_rate": 4.617401248590572e-05,
      "loss": 0.3551,
      "num_input_tokens_seen": 6485928,
      "step": 9945
    },
    {
      "epoch": 5.214884696016772,
      "grad_norm": 1.0998082160949707,
      "learning_rate": 4.616793116728955e-05,
      "loss": 0.3528,
      "num_input_tokens_seen": 6489768,
      "step": 9950
    },
    {
      "epoch": 5.217505241090147,
      "grad_norm": 1.4181941747665405,
      "learning_rate": 4.6161845420583685e-05,
      "loss": 0.3639,
      "num_input_tokens_seen": 6493352,
      "step": 9955
    },
    {
      "epoch": 5.220125786163522,
      "grad_norm": 1.9535163640975952,
      "learning_rate": 4.6155755247061174e-05,
      "loss": 0.308,
      "num_input_tokens_seen": 6495944,
      "step": 9960
    },
    {
      "epoch": 5.222746331236897,
      "grad_norm": 1.2738518714904785,
      "learning_rate": 4.614966064799603e-05,
      "loss": 0.3142,
      "num_input_tokens_seen": 6498472,
      "step": 9965
    },
    {
      "epoch": 5.2253668763102725,
      "grad_norm": 1.5668439865112305,
      "learning_rate": 4.6143561624663166e-05,
      "loss": 0.4005,
      "num_input_tokens_seen": 6502088,
      "step": 9970
    },
    {
      "epoch": 5.227987421383648,
      "grad_norm": 0.9069252610206604,
      "learning_rate": 4.6137458178338425e-05,
      "loss": 0.2499,
      "num_input_tokens_seen": 6505032,
      "step": 9975
    },
    {
      "epoch": 5.230607966457023,
      "grad_norm": 1.5895514488220215,
      "learning_rate": 4.613135031029859e-05,
      "loss": 0.2374,
      "num_input_tokens_seen": 6508488,
      "step": 9980
    },
    {
      "epoch": 5.233228511530398,
      "grad_norm": 1.4269829988479614,
      "learning_rate": 4.612523802182134e-05,
      "loss": 0.311,
      "num_input_tokens_seen": 6510984,
      "step": 9985
    },
    {
      "epoch": 5.235849056603773,
      "grad_norm": 2.2087831497192383,
      "learning_rate": 4.611912131418531e-05,
      "loss": 0.4038,
      "num_input_tokens_seen": 6514376,
      "step": 9990
    },
    {
      "epoch": 5.238469601677149,
      "grad_norm": 1.2325485944747925,
      "learning_rate": 4.611300018867004e-05,
      "loss": 0.3777,
      "num_input_tokens_seen": 6517192,
      "step": 9995
    },
    {
      "epoch": 5.241090146750524,
      "grad_norm": 4.315021514892578,
      "learning_rate": 4.6106874646556e-05,
      "loss": 0.2631,
      "num_input_tokens_seen": 6520424,
      "step": 10000
    },
    {
      "epoch": 5.2437106918239,
      "grad_norm": 0.598301351070404,
      "learning_rate": 4.6100744689124584e-05,
      "loss": 0.1499,
      "num_input_tokens_seen": 6523976,
      "step": 10005
    },
    {
      "epoch": 5.246331236897275,
      "grad_norm": 1.260893702507019,
      "learning_rate": 4.6094610317658116e-05,
      "loss": 0.2382,
      "num_input_tokens_seen": 6526408,
      "step": 10010
    },
    {
      "epoch": 5.24895178197065,
      "grad_norm": 2.123931646347046,
      "learning_rate": 4.608847153343982e-05,
      "loss": 0.3573,
      "num_input_tokens_seen": 6529288,
      "step": 10015
    },
    {
      "epoch": 5.251572327044025,
      "grad_norm": 1.5394411087036133,
      "learning_rate": 4.6082328337753874e-05,
      "loss": 0.3518,
      "num_input_tokens_seen": 6532072,
      "step": 10020
    },
    {
      "epoch": 5.2541928721174,
      "grad_norm": 4.442552089691162,
      "learning_rate": 4.6076180731885353e-05,
      "loss": 0.2667,
      "num_input_tokens_seen": 6536520,
      "step": 10025
    },
    {
      "epoch": 5.256813417190775,
      "grad_norm": 2.5205471515655518,
      "learning_rate": 4.6070028717120286e-05,
      "loss": 0.3581,
      "num_input_tokens_seen": 6539880,
      "step": 10030
    },
    {
      "epoch": 5.259433962264151,
      "grad_norm": 6.978264808654785,
      "learning_rate": 4.606387229474558e-05,
      "loss": 0.3163,
      "num_input_tokens_seen": 6542632,
      "step": 10035
    },
    {
      "epoch": 5.262054507337526,
      "grad_norm": 1.9724252223968506,
      "learning_rate": 4.60577114660491e-05,
      "loss": 0.2771,
      "num_input_tokens_seen": 6545416,
      "step": 10040
    },
    {
      "epoch": 5.264675052410902,
      "grad_norm": 1.048106074333191,
      "learning_rate": 4.605154623231962e-05,
      "loss": 0.2706,
      "num_input_tokens_seen": 6548680,
      "step": 10045
    },
    {
      "epoch": 5.267295597484277,
      "grad_norm": 1.4493136405944824,
      "learning_rate": 4.604537659484684e-05,
      "loss": 0.3259,
      "num_input_tokens_seen": 6552744,
      "step": 10050
    },
    {
      "epoch": 5.269916142557652,
      "grad_norm": 1.1528898477554321,
      "learning_rate": 4.603920255492138e-05,
      "loss": 0.241,
      "num_input_tokens_seen": 6557128,
      "step": 10055
    },
    {
      "epoch": 5.272536687631027,
      "grad_norm": 1.1468499898910522,
      "learning_rate": 4.603302411383477e-05,
      "loss": 0.3578,
      "num_input_tokens_seen": 6565352,
      "step": 10060
    },
    {
      "epoch": 5.2751572327044025,
      "grad_norm": 1.203147292137146,
      "learning_rate": 4.602684127287946e-05,
      "loss": 0.3338,
      "num_input_tokens_seen": 6568680,
      "step": 10065
    },
    {
      "epoch": 5.277777777777778,
      "grad_norm": 1.2180508375167847,
      "learning_rate": 4.602065403334885e-05,
      "loss": 0.2572,
      "num_input_tokens_seen": 6572520,
      "step": 10070
    },
    {
      "epoch": 5.280398322851153,
      "grad_norm": 1.11844801902771,
      "learning_rate": 4.601446239653724e-05,
      "loss": 0.3708,
      "num_input_tokens_seen": 6576232,
      "step": 10075
    },
    {
      "epoch": 5.283018867924528,
      "grad_norm": 1.9403374195098877,
      "learning_rate": 4.600826636373984e-05,
      "loss": 0.3156,
      "num_input_tokens_seen": 6579016,
      "step": 10080
    },
    {
      "epoch": 5.285639412997903,
      "grad_norm": 4.304579734802246,
      "learning_rate": 4.600206593625279e-05,
      "loss": 0.3497,
      "num_input_tokens_seen": 6581864,
      "step": 10085
    },
    {
      "epoch": 5.288259958071279,
      "grad_norm": 0.8834447860717773,
      "learning_rate": 4.599586111537315e-05,
      "loss": 0.3649,
      "num_input_tokens_seen": 6585256,
      "step": 10090
    },
    {
      "epoch": 5.290880503144654,
      "grad_norm": 0.8929767608642578,
      "learning_rate": 4.59896519023989e-05,
      "loss": 0.3712,
      "num_input_tokens_seen": 6587976,
      "step": 10095
    },
    {
      "epoch": 5.29350104821803,
      "grad_norm": 1.167303204536438,
      "learning_rate": 4.5983438298628935e-05,
      "loss": 0.4328,
      "num_input_tokens_seen": 6591080,
      "step": 10100
    },
    {
      "epoch": 5.296121593291405,
      "grad_norm": 1.3566339015960693,
      "learning_rate": 4.597722030536306e-05,
      "loss": 0.2996,
      "num_input_tokens_seen": 6594088,
      "step": 10105
    },
    {
      "epoch": 5.29874213836478,
      "grad_norm": 1.4358527660369873,
      "learning_rate": 4.597099792390202e-05,
      "loss": 0.2417,
      "num_input_tokens_seen": 6597544,
      "step": 10110
    },
    {
      "epoch": 5.301362683438155,
      "grad_norm": 1.483044981956482,
      "learning_rate": 4.596477115554748e-05,
      "loss": 0.3177,
      "num_input_tokens_seen": 6600552,
      "step": 10115
    },
    {
      "epoch": 5.30398322851153,
      "grad_norm": 1.9273176193237305,
      "learning_rate": 4.5958540001601976e-05,
      "loss": 0.3336,
      "num_input_tokens_seen": 6602888,
      "step": 10120
    },
    {
      "epoch": 5.306603773584905,
      "grad_norm": 1.4735335111618042,
      "learning_rate": 4.595230446336902e-05,
      "loss": 0.3188,
      "num_input_tokens_seen": 6606344,
      "step": 10125
    },
    {
      "epoch": 5.309224318658281,
      "grad_norm": 3.0081095695495605,
      "learning_rate": 4.5946064542153005e-05,
      "loss": 0.3685,
      "num_input_tokens_seen": 6609864,
      "step": 10130
    },
    {
      "epoch": 5.311844863731656,
      "grad_norm": 1.2469996213912964,
      "learning_rate": 4.593982023925926e-05,
      "loss": 0.2526,
      "num_input_tokens_seen": 6612616,
      "step": 10135
    },
    {
      "epoch": 5.314465408805032,
      "grad_norm": 1.3762998580932617,
      "learning_rate": 4.5933571555994e-05,
      "loss": 0.253,
      "num_input_tokens_seen": 6616072,
      "step": 10140
    },
    {
      "epoch": 5.317085953878407,
      "grad_norm": 1.2619081735610962,
      "learning_rate": 4.59273184936644e-05,
      "loss": 0.4154,
      "num_input_tokens_seen": 6619528,
      "step": 10145
    },
    {
      "epoch": 5.319706498951782,
      "grad_norm": 1.0290052890777588,
      "learning_rate": 4.5921061053578523e-05,
      "loss": 0.3076,
      "num_input_tokens_seen": 6622824,
      "step": 10150
    },
    {
      "epoch": 5.322327044025157,
      "grad_norm": 1.8090535402297974,
      "learning_rate": 4.5914799237045355e-05,
      "loss": 0.3208,
      "num_input_tokens_seen": 6626152,
      "step": 10155
    },
    {
      "epoch": 5.3249475890985325,
      "grad_norm": 1.5071126222610474,
      "learning_rate": 4.5908533045374796e-05,
      "loss": 0.3547,
      "num_input_tokens_seen": 6628808,
      "step": 10160
    },
    {
      "epoch": 5.327568134171908,
      "grad_norm": 1.2078396081924438,
      "learning_rate": 4.590226247987766e-05,
      "loss": 0.3326,
      "num_input_tokens_seen": 6632296,
      "step": 10165
    },
    {
      "epoch": 5.330188679245283,
      "grad_norm": 1.112053394317627,
      "learning_rate": 4.589598754186568e-05,
      "loss": 0.3583,
      "num_input_tokens_seen": 6635400,
      "step": 10170
    },
    {
      "epoch": 5.332809224318658,
      "grad_norm": 1.647094488143921,
      "learning_rate": 4.5889708232651494e-05,
      "loss": 0.3665,
      "num_input_tokens_seen": 6639432,
      "step": 10175
    },
    {
      "epoch": 5.335429769392033,
      "grad_norm": 0.908273458480835,
      "learning_rate": 4.588342455354867e-05,
      "loss": 0.3361,
      "num_input_tokens_seen": 6643336,
      "step": 10180
    },
    {
      "epoch": 5.338050314465409,
      "grad_norm": 1.3065992593765259,
      "learning_rate": 4.5877136505871685e-05,
      "loss": 0.3521,
      "num_input_tokens_seen": 6646088,
      "step": 10185
    },
    {
      "epoch": 5.340670859538784,
      "grad_norm": 1.4400593042373657,
      "learning_rate": 4.5870844090935914e-05,
      "loss": 0.2127,
      "num_input_tokens_seen": 6649384,
      "step": 10190
    },
    {
      "epoch": 5.34329140461216,
      "grad_norm": 1.7831231355667114,
      "learning_rate": 4.586454731005768e-05,
      "loss": 0.2862,
      "num_input_tokens_seen": 6652424,
      "step": 10195
    },
    {
      "epoch": 5.345911949685535,
      "grad_norm": 1.3351590633392334,
      "learning_rate": 4.5858246164554175e-05,
      "loss": 0.3112,
      "num_input_tokens_seen": 6655464,
      "step": 10200
    },
    {
      "epoch": 5.34853249475891,
      "grad_norm": 1.5068110227584839,
      "learning_rate": 4.5851940655743536e-05,
      "loss": 0.3278,
      "num_input_tokens_seen": 6659112,
      "step": 10205
    },
    {
      "epoch": 5.351153039832285,
      "grad_norm": 1.9628987312316895,
      "learning_rate": 4.5845630784944804e-05,
      "loss": 0.3109,
      "num_input_tokens_seen": 6663304,
      "step": 10210
    },
    {
      "epoch": 5.35377358490566,
      "grad_norm": 1.2249248027801514,
      "learning_rate": 4.583931655347793e-05,
      "loss": 0.2329,
      "num_input_tokens_seen": 6666888,
      "step": 10215
    },
    {
      "epoch": 5.356394129979035,
      "grad_norm": 1.555873990058899,
      "learning_rate": 4.5832997962663785e-05,
      "loss": 0.2494,
      "num_input_tokens_seen": 6670504,
      "step": 10220
    },
    {
      "epoch": 5.359014675052411,
      "grad_norm": 0.8773128390312195,
      "learning_rate": 4.582667501382414e-05,
      "loss": 0.2999,
      "num_input_tokens_seen": 6673832,
      "step": 10225
    },
    {
      "epoch": 5.361635220125786,
      "grad_norm": 2.5201938152313232,
      "learning_rate": 4.582034770828169e-05,
      "loss": 0.3458,
      "num_input_tokens_seen": 6677160,
      "step": 10230
    },
    {
      "epoch": 5.364255765199162,
      "grad_norm": 1.1089138984680176,
      "learning_rate": 4.581401604736002e-05,
      "loss": 0.327,
      "num_input_tokens_seen": 6681160,
      "step": 10235
    },
    {
      "epoch": 5.366876310272537,
      "grad_norm": 1.312789797782898,
      "learning_rate": 4.580768003238365e-05,
      "loss": 0.3046,
      "num_input_tokens_seen": 6684424,
      "step": 10240
    },
    {
      "epoch": 5.369496855345912,
      "grad_norm": 1.728586196899414,
      "learning_rate": 4.5801339664678014e-05,
      "loss": 0.2967,
      "num_input_tokens_seen": 6687528,
      "step": 10245
    },
    {
      "epoch": 5.372117400419287,
      "grad_norm": 3.470937967300415,
      "learning_rate": 4.579499494556943e-05,
      "loss": 0.3298,
      "num_input_tokens_seen": 6690120,
      "step": 10250
    },
    {
      "epoch": 5.3747379454926625,
      "grad_norm": 2.058906078338623,
      "learning_rate": 4.5788645876385147e-05,
      "loss": 0.3852,
      "num_input_tokens_seen": 6693992,
      "step": 10255
    },
    {
      "epoch": 5.377358490566038,
      "grad_norm": 2.0742669105529785,
      "learning_rate": 4.578229245845331e-05,
      "loss": 0.3125,
      "num_input_tokens_seen": 6698568,
      "step": 10260
    },
    {
      "epoch": 5.379979035639413,
      "grad_norm": 1.5375016927719116,
      "learning_rate": 4.5775934693102994e-05,
      "loss": 0.2986,
      "num_input_tokens_seen": 6701320,
      "step": 10265
    },
    {
      "epoch": 5.382599580712788,
      "grad_norm": 1.2372561693191528,
      "learning_rate": 4.5769572581664165e-05,
      "loss": 0.265,
      "num_input_tokens_seen": 6704168,
      "step": 10270
    },
    {
      "epoch": 5.385220125786163,
      "grad_norm": 1.057948350906372,
      "learning_rate": 4.57632061254677e-05,
      "loss": 0.3468,
      "num_input_tokens_seen": 6707688,
      "step": 10275
    },
    {
      "epoch": 5.387840670859539,
      "grad_norm": 6.953103065490723,
      "learning_rate": 4.5756835325845394e-05,
      "loss": 0.3118,
      "num_input_tokens_seen": 6711048,
      "step": 10280
    },
    {
      "epoch": 5.390461215932914,
      "grad_norm": 3.1243443489074707,
      "learning_rate": 4.575046018412995e-05,
      "loss": 0.3032,
      "num_input_tokens_seen": 6715016,
      "step": 10285
    },
    {
      "epoch": 5.3930817610062896,
      "grad_norm": 2.3201074600219727,
      "learning_rate": 4.5744080701654955e-05,
      "loss": 0.3133,
      "num_input_tokens_seen": 6717448,
      "step": 10290
    },
    {
      "epoch": 5.395702306079665,
      "grad_norm": 1.2550439834594727,
      "learning_rate": 4.573769687975495e-05,
      "loss": 0.2351,
      "num_input_tokens_seen": 6721032,
      "step": 10295
    },
    {
      "epoch": 5.39832285115304,
      "grad_norm": 1.2190262079238892,
      "learning_rate": 4.573130871976534e-05,
      "loss": 0.2825,
      "num_input_tokens_seen": 6723304,
      "step": 10300
    },
    {
      "epoch": 5.400943396226415,
      "grad_norm": 1.360213041305542,
      "learning_rate": 4.572491622302247e-05,
      "loss": 0.2808,
      "num_input_tokens_seen": 6726344,
      "step": 10305
    },
    {
      "epoch": 5.40356394129979,
      "grad_norm": 1.7505428791046143,
      "learning_rate": 4.571851939086356e-05,
      "loss": 0.3163,
      "num_input_tokens_seen": 6730248,
      "step": 10310
    },
    {
      "epoch": 5.406184486373165,
      "grad_norm": 1.4122819900512695,
      "learning_rate": 4.571211822462676e-05,
      "loss": 0.339,
      "num_input_tokens_seen": 6733352,
      "step": 10315
    },
    {
      "epoch": 5.408805031446541,
      "grad_norm": 1.9480912685394287,
      "learning_rate": 4.570571272565114e-05,
      "loss": 0.2448,
      "num_input_tokens_seen": 6736104,
      "step": 10320
    },
    {
      "epoch": 5.411425576519916,
      "grad_norm": 1.0833966732025146,
      "learning_rate": 4.569930289527663e-05,
      "loss": 0.3373,
      "num_input_tokens_seen": 6739336,
      "step": 10325
    },
    {
      "epoch": 5.414046121593292,
      "grad_norm": 1.2635811567306519,
      "learning_rate": 4.56928887348441e-05,
      "loss": 0.3066,
      "num_input_tokens_seen": 6742216,
      "step": 10330
    },
    {
      "epoch": 5.416666666666667,
      "grad_norm": 1.1482497453689575,
      "learning_rate": 4.568647024569533e-05,
      "loss": 0.2942,
      "num_input_tokens_seen": 6745320,
      "step": 10335
    },
    {
      "epoch": 5.419287211740042,
      "grad_norm": 0.9567716121673584,
      "learning_rate": 4.568004742917299e-05,
      "loss": 0.212,
      "num_input_tokens_seen": 6748328,
      "step": 10340
    },
    {
      "epoch": 5.421907756813417,
      "grad_norm": 1.649071455001831,
      "learning_rate": 4.567362028662065e-05,
      "loss": 0.3375,
      "num_input_tokens_seen": 6751240,
      "step": 10345
    },
    {
      "epoch": 5.4245283018867925,
      "grad_norm": 0.8842084407806396,
      "learning_rate": 4.566718881938281e-05,
      "loss": 0.2706,
      "num_input_tokens_seen": 6754344,
      "step": 10350
    },
    {
      "epoch": 5.427148846960168,
      "grad_norm": 1.4915287494659424,
      "learning_rate": 4.5660753028804846e-05,
      "loss": 0.2921,
      "num_input_tokens_seen": 6757320,
      "step": 10355
    },
    {
      "epoch": 5.429769392033543,
      "grad_norm": 1.310112714767456,
      "learning_rate": 4.5654312916233066e-05,
      "loss": 0.2998,
      "num_input_tokens_seen": 6760744,
      "step": 10360
    },
    {
      "epoch": 5.432389937106918,
      "grad_norm": 1.1327670812606812,
      "learning_rate": 4.564786848301466e-05,
      "loss": 0.3688,
      "num_input_tokens_seen": 6763848,
      "step": 10365
    },
    {
      "epoch": 5.435010482180293,
      "grad_norm": 1.84004545211792,
      "learning_rate": 4.564141973049773e-05,
      "loss": 0.3521,
      "num_input_tokens_seen": 6766792,
      "step": 10370
    },
    {
      "epoch": 5.437631027253669,
      "grad_norm": 1.3951002359390259,
      "learning_rate": 4.563496666003128e-05,
      "loss": 0.2847,
      "num_input_tokens_seen": 6770024,
      "step": 10375
    },
    {
      "epoch": 5.440251572327044,
      "grad_norm": 2.391309976577759,
      "learning_rate": 4.562850927296523e-05,
      "loss": 0.3664,
      "num_input_tokens_seen": 6772584,
      "step": 10380
    },
    {
      "epoch": 5.4428721174004195,
      "grad_norm": 1.2996902465820312,
      "learning_rate": 4.5622047570650365e-05,
      "loss": 0.302,
      "num_input_tokens_seen": 6776040,
      "step": 10385
    },
    {
      "epoch": 5.445492662473795,
      "grad_norm": 1.3547961711883545,
      "learning_rate": 4.561558155443843e-05,
      "loss": 0.3181,
      "num_input_tokens_seen": 6780616,
      "step": 10390
    },
    {
      "epoch": 5.44811320754717,
      "grad_norm": 1.1878732442855835,
      "learning_rate": 4.560911122568202e-05,
      "loss": 0.3631,
      "num_input_tokens_seen": 6783720,
      "step": 10395
    },
    {
      "epoch": 5.450733752620545,
      "grad_norm": 1.0792449712753296,
      "learning_rate": 4.560263658573467e-05,
      "loss": 0.318,
      "num_input_tokens_seen": 6786824,
      "step": 10400
    },
    {
      "epoch": 5.45335429769392,
      "grad_norm": 1.1233068704605103,
      "learning_rate": 4.559615763595079e-05,
      "loss": 0.3707,
      "num_input_tokens_seen": 6790696,
      "step": 10405
    },
    {
      "epoch": 5.455974842767295,
      "grad_norm": 0.9501178860664368,
      "learning_rate": 4.558967437768571e-05,
      "loss": 0.287,
      "num_input_tokens_seen": 6794408,
      "step": 10410
    },
    {
      "epoch": 5.4585953878406706,
      "grad_norm": 1.596940040588379,
      "learning_rate": 4.558318681229565e-05,
      "loss": 0.3404,
      "num_input_tokens_seen": 6798184,
      "step": 10415
    },
    {
      "epoch": 5.461215932914046,
      "grad_norm": 1.4567989110946655,
      "learning_rate": 4.557669494113773e-05,
      "loss": 0.2852,
      "num_input_tokens_seen": 6801928,
      "step": 10420
    },
    {
      "epoch": 5.463836477987422,
      "grad_norm": 1.1372298002243042,
      "learning_rate": 4.557019876556998e-05,
      "loss": 0.2887,
      "num_input_tokens_seen": 6805640,
      "step": 10425
    },
    {
      "epoch": 5.466457023060797,
      "grad_norm": 1.2761147022247314,
      "learning_rate": 4.556369828695133e-05,
      "loss": 0.2528,
      "num_input_tokens_seen": 6808680,
      "step": 10430
    },
    {
      "epoch": 5.469077568134172,
      "grad_norm": 1.6922760009765625,
      "learning_rate": 4.5557193506641596e-05,
      "loss": 0.3042,
      "num_input_tokens_seen": 6811752,
      "step": 10435
    },
    {
      "epoch": 5.471698113207547,
      "grad_norm": 1.2880842685699463,
      "learning_rate": 4.555068442600151e-05,
      "loss": 0.3423,
      "num_input_tokens_seen": 6814888,
      "step": 10440
    },
    {
      "epoch": 5.4743186582809225,
      "grad_norm": 0.962494432926178,
      "learning_rate": 4.5544171046392703e-05,
      "loss": 0.3127,
      "num_input_tokens_seen": 6819048,
      "step": 10445
    },
    {
      "epoch": 5.476939203354298,
      "grad_norm": 4.131008625030518,
      "learning_rate": 4.5537653369177685e-05,
      "loss": 0.2264,
      "num_input_tokens_seen": 6822792,
      "step": 10450
    },
    {
      "epoch": 5.479559748427673,
      "grad_norm": 0.9893204569816589,
      "learning_rate": 4.5531131395719895e-05,
      "loss": 0.3887,
      "num_input_tokens_seen": 6826248,
      "step": 10455
    },
    {
      "epoch": 5.482180293501048,
      "grad_norm": 2.292343854904175,
      "learning_rate": 4.552460512738364e-05,
      "loss": 0.3057,
      "num_input_tokens_seen": 6830248,
      "step": 10460
    },
    {
      "epoch": 5.484800838574423,
      "grad_norm": 1.1687873601913452,
      "learning_rate": 4.551807456553416e-05,
      "loss": 0.2576,
      "num_input_tokens_seen": 6833160,
      "step": 10465
    },
    {
      "epoch": 5.487421383647799,
      "grad_norm": 1.686363697052002,
      "learning_rate": 4.551153971153755e-05,
      "loss": 0.4477,
      "num_input_tokens_seen": 6836392,
      "step": 10470
    },
    {
      "epoch": 5.490041928721174,
      "grad_norm": 1.0741758346557617,
      "learning_rate": 4.550500056676085e-05,
      "loss": 0.3215,
      "num_input_tokens_seen": 6840872,
      "step": 10475
    },
    {
      "epoch": 5.4926624737945495,
      "grad_norm": 1.2015544176101685,
      "learning_rate": 4.5498457132571966e-05,
      "loss": 0.2885,
      "num_input_tokens_seen": 6843432,
      "step": 10480
    },
    {
      "epoch": 5.495283018867925,
      "grad_norm": 0.907842218875885,
      "learning_rate": 4.5491909410339706e-05,
      "loss": 0.2248,
      "num_input_tokens_seen": 6846760,
      "step": 10485
    },
    {
      "epoch": 5.4979035639413,
      "grad_norm": 1.8864216804504395,
      "learning_rate": 4.548535740143378e-05,
      "loss": 0.4075,
      "num_input_tokens_seen": 6849256,
      "step": 10490
    },
    {
      "epoch": 5.500524109014675,
      "grad_norm": 1.2053743600845337,
      "learning_rate": 4.54788011072248e-05,
      "loss": 0.2901,
      "num_input_tokens_seen": 6853256,
      "step": 10495
    },
    {
      "epoch": 5.50314465408805,
      "grad_norm": 2.0029118061065674,
      "learning_rate": 4.5472240529084254e-05,
      "loss": 0.3597,
      "num_input_tokens_seen": 6859688,
      "step": 10500
    },
    {
      "epoch": 5.505765199161425,
      "grad_norm": 1.863690972328186,
      "learning_rate": 4.5465675668384557e-05,
      "loss": 0.3023,
      "num_input_tokens_seen": 6863080,
      "step": 10505
    },
    {
      "epoch": 5.5083857442348005,
      "grad_norm": 1.386603593826294,
      "learning_rate": 4.545910652649899e-05,
      "loss": 0.3048,
      "num_input_tokens_seen": 6865704,
      "step": 10510
    },
    {
      "epoch": 5.511006289308176,
      "grad_norm": 1.957848072052002,
      "learning_rate": 4.5452533104801756e-05,
      "loss": 0.3562,
      "num_input_tokens_seen": 6868392,
      "step": 10515
    },
    {
      "epoch": 5.513626834381551,
      "grad_norm": 1.6525640487670898,
      "learning_rate": 4.544595540466792e-05,
      "loss": 0.3234,
      "num_input_tokens_seen": 6871624,
      "step": 10520
    },
    {
      "epoch": 5.516247379454927,
      "grad_norm": 1.11869478225708,
      "learning_rate": 4.5439373427473476e-05,
      "loss": 0.2632,
      "num_input_tokens_seen": 6875752,
      "step": 10525
    },
    {
      "epoch": 5.518867924528302,
      "grad_norm": 2.1073336601257324,
      "learning_rate": 4.54327871745953e-05,
      "loss": 0.4245,
      "num_input_tokens_seen": 6878728,
      "step": 10530
    },
    {
      "epoch": 5.521488469601677,
      "grad_norm": 1.3087178468704224,
      "learning_rate": 4.542619664741116e-05,
      "loss": 0.3358,
      "num_input_tokens_seen": 6882408,
      "step": 10535
    },
    {
      "epoch": 5.524109014675052,
      "grad_norm": 1.8070791959762573,
      "learning_rate": 4.5419601847299706e-05,
      "loss": 0.4028,
      "num_input_tokens_seen": 6885448,
      "step": 10540
    },
    {
      "epoch": 5.526729559748428,
      "grad_norm": 1.4978888034820557,
      "learning_rate": 4.5413002775640503e-05,
      "loss": 0.3455,
      "num_input_tokens_seen": 6889192,
      "step": 10545
    },
    {
      "epoch": 5.529350104821803,
      "grad_norm": 1.45793879032135,
      "learning_rate": 4.540639943381401e-05,
      "loss": 0.3417,
      "num_input_tokens_seen": 6892424,
      "step": 10550
    },
    {
      "epoch": 5.531970649895178,
      "grad_norm": 1.2321187257766724,
      "learning_rate": 4.539979182320157e-05,
      "loss": 0.2885,
      "num_input_tokens_seen": 6895368,
      "step": 10555
    },
    {
      "epoch": 5.534591194968553,
      "grad_norm": 1.4002255201339722,
      "learning_rate": 4.5393179945185404e-05,
      "loss": 0.3222,
      "num_input_tokens_seen": 6899784,
      "step": 10560
    },
    {
      "epoch": 5.537211740041929,
      "grad_norm": 2.8137898445129395,
      "learning_rate": 4.538656380114865e-05,
      "loss": 0.3485,
      "num_input_tokens_seen": 6902504,
      "step": 10565
    },
    {
      "epoch": 5.539832285115304,
      "grad_norm": 0.8336761593818665,
      "learning_rate": 4.537994339247533e-05,
      "loss": 0.3387,
      "num_input_tokens_seen": 6906184,
      "step": 10570
    },
    {
      "epoch": 5.5424528301886795,
      "grad_norm": 0.9983073472976685,
      "learning_rate": 4.5373318720550366e-05,
      "loss": 0.3509,
      "num_input_tokens_seen": 6909320,
      "step": 10575
    },
    {
      "epoch": 5.545073375262055,
      "grad_norm": 1.0255939960479736,
      "learning_rate": 4.536668978675956e-05,
      "loss": 0.3646,
      "num_input_tokens_seen": 6912168,
      "step": 10580
    },
    {
      "epoch": 5.54769392033543,
      "grad_norm": 1.6788657903671265,
      "learning_rate": 4.53600565924896e-05,
      "loss": 0.3444,
      "num_input_tokens_seen": 6915784,
      "step": 10585
    },
    {
      "epoch": 5.550314465408805,
      "grad_norm": 1.9301459789276123,
      "learning_rate": 4.535341913912808e-05,
      "loss": 0.3305,
      "num_input_tokens_seen": 6918920,
      "step": 10590
    },
    {
      "epoch": 5.55293501048218,
      "grad_norm": 1.5783076286315918,
      "learning_rate": 4.534677742806348e-05,
      "loss": 0.274,
      "num_input_tokens_seen": 6921992,
      "step": 10595
    },
    {
      "epoch": 5.555555555555555,
      "grad_norm": 2.868910312652588,
      "learning_rate": 4.5340131460685175e-05,
      "loss": 0.3063,
      "num_input_tokens_seen": 6924840,
      "step": 10600
    },
    {
      "epoch": 5.5581761006289305,
      "grad_norm": 1.049315094947815,
      "learning_rate": 4.533348123838341e-05,
      "loss": 0.2652,
      "num_input_tokens_seen": 6927848,
      "step": 10605
    },
    {
      "epoch": 5.560796645702306,
      "grad_norm": 1.8983176946640015,
      "learning_rate": 4.532682676254936e-05,
      "loss": 0.4213,
      "num_input_tokens_seen": 6930568,
      "step": 10610
    },
    {
      "epoch": 5.563417190775681,
      "grad_norm": 1.813103199005127,
      "learning_rate": 4.5320168034575044e-05,
      "loss": 0.2323,
      "num_input_tokens_seen": 6933992,
      "step": 10615
    },
    {
      "epoch": 5.566037735849057,
      "grad_norm": 1.6515048742294312,
      "learning_rate": 4.53135050558534e-05,
      "loss": 0.2851,
      "num_input_tokens_seen": 6936680,
      "step": 10620
    },
    {
      "epoch": 5.568658280922432,
      "grad_norm": 2.062972068786621,
      "learning_rate": 4.530683782777825e-05,
      "loss": 0.3589,
      "num_input_tokens_seen": 6938792,
      "step": 10625
    },
    {
      "epoch": 5.571278825995807,
      "grad_norm": 1.364121437072754,
      "learning_rate": 4.53001663517443e-05,
      "loss": 0.3078,
      "num_input_tokens_seen": 6942664,
      "step": 10630
    },
    {
      "epoch": 5.573899371069182,
      "grad_norm": 1.9626355171203613,
      "learning_rate": 4.529349062914714e-05,
      "loss": 0.3896,
      "num_input_tokens_seen": 6944968,
      "step": 10635
    },
    {
      "epoch": 5.576519916142558,
      "grad_norm": 1.1709671020507812,
      "learning_rate": 4.528681066138326e-05,
      "loss": 0.3215,
      "num_input_tokens_seen": 6948264,
      "step": 10640
    },
    {
      "epoch": 5.579140461215933,
      "grad_norm": 1.5955780744552612,
      "learning_rate": 4.528012644985005e-05,
      "loss": 0.286,
      "num_input_tokens_seen": 6951304,
      "step": 10645
    },
    {
      "epoch": 5.581761006289308,
      "grad_norm": 0.8979138731956482,
      "learning_rate": 4.527343799594574e-05,
      "loss": 0.2562,
      "num_input_tokens_seen": 6954184,
      "step": 10650
    },
    {
      "epoch": 5.584381551362683,
      "grad_norm": 1.141779899597168,
      "learning_rate": 4.5266745301069494e-05,
      "loss": 0.4066,
      "num_input_tokens_seen": 6956552,
      "step": 10655
    },
    {
      "epoch": 5.587002096436059,
      "grad_norm": 2.8700904846191406,
      "learning_rate": 4.526004836662135e-05,
      "loss": 0.3173,
      "num_input_tokens_seen": 6959496,
      "step": 10660
    },
    {
      "epoch": 5.589622641509434,
      "grad_norm": 1.5063928365707397,
      "learning_rate": 4.525334719400223e-05,
      "loss": 0.4626,
      "num_input_tokens_seen": 6962600,
      "step": 10665
    },
    {
      "epoch": 5.5922431865828095,
      "grad_norm": 1.3815780878067017,
      "learning_rate": 4.524664178461393e-05,
      "loss": 0.2932,
      "num_input_tokens_seen": 6965160,
      "step": 10670
    },
    {
      "epoch": 5.594863731656185,
      "grad_norm": 1.2049682140350342,
      "learning_rate": 4.5239932139859154e-05,
      "loss": 0.2846,
      "num_input_tokens_seen": 6967880,
      "step": 10675
    },
    {
      "epoch": 5.59748427672956,
      "grad_norm": 0.9883308410644531,
      "learning_rate": 4.523321826114149e-05,
      "loss": 0.3684,
      "num_input_tokens_seen": 6971624,
      "step": 10680
    },
    {
      "epoch": 5.600104821802935,
      "grad_norm": 1.2140092849731445,
      "learning_rate": 4.5226500149865395e-05,
      "loss": 0.3605,
      "num_input_tokens_seen": 6974696,
      "step": 10685
    },
    {
      "epoch": 5.60272536687631,
      "grad_norm": 2.0628669261932373,
      "learning_rate": 4.5219777807436224e-05,
      "loss": 0.4213,
      "num_input_tokens_seen": 6977544,
      "step": 10690
    },
    {
      "epoch": 5.605345911949685,
      "grad_norm": 0.6720227599143982,
      "learning_rate": 4.5213051235260225e-05,
      "loss": 0.2012,
      "num_input_tokens_seen": 6983688,
      "step": 10695
    },
    {
      "epoch": 5.6079664570230605,
      "grad_norm": 1.2206945419311523,
      "learning_rate": 4.52063204347445e-05,
      "loss": 0.3028,
      "num_input_tokens_seen": 6987112,
      "step": 10700
    },
    {
      "epoch": 5.610587002096436,
      "grad_norm": 1.5186998844146729,
      "learning_rate": 4.519958540729707e-05,
      "loss": 0.4317,
      "num_input_tokens_seen": 6990312,
      "step": 10705
    },
    {
      "epoch": 5.613207547169811,
      "grad_norm": 0.9525564908981323,
      "learning_rate": 4.519284615432682e-05,
      "loss": 0.2794,
      "num_input_tokens_seen": 6992456,
      "step": 10710
    },
    {
      "epoch": 5.615828092243187,
      "grad_norm": 1.2789214849472046,
      "learning_rate": 4.518610267724352e-05,
      "loss": 0.3022,
      "num_input_tokens_seen": 6994792,
      "step": 10715
    },
    {
      "epoch": 5.618448637316562,
      "grad_norm": 1.3008570671081543,
      "learning_rate": 4.517935497745786e-05,
      "loss": 0.3356,
      "num_input_tokens_seen": 6997960,
      "step": 10720
    },
    {
      "epoch": 5.621069182389937,
      "grad_norm": 1.729266881942749,
      "learning_rate": 4.517260305638133e-05,
      "loss": 0.4004,
      "num_input_tokens_seen": 7001640,
      "step": 10725
    },
    {
      "epoch": 5.623689727463312,
      "grad_norm": 1.5321162939071655,
      "learning_rate": 4.516584691542639e-05,
      "loss": 0.2894,
      "num_input_tokens_seen": 7004456,
      "step": 10730
    },
    {
      "epoch": 5.626310272536688,
      "grad_norm": 2.3364713191986084,
      "learning_rate": 4.515908655600634e-05,
      "loss": 0.3021,
      "num_input_tokens_seen": 7006888,
      "step": 10735
    },
    {
      "epoch": 5.628930817610063,
      "grad_norm": 1.0851240158081055,
      "learning_rate": 4.515232197953537e-05,
      "loss": 0.3107,
      "num_input_tokens_seen": 7010600,
      "step": 10740
    },
    {
      "epoch": 5.631551362683438,
      "grad_norm": 1.3663978576660156,
      "learning_rate": 4.514555318742856e-05,
      "loss": 0.3227,
      "num_input_tokens_seen": 7014312,
      "step": 10745
    },
    {
      "epoch": 5.634171907756813,
      "grad_norm": 1.0052865743637085,
      "learning_rate": 4.513878018110184e-05,
      "loss": 0.2616,
      "num_input_tokens_seen": 7017896,
      "step": 10750
    },
    {
      "epoch": 5.636792452830189,
      "grad_norm": 4.101335525512695,
      "learning_rate": 4.513200296197207e-05,
      "loss": 0.3751,
      "num_input_tokens_seen": 7021256,
      "step": 10755
    },
    {
      "epoch": 5.639412997903564,
      "grad_norm": 0.9036024808883667,
      "learning_rate": 4.512522153145696e-05,
      "loss": 0.3269,
      "num_input_tokens_seen": 7024968,
      "step": 10760
    },
    {
      "epoch": 5.6420335429769395,
      "grad_norm": 1.6076247692108154,
      "learning_rate": 4.5118435890975106e-05,
      "loss": 0.3531,
      "num_input_tokens_seen": 7027528,
      "step": 10765
    },
    {
      "epoch": 5.644654088050315,
      "grad_norm": 2.5671379566192627,
      "learning_rate": 4.511164604194598e-05,
      "loss": 0.3019,
      "num_input_tokens_seen": 7030632,
      "step": 10770
    },
    {
      "epoch": 5.64727463312369,
      "grad_norm": 1.0710910558700562,
      "learning_rate": 4.5104851985789956e-05,
      "loss": 0.2803,
      "num_input_tokens_seen": 7034568,
      "step": 10775
    },
    {
      "epoch": 5.649895178197065,
      "grad_norm": 1.2586075067520142,
      "learning_rate": 4.5098053723928254e-05,
      "loss": 0.3737,
      "num_input_tokens_seen": 7037544,
      "step": 10780
    },
    {
      "epoch": 5.65251572327044,
      "grad_norm": 1.450269103050232,
      "learning_rate": 4.509125125778302e-05,
      "loss": 0.2719,
      "num_input_tokens_seen": 7041256,
      "step": 10785
    },
    {
      "epoch": 5.655136268343815,
      "grad_norm": 1.696509599685669,
      "learning_rate": 4.508444458877722e-05,
      "loss": 0.2306,
      "num_input_tokens_seen": 7043528,
      "step": 10790
    },
    {
      "epoch": 5.6577568134171905,
      "grad_norm": 0.8659370541572571,
      "learning_rate": 4.5077633718334755e-05,
      "loss": 0.3495,
      "num_input_tokens_seen": 7046696,
      "step": 10795
    },
    {
      "epoch": 5.660377358490566,
      "grad_norm": 1.1598825454711914,
      "learning_rate": 4.5070818647880375e-05,
      "loss": 0.4161,
      "num_input_tokens_seen": 7050216,
      "step": 10800
    },
    {
      "epoch": 5.662997903563941,
      "grad_norm": 1.0776852369308472,
      "learning_rate": 4.506399937883972e-05,
      "loss": 0.4665,
      "num_input_tokens_seen": 7053832,
      "step": 10805
    },
    {
      "epoch": 5.665618448637317,
      "grad_norm": 1.3113939762115479,
      "learning_rate": 4.5057175912639286e-05,
      "loss": 0.3131,
      "num_input_tokens_seen": 7056584,
      "step": 10810
    },
    {
      "epoch": 5.668238993710692,
      "grad_norm": 0.7802664041519165,
      "learning_rate": 4.5050348250706475e-05,
      "loss": 0.2541,
      "num_input_tokens_seen": 7059240,
      "step": 10815
    },
    {
      "epoch": 5.670859538784067,
      "grad_norm": 2.29962158203125,
      "learning_rate": 4.5043516394469566e-05,
      "loss": 0.3173,
      "num_input_tokens_seen": 7062280,
      "step": 10820
    },
    {
      "epoch": 5.673480083857442,
      "grad_norm": 0.8850539326667786,
      "learning_rate": 4.503668034535768e-05,
      "loss": 0.3815,
      "num_input_tokens_seen": 7065288,
      "step": 10825
    },
    {
      "epoch": 5.676100628930818,
      "grad_norm": 0.8735002875328064,
      "learning_rate": 4.5029840104800876e-05,
      "loss": 0.3664,
      "num_input_tokens_seen": 7069160,
      "step": 10830
    },
    {
      "epoch": 5.678721174004193,
      "grad_norm": 1.6253010034561157,
      "learning_rate": 4.5022995674230025e-05,
      "loss": 0.3452,
      "num_input_tokens_seen": 7072552,
      "step": 10835
    },
    {
      "epoch": 5.681341719077568,
      "grad_norm": 1.1932984590530396,
      "learning_rate": 4.5016147055076914e-05,
      "loss": 0.2612,
      "num_input_tokens_seen": 7076040,
      "step": 10840
    },
    {
      "epoch": 5.683962264150943,
      "grad_norm": 1.532172679901123,
      "learning_rate": 4.500929424877419e-05,
      "loss": 0.3839,
      "num_input_tokens_seen": 7079368,
      "step": 10845
    },
    {
      "epoch": 5.686582809224319,
      "grad_norm": 1.1424599885940552,
      "learning_rate": 4.5002437256755394e-05,
      "loss": 0.2846,
      "num_input_tokens_seen": 7082632,
      "step": 10850
    },
    {
      "epoch": 5.689203354297694,
      "grad_norm": 1.182997703552246,
      "learning_rate": 4.4995576080454924e-05,
      "loss": 0.3612,
      "num_input_tokens_seen": 7085192,
      "step": 10855
    },
    {
      "epoch": 5.6918238993710695,
      "grad_norm": 2.2382121086120605,
      "learning_rate": 4.4988710721308056e-05,
      "loss": 0.2785,
      "num_input_tokens_seen": 7088040,
      "step": 10860
    },
    {
      "epoch": 5.694444444444445,
      "grad_norm": 1.9084066152572632,
      "learning_rate": 4.498184118075095e-05,
      "loss": 0.359,
      "num_input_tokens_seen": 7090664,
      "step": 10865
    },
    {
      "epoch": 5.69706498951782,
      "grad_norm": 1.6380512714385986,
      "learning_rate": 4.497496746022063e-05,
      "loss": 0.3146,
      "num_input_tokens_seen": 7093224,
      "step": 10870
    },
    {
      "epoch": 5.699685534591195,
      "grad_norm": 1.281823992729187,
      "learning_rate": 4.4968089561155015e-05,
      "loss": 0.3213,
      "num_input_tokens_seen": 7096008,
      "step": 10875
    },
    {
      "epoch": 5.70230607966457,
      "grad_norm": 1.4034706354141235,
      "learning_rate": 4.496120748499286e-05,
      "loss": 0.3177,
      "num_input_tokens_seen": 7098760,
      "step": 10880
    },
    {
      "epoch": 5.704926624737945,
      "grad_norm": 2.5132248401641846,
      "learning_rate": 4.4954321233173833e-05,
      "loss": 0.3705,
      "num_input_tokens_seen": 7102120,
      "step": 10885
    },
    {
      "epoch": 5.7075471698113205,
      "grad_norm": 1.1210544109344482,
      "learning_rate": 4.494743080713846e-05,
      "loss": 0.3374,
      "num_input_tokens_seen": 7105672,
      "step": 10890
    },
    {
      "epoch": 5.710167714884696,
      "grad_norm": 1.3897385597229004,
      "learning_rate": 4.494053620832812e-05,
      "loss": 0.3155,
      "num_input_tokens_seen": 7109288,
      "step": 10895
    },
    {
      "epoch": 5.712788259958071,
      "grad_norm": 1.429527759552002,
      "learning_rate": 4.493363743818511e-05,
      "loss": 0.2764,
      "num_input_tokens_seen": 7112264,
      "step": 10900
    },
    {
      "epoch": 5.715408805031447,
      "grad_norm": 2.4984707832336426,
      "learning_rate": 4.492673449815257e-05,
      "loss": 0.2694,
      "num_input_tokens_seen": 7114984,
      "step": 10905
    },
    {
      "epoch": 5.718029350104822,
      "grad_norm": 1.2739332914352417,
      "learning_rate": 4.4919827389674496e-05,
      "loss": 0.334,
      "num_input_tokens_seen": 7117896,
      "step": 10910
    },
    {
      "epoch": 5.720649895178197,
      "grad_norm": 1.169013261795044,
      "learning_rate": 4.491291611419579e-05,
      "loss": 0.2514,
      "num_input_tokens_seen": 7121192,
      "step": 10915
    },
    {
      "epoch": 5.723270440251572,
      "grad_norm": 1.1820812225341797,
      "learning_rate": 4.4906000673162216e-05,
      "loss": 0.3174,
      "num_input_tokens_seen": 7124680,
      "step": 10920
    },
    {
      "epoch": 5.725890985324948,
      "grad_norm": 1.6089140176773071,
      "learning_rate": 4.48990810680204e-05,
      "loss": 0.3211,
      "num_input_tokens_seen": 7127624,
      "step": 10925
    },
    {
      "epoch": 5.728511530398323,
      "grad_norm": 1.4587299823760986,
      "learning_rate": 4.4892157300217844e-05,
      "loss": 0.2963,
      "num_input_tokens_seen": 7134792,
      "step": 10930
    },
    {
      "epoch": 5.731132075471698,
      "grad_norm": 1.3536300659179688,
      "learning_rate": 4.488522937120293e-05,
      "loss": 0.2423,
      "num_input_tokens_seen": 7137448,
      "step": 10935
    },
    {
      "epoch": 5.733752620545073,
      "grad_norm": 2.3886196613311768,
      "learning_rate": 4.4878297282424886e-05,
      "loss": 0.3686,
      "num_input_tokens_seen": 7141384,
      "step": 10940
    },
    {
      "epoch": 5.736373165618449,
      "grad_norm": 2.8371055126190186,
      "learning_rate": 4.4871361035333836e-05,
      "loss": 0.3533,
      "num_input_tokens_seen": 7144296,
      "step": 10945
    },
    {
      "epoch": 5.738993710691824,
      "grad_norm": 1.5477328300476074,
      "learning_rate": 4.486442063138076e-05,
      "loss": 0.4476,
      "num_input_tokens_seen": 7148328,
      "step": 10950
    },
    {
      "epoch": 5.7416142557651995,
      "grad_norm": 1.175310492515564,
      "learning_rate": 4.4857476072017516e-05,
      "loss": 0.382,
      "num_input_tokens_seen": 7151304,
      "step": 10955
    },
    {
      "epoch": 5.744234800838575,
      "grad_norm": 1.2798203229904175,
      "learning_rate": 4.485052735869682e-05,
      "loss": 0.2643,
      "num_input_tokens_seen": 7155240,
      "step": 10960
    },
    {
      "epoch": 5.74685534591195,
      "grad_norm": 1.018046498298645,
      "learning_rate": 4.484357449287228e-05,
      "loss": 0.2617,
      "num_input_tokens_seen": 7159496,
      "step": 10965
    },
    {
      "epoch": 5.749475890985325,
      "grad_norm": 1.3549659252166748,
      "learning_rate": 4.483661747599833e-05,
      "loss": 0.2692,
      "num_input_tokens_seen": 7163688,
      "step": 10970
    },
    {
      "epoch": 5.7520964360587,
      "grad_norm": 0.9913578033447266,
      "learning_rate": 4.482965630953031e-05,
      "loss": 0.2746,
      "num_input_tokens_seen": 7166728,
      "step": 10975
    },
    {
      "epoch": 5.754716981132075,
      "grad_norm": 1.4906319379806519,
      "learning_rate": 4.4822690994924425e-05,
      "loss": 0.3642,
      "num_input_tokens_seen": 7169928,
      "step": 10980
    },
    {
      "epoch": 5.7573375262054505,
      "grad_norm": 1.0548086166381836,
      "learning_rate": 4.481572153363773e-05,
      "loss": 0.3714,
      "num_input_tokens_seen": 7172936,
      "step": 10985
    },
    {
      "epoch": 5.759958071278826,
      "grad_norm": 3.5625078678131104,
      "learning_rate": 4.480874792712817e-05,
      "loss": 0.306,
      "num_input_tokens_seen": 7175240,
      "step": 10990
    },
    {
      "epoch": 5.762578616352201,
      "grad_norm": 0.9207141995429993,
      "learning_rate": 4.480177017685452e-05,
      "loss": 0.3131,
      "num_input_tokens_seen": 7178216,
      "step": 10995
    },
    {
      "epoch": 5.765199161425577,
      "grad_norm": 1.7896889448165894,
      "learning_rate": 4.479478828427647e-05,
      "loss": 0.3672,
      "num_input_tokens_seen": 7182056,
      "step": 11000
    },
    {
      "epoch": 5.767819706498952,
      "grad_norm": 2.0308330059051514,
      "learning_rate": 4.478780225085454e-05,
      "loss": 0.3171,
      "num_input_tokens_seen": 7186056,
      "step": 11005
    },
    {
      "epoch": 5.770440251572327,
      "grad_norm": 1.600053310394287,
      "learning_rate": 4.4780812078050125e-05,
      "loss": 0.3763,
      "num_input_tokens_seen": 7188808,
      "step": 11010
    },
    {
      "epoch": 5.773060796645702,
      "grad_norm": 0.7395328283309937,
      "learning_rate": 4.47738177673255e-05,
      "loss": 0.3055,
      "num_input_tokens_seen": 7192328,
      "step": 11015
    },
    {
      "epoch": 5.7756813417190775,
      "grad_norm": 1.9852755069732666,
      "learning_rate": 4.4766819320143785e-05,
      "loss": 0.3477,
      "num_input_tokens_seen": 7195368,
      "step": 11020
    },
    {
      "epoch": 5.778301886792453,
      "grad_norm": 1.2217116355895996,
      "learning_rate": 4.475981673796899e-05,
      "loss": 0.3288,
      "num_input_tokens_seen": 7199752,
      "step": 11025
    },
    {
      "epoch": 5.780922431865828,
      "grad_norm": 1.611814022064209,
      "learning_rate": 4.475281002226596e-05,
      "loss": 0.4754,
      "num_input_tokens_seen": 7203432,
      "step": 11030
    },
    {
      "epoch": 5.783542976939203,
      "grad_norm": 1.5557372570037842,
      "learning_rate": 4.4745799174500414e-05,
      "loss": 0.291,
      "num_input_tokens_seen": 7205864,
      "step": 11035
    },
    {
      "epoch": 5.786163522012579,
      "grad_norm": 0.9674282670021057,
      "learning_rate": 4.473878419613896e-05,
      "loss": 0.2701,
      "num_input_tokens_seen": 7208424,
      "step": 11040
    },
    {
      "epoch": 5.788784067085954,
      "grad_norm": 1.018232822418213,
      "learning_rate": 4.473176508864905e-05,
      "loss": 0.589,
      "num_input_tokens_seen": 7211464,
      "step": 11045
    },
    {
      "epoch": 5.7914046121593294,
      "grad_norm": 1.6476027965545654,
      "learning_rate": 4.4724741853498995e-05,
      "loss": 0.3785,
      "num_input_tokens_seen": 7214472,
      "step": 11050
    },
    {
      "epoch": 5.794025157232705,
      "grad_norm": 1.9294054508209229,
      "learning_rate": 4.471771449215797e-05,
      "loss": 0.3445,
      "num_input_tokens_seen": 7218184,
      "step": 11055
    },
    {
      "epoch": 5.79664570230608,
      "grad_norm": 1.0922529697418213,
      "learning_rate": 4.471068300609603e-05,
      "loss": 0.2823,
      "num_input_tokens_seen": 7221480,
      "step": 11060
    },
    {
      "epoch": 5.799266247379455,
      "grad_norm": 1.23745596408844,
      "learning_rate": 4.4703647396784065e-05,
      "loss": 0.2177,
      "num_input_tokens_seen": 7224424,
      "step": 11065
    },
    {
      "epoch": 5.80188679245283,
      "grad_norm": 1.0895531177520752,
      "learning_rate": 4.469660766569387e-05,
      "loss": 0.3189,
      "num_input_tokens_seen": 7227144,
      "step": 11070
    },
    {
      "epoch": 5.804507337526205,
      "grad_norm": 2.9336771965026855,
      "learning_rate": 4.4689563814298057e-05,
      "loss": 0.32,
      "num_input_tokens_seen": 7229768,
      "step": 11075
    },
    {
      "epoch": 5.8071278825995805,
      "grad_norm": 2.0102365016937256,
      "learning_rate": 4.468251584407012e-05,
      "loss": 0.2726,
      "num_input_tokens_seen": 7233352,
      "step": 11080
    },
    {
      "epoch": 5.809748427672956,
      "grad_norm": 1.417602300643921,
      "learning_rate": 4.467546375648443e-05,
      "loss": 0.3288,
      "num_input_tokens_seen": 7236328,
      "step": 11085
    },
    {
      "epoch": 5.812368972746331,
      "grad_norm": 1.7046465873718262,
      "learning_rate": 4.4668407553016176e-05,
      "loss": 0.4364,
      "num_input_tokens_seen": 7238952,
      "step": 11090
    },
    {
      "epoch": 5.814989517819707,
      "grad_norm": 1.3114043474197388,
      "learning_rate": 4.466134723514146e-05,
      "loss": 0.3503,
      "num_input_tokens_seen": 7242568,
      "step": 11095
    },
    {
      "epoch": 5.817610062893082,
      "grad_norm": 2.1698760986328125,
      "learning_rate": 4.465428280433721e-05,
      "loss": 0.3529,
      "num_input_tokens_seen": 7245320,
      "step": 11100
    },
    {
      "epoch": 5.820230607966457,
      "grad_norm": 1.7751545906066895,
      "learning_rate": 4.464721426208123e-05,
      "loss": 0.2499,
      "num_input_tokens_seen": 7248264,
      "step": 11105
    },
    {
      "epoch": 5.822851153039832,
      "grad_norm": 1.4914757013320923,
      "learning_rate": 4.464014160985217e-05,
      "loss": 0.4053,
      "num_input_tokens_seen": 7251176,
      "step": 11110
    },
    {
      "epoch": 5.8254716981132075,
      "grad_norm": 2.155690908432007,
      "learning_rate": 4.463306484912955e-05,
      "loss": 0.3392,
      "num_input_tokens_seen": 7253864,
      "step": 11115
    },
    {
      "epoch": 5.828092243186583,
      "grad_norm": 1.9045162200927734,
      "learning_rate": 4.462598398139375e-05,
      "loss": 0.2589,
      "num_input_tokens_seen": 7257704,
      "step": 11120
    },
    {
      "epoch": 5.830712788259958,
      "grad_norm": 0.9985941052436829,
      "learning_rate": 4.4618899008126005e-05,
      "loss": 0.2879,
      "num_input_tokens_seen": 7260936,
      "step": 11125
    },
    {
      "epoch": 5.833333333333333,
      "grad_norm": 1.301242470741272,
      "learning_rate": 4.461180993080841e-05,
      "loss": 0.3441,
      "num_input_tokens_seen": 7264456,
      "step": 11130
    },
    {
      "epoch": 5.835953878406709,
      "grad_norm": 1.2154499292373657,
      "learning_rate": 4.460471675092393e-05,
      "loss": 0.4791,
      "num_input_tokens_seen": 7267624,
      "step": 11135
    },
    {
      "epoch": 5.838574423480084,
      "grad_norm": 0.8753020763397217,
      "learning_rate": 4.4597619469956356e-05,
      "loss": 0.3025,
      "num_input_tokens_seen": 7270760,
      "step": 11140
    },
    {
      "epoch": 5.841194968553459,
      "grad_norm": 1.103882908821106,
      "learning_rate": 4.459051808939038e-05,
      "loss": 0.2892,
      "num_input_tokens_seen": 7273608,
      "step": 11145
    },
    {
      "epoch": 5.843815513626835,
      "grad_norm": 1.6352866888046265,
      "learning_rate": 4.458341261071151e-05,
      "loss": 0.3353,
      "num_input_tokens_seen": 7276264,
      "step": 11150
    },
    {
      "epoch": 5.84643605870021,
      "grad_norm": 1.2236723899841309,
      "learning_rate": 4.457630303540615e-05,
      "loss": 0.2919,
      "num_input_tokens_seen": 7279336,
      "step": 11155
    },
    {
      "epoch": 5.849056603773585,
      "grad_norm": 1.00264310836792,
      "learning_rate": 4.456918936496154e-05,
      "loss": 0.3443,
      "num_input_tokens_seen": 7282376,
      "step": 11160
    },
    {
      "epoch": 5.85167714884696,
      "grad_norm": 1.2550063133239746,
      "learning_rate": 4.456207160086575e-05,
      "loss": 0.3058,
      "num_input_tokens_seen": 7285448,
      "step": 11165
    },
    {
      "epoch": 5.854297693920335,
      "grad_norm": 1.502326250076294,
      "learning_rate": 4.4554949744607775e-05,
      "loss": 0.352,
      "num_input_tokens_seen": 7288648,
      "step": 11170
    },
    {
      "epoch": 5.8569182389937104,
      "grad_norm": 1.5313043594360352,
      "learning_rate": 4.4547823797677405e-05,
      "loss": 0.294,
      "num_input_tokens_seen": 7291432,
      "step": 11175
    },
    {
      "epoch": 5.859538784067086,
      "grad_norm": 1.4439337253570557,
      "learning_rate": 4.45406937615653e-05,
      "loss": 0.2219,
      "num_input_tokens_seen": 7296744,
      "step": 11180
    },
    {
      "epoch": 5.862159329140461,
      "grad_norm": 1.782731533050537,
      "learning_rate": 4.4533559637763006e-05,
      "loss": 0.2589,
      "num_input_tokens_seen": 7299496,
      "step": 11185
    },
    {
      "epoch": 5.864779874213837,
      "grad_norm": 1.6291608810424805,
      "learning_rate": 4.452642142776288e-05,
      "loss": 0.2754,
      "num_input_tokens_seen": 7302440,
      "step": 11190
    },
    {
      "epoch": 5.867400419287212,
      "grad_norm": 1.1376134157180786,
      "learning_rate": 4.451927913305816e-05,
      "loss": 0.207,
      "num_input_tokens_seen": 7305864,
      "step": 11195
    },
    {
      "epoch": 5.870020964360587,
      "grad_norm": 1.4263004064559937,
      "learning_rate": 4.4512132755142935e-05,
      "loss": 0.3533,
      "num_input_tokens_seen": 7309096,
      "step": 11200
    },
    {
      "epoch": 5.872641509433962,
      "grad_norm": 2.019026756286621,
      "learning_rate": 4.450498229551214e-05,
      "loss": 0.4739,
      "num_input_tokens_seen": 7312648,
      "step": 11205
    },
    {
      "epoch": 5.8752620545073375,
      "grad_norm": 2.6007168292999268,
      "learning_rate": 4.449782775566158e-05,
      "loss": 0.3111,
      "num_input_tokens_seen": 7316264,
      "step": 11210
    },
    {
      "epoch": 5.877882599580713,
      "grad_norm": 1.0774433612823486,
      "learning_rate": 4.449066913708789e-05,
      "loss": 0.3773,
      "num_input_tokens_seen": 7320584,
      "step": 11215
    },
    {
      "epoch": 5.880503144654088,
      "grad_norm": 0.8343318104743958,
      "learning_rate": 4.448350644128858e-05,
      "loss": 0.4718,
      "num_input_tokens_seen": 7323880,
      "step": 11220
    },
    {
      "epoch": 5.883123689727463,
      "grad_norm": 3.132448673248291,
      "learning_rate": 4.4476339669762e-05,
      "loss": 0.3615,
      "num_input_tokens_seen": 7326984,
      "step": 11225
    },
    {
      "epoch": 5.885744234800838,
      "grad_norm": 1.2120476961135864,
      "learning_rate": 4.446916882400737e-05,
      "loss": 0.3053,
      "num_input_tokens_seen": 7330504,
      "step": 11230
    },
    {
      "epoch": 5.888364779874214,
      "grad_norm": 0.9504019021987915,
      "learning_rate": 4.446199390552472e-05,
      "loss": 0.3478,
      "num_input_tokens_seen": 7333480,
      "step": 11235
    },
    {
      "epoch": 5.890985324947589,
      "grad_norm": 0.9402258396148682,
      "learning_rate": 4.445481491581499e-05,
      "loss": 0.278,
      "num_input_tokens_seen": 7336232,
      "step": 11240
    },
    {
      "epoch": 5.893605870020965,
      "grad_norm": 1.3564811944961548,
      "learning_rate": 4.444763185637993e-05,
      "loss": 0.2636,
      "num_input_tokens_seen": 7338920,
      "step": 11245
    },
    {
      "epoch": 5.89622641509434,
      "grad_norm": 1.5410864353179932,
      "learning_rate": 4.4440444728722166e-05,
      "loss": 0.2768,
      "num_input_tokens_seen": 7341512,
      "step": 11250
    },
    {
      "epoch": 5.898846960167715,
      "grad_norm": 2.0232040882110596,
      "learning_rate": 4.4433253534345136e-05,
      "loss": 0.4213,
      "num_input_tokens_seen": 7345128,
      "step": 11255
    },
    {
      "epoch": 5.90146750524109,
      "grad_norm": 0.8785939812660217,
      "learning_rate": 4.442605827475318e-05,
      "loss": 0.2912,
      "num_input_tokens_seen": 7348328,
      "step": 11260
    },
    {
      "epoch": 5.904088050314465,
      "grad_norm": 1.467207431793213,
      "learning_rate": 4.4418858951451457e-05,
      "loss": 0.4392,
      "num_input_tokens_seen": 7351336,
      "step": 11265
    },
    {
      "epoch": 5.90670859538784,
      "grad_norm": 1.7216835021972656,
      "learning_rate": 4.441165556594599e-05,
      "loss": 0.3917,
      "num_input_tokens_seen": 7355208,
      "step": 11270
    },
    {
      "epoch": 5.909329140461216,
      "grad_norm": 1.9636768102645874,
      "learning_rate": 4.440444811974363e-05,
      "loss": 0.2387,
      "num_input_tokens_seen": 7357768,
      "step": 11275
    },
    {
      "epoch": 5.911949685534591,
      "grad_norm": 2.216000556945801,
      "learning_rate": 4.43972366143521e-05,
      "loss": 0.3063,
      "num_input_tokens_seen": 7360808,
      "step": 11280
    },
    {
      "epoch": 5.914570230607967,
      "grad_norm": 1.6070836782455444,
      "learning_rate": 4.4390021051279975e-05,
      "loss": 0.2951,
      "num_input_tokens_seen": 7363208,
      "step": 11285
    },
    {
      "epoch": 5.917190775681342,
      "grad_norm": 1.2246612310409546,
      "learning_rate": 4.438280143203665e-05,
      "loss": 0.2615,
      "num_input_tokens_seen": 7366504,
      "step": 11290
    },
    {
      "epoch": 5.919811320754717,
      "grad_norm": 1.4197686910629272,
      "learning_rate": 4.43755777581324e-05,
      "loss": 0.2682,
      "num_input_tokens_seen": 7369800,
      "step": 11295
    },
    {
      "epoch": 5.922431865828092,
      "grad_norm": 1.4167253971099854,
      "learning_rate": 4.4368350031078334e-05,
      "loss": 0.3431,
      "num_input_tokens_seen": 7372680,
      "step": 11300
    },
    {
      "epoch": 5.9250524109014675,
      "grad_norm": 1.029166579246521,
      "learning_rate": 4.43611182523864e-05,
      "loss": 0.3707,
      "num_input_tokens_seen": 7377480,
      "step": 11305
    },
    {
      "epoch": 5.927672955974843,
      "grad_norm": 2.223362922668457,
      "learning_rate": 4.4353882423569416e-05,
      "loss": 0.3121,
      "num_input_tokens_seen": 7380456,
      "step": 11310
    },
    {
      "epoch": 5.930293501048218,
      "grad_norm": 1.95270836353302,
      "learning_rate": 4.434664254614103e-05,
      "loss": 0.3437,
      "num_input_tokens_seen": 7383208,
      "step": 11315
    },
    {
      "epoch": 5.932914046121593,
      "grad_norm": 1.1793498992919922,
      "learning_rate": 4.4339398621615734e-05,
      "loss": 0.2771,
      "num_input_tokens_seen": 7385576,
      "step": 11320
    },
    {
      "epoch": 5.935534591194968,
      "grad_norm": 1.21455979347229,
      "learning_rate": 4.433215065150889e-05,
      "loss": 0.256,
      "num_input_tokens_seen": 7388296,
      "step": 11325
    },
    {
      "epoch": 5.938155136268344,
      "grad_norm": 1.4287248849868774,
      "learning_rate": 4.4324898637336674e-05,
      "loss": 0.4286,
      "num_input_tokens_seen": 7391784,
      "step": 11330
    },
    {
      "epoch": 5.940775681341719,
      "grad_norm": 1.539924144744873,
      "learning_rate": 4.431764258061615e-05,
      "loss": 0.3003,
      "num_input_tokens_seen": 7395016,
      "step": 11335
    },
    {
      "epoch": 5.943396226415095,
      "grad_norm": 1.0805860757827759,
      "learning_rate": 4.431038248286517e-05,
      "loss": 0.2253,
      "num_input_tokens_seen": 7398856,
      "step": 11340
    },
    {
      "epoch": 5.94601677148847,
      "grad_norm": 1.4659391641616821,
      "learning_rate": 4.430311834560248e-05,
      "loss": 0.384,
      "num_input_tokens_seen": 7401864,
      "step": 11345
    },
    {
      "epoch": 5.948637316561845,
      "grad_norm": 1.4317209720611572,
      "learning_rate": 4.4295850170347664e-05,
      "loss": 0.4647,
      "num_input_tokens_seen": 7404360,
      "step": 11350
    },
    {
      "epoch": 5.95125786163522,
      "grad_norm": 1.4155851602554321,
      "learning_rate": 4.428857795862113e-05,
      "loss": 0.363,
      "num_input_tokens_seen": 7407464,
      "step": 11355
    },
    {
      "epoch": 5.953878406708595,
      "grad_norm": 1.6002193689346313,
      "learning_rate": 4.428130171194414e-05,
      "loss": 0.3314,
      "num_input_tokens_seen": 7411144,
      "step": 11360
    },
    {
      "epoch": 5.95649895178197,
      "grad_norm": 1.5346717834472656,
      "learning_rate": 4.427402143183879e-05,
      "loss": 0.3271,
      "num_input_tokens_seen": 7414376,
      "step": 11365
    },
    {
      "epoch": 5.959119496855346,
      "grad_norm": 1.3990347385406494,
      "learning_rate": 4.426673711982806e-05,
      "loss": 0.4012,
      "num_input_tokens_seen": 7417672,
      "step": 11370
    },
    {
      "epoch": 5.961740041928721,
      "grad_norm": 1.5859206914901733,
      "learning_rate": 4.425944877743573e-05,
      "loss": 0.3086,
      "num_input_tokens_seen": 7420392,
      "step": 11375
    },
    {
      "epoch": 5.964360587002097,
      "grad_norm": 1.050025224685669,
      "learning_rate": 4.425215640618644e-05,
      "loss": 0.2768,
      "num_input_tokens_seen": 7423880,
      "step": 11380
    },
    {
      "epoch": 5.966981132075472,
      "grad_norm": 1.2002458572387695,
      "learning_rate": 4.424486000760567e-05,
      "loss": 0.3621,
      "num_input_tokens_seen": 7427624,
      "step": 11385
    },
    {
      "epoch": 5.969601677148847,
      "grad_norm": 1.8675642013549805,
      "learning_rate": 4.423755958321974e-05,
      "loss": 0.2929,
      "num_input_tokens_seen": 7430536,
      "step": 11390
    },
    {
      "epoch": 5.972222222222222,
      "grad_norm": 1.0060263872146606,
      "learning_rate": 4.423025513455582e-05,
      "loss": 0.4199,
      "num_input_tokens_seen": 7433192,
      "step": 11395
    },
    {
      "epoch": 5.9748427672955975,
      "grad_norm": 1.3954743146896362,
      "learning_rate": 4.422294666314192e-05,
      "loss": 0.2734,
      "num_input_tokens_seen": 7436104,
      "step": 11400
    },
    {
      "epoch": 5.977463312368973,
      "grad_norm": 1.7244462966918945,
      "learning_rate": 4.421563417050688e-05,
      "loss": 0.2503,
      "num_input_tokens_seen": 7439432,
      "step": 11405
    },
    {
      "epoch": 5.980083857442348,
      "grad_norm": 2.0097808837890625,
      "learning_rate": 4.42083176581804e-05,
      "loss": 0.3106,
      "num_input_tokens_seen": 7441960,
      "step": 11410
    },
    {
      "epoch": 5.982704402515723,
      "grad_norm": 0.8613712787628174,
      "learning_rate": 4.4200997127693e-05,
      "loss": 0.2537,
      "num_input_tokens_seen": 7445576,
      "step": 11415
    },
    {
      "epoch": 5.985324947589098,
      "grad_norm": 1.3441483974456787,
      "learning_rate": 4.419367258057606e-05,
      "loss": 0.3174,
      "num_input_tokens_seen": 7448104,
      "step": 11420
    },
    {
      "epoch": 5.987945492662474,
      "grad_norm": 1.0121376514434814,
      "learning_rate": 4.418634401836179e-05,
      "loss": 0.3192,
      "num_input_tokens_seen": 7451976,
      "step": 11425
    },
    {
      "epoch": 5.990566037735849,
      "grad_norm": 1.161160945892334,
      "learning_rate": 4.417901144258324e-05,
      "loss": 0.3103,
      "num_input_tokens_seen": 7456744,
      "step": 11430
    },
    {
      "epoch": 5.993186582809225,
      "grad_norm": 1.4588578939437866,
      "learning_rate": 4.4171674854774305e-05,
      "loss": 0.3933,
      "num_input_tokens_seen": 7460392,
      "step": 11435
    },
    {
      "epoch": 5.9958071278826,
      "grad_norm": 1.224139928817749,
      "learning_rate": 4.416433425646971e-05,
      "loss": 0.3392,
      "num_input_tokens_seen": 7463496,
      "step": 11440
    },
    {
      "epoch": 5.998427672955975,
      "grad_norm": 1.9086723327636719,
      "learning_rate": 4.415698964920503e-05,
      "loss": 0.3863,
      "num_input_tokens_seen": 7466376,
      "step": 11445
    },
    {
      "epoch": 6.0,
      "eval_loss": 0.5028393268585205,
      "eval_runtime": 11.8825,
      "eval_samples_per_second": 71.365,
      "eval_steps_per_second": 17.841,
      "num_input_tokens_seen": 7467968,
      "step": 11448
    },
    {
      "epoch": 6.00104821802935,
      "grad_norm": 1.4418452978134155,
      "learning_rate": 4.414964103451667e-05,
      "loss": 0.4069,
      "num_input_tokens_seen": 7468896,
      "step": 11450
    },
    {
      "epoch": 6.003668763102725,
      "grad_norm": 1.3892951011657715,
      "learning_rate": 4.414228841394187e-05,
      "loss": 0.2639,
      "num_input_tokens_seen": 7472640,
      "step": 11455
    },
    {
      "epoch": 6.0062893081761,
      "grad_norm": 0.8499836325645447,
      "learning_rate": 4.413493178901872e-05,
      "loss": 0.2516,
      "num_input_tokens_seen": 7476192,
      "step": 11460
    },
    {
      "epoch": 6.008909853249476,
      "grad_norm": 1.2180339097976685,
      "learning_rate": 4.412757116128614e-05,
      "loss": 0.2482,
      "num_input_tokens_seen": 7479712,
      "step": 11465
    },
    {
      "epoch": 6.011530398322851,
      "grad_norm": 0.9744583368301392,
      "learning_rate": 4.41202065322839e-05,
      "loss": 0.2844,
      "num_input_tokens_seen": 7482816,
      "step": 11470
    },
    {
      "epoch": 6.014150943396227,
      "grad_norm": 1.0086109638214111,
      "learning_rate": 4.411283790355259e-05,
      "loss": 0.2835,
      "num_input_tokens_seen": 7485760,
      "step": 11475
    },
    {
      "epoch": 6.016771488469602,
      "grad_norm": 1.6496813297271729,
      "learning_rate": 4.410546527663364e-05,
      "loss": 0.3323,
      "num_input_tokens_seen": 7489440,
      "step": 11480
    },
    {
      "epoch": 6.019392033542977,
      "grad_norm": 1.3064777851104736,
      "learning_rate": 4.409808865306931e-05,
      "loss": 0.1816,
      "num_input_tokens_seen": 7492704,
      "step": 11485
    },
    {
      "epoch": 6.022012578616352,
      "grad_norm": 2.089953899383545,
      "learning_rate": 4.409070803440273e-05,
      "loss": 0.27,
      "num_input_tokens_seen": 7495776,
      "step": 11490
    },
    {
      "epoch": 6.0246331236897275,
      "grad_norm": 1.5979270935058594,
      "learning_rate": 4.4083323422177824e-05,
      "loss": 0.2583,
      "num_input_tokens_seen": 7498560,
      "step": 11495
    },
    {
      "epoch": 6.027253668763103,
      "grad_norm": 1.3977192640304565,
      "learning_rate": 4.407593481793937e-05,
      "loss": 0.242,
      "num_input_tokens_seen": 7500992,
      "step": 11500
    },
    {
      "epoch": 6.029874213836478,
      "grad_norm": 4.063703536987305,
      "learning_rate": 4.406854222323298e-05,
      "loss": 0.3193,
      "num_input_tokens_seen": 7504192,
      "step": 11505
    },
    {
      "epoch": 6.032494758909853,
      "grad_norm": 1.367274284362793,
      "learning_rate": 4.406114563960511e-05,
      "loss": 0.2661,
      "num_input_tokens_seen": 7507552,
      "step": 11510
    },
    {
      "epoch": 6.035115303983228,
      "grad_norm": 2.464358329772949,
      "learning_rate": 4.4053745068603026e-05,
      "loss": 0.2944,
      "num_input_tokens_seen": 7510592,
      "step": 11515
    },
    {
      "epoch": 6.037735849056604,
      "grad_norm": 1.5620992183685303,
      "learning_rate": 4.404634051177485e-05,
      "loss": 0.3082,
      "num_input_tokens_seen": 7513664,
      "step": 11520
    },
    {
      "epoch": 6.040356394129979,
      "grad_norm": 3.3030028343200684,
      "learning_rate": 4.403893197066953e-05,
      "loss": 0.2858,
      "num_input_tokens_seen": 7516320,
      "step": 11525
    },
    {
      "epoch": 6.0429769392033545,
      "grad_norm": 1.6613144874572754,
      "learning_rate": 4.4031519446836855e-05,
      "loss": 0.2456,
      "num_input_tokens_seen": 7518944,
      "step": 11530
    },
    {
      "epoch": 6.04559748427673,
      "grad_norm": 1.71934175491333,
      "learning_rate": 4.402410294182743e-05,
      "loss": 0.2359,
      "num_input_tokens_seen": 7521376,
      "step": 11535
    },
    {
      "epoch": 6.048218029350105,
      "grad_norm": 0.9962888360023499,
      "learning_rate": 4.4016682457192714e-05,
      "loss": 0.304,
      "num_input_tokens_seen": 7526432,
      "step": 11540
    },
    {
      "epoch": 6.05083857442348,
      "grad_norm": 1.982828974723816,
      "learning_rate": 4.400925799448497e-05,
      "loss": 0.2051,
      "num_input_tokens_seen": 7532064,
      "step": 11545
    },
    {
      "epoch": 6.053459119496855,
      "grad_norm": 1.6212459802627563,
      "learning_rate": 4.400182955525733e-05,
      "loss": 0.248,
      "num_input_tokens_seen": 7535360,
      "step": 11550
    },
    {
      "epoch": 6.05607966457023,
      "grad_norm": 1.0463464260101318,
      "learning_rate": 4.3994397141063734e-05,
      "loss": 0.2185,
      "num_input_tokens_seen": 7539232,
      "step": 11555
    },
    {
      "epoch": 6.058700209643606,
      "grad_norm": 1.4498997926712036,
      "learning_rate": 4.398696075345896e-05,
      "loss": 0.3514,
      "num_input_tokens_seen": 7542464,
      "step": 11560
    },
    {
      "epoch": 6.061320754716981,
      "grad_norm": 2.013235092163086,
      "learning_rate": 4.39795203939986e-05,
      "loss": 0.1975,
      "num_input_tokens_seen": 7546176,
      "step": 11565
    },
    {
      "epoch": 6.063941299790357,
      "grad_norm": 1.6690424680709839,
      "learning_rate": 4.397207606423911e-05,
      "loss": 0.3737,
      "num_input_tokens_seen": 7549696,
      "step": 11570
    },
    {
      "epoch": 6.066561844863732,
      "grad_norm": 1.0005443096160889,
      "learning_rate": 4.396462776573775e-05,
      "loss": 0.2166,
      "num_input_tokens_seen": 7552544,
      "step": 11575
    },
    {
      "epoch": 6.069182389937107,
      "grad_norm": 3.303757905960083,
      "learning_rate": 4.395717550005262e-05,
      "loss": 0.2125,
      "num_input_tokens_seen": 7555168,
      "step": 11580
    },
    {
      "epoch": 6.071802935010482,
      "grad_norm": 1.3814829587936401,
      "learning_rate": 4.394971926874265e-05,
      "loss": 0.3501,
      "num_input_tokens_seen": 7558528,
      "step": 11585
    },
    {
      "epoch": 6.0744234800838575,
      "grad_norm": 1.4147686958312988,
      "learning_rate": 4.3942259073367604e-05,
      "loss": 0.3137,
      "num_input_tokens_seen": 7561760,
      "step": 11590
    },
    {
      "epoch": 6.077044025157233,
      "grad_norm": 1.3780230283737183,
      "learning_rate": 4.3934794915488065e-05,
      "loss": 0.3105,
      "num_input_tokens_seen": 7565312,
      "step": 11595
    },
    {
      "epoch": 6.079664570230608,
      "grad_norm": 1.3075501918792725,
      "learning_rate": 4.392732679666544e-05,
      "loss": 0.3092,
      "num_input_tokens_seen": 7568640,
      "step": 11600
    },
    {
      "epoch": 6.082285115303983,
      "grad_norm": 1.9719386100769043,
      "learning_rate": 4.3919854718461986e-05,
      "loss": 0.2875,
      "num_input_tokens_seen": 7572096,
      "step": 11605
    },
    {
      "epoch": 6.084905660377358,
      "grad_norm": 2.068328857421875,
      "learning_rate": 4.391237868244078e-05,
      "loss": 0.2817,
      "num_input_tokens_seen": 7574944,
      "step": 11610
    },
    {
      "epoch": 6.087526205450734,
      "grad_norm": 2.083097457885742,
      "learning_rate": 4.390489869016571e-05,
      "loss": 0.4101,
      "num_input_tokens_seen": 7578080,
      "step": 11615
    },
    {
      "epoch": 6.090146750524109,
      "grad_norm": 3.632598638534546,
      "learning_rate": 4.389741474320152e-05,
      "loss": 0.2445,
      "num_input_tokens_seen": 7580128,
      "step": 11620
    },
    {
      "epoch": 6.0927672955974845,
      "grad_norm": 1.1102230548858643,
      "learning_rate": 4.388992684311376e-05,
      "loss": 0.2898,
      "num_input_tokens_seen": 7583968,
      "step": 11625
    },
    {
      "epoch": 6.09538784067086,
      "grad_norm": 3.375929355621338,
      "learning_rate": 4.388243499146881e-05,
      "loss": 0.1707,
      "num_input_tokens_seen": 7586368,
      "step": 11630
    },
    {
      "epoch": 6.098008385744235,
      "grad_norm": 1.7745407819747925,
      "learning_rate": 4.3874939189833876e-05,
      "loss": 0.2835,
      "num_input_tokens_seen": 7589728,
      "step": 11635
    },
    {
      "epoch": 6.10062893081761,
      "grad_norm": 1.8934129476547241,
      "learning_rate": 4.3867439439777005e-05,
      "loss": 0.3593,
      "num_input_tokens_seen": 7592224,
      "step": 11640
    },
    {
      "epoch": 6.103249475890985,
      "grad_norm": 1.6052181720733643,
      "learning_rate": 4.385993574286705e-05,
      "loss": 0.4025,
      "num_input_tokens_seen": 7595648,
      "step": 11645
    },
    {
      "epoch": 6.10587002096436,
      "grad_norm": 2.7473108768463135,
      "learning_rate": 4.3852428100673705e-05,
      "loss": 0.2343,
      "num_input_tokens_seen": 7598784,
      "step": 11650
    },
    {
      "epoch": 6.1084905660377355,
      "grad_norm": 1.8646758794784546,
      "learning_rate": 4.3844916514767484e-05,
      "loss": 0.2167,
      "num_input_tokens_seen": 7602112,
      "step": 11655
    },
    {
      "epoch": 6.111111111111111,
      "grad_norm": 1.6136879920959473,
      "learning_rate": 4.383740098671972e-05,
      "loss": 0.2772,
      "num_input_tokens_seen": 7608768,
      "step": 11660
    },
    {
      "epoch": 6.113731656184487,
      "grad_norm": 1.5906686782836914,
      "learning_rate": 4.3829881518102576e-05,
      "loss": 0.3409,
      "num_input_tokens_seen": 7612224,
      "step": 11665
    },
    {
      "epoch": 6.116352201257862,
      "grad_norm": 1.3128225803375244,
      "learning_rate": 4.3822358110489045e-05,
      "loss": 0.232,
      "num_input_tokens_seen": 7616768,
      "step": 11670
    },
    {
      "epoch": 6.118972746331237,
      "grad_norm": 1.6483548879623413,
      "learning_rate": 4.381483076545294e-05,
      "loss": 0.2586,
      "num_input_tokens_seen": 7619840,
      "step": 11675
    },
    {
      "epoch": 6.121593291404612,
      "grad_norm": 1.0802711248397827,
      "learning_rate": 4.380729948456888e-05,
      "loss": 0.2732,
      "num_input_tokens_seen": 7623104,
      "step": 11680
    },
    {
      "epoch": 6.1242138364779874,
      "grad_norm": 1.5477356910705566,
      "learning_rate": 4.3799764269412336e-05,
      "loss": 0.3259,
      "num_input_tokens_seen": 7625856,
      "step": 11685
    },
    {
      "epoch": 6.126834381551363,
      "grad_norm": 1.5887620449066162,
      "learning_rate": 4.379222512155958e-05,
      "loss": 0.1839,
      "num_input_tokens_seen": 7629600,
      "step": 11690
    },
    {
      "epoch": 6.129454926624738,
      "grad_norm": 2.1062562465667725,
      "learning_rate": 4.378468204258773e-05,
      "loss": 0.3093,
      "num_input_tokens_seen": 7632960,
      "step": 11695
    },
    {
      "epoch": 6.132075471698113,
      "grad_norm": 2.226646661758423,
      "learning_rate": 4.37771350340747e-05,
      "loss": 0.2889,
      "num_input_tokens_seen": 7635840,
      "step": 11700
    },
    {
      "epoch": 6.134696016771488,
      "grad_norm": 1.3015698194503784,
      "learning_rate": 4.376958409759925e-05,
      "loss": 0.3593,
      "num_input_tokens_seen": 7639232,
      "step": 11705
    },
    {
      "epoch": 6.137316561844864,
      "grad_norm": 1.490415096282959,
      "learning_rate": 4.376202923474093e-05,
      "loss": 0.2223,
      "num_input_tokens_seen": 7645888,
      "step": 11710
    },
    {
      "epoch": 6.139937106918239,
      "grad_norm": 1.8104294538497925,
      "learning_rate": 4.3754470447080145e-05,
      "loss": 0.1898,
      "num_input_tokens_seen": 7648608,
      "step": 11715
    },
    {
      "epoch": 6.1425576519916145,
      "grad_norm": 2.2107717990875244,
      "learning_rate": 4.374690773619811e-05,
      "loss": 0.311,
      "num_input_tokens_seen": 7651072,
      "step": 11720
    },
    {
      "epoch": 6.14517819706499,
      "grad_norm": 1.9735503196716309,
      "learning_rate": 4.373934110367685e-05,
      "loss": 0.4019,
      "num_input_tokens_seen": 7654304,
      "step": 11725
    },
    {
      "epoch": 6.147798742138365,
      "grad_norm": 1.2110239267349243,
      "learning_rate": 4.373177055109922e-05,
      "loss": 0.2781,
      "num_input_tokens_seen": 7657568,
      "step": 11730
    },
    {
      "epoch": 6.15041928721174,
      "grad_norm": 1.1968103647232056,
      "learning_rate": 4.3724196080048896e-05,
      "loss": 0.2119,
      "num_input_tokens_seen": 7661056,
      "step": 11735
    },
    {
      "epoch": 6.153039832285115,
      "grad_norm": 1.4003586769104004,
      "learning_rate": 4.3716617692110364e-05,
      "loss": 0.2783,
      "num_input_tokens_seen": 7667360,
      "step": 11740
    },
    {
      "epoch": 6.15566037735849,
      "grad_norm": 1.2691094875335693,
      "learning_rate": 4.3709035388868943e-05,
      "loss": 0.2552,
      "num_input_tokens_seen": 7671200,
      "step": 11745
    },
    {
      "epoch": 6.1582809224318655,
      "grad_norm": 1.6586878299713135,
      "learning_rate": 4.3701449171910767e-05,
      "loss": 0.4041,
      "num_input_tokens_seen": 7674144,
      "step": 11750
    },
    {
      "epoch": 6.160901467505241,
      "grad_norm": 1.951785922050476,
      "learning_rate": 4.3693859042822774e-05,
      "loss": 0.2412,
      "num_input_tokens_seen": 7677408,
      "step": 11755
    },
    {
      "epoch": 6.163522012578617,
      "grad_norm": 1.4431945085525513,
      "learning_rate": 4.368626500319275e-05,
      "loss": 0.3443,
      "num_input_tokens_seen": 7680160,
      "step": 11760
    },
    {
      "epoch": 6.166142557651992,
      "grad_norm": 1.624428153038025,
      "learning_rate": 4.367866705460927e-05,
      "loss": 0.2978,
      "num_input_tokens_seen": 7682848,
      "step": 11765
    },
    {
      "epoch": 6.168763102725367,
      "grad_norm": 2.7955901622772217,
      "learning_rate": 4.367106519866173e-05,
      "loss": 0.2461,
      "num_input_tokens_seen": 7685728,
      "step": 11770
    },
    {
      "epoch": 6.171383647798742,
      "grad_norm": 1.4802969694137573,
      "learning_rate": 4.366345943694037e-05,
      "loss": 0.2754,
      "num_input_tokens_seen": 7688960,
      "step": 11775
    },
    {
      "epoch": 6.174004192872117,
      "grad_norm": 2.3285796642303467,
      "learning_rate": 4.365584977103622e-05,
      "loss": 0.2846,
      "num_input_tokens_seen": 7691872,
      "step": 11780
    },
    {
      "epoch": 6.176624737945493,
      "grad_norm": 1.650460124015808,
      "learning_rate": 4.364823620254113e-05,
      "loss": 0.2486,
      "num_input_tokens_seen": 7694336,
      "step": 11785
    },
    {
      "epoch": 6.179245283018868,
      "grad_norm": 1.7184226512908936,
      "learning_rate": 4.364061873304778e-05,
      "loss": 0.2542,
      "num_input_tokens_seen": 7697280,
      "step": 11790
    },
    {
      "epoch": 6.181865828092243,
      "grad_norm": 2.7151877880096436,
      "learning_rate": 4.363299736414965e-05,
      "loss": 0.3135,
      "num_input_tokens_seen": 7700640,
      "step": 11795
    },
    {
      "epoch": 6.184486373165618,
      "grad_norm": 2.050933361053467,
      "learning_rate": 4.3625372097441045e-05,
      "loss": 0.2678,
      "num_input_tokens_seen": 7704064,
      "step": 11800
    },
    {
      "epoch": 6.187106918238993,
      "grad_norm": 1.2223414182662964,
      "learning_rate": 4.3617742934517094e-05,
      "loss": 0.2502,
      "num_input_tokens_seen": 7707712,
      "step": 11805
    },
    {
      "epoch": 6.189727463312369,
      "grad_norm": 1.4785218238830566,
      "learning_rate": 4.361010987697373e-05,
      "loss": 0.293,
      "num_input_tokens_seen": 7710656,
      "step": 11810
    },
    {
      "epoch": 6.1923480083857445,
      "grad_norm": 1.1427435874938965,
      "learning_rate": 4.360247292640768e-05,
      "loss": 0.2823,
      "num_input_tokens_seen": 7713664,
      "step": 11815
    },
    {
      "epoch": 6.19496855345912,
      "grad_norm": 2.284151792526245,
      "learning_rate": 4.359483208441653e-05,
      "loss": 0.2791,
      "num_input_tokens_seen": 7716000,
      "step": 11820
    },
    {
      "epoch": 6.197589098532495,
      "grad_norm": 1.17288076877594,
      "learning_rate": 4.3587187352598646e-05,
      "loss": 0.2564,
      "num_input_tokens_seen": 7718432,
      "step": 11825
    },
    {
      "epoch": 6.20020964360587,
      "grad_norm": 4.688851833343506,
      "learning_rate": 4.357953873255323e-05,
      "loss": 0.2777,
      "num_input_tokens_seen": 7721216,
      "step": 11830
    },
    {
      "epoch": 6.202830188679245,
      "grad_norm": 1.5956714153289795,
      "learning_rate": 4.357188622588027e-05,
      "loss": 0.2631,
      "num_input_tokens_seen": 7724128,
      "step": 11835
    },
    {
      "epoch": 6.20545073375262,
      "grad_norm": 1.9839093685150146,
      "learning_rate": 4.356422983418059e-05,
      "loss": 0.2179,
      "num_input_tokens_seen": 7727744,
      "step": 11840
    },
    {
      "epoch": 6.2080712788259955,
      "grad_norm": 1.1742666959762573,
      "learning_rate": 4.3556569559055824e-05,
      "loss": 0.3155,
      "num_input_tokens_seen": 7730880,
      "step": 11845
    },
    {
      "epoch": 6.210691823899371,
      "grad_norm": 3.6291897296905518,
      "learning_rate": 4.3548905402108406e-05,
      "loss": 0.2924,
      "num_input_tokens_seen": 7734048,
      "step": 11850
    },
    {
      "epoch": 6.213312368972747,
      "grad_norm": 2.452671766281128,
      "learning_rate": 4.35412373649416e-05,
      "loss": 0.1934,
      "num_input_tokens_seen": 7736832,
      "step": 11855
    },
    {
      "epoch": 6.215932914046122,
      "grad_norm": 3.293137550354004,
      "learning_rate": 4.353356544915947e-05,
      "loss": 0.2787,
      "num_input_tokens_seen": 7739808,
      "step": 11860
    },
    {
      "epoch": 6.218553459119497,
      "grad_norm": 1.5567700862884521,
      "learning_rate": 4.352588965636689e-05,
      "loss": 0.2358,
      "num_input_tokens_seen": 7743072,
      "step": 11865
    },
    {
      "epoch": 6.221174004192872,
      "grad_norm": 1.4387614727020264,
      "learning_rate": 4.351820998816954e-05,
      "loss": 0.2848,
      "num_input_tokens_seen": 7746336,
      "step": 11870
    },
    {
      "epoch": 6.223794549266247,
      "grad_norm": 1.7450816631317139,
      "learning_rate": 4.3510526446173935e-05,
      "loss": 0.2976,
      "num_input_tokens_seen": 7748864,
      "step": 11875
    },
    {
      "epoch": 6.226415094339623,
      "grad_norm": 2.1032276153564453,
      "learning_rate": 4.350283903198737e-05,
      "loss": 0.2587,
      "num_input_tokens_seen": 7751424,
      "step": 11880
    },
    {
      "epoch": 6.229035639412998,
      "grad_norm": 2.6152961254119873,
      "learning_rate": 4.349514774721797e-05,
      "loss": 0.3045,
      "num_input_tokens_seen": 7754272,
      "step": 11885
    },
    {
      "epoch": 6.231656184486373,
      "grad_norm": 1.6091762781143188,
      "learning_rate": 4.348745259347467e-05,
      "loss": 0.2406,
      "num_input_tokens_seen": 7757920,
      "step": 11890
    },
    {
      "epoch": 6.234276729559748,
      "grad_norm": 2.1887075901031494,
      "learning_rate": 4.347975357236721e-05,
      "loss": 0.25,
      "num_input_tokens_seen": 7761792,
      "step": 11895
    },
    {
      "epoch": 6.236897274633123,
      "grad_norm": 1.2929718494415283,
      "learning_rate": 4.347205068550612e-05,
      "loss": 0.1976,
      "num_input_tokens_seen": 7764832,
      "step": 11900
    },
    {
      "epoch": 6.239517819706499,
      "grad_norm": 1.2429643869400024,
      "learning_rate": 4.3464343934502774e-05,
      "loss": 0.2942,
      "num_input_tokens_seen": 7768640,
      "step": 11905
    },
    {
      "epoch": 6.2421383647798745,
      "grad_norm": 0.7937080264091492,
      "learning_rate": 4.345663332096932e-05,
      "loss": 0.1731,
      "num_input_tokens_seen": 7771360,
      "step": 11910
    },
    {
      "epoch": 6.24475890985325,
      "grad_norm": 1.2751002311706543,
      "learning_rate": 4.344891884651875e-05,
      "loss": 0.1835,
      "num_input_tokens_seen": 7775104,
      "step": 11915
    },
    {
      "epoch": 6.247379454926625,
      "grad_norm": 3.0905232429504395,
      "learning_rate": 4.344120051276483e-05,
      "loss": 0.284,
      "num_input_tokens_seen": 7782720,
      "step": 11920
    },
    {
      "epoch": 6.25,
      "grad_norm": 1.673064947128296,
      "learning_rate": 4.3433478321322155e-05,
      "loss": 0.2237,
      "num_input_tokens_seen": 7785568,
      "step": 11925
    },
    {
      "epoch": 6.252620545073375,
      "grad_norm": 1.7209670543670654,
      "learning_rate": 4.3425752273806106e-05,
      "loss": 0.2848,
      "num_input_tokens_seen": 7789248,
      "step": 11930
    },
    {
      "epoch": 6.25524109014675,
      "grad_norm": 2.154132127761841,
      "learning_rate": 4.34180223718329e-05,
      "loss": 0.2254,
      "num_input_tokens_seen": 7791744,
      "step": 11935
    },
    {
      "epoch": 6.2578616352201255,
      "grad_norm": 1.2871052026748657,
      "learning_rate": 4.341028861701953e-05,
      "loss": 0.3661,
      "num_input_tokens_seen": 7795008,
      "step": 11940
    },
    {
      "epoch": 6.260482180293501,
      "grad_norm": 1.3545353412628174,
      "learning_rate": 4.3402551010983825e-05,
      "loss": 0.2717,
      "num_input_tokens_seen": 7798144,
      "step": 11945
    },
    {
      "epoch": 6.263102725366876,
      "grad_norm": 1.0180639028549194,
      "learning_rate": 4.33948095553444e-05,
      "loss": 0.2409,
      "num_input_tokens_seen": 7800608,
      "step": 11950
    },
    {
      "epoch": 6.265723270440252,
      "grad_norm": 1.4765219688415527,
      "learning_rate": 4.338706425172067e-05,
      "loss": 0.4261,
      "num_input_tokens_seen": 7804000,
      "step": 11955
    },
    {
      "epoch": 6.268343815513627,
      "grad_norm": 1.1862069368362427,
      "learning_rate": 4.3379315101732866e-05,
      "loss": 0.2207,
      "num_input_tokens_seen": 7807296,
      "step": 11960
    },
    {
      "epoch": 6.270964360587002,
      "grad_norm": 2.099250555038452,
      "learning_rate": 4.3371562107002026e-05,
      "loss": 0.248,
      "num_input_tokens_seen": 7810816,
      "step": 11965
    },
    {
      "epoch": 6.273584905660377,
      "grad_norm": 1.2212028503417969,
      "learning_rate": 4.3363805269149984e-05,
      "loss": 0.2689,
      "num_input_tokens_seen": 7814976,
      "step": 11970
    },
    {
      "epoch": 6.276205450733753,
      "grad_norm": 2.309629201889038,
      "learning_rate": 4.335604458979939e-05,
      "loss": 0.4549,
      "num_input_tokens_seen": 7818656,
      "step": 11975
    },
    {
      "epoch": 6.278825995807128,
      "grad_norm": 2.444908857345581,
      "learning_rate": 4.3348280070573686e-05,
      "loss": 0.3105,
      "num_input_tokens_seen": 7821664,
      "step": 11980
    },
    {
      "epoch": 6.281446540880503,
      "grad_norm": 1.0255377292633057,
      "learning_rate": 4.334051171309712e-05,
      "loss": 0.1909,
      "num_input_tokens_seen": 7824800,
      "step": 11985
    },
    {
      "epoch": 6.284067085953878,
      "grad_norm": 1.5435017347335815,
      "learning_rate": 4.333273951899474e-05,
      "loss": 0.3021,
      "num_input_tokens_seen": 7827456,
      "step": 11990
    },
    {
      "epoch": 6.286687631027253,
      "grad_norm": 1.275441288948059,
      "learning_rate": 4.33249634898924e-05,
      "loss": 0.2791,
      "num_input_tokens_seen": 7832064,
      "step": 11995
    },
    {
      "epoch": 6.289308176100629,
      "grad_norm": 1.8096617460250854,
      "learning_rate": 4.331718362741677e-05,
      "loss": 0.2617,
      "num_input_tokens_seen": 7834976,
      "step": 12000
    },
    {
      "epoch": 6.2919287211740045,
      "grad_norm": 1.3468962907791138,
      "learning_rate": 4.330939993319529e-05,
      "loss": 0.2326,
      "num_input_tokens_seen": 7837920,
      "step": 12005
    },
    {
      "epoch": 6.29454926624738,
      "grad_norm": 3.267305850982666,
      "learning_rate": 4.330161240885623e-05,
      "loss": 0.2909,
      "num_input_tokens_seen": 7840448,
      "step": 12010
    },
    {
      "epoch": 6.297169811320755,
      "grad_norm": 1.1302261352539062,
      "learning_rate": 4.3293821056028654e-05,
      "loss": 0.2576,
      "num_input_tokens_seen": 7843680,
      "step": 12015
    },
    {
      "epoch": 6.29979035639413,
      "grad_norm": 1.2964024543762207,
      "learning_rate": 4.3286025876342415e-05,
      "loss": 0.1956,
      "num_input_tokens_seen": 7847264,
      "step": 12020
    },
    {
      "epoch": 6.302410901467505,
      "grad_norm": 1.464773178100586,
      "learning_rate": 4.327822687142819e-05,
      "loss": 0.246,
      "num_input_tokens_seen": 7850144,
      "step": 12025
    },
    {
      "epoch": 6.30503144654088,
      "grad_norm": 1.4522123336791992,
      "learning_rate": 4.327042404291742e-05,
      "loss": 0.2205,
      "num_input_tokens_seen": 7853696,
      "step": 12030
    },
    {
      "epoch": 6.3076519916142555,
      "grad_norm": 1.4478459358215332,
      "learning_rate": 4.326261739244239e-05,
      "loss": 0.2538,
      "num_input_tokens_seen": 7856736,
      "step": 12035
    },
    {
      "epoch": 6.310272536687631,
      "grad_norm": 2.5391132831573486,
      "learning_rate": 4.325480692163615e-05,
      "loss": 0.2565,
      "num_input_tokens_seen": 7860640,
      "step": 12040
    },
    {
      "epoch": 6.312893081761006,
      "grad_norm": 1.133668303489685,
      "learning_rate": 4.324699263213256e-05,
      "loss": 0.1759,
      "num_input_tokens_seen": 7863680,
      "step": 12045
    },
    {
      "epoch": 6.315513626834382,
      "grad_norm": 1.7701354026794434,
      "learning_rate": 4.3239174525566285e-05,
      "loss": 0.3638,
      "num_input_tokens_seen": 7866688,
      "step": 12050
    },
    {
      "epoch": 6.318134171907757,
      "grad_norm": 2.179250478744507,
      "learning_rate": 4.3231352603572784e-05,
      "loss": 0.2107,
      "num_input_tokens_seen": 7869824,
      "step": 12055
    },
    {
      "epoch": 6.320754716981132,
      "grad_norm": 1.6265393495559692,
      "learning_rate": 4.322352686778832e-05,
      "loss": 0.2719,
      "num_input_tokens_seen": 7872640,
      "step": 12060
    },
    {
      "epoch": 6.323375262054507,
      "grad_norm": 1.8993830680847168,
      "learning_rate": 4.3215697319849926e-05,
      "loss": 0.2794,
      "num_input_tokens_seen": 7875424,
      "step": 12065
    },
    {
      "epoch": 6.325995807127883,
      "grad_norm": 0.8665585517883301,
      "learning_rate": 4.3207863961395476e-05,
      "loss": 0.2739,
      "num_input_tokens_seen": 7879360,
      "step": 12070
    },
    {
      "epoch": 6.328616352201258,
      "grad_norm": 1.3843910694122314,
      "learning_rate": 4.320002679406361e-05,
      "loss": 0.2276,
      "num_input_tokens_seen": 7882144,
      "step": 12075
    },
    {
      "epoch": 6.331236897274633,
      "grad_norm": 1.593367338180542,
      "learning_rate": 4.319218581949378e-05,
      "loss": 0.2493,
      "num_input_tokens_seen": 7886176,
      "step": 12080
    },
    {
      "epoch": 6.333857442348008,
      "grad_norm": 1.860529899597168,
      "learning_rate": 4.318434103932622e-05,
      "loss": 0.2634,
      "num_input_tokens_seen": 7889760,
      "step": 12085
    },
    {
      "epoch": 6.336477987421383,
      "grad_norm": 1.372369647026062,
      "learning_rate": 4.317649245520197e-05,
      "loss": 0.3359,
      "num_input_tokens_seen": 7892448,
      "step": 12090
    },
    {
      "epoch": 6.339098532494759,
      "grad_norm": 1.3495707511901855,
      "learning_rate": 4.3168640068762875e-05,
      "loss": 0.2351,
      "num_input_tokens_seen": 7896608,
      "step": 12095
    },
    {
      "epoch": 6.3417190775681345,
      "grad_norm": 2.1286797523498535,
      "learning_rate": 4.316078388165155e-05,
      "loss": 0.2777,
      "num_input_tokens_seen": 7899584,
      "step": 12100
    },
    {
      "epoch": 6.34433962264151,
      "grad_norm": 2.285874843597412,
      "learning_rate": 4.315292389551143e-05,
      "loss": 0.3217,
      "num_input_tokens_seen": 7903008,
      "step": 12105
    },
    {
      "epoch": 6.346960167714885,
      "grad_norm": 2.294077157974243,
      "learning_rate": 4.3145060111986726e-05,
      "loss": 0.3064,
      "num_input_tokens_seen": 7905696,
      "step": 12110
    },
    {
      "epoch": 6.34958071278826,
      "grad_norm": 1.3176958560943604,
      "learning_rate": 4.313719253272246e-05,
      "loss": 0.2959,
      "num_input_tokens_seen": 7908832,
      "step": 12115
    },
    {
      "epoch": 6.352201257861635,
      "grad_norm": 1.7184319496154785,
      "learning_rate": 4.3129321159364436e-05,
      "loss": 0.2699,
      "num_input_tokens_seen": 7911520,
      "step": 12120
    },
    {
      "epoch": 6.35482180293501,
      "grad_norm": 2.1651687622070312,
      "learning_rate": 4.312144599355925e-05,
      "loss": 0.2875,
      "num_input_tokens_seen": 7914848,
      "step": 12125
    },
    {
      "epoch": 6.3574423480083855,
      "grad_norm": 1.545786738395691,
      "learning_rate": 4.31135670369543e-05,
      "loss": 0.2426,
      "num_input_tokens_seen": 7918400,
      "step": 12130
    },
    {
      "epoch": 6.360062893081761,
      "grad_norm": 1.6188892126083374,
      "learning_rate": 4.310568429119779e-05,
      "loss": 0.4564,
      "num_input_tokens_seen": 7921248,
      "step": 12135
    },
    {
      "epoch": 6.362683438155136,
      "grad_norm": 2.5931968688964844,
      "learning_rate": 4.309779775793868e-05,
      "loss": 0.3331,
      "num_input_tokens_seen": 7926240,
      "step": 12140
    },
    {
      "epoch": 6.365303983228512,
      "grad_norm": 1.0657771825790405,
      "learning_rate": 4.308990743882675e-05,
      "loss": 0.3367,
      "num_input_tokens_seen": 7930176,
      "step": 12145
    },
    {
      "epoch": 6.367924528301887,
      "grad_norm": 2.1069164276123047,
      "learning_rate": 4.308201333551256e-05,
      "loss": 0.2678,
      "num_input_tokens_seen": 7933120,
      "step": 12150
    },
    {
      "epoch": 6.370545073375262,
      "grad_norm": 1.4457530975341797,
      "learning_rate": 4.3074115449647476e-05,
      "loss": 0.3705,
      "num_input_tokens_seen": 7935552,
      "step": 12155
    },
    {
      "epoch": 6.373165618448637,
      "grad_norm": 1.456567406654358,
      "learning_rate": 4.306621378288364e-05,
      "loss": 0.1958,
      "num_input_tokens_seen": 7937824,
      "step": 12160
    },
    {
      "epoch": 6.3757861635220126,
      "grad_norm": 1.0060489177703857,
      "learning_rate": 4.305830833687399e-05,
      "loss": 0.3232,
      "num_input_tokens_seen": 7940128,
      "step": 12165
    },
    {
      "epoch": 6.378406708595388,
      "grad_norm": 0.9847808480262756,
      "learning_rate": 4.305039911327225e-05,
      "loss": 0.1764,
      "num_input_tokens_seen": 7943008,
      "step": 12170
    },
    {
      "epoch": 6.381027253668763,
      "grad_norm": 2.4240243434906006,
      "learning_rate": 4.304248611373296e-05,
      "loss": 0.2591,
      "num_input_tokens_seen": 7945472,
      "step": 12175
    },
    {
      "epoch": 6.383647798742138,
      "grad_norm": 1.8314944505691528,
      "learning_rate": 4.30345693399114e-05,
      "loss": 0.1893,
      "num_input_tokens_seen": 7948608,
      "step": 12180
    },
    {
      "epoch": 6.386268343815513,
      "grad_norm": 1.3727045059204102,
      "learning_rate": 4.30266487934637e-05,
      "loss": 0.206,
      "num_input_tokens_seen": 7951296,
      "step": 12185
    },
    {
      "epoch": 6.388888888888889,
      "grad_norm": 1.5476412773132324,
      "learning_rate": 4.3018724476046714e-05,
      "loss": 0.2188,
      "num_input_tokens_seen": 7955488,
      "step": 12190
    },
    {
      "epoch": 6.3915094339622645,
      "grad_norm": 6.333277702331543,
      "learning_rate": 4.301079638931814e-05,
      "loss": 0.2476,
      "num_input_tokens_seen": 7958880,
      "step": 12195
    },
    {
      "epoch": 6.39412997903564,
      "grad_norm": 1.5787358283996582,
      "learning_rate": 4.3002864534936444e-05,
      "loss": 0.2957,
      "num_input_tokens_seen": 7961888,
      "step": 12200
    },
    {
      "epoch": 6.396750524109015,
      "grad_norm": 1.1007940769195557,
      "learning_rate": 4.2994928914560874e-05,
      "loss": 0.3026,
      "num_input_tokens_seen": 7964896,
      "step": 12205
    },
    {
      "epoch": 6.39937106918239,
      "grad_norm": 1.2818793058395386,
      "learning_rate": 4.298698952985147e-05,
      "loss": 0.253,
      "num_input_tokens_seen": 7968032,
      "step": 12210
    },
    {
      "epoch": 6.401991614255765,
      "grad_norm": 2.0945048332214355,
      "learning_rate": 4.297904638246906e-05,
      "loss": 0.238,
      "num_input_tokens_seen": 7971040,
      "step": 12215
    },
    {
      "epoch": 6.40461215932914,
      "grad_norm": 1.6624325513839722,
      "learning_rate": 4.297109947407527e-05,
      "loss": 0.2292,
      "num_input_tokens_seen": 7974560,
      "step": 12220
    },
    {
      "epoch": 6.4072327044025155,
      "grad_norm": 2.1282968521118164,
      "learning_rate": 4.2963148806332486e-05,
      "loss": 0.2222,
      "num_input_tokens_seen": 7978144,
      "step": 12225
    },
    {
      "epoch": 6.409853249475891,
      "grad_norm": 1.3614962100982666,
      "learning_rate": 4.2955194380903915e-05,
      "loss": 0.2171,
      "num_input_tokens_seen": 7981024,
      "step": 12230
    },
    {
      "epoch": 6.412473794549266,
      "grad_norm": 1.2412337064743042,
      "learning_rate": 4.2947236199453524e-05,
      "loss": 0.2826,
      "num_input_tokens_seen": 7984448,
      "step": 12235
    },
    {
      "epoch": 6.415094339622642,
      "grad_norm": 1.9100422859191895,
      "learning_rate": 4.2939274263646065e-05,
      "loss": 0.2751,
      "num_input_tokens_seen": 7987520,
      "step": 12240
    },
    {
      "epoch": 6.417714884696017,
      "grad_norm": 2.5168569087982178,
      "learning_rate": 4.29313085751471e-05,
      "loss": 0.2436,
      "num_input_tokens_seen": 7990784,
      "step": 12245
    },
    {
      "epoch": 6.420335429769392,
      "grad_norm": 1.290651798248291,
      "learning_rate": 4.292333913562294e-05,
      "loss": 0.2312,
      "num_input_tokens_seen": 7993792,
      "step": 12250
    },
    {
      "epoch": 6.422955974842767,
      "grad_norm": 1.04929780960083,
      "learning_rate": 4.291536594674073e-05,
      "loss": 0.2717,
      "num_input_tokens_seen": 7996768,
      "step": 12255
    },
    {
      "epoch": 6.4255765199161425,
      "grad_norm": 1.6918847560882568,
      "learning_rate": 4.2907389010168344e-05,
      "loss": 0.1875,
      "num_input_tokens_seen": 7999424,
      "step": 12260
    },
    {
      "epoch": 6.428197064989518,
      "grad_norm": 1.0996322631835938,
      "learning_rate": 4.2899408327574475e-05,
      "loss": 0.2051,
      "num_input_tokens_seen": 8003456,
      "step": 12265
    },
    {
      "epoch": 6.430817610062893,
      "grad_norm": 1.9622294902801514,
      "learning_rate": 4.289142390062859e-05,
      "loss": 0.2028,
      "num_input_tokens_seen": 8007360,
      "step": 12270
    },
    {
      "epoch": 6.433438155136268,
      "grad_norm": 1.451308012008667,
      "learning_rate": 4.2883435731000946e-05,
      "loss": 0.1841,
      "num_input_tokens_seen": 8010112,
      "step": 12275
    },
    {
      "epoch": 6.436058700209643,
      "grad_norm": 1.2492886781692505,
      "learning_rate": 4.287544382036256e-05,
      "loss": 0.2318,
      "num_input_tokens_seen": 8013024,
      "step": 12280
    },
    {
      "epoch": 6.438679245283019,
      "grad_norm": 3.6936769485473633,
      "learning_rate": 4.286744817038527e-05,
      "loss": 0.2909,
      "num_input_tokens_seen": 8015584,
      "step": 12285
    },
    {
      "epoch": 6.441299790356394,
      "grad_norm": 1.1243083477020264,
      "learning_rate": 4.2859448782741666e-05,
      "loss": 0.2823,
      "num_input_tokens_seen": 8018720,
      "step": 12290
    },
    {
      "epoch": 6.44392033542977,
      "grad_norm": 1.1909661293029785,
      "learning_rate": 4.2851445659105126e-05,
      "loss": 0.2434,
      "num_input_tokens_seen": 8021184,
      "step": 12295
    },
    {
      "epoch": 6.446540880503145,
      "grad_norm": 2.8213114738464355,
      "learning_rate": 4.284343880114982e-05,
      "loss": 0.3227,
      "num_input_tokens_seen": 8024416,
      "step": 12300
    },
    {
      "epoch": 6.44916142557652,
      "grad_norm": 4.180333614349365,
      "learning_rate": 4.2835428210550676e-05,
      "loss": 0.2238,
      "num_input_tokens_seen": 8028160,
      "step": 12305
    },
    {
      "epoch": 6.451781970649895,
      "grad_norm": 2.361889123916626,
      "learning_rate": 4.282741388898344e-05,
      "loss": 0.2055,
      "num_input_tokens_seen": 8031296,
      "step": 12310
    },
    {
      "epoch": 6.45440251572327,
      "grad_norm": 2.330691337585449,
      "learning_rate": 4.281939583812459e-05,
      "loss": 0.3269,
      "num_input_tokens_seen": 8034624,
      "step": 12315
    },
    {
      "epoch": 6.4570230607966455,
      "grad_norm": 1.2345736026763916,
      "learning_rate": 4.281137405965143e-05,
      "loss": 0.2207,
      "num_input_tokens_seen": 8038176,
      "step": 12320
    },
    {
      "epoch": 6.459643605870021,
      "grad_norm": 1.3463741540908813,
      "learning_rate": 4.280334855524203e-05,
      "loss": 0.3014,
      "num_input_tokens_seen": 8041920,
      "step": 12325
    },
    {
      "epoch": 6.462264150943396,
      "grad_norm": 1.3491078615188599,
      "learning_rate": 4.279531932657521e-05,
      "loss": 0.2734,
      "num_input_tokens_seen": 8044992,
      "step": 12330
    },
    {
      "epoch": 6.464884696016772,
      "grad_norm": 2.0974559783935547,
      "learning_rate": 4.278728637533062e-05,
      "loss": 0.2668,
      "num_input_tokens_seen": 8048352,
      "step": 12335
    },
    {
      "epoch": 6.467505241090147,
      "grad_norm": 1.7316374778747559,
      "learning_rate": 4.2779249703188626e-05,
      "loss": 0.2928,
      "num_input_tokens_seen": 8051168,
      "step": 12340
    },
    {
      "epoch": 6.470125786163522,
      "grad_norm": 1.589621663093567,
      "learning_rate": 4.277120931183045e-05,
      "loss": 0.2883,
      "num_input_tokens_seen": 8054848,
      "step": 12345
    },
    {
      "epoch": 6.472746331236897,
      "grad_norm": 2.619861125946045,
      "learning_rate": 4.276316520293803e-05,
      "loss": 0.226,
      "num_input_tokens_seen": 8057952,
      "step": 12350
    },
    {
      "epoch": 6.4753668763102725,
      "grad_norm": 1.0933748483657837,
      "learning_rate": 4.27551173781941e-05,
      "loss": 0.2754,
      "num_input_tokens_seen": 8061344,
      "step": 12355
    },
    {
      "epoch": 6.477987421383648,
      "grad_norm": 2.785205841064453,
      "learning_rate": 4.274706583928216e-05,
      "loss": 0.2918,
      "num_input_tokens_seen": 8064704,
      "step": 12360
    },
    {
      "epoch": 6.480607966457023,
      "grad_norm": 2.126943826675415,
      "learning_rate": 4.273901058788654e-05,
      "loss": 0.3564,
      "num_input_tokens_seen": 8067360,
      "step": 12365
    },
    {
      "epoch": 6.483228511530398,
      "grad_norm": 1.53756582736969,
      "learning_rate": 4.2730951625692275e-05,
      "loss": 0.2921,
      "num_input_tokens_seen": 8071520,
      "step": 12370
    },
    {
      "epoch": 6.485849056603773,
      "grad_norm": 1.2479348182678223,
      "learning_rate": 4.27228889543852e-05,
      "loss": 0.2918,
      "num_input_tokens_seen": 8075680,
      "step": 12375
    },
    {
      "epoch": 6.488469601677149,
      "grad_norm": 3.1489369869232178,
      "learning_rate": 4.2714822575651966e-05,
      "loss": 0.5702,
      "num_input_tokens_seen": 8077600,
      "step": 12380
    },
    {
      "epoch": 6.491090146750524,
      "grad_norm": 1.7920591831207275,
      "learning_rate": 4.270675249117994e-05,
      "loss": 0.2311,
      "num_input_tokens_seen": 8079872,
      "step": 12385
    },
    {
      "epoch": 6.4937106918239,
      "grad_norm": 1.231293797492981,
      "learning_rate": 4.269867870265731e-05,
      "loss": 0.2778,
      "num_input_tokens_seen": 8082720,
      "step": 12390
    },
    {
      "epoch": 6.496331236897275,
      "grad_norm": 1.6301442384719849,
      "learning_rate": 4.269060121177301e-05,
      "loss": 0.3111,
      "num_input_tokens_seen": 8085728,
      "step": 12395
    },
    {
      "epoch": 6.49895178197065,
      "grad_norm": 1.714225172996521,
      "learning_rate": 4.2682520020216764e-05,
      "loss": 0.2802,
      "num_input_tokens_seen": 8089280,
      "step": 12400
    },
    {
      "epoch": 6.501572327044025,
      "grad_norm": 1.4979337453842163,
      "learning_rate": 4.267443512967906e-05,
      "loss": 0.2705,
      "num_input_tokens_seen": 8092576,
      "step": 12405
    },
    {
      "epoch": 6.5041928721174,
      "grad_norm": 2.5358798503875732,
      "learning_rate": 4.2666346541851165e-05,
      "loss": 0.3064,
      "num_input_tokens_seen": 8096032,
      "step": 12410
    },
    {
      "epoch": 6.506813417190775,
      "grad_norm": 2.0106256008148193,
      "learning_rate": 4.265825425842512e-05,
      "loss": 0.3161,
      "num_input_tokens_seen": 8099424,
      "step": 12415
    },
    {
      "epoch": 6.509433962264151,
      "grad_norm": 2.3867034912109375,
      "learning_rate": 4.265015828109375e-05,
      "loss": 0.2802,
      "num_input_tokens_seen": 8101952,
      "step": 12420
    },
    {
      "epoch": 6.512054507337526,
      "grad_norm": 2.1952202320098877,
      "learning_rate": 4.264205861155063e-05,
      "loss": 0.2715,
      "num_input_tokens_seen": 8105472,
      "step": 12425
    },
    {
      "epoch": 6.514675052410902,
      "grad_norm": 1.9681061506271362,
      "learning_rate": 4.263395525149011e-05,
      "loss": 0.422,
      "num_input_tokens_seen": 8108256,
      "step": 12430
    },
    {
      "epoch": 6.517295597484277,
      "grad_norm": 1.6629048585891724,
      "learning_rate": 4.262584820260734e-05,
      "loss": 0.2508,
      "num_input_tokens_seen": 8111008,
      "step": 12435
    },
    {
      "epoch": 6.519916142557652,
      "grad_norm": 1.9525985717773438,
      "learning_rate": 4.261773746659821e-05,
      "loss": 0.283,
      "num_input_tokens_seen": 8114208,
      "step": 12440
    },
    {
      "epoch": 6.522536687631027,
      "grad_norm": 0.9609215259552002,
      "learning_rate": 4.260962304515939e-05,
      "loss": 0.2297,
      "num_input_tokens_seen": 8117376,
      "step": 12445
    },
    {
      "epoch": 6.5251572327044025,
      "grad_norm": 4.677395820617676,
      "learning_rate": 4.2601504939988335e-05,
      "loss": 0.3702,
      "num_input_tokens_seen": 8120256,
      "step": 12450
    },
    {
      "epoch": 6.527777777777778,
      "grad_norm": 3.142465353012085,
      "learning_rate": 4.259338315278325e-05,
      "loss": 0.2122,
      "num_input_tokens_seen": 8122848,
      "step": 12455
    },
    {
      "epoch": 6.530398322851153,
      "grad_norm": 1.662632942199707,
      "learning_rate": 4.258525768524313e-05,
      "loss": 0.2289,
      "num_input_tokens_seen": 8126336,
      "step": 12460
    },
    {
      "epoch": 6.533018867924528,
      "grad_norm": 0.8758715391159058,
      "learning_rate": 4.2577128539067725e-05,
      "loss": 0.1475,
      "num_input_tokens_seen": 8129760,
      "step": 12465
    },
    {
      "epoch": 6.535639412997903,
      "grad_norm": 0.9455729722976685,
      "learning_rate": 4.256899571595756e-05,
      "loss": 0.248,
      "num_input_tokens_seen": 8133248,
      "step": 12470
    },
    {
      "epoch": 6.538259958071279,
      "grad_norm": 1.3326200246810913,
      "learning_rate": 4.256085921761393e-05,
      "loss": 0.259,
      "num_input_tokens_seen": 8136032,
      "step": 12475
    },
    {
      "epoch": 6.540880503144654,
      "grad_norm": 0.9004560112953186,
      "learning_rate": 4.25527190457389e-05,
      "loss": 0.1868,
      "num_input_tokens_seen": 8140224,
      "step": 12480
    },
    {
      "epoch": 6.54350104821803,
      "grad_norm": 1.0572112798690796,
      "learning_rate": 4.2544575202035285e-05,
      "loss": 0.2721,
      "num_input_tokens_seen": 8143424,
      "step": 12485
    },
    {
      "epoch": 6.546121593291405,
      "grad_norm": 1.4689750671386719,
      "learning_rate": 4.253642768820671e-05,
      "loss": 0.281,
      "num_input_tokens_seen": 8146336,
      "step": 12490
    },
    {
      "epoch": 6.54874213836478,
      "grad_norm": 1.9257465600967407,
      "learning_rate": 4.252827650595752e-05,
      "loss": 0.2913,
      "num_input_tokens_seen": 8148768,
      "step": 12495
    },
    {
      "epoch": 6.551362683438155,
      "grad_norm": 2.3331072330474854,
      "learning_rate": 4.252012165699286e-05,
      "loss": 0.1604,
      "num_input_tokens_seen": 8151840,
      "step": 12500
    },
    {
      "epoch": 6.55398322851153,
      "grad_norm": 1.4477580785751343,
      "learning_rate": 4.2511963143018634e-05,
      "loss": 0.2554,
      "num_input_tokens_seen": 8154560,
      "step": 12505
    },
    {
      "epoch": 6.556603773584905,
      "grad_norm": 1.1758997440338135,
      "learning_rate": 4.25038009657415e-05,
      "loss": 0.2734,
      "num_input_tokens_seen": 8158144,
      "step": 12510
    },
    {
      "epoch": 6.559224318658281,
      "grad_norm": 3.615556240081787,
      "learning_rate": 4.24956351268689e-05,
      "loss": 0.3446,
      "num_input_tokens_seen": 8160672,
      "step": 12515
    },
    {
      "epoch": 6.561844863731656,
      "grad_norm": 1.533805251121521,
      "learning_rate": 4.2487465628109036e-05,
      "loss": 0.2941,
      "num_input_tokens_seen": 8163744,
      "step": 12520
    },
    {
      "epoch": 6.564465408805032,
      "grad_norm": 3.0559921264648438,
      "learning_rate": 4.247929247117087e-05,
      "loss": 0.3951,
      "num_input_tokens_seen": 8167104,
      "step": 12525
    },
    {
      "epoch": 6.567085953878407,
      "grad_norm": 1.3763494491577148,
      "learning_rate": 4.247111565776413e-05,
      "loss": 0.2451,
      "num_input_tokens_seen": 8170304,
      "step": 12530
    },
    {
      "epoch": 6.569706498951782,
      "grad_norm": 1.836594581604004,
      "learning_rate": 4.246293518959932e-05,
      "loss": 0.29,
      "num_input_tokens_seen": 8173184,
      "step": 12535
    },
    {
      "epoch": 6.572327044025157,
      "grad_norm": 1.6868796348571777,
      "learning_rate": 4.245475106838769e-05,
      "loss": 0.3108,
      "num_input_tokens_seen": 8177472,
      "step": 12540
    },
    {
      "epoch": 6.5749475890985325,
      "grad_norm": 1.052030086517334,
      "learning_rate": 4.2446563295841266e-05,
      "loss": 0.2487,
      "num_input_tokens_seen": 8180480,
      "step": 12545
    },
    {
      "epoch": 6.577568134171908,
      "grad_norm": 1.504075288772583,
      "learning_rate": 4.243837187367285e-05,
      "loss": 0.2193,
      "num_input_tokens_seen": 8185088,
      "step": 12550
    },
    {
      "epoch": 6.580188679245283,
      "grad_norm": 1.0440324544906616,
      "learning_rate": 4.2430176803595987e-05,
      "loss": 0.3325,
      "num_input_tokens_seen": 8188384,
      "step": 12555
    },
    {
      "epoch": 6.582809224318658,
      "grad_norm": 3.004716396331787,
      "learning_rate": 4.2421978087324985e-05,
      "loss": 0.352,
      "num_input_tokens_seen": 8191360,
      "step": 12560
    },
    {
      "epoch": 6.585429769392033,
      "grad_norm": 2.6273353099823,
      "learning_rate": 4.241377572657493e-05,
      "loss": 0.2219,
      "num_input_tokens_seen": 8194816,
      "step": 12565
    },
    {
      "epoch": 6.588050314465409,
      "grad_norm": 1.6624966859817505,
      "learning_rate": 4.2405569723061654e-05,
      "loss": 0.2233,
      "num_input_tokens_seen": 8198016,
      "step": 12570
    },
    {
      "epoch": 6.590670859538784,
      "grad_norm": 1.7353001832962036,
      "learning_rate": 4.239736007850178e-05,
      "loss": 0.2898,
      "num_input_tokens_seen": 8200448,
      "step": 12575
    },
    {
      "epoch": 6.59329140461216,
      "grad_norm": 1.9935965538024902,
      "learning_rate": 4.238914679461264e-05,
      "loss": 0.3547,
      "num_input_tokens_seen": 8203488,
      "step": 12580
    },
    {
      "epoch": 6.595911949685535,
      "grad_norm": 2.376986026763916,
      "learning_rate": 4.2380929873112386e-05,
      "loss": 0.2705,
      "num_input_tokens_seen": 8206464,
      "step": 12585
    },
    {
      "epoch": 6.59853249475891,
      "grad_norm": 3.0982882976531982,
      "learning_rate": 4.2372709315719896e-05,
      "loss": 0.2228,
      "num_input_tokens_seen": 8212928,
      "step": 12590
    },
    {
      "epoch": 6.601153039832285,
      "grad_norm": 1.1913560628890991,
      "learning_rate": 4.2364485124154806e-05,
      "loss": 0.2614,
      "num_input_tokens_seen": 8216128,
      "step": 12595
    },
    {
      "epoch": 6.60377358490566,
      "grad_norm": 5.520634651184082,
      "learning_rate": 4.235625730013754e-05,
      "loss": 0.3325,
      "num_input_tokens_seen": 8218496,
      "step": 12600
    },
    {
      "epoch": 6.606394129979035,
      "grad_norm": 1.1643708944320679,
      "learning_rate": 4.234802584538926e-05,
      "loss": 0.1761,
      "num_input_tokens_seen": 8221824,
      "step": 12605
    },
    {
      "epoch": 6.609014675052411,
      "grad_norm": 1.6462469100952148,
      "learning_rate": 4.2339790761631884e-05,
      "loss": 0.1941,
      "num_input_tokens_seen": 8225376,
      "step": 12610
    },
    {
      "epoch": 6.611635220125786,
      "grad_norm": 1.842877984046936,
      "learning_rate": 4.233155205058811e-05,
      "loss": 0.3459,
      "num_input_tokens_seen": 8228224,
      "step": 12615
    },
    {
      "epoch": 6.614255765199162,
      "grad_norm": 1.7255979776382446,
      "learning_rate": 4.232330971398137e-05,
      "loss": 0.3149,
      "num_input_tokens_seen": 8230656,
      "step": 12620
    },
    {
      "epoch": 6.616876310272537,
      "grad_norm": 1.710070252418518,
      "learning_rate": 4.2315063753535875e-05,
      "loss": 0.2487,
      "num_input_tokens_seen": 8233408,
      "step": 12625
    },
    {
      "epoch": 6.619496855345912,
      "grad_norm": 1.0971312522888184,
      "learning_rate": 4.230681417097658e-05,
      "loss": 0.294,
      "num_input_tokens_seen": 8237216,
      "step": 12630
    },
    {
      "epoch": 6.622117400419287,
      "grad_norm": 1.9525904655456543,
      "learning_rate": 4.229856096802921e-05,
      "loss": 0.3158,
      "num_input_tokens_seen": 8240000,
      "step": 12635
    },
    {
      "epoch": 6.6247379454926625,
      "grad_norm": 1.892961025238037,
      "learning_rate": 4.229030414642024e-05,
      "loss": 0.3057,
      "num_input_tokens_seen": 8242848,
      "step": 12640
    },
    {
      "epoch": 6.627358490566038,
      "grad_norm": 1.9167678356170654,
      "learning_rate": 4.2282043707876896e-05,
      "loss": 0.2548,
      "num_input_tokens_seen": 8246304,
      "step": 12645
    },
    {
      "epoch": 6.629979035639413,
      "grad_norm": 1.152784824371338,
      "learning_rate": 4.2273779654127175e-05,
      "loss": 0.2659,
      "num_input_tokens_seen": 8249984,
      "step": 12650
    },
    {
      "epoch": 6.632599580712788,
      "grad_norm": 1.3592453002929688,
      "learning_rate": 4.226551198689982e-05,
      "loss": 0.2901,
      "num_input_tokens_seen": 8254240,
      "step": 12655
    },
    {
      "epoch": 6.635220125786163,
      "grad_norm": 1.3299713134765625,
      "learning_rate": 4.225724070792434e-05,
      "loss": 0.23,
      "num_input_tokens_seen": 8257312,
      "step": 12660
    },
    {
      "epoch": 6.637840670859539,
      "grad_norm": 6.335573196411133,
      "learning_rate": 4.2248965818930975e-05,
      "loss": 0.2564,
      "num_input_tokens_seen": 8261216,
      "step": 12665
    },
    {
      "epoch": 6.640461215932914,
      "grad_norm": 1.7008898258209229,
      "learning_rate": 4.224068732165075e-05,
      "loss": 0.2867,
      "num_input_tokens_seen": 8264608,
      "step": 12670
    },
    {
      "epoch": 6.6430817610062896,
      "grad_norm": 2.2000136375427246,
      "learning_rate": 4.223240521781543e-05,
      "loss": 0.2929,
      "num_input_tokens_seen": 8268384,
      "step": 12675
    },
    {
      "epoch": 6.645702306079665,
      "grad_norm": 1.2447577714920044,
      "learning_rate": 4.2224119509157535e-05,
      "loss": 0.2877,
      "num_input_tokens_seen": 8272000,
      "step": 12680
    },
    {
      "epoch": 6.64832285115304,
      "grad_norm": 1.6186343431472778,
      "learning_rate": 4.221583019741033e-05,
      "loss": 0.2497,
      "num_input_tokens_seen": 8275776,
      "step": 12685
    },
    {
      "epoch": 6.650943396226415,
      "grad_norm": 2.997396469116211,
      "learning_rate": 4.2207537284307876e-05,
      "loss": 0.2472,
      "num_input_tokens_seen": 8279008,
      "step": 12690
    },
    {
      "epoch": 6.65356394129979,
      "grad_norm": 1.1802335977554321,
      "learning_rate": 4.219924077158493e-05,
      "loss": 0.3066,
      "num_input_tokens_seen": 8281952,
      "step": 12695
    },
    {
      "epoch": 6.656184486373165,
      "grad_norm": 0.7057392597198486,
      "learning_rate": 4.219094066097703e-05,
      "loss": 0.2535,
      "num_input_tokens_seen": 8285472,
      "step": 12700
    },
    {
      "epoch": 6.658805031446541,
      "grad_norm": 2.8227498531341553,
      "learning_rate": 4.218263695422047e-05,
      "loss": 0.315,
      "num_input_tokens_seen": 8288416,
      "step": 12705
    },
    {
      "epoch": 6.661425576519916,
      "grad_norm": 1.196879506111145,
      "learning_rate": 4.217432965305229e-05,
      "loss": 0.3009,
      "num_input_tokens_seen": 8292064,
      "step": 12710
    },
    {
      "epoch": 6.664046121593291,
      "grad_norm": 2.2514185905456543,
      "learning_rate": 4.2166018759210286e-05,
      "loss": 0.2787,
      "num_input_tokens_seen": 8295008,
      "step": 12715
    },
    {
      "epoch": 6.666666666666667,
      "grad_norm": 1.1265767812728882,
      "learning_rate": 4.215770427443299e-05,
      "loss": 0.2608,
      "num_input_tokens_seen": 8298432,
      "step": 12720
    },
    {
      "epoch": 6.669287211740042,
      "grad_norm": 1.0293902158737183,
      "learning_rate": 4.214938620045971e-05,
      "loss": 0.2551,
      "num_input_tokens_seen": 8302240,
      "step": 12725
    },
    {
      "epoch": 6.671907756813417,
      "grad_norm": 1.5458426475524902,
      "learning_rate": 4.214106453903048e-05,
      "loss": 0.3413,
      "num_input_tokens_seen": 8304960,
      "step": 12730
    },
    {
      "epoch": 6.6745283018867925,
      "grad_norm": 3.169036626815796,
      "learning_rate": 4.2132739291886114e-05,
      "loss": 0.2341,
      "num_input_tokens_seen": 8308320,
      "step": 12735
    },
    {
      "epoch": 6.677148846960168,
      "grad_norm": 1.5716984272003174,
      "learning_rate": 4.212441046076814e-05,
      "loss": 0.2771,
      "num_input_tokens_seen": 8311360,
      "step": 12740
    },
    {
      "epoch": 6.679769392033543,
      "grad_norm": 6.29597806930542,
      "learning_rate": 4.211607804741887e-05,
      "loss": 0.3057,
      "num_input_tokens_seen": 8315680,
      "step": 12745
    },
    {
      "epoch": 6.682389937106918,
      "grad_norm": 2.002016067504883,
      "learning_rate": 4.2107742053581344e-05,
      "loss": 0.3783,
      "num_input_tokens_seen": 8319200,
      "step": 12750
    },
    {
      "epoch": 6.685010482180293,
      "grad_norm": 1.4074193239212036,
      "learning_rate": 4.209940248099935e-05,
      "loss": 0.3324,
      "num_input_tokens_seen": 8322816,
      "step": 12755
    },
    {
      "epoch": 6.687631027253669,
      "grad_norm": 1.7147046327590942,
      "learning_rate": 4.2091059331417445e-05,
      "loss": 0.2937,
      "num_input_tokens_seen": 8325408,
      "step": 12760
    },
    {
      "epoch": 6.690251572327044,
      "grad_norm": 0.8165541887283325,
      "learning_rate": 4.208271260658091e-05,
      "loss": 0.2125,
      "num_input_tokens_seen": 8329408,
      "step": 12765
    },
    {
      "epoch": 6.6928721174004195,
      "grad_norm": 1.9113242626190186,
      "learning_rate": 4.2074362308235785e-05,
      "loss": 0.3601,
      "num_input_tokens_seen": 8332736,
      "step": 12770
    },
    {
      "epoch": 6.695492662473795,
      "grad_norm": 1.43397057056427,
      "learning_rate": 4.206600843812886e-05,
      "loss": 0.2071,
      "num_input_tokens_seen": 8338272,
      "step": 12775
    },
    {
      "epoch": 6.69811320754717,
      "grad_norm": 2.61269211769104,
      "learning_rate": 4.205765099800767e-05,
      "loss": 0.3519,
      "num_input_tokens_seen": 8341472,
      "step": 12780
    },
    {
      "epoch": 6.700733752620545,
      "grad_norm": 1.6211555004119873,
      "learning_rate": 4.204928998962049e-05,
      "loss": 0.2433,
      "num_input_tokens_seen": 8344288,
      "step": 12785
    },
    {
      "epoch": 6.70335429769392,
      "grad_norm": 1.3563259840011597,
      "learning_rate": 4.204092541471636e-05,
      "loss": 0.2253,
      "num_input_tokens_seen": 8347392,
      "step": 12790
    },
    {
      "epoch": 6.705974842767295,
      "grad_norm": 1.4290900230407715,
      "learning_rate": 4.203255727504504e-05,
      "loss": 0.2747,
      "num_input_tokens_seen": 8350656,
      "step": 12795
    },
    {
      "epoch": 6.7085953878406706,
      "grad_norm": 1.3434468507766724,
      "learning_rate": 4.202418557235706e-05,
      "loss": 0.2986,
      "num_input_tokens_seen": 8353888,
      "step": 12800
    },
    {
      "epoch": 6.711215932914046,
      "grad_norm": 1.5903042554855347,
      "learning_rate": 4.201581030840368e-05,
      "loss": 0.287,
      "num_input_tokens_seen": 8356736,
      "step": 12805
    },
    {
      "epoch": 6.713836477987421,
      "grad_norm": 1.9508787393569946,
      "learning_rate": 4.200743148493691e-05,
      "loss": 0.2592,
      "num_input_tokens_seen": 8360992,
      "step": 12810
    },
    {
      "epoch": 6.716457023060797,
      "grad_norm": 2.0070931911468506,
      "learning_rate": 4.199904910370951e-05,
      "loss": 0.2817,
      "num_input_tokens_seen": 8364000,
      "step": 12815
    },
    {
      "epoch": 6.719077568134172,
      "grad_norm": 2.1847259998321533,
      "learning_rate": 4.1990663166474964e-05,
      "loss": 0.2243,
      "num_input_tokens_seen": 8366304,
      "step": 12820
    },
    {
      "epoch": 6.721698113207547,
      "grad_norm": 1.6391090154647827,
      "learning_rate": 4.1982273674987524e-05,
      "loss": 0.2229,
      "num_input_tokens_seen": 8369792,
      "step": 12825
    },
    {
      "epoch": 6.7243186582809225,
      "grad_norm": 2.3195741176605225,
      "learning_rate": 4.197388063100218e-05,
      "loss": 0.2507,
      "num_input_tokens_seen": 8372384,
      "step": 12830
    },
    {
      "epoch": 6.726939203354298,
      "grad_norm": 1.6839518547058105,
      "learning_rate": 4.196548403627465e-05,
      "loss": 0.2425,
      "num_input_tokens_seen": 8374720,
      "step": 12835
    },
    {
      "epoch": 6.729559748427673,
      "grad_norm": 1.5448899269104004,
      "learning_rate": 4.195708389256142e-05,
      "loss": 0.3182,
      "num_input_tokens_seen": 8378144,
      "step": 12840
    },
    {
      "epoch": 6.732180293501048,
      "grad_norm": 1.0813034772872925,
      "learning_rate": 4.194868020161969e-05,
      "loss": 0.2313,
      "num_input_tokens_seen": 8381536,
      "step": 12845
    },
    {
      "epoch": 6.734800838574423,
      "grad_norm": 1.3519681692123413,
      "learning_rate": 4.1940272965207425e-05,
      "loss": 0.3097,
      "num_input_tokens_seen": 8384416,
      "step": 12850
    },
    {
      "epoch": 6.737421383647799,
      "grad_norm": 1.236314058303833,
      "learning_rate": 4.1931862185083326e-05,
      "loss": 0.2179,
      "num_input_tokens_seen": 8387616,
      "step": 12855
    },
    {
      "epoch": 6.740041928721174,
      "grad_norm": 1.8865394592285156,
      "learning_rate": 4.192344786300682e-05,
      "loss": 0.2659,
      "num_input_tokens_seen": 8390208,
      "step": 12860
    },
    {
      "epoch": 6.7426624737945495,
      "grad_norm": 1.79462730884552,
      "learning_rate": 4.19150300007381e-05,
      "loss": 0.2411,
      "num_input_tokens_seen": 8393792,
      "step": 12865
    },
    {
      "epoch": 6.745283018867925,
      "grad_norm": 1.1903326511383057,
      "learning_rate": 4.190660860003806e-05,
      "loss": 0.3817,
      "num_input_tokens_seen": 8397120,
      "step": 12870
    },
    {
      "epoch": 6.7479035639413,
      "grad_norm": 2.418393611907959,
      "learning_rate": 4.189818366266841e-05,
      "loss": 0.3276,
      "num_input_tokens_seen": 8400160,
      "step": 12875
    },
    {
      "epoch": 6.750524109014675,
      "grad_norm": 1.7452101707458496,
      "learning_rate": 4.188975519039151e-05,
      "loss": 0.3961,
      "num_input_tokens_seen": 8404032,
      "step": 12880
    },
    {
      "epoch": 6.75314465408805,
      "grad_norm": 3.2177093029022217,
      "learning_rate": 4.188132318497051e-05,
      "loss": 0.2786,
      "num_input_tokens_seen": 8407392,
      "step": 12885
    },
    {
      "epoch": 6.755765199161425,
      "grad_norm": 2.300633668899536,
      "learning_rate": 4.1872887648169304e-05,
      "loss": 0.3082,
      "num_input_tokens_seen": 8410656,
      "step": 12890
    },
    {
      "epoch": 6.7583857442348005,
      "grad_norm": 3.4871034622192383,
      "learning_rate": 4.186444858175248e-05,
      "loss": 0.2326,
      "num_input_tokens_seen": 8413344,
      "step": 12895
    },
    {
      "epoch": 6.761006289308176,
      "grad_norm": 1.1209297180175781,
      "learning_rate": 4.1856005987485424e-05,
      "loss": 0.2678,
      "num_input_tokens_seen": 8416960,
      "step": 12900
    },
    {
      "epoch": 6.763626834381551,
      "grad_norm": 1.5882409811019897,
      "learning_rate": 4.1847559867134215e-05,
      "loss": 0.362,
      "num_input_tokens_seen": 8420352,
      "step": 12905
    },
    {
      "epoch": 6.766247379454927,
      "grad_norm": 1.4353889226913452,
      "learning_rate": 4.1839110222465693e-05,
      "loss": 0.3407,
      "num_input_tokens_seen": 8423360,
      "step": 12910
    },
    {
      "epoch": 6.768867924528302,
      "grad_norm": 1.3728383779525757,
      "learning_rate": 4.183065705524742e-05,
      "loss": 0.167,
      "num_input_tokens_seen": 8426816,
      "step": 12915
    },
    {
      "epoch": 6.771488469601677,
      "grad_norm": 1.0429383516311646,
      "learning_rate": 4.1822200367247696e-05,
      "loss": 0.3739,
      "num_input_tokens_seen": 8429856,
      "step": 12920
    },
    {
      "epoch": 6.774109014675052,
      "grad_norm": 1.0795152187347412,
      "learning_rate": 4.181374016023559e-05,
      "loss": 0.3671,
      "num_input_tokens_seen": 8433056,
      "step": 12925
    },
    {
      "epoch": 6.776729559748428,
      "grad_norm": 1.2663935422897339,
      "learning_rate": 4.1805276435980854e-05,
      "loss": 0.3753,
      "num_input_tokens_seen": 8436448,
      "step": 12930
    },
    {
      "epoch": 6.779350104821803,
      "grad_norm": 1.1647248268127441,
      "learning_rate": 4.1796809196254014e-05,
      "loss": 0.1996,
      "num_input_tokens_seen": 8439200,
      "step": 12935
    },
    {
      "epoch": 6.781970649895178,
      "grad_norm": 4.90639066696167,
      "learning_rate": 4.178833844282632e-05,
      "loss": 0.2706,
      "num_input_tokens_seen": 8443072,
      "step": 12940
    },
    {
      "epoch": 6.784591194968553,
      "grad_norm": 1.3174314498901367,
      "learning_rate": 4.177986417746976e-05,
      "loss": 0.2476,
      "num_input_tokens_seen": 8445952,
      "step": 12945
    },
    {
      "epoch": 6.787211740041929,
      "grad_norm": 2.134330987930298,
      "learning_rate": 4.177138640195704e-05,
      "loss": 0.2863,
      "num_input_tokens_seen": 8448768,
      "step": 12950
    },
    {
      "epoch": 6.789832285115304,
      "grad_norm": 1.809795618057251,
      "learning_rate": 4.176290511806162e-05,
      "loss": 0.2142,
      "num_input_tokens_seen": 8451232,
      "step": 12955
    },
    {
      "epoch": 6.7924528301886795,
      "grad_norm": 1.083619236946106,
      "learning_rate": 4.1754420327557706e-05,
      "loss": 0.2201,
      "num_input_tokens_seen": 8454688,
      "step": 12960
    },
    {
      "epoch": 6.795073375262055,
      "grad_norm": 2.1375339031219482,
      "learning_rate": 4.1745932032220195e-05,
      "loss": 0.1959,
      "num_input_tokens_seen": 8458048,
      "step": 12965
    },
    {
      "epoch": 6.79769392033543,
      "grad_norm": 1.7976901531219482,
      "learning_rate": 4.173744023382474e-05,
      "loss": 0.2735,
      "num_input_tokens_seen": 8460960,
      "step": 12970
    },
    {
      "epoch": 6.800314465408805,
      "grad_norm": 1.7747963666915894,
      "learning_rate": 4.172894493414776e-05,
      "loss": 0.3089,
      "num_input_tokens_seen": 8464448,
      "step": 12975
    },
    {
      "epoch": 6.80293501048218,
      "grad_norm": 2.078061819076538,
      "learning_rate": 4.172044613496634e-05,
      "loss": 0.1969,
      "num_input_tokens_seen": 8468000,
      "step": 12980
    },
    {
      "epoch": 6.805555555555555,
      "grad_norm": 1.2175238132476807,
      "learning_rate": 4.171194383805835e-05,
      "loss": 0.3087,
      "num_input_tokens_seen": 8472288,
      "step": 12985
    },
    {
      "epoch": 6.8081761006289305,
      "grad_norm": 1.7194043397903442,
      "learning_rate": 4.170343804520237e-05,
      "loss": 0.2262,
      "num_input_tokens_seen": 8475616,
      "step": 12990
    },
    {
      "epoch": 6.810796645702306,
      "grad_norm": 2.546626091003418,
      "learning_rate": 4.169492875817771e-05,
      "loss": 0.334,
      "num_input_tokens_seen": 8477984,
      "step": 12995
    },
    {
      "epoch": 6.813417190775681,
      "grad_norm": 1.9283674955368042,
      "learning_rate": 4.1686415978764415e-05,
      "loss": 0.2881,
      "num_input_tokens_seen": 8480864,
      "step": 13000
    },
    {
      "epoch": 6.816037735849057,
      "grad_norm": 1.8311845064163208,
      "learning_rate": 4.167789970874327e-05,
      "loss": 0.3074,
      "num_input_tokens_seen": 8483680,
      "step": 13005
    },
    {
      "epoch": 6.818658280922432,
      "grad_norm": 4.01028299331665,
      "learning_rate": 4.166937994989578e-05,
      "loss": 0.2456,
      "num_input_tokens_seen": 8487296,
      "step": 13010
    },
    {
      "epoch": 6.821278825995807,
      "grad_norm": 1.585087537765503,
      "learning_rate": 4.166085670400418e-05,
      "loss": 0.2285,
      "num_input_tokens_seen": 8490528,
      "step": 13015
    },
    {
      "epoch": 6.823899371069182,
      "grad_norm": 1.7582452297210693,
      "learning_rate": 4.165232997285143e-05,
      "loss": 0.2975,
      "num_input_tokens_seen": 8493536,
      "step": 13020
    },
    {
      "epoch": 6.826519916142558,
      "grad_norm": 5.006033420562744,
      "learning_rate": 4.164379975822124e-05,
      "loss": 0.3657,
      "num_input_tokens_seen": 8497472,
      "step": 13025
    },
    {
      "epoch": 6.829140461215933,
      "grad_norm": 1.693211555480957,
      "learning_rate": 4.1635266061898015e-05,
      "loss": 0.3719,
      "num_input_tokens_seen": 8500288,
      "step": 13030
    },
    {
      "epoch": 6.831761006289308,
      "grad_norm": 1.549646258354187,
      "learning_rate": 4.162672888566691e-05,
      "loss": 0.2841,
      "num_input_tokens_seen": 8503136,
      "step": 13035
    },
    {
      "epoch": 6.834381551362683,
      "grad_norm": 1.3273216485977173,
      "learning_rate": 4.1618188231313816e-05,
      "loss": 0.2136,
      "num_input_tokens_seen": 8507168,
      "step": 13040
    },
    {
      "epoch": 6.837002096436059,
      "grad_norm": 2.1101531982421875,
      "learning_rate": 4.1609644100625335e-05,
      "loss": 0.2856,
      "num_input_tokens_seen": 8509760,
      "step": 13045
    },
    {
      "epoch": 6.839622641509434,
      "grad_norm": 1.5490013360977173,
      "learning_rate": 4.16010964953888e-05,
      "loss": 0.2606,
      "num_input_tokens_seen": 8512672,
      "step": 13050
    },
    {
      "epoch": 6.8422431865828095,
      "grad_norm": 2.2924578189849854,
      "learning_rate": 4.159254541739227e-05,
      "loss": 0.2385,
      "num_input_tokens_seen": 8516320,
      "step": 13055
    },
    {
      "epoch": 6.844863731656185,
      "grad_norm": 1.8667104244232178,
      "learning_rate": 4.158399086842454e-05,
      "loss": 0.2962,
      "num_input_tokens_seen": 8519456,
      "step": 13060
    },
    {
      "epoch": 6.84748427672956,
      "grad_norm": 3.3148672580718994,
      "learning_rate": 4.157543285027512e-05,
      "loss": 0.3517,
      "num_input_tokens_seen": 8522560,
      "step": 13065
    },
    {
      "epoch": 6.850104821802935,
      "grad_norm": 1.353354811668396,
      "learning_rate": 4.156687136473425e-05,
      "loss": 0.2624,
      "num_input_tokens_seen": 8525792,
      "step": 13070
    },
    {
      "epoch": 6.85272536687631,
      "grad_norm": 1.654451608657837,
      "learning_rate": 4.1558306413592894e-05,
      "loss": 0.2423,
      "num_input_tokens_seen": 8528704,
      "step": 13075
    },
    {
      "epoch": 6.855345911949685,
      "grad_norm": 1.9832239151000977,
      "learning_rate": 4.1549737998642745e-05,
      "loss": 0.2656,
      "num_input_tokens_seen": 8532000,
      "step": 13080
    },
    {
      "epoch": 6.8579664570230605,
      "grad_norm": 1.4271063804626465,
      "learning_rate": 4.154116612167622e-05,
      "loss": 0.3394,
      "num_input_tokens_seen": 8535136,
      "step": 13085
    },
    {
      "epoch": 6.860587002096436,
      "grad_norm": 2.666250705718994,
      "learning_rate": 4.1532590784486444e-05,
      "loss": 0.3322,
      "num_input_tokens_seen": 8538592,
      "step": 13090
    },
    {
      "epoch": 6.863207547169811,
      "grad_norm": 1.5839186906814575,
      "learning_rate": 4.1524011988867294e-05,
      "loss": 0.2947,
      "num_input_tokens_seen": 8541344,
      "step": 13095
    },
    {
      "epoch": 6.865828092243187,
      "grad_norm": 1.8639219999313354,
      "learning_rate": 4.151542973661336e-05,
      "loss": 0.2721,
      "num_input_tokens_seen": 8544576,
      "step": 13100
    },
    {
      "epoch": 6.868448637316562,
      "grad_norm": 1.6025460958480835,
      "learning_rate": 4.150684402951994e-05,
      "loss": 0.252,
      "num_input_tokens_seen": 8547680,
      "step": 13105
    },
    {
      "epoch": 6.871069182389937,
      "grad_norm": 1.7305530309677124,
      "learning_rate": 4.149825486938306e-05,
      "loss": 0.3193,
      "num_input_tokens_seen": 8550720,
      "step": 13110
    },
    {
      "epoch": 6.873689727463312,
      "grad_norm": 2.7987844944000244,
      "learning_rate": 4.148966225799949e-05,
      "loss": 0.2629,
      "num_input_tokens_seen": 8554112,
      "step": 13115
    },
    {
      "epoch": 6.876310272536688,
      "grad_norm": 1.1771129369735718,
      "learning_rate": 4.14810661971667e-05,
      "loss": 0.231,
      "num_input_tokens_seen": 8557856,
      "step": 13120
    },
    {
      "epoch": 6.878930817610063,
      "grad_norm": 1.5492831468582153,
      "learning_rate": 4.147246668868289e-05,
      "loss": 0.3161,
      "num_input_tokens_seen": 8560960,
      "step": 13125
    },
    {
      "epoch": 6.881551362683438,
      "grad_norm": 1.655911922454834,
      "learning_rate": 4.1463863734346964e-05,
      "loss": 0.2703,
      "num_input_tokens_seen": 8564288,
      "step": 13130
    },
    {
      "epoch": 6.884171907756813,
      "grad_norm": 1.284380555152893,
      "learning_rate": 4.145525733595859e-05,
      "loss": 0.272,
      "num_input_tokens_seen": 8568608,
      "step": 13135
    },
    {
      "epoch": 6.886792452830189,
      "grad_norm": 1.2129749059677124,
      "learning_rate": 4.14466474953181e-05,
      "loss": 0.257,
      "num_input_tokens_seen": 8571872,
      "step": 13140
    },
    {
      "epoch": 6.889412997903564,
      "grad_norm": 2.0156476497650146,
      "learning_rate": 4.14380342142266e-05,
      "loss": 0.3464,
      "num_input_tokens_seen": 8574336,
      "step": 13145
    },
    {
      "epoch": 6.8920335429769395,
      "grad_norm": 1.5318408012390137,
      "learning_rate": 4.142941749448587e-05,
      "loss": 0.2578,
      "num_input_tokens_seen": 8577248,
      "step": 13150
    },
    {
      "epoch": 6.894654088050315,
      "grad_norm": 1.9030349254608154,
      "learning_rate": 4.142079733789843e-05,
      "loss": 0.261,
      "num_input_tokens_seen": 8581120,
      "step": 13155
    },
    {
      "epoch": 6.89727463312369,
      "grad_norm": 2.3719260692596436,
      "learning_rate": 4.141217374626753e-05,
      "loss": 0.1865,
      "num_input_tokens_seen": 8585280,
      "step": 13160
    },
    {
      "epoch": 6.899895178197065,
      "grad_norm": 1.2238625288009644,
      "learning_rate": 4.140354672139713e-05,
      "loss": 0.2527,
      "num_input_tokens_seen": 8590464,
      "step": 13165
    },
    {
      "epoch": 6.90251572327044,
      "grad_norm": 0.9663563370704651,
      "learning_rate": 4.1394916265091886e-05,
      "loss": 0.2469,
      "num_input_tokens_seen": 8593408,
      "step": 13170
    },
    {
      "epoch": 6.905136268343815,
      "grad_norm": 2.684237241744995,
      "learning_rate": 4.1386282379157214e-05,
      "loss": 0.2744,
      "num_input_tokens_seen": 8596832,
      "step": 13175
    },
    {
      "epoch": 6.9077568134171905,
      "grad_norm": 1.3079127073287964,
      "learning_rate": 4.137764506539921e-05,
      "loss": 0.1927,
      "num_input_tokens_seen": 8600000,
      "step": 13180
    },
    {
      "epoch": 6.910377358490566,
      "grad_norm": 2.0573501586914062,
      "learning_rate": 4.136900432562471e-05,
      "loss": 0.2986,
      "num_input_tokens_seen": 8602560,
      "step": 13185
    },
    {
      "epoch": 6.912997903563941,
      "grad_norm": 3.2381765842437744,
      "learning_rate": 4.136036016164123e-05,
      "loss": 0.3812,
      "num_input_tokens_seen": 8605088,
      "step": 13190
    },
    {
      "epoch": 6.915618448637317,
      "grad_norm": 1.6440364122390747,
      "learning_rate": 4.135171257525707e-05,
      "loss": 0.3185,
      "num_input_tokens_seen": 8608704,
      "step": 13195
    },
    {
      "epoch": 6.918238993710692,
      "grad_norm": 1.0493086576461792,
      "learning_rate": 4.1343061568281195e-05,
      "loss": 0.2784,
      "num_input_tokens_seen": 8612320,
      "step": 13200
    },
    {
      "epoch": 6.920859538784067,
      "grad_norm": 1.5678907632827759,
      "learning_rate": 4.133440714252328e-05,
      "loss": 0.3229,
      "num_input_tokens_seen": 8615360,
      "step": 13205
    },
    {
      "epoch": 6.923480083857442,
      "grad_norm": 1.5349470376968384,
      "learning_rate": 4.132574929979375e-05,
      "loss": 0.2349,
      "num_input_tokens_seen": 8618464,
      "step": 13210
    },
    {
      "epoch": 6.926100628930818,
      "grad_norm": 2.1689209938049316,
      "learning_rate": 4.1317088041903714e-05,
      "loss": 0.1969,
      "num_input_tokens_seen": 8621664,
      "step": 13215
    },
    {
      "epoch": 6.928721174004193,
      "grad_norm": 1.141116738319397,
      "learning_rate": 4.1308423370665015e-05,
      "loss": 0.253,
      "num_input_tokens_seen": 8624608,
      "step": 13220
    },
    {
      "epoch": 6.931341719077568,
      "grad_norm": 2.0336503982543945,
      "learning_rate": 4.1299755287890206e-05,
      "loss": 0.1698,
      "num_input_tokens_seen": 8626880,
      "step": 13225
    },
    {
      "epoch": 6.933962264150943,
      "grad_norm": 2.5355498790740967,
      "learning_rate": 4.129108379539255e-05,
      "loss": 0.3189,
      "num_input_tokens_seen": 8630880,
      "step": 13230
    },
    {
      "epoch": 6.936582809224319,
      "grad_norm": 2.3492610454559326,
      "learning_rate": 4.128240889498602e-05,
      "loss": 0.3059,
      "num_input_tokens_seen": 8633856,
      "step": 13235
    },
    {
      "epoch": 6.939203354297694,
      "grad_norm": 1.426665186882019,
      "learning_rate": 4.127373058848531e-05,
      "loss": 0.3186,
      "num_input_tokens_seen": 8636288,
      "step": 13240
    },
    {
      "epoch": 6.9418238993710695,
      "grad_norm": 2.5004096031188965,
      "learning_rate": 4.126504887770582e-05,
      "loss": 0.4111,
      "num_input_tokens_seen": 8638880,
      "step": 13245
    },
    {
      "epoch": 6.944444444444445,
      "grad_norm": 1.7064517736434937,
      "learning_rate": 4.1256363764463666e-05,
      "loss": 0.302,
      "num_input_tokens_seen": 8642080,
      "step": 13250
    },
    {
      "epoch": 6.94706498951782,
      "grad_norm": 1.410300612449646,
      "learning_rate": 4.1247675250575676e-05,
      "loss": 0.2274,
      "num_input_tokens_seen": 8644672,
      "step": 13255
    },
    {
      "epoch": 6.949685534591195,
      "grad_norm": 1.6319754123687744,
      "learning_rate": 4.1238983337859396e-05,
      "loss": 0.2466,
      "num_input_tokens_seen": 8648160,
      "step": 13260
    },
    {
      "epoch": 6.95230607966457,
      "grad_norm": 1.2973988056182861,
      "learning_rate": 4.123028802813305e-05,
      "loss": 0.2718,
      "num_input_tokens_seen": 8651168,
      "step": 13265
    },
    {
      "epoch": 6.954926624737945,
      "grad_norm": 1.1536232233047485,
      "learning_rate": 4.1221589323215625e-05,
      "loss": 0.2603,
      "num_input_tokens_seen": 8653984,
      "step": 13270
    },
    {
      "epoch": 6.9575471698113205,
      "grad_norm": 1.3308331966400146,
      "learning_rate": 4.121288722492678e-05,
      "loss": 0.213,
      "num_input_tokens_seen": 8656544,
      "step": 13275
    },
    {
      "epoch": 6.960167714884696,
      "grad_norm": 3.5062692165374756,
      "learning_rate": 4.120418173508688e-05,
      "loss": 0.2016,
      "num_input_tokens_seen": 8659456,
      "step": 13280
    },
    {
      "epoch": 6.962788259958071,
      "grad_norm": 2.062352180480957,
      "learning_rate": 4.1195472855517045e-05,
      "loss": 0.2454,
      "num_input_tokens_seen": 8662112,
      "step": 13285
    },
    {
      "epoch": 6.965408805031447,
      "grad_norm": 5.725493431091309,
      "learning_rate": 4.1186760588039045e-05,
      "loss": 0.3273,
      "num_input_tokens_seen": 8665760,
      "step": 13290
    },
    {
      "epoch": 6.968029350104822,
      "grad_norm": 1.682883381843567,
      "learning_rate": 4.11780449344754e-05,
      "loss": 0.3076,
      "num_input_tokens_seen": 8669152,
      "step": 13295
    },
    {
      "epoch": 6.970649895178197,
      "grad_norm": 2.6870994567871094,
      "learning_rate": 4.1169325896649326e-05,
      "loss": 0.2999,
      "num_input_tokens_seen": 8671648,
      "step": 13300
    },
    {
      "epoch": 6.973270440251572,
      "grad_norm": 1.1448583602905273,
      "learning_rate": 4.116060347638474e-05,
      "loss": 0.2996,
      "num_input_tokens_seen": 8676192,
      "step": 13305
    },
    {
      "epoch": 6.975890985324948,
      "grad_norm": 2.2946043014526367,
      "learning_rate": 4.115187767550628e-05,
      "loss": 0.332,
      "num_input_tokens_seen": 8679008,
      "step": 13310
    },
    {
      "epoch": 6.978511530398323,
      "grad_norm": 1.8064621686935425,
      "learning_rate": 4.114314849583928e-05,
      "loss": 0.2476,
      "num_input_tokens_seen": 8682848,
      "step": 13315
    },
    {
      "epoch": 6.981132075471698,
      "grad_norm": 2.613740921020508,
      "learning_rate": 4.113441593920979e-05,
      "loss": 0.2224,
      "num_input_tokens_seen": 8685376,
      "step": 13320
    },
    {
      "epoch": 6.983752620545073,
      "grad_norm": 2.18414044380188,
      "learning_rate": 4.112568000744454e-05,
      "loss": 0.3673,
      "num_input_tokens_seen": 8688672,
      "step": 13325
    },
    {
      "epoch": 6.986373165618449,
      "grad_norm": 1.091057538986206,
      "learning_rate": 4.1116940702371025e-05,
      "loss": 0.3237,
      "num_input_tokens_seen": 8692256,
      "step": 13330
    },
    {
      "epoch": 6.988993710691824,
      "grad_norm": 4.1842122077941895,
      "learning_rate": 4.1108198025817376e-05,
      "loss": 0.32,
      "num_input_tokens_seen": 8695968,
      "step": 13335
    },
    {
      "epoch": 6.9916142557651995,
      "grad_norm": 1.3865710496902466,
      "learning_rate": 4.1099451979612476e-05,
      "loss": 0.2383,
      "num_input_tokens_seen": 8699808,
      "step": 13340
    },
    {
      "epoch": 6.994234800838575,
      "grad_norm": 2.2314980030059814,
      "learning_rate": 4.109070256558589e-05,
      "loss": 0.3224,
      "num_input_tokens_seen": 8702432,
      "step": 13345
    },
    {
      "epoch": 6.99685534591195,
      "grad_norm": 3.194265842437744,
      "learning_rate": 4.108194978556791e-05,
      "loss": 0.3304,
      "num_input_tokens_seen": 8705792,
      "step": 13350
    },
    {
      "epoch": 6.999475890985325,
      "grad_norm": 1.2680060863494873,
      "learning_rate": 4.1073193641389495e-05,
      "loss": 0.2981,
      "num_input_tokens_seen": 8709056,
      "step": 13355
    },
    {
      "epoch": 7.0,
      "eval_loss": 0.5451971888542175,
      "eval_runtime": 12.1184,
      "eval_samples_per_second": 69.976,
      "eval_steps_per_second": 17.494,
      "num_input_tokens_seen": 8709168,
      "step": 13356
    },
    {
      "epoch": 7.0020964360587,
      "grad_norm": 0.7639363408088684,
      "learning_rate": 4.1064434134882345e-05,
      "loss": 0.254,
      "num_input_tokens_seen": 8711536,
      "step": 13360
    },
    {
      "epoch": 7.004716981132075,
      "grad_norm": 1.2426396608352661,
      "learning_rate": 4.1055671267878847e-05,
      "loss": 0.2719,
      "num_input_tokens_seen": 8715728,
      "step": 13365
    },
    {
      "epoch": 7.0073375262054505,
      "grad_norm": 1.097050428390503,
      "learning_rate": 4.1046905042212106e-05,
      "loss": 0.2329,
      "num_input_tokens_seen": 8719216,
      "step": 13370
    },
    {
      "epoch": 7.009958071278826,
      "grad_norm": 1.5068910121917725,
      "learning_rate": 4.103813545971589e-05,
      "loss": 0.1829,
      "num_input_tokens_seen": 8722160,
      "step": 13375
    },
    {
      "epoch": 7.012578616352202,
      "grad_norm": 1.3251522779464722,
      "learning_rate": 4.10293625222247e-05,
      "loss": 0.1638,
      "num_input_tokens_seen": 8725584,
      "step": 13380
    },
    {
      "epoch": 7.015199161425577,
      "grad_norm": 2.9109485149383545,
      "learning_rate": 4.102058623157376e-05,
      "loss": 0.141,
      "num_input_tokens_seen": 8728272,
      "step": 13385
    },
    {
      "epoch": 7.017819706498952,
      "grad_norm": 2.70029354095459,
      "learning_rate": 4.101180658959894e-05,
      "loss": 0.1983,
      "num_input_tokens_seen": 8731568,
      "step": 13390
    },
    {
      "epoch": 7.020440251572327,
      "grad_norm": 6.843349456787109,
      "learning_rate": 4.1003023598136866e-05,
      "loss": 0.2523,
      "num_input_tokens_seen": 8735120,
      "step": 13395
    },
    {
      "epoch": 7.023060796645702,
      "grad_norm": 0.9017950892448425,
      "learning_rate": 4.099423725902482e-05,
      "loss": 0.1721,
      "num_input_tokens_seen": 8737552,
      "step": 13400
    },
    {
      "epoch": 7.0256813417190775,
      "grad_norm": 1.9201719760894775,
      "learning_rate": 4.098544757410081e-05,
      "loss": 0.1675,
      "num_input_tokens_seen": 8741712,
      "step": 13405
    },
    {
      "epoch": 7.028301886792453,
      "grad_norm": 1.4226871728897095,
      "learning_rate": 4.097665454520354e-05,
      "loss": 0.2332,
      "num_input_tokens_seen": 8744464,
      "step": 13410
    },
    {
      "epoch": 7.030922431865828,
      "grad_norm": 1.1078593730926514,
      "learning_rate": 4.096785817417241e-05,
      "loss": 0.2108,
      "num_input_tokens_seen": 8748048,
      "step": 13415
    },
    {
      "epoch": 7.033542976939203,
      "grad_norm": 2.351576805114746,
      "learning_rate": 4.0959058462847524e-05,
      "loss": 0.2048,
      "num_input_tokens_seen": 8751024,
      "step": 13420
    },
    {
      "epoch": 7.036163522012578,
      "grad_norm": 2.1917684078216553,
      "learning_rate": 4.095025541306968e-05,
      "loss": 0.2154,
      "num_input_tokens_seen": 8753264,
      "step": 13425
    },
    {
      "epoch": 7.038784067085954,
      "grad_norm": 1.3229292631149292,
      "learning_rate": 4.0941449026680374e-05,
      "loss": 0.1905,
      "num_input_tokens_seen": 8756240,
      "step": 13430
    },
    {
      "epoch": 7.0414046121593294,
      "grad_norm": 1.9706674814224243,
      "learning_rate": 4.09326393055218e-05,
      "loss": 0.1818,
      "num_input_tokens_seen": 8759184,
      "step": 13435
    },
    {
      "epoch": 7.044025157232705,
      "grad_norm": 1.9330073595046997,
      "learning_rate": 4.0923826251436845e-05,
      "loss": 0.2872,
      "num_input_tokens_seen": 8762864,
      "step": 13440
    },
    {
      "epoch": 7.04664570230608,
      "grad_norm": 1.8820875883102417,
      "learning_rate": 4.0915009866269115e-05,
      "loss": 0.1485,
      "num_input_tokens_seen": 8767856,
      "step": 13445
    },
    {
      "epoch": 7.049266247379455,
      "grad_norm": 2.1625800132751465,
      "learning_rate": 4.090619015186288e-05,
      "loss": 0.2101,
      "num_input_tokens_seen": 8771344,
      "step": 13450
    },
    {
      "epoch": 7.05188679245283,
      "grad_norm": 1.0126672983169556,
      "learning_rate": 4.0897367110063145e-05,
      "loss": 0.1891,
      "num_input_tokens_seen": 8775216,
      "step": 13455
    },
    {
      "epoch": 7.054507337526205,
      "grad_norm": 1.4764630794525146,
      "learning_rate": 4.088854074271557e-05,
      "loss": 0.2384,
      "num_input_tokens_seen": 8779536,
      "step": 13460
    },
    {
      "epoch": 7.0571278825995805,
      "grad_norm": 3.0591564178466797,
      "learning_rate": 4.0879711051666534e-05,
      "loss": 0.2997,
      "num_input_tokens_seen": 8782640,
      "step": 13465
    },
    {
      "epoch": 7.059748427672956,
      "grad_norm": 1.6695927381515503,
      "learning_rate": 4.087087803876313e-05,
      "loss": 0.2786,
      "num_input_tokens_seen": 8786384,
      "step": 13470
    },
    {
      "epoch": 7.062368972746331,
      "grad_norm": 2.186532974243164,
      "learning_rate": 4.086204170585308e-05,
      "loss": 0.2306,
      "num_input_tokens_seen": 8789616,
      "step": 13475
    },
    {
      "epoch": 7.064989517819707,
      "grad_norm": 2.6056010723114014,
      "learning_rate": 4.085320205478488e-05,
      "loss": 0.198,
      "num_input_tokens_seen": 8792816,
      "step": 13480
    },
    {
      "epoch": 7.067610062893082,
      "grad_norm": 1.70539391040802,
      "learning_rate": 4.084435908740768e-05,
      "loss": 0.2114,
      "num_input_tokens_seen": 8796784,
      "step": 13485
    },
    {
      "epoch": 7.070230607966457,
      "grad_norm": 1.6465797424316406,
      "learning_rate": 4.0835512805571306e-05,
      "loss": 0.2369,
      "num_input_tokens_seen": 8800016,
      "step": 13490
    },
    {
      "epoch": 7.072851153039832,
      "grad_norm": 1.5607553720474243,
      "learning_rate": 4.082666321112633e-05,
      "loss": 0.2439,
      "num_input_tokens_seen": 8803088,
      "step": 13495
    },
    {
      "epoch": 7.0754716981132075,
      "grad_norm": 1.5780869722366333,
      "learning_rate": 4.081781030592396e-05,
      "loss": 0.1818,
      "num_input_tokens_seen": 8806096,
      "step": 13500
    },
    {
      "epoch": 7.078092243186583,
      "grad_norm": 2.128662109375,
      "learning_rate": 4.080895409181613e-05,
      "loss": 0.194,
      "num_input_tokens_seen": 8808624,
      "step": 13505
    },
    {
      "epoch": 7.080712788259958,
      "grad_norm": 5.227298259735107,
      "learning_rate": 4.080009457065547e-05,
      "loss": 0.2866,
      "num_input_tokens_seen": 8811920,
      "step": 13510
    },
    {
      "epoch": 7.083333333333333,
      "grad_norm": 0.3811264634132385,
      "learning_rate": 4.0791231744295286e-05,
      "loss": 0.1622,
      "num_input_tokens_seen": 8817904,
      "step": 13515
    },
    {
      "epoch": 7.085953878406708,
      "grad_norm": 2.2999589443206787,
      "learning_rate": 4.078236561458957e-05,
      "loss": 0.1766,
      "num_input_tokens_seen": 8820496,
      "step": 13520
    },
    {
      "epoch": 7.088574423480084,
      "grad_norm": 4.154388427734375,
      "learning_rate": 4.077349618339302e-05,
      "loss": 0.2265,
      "num_input_tokens_seen": 8823120,
      "step": 13525
    },
    {
      "epoch": 7.091194968553459,
      "grad_norm": 1.3012363910675049,
      "learning_rate": 4.076462345256103e-05,
      "loss": 0.215,
      "num_input_tokens_seen": 8826192,
      "step": 13530
    },
    {
      "epoch": 7.093815513626835,
      "grad_norm": 1.4729307889938354,
      "learning_rate": 4.0755747423949655e-05,
      "loss": 0.2137,
      "num_input_tokens_seen": 8829456,
      "step": 13535
    },
    {
      "epoch": 7.09643605870021,
      "grad_norm": 1.3709092140197754,
      "learning_rate": 4.074686809941568e-05,
      "loss": 0.3422,
      "num_input_tokens_seen": 8832112,
      "step": 13540
    },
    {
      "epoch": 7.099056603773585,
      "grad_norm": 2.1467576026916504,
      "learning_rate": 4.0737985480816546e-05,
      "loss": 0.3352,
      "num_input_tokens_seen": 8834832,
      "step": 13545
    },
    {
      "epoch": 7.10167714884696,
      "grad_norm": 1.3972066640853882,
      "learning_rate": 4.0729099570010395e-05,
      "loss": 0.2224,
      "num_input_tokens_seen": 8838480,
      "step": 13550
    },
    {
      "epoch": 7.104297693920335,
      "grad_norm": 3.14361572265625,
      "learning_rate": 4.072021036885607e-05,
      "loss": 0.1646,
      "num_input_tokens_seen": 8842128,
      "step": 13555
    },
    {
      "epoch": 7.1069182389937104,
      "grad_norm": 3.143906354904175,
      "learning_rate": 4.0711317879213075e-05,
      "loss": 0.1893,
      "num_input_tokens_seen": 8845872,
      "step": 13560
    },
    {
      "epoch": 7.109538784067086,
      "grad_norm": 2.047757863998413,
      "learning_rate": 4.070242210294162e-05,
      "loss": 0.2284,
      "num_input_tokens_seen": 8848720,
      "step": 13565
    },
    {
      "epoch": 7.112159329140461,
      "grad_norm": 3.295119047164917,
      "learning_rate": 4.069352304190261e-05,
      "loss": 0.1679,
      "num_input_tokens_seen": 8853072,
      "step": 13570
    },
    {
      "epoch": 7.114779874213837,
      "grad_norm": 3.1292312145233154,
      "learning_rate": 4.068462069795761e-05,
      "loss": 0.1697,
      "num_input_tokens_seen": 8856208,
      "step": 13575
    },
    {
      "epoch": 7.117400419287212,
      "grad_norm": 2.560983896255493,
      "learning_rate": 4.067571507296891e-05,
      "loss": 0.2666,
      "num_input_tokens_seen": 8859600,
      "step": 13580
    },
    {
      "epoch": 7.120020964360587,
      "grad_norm": 2.182321548461914,
      "learning_rate": 4.0666806168799455e-05,
      "loss": 0.292,
      "num_input_tokens_seen": 8862576,
      "step": 13585
    },
    {
      "epoch": 7.122641509433962,
      "grad_norm": 1.532529592514038,
      "learning_rate": 4.0657893987312876e-05,
      "loss": 0.1881,
      "num_input_tokens_seen": 8866704,
      "step": 13590
    },
    {
      "epoch": 7.1252620545073375,
      "grad_norm": 1.6989258527755737,
      "learning_rate": 4.064897853037352e-05,
      "loss": 0.2199,
      "num_input_tokens_seen": 8869584,
      "step": 13595
    },
    {
      "epoch": 7.127882599580713,
      "grad_norm": 1.4087063074111938,
      "learning_rate": 4.064005979984639e-05,
      "loss": 0.2175,
      "num_input_tokens_seen": 8873040,
      "step": 13600
    },
    {
      "epoch": 7.130503144654088,
      "grad_norm": 1.8534421920776367,
      "learning_rate": 4.0631137797597176e-05,
      "loss": 0.2305,
      "num_input_tokens_seen": 8876208,
      "step": 13605
    },
    {
      "epoch": 7.133123689727463,
      "grad_norm": 1.0615628957748413,
      "learning_rate": 4.0622212525492266e-05,
      "loss": 0.1826,
      "num_input_tokens_seen": 8880240,
      "step": 13610
    },
    {
      "epoch": 7.135744234800838,
      "grad_norm": 2.1478822231292725,
      "learning_rate": 4.061328398539872e-05,
      "loss": 0.207,
      "num_input_tokens_seen": 8883280,
      "step": 13615
    },
    {
      "epoch": 7.138364779874214,
      "grad_norm": 1.5723048448562622,
      "learning_rate": 4.060435217918429e-05,
      "loss": 0.2107,
      "num_input_tokens_seen": 8886192,
      "step": 13620
    },
    {
      "epoch": 7.140985324947589,
      "grad_norm": 1.2255297899246216,
      "learning_rate": 4.0595417108717405e-05,
      "loss": 0.1664,
      "num_input_tokens_seen": 8889456,
      "step": 13625
    },
    {
      "epoch": 7.143605870020965,
      "grad_norm": 1.999008059501648,
      "learning_rate": 4.0586478775867185e-05,
      "loss": 0.1675,
      "num_input_tokens_seen": 8892528,
      "step": 13630
    },
    {
      "epoch": 7.14622641509434,
      "grad_norm": 2.0375139713287354,
      "learning_rate": 4.057753718250343e-05,
      "loss": 0.2031,
      "num_input_tokens_seen": 8895824,
      "step": 13635
    },
    {
      "epoch": 7.148846960167715,
      "grad_norm": 2.534358024597168,
      "learning_rate": 4.056859233049661e-05,
      "loss": 0.2705,
      "num_input_tokens_seen": 8898544,
      "step": 13640
    },
    {
      "epoch": 7.15146750524109,
      "grad_norm": 1.6390739679336548,
      "learning_rate": 4.05596442217179e-05,
      "loss": 0.3023,
      "num_input_tokens_seen": 8902480,
      "step": 13645
    },
    {
      "epoch": 7.154088050314465,
      "grad_norm": 1.5913687944412231,
      "learning_rate": 4.055069285803912e-05,
      "loss": 0.2548,
      "num_input_tokens_seen": 8905968,
      "step": 13650
    },
    {
      "epoch": 7.15670859538784,
      "grad_norm": 2.5100839138031006,
      "learning_rate": 4.0541738241332815e-05,
      "loss": 0.221,
      "num_input_tokens_seen": 8909296,
      "step": 13655
    },
    {
      "epoch": 7.159329140461216,
      "grad_norm": 1.8378961086273193,
      "learning_rate": 4.053278037347217e-05,
      "loss": 0.1603,
      "num_input_tokens_seen": 8911792,
      "step": 13660
    },
    {
      "epoch": 7.161949685534591,
      "grad_norm": 1.8061968088150024,
      "learning_rate": 4.052381925633109e-05,
      "loss": 0.254,
      "num_input_tokens_seen": 8915280,
      "step": 13665
    },
    {
      "epoch": 7.164570230607967,
      "grad_norm": 2.102419853210449,
      "learning_rate": 4.051485489178412e-05,
      "loss": 0.195,
      "num_input_tokens_seen": 8918736,
      "step": 13670
    },
    {
      "epoch": 7.167190775681342,
      "grad_norm": 2.046360492706299,
      "learning_rate": 4.05058872817065e-05,
      "loss": 0.2701,
      "num_input_tokens_seen": 8921776,
      "step": 13675
    },
    {
      "epoch": 7.169811320754717,
      "grad_norm": 0.8671771287918091,
      "learning_rate": 4.0496916427974177e-05,
      "loss": 0.3166,
      "num_input_tokens_seen": 8925264,
      "step": 13680
    },
    {
      "epoch": 7.172431865828092,
      "grad_norm": 0.9392975568771362,
      "learning_rate": 4.048794233246373e-05,
      "loss": 0.2536,
      "num_input_tokens_seen": 8928400,
      "step": 13685
    },
    {
      "epoch": 7.1750524109014675,
      "grad_norm": 1.6353472471237183,
      "learning_rate": 4.0478964997052435e-05,
      "loss": 0.2694,
      "num_input_tokens_seen": 8931792,
      "step": 13690
    },
    {
      "epoch": 7.177672955974843,
      "grad_norm": 1.492975115776062,
      "learning_rate": 4.046998442361827e-05,
      "loss": 0.1922,
      "num_input_tokens_seen": 8934768,
      "step": 13695
    },
    {
      "epoch": 7.180293501048218,
      "grad_norm": 4.9950852394104,
      "learning_rate": 4.046100061403984e-05,
      "loss": 0.1336,
      "num_input_tokens_seen": 8937552,
      "step": 13700
    },
    {
      "epoch": 7.182914046121593,
      "grad_norm": 2.499849557876587,
      "learning_rate": 4.045201357019648e-05,
      "loss": 0.2878,
      "num_input_tokens_seen": 8941520,
      "step": 13705
    },
    {
      "epoch": 7.185534591194968,
      "grad_norm": 1.7634708881378174,
      "learning_rate": 4.044302329396816e-05,
      "loss": 0.1806,
      "num_input_tokens_seen": 8945616,
      "step": 13710
    },
    {
      "epoch": 7.188155136268344,
      "grad_norm": 2.435648202896118,
      "learning_rate": 4.0434029787235564e-05,
      "loss": 0.268,
      "num_input_tokens_seen": 8948592,
      "step": 13715
    },
    {
      "epoch": 7.190775681341719,
      "grad_norm": 2.484220027923584,
      "learning_rate": 4.042503305188e-05,
      "loss": 0.179,
      "num_input_tokens_seen": 8951344,
      "step": 13720
    },
    {
      "epoch": 7.193396226415095,
      "grad_norm": 2.787581205368042,
      "learning_rate": 4.041603308978351e-05,
      "loss": 0.2403,
      "num_input_tokens_seen": 8955056,
      "step": 13725
    },
    {
      "epoch": 7.19601677148847,
      "grad_norm": 1.3582109212875366,
      "learning_rate": 4.040702990282878e-05,
      "loss": 0.1762,
      "num_input_tokens_seen": 8958128,
      "step": 13730
    },
    {
      "epoch": 7.198637316561845,
      "grad_norm": 1.628902792930603,
      "learning_rate": 4.039802349289914e-05,
      "loss": 0.1815,
      "num_input_tokens_seen": 8961840,
      "step": 13735
    },
    {
      "epoch": 7.20125786163522,
      "grad_norm": 1.5942866802215576,
      "learning_rate": 4.038901386187868e-05,
      "loss": 0.2508,
      "num_input_tokens_seen": 8965040,
      "step": 13740
    },
    {
      "epoch": 7.203878406708595,
      "grad_norm": 1.2765109539031982,
      "learning_rate": 4.038000101165208e-05,
      "loss": 0.1935,
      "num_input_tokens_seen": 8969200,
      "step": 13745
    },
    {
      "epoch": 7.20649895178197,
      "grad_norm": 2.0738654136657715,
      "learning_rate": 4.0370984944104725e-05,
      "loss": 0.3238,
      "num_input_tokens_seen": 8972624,
      "step": 13750
    },
    {
      "epoch": 7.209119496855346,
      "grad_norm": 1.1222496032714844,
      "learning_rate": 4.0361965661122695e-05,
      "loss": 0.1843,
      "num_input_tokens_seen": 8975728,
      "step": 13755
    },
    {
      "epoch": 7.211740041928721,
      "grad_norm": 1.6071982383728027,
      "learning_rate": 4.035294316459271e-05,
      "loss": 0.2915,
      "num_input_tokens_seen": 8979312,
      "step": 13760
    },
    {
      "epoch": 7.214360587002097,
      "grad_norm": 2.140167713165283,
      "learning_rate": 4.034391745640215e-05,
      "loss": 0.2036,
      "num_input_tokens_seen": 8982032,
      "step": 13765
    },
    {
      "epoch": 7.216981132075472,
      "grad_norm": 1.9816468954086304,
      "learning_rate": 4.033488853843913e-05,
      "loss": 0.1956,
      "num_input_tokens_seen": 8984432,
      "step": 13770
    },
    {
      "epoch": 7.219601677148847,
      "grad_norm": 2.2526237964630127,
      "learning_rate": 4.0325856412592363e-05,
      "loss": 0.2478,
      "num_input_tokens_seen": 8988624,
      "step": 13775
    },
    {
      "epoch": 7.222222222222222,
      "grad_norm": 3.103212594985962,
      "learning_rate": 4.031682108075129e-05,
      "loss": 0.3659,
      "num_input_tokens_seen": 8991664,
      "step": 13780
    },
    {
      "epoch": 7.2248427672955975,
      "grad_norm": 1.7723997831344604,
      "learning_rate": 4.030778254480599e-05,
      "loss": 0.1833,
      "num_input_tokens_seen": 8994928,
      "step": 13785
    },
    {
      "epoch": 7.227463312368973,
      "grad_norm": 16.246395111083984,
      "learning_rate": 4.029874080664722e-05,
      "loss": 0.2755,
      "num_input_tokens_seen": 8997872,
      "step": 13790
    },
    {
      "epoch": 7.230083857442348,
      "grad_norm": 1.2908791303634644,
      "learning_rate": 4.0289695868166414e-05,
      "loss": 0.2129,
      "num_input_tokens_seen": 9002384,
      "step": 13795
    },
    {
      "epoch": 7.232704402515723,
      "grad_norm": 2.7727770805358887,
      "learning_rate": 4.028064773125566e-05,
      "loss": 0.1938,
      "num_input_tokens_seen": 9005296,
      "step": 13800
    },
    {
      "epoch": 7.235324947589098,
      "grad_norm": 1.2719744443893433,
      "learning_rate": 4.0271596397807734e-05,
      "loss": 0.2309,
      "num_input_tokens_seen": 9009968,
      "step": 13805
    },
    {
      "epoch": 7.237945492662474,
      "grad_norm": 2.5994532108306885,
      "learning_rate": 4.026254186971607e-05,
      "loss": 0.2218,
      "num_input_tokens_seen": 9013776,
      "step": 13810
    },
    {
      "epoch": 7.240566037735849,
      "grad_norm": 2.457440137863159,
      "learning_rate": 4.025348414887477e-05,
      "loss": 0.1452,
      "num_input_tokens_seen": 9017072,
      "step": 13815
    },
    {
      "epoch": 7.243186582809225,
      "grad_norm": 2.6577954292297363,
      "learning_rate": 4.024442323717861e-05,
      "loss": 0.3654,
      "num_input_tokens_seen": 9020272,
      "step": 13820
    },
    {
      "epoch": 7.2458071278826,
      "grad_norm": 1.3933945894241333,
      "learning_rate": 4.023535913652302e-05,
      "loss": 0.209,
      "num_input_tokens_seen": 9024528,
      "step": 13825
    },
    {
      "epoch": 7.248427672955975,
      "grad_norm": 1.7841178178787231,
      "learning_rate": 4.022629184880412e-05,
      "loss": 0.1725,
      "num_input_tokens_seen": 9027376,
      "step": 13830
    },
    {
      "epoch": 7.25104821802935,
      "grad_norm": 4.061572074890137,
      "learning_rate": 4.021722137591867e-05,
      "loss": 0.2722,
      "num_input_tokens_seen": 9030832,
      "step": 13835
    },
    {
      "epoch": 7.253668763102725,
      "grad_norm": 1.9582170248031616,
      "learning_rate": 4.020814771976412e-05,
      "loss": 0.3165,
      "num_input_tokens_seen": 9034064,
      "step": 13840
    },
    {
      "epoch": 7.2562893081761,
      "grad_norm": 2.1095242500305176,
      "learning_rate": 4.019907088223857e-05,
      "loss": 0.213,
      "num_input_tokens_seen": 9037072,
      "step": 13845
    },
    {
      "epoch": 7.258909853249476,
      "grad_norm": 0.9060115218162537,
      "learning_rate": 4.018999086524079e-05,
      "loss": 0.1832,
      "num_input_tokens_seen": 9040624,
      "step": 13850
    },
    {
      "epoch": 7.261530398322851,
      "grad_norm": 2.016829013824463,
      "learning_rate": 4.018090767067022e-05,
      "loss": 0.2036,
      "num_input_tokens_seen": 9043792,
      "step": 13855
    },
    {
      "epoch": 7.264150943396227,
      "grad_norm": 1.581013798713684,
      "learning_rate": 4.017182130042696e-05,
      "loss": 0.2418,
      "num_input_tokens_seen": 9046640,
      "step": 13860
    },
    {
      "epoch": 7.266771488469602,
      "grad_norm": 2.362001895904541,
      "learning_rate": 4.0162731756411764e-05,
      "loss": 0.1812,
      "num_input_tokens_seen": 9049328,
      "step": 13865
    },
    {
      "epoch": 7.269392033542977,
      "grad_norm": 1.2251133918762207,
      "learning_rate": 4.015363904052608e-05,
      "loss": 0.2426,
      "num_input_tokens_seen": 9052944,
      "step": 13870
    },
    {
      "epoch": 7.272012578616352,
      "grad_norm": 1.9430924654006958,
      "learning_rate": 4.014454315467199e-05,
      "loss": 0.1436,
      "num_input_tokens_seen": 9055344,
      "step": 13875
    },
    {
      "epoch": 7.2746331236897275,
      "grad_norm": 1.6103882789611816,
      "learning_rate": 4.013544410075224e-05,
      "loss": 0.2061,
      "num_input_tokens_seen": 9058928,
      "step": 13880
    },
    {
      "epoch": 7.277253668763103,
      "grad_norm": 2.4307589530944824,
      "learning_rate": 4.0126341880670273e-05,
      "loss": 0.2344,
      "num_input_tokens_seen": 9062384,
      "step": 13885
    },
    {
      "epoch": 7.279874213836478,
      "grad_norm": 2.286276340484619,
      "learning_rate": 4.011723649633014e-05,
      "loss": 0.2106,
      "num_input_tokens_seen": 9066288,
      "step": 13890
    },
    {
      "epoch": 7.282494758909853,
      "grad_norm": 1.6443082094192505,
      "learning_rate": 4.0108127949636606e-05,
      "loss": 0.2204,
      "num_input_tokens_seen": 9070000,
      "step": 13895
    },
    {
      "epoch": 7.285115303983228,
      "grad_norm": 1.7878854274749756,
      "learning_rate": 4.009901624249507e-05,
      "loss": 0.1835,
      "num_input_tokens_seen": 9072784,
      "step": 13900
    },
    {
      "epoch": 7.287735849056604,
      "grad_norm": 1.8023215532302856,
      "learning_rate": 4.008990137681159e-05,
      "loss": 0.1643,
      "num_input_tokens_seen": 9075632,
      "step": 13905
    },
    {
      "epoch": 7.290356394129979,
      "grad_norm": 2.038156747817993,
      "learning_rate": 4.00807833544929e-05,
      "loss": 0.3614,
      "num_input_tokens_seen": 9078448,
      "step": 13910
    },
    {
      "epoch": 7.2929769392033545,
      "grad_norm": 2.4355037212371826,
      "learning_rate": 4.0071662177446376e-05,
      "loss": 0.2092,
      "num_input_tokens_seen": 9082352,
      "step": 13915
    },
    {
      "epoch": 7.29559748427673,
      "grad_norm": 2.7543513774871826,
      "learning_rate": 4.006253784758008e-05,
      "loss": 0.287,
      "num_input_tokens_seen": 9085104,
      "step": 13920
    },
    {
      "epoch": 7.298218029350105,
      "grad_norm": 0.674187421798706,
      "learning_rate": 4.0053410366802704e-05,
      "loss": 0.2451,
      "num_input_tokens_seen": 9087856,
      "step": 13925
    },
    {
      "epoch": 7.30083857442348,
      "grad_norm": 0.9020017385482788,
      "learning_rate": 4.004427973702362e-05,
      "loss": 0.1712,
      "num_input_tokens_seen": 9091216,
      "step": 13930
    },
    {
      "epoch": 7.303459119496855,
      "grad_norm": 3.06016206741333,
      "learning_rate": 4.003514596015285e-05,
      "loss": 0.2161,
      "num_input_tokens_seen": 9094256,
      "step": 13935
    },
    {
      "epoch": 7.30607966457023,
      "grad_norm": 1.3625845909118652,
      "learning_rate": 4.002600903810108e-05,
      "loss": 0.1844,
      "num_input_tokens_seen": 9098192,
      "step": 13940
    },
    {
      "epoch": 7.308700209643606,
      "grad_norm": 1.8629388809204102,
      "learning_rate": 4.001686897277963e-05,
      "loss": 0.2041,
      "num_input_tokens_seen": 9100432,
      "step": 13945
    },
    {
      "epoch": 7.311320754716981,
      "grad_norm": 1.387587547302246,
      "learning_rate": 4.000772576610052e-05,
      "loss": 0.2177,
      "num_input_tokens_seen": 9104016,
      "step": 13950
    },
    {
      "epoch": 7.313941299790357,
      "grad_norm": 1.9933652877807617,
      "learning_rate": 3.99985794199764e-05,
      "loss": 0.4481,
      "num_input_tokens_seen": 9107088,
      "step": 13955
    },
    {
      "epoch": 7.316561844863732,
      "grad_norm": 1.6633774042129517,
      "learning_rate": 3.998942993632058e-05,
      "loss": 0.2608,
      "num_input_tokens_seen": 9110064,
      "step": 13960
    },
    {
      "epoch": 7.319182389937107,
      "grad_norm": 3.479335069656372,
      "learning_rate": 3.998027731704702e-05,
      "loss": 0.2806,
      "num_input_tokens_seen": 9113040,
      "step": 13965
    },
    {
      "epoch": 7.321802935010482,
      "grad_norm": 1.7378255128860474,
      "learning_rate": 3.9971121564070354e-05,
      "loss": 0.1948,
      "num_input_tokens_seen": 9116208,
      "step": 13970
    },
    {
      "epoch": 7.3244234800838575,
      "grad_norm": 1.9431320428848267,
      "learning_rate": 3.996196267930585e-05,
      "loss": 0.215,
      "num_input_tokens_seen": 9119280,
      "step": 13975
    },
    {
      "epoch": 7.327044025157233,
      "grad_norm": 2.426982879638672,
      "learning_rate": 3.995280066466945e-05,
      "loss": 0.2862,
      "num_input_tokens_seen": 9124048,
      "step": 13980
    },
    {
      "epoch": 7.329664570230608,
      "grad_norm": 3.8441011905670166,
      "learning_rate": 3.9943635522077746e-05,
      "loss": 0.2418,
      "num_input_tokens_seen": 9128080,
      "step": 13985
    },
    {
      "epoch": 7.332285115303983,
      "grad_norm": 1.0955395698547363,
      "learning_rate": 3.993446725344797e-05,
      "loss": 0.202,
      "num_input_tokens_seen": 9131120,
      "step": 13990
    },
    {
      "epoch": 7.334905660377358,
      "grad_norm": 1.1766636371612549,
      "learning_rate": 3.992529586069803e-05,
      "loss": 0.2796,
      "num_input_tokens_seen": 9133776,
      "step": 13995
    },
    {
      "epoch": 7.337526205450734,
      "grad_norm": 2.5023579597473145,
      "learning_rate": 3.991612134574648e-05,
      "loss": 0.2992,
      "num_input_tokens_seen": 9137680,
      "step": 14000
    },
    {
      "epoch": 7.340146750524109,
      "grad_norm": 3.852931499481201,
      "learning_rate": 3.99069437105125e-05,
      "loss": 0.1835,
      "num_input_tokens_seen": 9140912,
      "step": 14005
    },
    {
      "epoch": 7.3427672955974845,
      "grad_norm": 2.4733686447143555,
      "learning_rate": 3.9897762956915976e-05,
      "loss": 0.2218,
      "num_input_tokens_seen": 9144432,
      "step": 14010
    },
    {
      "epoch": 7.34538784067086,
      "grad_norm": 3.053619861602783,
      "learning_rate": 3.9888579086877403e-05,
      "loss": 0.297,
      "num_input_tokens_seen": 9147504,
      "step": 14015
    },
    {
      "epoch": 7.348008385744235,
      "grad_norm": 1.5544705390930176,
      "learning_rate": 3.987939210231794e-05,
      "loss": 0.2119,
      "num_input_tokens_seen": 9150256,
      "step": 14020
    },
    {
      "epoch": 7.35062893081761,
      "grad_norm": 2.1152143478393555,
      "learning_rate": 3.987020200515941e-05,
      "loss": 0.2006,
      "num_input_tokens_seen": 9153552,
      "step": 14025
    },
    {
      "epoch": 7.353249475890985,
      "grad_norm": 1.796085238456726,
      "learning_rate": 3.9861008797324255e-05,
      "loss": 0.1808,
      "num_input_tokens_seen": 9156624,
      "step": 14030
    },
    {
      "epoch": 7.35587002096436,
      "grad_norm": 13.842198371887207,
      "learning_rate": 3.9851812480735605e-05,
      "loss": 0.1853,
      "num_input_tokens_seen": 9159792,
      "step": 14035
    },
    {
      "epoch": 7.3584905660377355,
      "grad_norm": 2.437988519668579,
      "learning_rate": 3.9842613057317235e-05,
      "loss": 0.2425,
      "num_input_tokens_seen": 9162960,
      "step": 14040
    },
    {
      "epoch": 7.361111111111111,
      "grad_norm": 2.444485664367676,
      "learning_rate": 3.9833410528993534e-05,
      "loss": 0.2472,
      "num_input_tokens_seen": 9166416,
      "step": 14045
    },
    {
      "epoch": 7.363731656184487,
      "grad_norm": 2.9118123054504395,
      "learning_rate": 3.9824204897689585e-05,
      "loss": 0.2688,
      "num_input_tokens_seen": 9169712,
      "step": 14050
    },
    {
      "epoch": 7.366352201257862,
      "grad_norm": 2.169670581817627,
      "learning_rate": 3.98149961653311e-05,
      "loss": 0.2397,
      "num_input_tokens_seen": 9172208,
      "step": 14055
    },
    {
      "epoch": 7.368972746331237,
      "grad_norm": 1.7796794176101685,
      "learning_rate": 3.980578433384442e-05,
      "loss": 0.2082,
      "num_input_tokens_seen": 9174640,
      "step": 14060
    },
    {
      "epoch": 7.371593291404612,
      "grad_norm": 2.7992019653320312,
      "learning_rate": 3.979656940515659e-05,
      "loss": 0.2372,
      "num_input_tokens_seen": 9178416,
      "step": 14065
    },
    {
      "epoch": 7.3742138364779874,
      "grad_norm": 4.276182174682617,
      "learning_rate": 3.9787351381195235e-05,
      "loss": 0.2069,
      "num_input_tokens_seen": 9181040,
      "step": 14070
    },
    {
      "epoch": 7.376834381551363,
      "grad_norm": 1.6276941299438477,
      "learning_rate": 3.977813026388868e-05,
      "loss": 0.2381,
      "num_input_tokens_seen": 9184048,
      "step": 14075
    },
    {
      "epoch": 7.379454926624738,
      "grad_norm": 1.3604638576507568,
      "learning_rate": 3.976890605516587e-05,
      "loss": 0.1497,
      "num_input_tokens_seen": 9187344,
      "step": 14080
    },
    {
      "epoch": 7.382075471698113,
      "grad_norm": 1.4200822114944458,
      "learning_rate": 3.97596787569564e-05,
      "loss": 0.1745,
      "num_input_tokens_seen": 9190608,
      "step": 14085
    },
    {
      "epoch": 7.384696016771488,
      "grad_norm": 2.159576177597046,
      "learning_rate": 3.975044837119053e-05,
      "loss": 0.1954,
      "num_input_tokens_seen": 9194160,
      "step": 14090
    },
    {
      "epoch": 7.387316561844864,
      "grad_norm": 3.758052349090576,
      "learning_rate": 3.974121489979914e-05,
      "loss": 0.196,
      "num_input_tokens_seen": 9196688,
      "step": 14095
    },
    {
      "epoch": 7.389937106918239,
      "grad_norm": 1.2957630157470703,
      "learning_rate": 3.973197834471377e-05,
      "loss": 0.248,
      "num_input_tokens_seen": 9200880,
      "step": 14100
    },
    {
      "epoch": 7.3925576519916145,
      "grad_norm": 4.429099082946777,
      "learning_rate": 3.9722738707866595e-05,
      "loss": 0.1883,
      "num_input_tokens_seen": 9204720,
      "step": 14105
    },
    {
      "epoch": 7.39517819706499,
      "grad_norm": 2.5413148403167725,
      "learning_rate": 3.971349599119046e-05,
      "loss": 0.1908,
      "num_input_tokens_seen": 9207184,
      "step": 14110
    },
    {
      "epoch": 7.397798742138365,
      "grad_norm": 1.5990183353424072,
      "learning_rate": 3.970425019661881e-05,
      "loss": 0.2178,
      "num_input_tokens_seen": 9210224,
      "step": 14115
    },
    {
      "epoch": 7.40041928721174,
      "grad_norm": 1.9078510999679565,
      "learning_rate": 3.9695001326085785e-05,
      "loss": 0.2193,
      "num_input_tokens_seen": 9213776,
      "step": 14120
    },
    {
      "epoch": 7.403039832285115,
      "grad_norm": 2.1381630897521973,
      "learning_rate": 3.968574938152613e-05,
      "loss": 0.2524,
      "num_input_tokens_seen": 9216368,
      "step": 14125
    },
    {
      "epoch": 7.40566037735849,
      "grad_norm": 2.47835373878479,
      "learning_rate": 3.9676494364875247e-05,
      "loss": 0.4009,
      "num_input_tokens_seen": 9220112,
      "step": 14130
    },
    {
      "epoch": 7.4082809224318655,
      "grad_norm": 2.322624921798706,
      "learning_rate": 3.966723627806918e-05,
      "loss": 0.2398,
      "num_input_tokens_seen": 9223120,
      "step": 14135
    },
    {
      "epoch": 7.410901467505241,
      "grad_norm": 3.3963851928710938,
      "learning_rate": 3.9657975123044624e-05,
      "loss": 0.2903,
      "num_input_tokens_seen": 9225744,
      "step": 14140
    },
    {
      "epoch": 7.413522012578617,
      "grad_norm": 1.8198227882385254,
      "learning_rate": 3.964871090173888e-05,
      "loss": 0.2023,
      "num_input_tokens_seen": 9229200,
      "step": 14145
    },
    {
      "epoch": 7.416142557651992,
      "grad_norm": 0.8675702810287476,
      "learning_rate": 3.963944361608995e-05,
      "loss": 0.2391,
      "num_input_tokens_seen": 9232752,
      "step": 14150
    },
    {
      "epoch": 7.418763102725367,
      "grad_norm": 2.026615619659424,
      "learning_rate": 3.963017326803644e-05,
      "loss": 0.2275,
      "num_input_tokens_seen": 9238704,
      "step": 14155
    },
    {
      "epoch": 7.421383647798742,
      "grad_norm": 2.064737319946289,
      "learning_rate": 3.962089985951758e-05,
      "loss": 0.4086,
      "num_input_tokens_seen": 9241648,
      "step": 14160
    },
    {
      "epoch": 7.424004192872117,
      "grad_norm": 2.450679063796997,
      "learning_rate": 3.961162339247327e-05,
      "loss": 0.2496,
      "num_input_tokens_seen": 9244944,
      "step": 14165
    },
    {
      "epoch": 7.426624737945493,
      "grad_norm": 2.8348300457000732,
      "learning_rate": 3.9602343868844055e-05,
      "loss": 0.2052,
      "num_input_tokens_seen": 9248144,
      "step": 14170
    },
    {
      "epoch": 7.429245283018868,
      "grad_norm": 1.2480099201202393,
      "learning_rate": 3.959306129057109e-05,
      "loss": 0.3001,
      "num_input_tokens_seen": 9251792,
      "step": 14175
    },
    {
      "epoch": 7.431865828092243,
      "grad_norm": 1.1763139963150024,
      "learning_rate": 3.958377565959619e-05,
      "loss": 0.3159,
      "num_input_tokens_seen": 9255248,
      "step": 14180
    },
    {
      "epoch": 7.434486373165618,
      "grad_norm": 1.4461638927459717,
      "learning_rate": 3.95744869778618e-05,
      "loss": 0.2698,
      "num_input_tokens_seen": 9258096,
      "step": 14185
    },
    {
      "epoch": 7.437106918238994,
      "grad_norm": 2.4268603324890137,
      "learning_rate": 3.956519524731101e-05,
      "loss": 0.2086,
      "num_input_tokens_seen": 9260816,
      "step": 14190
    },
    {
      "epoch": 7.439727463312369,
      "grad_norm": 2.6065731048583984,
      "learning_rate": 3.9555900469887555e-05,
      "loss": 0.2056,
      "num_input_tokens_seen": 9263472,
      "step": 14195
    },
    {
      "epoch": 7.4423480083857445,
      "grad_norm": 4.8350443840026855,
      "learning_rate": 3.954660264753577e-05,
      "loss": 0.2059,
      "num_input_tokens_seen": 9266832,
      "step": 14200
    },
    {
      "epoch": 7.44496855345912,
      "grad_norm": 3.442129373550415,
      "learning_rate": 3.953730178220067e-05,
      "loss": 0.2557,
      "num_input_tokens_seen": 9270512,
      "step": 14205
    },
    {
      "epoch": 7.447589098532495,
      "grad_norm": 2.0859920978546143,
      "learning_rate": 3.952799787582789e-05,
      "loss": 0.2074,
      "num_input_tokens_seen": 9274032,
      "step": 14210
    },
    {
      "epoch": 7.45020964360587,
      "grad_norm": 1.7156943082809448,
      "learning_rate": 3.9518690930363694e-05,
      "loss": 0.2262,
      "num_input_tokens_seen": 9276720,
      "step": 14215
    },
    {
      "epoch": 7.452830188679245,
      "grad_norm": 2.170063018798828,
      "learning_rate": 3.9509380947754994e-05,
      "loss": 0.1774,
      "num_input_tokens_seen": 9279504,
      "step": 14220
    },
    {
      "epoch": 7.45545073375262,
      "grad_norm": 2.6883609294891357,
      "learning_rate": 3.9500067929949336e-05,
      "loss": 0.2081,
      "num_input_tokens_seen": 9282864,
      "step": 14225
    },
    {
      "epoch": 7.4580712788259955,
      "grad_norm": 1.4027546644210815,
      "learning_rate": 3.9490751878894895e-05,
      "loss": 0.3047,
      "num_input_tokens_seen": 9285520,
      "step": 14230
    },
    {
      "epoch": 7.460691823899371,
      "grad_norm": 1.1647149324417114,
      "learning_rate": 3.948143279654047e-05,
      "loss": 0.1895,
      "num_input_tokens_seen": 9289840,
      "step": 14235
    },
    {
      "epoch": 7.463312368972747,
      "grad_norm": 2.1960554122924805,
      "learning_rate": 3.947211068483553e-05,
      "loss": 0.2058,
      "num_input_tokens_seen": 9292016,
      "step": 14240
    },
    {
      "epoch": 7.465932914046122,
      "grad_norm": 1.344278335571289,
      "learning_rate": 3.946278554573013e-05,
      "loss": 0.1956,
      "num_input_tokens_seen": 9295312,
      "step": 14245
    },
    {
      "epoch": 7.468553459119497,
      "grad_norm": 4.611881732940674,
      "learning_rate": 3.9453457381175e-05,
      "loss": 0.3387,
      "num_input_tokens_seen": 9298032,
      "step": 14250
    },
    {
      "epoch": 7.471174004192872,
      "grad_norm": 2.9395461082458496,
      "learning_rate": 3.9444126193121474e-05,
      "loss": 0.2814,
      "num_input_tokens_seen": 9301552,
      "step": 14255
    },
    {
      "epoch": 7.473794549266247,
      "grad_norm": 2.375682830810547,
      "learning_rate": 3.943479198352153e-05,
      "loss": 0.1587,
      "num_input_tokens_seen": 9304688,
      "step": 14260
    },
    {
      "epoch": 7.476415094339623,
      "grad_norm": 1.6752266883850098,
      "learning_rate": 3.94254547543278e-05,
      "loss": 0.315,
      "num_input_tokens_seen": 9307664,
      "step": 14265
    },
    {
      "epoch": 7.479035639412998,
      "grad_norm": 2.4198222160339355,
      "learning_rate": 3.9416114507493504e-05,
      "loss": 0.2033,
      "num_input_tokens_seen": 9310736,
      "step": 14270
    },
    {
      "epoch": 7.481656184486373,
      "grad_norm": 2.433805465698242,
      "learning_rate": 3.940677124497252e-05,
      "loss": 0.233,
      "num_input_tokens_seen": 9315056,
      "step": 14275
    },
    {
      "epoch": 7.484276729559748,
      "grad_norm": 1.4351667165756226,
      "learning_rate": 3.939742496871935e-05,
      "loss": 0.1723,
      "num_input_tokens_seen": 9318672,
      "step": 14280
    },
    {
      "epoch": 7.486897274633124,
      "grad_norm": 2.1876988410949707,
      "learning_rate": 3.9388075680689136e-05,
      "loss": 0.2755,
      "num_input_tokens_seen": 9325040,
      "step": 14285
    },
    {
      "epoch": 7.489517819706499,
      "grad_norm": 1.8818638324737549,
      "learning_rate": 3.9378723382837634e-05,
      "loss": 0.1699,
      "num_input_tokens_seen": 9328496,
      "step": 14290
    },
    {
      "epoch": 7.4921383647798745,
      "grad_norm": 1.4839558601379395,
      "learning_rate": 3.936936807712124e-05,
      "loss": 0.2134,
      "num_input_tokens_seen": 9331824,
      "step": 14295
    },
    {
      "epoch": 7.49475890985325,
      "grad_norm": 2.2293365001678467,
      "learning_rate": 3.936000976549698e-05,
      "loss": 0.1515,
      "num_input_tokens_seen": 9334768,
      "step": 14300
    },
    {
      "epoch": 7.497379454926625,
      "grad_norm": 1.8333015441894531,
      "learning_rate": 3.935064844992251e-05,
      "loss": 0.2292,
      "num_input_tokens_seen": 9337616,
      "step": 14305
    },
    {
      "epoch": 7.5,
      "grad_norm": 1.6944479942321777,
      "learning_rate": 3.93412841323561e-05,
      "loss": 0.239,
      "num_input_tokens_seen": 9340304,
      "step": 14310
    },
    {
      "epoch": 7.502620545073375,
      "grad_norm": 3.1797852516174316,
      "learning_rate": 3.933191681475666e-05,
      "loss": 0.2804,
      "num_input_tokens_seen": 9343440,
      "step": 14315
    },
    {
      "epoch": 7.50524109014675,
      "grad_norm": 1.974854826927185,
      "learning_rate": 3.932254649908373e-05,
      "loss": 0.2267,
      "num_input_tokens_seen": 9346864,
      "step": 14320
    },
    {
      "epoch": 7.5078616352201255,
      "grad_norm": 2.3409018516540527,
      "learning_rate": 3.931317318729748e-05,
      "loss": 0.2691,
      "num_input_tokens_seen": 9350384,
      "step": 14325
    },
    {
      "epoch": 7.510482180293501,
      "grad_norm": 2.4580140113830566,
      "learning_rate": 3.930379688135869e-05,
      "loss": 0.2098,
      "num_input_tokens_seen": 9354256,
      "step": 14330
    },
    {
      "epoch": 7.513102725366876,
      "grad_norm": 1.6929426193237305,
      "learning_rate": 3.929441758322878e-05,
      "loss": 0.1622,
      "num_input_tokens_seen": 9357456,
      "step": 14335
    },
    {
      "epoch": 7.515723270440252,
      "grad_norm": 1.6775166988372803,
      "learning_rate": 3.928503529486979e-05,
      "loss": 0.1569,
      "num_input_tokens_seen": 9360560,
      "step": 14340
    },
    {
      "epoch": 7.518343815513627,
      "grad_norm": 3.628351926803589,
      "learning_rate": 3.927565001824438e-05,
      "loss": 0.2137,
      "num_input_tokens_seen": 9363248,
      "step": 14345
    },
    {
      "epoch": 7.520964360587002,
      "grad_norm": 1.1954432725906372,
      "learning_rate": 3.926626175531586e-05,
      "loss": 0.1874,
      "num_input_tokens_seen": 9367088,
      "step": 14350
    },
    {
      "epoch": 7.523584905660377,
      "grad_norm": 1.1947107315063477,
      "learning_rate": 3.925687050804814e-05,
      "loss": 0.2231,
      "num_input_tokens_seen": 9370960,
      "step": 14355
    },
    {
      "epoch": 7.526205450733753,
      "grad_norm": 1.5075372457504272,
      "learning_rate": 3.924747627840575e-05,
      "loss": 0.2166,
      "num_input_tokens_seen": 9373936,
      "step": 14360
    },
    {
      "epoch": 7.528825995807128,
      "grad_norm": 1.372430443763733,
      "learning_rate": 3.923807906835388e-05,
      "loss": 0.1684,
      "num_input_tokens_seen": 9377264,
      "step": 14365
    },
    {
      "epoch": 7.531446540880503,
      "grad_norm": 1.8612171411514282,
      "learning_rate": 3.92286788798583e-05,
      "loss": 0.2225,
      "num_input_tokens_seen": 9380240,
      "step": 14370
    },
    {
      "epoch": 7.534067085953878,
      "grad_norm": 2.1945431232452393,
      "learning_rate": 3.921927571488543e-05,
      "loss": 0.168,
      "num_input_tokens_seen": 9385136,
      "step": 14375
    },
    {
      "epoch": 7.536687631027254,
      "grad_norm": 1.5767545700073242,
      "learning_rate": 3.9209869575402296e-05,
      "loss": 0.2644,
      "num_input_tokens_seen": 9387760,
      "step": 14380
    },
    {
      "epoch": 7.539308176100629,
      "grad_norm": 2.0403635501861572,
      "learning_rate": 3.9200460463376566e-05,
      "loss": 0.2601,
      "num_input_tokens_seen": 9391184,
      "step": 14385
    },
    {
      "epoch": 7.5419287211740045,
      "grad_norm": 1.1121889352798462,
      "learning_rate": 3.9191048380776515e-05,
      "loss": 0.2275,
      "num_input_tokens_seen": 9393776,
      "step": 14390
    },
    {
      "epoch": 7.54454926624738,
      "grad_norm": 1.5569560527801514,
      "learning_rate": 3.918163332957104e-05,
      "loss": 0.2573,
      "num_input_tokens_seen": 9397008,
      "step": 14395
    },
    {
      "epoch": 7.547169811320755,
      "grad_norm": 1.953513264656067,
      "learning_rate": 3.9172215311729665e-05,
      "loss": 0.3261,
      "num_input_tokens_seen": 9400144,
      "step": 14400
    },
    {
      "epoch": 7.54979035639413,
      "grad_norm": 1.9644505977630615,
      "learning_rate": 3.916279432922254e-05,
      "loss": 0.1814,
      "num_input_tokens_seen": 9403728,
      "step": 14405
    },
    {
      "epoch": 7.552410901467505,
      "grad_norm": 4.973997592926025,
      "learning_rate": 3.915337038402041e-05,
      "loss": 0.3178,
      "num_input_tokens_seen": 9408112,
      "step": 14410
    },
    {
      "epoch": 7.55503144654088,
      "grad_norm": 2.289541244506836,
      "learning_rate": 3.914394347809467e-05,
      "loss": 0.2583,
      "num_input_tokens_seen": 9411152,
      "step": 14415
    },
    {
      "epoch": 7.5576519916142555,
      "grad_norm": 1.5340603590011597,
      "learning_rate": 3.9134513613417315e-05,
      "loss": 0.1731,
      "num_input_tokens_seen": 9414416,
      "step": 14420
    },
    {
      "epoch": 7.560272536687631,
      "grad_norm": 1.829518437385559,
      "learning_rate": 3.912508079196097e-05,
      "loss": 0.2143,
      "num_input_tokens_seen": 9417168,
      "step": 14425
    },
    {
      "epoch": 7.562893081761006,
      "grad_norm": 3.5813825130462646,
      "learning_rate": 3.9115645015698866e-05,
      "loss": 0.1774,
      "num_input_tokens_seen": 9420880,
      "step": 14430
    },
    {
      "epoch": 7.565513626834382,
      "grad_norm": 1.1689492464065552,
      "learning_rate": 3.9106206286604865e-05,
      "loss": 0.2087,
      "num_input_tokens_seen": 9424144,
      "step": 14435
    },
    {
      "epoch": 7.568134171907757,
      "grad_norm": 2.6850719451904297,
      "learning_rate": 3.909676460665345e-05,
      "loss": 0.2244,
      "num_input_tokens_seen": 9426448,
      "step": 14440
    },
    {
      "epoch": 7.570754716981132,
      "grad_norm": 2.7344777584075928,
      "learning_rate": 3.908731997781969e-05,
      "loss": 0.2475,
      "num_input_tokens_seen": 9429456,
      "step": 14445
    },
    {
      "epoch": 7.573375262054507,
      "grad_norm": 1.717820644378662,
      "learning_rate": 3.907787240207932e-05,
      "loss": 0.2618,
      "num_input_tokens_seen": 9432720,
      "step": 14450
    },
    {
      "epoch": 7.575995807127883,
      "grad_norm": 2.0837314128875732,
      "learning_rate": 3.9068421881408645e-05,
      "loss": 0.3012,
      "num_input_tokens_seen": 9435344,
      "step": 14455
    },
    {
      "epoch": 7.578616352201258,
      "grad_norm": 1.5172303915023804,
      "learning_rate": 3.9058968417784614e-05,
      "loss": 0.2539,
      "num_input_tokens_seen": 9438416,
      "step": 14460
    },
    {
      "epoch": 7.581236897274633,
      "grad_norm": 1.472683072090149,
      "learning_rate": 3.9049512013184785e-05,
      "loss": 0.2095,
      "num_input_tokens_seen": 9440912,
      "step": 14465
    },
    {
      "epoch": 7.583857442348008,
      "grad_norm": 0.9843655228614807,
      "learning_rate": 3.9040052669587325e-05,
      "loss": 0.1456,
      "num_input_tokens_seen": 9444272,
      "step": 14470
    },
    {
      "epoch": 7.586477987421384,
      "grad_norm": 1.6828161478042603,
      "learning_rate": 3.9030590388971026e-05,
      "loss": 0.3247,
      "num_input_tokens_seen": 9446704,
      "step": 14475
    },
    {
      "epoch": 7.589098532494759,
      "grad_norm": 2.1950273513793945,
      "learning_rate": 3.902112517331529e-05,
      "loss": 0.1628,
      "num_input_tokens_seen": 9450128,
      "step": 14480
    },
    {
      "epoch": 7.5917190775681345,
      "grad_norm": 4.0419769287109375,
      "learning_rate": 3.901165702460012e-05,
      "loss": 0.2768,
      "num_input_tokens_seen": 9453072,
      "step": 14485
    },
    {
      "epoch": 7.59433962264151,
      "grad_norm": 3.166477918624878,
      "learning_rate": 3.900218594480616e-05,
      "loss": 0.2027,
      "num_input_tokens_seen": 9455856,
      "step": 14490
    },
    {
      "epoch": 7.596960167714885,
      "grad_norm": 1.7003026008605957,
      "learning_rate": 3.899271193591465e-05,
      "loss": 0.2801,
      "num_input_tokens_seen": 9458736,
      "step": 14495
    },
    {
      "epoch": 7.59958071278826,
      "grad_norm": 1.8741698265075684,
      "learning_rate": 3.8983234999907435e-05,
      "loss": 0.1883,
      "num_input_tokens_seen": 9461840,
      "step": 14500
    },
    {
      "epoch": 7.602201257861635,
      "grad_norm": 1.3050591945648193,
      "learning_rate": 3.897375513876699e-05,
      "loss": 0.2309,
      "num_input_tokens_seen": 9468368,
      "step": 14505
    },
    {
      "epoch": 7.60482180293501,
      "grad_norm": 2.3612563610076904,
      "learning_rate": 3.8964272354476396e-05,
      "loss": 0.3265,
      "num_input_tokens_seen": 9470928,
      "step": 14510
    },
    {
      "epoch": 7.6074423480083855,
      "grad_norm": 3.4868502616882324,
      "learning_rate": 3.895478664901933e-05,
      "loss": 0.3866,
      "num_input_tokens_seen": 9475024,
      "step": 14515
    },
    {
      "epoch": 7.610062893081761,
      "grad_norm": 1.105027675628662,
      "learning_rate": 3.89452980243801e-05,
      "loss": 0.2898,
      "num_input_tokens_seen": 9479408,
      "step": 14520
    },
    {
      "epoch": 7.612683438155136,
      "grad_norm": 1.7434455156326294,
      "learning_rate": 3.893580648254362e-05,
      "loss": 0.2086,
      "num_input_tokens_seen": 9482480,
      "step": 14525
    },
    {
      "epoch": 7.615303983228512,
      "grad_norm": 1.7417292594909668,
      "learning_rate": 3.8926312025495414e-05,
      "loss": 0.1757,
      "num_input_tokens_seen": 9484816,
      "step": 14530
    },
    {
      "epoch": 7.617924528301887,
      "grad_norm": 2.8507683277130127,
      "learning_rate": 3.891681465522161e-05,
      "loss": 0.1727,
      "num_input_tokens_seen": 9487728,
      "step": 14535
    },
    {
      "epoch": 7.620545073375262,
      "grad_norm": 2.2671964168548584,
      "learning_rate": 3.890731437370896e-05,
      "loss": 0.2983,
      "num_input_tokens_seen": 9490448,
      "step": 14540
    },
    {
      "epoch": 7.623165618448637,
      "grad_norm": 1.6775243282318115,
      "learning_rate": 3.88978111829448e-05,
      "loss": 0.2581,
      "num_input_tokens_seen": 9493360,
      "step": 14545
    },
    {
      "epoch": 7.6257861635220126,
      "grad_norm": 1.3705040216445923,
      "learning_rate": 3.88883050849171e-05,
      "loss": 0.1978,
      "num_input_tokens_seen": 9496560,
      "step": 14550
    },
    {
      "epoch": 7.628406708595388,
      "grad_norm": 5.104020595550537,
      "learning_rate": 3.8878796081614417e-05,
      "loss": 0.1256,
      "num_input_tokens_seen": 9499632,
      "step": 14555
    },
    {
      "epoch": 7.631027253668763,
      "grad_norm": 1.2754817008972168,
      "learning_rate": 3.8869284175025935e-05,
      "loss": 0.2697,
      "num_input_tokens_seen": 9503696,
      "step": 14560
    },
    {
      "epoch": 7.633647798742138,
      "grad_norm": 1.4848930835723877,
      "learning_rate": 3.885976936714143e-05,
      "loss": 0.2167,
      "num_input_tokens_seen": 9506768,
      "step": 14565
    },
    {
      "epoch": 7.636268343815514,
      "grad_norm": 1.9559454917907715,
      "learning_rate": 3.885025165995129e-05,
      "loss": 0.1888,
      "num_input_tokens_seen": 9509488,
      "step": 14570
    },
    {
      "epoch": 7.638888888888889,
      "grad_norm": 1.3186558485031128,
      "learning_rate": 3.884073105544652e-05,
      "loss": 0.2011,
      "num_input_tokens_seen": 9512336,
      "step": 14575
    },
    {
      "epoch": 7.6415094339622645,
      "grad_norm": 2.5971007347106934,
      "learning_rate": 3.8831207555618723e-05,
      "loss": 0.2815,
      "num_input_tokens_seen": 9515376,
      "step": 14580
    },
    {
      "epoch": 7.64412997903564,
      "grad_norm": 2.0331945419311523,
      "learning_rate": 3.882168116246009e-05,
      "loss": 0.1818,
      "num_input_tokens_seen": 9518160,
      "step": 14585
    },
    {
      "epoch": 7.646750524109015,
      "grad_norm": 1.3507928848266602,
      "learning_rate": 3.881215187796344e-05,
      "loss": 0.2402,
      "num_input_tokens_seen": 9520912,
      "step": 14590
    },
    {
      "epoch": 7.64937106918239,
      "grad_norm": 1.836530089378357,
      "learning_rate": 3.8802619704122204e-05,
      "loss": 0.204,
      "num_input_tokens_seen": 9523728,
      "step": 14595
    },
    {
      "epoch": 7.651991614255765,
      "grad_norm": 3.5325801372528076,
      "learning_rate": 3.8793084642930387e-05,
      "loss": 0.3603,
      "num_input_tokens_seen": 9526416,
      "step": 14600
    },
    {
      "epoch": 7.65461215932914,
      "grad_norm": 1.7771668434143066,
      "learning_rate": 3.878354669638263e-05,
      "loss": 0.2508,
      "num_input_tokens_seen": 9529936,
      "step": 14605
    },
    {
      "epoch": 7.6572327044025155,
      "grad_norm": 1.762568712234497,
      "learning_rate": 3.8774005866474155e-05,
      "loss": 0.241,
      "num_input_tokens_seen": 9533232,
      "step": 14610
    },
    {
      "epoch": 7.659853249475891,
      "grad_norm": 1.1224374771118164,
      "learning_rate": 3.876446215520079e-05,
      "loss": 0.1812,
      "num_input_tokens_seen": 9536656,
      "step": 14615
    },
    {
      "epoch": 7.662473794549266,
      "grad_norm": 1.6653369665145874,
      "learning_rate": 3.875491556455898e-05,
      "loss": 0.2787,
      "num_input_tokens_seen": 9539344,
      "step": 14620
    },
    {
      "epoch": 7.665094339622642,
      "grad_norm": 2.4487318992614746,
      "learning_rate": 3.8745366096545754e-05,
      "loss": 0.2341,
      "num_input_tokens_seen": 9542832,
      "step": 14625
    },
    {
      "epoch": 7.667714884696017,
      "grad_norm": 3.5497257709503174,
      "learning_rate": 3.873581375315876e-05,
      "loss": 0.2688,
      "num_input_tokens_seen": 9546640,
      "step": 14630
    },
    {
      "epoch": 7.670335429769392,
      "grad_norm": 1.1627323627471924,
      "learning_rate": 3.872625853639623e-05,
      "loss": 0.154,
      "num_input_tokens_seen": 9550000,
      "step": 14635
    },
    {
      "epoch": 7.672955974842767,
      "grad_norm": 3.8152449131011963,
      "learning_rate": 3.871670044825701e-05,
      "loss": 0.2463,
      "num_input_tokens_seen": 9553424,
      "step": 14640
    },
    {
      "epoch": 7.6755765199161425,
      "grad_norm": 1.488888144493103,
      "learning_rate": 3.870713949074056e-05,
      "loss": 0.2106,
      "num_input_tokens_seen": 9556144,
      "step": 14645
    },
    {
      "epoch": 7.678197064989518,
      "grad_norm": 2.0941104888916016,
      "learning_rate": 3.86975756658469e-05,
      "loss": 0.2248,
      "num_input_tokens_seen": 9559024,
      "step": 14650
    },
    {
      "epoch": 7.680817610062893,
      "grad_norm": 1.4716145992279053,
      "learning_rate": 3.868800897557667e-05,
      "loss": 0.2436,
      "num_input_tokens_seen": 9562640,
      "step": 14655
    },
    {
      "epoch": 7.683438155136268,
      "grad_norm": 1.4696083068847656,
      "learning_rate": 3.867843942193113e-05,
      "loss": 0.1987,
      "num_input_tokens_seen": 9565360,
      "step": 14660
    },
    {
      "epoch": 7.686058700209644,
      "grad_norm": 1.9454989433288574,
      "learning_rate": 3.866886700691212e-05,
      "loss": 0.2483,
      "num_input_tokens_seen": 9568688,
      "step": 14665
    },
    {
      "epoch": 7.688679245283019,
      "grad_norm": 2.699202537536621,
      "learning_rate": 3.8659291732522075e-05,
      "loss": 0.1432,
      "num_input_tokens_seen": 9571568,
      "step": 14670
    },
    {
      "epoch": 7.691299790356394,
      "grad_norm": 3.0131328105926514,
      "learning_rate": 3.8649713600764026e-05,
      "loss": 0.2544,
      "num_input_tokens_seen": 9574128,
      "step": 14675
    },
    {
      "epoch": 7.69392033542977,
      "grad_norm": 2.120142936706543,
      "learning_rate": 3.864013261364162e-05,
      "loss": 0.2518,
      "num_input_tokens_seen": 9577680,
      "step": 14680
    },
    {
      "epoch": 7.696540880503145,
      "grad_norm": 6.804782390594482,
      "learning_rate": 3.8630548773159084e-05,
      "loss": 0.2521,
      "num_input_tokens_seen": 9580656,
      "step": 14685
    },
    {
      "epoch": 7.69916142557652,
      "grad_norm": 5.149459362030029,
      "learning_rate": 3.862096208132125e-05,
      "loss": 0.291,
      "num_input_tokens_seen": 9584496,
      "step": 14690
    },
    {
      "epoch": 7.701781970649895,
      "grad_norm": 2.012380361557007,
      "learning_rate": 3.861137254013355e-05,
      "loss": 0.3121,
      "num_input_tokens_seen": 9591056,
      "step": 14695
    },
    {
      "epoch": 7.70440251572327,
      "grad_norm": 1.881935954093933,
      "learning_rate": 3.8601780151601996e-05,
      "loss": 0.1903,
      "num_input_tokens_seen": 9594320,
      "step": 14700
    },
    {
      "epoch": 7.7070230607966455,
      "grad_norm": 1.3542720079421997,
      "learning_rate": 3.859218491773321e-05,
      "loss": 0.2172,
      "num_input_tokens_seen": 9597616,
      "step": 14705
    },
    {
      "epoch": 7.709643605870021,
      "grad_norm": 1.8993231058120728,
      "learning_rate": 3.858258684053441e-05,
      "loss": 0.28,
      "num_input_tokens_seen": 9601072,
      "step": 14710
    },
    {
      "epoch": 7.712264150943396,
      "grad_norm": 1.7361910343170166,
      "learning_rate": 3.8572985922013405e-05,
      "loss": 0.2302,
      "num_input_tokens_seen": 9604752,
      "step": 14715
    },
    {
      "epoch": 7.714884696016772,
      "grad_norm": 1.5132588148117065,
      "learning_rate": 3.856338216417859e-05,
      "loss": 0.1896,
      "num_input_tokens_seen": 9607600,
      "step": 14720
    },
    {
      "epoch": 7.717505241090147,
      "grad_norm": 1.7055997848510742,
      "learning_rate": 3.855377556903897e-05,
      "loss": 0.282,
      "num_input_tokens_seen": 9611184,
      "step": 14725
    },
    {
      "epoch": 7.720125786163522,
      "grad_norm": 1.710167646408081,
      "learning_rate": 3.8544166138604124e-05,
      "loss": 0.2564,
      "num_input_tokens_seen": 9614288,
      "step": 14730
    },
    {
      "epoch": 7.722746331236897,
      "grad_norm": 2.136579990386963,
      "learning_rate": 3.8534553874884244e-05,
      "loss": 0.2548,
      "num_input_tokens_seen": 9617680,
      "step": 14735
    },
    {
      "epoch": 7.7253668763102725,
      "grad_norm": 1.5407735109329224,
      "learning_rate": 3.852493877989011e-05,
      "loss": 0.1788,
      "num_input_tokens_seen": 9620944,
      "step": 14740
    },
    {
      "epoch": 7.727987421383648,
      "grad_norm": 2.4433939456939697,
      "learning_rate": 3.851532085563308e-05,
      "loss": 0.2425,
      "num_input_tokens_seen": 9627536,
      "step": 14745
    },
    {
      "epoch": 7.730607966457023,
      "grad_norm": 1.4518901109695435,
      "learning_rate": 3.850570010412511e-05,
      "loss": 0.2765,
      "num_input_tokens_seen": 9631216,
      "step": 14750
    },
    {
      "epoch": 7.733228511530398,
      "grad_norm": 2.0824790000915527,
      "learning_rate": 3.8496076527378766e-05,
      "loss": 0.1698,
      "num_input_tokens_seen": 9634160,
      "step": 14755
    },
    {
      "epoch": 7.735849056603773,
      "grad_norm": 1.7227952480316162,
      "learning_rate": 3.848645012740718e-05,
      "loss": 0.21,
      "num_input_tokens_seen": 9637456,
      "step": 14760
    },
    {
      "epoch": 7.738469601677149,
      "grad_norm": 2.031100273132324,
      "learning_rate": 3.8476820906224096e-05,
      "loss": 0.2307,
      "num_input_tokens_seen": 9641520,
      "step": 14765
    },
    {
      "epoch": 7.741090146750524,
      "grad_norm": 1.2423291206359863,
      "learning_rate": 3.846718886584382e-05,
      "loss": 0.2458,
      "num_input_tokens_seen": 9644912,
      "step": 14770
    },
    {
      "epoch": 7.7437106918239,
      "grad_norm": 4.876009941101074,
      "learning_rate": 3.845755400828128e-05,
      "loss": 0.1935,
      "num_input_tokens_seen": 9647568,
      "step": 14775
    },
    {
      "epoch": 7.746331236897275,
      "grad_norm": 1.4746315479278564,
      "learning_rate": 3.844791633555198e-05,
      "loss": 0.1934,
      "num_input_tokens_seen": 9651600,
      "step": 14780
    },
    {
      "epoch": 7.74895178197065,
      "grad_norm": 2.249300003051758,
      "learning_rate": 3.8438275849671996e-05,
      "loss": 0.1767,
      "num_input_tokens_seen": 9654576,
      "step": 14785
    },
    {
      "epoch": 7.751572327044025,
      "grad_norm": 4.117695331573486,
      "learning_rate": 3.842863255265802e-05,
      "loss": 0.2203,
      "num_input_tokens_seen": 9658000,
      "step": 14790
    },
    {
      "epoch": 7.7541928721174,
      "grad_norm": 1.0779619216918945,
      "learning_rate": 3.841898644652732e-05,
      "loss": 0.2154,
      "num_input_tokens_seen": 9660880,
      "step": 14795
    },
    {
      "epoch": 7.756813417190775,
      "grad_norm": 1.408279538154602,
      "learning_rate": 3.840933753329774e-05,
      "loss": 0.1505,
      "num_input_tokens_seen": 9663920,
      "step": 14800
    },
    {
      "epoch": 7.759433962264151,
      "grad_norm": 3.080562114715576,
      "learning_rate": 3.839968581498773e-05,
      "loss": 0.2229,
      "num_input_tokens_seen": 9667792,
      "step": 14805
    },
    {
      "epoch": 7.762054507337526,
      "grad_norm": 2.090761423110962,
      "learning_rate": 3.839003129361632e-05,
      "loss": 0.2125,
      "num_input_tokens_seen": 9671120,
      "step": 14810
    },
    {
      "epoch": 7.764675052410902,
      "grad_norm": 1.5333328247070312,
      "learning_rate": 3.8380373971203134e-05,
      "loss": 0.166,
      "num_input_tokens_seen": 9674608,
      "step": 14815
    },
    {
      "epoch": 7.767295597484277,
      "grad_norm": 1.9252032041549683,
      "learning_rate": 3.8370713849768346e-05,
      "loss": 0.1833,
      "num_input_tokens_seen": 9677904,
      "step": 14820
    },
    {
      "epoch": 7.769916142557652,
      "grad_norm": 2.2578186988830566,
      "learning_rate": 3.8361050931332774e-05,
      "loss": 0.2564,
      "num_input_tokens_seen": 9681648,
      "step": 14825
    },
    {
      "epoch": 7.772536687631027,
      "grad_norm": 2.8994033336639404,
      "learning_rate": 3.835138521791778e-05,
      "loss": 0.2641,
      "num_input_tokens_seen": 9684208,
      "step": 14830
    },
    {
      "epoch": 7.7751572327044025,
      "grad_norm": 2.352085828781128,
      "learning_rate": 3.8341716711545314e-05,
      "loss": 0.2744,
      "num_input_tokens_seen": 9686800,
      "step": 14835
    },
    {
      "epoch": 7.777777777777778,
      "grad_norm": 1.756245493888855,
      "learning_rate": 3.833204541423793e-05,
      "loss": 0.2078,
      "num_input_tokens_seen": 9689424,
      "step": 14840
    },
    {
      "epoch": 7.780398322851153,
      "grad_norm": 1.6731367111206055,
      "learning_rate": 3.832237132801874e-05,
      "loss": 0.1969,
      "num_input_tokens_seen": 9692368,
      "step": 14845
    },
    {
      "epoch": 7.783018867924528,
      "grad_norm": 1.1076081991195679,
      "learning_rate": 3.831269445491146e-05,
      "loss": 0.1614,
      "num_input_tokens_seen": 9695856,
      "step": 14850
    },
    {
      "epoch": 7.785639412997903,
      "grad_norm": 1.378661870956421,
      "learning_rate": 3.8303014796940375e-05,
      "loss": 0.2016,
      "num_input_tokens_seen": 9699280,
      "step": 14855
    },
    {
      "epoch": 7.788259958071279,
      "grad_norm": 1.8571550846099854,
      "learning_rate": 3.829333235613036e-05,
      "loss": 0.2881,
      "num_input_tokens_seen": 9701840,
      "step": 14860
    },
    {
      "epoch": 7.790880503144654,
      "grad_norm": 2.4774746894836426,
      "learning_rate": 3.828364713450688e-05,
      "loss": 0.2853,
      "num_input_tokens_seen": 9704400,
      "step": 14865
    },
    {
      "epoch": 7.79350104821803,
      "grad_norm": 1.5231379270553589,
      "learning_rate": 3.827395913409597e-05,
      "loss": 0.2185,
      "num_input_tokens_seen": 9707664,
      "step": 14870
    },
    {
      "epoch": 7.796121593291405,
      "grad_norm": 1.7686418294906616,
      "learning_rate": 3.826426835692424e-05,
      "loss": 0.2611,
      "num_input_tokens_seen": 9710672,
      "step": 14875
    },
    {
      "epoch": 7.79874213836478,
      "grad_norm": 1.6275877952575684,
      "learning_rate": 3.825457480501891e-05,
      "loss": 0.248,
      "num_input_tokens_seen": 9713872,
      "step": 14880
    },
    {
      "epoch": 7.801362683438155,
      "grad_norm": 1.5106837749481201,
      "learning_rate": 3.824487848040773e-05,
      "loss": 0.1783,
      "num_input_tokens_seen": 9717008,
      "step": 14885
    },
    {
      "epoch": 7.80398322851153,
      "grad_norm": 3.271724224090576,
      "learning_rate": 3.823517938511908e-05,
      "loss": 0.3299,
      "num_input_tokens_seen": 9720560,
      "step": 14890
    },
    {
      "epoch": 7.806603773584905,
      "grad_norm": 0.982873260974884,
      "learning_rate": 3.8225477521181906e-05,
      "loss": 0.1747,
      "num_input_tokens_seen": 9723792,
      "step": 14895
    },
    {
      "epoch": 7.809224318658281,
      "grad_norm": 1.8118162155151367,
      "learning_rate": 3.821577289062571e-05,
      "loss": 0.1881,
      "num_input_tokens_seen": 9727184,
      "step": 14900
    },
    {
      "epoch": 7.811844863731656,
      "grad_norm": 1.934013843536377,
      "learning_rate": 3.82060654954806e-05,
      "loss": 0.1856,
      "num_input_tokens_seen": 9730832,
      "step": 14905
    },
    {
      "epoch": 7.814465408805032,
      "grad_norm": 2.0431387424468994,
      "learning_rate": 3.819635533777726e-05,
      "loss": 0.253,
      "num_input_tokens_seen": 9734640,
      "step": 14910
    },
    {
      "epoch": 7.817085953878407,
      "grad_norm": 1.8347101211547852,
      "learning_rate": 3.8186642419546924e-05,
      "loss": 0.1505,
      "num_input_tokens_seen": 9737520,
      "step": 14915
    },
    {
      "epoch": 7.819706498951782,
      "grad_norm": 2.281080722808838,
      "learning_rate": 3.817692674282145e-05,
      "loss": 0.1722,
      "num_input_tokens_seen": 9740880,
      "step": 14920
    },
    {
      "epoch": 7.822327044025157,
      "grad_norm": 1.5864211320877075,
      "learning_rate": 3.816720830963322e-05,
      "loss": 0.161,
      "num_input_tokens_seen": 9744080,
      "step": 14925
    },
    {
      "epoch": 7.8249475890985325,
      "grad_norm": 2.435837507247925,
      "learning_rate": 3.815748712201523e-05,
      "loss": 0.2152,
      "num_input_tokens_seen": 9747184,
      "step": 14930
    },
    {
      "epoch": 7.827568134171908,
      "grad_norm": 2.3130640983581543,
      "learning_rate": 3.814776318200105e-05,
      "loss": 0.2952,
      "num_input_tokens_seen": 9749648,
      "step": 14935
    },
    {
      "epoch": 7.830188679245283,
      "grad_norm": 2.216416597366333,
      "learning_rate": 3.8138036491624805e-05,
      "loss": 0.2671,
      "num_input_tokens_seen": 9751984,
      "step": 14940
    },
    {
      "epoch": 7.832809224318658,
      "grad_norm": 1.7605985403060913,
      "learning_rate": 3.812830705292122e-05,
      "loss": 0.1822,
      "num_input_tokens_seen": 9755152,
      "step": 14945
    },
    {
      "epoch": 7.835429769392033,
      "grad_norm": 1.4788818359375,
      "learning_rate": 3.8118574867925564e-05,
      "loss": 0.2281,
      "num_input_tokens_seen": 9758448,
      "step": 14950
    },
    {
      "epoch": 7.838050314465409,
      "grad_norm": 1.3942692279815674,
      "learning_rate": 3.810883993867371e-05,
      "loss": 0.1847,
      "num_input_tokens_seen": 9761264,
      "step": 14955
    },
    {
      "epoch": 7.840670859538784,
      "grad_norm": 2.830152988433838,
      "learning_rate": 3.80991022672021e-05,
      "loss": 0.2891,
      "num_input_tokens_seen": 9764976,
      "step": 14960
    },
    {
      "epoch": 7.84329140461216,
      "grad_norm": 1.9600446224212646,
      "learning_rate": 3.8089361855547744e-05,
      "loss": 0.2842,
      "num_input_tokens_seen": 9768208,
      "step": 14965
    },
    {
      "epoch": 7.845911949685535,
      "grad_norm": 1.5528854131698608,
      "learning_rate": 3.8079618705748214e-05,
      "loss": 0.1566,
      "num_input_tokens_seen": 9771248,
      "step": 14970
    },
    {
      "epoch": 7.84853249475891,
      "grad_norm": 2.5944056510925293,
      "learning_rate": 3.806987281984167e-05,
      "loss": 0.2862,
      "num_input_tokens_seen": 9773776,
      "step": 14975
    },
    {
      "epoch": 7.851153039832285,
      "grad_norm": 1.933088779449463,
      "learning_rate": 3.806012419986684e-05,
      "loss": 0.2482,
      "num_input_tokens_seen": 9777360,
      "step": 14980
    },
    {
      "epoch": 7.85377358490566,
      "grad_norm": 1.8827619552612305,
      "learning_rate": 3.805037284786303e-05,
      "loss": 0.2176,
      "num_input_tokens_seen": 9779984,
      "step": 14985
    },
    {
      "epoch": 7.856394129979035,
      "grad_norm": 2.7383642196655273,
      "learning_rate": 3.804061876587009e-05,
      "loss": 0.2465,
      "num_input_tokens_seen": 9782480,
      "step": 14990
    },
    {
      "epoch": 7.859014675052411,
      "grad_norm": 3.105362892150879,
      "learning_rate": 3.80308619559285e-05,
      "loss": 0.3091,
      "num_input_tokens_seen": 9786608,
      "step": 14995
    },
    {
      "epoch": 7.861635220125786,
      "grad_norm": 1.871299386024475,
      "learning_rate": 3.802110242007925e-05,
      "loss": 0.2547,
      "num_input_tokens_seen": 9790096,
      "step": 15000
    },
    {
      "epoch": 7.864255765199162,
      "grad_norm": 2.0892629623413086,
      "learning_rate": 3.801134016036392e-05,
      "loss": 0.2769,
      "num_input_tokens_seen": 9793072,
      "step": 15005
    },
    {
      "epoch": 7.866876310272537,
      "grad_norm": 2.286400079727173,
      "learning_rate": 3.8001575178824665e-05,
      "loss": 0.1682,
      "num_input_tokens_seen": 9796528,
      "step": 15010
    },
    {
      "epoch": 7.869496855345912,
      "grad_norm": 3.790745496749878,
      "learning_rate": 3.799180747750422e-05,
      "loss": 0.2525,
      "num_input_tokens_seen": 9800048,
      "step": 15015
    },
    {
      "epoch": 7.872117400419287,
      "grad_norm": 1.595228910446167,
      "learning_rate": 3.798203705844586e-05,
      "loss": 0.2113,
      "num_input_tokens_seen": 9802672,
      "step": 15020
    },
    {
      "epoch": 7.8747379454926625,
      "grad_norm": 1.9611806869506836,
      "learning_rate": 3.7972263923693476e-05,
      "loss": 0.3111,
      "num_input_tokens_seen": 9806352,
      "step": 15025
    },
    {
      "epoch": 7.877358490566038,
      "grad_norm": 3.5817980766296387,
      "learning_rate": 3.796248807529145e-05,
      "loss": 0.177,
      "num_input_tokens_seen": 9809200,
      "step": 15030
    },
    {
      "epoch": 7.879979035639413,
      "grad_norm": 3.847219944000244,
      "learning_rate": 3.795270951528482e-05,
      "loss": 0.32,
      "num_input_tokens_seen": 9811568,
      "step": 15035
    },
    {
      "epoch": 7.882599580712788,
      "grad_norm": 2.147531032562256,
      "learning_rate": 3.7942928245719124e-05,
      "loss": 0.283,
      "num_input_tokens_seen": 9814480,
      "step": 15040
    },
    {
      "epoch": 7.885220125786163,
      "grad_norm": 1.6564116477966309,
      "learning_rate": 3.7933144268640495e-05,
      "loss": 0.1953,
      "num_input_tokens_seen": 9817200,
      "step": 15045
    },
    {
      "epoch": 7.887840670859539,
      "grad_norm": 2.847020149230957,
      "learning_rate": 3.7923357586095645e-05,
      "loss": 0.1836,
      "num_input_tokens_seen": 9820336,
      "step": 15050
    },
    {
      "epoch": 7.890461215932914,
      "grad_norm": 2.306204080581665,
      "learning_rate": 3.7913568200131823e-05,
      "loss": 0.2603,
      "num_input_tokens_seen": 9823664,
      "step": 15055
    },
    {
      "epoch": 7.8930817610062896,
      "grad_norm": 1.719594955444336,
      "learning_rate": 3.790377611279685e-05,
      "loss": 0.2803,
      "num_input_tokens_seen": 9827280,
      "step": 15060
    },
    {
      "epoch": 7.895702306079665,
      "grad_norm": 5.150516510009766,
      "learning_rate": 3.789398132613914e-05,
      "loss": 0.2238,
      "num_input_tokens_seen": 9830192,
      "step": 15065
    },
    {
      "epoch": 7.89832285115304,
      "grad_norm": 4.0669097900390625,
      "learning_rate": 3.788418384220763e-05,
      "loss": 0.2421,
      "num_input_tokens_seen": 9832816,
      "step": 15070
    },
    {
      "epoch": 7.900943396226415,
      "grad_norm": 1.7152570486068726,
      "learning_rate": 3.7874383663051846e-05,
      "loss": 0.1388,
      "num_input_tokens_seen": 9836080,
      "step": 15075
    },
    {
      "epoch": 7.90356394129979,
      "grad_norm": 2.202601909637451,
      "learning_rate": 3.7864580790721894e-05,
      "loss": 0.2123,
      "num_input_tokens_seen": 9839152,
      "step": 15080
    },
    {
      "epoch": 7.906184486373165,
      "grad_norm": 1.5032389163970947,
      "learning_rate": 3.785477522726839e-05,
      "loss": 0.201,
      "num_input_tokens_seen": 9843120,
      "step": 15085
    },
    {
      "epoch": 7.908805031446541,
      "grad_norm": 2.3780906200408936,
      "learning_rate": 3.784496697474258e-05,
      "loss": 0.2375,
      "num_input_tokens_seen": 9846608,
      "step": 15090
    },
    {
      "epoch": 7.911425576519916,
      "grad_norm": 13.0897798538208,
      "learning_rate": 3.783515603519622e-05,
      "loss": 0.2227,
      "num_input_tokens_seen": 9849552,
      "step": 15095
    },
    {
      "epoch": 7.914046121593291,
      "grad_norm": 2.620283842086792,
      "learning_rate": 3.782534241068164e-05,
      "loss": 0.2347,
      "num_input_tokens_seen": 9852336,
      "step": 15100
    },
    {
      "epoch": 7.916666666666667,
      "grad_norm": 1.1230719089508057,
      "learning_rate": 3.781552610325175e-05,
      "loss": 0.2313,
      "num_input_tokens_seen": 9855376,
      "step": 15105
    },
    {
      "epoch": 7.919287211740042,
      "grad_norm": 1.429038643836975,
      "learning_rate": 3.780570711496001e-05,
      "loss": 0.2923,
      "num_input_tokens_seen": 9858608,
      "step": 15110
    },
    {
      "epoch": 7.921907756813417,
      "grad_norm": 2.6509652137756348,
      "learning_rate": 3.779588544786044e-05,
      "loss": 0.2267,
      "num_input_tokens_seen": 9861776,
      "step": 15115
    },
    {
      "epoch": 7.9245283018867925,
      "grad_norm": 3.327876329421997,
      "learning_rate": 3.778606110400761e-05,
      "loss": 0.287,
      "num_input_tokens_seen": 9864144,
      "step": 15120
    },
    {
      "epoch": 7.927148846960168,
      "grad_norm": 2.2496585845947266,
      "learning_rate": 3.7776234085456665e-05,
      "loss": 0.2675,
      "num_input_tokens_seen": 9866512,
      "step": 15125
    },
    {
      "epoch": 7.929769392033543,
      "grad_norm": 1.7331331968307495,
      "learning_rate": 3.776640439426332e-05,
      "loss": 0.2596,
      "num_input_tokens_seen": 9869936,
      "step": 15130
    },
    {
      "epoch": 7.932389937106918,
      "grad_norm": 1.718220829963684,
      "learning_rate": 3.775657203248381e-05,
      "loss": 0.1906,
      "num_input_tokens_seen": 9873168,
      "step": 15135
    },
    {
      "epoch": 7.935010482180293,
      "grad_norm": 2.5019240379333496,
      "learning_rate": 3.774673700217497e-05,
      "loss": 0.329,
      "num_input_tokens_seen": 9875888,
      "step": 15140
    },
    {
      "epoch": 7.937631027253669,
      "grad_norm": 1.6498503684997559,
      "learning_rate": 3.773689930539418e-05,
      "loss": 0.3037,
      "num_input_tokens_seen": 9880432,
      "step": 15145
    },
    {
      "epoch": 7.940251572327044,
      "grad_norm": 2.218024730682373,
      "learning_rate": 3.7727058944199354e-05,
      "loss": 0.2223,
      "num_input_tokens_seen": 9885136,
      "step": 15150
    },
    {
      "epoch": 7.9428721174004195,
      "grad_norm": 2.967679500579834,
      "learning_rate": 3.7717215920648994e-05,
      "loss": 0.2317,
      "num_input_tokens_seen": 9887920,
      "step": 15155
    },
    {
      "epoch": 7.945492662473795,
      "grad_norm": 1.8630698919296265,
      "learning_rate": 3.770737023680215e-05,
      "loss": 0.2336,
      "num_input_tokens_seen": 9892464,
      "step": 15160
    },
    {
      "epoch": 7.94811320754717,
      "grad_norm": 2.508331775665283,
      "learning_rate": 3.7697521894718436e-05,
      "loss": 0.2212,
      "num_input_tokens_seen": 9895344,
      "step": 15165
    },
    {
      "epoch": 7.950733752620545,
      "grad_norm": 1.3251688480377197,
      "learning_rate": 3.7687670896458e-05,
      "loss": 0.2776,
      "num_input_tokens_seen": 9898032,
      "step": 15170
    },
    {
      "epoch": 7.95335429769392,
      "grad_norm": 3.0694518089294434,
      "learning_rate": 3.7677817244081556e-05,
      "loss": 0.1735,
      "num_input_tokens_seen": 9902160,
      "step": 15175
    },
    {
      "epoch": 7.955974842767295,
      "grad_norm": 1.8541088104248047,
      "learning_rate": 3.766796093965038e-05,
      "loss": 0.2258,
      "num_input_tokens_seen": 9906160,
      "step": 15180
    },
    {
      "epoch": 7.9585953878406706,
      "grad_norm": 1.78255295753479,
      "learning_rate": 3.76581019852263e-05,
      "loss": 0.2219,
      "num_input_tokens_seen": 9909296,
      "step": 15185
    },
    {
      "epoch": 7.961215932914046,
      "grad_norm": 1.6418795585632324,
      "learning_rate": 3.7648240382871705e-05,
      "loss": 0.274,
      "num_input_tokens_seen": 9911952,
      "step": 15190
    },
    {
      "epoch": 7.963836477987421,
      "grad_norm": 1.921459436416626,
      "learning_rate": 3.763837613464951e-05,
      "loss": 0.2831,
      "num_input_tokens_seen": 9914992,
      "step": 15195
    },
    {
      "epoch": 7.966457023060797,
      "grad_norm": 1.902962565422058,
      "learning_rate": 3.7628509242623215e-05,
      "loss": 0.1976,
      "num_input_tokens_seen": 9919184,
      "step": 15200
    },
    {
      "epoch": 7.969077568134172,
      "grad_norm": 2.18902325630188,
      "learning_rate": 3.761863970885686e-05,
      "loss": 0.3415,
      "num_input_tokens_seen": 9921744,
      "step": 15205
    },
    {
      "epoch": 7.971698113207547,
      "grad_norm": 1.6919145584106445,
      "learning_rate": 3.760876753541504e-05,
      "loss": 0.2502,
      "num_input_tokens_seen": 9924912,
      "step": 15210
    },
    {
      "epoch": 7.9743186582809225,
      "grad_norm": 1.3802686929702759,
      "learning_rate": 3.7598892724362907e-05,
      "loss": 0.2012,
      "num_input_tokens_seen": 9927728,
      "step": 15215
    },
    {
      "epoch": 7.976939203354298,
      "grad_norm": 2.0740044116973877,
      "learning_rate": 3.758901527776615e-05,
      "loss": 0.2098,
      "num_input_tokens_seen": 9930448,
      "step": 15220
    },
    {
      "epoch": 7.979559748427673,
      "grad_norm": 2.712158679962158,
      "learning_rate": 3.7579135197691015e-05,
      "loss": 0.2648,
      "num_input_tokens_seen": 9934608,
      "step": 15225
    },
    {
      "epoch": 7.982180293501048,
      "grad_norm": 2.116950750350952,
      "learning_rate": 3.7569252486204306e-05,
      "loss": 0.1785,
      "num_input_tokens_seen": 9937456,
      "step": 15230
    },
    {
      "epoch": 7.984800838574423,
      "grad_norm": 2.7488365173339844,
      "learning_rate": 3.755936714537338e-05,
      "loss": 0.2855,
      "num_input_tokens_seen": 9940816,
      "step": 15235
    },
    {
      "epoch": 7.987421383647799,
      "grad_norm": 2.327174663543701,
      "learning_rate": 3.754947917726612e-05,
      "loss": 0.2465,
      "num_input_tokens_seen": 9944080,
      "step": 15240
    },
    {
      "epoch": 7.990041928721174,
      "grad_norm": 3.299860715866089,
      "learning_rate": 3.7539588583951e-05,
      "loss": 0.1877,
      "num_input_tokens_seen": 9946480,
      "step": 15245
    },
    {
      "epoch": 7.9926624737945495,
      "grad_norm": 1.578603982925415,
      "learning_rate": 3.752969536749701e-05,
      "loss": 0.2304,
      "num_input_tokens_seen": 9949872,
      "step": 15250
    },
    {
      "epoch": 7.995283018867925,
      "grad_norm": 2.3856253623962402,
      "learning_rate": 3.7519799529973685e-05,
      "loss": 0.2217,
      "num_input_tokens_seen": 9953520,
      "step": 15255
    },
    {
      "epoch": 7.9979035639413,
      "grad_norm": 2.1299054622650146,
      "learning_rate": 3.7509901073451134e-05,
      "loss": 0.2559,
      "num_input_tokens_seen": 9956624,
      "step": 15260
    },
    {
      "epoch": 8.0,
      "eval_loss": 0.619249165058136,
      "eval_runtime": 12.2208,
      "eval_samples_per_second": 69.39,
      "eval_steps_per_second": 17.348,
      "num_input_tokens_seen": 9958360,
      "step": 15264
    },
    {
      "epoch": 8.000524109014675,
      "grad_norm": 0.9792529344558716,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.2407,
      "num_input_tokens_seen": 9958968,
      "step": 15265
    },
    {
      "epoch": 8.00314465408805,
      "grad_norm": 1.6599009037017822,
      "learning_rate": 3.749009631169147e-05,
      "loss": 0.169,
      "num_input_tokens_seen": 9961624,
      "step": 15270
    },
    {
      "epoch": 8.005765199161425,
      "grad_norm": 2.2836098670959473,
      "learning_rate": 3.74801900105973e-05,
      "loss": 0.1945,
      "num_input_tokens_seen": 9964600,
      "step": 15275
    },
    {
      "epoch": 8.0083857442348,
      "grad_norm": 1.6185920238494873,
      "learning_rate": 3.747028109878975e-05,
      "loss": 0.1275,
      "num_input_tokens_seen": 9967096,
      "step": 15280
    },
    {
      "epoch": 8.011006289308176,
      "grad_norm": 2.3193094730377197,
      "learning_rate": 3.746036957834167e-05,
      "loss": 0.1278,
      "num_input_tokens_seen": 9969848,
      "step": 15285
    },
    {
      "epoch": 8.01362683438155,
      "grad_norm": 2.609950304031372,
      "learning_rate": 3.7450455451326415e-05,
      "loss": 0.1786,
      "num_input_tokens_seen": 9973272,
      "step": 15290
    },
    {
      "epoch": 8.016247379454926,
      "grad_norm": 2.1599223613739014,
      "learning_rate": 3.7440538719817935e-05,
      "loss": 0.1842,
      "num_input_tokens_seen": 9975992,
      "step": 15295
    },
    {
      "epoch": 8.018867924528301,
      "grad_norm": 1.738405704498291,
      "learning_rate": 3.743061938589068e-05,
      "loss": 0.1812,
      "num_input_tokens_seen": 9979704,
      "step": 15300
    },
    {
      "epoch": 8.021488469601676,
      "grad_norm": 2.288848876953125,
      "learning_rate": 3.742069745161966e-05,
      "loss": 0.2026,
      "num_input_tokens_seen": 9982712,
      "step": 15305
    },
    {
      "epoch": 8.024109014675052,
      "grad_norm": 1.969607949256897,
      "learning_rate": 3.741077291908045e-05,
      "loss": 0.092,
      "num_input_tokens_seen": 9985560,
      "step": 15310
    },
    {
      "epoch": 8.026729559748428,
      "grad_norm": 2.7487454414367676,
      "learning_rate": 3.740084579034913e-05,
      "loss": 0.1637,
      "num_input_tokens_seen": 9989272,
      "step": 15315
    },
    {
      "epoch": 8.029350104821804,
      "grad_norm": 2.0313467979431152,
      "learning_rate": 3.739091606750234e-05,
      "loss": 0.1797,
      "num_input_tokens_seen": 9992088,
      "step": 15320
    },
    {
      "epoch": 8.031970649895179,
      "grad_norm": 2.5419957637786865,
      "learning_rate": 3.7380983752617275e-05,
      "loss": 0.1995,
      "num_input_tokens_seen": 9995032,
      "step": 15325
    },
    {
      "epoch": 8.034591194968554,
      "grad_norm": 3.2396798133850098,
      "learning_rate": 3.737104884777166e-05,
      "loss": 0.1407,
      "num_input_tokens_seen": 9998008,
      "step": 15330
    },
    {
      "epoch": 8.03721174004193,
      "grad_norm": 1.2174566984176636,
      "learning_rate": 3.736111135504376e-05,
      "loss": 0.2358,
      "num_input_tokens_seen": 10001464,
      "step": 15335
    },
    {
      "epoch": 8.039832285115304,
      "grad_norm": 1.1038330793380737,
      "learning_rate": 3.735117127651239e-05,
      "loss": 0.1613,
      "num_input_tokens_seen": 10004472,
      "step": 15340
    },
    {
      "epoch": 8.04245283018868,
      "grad_norm": 2.531731605529785,
      "learning_rate": 3.734122861425689e-05,
      "loss": 0.1708,
      "num_input_tokens_seen": 10007160,
      "step": 15345
    },
    {
      "epoch": 8.045073375262055,
      "grad_norm": 1.4272749423980713,
      "learning_rate": 3.7331283370357173e-05,
      "loss": 0.2055,
      "num_input_tokens_seen": 10010136,
      "step": 15350
    },
    {
      "epoch": 8.04769392033543,
      "grad_norm": 2.5909526348114014,
      "learning_rate": 3.732133554689365e-05,
      "loss": 0.1998,
      "num_input_tokens_seen": 10013560,
      "step": 15355
    },
    {
      "epoch": 8.050314465408805,
      "grad_norm": 2.2174482345581055,
      "learning_rate": 3.73113851459473e-05,
      "loss": 0.1224,
      "num_input_tokens_seen": 10016280,
      "step": 15360
    },
    {
      "epoch": 8.05293501048218,
      "grad_norm": 2.413421154022217,
      "learning_rate": 3.730143216959964e-05,
      "loss": 0.2331,
      "num_input_tokens_seen": 10019096,
      "step": 15365
    },
    {
      "epoch": 8.055555555555555,
      "grad_norm": 2.8074445724487305,
      "learning_rate": 3.729147661993271e-05,
      "loss": 0.2342,
      "num_input_tokens_seen": 10022200,
      "step": 15370
    },
    {
      "epoch": 8.05817610062893,
      "grad_norm": 1.8107343912124634,
      "learning_rate": 3.728151849902909e-05,
      "loss": 0.1483,
      "num_input_tokens_seen": 10025528,
      "step": 15375
    },
    {
      "epoch": 8.060796645702306,
      "grad_norm": 11.123870849609375,
      "learning_rate": 3.727155780897193e-05,
      "loss": 0.2451,
      "num_input_tokens_seen": 10030104,
      "step": 15380
    },
    {
      "epoch": 8.06341719077568,
      "grad_norm": 4.0177321434021,
      "learning_rate": 3.7261594551844865e-05,
      "loss": 0.1908,
      "num_input_tokens_seen": 10033112,
      "step": 15385
    },
    {
      "epoch": 8.066037735849056,
      "grad_norm": 1.5235475301742554,
      "learning_rate": 3.7251628729732116e-05,
      "loss": 0.157,
      "num_input_tokens_seen": 10036280,
      "step": 15390
    },
    {
      "epoch": 8.068658280922431,
      "grad_norm": 1.4313386678695679,
      "learning_rate": 3.724166034471841e-05,
      "loss": 0.1243,
      "num_input_tokens_seen": 10039160,
      "step": 15395
    },
    {
      "epoch": 8.071278825995806,
      "grad_norm": 1.2548720836639404,
      "learning_rate": 3.723168939888901e-05,
      "loss": 0.146,
      "num_input_tokens_seen": 10043224,
      "step": 15400
    },
    {
      "epoch": 8.073899371069182,
      "grad_norm": 1.8209854364395142,
      "learning_rate": 3.7221715894329747e-05,
      "loss": 0.132,
      "num_input_tokens_seen": 10045976,
      "step": 15405
    },
    {
      "epoch": 8.076519916142558,
      "grad_norm": 2.290376901626587,
      "learning_rate": 3.7211739833126945e-05,
      "loss": 0.1524,
      "num_input_tokens_seen": 10048824,
      "step": 15410
    },
    {
      "epoch": 8.079140461215934,
      "grad_norm": 3.474961996078491,
      "learning_rate": 3.720176121736748e-05,
      "loss": 0.149,
      "num_input_tokens_seen": 10051512,
      "step": 15415
    },
    {
      "epoch": 8.081761006289309,
      "grad_norm": 1.8648091554641724,
      "learning_rate": 3.719178004913879e-05,
      "loss": 0.2173,
      "num_input_tokens_seen": 10054712,
      "step": 15420
    },
    {
      "epoch": 8.084381551362684,
      "grad_norm": 1.9159284830093384,
      "learning_rate": 3.7181796330528796e-05,
      "loss": 0.2546,
      "num_input_tokens_seen": 10058232,
      "step": 15425
    },
    {
      "epoch": 8.08700209643606,
      "grad_norm": 1.0637905597686768,
      "learning_rate": 3.717181006362599e-05,
      "loss": 0.1756,
      "num_input_tokens_seen": 10061880,
      "step": 15430
    },
    {
      "epoch": 8.089622641509434,
      "grad_norm": 2.159672975540161,
      "learning_rate": 3.7161821250519384e-05,
      "loss": 0.1859,
      "num_input_tokens_seen": 10064984,
      "step": 15435
    },
    {
      "epoch": 8.09224318658281,
      "grad_norm": 3.1562259197235107,
      "learning_rate": 3.7151829893298515e-05,
      "loss": 0.14,
      "num_input_tokens_seen": 10067384,
      "step": 15440
    },
    {
      "epoch": 8.094863731656185,
      "grad_norm": 1.2812689542770386,
      "learning_rate": 3.714183599405347e-05,
      "loss": 0.1674,
      "num_input_tokens_seen": 10070776,
      "step": 15445
    },
    {
      "epoch": 8.09748427672956,
      "grad_norm": 1.2421578168869019,
      "learning_rate": 3.713183955487487e-05,
      "loss": 0.1487,
      "num_input_tokens_seen": 10074104,
      "step": 15450
    },
    {
      "epoch": 8.100104821802935,
      "grad_norm": 1.9206558465957642,
      "learning_rate": 3.7121840577853815e-05,
      "loss": 0.1561,
      "num_input_tokens_seen": 10077784,
      "step": 15455
    },
    {
      "epoch": 8.10272536687631,
      "grad_norm": 1.8149635791778564,
      "learning_rate": 3.7111839065082027e-05,
      "loss": 0.1961,
      "num_input_tokens_seen": 10082392,
      "step": 15460
    },
    {
      "epoch": 8.105345911949685,
      "grad_norm": 1.5419586896896362,
      "learning_rate": 3.710183501865168e-05,
      "loss": 0.1462,
      "num_input_tokens_seen": 10085656,
      "step": 15465
    },
    {
      "epoch": 8.10796645702306,
      "grad_norm": 2.687856674194336,
      "learning_rate": 3.7091828440655514e-05,
      "loss": 0.1815,
      "num_input_tokens_seen": 10089688,
      "step": 15470
    },
    {
      "epoch": 8.110587002096436,
      "grad_norm": 2.20747971534729,
      "learning_rate": 3.708181933318678e-05,
      "loss": 0.2846,
      "num_input_tokens_seen": 10092632,
      "step": 15475
    },
    {
      "epoch": 8.11320754716981,
      "grad_norm": 2.5756919384002686,
      "learning_rate": 3.707180769833929e-05,
      "loss": 0.1428,
      "num_input_tokens_seen": 10095384,
      "step": 15480
    },
    {
      "epoch": 8.115828092243186,
      "grad_norm": 15.717047691345215,
      "learning_rate": 3.706179353820734e-05,
      "loss": 0.1808,
      "num_input_tokens_seen": 10101272,
      "step": 15485
    },
    {
      "epoch": 8.118448637316561,
      "grad_norm": 1.633732557296753,
      "learning_rate": 3.7051776854885794e-05,
      "loss": 0.1513,
      "num_input_tokens_seen": 10104920,
      "step": 15490
    },
    {
      "epoch": 8.121069182389936,
      "grad_norm": 3.971987247467041,
      "learning_rate": 3.7041757650470025e-05,
      "loss": 0.173,
      "num_input_tokens_seen": 10108440,
      "step": 15495
    },
    {
      "epoch": 8.123689727463312,
      "grad_norm": 2.3423004150390625,
      "learning_rate": 3.703173592705594e-05,
      "loss": 0.184,
      "num_input_tokens_seen": 10111672,
      "step": 15500
    },
    {
      "epoch": 8.126310272536688,
      "grad_norm": 1.7297232151031494,
      "learning_rate": 3.702171168673996e-05,
      "loss": 0.227,
      "num_input_tokens_seen": 10115448,
      "step": 15505
    },
    {
      "epoch": 8.128930817610064,
      "grad_norm": 1.085051417350769,
      "learning_rate": 3.701168493161905e-05,
      "loss": 0.1354,
      "num_input_tokens_seen": 10118328,
      "step": 15510
    },
    {
      "epoch": 8.131551362683439,
      "grad_norm": 2.8746824264526367,
      "learning_rate": 3.7001655663790676e-05,
      "loss": 0.1594,
      "num_input_tokens_seen": 10122328,
      "step": 15515
    },
    {
      "epoch": 8.134171907756814,
      "grad_norm": 1.9518284797668457,
      "learning_rate": 3.699162388535288e-05,
      "loss": 0.2259,
      "num_input_tokens_seen": 10125112,
      "step": 15520
    },
    {
      "epoch": 8.13679245283019,
      "grad_norm": 1.9318794012069702,
      "learning_rate": 3.698158959840416e-05,
      "loss": 0.1405,
      "num_input_tokens_seen": 10128280,
      "step": 15525
    },
    {
      "epoch": 8.139412997903564,
      "grad_norm": 2.1163628101348877,
      "learning_rate": 3.6971552805043585e-05,
      "loss": 0.1416,
      "num_input_tokens_seen": 10131320,
      "step": 15530
    },
    {
      "epoch": 8.14203354297694,
      "grad_norm": 2.738936185836792,
      "learning_rate": 3.6961513507370754e-05,
      "loss": 0.1488,
      "num_input_tokens_seen": 10135384,
      "step": 15535
    },
    {
      "epoch": 8.144654088050315,
      "grad_norm": 2.0196638107299805,
      "learning_rate": 3.695147170748576e-05,
      "loss": 0.2305,
      "num_input_tokens_seen": 10138520,
      "step": 15540
    },
    {
      "epoch": 8.14727463312369,
      "grad_norm": 1.8618966341018677,
      "learning_rate": 3.694142740748923e-05,
      "loss": 0.2023,
      "num_input_tokens_seen": 10141528,
      "step": 15545
    },
    {
      "epoch": 8.149895178197065,
      "grad_norm": 1.5964767932891846,
      "learning_rate": 3.693138060948232e-05,
      "loss": 0.1916,
      "num_input_tokens_seen": 10144568,
      "step": 15550
    },
    {
      "epoch": 8.15251572327044,
      "grad_norm": 1.7206323146820068,
      "learning_rate": 3.692133131556671e-05,
      "loss": 0.1802,
      "num_input_tokens_seen": 10148504,
      "step": 15555
    },
    {
      "epoch": 8.155136268343815,
      "grad_norm": 1.0969326496124268,
      "learning_rate": 3.6911279527844596e-05,
      "loss": 0.1887,
      "num_input_tokens_seen": 10151288,
      "step": 15560
    },
    {
      "epoch": 8.15775681341719,
      "grad_norm": 2.325395345687866,
      "learning_rate": 3.69012252484187e-05,
      "loss": 0.1833,
      "num_input_tokens_seen": 10154744,
      "step": 15565
    },
    {
      "epoch": 8.160377358490566,
      "grad_norm": 1.8252050876617432,
      "learning_rate": 3.6891168479392245e-05,
      "loss": 0.1346,
      "num_input_tokens_seen": 10158136,
      "step": 15570
    },
    {
      "epoch": 8.16299790356394,
      "grad_norm": 2.094594717025757,
      "learning_rate": 3.688110922286901e-05,
      "loss": 0.1778,
      "num_input_tokens_seen": 10161560,
      "step": 15575
    },
    {
      "epoch": 8.165618448637316,
      "grad_norm": 2.499248504638672,
      "learning_rate": 3.687104748095328e-05,
      "loss": 0.1806,
      "num_input_tokens_seen": 10163960,
      "step": 15580
    },
    {
      "epoch": 8.168238993710691,
      "grad_norm": 4.317614555358887,
      "learning_rate": 3.686098325574984e-05,
      "loss": 0.1301,
      "num_input_tokens_seen": 10166776,
      "step": 15585
    },
    {
      "epoch": 8.170859538784066,
      "grad_norm": 2.1915440559387207,
      "learning_rate": 3.685091654936403e-05,
      "loss": 0.1238,
      "num_input_tokens_seen": 10169592,
      "step": 15590
    },
    {
      "epoch": 8.173480083857442,
      "grad_norm": 1.5925642251968384,
      "learning_rate": 3.6840847363901675e-05,
      "loss": 0.1309,
      "num_input_tokens_seen": 10172760,
      "step": 15595
    },
    {
      "epoch": 8.176100628930818,
      "grad_norm": 1.0690524578094482,
      "learning_rate": 3.683077570146914e-05,
      "loss": 0.1962,
      "num_input_tokens_seen": 10177496,
      "step": 15600
    },
    {
      "epoch": 8.178721174004194,
      "grad_norm": 3.4649503231048584,
      "learning_rate": 3.68207015641733e-05,
      "loss": 0.1748,
      "num_input_tokens_seen": 10181112,
      "step": 15605
    },
    {
      "epoch": 8.181341719077569,
      "grad_norm": 1.5991069078445435,
      "learning_rate": 3.681062495412155e-05,
      "loss": 0.2002,
      "num_input_tokens_seen": 10184632,
      "step": 15610
    },
    {
      "epoch": 8.183962264150944,
      "grad_norm": 1.4389632940292358,
      "learning_rate": 3.68005458734218e-05,
      "loss": 0.1757,
      "num_input_tokens_seen": 10187576,
      "step": 15615
    },
    {
      "epoch": 8.18658280922432,
      "grad_norm": 2.725731134414673,
      "learning_rate": 3.679046432418248e-05,
      "loss": 0.2439,
      "num_input_tokens_seen": 10190648,
      "step": 15620
    },
    {
      "epoch": 8.189203354297694,
      "grad_norm": 1.7787295579910278,
      "learning_rate": 3.6780380308512543e-05,
      "loss": 0.1162,
      "num_input_tokens_seen": 10193944,
      "step": 15625
    },
    {
      "epoch": 8.19182389937107,
      "grad_norm": 1.1966735124588013,
      "learning_rate": 3.677029382852144e-05,
      "loss": 0.1712,
      "num_input_tokens_seen": 10196824,
      "step": 15630
    },
    {
      "epoch": 8.194444444444445,
      "grad_norm": 2.586300849914551,
      "learning_rate": 3.676020488631915e-05,
      "loss": 0.2164,
      "num_input_tokens_seen": 10200312,
      "step": 15635
    },
    {
      "epoch": 8.19706498951782,
      "grad_norm": 1.1005750894546509,
      "learning_rate": 3.675011348401617e-05,
      "loss": 0.2241,
      "num_input_tokens_seen": 10203992,
      "step": 15640
    },
    {
      "epoch": 8.199685534591195,
      "grad_norm": 4.6011223793029785,
      "learning_rate": 3.6740019623723494e-05,
      "loss": 0.1579,
      "num_input_tokens_seen": 10206904,
      "step": 15645
    },
    {
      "epoch": 8.20230607966457,
      "grad_norm": 2.62819504737854,
      "learning_rate": 3.672992330755265e-05,
      "loss": 0.1706,
      "num_input_tokens_seen": 10209880,
      "step": 15650
    },
    {
      "epoch": 8.204926624737945,
      "grad_norm": 1.6158723831176758,
      "learning_rate": 3.671982453761568e-05,
      "loss": 0.1495,
      "num_input_tokens_seen": 10215864,
      "step": 15655
    },
    {
      "epoch": 8.20754716981132,
      "grad_norm": 2.359524726867676,
      "learning_rate": 3.6709723316025114e-05,
      "loss": 0.154,
      "num_input_tokens_seen": 10218904,
      "step": 15660
    },
    {
      "epoch": 8.210167714884696,
      "grad_norm": 2.2631478309631348,
      "learning_rate": 3.669961964489403e-05,
      "loss": 0.1325,
      "num_input_tokens_seen": 10225048,
      "step": 15665
    },
    {
      "epoch": 8.21278825995807,
      "grad_norm": 1.1748558282852173,
      "learning_rate": 3.668951352633599e-05,
      "loss": 0.1834,
      "num_input_tokens_seen": 10227576,
      "step": 15670
    },
    {
      "epoch": 8.215408805031446,
      "grad_norm": 18.56480598449707,
      "learning_rate": 3.6679404962465084e-05,
      "loss": 0.2951,
      "num_input_tokens_seen": 10230712,
      "step": 15675
    },
    {
      "epoch": 8.218029350104821,
      "grad_norm": 14.888421058654785,
      "learning_rate": 3.66692939553959e-05,
      "loss": 0.1932,
      "num_input_tokens_seen": 10233848,
      "step": 15680
    },
    {
      "epoch": 8.220649895178196,
      "grad_norm": 1.6420574188232422,
      "learning_rate": 3.6659180507243554e-05,
      "loss": 0.2028,
      "num_input_tokens_seen": 10237816,
      "step": 15685
    },
    {
      "epoch": 8.223270440251572,
      "grad_norm": 1.2742542028427124,
      "learning_rate": 3.664906462012366e-05,
      "loss": 0.1711,
      "num_input_tokens_seen": 10241272,
      "step": 15690
    },
    {
      "epoch": 8.225890985324948,
      "grad_norm": 1.9400261640548706,
      "learning_rate": 3.663894629615234e-05,
      "loss": 0.1512,
      "num_input_tokens_seen": 10243864,
      "step": 15695
    },
    {
      "epoch": 8.228511530398324,
      "grad_norm": 1.2280672788619995,
      "learning_rate": 3.6628825537446246e-05,
      "loss": 0.1234,
      "num_input_tokens_seen": 10246712,
      "step": 15700
    },
    {
      "epoch": 8.231132075471699,
      "grad_norm": 0.8906325101852417,
      "learning_rate": 3.6618702346122525e-05,
      "loss": 0.2283,
      "num_input_tokens_seen": 10250648,
      "step": 15705
    },
    {
      "epoch": 8.233752620545074,
      "grad_norm": 2.339695930480957,
      "learning_rate": 3.6608576724298817e-05,
      "loss": 0.1647,
      "num_input_tokens_seen": 10253624,
      "step": 15710
    },
    {
      "epoch": 8.23637316561845,
      "grad_norm": 4.0242438316345215,
      "learning_rate": 3.65984486740933e-05,
      "loss": 0.1709,
      "num_input_tokens_seen": 10256536,
      "step": 15715
    },
    {
      "epoch": 8.238993710691824,
      "grad_norm": 5.297937393188477,
      "learning_rate": 3.6588318197624644e-05,
      "loss": 0.1955,
      "num_input_tokens_seen": 10259448,
      "step": 15720
    },
    {
      "epoch": 8.2416142557652,
      "grad_norm": 3.6101534366607666,
      "learning_rate": 3.6578185297012015e-05,
      "loss": 0.1617,
      "num_input_tokens_seen": 10263768,
      "step": 15725
    },
    {
      "epoch": 8.244234800838575,
      "grad_norm": 2.4257395267486572,
      "learning_rate": 3.656804997437512e-05,
      "loss": 0.1506,
      "num_input_tokens_seen": 10266712,
      "step": 15730
    },
    {
      "epoch": 8.24685534591195,
      "grad_norm": 3.1558778285980225,
      "learning_rate": 3.6557912231834155e-05,
      "loss": 0.2637,
      "num_input_tokens_seen": 10270712,
      "step": 15735
    },
    {
      "epoch": 8.249475890985325,
      "grad_norm": 3.7028324604034424,
      "learning_rate": 3.65477720715098e-05,
      "loss": 0.2559,
      "num_input_tokens_seen": 10274456,
      "step": 15740
    },
    {
      "epoch": 8.2520964360587,
      "grad_norm": 1.1029785871505737,
      "learning_rate": 3.653762949552328e-05,
      "loss": 0.2006,
      "num_input_tokens_seen": 10278840,
      "step": 15745
    },
    {
      "epoch": 8.254716981132075,
      "grad_norm": 1.3886858224868774,
      "learning_rate": 3.6527484505996295e-05,
      "loss": 0.1453,
      "num_input_tokens_seen": 10282104,
      "step": 15750
    },
    {
      "epoch": 8.25733752620545,
      "grad_norm": 3.158134937286377,
      "learning_rate": 3.651733710505106e-05,
      "loss": 0.2494,
      "num_input_tokens_seen": 10285272,
      "step": 15755
    },
    {
      "epoch": 8.259958071278826,
      "grad_norm": 3.0508177280426025,
      "learning_rate": 3.6507187294810305e-05,
      "loss": 0.1466,
      "num_input_tokens_seen": 10288472,
      "step": 15760
    },
    {
      "epoch": 8.2625786163522,
      "grad_norm": 2.3618597984313965,
      "learning_rate": 3.649703507739724e-05,
      "loss": 0.2533,
      "num_input_tokens_seen": 10291128,
      "step": 15765
    },
    {
      "epoch": 8.265199161425576,
      "grad_norm": 1.8595223426818848,
      "learning_rate": 3.6486880454935606e-05,
      "loss": 0.1722,
      "num_input_tokens_seen": 10295288,
      "step": 15770
    },
    {
      "epoch": 8.267819706498951,
      "grad_norm": 7.713320732116699,
      "learning_rate": 3.647672342954963e-05,
      "loss": 0.1917,
      "num_input_tokens_seen": 10298680,
      "step": 15775
    },
    {
      "epoch": 8.270440251572326,
      "grad_norm": 5.135886192321777,
      "learning_rate": 3.646656400336405e-05,
      "loss": 0.1932,
      "num_input_tokens_seen": 10301496,
      "step": 15780
    },
    {
      "epoch": 8.273060796645701,
      "grad_norm": 3.5841283798217773,
      "learning_rate": 3.6456402178504094e-05,
      "loss": 0.1204,
      "num_input_tokens_seen": 10303992,
      "step": 15785
    },
    {
      "epoch": 8.275681341719078,
      "grad_norm": 1.509262204170227,
      "learning_rate": 3.6446237957095505e-05,
      "loss": 0.1742,
      "num_input_tokens_seen": 10306808,
      "step": 15790
    },
    {
      "epoch": 8.278301886792454,
      "grad_norm": 0.8338755965232849,
      "learning_rate": 3.643607134126452e-05,
      "loss": 0.1494,
      "num_input_tokens_seen": 10310584,
      "step": 15795
    },
    {
      "epoch": 8.280922431865829,
      "grad_norm": 0.8745976686477661,
      "learning_rate": 3.642590233313789e-05,
      "loss": 0.1437,
      "num_input_tokens_seen": 10314872,
      "step": 15800
    },
    {
      "epoch": 8.283542976939204,
      "grad_norm": 2.7725813388824463,
      "learning_rate": 3.6415730934842827e-05,
      "loss": 0.1416,
      "num_input_tokens_seen": 10317624,
      "step": 15805
    },
    {
      "epoch": 8.286163522012579,
      "grad_norm": 0.9383116960525513,
      "learning_rate": 3.64055571485071e-05,
      "loss": 0.1658,
      "num_input_tokens_seen": 10320408,
      "step": 15810
    },
    {
      "epoch": 8.288784067085954,
      "grad_norm": 2.282492160797119,
      "learning_rate": 3.639538097625894e-05,
      "loss": 0.1469,
      "num_input_tokens_seen": 10323320,
      "step": 15815
    },
    {
      "epoch": 8.29140461215933,
      "grad_norm": 2.0073347091674805,
      "learning_rate": 3.638520242022708e-05,
      "loss": 0.1581,
      "num_input_tokens_seen": 10326776,
      "step": 15820
    },
    {
      "epoch": 8.294025157232705,
      "grad_norm": 1.733025074005127,
      "learning_rate": 3.637502148254077e-05,
      "loss": 0.103,
      "num_input_tokens_seen": 10330584,
      "step": 15825
    },
    {
      "epoch": 8.29664570230608,
      "grad_norm": 1.7902544736862183,
      "learning_rate": 3.636483816532974e-05,
      "loss": 0.162,
      "num_input_tokens_seen": 10332824,
      "step": 15830
    },
    {
      "epoch": 8.299266247379455,
      "grad_norm": 2.702010154724121,
      "learning_rate": 3.6354652470724217e-05,
      "loss": 0.242,
      "num_input_tokens_seen": 10335480,
      "step": 15835
    },
    {
      "epoch": 8.30188679245283,
      "grad_norm": 1.5783493518829346,
      "learning_rate": 3.634446440085494e-05,
      "loss": 0.1991,
      "num_input_tokens_seen": 10338424,
      "step": 15840
    },
    {
      "epoch": 8.304507337526205,
      "grad_norm": 1.5931347608566284,
      "learning_rate": 3.633427395785313e-05,
      "loss": 0.1223,
      "num_input_tokens_seen": 10341784,
      "step": 15845
    },
    {
      "epoch": 8.30712788259958,
      "grad_norm": 1.4037725925445557,
      "learning_rate": 3.6324081143850516e-05,
      "loss": 0.21,
      "num_input_tokens_seen": 10345144,
      "step": 15850
    },
    {
      "epoch": 8.309748427672956,
      "grad_norm": 1.428037166595459,
      "learning_rate": 3.6313885960979326e-05,
      "loss": 0.1674,
      "num_input_tokens_seen": 10348984,
      "step": 15855
    },
    {
      "epoch": 8.31236897274633,
      "grad_norm": 2.098564624786377,
      "learning_rate": 3.630368841137226e-05,
      "loss": 0.2154,
      "num_input_tokens_seen": 10352152,
      "step": 15860
    },
    {
      "epoch": 8.314989517819706,
      "grad_norm": 5.556021690368652,
      "learning_rate": 3.6293488497162554e-05,
      "loss": 0.2566,
      "num_input_tokens_seen": 10354392,
      "step": 15865
    },
    {
      "epoch": 8.317610062893081,
      "grad_norm": 1.8836560249328613,
      "learning_rate": 3.628328622048388e-05,
      "loss": 0.4152,
      "num_input_tokens_seen": 10357368,
      "step": 15870
    },
    {
      "epoch": 8.320230607966456,
      "grad_norm": 3.221963882446289,
      "learning_rate": 3.627308158347046e-05,
      "loss": 0.1548,
      "num_input_tokens_seen": 10359864,
      "step": 15875
    },
    {
      "epoch": 8.322851153039831,
      "grad_norm": 1.9035792350769043,
      "learning_rate": 3.626287458825699e-05,
      "loss": 0.2561,
      "num_input_tokens_seen": 10364536,
      "step": 15880
    },
    {
      "epoch": 8.325471698113208,
      "grad_norm": 1.862870454788208,
      "learning_rate": 3.625266523697864e-05,
      "loss": 0.1811,
      "num_input_tokens_seen": 10367864,
      "step": 15885
    },
    {
      "epoch": 8.328092243186584,
      "grad_norm": 1.3576912879943848,
      "learning_rate": 3.62424535317711e-05,
      "loss": 0.2203,
      "num_input_tokens_seen": 10371736,
      "step": 15890
    },
    {
      "epoch": 8.330712788259959,
      "grad_norm": 1.8089543581008911,
      "learning_rate": 3.623223947477055e-05,
      "loss": 0.1459,
      "num_input_tokens_seen": 10375192,
      "step": 15895
    },
    {
      "epoch": 8.333333333333334,
      "grad_norm": 1.0645800828933716,
      "learning_rate": 3.6222023068113635e-05,
      "loss": 0.1771,
      "num_input_tokens_seen": 10383288,
      "step": 15900
    },
    {
      "epoch": 8.335953878406709,
      "grad_norm": 2.0468809604644775,
      "learning_rate": 3.621180431393754e-05,
      "loss": 0.2186,
      "num_input_tokens_seen": 10387192,
      "step": 15905
    },
    {
      "epoch": 8.338574423480084,
      "grad_norm": 2.9119255542755127,
      "learning_rate": 3.6201583214379874e-05,
      "loss": 0.1969,
      "num_input_tokens_seen": 10389400,
      "step": 15910
    },
    {
      "epoch": 8.34119496855346,
      "grad_norm": 1.8551273345947266,
      "learning_rate": 3.61913597715788e-05,
      "loss": 0.2342,
      "num_input_tokens_seen": 10392920,
      "step": 15915
    },
    {
      "epoch": 8.343815513626835,
      "grad_norm": 0.6295341849327087,
      "learning_rate": 3.6181133987672944e-05,
      "loss": 0.175,
      "num_input_tokens_seen": 10396280,
      "step": 15920
    },
    {
      "epoch": 8.34643605870021,
      "grad_norm": 4.256894588470459,
      "learning_rate": 3.617090586480142e-05,
      "loss": 0.1777,
      "num_input_tokens_seen": 10399960,
      "step": 15925
    },
    {
      "epoch": 8.349056603773585,
      "grad_norm": 2.3593928813934326,
      "learning_rate": 3.6160675405103826e-05,
      "loss": 0.1585,
      "num_input_tokens_seen": 10403128,
      "step": 15930
    },
    {
      "epoch": 8.35167714884696,
      "grad_norm": 3.064204454421997,
      "learning_rate": 3.615044261072027e-05,
      "loss": 0.1566,
      "num_input_tokens_seen": 10405976,
      "step": 15935
    },
    {
      "epoch": 8.354297693920335,
      "grad_norm": 2.4557743072509766,
      "learning_rate": 3.614020748379133e-05,
      "loss": 0.1562,
      "num_input_tokens_seen": 10408632,
      "step": 15940
    },
    {
      "epoch": 8.35691823899371,
      "grad_norm": 2.323655128479004,
      "learning_rate": 3.6129970026458076e-05,
      "loss": 0.2098,
      "num_input_tokens_seen": 10412216,
      "step": 15945
    },
    {
      "epoch": 8.359538784067086,
      "grad_norm": 2.899841547012329,
      "learning_rate": 3.611973024086207e-05,
      "loss": 0.1853,
      "num_input_tokens_seen": 10414680,
      "step": 15950
    },
    {
      "epoch": 8.36215932914046,
      "grad_norm": 2.9787800312042236,
      "learning_rate": 3.610948812914536e-05,
      "loss": 0.2098,
      "num_input_tokens_seen": 10417560,
      "step": 15955
    },
    {
      "epoch": 8.364779874213836,
      "grad_norm": 1.4984735250473022,
      "learning_rate": 3.609924369345048e-05,
      "loss": 0.1319,
      "num_input_tokens_seen": 10420056,
      "step": 15960
    },
    {
      "epoch": 8.367400419287211,
      "grad_norm": 3.045299530029297,
      "learning_rate": 3.6088996935920444e-05,
      "loss": 0.1309,
      "num_input_tokens_seen": 10423288,
      "step": 15965
    },
    {
      "epoch": 8.370020964360586,
      "grad_norm": 2.9724931716918945,
      "learning_rate": 3.607874785869876e-05,
      "loss": 0.2004,
      "num_input_tokens_seen": 10425752,
      "step": 15970
    },
    {
      "epoch": 8.372641509433961,
      "grad_norm": 1.605809211730957,
      "learning_rate": 3.6068496463929416e-05,
      "loss": 0.1383,
      "num_input_tokens_seen": 10429432,
      "step": 15975
    },
    {
      "epoch": 8.375262054507338,
      "grad_norm": 1.4927678108215332,
      "learning_rate": 3.605824275375689e-05,
      "loss": 0.1915,
      "num_input_tokens_seen": 10432856,
      "step": 15980
    },
    {
      "epoch": 8.377882599580714,
      "grad_norm": 2.150418758392334,
      "learning_rate": 3.604798673032613e-05,
      "loss": 0.1747,
      "num_input_tokens_seen": 10436600,
      "step": 15985
    },
    {
      "epoch": 8.380503144654089,
      "grad_norm": 1.2025349140167236,
      "learning_rate": 3.60377283957826e-05,
      "loss": 0.1644,
      "num_input_tokens_seen": 10439352,
      "step": 15990
    },
    {
      "epoch": 8.383123689727464,
      "grad_norm": 4.5889434814453125,
      "learning_rate": 3.6027467752272203e-05,
      "loss": 0.1936,
      "num_input_tokens_seen": 10441976,
      "step": 15995
    },
    {
      "epoch": 8.385744234800839,
      "grad_norm": 2.444249153137207,
      "learning_rate": 3.601720480194137e-05,
      "loss": 0.1564,
      "num_input_tokens_seen": 10445272,
      "step": 16000
    },
    {
      "epoch": 8.388364779874214,
      "grad_norm": 1.9813628196716309,
      "learning_rate": 3.600693954693698e-05,
      "loss": 0.2031,
      "num_input_tokens_seen": 10447864,
      "step": 16005
    },
    {
      "epoch": 8.39098532494759,
      "grad_norm": 3.2756500244140625,
      "learning_rate": 3.5996671989406414e-05,
      "loss": 0.14,
      "num_input_tokens_seen": 10451992,
      "step": 16010
    },
    {
      "epoch": 8.393605870020965,
      "grad_norm": 1.831421136856079,
      "learning_rate": 3.598640213149752e-05,
      "loss": 0.1935,
      "num_input_tokens_seen": 10454968,
      "step": 16015
    },
    {
      "epoch": 8.39622641509434,
      "grad_norm": 1.9325264692306519,
      "learning_rate": 3.597612997535864e-05,
      "loss": 0.1605,
      "num_input_tokens_seen": 10458392,
      "step": 16020
    },
    {
      "epoch": 8.398846960167715,
      "grad_norm": 1.0762895345687866,
      "learning_rate": 3.596585552313859e-05,
      "loss": 0.2234,
      "num_input_tokens_seen": 10461144,
      "step": 16025
    },
    {
      "epoch": 8.40146750524109,
      "grad_norm": 1.9716973304748535,
      "learning_rate": 3.595557877698666e-05,
      "loss": 0.1382,
      "num_input_tokens_seen": 10463768,
      "step": 16030
    },
    {
      "epoch": 8.404088050314465,
      "grad_norm": 2.370210886001587,
      "learning_rate": 3.594529973905266e-05,
      "loss": 0.1748,
      "num_input_tokens_seen": 10467128,
      "step": 16035
    },
    {
      "epoch": 8.40670859538784,
      "grad_norm": 5.195252418518066,
      "learning_rate": 3.593501841148681e-05,
      "loss": 0.1811,
      "num_input_tokens_seen": 10469240,
      "step": 16040
    },
    {
      "epoch": 8.409329140461216,
      "grad_norm": 1.8279451131820679,
      "learning_rate": 3.5924734796439864e-05,
      "loss": 0.2155,
      "num_input_tokens_seen": 10472728,
      "step": 16045
    },
    {
      "epoch": 8.41194968553459,
      "grad_norm": 1.8598625659942627,
      "learning_rate": 3.5914448896063044e-05,
      "loss": 0.1665,
      "num_input_tokens_seen": 10474936,
      "step": 16050
    },
    {
      "epoch": 8.414570230607966,
      "grad_norm": 1.8023585081100464,
      "learning_rate": 3.590416071250802e-05,
      "loss": 0.1839,
      "num_input_tokens_seen": 10478264,
      "step": 16055
    },
    {
      "epoch": 8.417190775681341,
      "grad_norm": 1.7969027757644653,
      "learning_rate": 3.589387024792699e-05,
      "loss": 0.2046,
      "num_input_tokens_seen": 10481432,
      "step": 16060
    },
    {
      "epoch": 8.419811320754716,
      "grad_norm": 1.859519124031067,
      "learning_rate": 3.5883577504472576e-05,
      "loss": 0.1773,
      "num_input_tokens_seen": 10483864,
      "step": 16065
    },
    {
      "epoch": 8.422431865828091,
      "grad_norm": 1.7100998163223267,
      "learning_rate": 3.587328248429792e-05,
      "loss": 0.2056,
      "num_input_tokens_seen": 10487480,
      "step": 16070
    },
    {
      "epoch": 8.425052410901468,
      "grad_norm": 2.336010217666626,
      "learning_rate": 3.586298518955661e-05,
      "loss": 0.1336,
      "num_input_tokens_seen": 10490616,
      "step": 16075
    },
    {
      "epoch": 8.427672955974844,
      "grad_norm": 1.8100587129592896,
      "learning_rate": 3.5852685622402746e-05,
      "loss": 0.1479,
      "num_input_tokens_seen": 10493528,
      "step": 16080
    },
    {
      "epoch": 8.430293501048219,
      "grad_norm": 2.935835361480713,
      "learning_rate": 3.584238378499085e-05,
      "loss": 0.2353,
      "num_input_tokens_seen": 10496568,
      "step": 16085
    },
    {
      "epoch": 8.432914046121594,
      "grad_norm": 1.7888193130493164,
      "learning_rate": 3.5832079679475974e-05,
      "loss": 0.1378,
      "num_input_tokens_seen": 10499992,
      "step": 16090
    },
    {
      "epoch": 8.435534591194969,
      "grad_norm": 1.5365256071090698,
      "learning_rate": 3.582177330801359e-05,
      "loss": 0.1767,
      "num_input_tokens_seen": 10502840,
      "step": 16095
    },
    {
      "epoch": 8.438155136268344,
      "grad_norm": 1.9328720569610596,
      "learning_rate": 3.58114646727597e-05,
      "loss": 0.1784,
      "num_input_tokens_seen": 10505752,
      "step": 16100
    },
    {
      "epoch": 8.44077568134172,
      "grad_norm": 1.1839603185653687,
      "learning_rate": 3.580115377587075e-05,
      "loss": 0.1394,
      "num_input_tokens_seen": 10508664,
      "step": 16105
    },
    {
      "epoch": 8.443396226415095,
      "grad_norm": 2.238865852355957,
      "learning_rate": 3.5790840619503654e-05,
      "loss": 0.2249,
      "num_input_tokens_seen": 10511416,
      "step": 16110
    },
    {
      "epoch": 8.44601677148847,
      "grad_norm": 1.856384515762329,
      "learning_rate": 3.57805252058158e-05,
      "loss": 0.1965,
      "num_input_tokens_seen": 10513944,
      "step": 16115
    },
    {
      "epoch": 8.448637316561845,
      "grad_norm": 1.7752645015716553,
      "learning_rate": 3.577020753696507e-05,
      "loss": 0.1702,
      "num_input_tokens_seen": 10517208,
      "step": 16120
    },
    {
      "epoch": 8.45125786163522,
      "grad_norm": 2.141605854034424,
      "learning_rate": 3.57598876151098e-05,
      "loss": 0.1515,
      "num_input_tokens_seen": 10519704,
      "step": 16125
    },
    {
      "epoch": 8.453878406708595,
      "grad_norm": 2.842458963394165,
      "learning_rate": 3.5749565442408796e-05,
      "loss": 0.2391,
      "num_input_tokens_seen": 10523160,
      "step": 16130
    },
    {
      "epoch": 8.45649895178197,
      "grad_norm": 1.807911992073059,
      "learning_rate": 3.573924102102133e-05,
      "loss": 0.1375,
      "num_input_tokens_seen": 10526168,
      "step": 16135
    },
    {
      "epoch": 8.459119496855346,
      "grad_norm": 1.4687801599502563,
      "learning_rate": 3.572891435310717e-05,
      "loss": 0.1734,
      "num_input_tokens_seen": 10528824,
      "step": 16140
    },
    {
      "epoch": 8.46174004192872,
      "grad_norm": 2.773615598678589,
      "learning_rate": 3.5718585440826527e-05,
      "loss": 0.1542,
      "num_input_tokens_seen": 10531640,
      "step": 16145
    },
    {
      "epoch": 8.464360587002096,
      "grad_norm": 2.6777358055114746,
      "learning_rate": 3.570825428634009e-05,
      "loss": 0.1592,
      "num_input_tokens_seen": 10534808,
      "step": 16150
    },
    {
      "epoch": 8.466981132075471,
      "grad_norm": 3.3147132396698,
      "learning_rate": 3.569792089180903e-05,
      "loss": 0.2069,
      "num_input_tokens_seen": 10537784,
      "step": 16155
    },
    {
      "epoch": 8.469601677148846,
      "grad_norm": 1.9958784580230713,
      "learning_rate": 3.568758525939497e-05,
      "loss": 0.1879,
      "num_input_tokens_seen": 10540760,
      "step": 16160
    },
    {
      "epoch": 8.472222222222221,
      "grad_norm": 1.1472814083099365,
      "learning_rate": 3.5677247391259995e-05,
      "loss": 0.1659,
      "num_input_tokens_seen": 10543928,
      "step": 16165
    },
    {
      "epoch": 8.474842767295598,
      "grad_norm": 1.9200501441955566,
      "learning_rate": 3.566690728956668e-05,
      "loss": 0.3179,
      "num_input_tokens_seen": 10546456,
      "step": 16170
    },
    {
      "epoch": 8.477463312368974,
      "grad_norm": 1.9826998710632324,
      "learning_rate": 3.5656564956478056e-05,
      "loss": 0.2133,
      "num_input_tokens_seen": 10549752,
      "step": 16175
    },
    {
      "epoch": 8.480083857442349,
      "grad_norm": 2.9855740070343018,
      "learning_rate": 3.5646220394157626e-05,
      "loss": 0.1745,
      "num_input_tokens_seen": 10553048,
      "step": 16180
    },
    {
      "epoch": 8.482704402515724,
      "grad_norm": 10.411952018737793,
      "learning_rate": 3.563587360476933e-05,
      "loss": 0.2143,
      "num_input_tokens_seen": 10555768,
      "step": 16185
    },
    {
      "epoch": 8.485324947589099,
      "grad_norm": 5.092311859130859,
      "learning_rate": 3.562552459047763e-05,
      "loss": 0.1849,
      "num_input_tokens_seen": 10559448,
      "step": 16190
    },
    {
      "epoch": 8.487945492662474,
      "grad_norm": 2.493594169616699,
      "learning_rate": 3.561517335344741e-05,
      "loss": 0.2313,
      "num_input_tokens_seen": 10562008,
      "step": 16195
    },
    {
      "epoch": 8.49056603773585,
      "grad_norm": 1.988491177558899,
      "learning_rate": 3.560481989584402e-05,
      "loss": 0.2186,
      "num_input_tokens_seen": 10564760,
      "step": 16200
    },
    {
      "epoch": 8.493186582809225,
      "grad_norm": 2.9098076820373535,
      "learning_rate": 3.55944642198333e-05,
      "loss": 0.2512,
      "num_input_tokens_seen": 10567512,
      "step": 16205
    },
    {
      "epoch": 8.4958071278826,
      "grad_norm": 4.289554595947266,
      "learning_rate": 3.5584106327581536e-05,
      "loss": 0.1405,
      "num_input_tokens_seen": 10570104,
      "step": 16210
    },
    {
      "epoch": 8.498427672955975,
      "grad_norm": 2.081965446472168,
      "learning_rate": 3.557374622125547e-05,
      "loss": 0.227,
      "num_input_tokens_seen": 10573304,
      "step": 16215
    },
    {
      "epoch": 8.50104821802935,
      "grad_norm": 2.281765937805176,
      "learning_rate": 3.556338390302234e-05,
      "loss": 0.1436,
      "num_input_tokens_seen": 10576408,
      "step": 16220
    },
    {
      "epoch": 8.503668763102725,
      "grad_norm": 2.4706315994262695,
      "learning_rate": 3.55530193750498e-05,
      "loss": 0.1076,
      "num_input_tokens_seen": 10579864,
      "step": 16225
    },
    {
      "epoch": 8.5062893081761,
      "grad_norm": 1.438325047492981,
      "learning_rate": 3.5542652639505996e-05,
      "loss": 0.254,
      "num_input_tokens_seen": 10582840,
      "step": 16230
    },
    {
      "epoch": 8.508909853249476,
      "grad_norm": 2.473665237426758,
      "learning_rate": 3.5532283698559556e-05,
      "loss": 0.2083,
      "num_input_tokens_seen": 10586392,
      "step": 16235
    },
    {
      "epoch": 8.51153039832285,
      "grad_norm": 1.9504570960998535,
      "learning_rate": 3.55219125543795e-05,
      "loss": 0.1729,
      "num_input_tokens_seen": 10591224,
      "step": 16240
    },
    {
      "epoch": 8.514150943396226,
      "grad_norm": 1.279008388519287,
      "learning_rate": 3.5511539209135396e-05,
      "loss": 0.2131,
      "num_input_tokens_seen": 10594296,
      "step": 16245
    },
    {
      "epoch": 8.516771488469601,
      "grad_norm": 1.96903657913208,
      "learning_rate": 3.550116366499721e-05,
      "loss": 0.1562,
      "num_input_tokens_seen": 10597912,
      "step": 16250
    },
    {
      "epoch": 8.519392033542976,
      "grad_norm": 5.480337142944336,
      "learning_rate": 3.549078592413538e-05,
      "loss": 0.1516,
      "num_input_tokens_seen": 10600632,
      "step": 16255
    },
    {
      "epoch": 8.522012578616351,
      "grad_norm": 2.075042963027954,
      "learning_rate": 3.548040598872083e-05,
      "loss": 0.1652,
      "num_input_tokens_seen": 10603672,
      "step": 16260
    },
    {
      "epoch": 8.524633123689728,
      "grad_norm": 2.420815944671631,
      "learning_rate": 3.5470023860924905e-05,
      "loss": 0.137,
      "num_input_tokens_seen": 10606648,
      "step": 16265
    },
    {
      "epoch": 8.527253668763104,
      "grad_norm": 0.9019947052001953,
      "learning_rate": 3.5459639542919444e-05,
      "loss": 0.1553,
      "num_input_tokens_seen": 10609432,
      "step": 16270
    },
    {
      "epoch": 8.529874213836479,
      "grad_norm": 1.7926181554794312,
      "learning_rate": 3.5449253036876715e-05,
      "loss": 0.1539,
      "num_input_tokens_seen": 10614424,
      "step": 16275
    },
    {
      "epoch": 8.532494758909854,
      "grad_norm": 1.3952322006225586,
      "learning_rate": 3.543886434496947e-05,
      "loss": 0.2157,
      "num_input_tokens_seen": 10617784,
      "step": 16280
    },
    {
      "epoch": 8.535115303983229,
      "grad_norm": 3.2416248321533203,
      "learning_rate": 3.542847346937089e-05,
      "loss": 0.2226,
      "num_input_tokens_seen": 10620216,
      "step": 16285
    },
    {
      "epoch": 8.537735849056604,
      "grad_norm": 1.3685145378112793,
      "learning_rate": 3.541808041225464e-05,
      "loss": 0.1956,
      "num_input_tokens_seen": 10622808,
      "step": 16290
    },
    {
      "epoch": 8.54035639412998,
      "grad_norm": 1.46420156955719,
      "learning_rate": 3.540768517579481e-05,
      "loss": 0.2117,
      "num_input_tokens_seen": 10625912,
      "step": 16295
    },
    {
      "epoch": 8.542976939203355,
      "grad_norm": 2.0649800300598145,
      "learning_rate": 3.539728776216599e-05,
      "loss": 0.2192,
      "num_input_tokens_seen": 10629016,
      "step": 16300
    },
    {
      "epoch": 8.54559748427673,
      "grad_norm": 1.3754719495773315,
      "learning_rate": 3.5386888173543185e-05,
      "loss": 0.2666,
      "num_input_tokens_seen": 10632920,
      "step": 16305
    },
    {
      "epoch": 8.548218029350105,
      "grad_norm": 2.4191737174987793,
      "learning_rate": 3.5376486412101875e-05,
      "loss": 0.1425,
      "num_input_tokens_seen": 10635800,
      "step": 16310
    },
    {
      "epoch": 8.55083857442348,
      "grad_norm": 2.66103458404541,
      "learning_rate": 3.5366082480017994e-05,
      "loss": 0.1641,
      "num_input_tokens_seen": 10638712,
      "step": 16315
    },
    {
      "epoch": 8.553459119496855,
      "grad_norm": 3.393319845199585,
      "learning_rate": 3.535567637946792e-05,
      "loss": 0.1471,
      "num_input_tokens_seen": 10641528,
      "step": 16320
    },
    {
      "epoch": 8.55607966457023,
      "grad_norm": 4.466306209564209,
      "learning_rate": 3.534526811262848e-05,
      "loss": 0.161,
      "num_input_tokens_seen": 10644152,
      "step": 16325
    },
    {
      "epoch": 8.558700209643606,
      "grad_norm": 1.5906344652175903,
      "learning_rate": 3.533485768167698e-05,
      "loss": 0.1662,
      "num_input_tokens_seen": 10647288,
      "step": 16330
    },
    {
      "epoch": 8.56132075471698,
      "grad_norm": 2.335606575012207,
      "learning_rate": 3.532444508879117e-05,
      "loss": 0.1533,
      "num_input_tokens_seen": 10650840,
      "step": 16335
    },
    {
      "epoch": 8.563941299790356,
      "grad_norm": 6.595895767211914,
      "learning_rate": 3.531403033614922e-05,
      "loss": 0.1349,
      "num_input_tokens_seen": 10654360,
      "step": 16340
    },
    {
      "epoch": 8.566561844863731,
      "grad_norm": 1.525862455368042,
      "learning_rate": 3.530361342592981e-05,
      "loss": 0.2106,
      "num_input_tokens_seen": 10659448,
      "step": 16345
    },
    {
      "epoch": 8.569182389937106,
      "grad_norm": 2.1614203453063965,
      "learning_rate": 3.5293194360312006e-05,
      "loss": 0.1909,
      "num_input_tokens_seen": 10662520,
      "step": 16350
    },
    {
      "epoch": 8.571802935010481,
      "grad_norm": 2.4439218044281006,
      "learning_rate": 3.528277314147538e-05,
      "loss": 0.1662,
      "num_input_tokens_seen": 10665176,
      "step": 16355
    },
    {
      "epoch": 8.574423480083858,
      "grad_norm": 2.2057037353515625,
      "learning_rate": 3.5272349771599925e-05,
      "loss": 0.1827,
      "num_input_tokens_seen": 10668984,
      "step": 16360
    },
    {
      "epoch": 8.577044025157234,
      "grad_norm": 0.8766621351242065,
      "learning_rate": 3.526192425286609e-05,
      "loss": 0.1413,
      "num_input_tokens_seen": 10673112,
      "step": 16365
    },
    {
      "epoch": 8.579664570230609,
      "grad_norm": 2.347872495651245,
      "learning_rate": 3.525149658745476e-05,
      "loss": 0.1179,
      "num_input_tokens_seen": 10676504,
      "step": 16370
    },
    {
      "epoch": 8.582285115303984,
      "grad_norm": 0.9631083607673645,
      "learning_rate": 3.5241066777547315e-05,
      "loss": 0.0765,
      "num_input_tokens_seen": 10679480,
      "step": 16375
    },
    {
      "epoch": 8.584905660377359,
      "grad_norm": 5.66265869140625,
      "learning_rate": 3.523063482532554e-05,
      "loss": 0.1428,
      "num_input_tokens_seen": 10682360,
      "step": 16380
    },
    {
      "epoch": 8.587526205450734,
      "grad_norm": 2.4019906520843506,
      "learning_rate": 3.5220200732971665e-05,
      "loss": 0.2175,
      "num_input_tokens_seen": 10686232,
      "step": 16385
    },
    {
      "epoch": 8.59014675052411,
      "grad_norm": 2.6605241298675537,
      "learning_rate": 3.52097645026684e-05,
      "loss": 0.172,
      "num_input_tokens_seen": 10689592,
      "step": 16390
    },
    {
      "epoch": 8.592767295597485,
      "grad_norm": 2.0312554836273193,
      "learning_rate": 3.519932613659889e-05,
      "loss": 0.2095,
      "num_input_tokens_seen": 10693272,
      "step": 16395
    },
    {
      "epoch": 8.59538784067086,
      "grad_norm": 3.0868027210235596,
      "learning_rate": 3.5188885636946696e-05,
      "loss": 0.1487,
      "num_input_tokens_seen": 10697560,
      "step": 16400
    },
    {
      "epoch": 8.598008385744235,
      "grad_norm": 1.8125090599060059,
      "learning_rate": 3.5178443005895884e-05,
      "loss": 0.3011,
      "num_input_tokens_seen": 10700664,
      "step": 16405
    },
    {
      "epoch": 8.60062893081761,
      "grad_norm": 2.760003089904785,
      "learning_rate": 3.51679982456309e-05,
      "loss": 0.2149,
      "num_input_tokens_seen": 10704696,
      "step": 16410
    },
    {
      "epoch": 8.603249475890985,
      "grad_norm": 1.544843077659607,
      "learning_rate": 3.51575513583367e-05,
      "loss": 0.2212,
      "num_input_tokens_seen": 10707512,
      "step": 16415
    },
    {
      "epoch": 8.60587002096436,
      "grad_norm": 1.9051544666290283,
      "learning_rate": 3.5147102346198635e-05,
      "loss": 0.132,
      "num_input_tokens_seen": 10711128,
      "step": 16420
    },
    {
      "epoch": 8.608490566037736,
      "grad_norm": 1.6339607238769531,
      "learning_rate": 3.5136651211402515e-05,
      "loss": 0.2149,
      "num_input_tokens_seen": 10714392,
      "step": 16425
    },
    {
      "epoch": 8.61111111111111,
      "grad_norm": 2.6701622009277344,
      "learning_rate": 3.5126197956134614e-05,
      "loss": 0.1762,
      "num_input_tokens_seen": 10718008,
      "step": 16430
    },
    {
      "epoch": 8.613731656184486,
      "grad_norm": 1.5887079238891602,
      "learning_rate": 3.5115742582581626e-05,
      "loss": 0.2113,
      "num_input_tokens_seen": 10721240,
      "step": 16435
    },
    {
      "epoch": 8.616352201257861,
      "grad_norm": 1.6315118074417114,
      "learning_rate": 3.51052850929307e-05,
      "loss": 0.2142,
      "num_input_tokens_seen": 10724312,
      "step": 16440
    },
    {
      "epoch": 8.618972746331236,
      "grad_norm": 2.9576468467712402,
      "learning_rate": 3.509482548936941e-05,
      "loss": 0.1628,
      "num_input_tokens_seen": 10726872,
      "step": 16445
    },
    {
      "epoch": 8.621593291404611,
      "grad_norm": 1.9392563104629517,
      "learning_rate": 3.5084363774085804e-05,
      "loss": 0.2227,
      "num_input_tokens_seen": 10730808,
      "step": 16450
    },
    {
      "epoch": 8.624213836477988,
      "grad_norm": 2.3267643451690674,
      "learning_rate": 3.507389994926834e-05,
      "loss": 0.165,
      "num_input_tokens_seen": 10733848,
      "step": 16455
    },
    {
      "epoch": 8.626834381551364,
      "grad_norm": 2.009117364883423,
      "learning_rate": 3.5063434017105946e-05,
      "loss": 0.1445,
      "num_input_tokens_seen": 10736952,
      "step": 16460
    },
    {
      "epoch": 8.629454926624739,
      "grad_norm": 4.922219276428223,
      "learning_rate": 3.505296597978795e-05,
      "loss": 0.1417,
      "num_input_tokens_seen": 10740024,
      "step": 16465
    },
    {
      "epoch": 8.632075471698114,
      "grad_norm": 1.7816259860992432,
      "learning_rate": 3.504249583950417e-05,
      "loss": 0.152,
      "num_input_tokens_seen": 10742680,
      "step": 16470
    },
    {
      "epoch": 8.634696016771489,
      "grad_norm": 1.6113933324813843,
      "learning_rate": 3.503202359844483e-05,
      "loss": 0.2069,
      "num_input_tokens_seen": 10746904,
      "step": 16475
    },
    {
      "epoch": 8.637316561844864,
      "grad_norm": 2.7270307540893555,
      "learning_rate": 3.50215492588006e-05,
      "loss": 0.1656,
      "num_input_tokens_seen": 10749656,
      "step": 16480
    },
    {
      "epoch": 8.63993710691824,
      "grad_norm": 2.9818520545959473,
      "learning_rate": 3.5011072822762605e-05,
      "loss": 0.3275,
      "num_input_tokens_seen": 10752280,
      "step": 16485
    },
    {
      "epoch": 8.642557651991615,
      "grad_norm": 1.6502419710159302,
      "learning_rate": 3.500059429252238e-05,
      "loss": 0.1325,
      "num_input_tokens_seen": 10755672,
      "step": 16490
    },
    {
      "epoch": 8.64517819706499,
      "grad_norm": 2.06479811668396,
      "learning_rate": 3.499011367027193e-05,
      "loss": 0.1741,
      "num_input_tokens_seen": 10759256,
      "step": 16495
    },
    {
      "epoch": 8.647798742138365,
      "grad_norm": 5.615596294403076,
      "learning_rate": 3.4979630958203665e-05,
      "loss": 0.1995,
      "num_input_tokens_seen": 10762840,
      "step": 16500
    },
    {
      "epoch": 8.65041928721174,
      "grad_norm": 3.10268497467041,
      "learning_rate": 3.496914615851046e-05,
      "loss": 0.1317,
      "num_input_tokens_seen": 10765592,
      "step": 16505
    },
    {
      "epoch": 8.653039832285115,
      "grad_norm": 2.9172141551971436,
      "learning_rate": 3.495865927338562e-05,
      "loss": 0.1137,
      "num_input_tokens_seen": 10767800,
      "step": 16510
    },
    {
      "epoch": 8.65566037735849,
      "grad_norm": 2.144268751144409,
      "learning_rate": 3.494817030502286e-05,
      "loss": 0.2051,
      "num_input_tokens_seen": 10775384,
      "step": 16515
    },
    {
      "epoch": 8.658280922431866,
      "grad_norm": 2.7620742321014404,
      "learning_rate": 3.493767925561638e-05,
      "loss": 0.253,
      "num_input_tokens_seen": 10777752,
      "step": 16520
    },
    {
      "epoch": 8.66090146750524,
      "grad_norm": 1.888041615486145,
      "learning_rate": 3.492718612736077e-05,
      "loss": 0.1641,
      "num_input_tokens_seen": 10780376,
      "step": 16525
    },
    {
      "epoch": 8.663522012578616,
      "grad_norm": 2.8630447387695312,
      "learning_rate": 3.491669092245108e-05,
      "loss": 0.1371,
      "num_input_tokens_seen": 10784344,
      "step": 16530
    },
    {
      "epoch": 8.666142557651991,
      "grad_norm": 2.720313549041748,
      "learning_rate": 3.4906193643082784e-05,
      "loss": 0.2607,
      "num_input_tokens_seen": 10787128,
      "step": 16535
    },
    {
      "epoch": 8.668763102725366,
      "grad_norm": 2.134101629257202,
      "learning_rate": 3.4895694291451795e-05,
      "loss": 0.165,
      "num_input_tokens_seen": 10790296,
      "step": 16540
    },
    {
      "epoch": 8.671383647798741,
      "grad_norm": 3.1846606731414795,
      "learning_rate": 3.4885192869754444e-05,
      "loss": 0.194,
      "num_input_tokens_seen": 10793144,
      "step": 16545
    },
    {
      "epoch": 8.674004192872118,
      "grad_norm": 6.911776065826416,
      "learning_rate": 3.4874689380187537e-05,
      "loss": 0.1932,
      "num_input_tokens_seen": 10796280,
      "step": 16550
    },
    {
      "epoch": 8.676624737945493,
      "grad_norm": 1.8955141305923462,
      "learning_rate": 3.486418382494826e-05,
      "loss": 0.163,
      "num_input_tokens_seen": 10799000,
      "step": 16555
    },
    {
      "epoch": 8.679245283018869,
      "grad_norm": 1.722265362739563,
      "learning_rate": 3.485367620623426e-05,
      "loss": 0.2082,
      "num_input_tokens_seen": 10802264,
      "step": 16560
    },
    {
      "epoch": 8.681865828092244,
      "grad_norm": 1.765910029411316,
      "learning_rate": 3.484316652624362e-05,
      "loss": 0.1837,
      "num_input_tokens_seen": 10805656,
      "step": 16565
    },
    {
      "epoch": 8.684486373165619,
      "grad_norm": 1.8916513919830322,
      "learning_rate": 3.483265478717484e-05,
      "loss": 0.1838,
      "num_input_tokens_seen": 10809496,
      "step": 16570
    },
    {
      "epoch": 8.687106918238994,
      "grad_norm": 1.8461982011795044,
      "learning_rate": 3.482214099122686e-05,
      "loss": 0.1752,
      "num_input_tokens_seen": 10812856,
      "step": 16575
    },
    {
      "epoch": 8.68972746331237,
      "grad_norm": 2.0245840549468994,
      "learning_rate": 3.481162514059903e-05,
      "loss": 0.1608,
      "num_input_tokens_seen": 10815544,
      "step": 16580
    },
    {
      "epoch": 8.692348008385745,
      "grad_norm": 2.2748146057128906,
      "learning_rate": 3.480110723749116e-05,
      "loss": 0.2019,
      "num_input_tokens_seen": 10818072,
      "step": 16585
    },
    {
      "epoch": 8.69496855345912,
      "grad_norm": 2.007603168487549,
      "learning_rate": 3.479058728410348e-05,
      "loss": 0.1629,
      "num_input_tokens_seen": 10821304,
      "step": 16590
    },
    {
      "epoch": 8.697589098532495,
      "grad_norm": 1.3765827417373657,
      "learning_rate": 3.478006528263663e-05,
      "loss": 0.1139,
      "num_input_tokens_seen": 10824984,
      "step": 16595
    },
    {
      "epoch": 8.70020964360587,
      "grad_norm": 1.9602636098861694,
      "learning_rate": 3.4769541235291704e-05,
      "loss": 0.2919,
      "num_input_tokens_seen": 10828024,
      "step": 16600
    },
    {
      "epoch": 8.702830188679245,
      "grad_norm": 3.205810070037842,
      "learning_rate": 3.475901514427021e-05,
      "loss": 0.2003,
      "num_input_tokens_seen": 10831640,
      "step": 16605
    },
    {
      "epoch": 8.70545073375262,
      "grad_norm": 1.4054512977600098,
      "learning_rate": 3.4748487011774075e-05,
      "loss": 0.1821,
      "num_input_tokens_seen": 10835160,
      "step": 16610
    },
    {
      "epoch": 8.708071278825996,
      "grad_norm": 2.156183958053589,
      "learning_rate": 3.473795684000569e-05,
      "loss": 0.1311,
      "num_input_tokens_seen": 10838232,
      "step": 16615
    },
    {
      "epoch": 8.71069182389937,
      "grad_norm": 3.6700611114501953,
      "learning_rate": 3.4727424631167815e-05,
      "loss": 0.1397,
      "num_input_tokens_seen": 10841240,
      "step": 16620
    },
    {
      "epoch": 8.713312368972746,
      "grad_norm": 1.975943922996521,
      "learning_rate": 3.471689038746369e-05,
      "loss": 0.1615,
      "num_input_tokens_seen": 10844312,
      "step": 16625
    },
    {
      "epoch": 8.715932914046121,
      "grad_norm": 2.42331862449646,
      "learning_rate": 3.470635411109695e-05,
      "loss": 0.1442,
      "num_input_tokens_seen": 10847320,
      "step": 16630
    },
    {
      "epoch": 8.718553459119496,
      "grad_norm": 1.4382776021957397,
      "learning_rate": 3.469581580427167e-05,
      "loss": 0.1549,
      "num_input_tokens_seen": 10850648,
      "step": 16635
    },
    {
      "epoch": 8.721174004192871,
      "grad_norm": 1.6527665853500366,
      "learning_rate": 3.4685275469192335e-05,
      "loss": 0.206,
      "num_input_tokens_seen": 10853912,
      "step": 16640
    },
    {
      "epoch": 8.723794549266248,
      "grad_norm": 2.0506768226623535,
      "learning_rate": 3.467473310806387e-05,
      "loss": 0.1574,
      "num_input_tokens_seen": 10856824,
      "step": 16645
    },
    {
      "epoch": 8.726415094339622,
      "grad_norm": 1.9088712930679321,
      "learning_rate": 3.466418872309161e-05,
      "loss": 0.1299,
      "num_input_tokens_seen": 10860280,
      "step": 16650
    },
    {
      "epoch": 8.729035639412999,
      "grad_norm": 1.7402173280715942,
      "learning_rate": 3.465364231648133e-05,
      "loss": 0.2203,
      "num_input_tokens_seen": 10863832,
      "step": 16655
    },
    {
      "epoch": 8.731656184486374,
      "grad_norm": 1.5259615182876587,
      "learning_rate": 3.464309389043921e-05,
      "loss": 0.1834,
      "num_input_tokens_seen": 10867576,
      "step": 16660
    },
    {
      "epoch": 8.734276729559749,
      "grad_norm": 1.758773922920227,
      "learning_rate": 3.4632543447171856e-05,
      "loss": 0.1584,
      "num_input_tokens_seen": 10871384,
      "step": 16665
    },
    {
      "epoch": 8.736897274633124,
      "grad_norm": 3.335587501525879,
      "learning_rate": 3.4621990988886315e-05,
      "loss": 0.2669,
      "num_input_tokens_seen": 10874584,
      "step": 16670
    },
    {
      "epoch": 8.7395178197065,
      "grad_norm": 1.3944416046142578,
      "learning_rate": 3.461143651779003e-05,
      "loss": 0.1528,
      "num_input_tokens_seen": 10878584,
      "step": 16675
    },
    {
      "epoch": 8.742138364779874,
      "grad_norm": 3.6720640659332275,
      "learning_rate": 3.4600880036090876e-05,
      "loss": 0.3508,
      "num_input_tokens_seen": 10881560,
      "step": 16680
    },
    {
      "epoch": 8.74475890985325,
      "grad_norm": 2.7533810138702393,
      "learning_rate": 3.459032154599715e-05,
      "loss": 0.2039,
      "num_input_tokens_seen": 10884600,
      "step": 16685
    },
    {
      "epoch": 8.747379454926625,
      "grad_norm": 1.8167364597320557,
      "learning_rate": 3.457976104971757e-05,
      "loss": 0.1406,
      "num_input_tokens_seen": 10888408,
      "step": 16690
    },
    {
      "epoch": 8.75,
      "grad_norm": 3.9501020908355713,
      "learning_rate": 3.4569198549461276e-05,
      "loss": 0.1908,
      "num_input_tokens_seen": 10891736,
      "step": 16695
    },
    {
      "epoch": 8.752620545073375,
      "grad_norm": 2.7442216873168945,
      "learning_rate": 3.455863404743781e-05,
      "loss": 0.0924,
      "num_input_tokens_seen": 10894648,
      "step": 16700
    },
    {
      "epoch": 8.75524109014675,
      "grad_norm": 1.3942025899887085,
      "learning_rate": 3.454806754585716e-05,
      "loss": 0.1174,
      "num_input_tokens_seen": 10897624,
      "step": 16705
    },
    {
      "epoch": 8.757861635220126,
      "grad_norm": 2.746417760848999,
      "learning_rate": 3.453749904692971e-05,
      "loss": 0.2053,
      "num_input_tokens_seen": 10900696,
      "step": 16710
    },
    {
      "epoch": 8.7604821802935,
      "grad_norm": 2.6897428035736084,
      "learning_rate": 3.452692855286626e-05,
      "loss": 0.2904,
      "num_input_tokens_seen": 10904280,
      "step": 16715
    },
    {
      "epoch": 8.763102725366876,
      "grad_norm": 1.3280359506607056,
      "learning_rate": 3.4516356065878055e-05,
      "loss": 0.1469,
      "num_input_tokens_seen": 10907896,
      "step": 16720
    },
    {
      "epoch": 8.765723270440251,
      "grad_norm": 2.3572680950164795,
      "learning_rate": 3.450578158817671e-05,
      "loss": 0.1591,
      "num_input_tokens_seen": 10911192,
      "step": 16725
    },
    {
      "epoch": 8.768343815513626,
      "grad_norm": 4.02177619934082,
      "learning_rate": 3.4495205121974325e-05,
      "loss": 0.2224,
      "num_input_tokens_seen": 10914488,
      "step": 16730
    },
    {
      "epoch": 8.770964360587001,
      "grad_norm": 2.8799967765808105,
      "learning_rate": 3.448462666948335e-05,
      "loss": 0.1823,
      "num_input_tokens_seen": 10918456,
      "step": 16735
    },
    {
      "epoch": 8.773584905660378,
      "grad_norm": 1.5609261989593506,
      "learning_rate": 3.4474046232916674e-05,
      "loss": 0.2177,
      "num_input_tokens_seen": 10922296,
      "step": 16740
    },
    {
      "epoch": 8.776205450733752,
      "grad_norm": 1.7708169221878052,
      "learning_rate": 3.4463463814487616e-05,
      "loss": 0.277,
      "num_input_tokens_seen": 10925400,
      "step": 16745
    },
    {
      "epoch": 8.778825995807129,
      "grad_norm": 2.565124750137329,
      "learning_rate": 3.44528794164099e-05,
      "loss": 0.1268,
      "num_input_tokens_seen": 10928120,
      "step": 16750
    },
    {
      "epoch": 8.781446540880504,
      "grad_norm": 3.92635178565979,
      "learning_rate": 3.4442293040897646e-05,
      "loss": 0.1774,
      "num_input_tokens_seen": 10932088,
      "step": 16755
    },
    {
      "epoch": 8.784067085953879,
      "grad_norm": 4.481942176818848,
      "learning_rate": 3.443170469016542e-05,
      "loss": 0.1753,
      "num_input_tokens_seen": 10935832,
      "step": 16760
    },
    {
      "epoch": 8.786687631027254,
      "grad_norm": 2.602611541748047,
      "learning_rate": 3.442111436642817e-05,
      "loss": 0.246,
      "num_input_tokens_seen": 10938552,
      "step": 16765
    },
    {
      "epoch": 8.78930817610063,
      "grad_norm": 1.8048266172409058,
      "learning_rate": 3.441052207190127e-05,
      "loss": 0.1928,
      "num_input_tokens_seen": 10942520,
      "step": 16770
    },
    {
      "epoch": 8.791928721174004,
      "grad_norm": 2.164816379547119,
      "learning_rate": 3.4399927808800526e-05,
      "loss": 0.1871,
      "num_input_tokens_seen": 10946008,
      "step": 16775
    },
    {
      "epoch": 8.79454926624738,
      "grad_norm": 2.1007909774780273,
      "learning_rate": 3.438933157934211e-05,
      "loss": 0.1893,
      "num_input_tokens_seen": 10949144,
      "step": 16780
    },
    {
      "epoch": 8.797169811320755,
      "grad_norm": 1.6794023513793945,
      "learning_rate": 3.4378733385742653e-05,
      "loss": 0.1498,
      "num_input_tokens_seen": 10953400,
      "step": 16785
    },
    {
      "epoch": 8.79979035639413,
      "grad_norm": 2.2920873165130615,
      "learning_rate": 3.436813323021918e-05,
      "loss": 0.1864,
      "num_input_tokens_seen": 10956472,
      "step": 16790
    },
    {
      "epoch": 8.802410901467505,
      "grad_norm": 2.404066801071167,
      "learning_rate": 3.43575311149891e-05,
      "loss": 0.2064,
      "num_input_tokens_seen": 10960312,
      "step": 16795
    },
    {
      "epoch": 8.80503144654088,
      "grad_norm": 1.9260566234588623,
      "learning_rate": 3.434692704227028e-05,
      "loss": 0.1578,
      "num_input_tokens_seen": 10962616,
      "step": 16800
    },
    {
      "epoch": 8.807651991614255,
      "grad_norm": 1.4109362363815308,
      "learning_rate": 3.4336321014280944e-05,
      "loss": 0.1405,
      "num_input_tokens_seen": 10966008,
      "step": 16805
    },
    {
      "epoch": 8.81027253668763,
      "grad_norm": 1.5579795837402344,
      "learning_rate": 3.432571303323977e-05,
      "loss": 0.2034,
      "num_input_tokens_seen": 10970040,
      "step": 16810
    },
    {
      "epoch": 8.812893081761006,
      "grad_norm": 2.1191866397857666,
      "learning_rate": 3.431510310136583e-05,
      "loss": 0.2079,
      "num_input_tokens_seen": 10972472,
      "step": 16815
    },
    {
      "epoch": 8.815513626834381,
      "grad_norm": 1.2393865585327148,
      "learning_rate": 3.4304491220878585e-05,
      "loss": 0.1494,
      "num_input_tokens_seen": 10976664,
      "step": 16820
    },
    {
      "epoch": 8.818134171907756,
      "grad_norm": 1.4207502603530884,
      "learning_rate": 3.429387739399793e-05,
      "loss": 0.1254,
      "num_input_tokens_seen": 10979192,
      "step": 16825
    },
    {
      "epoch": 8.820754716981131,
      "grad_norm": 3.305389165878296,
      "learning_rate": 3.428326162294416e-05,
      "loss": 0.1848,
      "num_input_tokens_seen": 10981688,
      "step": 16830
    },
    {
      "epoch": 8.823375262054507,
      "grad_norm": 2.259753942489624,
      "learning_rate": 3.427264390993795e-05,
      "loss": 0.216,
      "num_input_tokens_seen": 10985176,
      "step": 16835
    },
    {
      "epoch": 8.825995807127882,
      "grad_norm": 2.7835254669189453,
      "learning_rate": 3.426202425720044e-05,
      "loss": 0.1722,
      "num_input_tokens_seen": 10988536,
      "step": 16840
    },
    {
      "epoch": 8.828616352201259,
      "grad_norm": 3.02046537399292,
      "learning_rate": 3.425140266695312e-05,
      "loss": 0.1607,
      "num_input_tokens_seen": 10990936,
      "step": 16845
    },
    {
      "epoch": 8.831236897274634,
      "grad_norm": 2.216434955596924,
      "learning_rate": 3.4240779141417905e-05,
      "loss": 0.1892,
      "num_input_tokens_seen": 10994200,
      "step": 16850
    },
    {
      "epoch": 8.833857442348009,
      "grad_norm": 2.1403582096099854,
      "learning_rate": 3.423015368281711e-05,
      "loss": 0.1691,
      "num_input_tokens_seen": 10997240,
      "step": 16855
    },
    {
      "epoch": 8.836477987421384,
      "grad_norm": 1.9995548725128174,
      "learning_rate": 3.421952629337348e-05,
      "loss": 0.1596,
      "num_input_tokens_seen": 11000568,
      "step": 16860
    },
    {
      "epoch": 8.83909853249476,
      "grad_norm": 2.103300094604492,
      "learning_rate": 3.420889697531013e-05,
      "loss": 0.1453,
      "num_input_tokens_seen": 11005528,
      "step": 16865
    },
    {
      "epoch": 8.841719077568134,
      "grad_norm": 3.268489122390747,
      "learning_rate": 3.419826573085058e-05,
      "loss": 0.1872,
      "num_input_tokens_seen": 11009176,
      "step": 16870
    },
    {
      "epoch": 8.84433962264151,
      "grad_norm": 1.2854514122009277,
      "learning_rate": 3.418763256221879e-05,
      "loss": 0.1374,
      "num_input_tokens_seen": 11012504,
      "step": 16875
    },
    {
      "epoch": 8.846960167714885,
      "grad_norm": 2.0313193798065186,
      "learning_rate": 3.4176997471639085e-05,
      "loss": 0.1983,
      "num_input_tokens_seen": 11015992,
      "step": 16880
    },
    {
      "epoch": 8.84958071278826,
      "grad_norm": 1.1434868574142456,
      "learning_rate": 3.4166360461336206e-05,
      "loss": 0.1477,
      "num_input_tokens_seen": 11019928,
      "step": 16885
    },
    {
      "epoch": 8.852201257861635,
      "grad_norm": 1.5065962076187134,
      "learning_rate": 3.41557215335353e-05,
      "loss": 0.2261,
      "num_input_tokens_seen": 11023096,
      "step": 16890
    },
    {
      "epoch": 8.85482180293501,
      "grad_norm": 6.21032190322876,
      "learning_rate": 3.4145080690461887e-05,
      "loss": 0.2248,
      "num_input_tokens_seen": 11026232,
      "step": 16895
    },
    {
      "epoch": 8.857442348008385,
      "grad_norm": 3.544980049133301,
      "learning_rate": 3.4134437934341936e-05,
      "loss": 0.2681,
      "num_input_tokens_seen": 11029560,
      "step": 16900
    },
    {
      "epoch": 8.86006289308176,
      "grad_norm": 2.611959218978882,
      "learning_rate": 3.4123793267401784e-05,
      "loss": 0.163,
      "num_input_tokens_seen": 11032120,
      "step": 16905
    },
    {
      "epoch": 8.862683438155136,
      "grad_norm": 3.1268351078033447,
      "learning_rate": 3.411314669186816e-05,
      "loss": 0.2406,
      "num_input_tokens_seen": 11034968,
      "step": 16910
    },
    {
      "epoch": 8.865303983228511,
      "grad_norm": 1.463666319847107,
      "learning_rate": 3.410249820996823e-05,
      "loss": 0.1345,
      "num_input_tokens_seen": 11038456,
      "step": 16915
    },
    {
      "epoch": 8.867924528301886,
      "grad_norm": 3.017024278640747,
      "learning_rate": 3.4091847823929513e-05,
      "loss": 0.1999,
      "num_input_tokens_seen": 11042712,
      "step": 16920
    },
    {
      "epoch": 8.870545073375261,
      "grad_norm": 2.12963604927063,
      "learning_rate": 3.4081195535979946e-05,
      "loss": 0.2758,
      "num_input_tokens_seen": 11046008,
      "step": 16925
    },
    {
      "epoch": 8.873165618448636,
      "grad_norm": 5.529512882232666,
      "learning_rate": 3.40705413483479e-05,
      "loss": 0.2638,
      "num_input_tokens_seen": 11049336,
      "step": 16930
    },
    {
      "epoch": 8.875786163522012,
      "grad_norm": 2.3776028156280518,
      "learning_rate": 3.405988526326207e-05,
      "loss": 0.1708,
      "num_input_tokens_seen": 11053272,
      "step": 16935
    },
    {
      "epoch": 8.878406708595389,
      "grad_norm": 2.09501576423645,
      "learning_rate": 3.40492272829516e-05,
      "loss": 0.1768,
      "num_input_tokens_seen": 11057848,
      "step": 16940
    },
    {
      "epoch": 8.881027253668764,
      "grad_norm": 2.718872308731079,
      "learning_rate": 3.403856740964604e-05,
      "loss": 0.2449,
      "num_input_tokens_seen": 11060216,
      "step": 16945
    },
    {
      "epoch": 8.883647798742139,
      "grad_norm": 3.3216001987457275,
      "learning_rate": 3.4027905645575275e-05,
      "loss": 0.2054,
      "num_input_tokens_seen": 11063640,
      "step": 16950
    },
    {
      "epoch": 8.886268343815514,
      "grad_norm": 2.0116755962371826,
      "learning_rate": 3.401724199296966e-05,
      "loss": 0.1681,
      "num_input_tokens_seen": 11066872,
      "step": 16955
    },
    {
      "epoch": 8.88888888888889,
      "grad_norm": 2.1407673358917236,
      "learning_rate": 3.400657645405989e-05,
      "loss": 0.2103,
      "num_input_tokens_seen": 11069624,
      "step": 16960
    },
    {
      "epoch": 8.891509433962264,
      "grad_norm": 1.514816403388977,
      "learning_rate": 3.3995909031077065e-05,
      "loss": 0.1606,
      "num_input_tokens_seen": 11073048,
      "step": 16965
    },
    {
      "epoch": 8.89412997903564,
      "grad_norm": 1.1645246744155884,
      "learning_rate": 3.398523972625272e-05,
      "loss": 0.1542,
      "num_input_tokens_seen": 11076312,
      "step": 16970
    },
    {
      "epoch": 8.896750524109015,
      "grad_norm": 2.3308982849121094,
      "learning_rate": 3.3974568541818723e-05,
      "loss": 0.2102,
      "num_input_tokens_seen": 11079416,
      "step": 16975
    },
    {
      "epoch": 8.89937106918239,
      "grad_norm": 3.1289496421813965,
      "learning_rate": 3.396389548000738e-05,
      "loss": 0.1788,
      "num_input_tokens_seen": 11082168,
      "step": 16980
    },
    {
      "epoch": 8.901991614255765,
      "grad_norm": 3.0176351070404053,
      "learning_rate": 3.395322054305137e-05,
      "loss": 0.1228,
      "num_input_tokens_seen": 11084568,
      "step": 16985
    },
    {
      "epoch": 8.90461215932914,
      "grad_norm": 1.1951326131820679,
      "learning_rate": 3.394254373318376e-05,
      "loss": 0.1457,
      "num_input_tokens_seen": 11087736,
      "step": 16990
    },
    {
      "epoch": 8.907232704402515,
      "grad_norm": 1.8347675800323486,
      "learning_rate": 3.3931865052638026e-05,
      "loss": 0.2345,
      "num_input_tokens_seen": 11090648,
      "step": 16995
    },
    {
      "epoch": 8.90985324947589,
      "grad_norm": 5.383452415466309,
      "learning_rate": 3.392118450364802e-05,
      "loss": 0.2559,
      "num_input_tokens_seen": 11093240,
      "step": 17000
    },
    {
      "epoch": 8.912473794549266,
      "grad_norm": 2.7847094535827637,
      "learning_rate": 3.391050208844799e-05,
      "loss": 0.2117,
      "num_input_tokens_seen": 11096568,
      "step": 17005
    },
    {
      "epoch": 8.915094339622641,
      "grad_norm": 1.2303578853607178,
      "learning_rate": 3.389981780927259e-05,
      "loss": 0.1732,
      "num_input_tokens_seen": 11099384,
      "step": 17010
    },
    {
      "epoch": 8.917714884696016,
      "grad_norm": 3.3573267459869385,
      "learning_rate": 3.3889131668356834e-05,
      "loss": 0.1341,
      "num_input_tokens_seen": 11101944,
      "step": 17015
    },
    {
      "epoch": 8.920335429769391,
      "grad_norm": 3.3958067893981934,
      "learning_rate": 3.387844366793614e-05,
      "loss": 0.2842,
      "num_input_tokens_seen": 11104856,
      "step": 17020
    },
    {
      "epoch": 8.922955974842766,
      "grad_norm": 1.9361956119537354,
      "learning_rate": 3.3867753810246326e-05,
      "loss": 0.168,
      "num_input_tokens_seen": 11108440,
      "step": 17025
    },
    {
      "epoch": 8.925576519916142,
      "grad_norm": 3.558396339416504,
      "learning_rate": 3.3857062097523586e-05,
      "loss": 0.1832,
      "num_input_tokens_seen": 11112504,
      "step": 17030
    },
    {
      "epoch": 8.928197064989519,
      "grad_norm": 2.467500686645508,
      "learning_rate": 3.38463685320045e-05,
      "loss": 0.2378,
      "num_input_tokens_seen": 11116024,
      "step": 17035
    },
    {
      "epoch": 8.930817610062894,
      "grad_norm": 2.6376051902770996,
      "learning_rate": 3.3835673115926045e-05,
      "loss": 0.1949,
      "num_input_tokens_seen": 11119128,
      "step": 17040
    },
    {
      "epoch": 8.933438155136269,
      "grad_norm": 2.1148924827575684,
      "learning_rate": 3.382497585152557e-05,
      "loss": 0.2673,
      "num_input_tokens_seen": 11122584,
      "step": 17045
    },
    {
      "epoch": 8.936058700209644,
      "grad_norm": 1.460496425628662,
      "learning_rate": 3.3814276741040835e-05,
      "loss": 0.1575,
      "num_input_tokens_seen": 11125592,
      "step": 17050
    },
    {
      "epoch": 8.93867924528302,
      "grad_norm": 1.4431931972503662,
      "learning_rate": 3.380357578670996e-05,
      "loss": 0.1831,
      "num_input_tokens_seen": 11128312,
      "step": 17055
    },
    {
      "epoch": 8.941299790356394,
      "grad_norm": 2.0296051502227783,
      "learning_rate": 3.379287299077146e-05,
      "loss": 0.1691,
      "num_input_tokens_seen": 11132120,
      "step": 17060
    },
    {
      "epoch": 8.94392033542977,
      "grad_norm": 1.9733715057373047,
      "learning_rate": 3.3782168355464263e-05,
      "loss": 0.1738,
      "num_input_tokens_seen": 11135768,
      "step": 17065
    },
    {
      "epoch": 8.946540880503145,
      "grad_norm": 3.5197789669036865,
      "learning_rate": 3.377146188302762e-05,
      "loss": 0.1596,
      "num_input_tokens_seen": 11139128,
      "step": 17070
    },
    {
      "epoch": 8.94916142557652,
      "grad_norm": 2.2498207092285156,
      "learning_rate": 3.376075357570124e-05,
      "loss": 0.3164,
      "num_input_tokens_seen": 11142360,
      "step": 17075
    },
    {
      "epoch": 8.951781970649895,
      "grad_norm": 2.716498851776123,
      "learning_rate": 3.3750043435725144e-05,
      "loss": 0.1528,
      "num_input_tokens_seen": 11144984,
      "step": 17080
    },
    {
      "epoch": 8.95440251572327,
      "grad_norm": 5.333407402038574,
      "learning_rate": 3.373933146533979e-05,
      "loss": 0.1844,
      "num_input_tokens_seen": 11147672,
      "step": 17085
    },
    {
      "epoch": 8.957023060796645,
      "grad_norm": 3.6144089698791504,
      "learning_rate": 3.3728617666786e-05,
      "loss": 0.154,
      "num_input_tokens_seen": 11150872,
      "step": 17090
    },
    {
      "epoch": 8.95964360587002,
      "grad_norm": 2.0129168033599854,
      "learning_rate": 3.371790204230497e-05,
      "loss": 0.1833,
      "num_input_tokens_seen": 11154360,
      "step": 17095
    },
    {
      "epoch": 8.962264150943396,
      "grad_norm": 2.2981743812561035,
      "learning_rate": 3.3707184594138295e-05,
      "loss": 0.1746,
      "num_input_tokens_seen": 11156984,
      "step": 17100
    },
    {
      "epoch": 8.964884696016771,
      "grad_norm": 2.827409029006958,
      "learning_rate": 3.369646532452795e-05,
      "loss": 0.3447,
      "num_input_tokens_seen": 11160696,
      "step": 17105
    },
    {
      "epoch": 8.967505241090146,
      "grad_norm": 1.647047519683838,
      "learning_rate": 3.3685744235716254e-05,
      "loss": 0.2638,
      "num_input_tokens_seen": 11163800,
      "step": 17110
    },
    {
      "epoch": 8.970125786163521,
      "grad_norm": 2.038341999053955,
      "learning_rate": 3.367502132994598e-05,
      "loss": 0.2425,
      "num_input_tokens_seen": 11167480,
      "step": 17115
    },
    {
      "epoch": 8.972746331236896,
      "grad_norm": 5.310999870300293,
      "learning_rate": 3.366429660946019e-05,
      "loss": 0.1823,
      "num_input_tokens_seen": 11170232,
      "step": 17120
    },
    {
      "epoch": 8.975366876310272,
      "grad_norm": 3.7630577087402344,
      "learning_rate": 3.365357007650241e-05,
      "loss": 0.242,
      "num_input_tokens_seen": 11173400,
      "step": 17125
    },
    {
      "epoch": 8.977987421383649,
      "grad_norm": 2.206289291381836,
      "learning_rate": 3.364284173331649e-05,
      "loss": 0.2117,
      "num_input_tokens_seen": 11176472,
      "step": 17130
    },
    {
      "epoch": 8.980607966457024,
      "grad_norm": 1.4065853357315063,
      "learning_rate": 3.3632111582146686e-05,
      "loss": 0.1671,
      "num_input_tokens_seen": 11179288,
      "step": 17135
    },
    {
      "epoch": 8.983228511530399,
      "grad_norm": 3.935251235961914,
      "learning_rate": 3.362137962523761e-05,
      "loss": 0.1808,
      "num_input_tokens_seen": 11182136,
      "step": 17140
    },
    {
      "epoch": 8.985849056603774,
      "grad_norm": 1.9545191526412964,
      "learning_rate": 3.361064586483429e-05,
      "loss": 0.154,
      "num_input_tokens_seen": 11186296,
      "step": 17145
    },
    {
      "epoch": 8.98846960167715,
      "grad_norm": 1.419445276260376,
      "learning_rate": 3.359991030318208e-05,
      "loss": 0.1099,
      "num_input_tokens_seen": 11189304,
      "step": 17150
    },
    {
      "epoch": 8.991090146750524,
      "grad_norm": 2.074974536895752,
      "learning_rate": 3.358917294252675e-05,
      "loss": 0.1912,
      "num_input_tokens_seen": 11192024,
      "step": 17155
    },
    {
      "epoch": 8.9937106918239,
      "grad_norm": 2.120065450668335,
      "learning_rate": 3.357843378511443e-05,
      "loss": 0.2296,
      "num_input_tokens_seen": 11195160,
      "step": 17160
    },
    {
      "epoch": 8.996331236897275,
      "grad_norm": 2.39567232131958,
      "learning_rate": 3.3567692833191635e-05,
      "loss": 0.2184,
      "num_input_tokens_seen": 11198264,
      "step": 17165
    },
    {
      "epoch": 8.99895178197065,
      "grad_norm": 1.6834229230880737,
      "learning_rate": 3.355695008900525e-05,
      "loss": 0.2353,
      "num_input_tokens_seen": 11202840,
      "step": 17170
    },
    {
      "epoch": 9.0,
      "eval_loss": 0.6834951639175415,
      "eval_runtime": 11.9579,
      "eval_samples_per_second": 70.915,
      "eval_steps_per_second": 17.729,
      "num_input_tokens_seen": 11204000,
      "step": 17172
    },
    {
      "epoch": 9.001572327044025,
      "grad_norm": 2.250095844268799,
      "learning_rate": 3.354620555480251e-05,
      "loss": 0.1576,
      "num_input_tokens_seen": 11206048,
      "step": 17175
    },
    {
      "epoch": 9.0041928721174,
      "grad_norm": 1.3801934719085693,
      "learning_rate": 3.353545923283108e-05,
      "loss": 0.2278,
      "num_input_tokens_seen": 11210464,
      "step": 17180
    },
    {
      "epoch": 9.006813417190775,
      "grad_norm": 2.1393492221832275,
      "learning_rate": 3.352471112533895e-05,
      "loss": 0.1256,
      "num_input_tokens_seen": 11213312,
      "step": 17185
    },
    {
      "epoch": 9.00943396226415,
      "grad_norm": 1.5144145488739014,
      "learning_rate": 3.351396123457451e-05,
      "loss": 0.1172,
      "num_input_tokens_seen": 11216704,
      "step": 17190
    },
    {
      "epoch": 9.012054507337526,
      "grad_norm": 1.599195122718811,
      "learning_rate": 3.350320956278652e-05,
      "loss": 0.1312,
      "num_input_tokens_seen": 11221248,
      "step": 17195
    },
    {
      "epoch": 9.014675052410901,
      "grad_norm": 1.2620735168457031,
      "learning_rate": 3.349245611222409e-05,
      "loss": 0.1394,
      "num_input_tokens_seen": 11223904,
      "step": 17200
    },
    {
      "epoch": 9.017295597484276,
      "grad_norm": 1.548994779586792,
      "learning_rate": 3.348170088513673e-05,
      "loss": 0.1035,
      "num_input_tokens_seen": 11227648,
      "step": 17205
    },
    {
      "epoch": 9.019916142557651,
      "grad_norm": 2.1891653537750244,
      "learning_rate": 3.347094388377431e-05,
      "loss": 0.0998,
      "num_input_tokens_seen": 11230432,
      "step": 17210
    },
    {
      "epoch": 9.022536687631026,
      "grad_norm": 3.0112144947052,
      "learning_rate": 3.3460185110387076e-05,
      "loss": 0.1226,
      "num_input_tokens_seen": 11233536,
      "step": 17215
    },
    {
      "epoch": 9.025157232704403,
      "grad_norm": 3.091545581817627,
      "learning_rate": 3.344942456722563e-05,
      "loss": 0.1783,
      "num_input_tokens_seen": 11236512,
      "step": 17220
    },
    {
      "epoch": 9.027777777777779,
      "grad_norm": 3.107701301574707,
      "learning_rate": 3.343866225654096e-05,
      "loss": 0.1027,
      "num_input_tokens_seen": 11238656,
      "step": 17225
    },
    {
      "epoch": 9.030398322851154,
      "grad_norm": 1.2225643396377563,
      "learning_rate": 3.342789818058443e-05,
      "loss": 0.1305,
      "num_input_tokens_seen": 11242400,
      "step": 17230
    },
    {
      "epoch": 9.033018867924529,
      "grad_norm": 2.2505760192871094,
      "learning_rate": 3.341713234160774e-05,
      "loss": 0.1497,
      "num_input_tokens_seen": 11245824,
      "step": 17235
    },
    {
      "epoch": 9.035639412997904,
      "grad_norm": 1.95954167842865,
      "learning_rate": 3.3406364741862986e-05,
      "loss": 0.0945,
      "num_input_tokens_seen": 11248736,
      "step": 17240
    },
    {
      "epoch": 9.03825995807128,
      "grad_norm": 1.6536731719970703,
      "learning_rate": 3.3395595383602644e-05,
      "loss": 0.1341,
      "num_input_tokens_seen": 11252096,
      "step": 17245
    },
    {
      "epoch": 9.040880503144654,
      "grad_norm": 1.7871136665344238,
      "learning_rate": 3.338482426907953e-05,
      "loss": 0.1186,
      "num_input_tokens_seen": 11258496,
      "step": 17250
    },
    {
      "epoch": 9.04350104821803,
      "grad_norm": 3.2837584018707275,
      "learning_rate": 3.337405140054683e-05,
      "loss": 0.1007,
      "num_input_tokens_seen": 11260960,
      "step": 17255
    },
    {
      "epoch": 9.046121593291405,
      "grad_norm": 2.846165180206299,
      "learning_rate": 3.336327678025813e-05,
      "loss": 0.1192,
      "num_input_tokens_seen": 11264512,
      "step": 17260
    },
    {
      "epoch": 9.04874213836478,
      "grad_norm": 4.694727420806885,
      "learning_rate": 3.3352500410467324e-05,
      "loss": 0.1101,
      "num_input_tokens_seen": 11268256,
      "step": 17265
    },
    {
      "epoch": 9.051362683438155,
      "grad_norm": 1.570120096206665,
      "learning_rate": 3.3341722293428726e-05,
      "loss": 0.1866,
      "num_input_tokens_seen": 11271328,
      "step": 17270
    },
    {
      "epoch": 9.05398322851153,
      "grad_norm": 3.7258214950561523,
      "learning_rate": 3.3330942431397e-05,
      "loss": 0.1024,
      "num_input_tokens_seen": 11274048,
      "step": 17275
    },
    {
      "epoch": 9.056603773584905,
      "grad_norm": 1.7970521450042725,
      "learning_rate": 3.332016082662715e-05,
      "loss": 0.1534,
      "num_input_tokens_seen": 11277440,
      "step": 17280
    },
    {
      "epoch": 9.05922431865828,
      "grad_norm": 2.7491343021392822,
      "learning_rate": 3.3309377481374584e-05,
      "loss": 0.126,
      "num_input_tokens_seen": 11281120,
      "step": 17285
    },
    {
      "epoch": 9.061844863731656,
      "grad_norm": 0.9524456262588501,
      "learning_rate": 3.3298592397895046e-05,
      "loss": 0.1072,
      "num_input_tokens_seen": 11284064,
      "step": 17290
    },
    {
      "epoch": 9.064465408805031,
      "grad_norm": 1.5025001764297485,
      "learning_rate": 3.3287805578444645e-05,
      "loss": 0.1205,
      "num_input_tokens_seen": 11288096,
      "step": 17295
    },
    {
      "epoch": 9.067085953878406,
      "grad_norm": 3.6131985187530518,
      "learning_rate": 3.327701702527989e-05,
      "loss": 0.0842,
      "num_input_tokens_seen": 11292032,
      "step": 17300
    },
    {
      "epoch": 9.069706498951781,
      "grad_norm": 1.251287817955017,
      "learning_rate": 3.3266226740657584e-05,
      "loss": 0.117,
      "num_input_tokens_seen": 11294720,
      "step": 17305
    },
    {
      "epoch": 9.072327044025156,
      "grad_norm": 2.283154249191284,
      "learning_rate": 3.3255434726834954e-05,
      "loss": 0.1698,
      "num_input_tokens_seen": 11298880,
      "step": 17310
    },
    {
      "epoch": 9.074947589098532,
      "grad_norm": 0.8344667553901672,
      "learning_rate": 3.324464098606957e-05,
      "loss": 0.2357,
      "num_input_tokens_seen": 11302336,
      "step": 17315
    },
    {
      "epoch": 9.077568134171909,
      "grad_norm": 0.7410045862197876,
      "learning_rate": 3.323384552061934e-05,
      "loss": 0.0843,
      "num_input_tokens_seen": 11305216,
      "step": 17320
    },
    {
      "epoch": 9.080188679245284,
      "grad_norm": 1.664435863494873,
      "learning_rate": 3.322304833274257e-05,
      "loss": 0.2317,
      "num_input_tokens_seen": 11308512,
      "step": 17325
    },
    {
      "epoch": 9.082809224318659,
      "grad_norm": 1.4989041090011597,
      "learning_rate": 3.32122494246979e-05,
      "loss": 0.1712,
      "num_input_tokens_seen": 11312416,
      "step": 17330
    },
    {
      "epoch": 9.085429769392034,
      "grad_norm": 1.6165950298309326,
      "learning_rate": 3.320144879874434e-05,
      "loss": 0.1136,
      "num_input_tokens_seen": 11315904,
      "step": 17335
    },
    {
      "epoch": 9.08805031446541,
      "grad_norm": 0.4345267713069916,
      "learning_rate": 3.319064645714126e-05,
      "loss": 0.1085,
      "num_input_tokens_seen": 11321664,
      "step": 17340
    },
    {
      "epoch": 9.090670859538784,
      "grad_norm": 3.138026475906372,
      "learning_rate": 3.3179842402148386e-05,
      "loss": 0.1116,
      "num_input_tokens_seen": 11325728,
      "step": 17345
    },
    {
      "epoch": 9.09329140461216,
      "grad_norm": 1.4370298385620117,
      "learning_rate": 3.316903663602581e-05,
      "loss": 0.1324,
      "num_input_tokens_seen": 11329344,
      "step": 17350
    },
    {
      "epoch": 9.095911949685535,
      "grad_norm": 1.6861659288406372,
      "learning_rate": 3.315822916103395e-05,
      "loss": 0.1064,
      "num_input_tokens_seen": 11331840,
      "step": 17355
    },
    {
      "epoch": 9.09853249475891,
      "grad_norm": 5.235038757324219,
      "learning_rate": 3.3147419979433635e-05,
      "loss": 0.1549,
      "num_input_tokens_seen": 11334912,
      "step": 17360
    },
    {
      "epoch": 9.101153039832285,
      "grad_norm": 2.228163719177246,
      "learning_rate": 3.313660909348601e-05,
      "loss": 0.0957,
      "num_input_tokens_seen": 11337344,
      "step": 17365
    },
    {
      "epoch": 9.10377358490566,
      "grad_norm": 2.289100408554077,
      "learning_rate": 3.312579650545259e-05,
      "loss": 0.1809,
      "num_input_tokens_seen": 11340384,
      "step": 17370
    },
    {
      "epoch": 9.106394129979035,
      "grad_norm": 4.2935075759887695,
      "learning_rate": 3.311498221759524e-05,
      "loss": 0.146,
      "num_input_tokens_seen": 11342976,
      "step": 17375
    },
    {
      "epoch": 9.10901467505241,
      "grad_norm": 2.464327812194824,
      "learning_rate": 3.310416623217619e-05,
      "loss": 0.1033,
      "num_input_tokens_seen": 11345504,
      "step": 17380
    },
    {
      "epoch": 9.111635220125786,
      "grad_norm": 1.544812798500061,
      "learning_rate": 3.309334855145803e-05,
      "loss": 0.1405,
      "num_input_tokens_seen": 11349216,
      "step": 17385
    },
    {
      "epoch": 9.114255765199161,
      "grad_norm": 1.5659154653549194,
      "learning_rate": 3.3082529177703684e-05,
      "loss": 0.1251,
      "num_input_tokens_seen": 11352576,
      "step": 17390
    },
    {
      "epoch": 9.116876310272536,
      "grad_norm": 2.893026828765869,
      "learning_rate": 3.3071708113176435e-05,
      "loss": 0.1283,
      "num_input_tokens_seen": 11355104,
      "step": 17395
    },
    {
      "epoch": 9.119496855345911,
      "grad_norm": 3.371645450592041,
      "learning_rate": 3.306088536013995e-05,
      "loss": 0.1879,
      "num_input_tokens_seen": 11361504,
      "step": 17400
    },
    {
      "epoch": 9.122117400419286,
      "grad_norm": 1.7482514381408691,
      "learning_rate": 3.3050060920858206e-05,
      "loss": 0.1312,
      "num_input_tokens_seen": 11365120,
      "step": 17405
    },
    {
      "epoch": 9.124737945492662,
      "grad_norm": 3.254425287246704,
      "learning_rate": 3.3039234797595544e-05,
      "loss": 0.1752,
      "num_input_tokens_seen": 11368800,
      "step": 17410
    },
    {
      "epoch": 9.127358490566039,
      "grad_norm": 3.179640054702759,
      "learning_rate": 3.3028406992616695e-05,
      "loss": 0.1842,
      "num_input_tokens_seen": 11371616,
      "step": 17415
    },
    {
      "epoch": 9.129979035639414,
      "grad_norm": 2.188624143600464,
      "learning_rate": 3.301757750818669e-05,
      "loss": 0.2093,
      "num_input_tokens_seen": 11374496,
      "step": 17420
    },
    {
      "epoch": 9.132599580712789,
      "grad_norm": 2.1402106285095215,
      "learning_rate": 3.300674634657094e-05,
      "loss": 0.12,
      "num_input_tokens_seen": 11378624,
      "step": 17425
    },
    {
      "epoch": 9.135220125786164,
      "grad_norm": 6.978676795959473,
      "learning_rate": 3.2995913510035206e-05,
      "loss": 0.165,
      "num_input_tokens_seen": 11381312,
      "step": 17430
    },
    {
      "epoch": 9.13784067085954,
      "grad_norm": 2.7986247539520264,
      "learning_rate": 3.298507900084558e-05,
      "loss": 0.0825,
      "num_input_tokens_seen": 11383808,
      "step": 17435
    },
    {
      "epoch": 9.140461215932914,
      "grad_norm": 3.772329568862915,
      "learning_rate": 3.2974242821268524e-05,
      "loss": 0.1829,
      "num_input_tokens_seen": 11387840,
      "step": 17440
    },
    {
      "epoch": 9.14308176100629,
      "grad_norm": 2.4994430541992188,
      "learning_rate": 3.2963404973570855e-05,
      "loss": 0.0956,
      "num_input_tokens_seen": 11392800,
      "step": 17445
    },
    {
      "epoch": 9.145702306079665,
      "grad_norm": 2.856337547302246,
      "learning_rate": 3.2952565460019715e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 11396224,
      "step": 17450
    },
    {
      "epoch": 9.14832285115304,
      "grad_norm": 2.805652379989624,
      "learning_rate": 3.29417242828826e-05,
      "loss": 0.1261,
      "num_input_tokens_seen": 11399456,
      "step": 17455
    },
    {
      "epoch": 9.150943396226415,
      "grad_norm": 2.049990177154541,
      "learning_rate": 3.2930881444427374e-05,
      "loss": 0.1674,
      "num_input_tokens_seen": 11402304,
      "step": 17460
    },
    {
      "epoch": 9.15356394129979,
      "grad_norm": 3.660047769546509,
      "learning_rate": 3.292003694692223e-05,
      "loss": 0.1761,
      "num_input_tokens_seen": 11405440,
      "step": 17465
    },
    {
      "epoch": 9.156184486373165,
      "grad_norm": 1.8828208446502686,
      "learning_rate": 3.290919079263572e-05,
      "loss": 0.1385,
      "num_input_tokens_seen": 11408256,
      "step": 17470
    },
    {
      "epoch": 9.15880503144654,
      "grad_norm": 1.012287974357605,
      "learning_rate": 3.2898342983836715e-05,
      "loss": 0.1226,
      "num_input_tokens_seen": 11412288,
      "step": 17475
    },
    {
      "epoch": 9.161425576519916,
      "grad_norm": 1.7503124475479126,
      "learning_rate": 3.288749352279447e-05,
      "loss": 0.11,
      "num_input_tokens_seen": 11414816,
      "step": 17480
    },
    {
      "epoch": 9.164046121593291,
      "grad_norm": 1.2833404541015625,
      "learning_rate": 3.2876642411778573e-05,
      "loss": 0.1367,
      "num_input_tokens_seen": 11417632,
      "step": 17485
    },
    {
      "epoch": 9.166666666666666,
      "grad_norm": 2.198197841644287,
      "learning_rate": 3.286578965305894e-05,
      "loss": 0.1686,
      "num_input_tokens_seen": 11420384,
      "step": 17490
    },
    {
      "epoch": 9.169287211740041,
      "grad_norm": 1.5550103187561035,
      "learning_rate": 3.285493524890585e-05,
      "loss": 0.1652,
      "num_input_tokens_seen": 11423648,
      "step": 17495
    },
    {
      "epoch": 9.171907756813416,
      "grad_norm": 1.103234052658081,
      "learning_rate": 3.284407920158992e-05,
      "loss": 0.105,
      "num_input_tokens_seen": 11426976,
      "step": 17500
    },
    {
      "epoch": 9.174528301886792,
      "grad_norm": 3.375194787979126,
      "learning_rate": 3.283322151338211e-05,
      "loss": 0.1566,
      "num_input_tokens_seen": 11430304,
      "step": 17505
    },
    {
      "epoch": 9.177148846960169,
      "grad_norm": 8.970266342163086,
      "learning_rate": 3.282236218655373e-05,
      "loss": 0.1276,
      "num_input_tokens_seen": 11432704,
      "step": 17510
    },
    {
      "epoch": 9.179769392033544,
      "grad_norm": 1.4462352991104126,
      "learning_rate": 3.281150122337642e-05,
      "loss": 0.1227,
      "num_input_tokens_seen": 11436448,
      "step": 17515
    },
    {
      "epoch": 9.182389937106919,
      "grad_norm": 2.8130991458892822,
      "learning_rate": 3.280063862612216e-05,
      "loss": 0.1227,
      "num_input_tokens_seen": 11439872,
      "step": 17520
    },
    {
      "epoch": 9.185010482180294,
      "grad_norm": 3.358525037765503,
      "learning_rate": 3.2789774397063314e-05,
      "loss": 0.1405,
      "num_input_tokens_seen": 11442304,
      "step": 17525
    },
    {
      "epoch": 9.18763102725367,
      "grad_norm": 2.851137161254883,
      "learning_rate": 3.277890853847253e-05,
      "loss": 0.2779,
      "num_input_tokens_seen": 11445856,
      "step": 17530
    },
    {
      "epoch": 9.190251572327044,
      "grad_norm": 1.6956034898757935,
      "learning_rate": 3.2768041052622824e-05,
      "loss": 0.1088,
      "num_input_tokens_seen": 11452288,
      "step": 17535
    },
    {
      "epoch": 9.19287211740042,
      "grad_norm": 1.7097283601760864,
      "learning_rate": 3.275717194178755e-05,
      "loss": 0.1987,
      "num_input_tokens_seen": 11456032,
      "step": 17540
    },
    {
      "epoch": 9.195492662473795,
      "grad_norm": 1.3275208473205566,
      "learning_rate": 3.27463012082404e-05,
      "loss": 0.0948,
      "num_input_tokens_seen": 11459392,
      "step": 17545
    },
    {
      "epoch": 9.19811320754717,
      "grad_norm": 4.399864673614502,
      "learning_rate": 3.2735428854255416e-05,
      "loss": 0.1536,
      "num_input_tokens_seen": 11462336,
      "step": 17550
    },
    {
      "epoch": 9.200733752620545,
      "grad_norm": 3.970682144165039,
      "learning_rate": 3.272455488210697e-05,
      "loss": 0.1106,
      "num_input_tokens_seen": 11464896,
      "step": 17555
    },
    {
      "epoch": 9.20335429769392,
      "grad_norm": 1.7368220090866089,
      "learning_rate": 3.271367929406977e-05,
      "loss": 0.1472,
      "num_input_tokens_seen": 11472224,
      "step": 17560
    },
    {
      "epoch": 9.205974842767295,
      "grad_norm": 2.2335715293884277,
      "learning_rate": 3.270280209241886e-05,
      "loss": 0.1201,
      "num_input_tokens_seen": 11474912,
      "step": 17565
    },
    {
      "epoch": 9.20859538784067,
      "grad_norm": 3.9171605110168457,
      "learning_rate": 3.2691923279429624e-05,
      "loss": 0.2167,
      "num_input_tokens_seen": 11478240,
      "step": 17570
    },
    {
      "epoch": 9.211215932914046,
      "grad_norm": 1.913928747177124,
      "learning_rate": 3.2681042857377804e-05,
      "loss": 0.0992,
      "num_input_tokens_seen": 11481216,
      "step": 17575
    },
    {
      "epoch": 9.213836477987421,
      "grad_norm": 2.058962345123291,
      "learning_rate": 3.2670160828539435e-05,
      "loss": 0.1305,
      "num_input_tokens_seen": 11484384,
      "step": 17580
    },
    {
      "epoch": 9.216457023060796,
      "grad_norm": 1.6679760217666626,
      "learning_rate": 3.2659277195190934e-05,
      "loss": 0.1333,
      "num_input_tokens_seen": 11487104,
      "step": 17585
    },
    {
      "epoch": 9.219077568134171,
      "grad_norm": 2.7743287086486816,
      "learning_rate": 3.2648391959609026e-05,
      "loss": 0.1023,
      "num_input_tokens_seen": 11489696,
      "step": 17590
    },
    {
      "epoch": 9.221698113207546,
      "grad_norm": 1.6379570960998535,
      "learning_rate": 3.263750512407078e-05,
      "loss": 0.1043,
      "num_input_tokens_seen": 11492512,
      "step": 17595
    },
    {
      "epoch": 9.224318658280922,
      "grad_norm": 2.0864994525909424,
      "learning_rate": 3.262661669085359e-05,
      "loss": 0.1249,
      "num_input_tokens_seen": 11495648,
      "step": 17600
    },
    {
      "epoch": 9.226939203354299,
      "grad_norm": 2.329536199569702,
      "learning_rate": 3.2615726662235205e-05,
      "loss": 0.1399,
      "num_input_tokens_seen": 11498976,
      "step": 17605
    },
    {
      "epoch": 9.229559748427674,
      "grad_norm": 5.290849208831787,
      "learning_rate": 3.2604835040493676e-05,
      "loss": 0.1051,
      "num_input_tokens_seen": 11501888,
      "step": 17610
    },
    {
      "epoch": 9.232180293501049,
      "grad_norm": 3.223355293273926,
      "learning_rate": 3.259394182790742e-05,
      "loss": 0.163,
      "num_input_tokens_seen": 11504544,
      "step": 17615
    },
    {
      "epoch": 9.234800838574424,
      "grad_norm": 1.0863633155822754,
      "learning_rate": 3.2583047026755174e-05,
      "loss": 0.1098,
      "num_input_tokens_seen": 11507168,
      "step": 17620
    },
    {
      "epoch": 9.2374213836478,
      "grad_norm": 5.0414204597473145,
      "learning_rate": 3.2572150639316e-05,
      "loss": 0.0984,
      "num_input_tokens_seen": 11510272,
      "step": 17625
    },
    {
      "epoch": 9.240041928721174,
      "grad_norm": 4.873777389526367,
      "learning_rate": 3.256125266786931e-05,
      "loss": 0.1231,
      "num_input_tokens_seen": 11513280,
      "step": 17630
    },
    {
      "epoch": 9.24266247379455,
      "grad_norm": 1.7229077816009521,
      "learning_rate": 3.255035311469481e-05,
      "loss": 0.1066,
      "num_input_tokens_seen": 11516384,
      "step": 17635
    },
    {
      "epoch": 9.245283018867925,
      "grad_norm": 1.4446576833724976,
      "learning_rate": 3.2539451982072586e-05,
      "loss": 0.1727,
      "num_input_tokens_seen": 11520672,
      "step": 17640
    },
    {
      "epoch": 9.2479035639413,
      "grad_norm": 2.5970888137817383,
      "learning_rate": 3.2528549272283016e-05,
      "loss": 0.1552,
      "num_input_tokens_seen": 11523456,
      "step": 17645
    },
    {
      "epoch": 9.250524109014675,
      "grad_norm": 2.8878777027130127,
      "learning_rate": 3.251764498760683e-05,
      "loss": 0.0756,
      "num_input_tokens_seen": 11526240,
      "step": 17650
    },
    {
      "epoch": 9.25314465408805,
      "grad_norm": 2.403045654296875,
      "learning_rate": 3.2506739130325084e-05,
      "loss": 0.1509,
      "num_input_tokens_seen": 11531040,
      "step": 17655
    },
    {
      "epoch": 9.255765199161425,
      "grad_norm": 4.121960639953613,
      "learning_rate": 3.2495831702719146e-05,
      "loss": 0.102,
      "num_input_tokens_seen": 11533984,
      "step": 17660
    },
    {
      "epoch": 9.2583857442348,
      "grad_norm": 2.4743266105651855,
      "learning_rate": 3.248492270707074e-05,
      "loss": 0.1314,
      "num_input_tokens_seen": 11537248,
      "step": 17665
    },
    {
      "epoch": 9.261006289308176,
      "grad_norm": 4.385849475860596,
      "learning_rate": 3.247401214566189e-05,
      "loss": 0.1457,
      "num_input_tokens_seen": 11540384,
      "step": 17670
    },
    {
      "epoch": 9.26362683438155,
      "grad_norm": 1.930738925933838,
      "learning_rate": 3.2463100020774965e-05,
      "loss": 0.2068,
      "num_input_tokens_seen": 11543424,
      "step": 17675
    },
    {
      "epoch": 9.266247379454926,
      "grad_norm": 1.3210476636886597,
      "learning_rate": 3.245218633469267e-05,
      "loss": 0.1379,
      "num_input_tokens_seen": 11547136,
      "step": 17680
    },
    {
      "epoch": 9.268867924528301,
      "grad_norm": 3.3757121562957764,
      "learning_rate": 3.244127108969801e-05,
      "loss": 0.1873,
      "num_input_tokens_seen": 11550528,
      "step": 17685
    },
    {
      "epoch": 9.271488469601676,
      "grad_norm": 1.165698766708374,
      "learning_rate": 3.2430354288074314e-05,
      "loss": 0.1233,
      "num_input_tokens_seen": 11553440,
      "step": 17690
    },
    {
      "epoch": 9.274109014675052,
      "grad_norm": 1.7717565298080444,
      "learning_rate": 3.241943593210529e-05,
      "loss": 0.083,
      "num_input_tokens_seen": 11557184,
      "step": 17695
    },
    {
      "epoch": 9.276729559748428,
      "grad_norm": 2.515787363052368,
      "learning_rate": 3.2408516024074916e-05,
      "loss": 0.0849,
      "num_input_tokens_seen": 11560416,
      "step": 17700
    },
    {
      "epoch": 9.279350104821804,
      "grad_norm": 1.3140009641647339,
      "learning_rate": 3.2397594566267505e-05,
      "loss": 0.1608,
      "num_input_tokens_seen": 11563712,
      "step": 17705
    },
    {
      "epoch": 9.281970649895179,
      "grad_norm": 3.640801191329956,
      "learning_rate": 3.23866715609677e-05,
      "loss": 0.171,
      "num_input_tokens_seen": 11567776,
      "step": 17710
    },
    {
      "epoch": 9.284591194968554,
      "grad_norm": 1.4296791553497314,
      "learning_rate": 3.237574701046049e-05,
      "loss": 0.1664,
      "num_input_tokens_seen": 11571456,
      "step": 17715
    },
    {
      "epoch": 9.28721174004193,
      "grad_norm": 1.811365008354187,
      "learning_rate": 3.236482091703115e-05,
      "loss": 0.1547,
      "num_input_tokens_seen": 11575136,
      "step": 17720
    },
    {
      "epoch": 9.289832285115304,
      "grad_norm": 1.377213478088379,
      "learning_rate": 3.235389328296528e-05,
      "loss": 0.1769,
      "num_input_tokens_seen": 11578880,
      "step": 17725
    },
    {
      "epoch": 9.29245283018868,
      "grad_norm": 3.5521435737609863,
      "learning_rate": 3.2342964110548845e-05,
      "loss": 0.1144,
      "num_input_tokens_seen": 11581792,
      "step": 17730
    },
    {
      "epoch": 9.295073375262055,
      "grad_norm": 1.0356112718582153,
      "learning_rate": 3.2332033402068093e-05,
      "loss": 0.2358,
      "num_input_tokens_seen": 11584704,
      "step": 17735
    },
    {
      "epoch": 9.29769392033543,
      "grad_norm": 1.8978712558746338,
      "learning_rate": 3.2321101159809605e-05,
      "loss": 0.2091,
      "num_input_tokens_seen": 11587936,
      "step": 17740
    },
    {
      "epoch": 9.300314465408805,
      "grad_norm": 3.959731101989746,
      "learning_rate": 3.231016738606027e-05,
      "loss": 0.1596,
      "num_input_tokens_seen": 11591232,
      "step": 17745
    },
    {
      "epoch": 9.30293501048218,
      "grad_norm": 1.2250428199768066,
      "learning_rate": 3.2299232083107313e-05,
      "loss": 0.3207,
      "num_input_tokens_seen": 11594656,
      "step": 17750
    },
    {
      "epoch": 9.305555555555555,
      "grad_norm": 1.3546732664108276,
      "learning_rate": 3.228829525323829e-05,
      "loss": 0.1129,
      "num_input_tokens_seen": 11598016,
      "step": 17755
    },
    {
      "epoch": 9.30817610062893,
      "grad_norm": 2.5804190635681152,
      "learning_rate": 3.227735689874104e-05,
      "loss": 0.107,
      "num_input_tokens_seen": 11601312,
      "step": 17760
    },
    {
      "epoch": 9.310796645702306,
      "grad_norm": 1.9227062463760376,
      "learning_rate": 3.226641702190375e-05,
      "loss": 0.1728,
      "num_input_tokens_seen": 11604896,
      "step": 17765
    },
    {
      "epoch": 9.31341719077568,
      "grad_norm": 1.5068066120147705,
      "learning_rate": 3.225547562501493e-05,
      "loss": 0.1409,
      "num_input_tokens_seen": 11607840,
      "step": 17770
    },
    {
      "epoch": 9.316037735849056,
      "grad_norm": 2.0360960960388184,
      "learning_rate": 3.2244532710363375e-05,
      "loss": 0.1508,
      "num_input_tokens_seen": 11611392,
      "step": 17775
    },
    {
      "epoch": 9.318658280922431,
      "grad_norm": 1.574415922164917,
      "learning_rate": 3.2233588280238224e-05,
      "loss": 0.1364,
      "num_input_tokens_seen": 11614752,
      "step": 17780
    },
    {
      "epoch": 9.321278825995806,
      "grad_norm": 2.1936700344085693,
      "learning_rate": 3.2222642336928944e-05,
      "loss": 0.1251,
      "num_input_tokens_seen": 11617728,
      "step": 17785
    },
    {
      "epoch": 9.323899371069182,
      "grad_norm": 2.222100019454956,
      "learning_rate": 3.2211694882725275e-05,
      "loss": 0.1448,
      "num_input_tokens_seen": 11620736,
      "step": 17790
    },
    {
      "epoch": 9.326519916142558,
      "grad_norm": 1.9818115234375,
      "learning_rate": 3.220074591991731e-05,
      "loss": 0.1744,
      "num_input_tokens_seen": 11623744,
      "step": 17795
    },
    {
      "epoch": 9.329140461215934,
      "grad_norm": 1.6471054553985596,
      "learning_rate": 3.218979545079546e-05,
      "loss": 0.1174,
      "num_input_tokens_seen": 11626496,
      "step": 17800
    },
    {
      "epoch": 9.331761006289309,
      "grad_norm": 2.0081093311309814,
      "learning_rate": 3.217884347765042e-05,
      "loss": 0.1192,
      "num_input_tokens_seen": 11629504,
      "step": 17805
    },
    {
      "epoch": 9.334381551362684,
      "grad_norm": 4.164993762969971,
      "learning_rate": 3.216789000277323e-05,
      "loss": 0.1141,
      "num_input_tokens_seen": 11632384,
      "step": 17810
    },
    {
      "epoch": 9.33700209643606,
      "grad_norm": 2.324708938598633,
      "learning_rate": 3.2156935028455224e-05,
      "loss": 0.1447,
      "num_input_tokens_seen": 11635616,
      "step": 17815
    },
    {
      "epoch": 9.339622641509434,
      "grad_norm": 2.6693339347839355,
      "learning_rate": 3.2145978556988056e-05,
      "loss": 0.1203,
      "num_input_tokens_seen": 11638912,
      "step": 17820
    },
    {
      "epoch": 9.34224318658281,
      "grad_norm": 2.198150634765625,
      "learning_rate": 3.213502059066371e-05,
      "loss": 0.147,
      "num_input_tokens_seen": 11642432,
      "step": 17825
    },
    {
      "epoch": 9.344863731656185,
      "grad_norm": 1.8333990573883057,
      "learning_rate": 3.212406113177445e-05,
      "loss": 0.0752,
      "num_input_tokens_seen": 11645120,
      "step": 17830
    },
    {
      "epoch": 9.34748427672956,
      "grad_norm": 12.363137245178223,
      "learning_rate": 3.211310018261287e-05,
      "loss": 0.1126,
      "num_input_tokens_seen": 11647456,
      "step": 17835
    },
    {
      "epoch": 9.350104821802935,
      "grad_norm": 1.4427218437194824,
      "learning_rate": 3.21021377454719e-05,
      "loss": 0.2007,
      "num_input_tokens_seen": 11650976,
      "step": 17840
    },
    {
      "epoch": 9.35272536687631,
      "grad_norm": 3.7047038078308105,
      "learning_rate": 3.209117382264473e-05,
      "loss": 0.1535,
      "num_input_tokens_seen": 11654304,
      "step": 17845
    },
    {
      "epoch": 9.355345911949685,
      "grad_norm": 5.128296852111816,
      "learning_rate": 3.208020841642489e-05,
      "loss": 0.1502,
      "num_input_tokens_seen": 11658592,
      "step": 17850
    },
    {
      "epoch": 9.35796645702306,
      "grad_norm": 2.8467354774475098,
      "learning_rate": 3.206924152910623e-05,
      "loss": 0.1355,
      "num_input_tokens_seen": 11662464,
      "step": 17855
    },
    {
      "epoch": 9.360587002096436,
      "grad_norm": 2.430004119873047,
      "learning_rate": 3.205827316298289e-05,
      "loss": 0.1228,
      "num_input_tokens_seen": 11665248,
      "step": 17860
    },
    {
      "epoch": 9.36320754716981,
      "grad_norm": 5.916564464569092,
      "learning_rate": 3.204730332034934e-05,
      "loss": 0.2004,
      "num_input_tokens_seen": 11668256,
      "step": 17865
    },
    {
      "epoch": 9.365828092243186,
      "grad_norm": 1.4110318422317505,
      "learning_rate": 3.2036332003500324e-05,
      "loss": 0.1762,
      "num_input_tokens_seen": 11671264,
      "step": 17870
    },
    {
      "epoch": 9.368448637316561,
      "grad_norm": 2.7513835430145264,
      "learning_rate": 3.2025359214730924e-05,
      "loss": 0.1733,
      "num_input_tokens_seen": 11673728,
      "step": 17875
    },
    {
      "epoch": 9.371069182389936,
      "grad_norm": 0.8761131167411804,
      "learning_rate": 3.201438495633654e-05,
      "loss": 0.1173,
      "num_input_tokens_seen": 11676960,
      "step": 17880
    },
    {
      "epoch": 9.373689727463312,
      "grad_norm": 2.719813823699951,
      "learning_rate": 3.200340923061284e-05,
      "loss": 0.1696,
      "num_input_tokens_seen": 11681440,
      "step": 17885
    },
    {
      "epoch": 9.376310272536688,
      "grad_norm": 2.320495843887329,
      "learning_rate": 3.199243203985583e-05,
      "loss": 0.1082,
      "num_input_tokens_seen": 11684544,
      "step": 17890
    },
    {
      "epoch": 9.378930817610064,
      "grad_norm": 2.1970043182373047,
      "learning_rate": 3.1981453386361804e-05,
      "loss": 0.112,
      "num_input_tokens_seen": 11687584,
      "step": 17895
    },
    {
      "epoch": 9.381551362683439,
      "grad_norm": 1.2124595642089844,
      "learning_rate": 3.1970473272427386e-05,
      "loss": 0.1022,
      "num_input_tokens_seen": 11690944,
      "step": 17900
    },
    {
      "epoch": 9.384171907756814,
      "grad_norm": 6.031479358673096,
      "learning_rate": 3.195949170034948e-05,
      "loss": 0.1331,
      "num_input_tokens_seen": 11694112,
      "step": 17905
    },
    {
      "epoch": 9.38679245283019,
      "grad_norm": 2.7650582790374756,
      "learning_rate": 3.19485086724253e-05,
      "loss": 0.1539,
      "num_input_tokens_seen": 11697376,
      "step": 17910
    },
    {
      "epoch": 9.389412997903564,
      "grad_norm": 2.566190719604492,
      "learning_rate": 3.193752419095239e-05,
      "loss": 0.0974,
      "num_input_tokens_seen": 11700832,
      "step": 17915
    },
    {
      "epoch": 9.39203354297694,
      "grad_norm": 2.114842414855957,
      "learning_rate": 3.192653825822857e-05,
      "loss": 0.1459,
      "num_input_tokens_seen": 11703520,
      "step": 17920
    },
    {
      "epoch": 9.394654088050315,
      "grad_norm": 2.2278709411621094,
      "learning_rate": 3.1915550876551964e-05,
      "loss": 0.1652,
      "num_input_tokens_seen": 11707008,
      "step": 17925
    },
    {
      "epoch": 9.39727463312369,
      "grad_norm": 2.202223062515259,
      "learning_rate": 3.190456204822101e-05,
      "loss": 0.0805,
      "num_input_tokens_seen": 11709408,
      "step": 17930
    },
    {
      "epoch": 9.399895178197065,
      "grad_norm": 1.8661657571792603,
      "learning_rate": 3.1893571775534445e-05,
      "loss": 0.1224,
      "num_input_tokens_seen": 11712000,
      "step": 17935
    },
    {
      "epoch": 9.40251572327044,
      "grad_norm": 3.0805139541625977,
      "learning_rate": 3.1882580060791315e-05,
      "loss": 0.1554,
      "num_input_tokens_seen": 11716064,
      "step": 17940
    },
    {
      "epoch": 9.405136268343815,
      "grad_norm": 2.213282585144043,
      "learning_rate": 3.187158690629095e-05,
      "loss": 0.1048,
      "num_input_tokens_seen": 11719776,
      "step": 17945
    },
    {
      "epoch": 9.40775681341719,
      "grad_norm": 2.216061592102051,
      "learning_rate": 3.1860592314333e-05,
      "loss": 0.1161,
      "num_input_tokens_seen": 11722880,
      "step": 17950
    },
    {
      "epoch": 9.410377358490566,
      "grad_norm": 2.5485494136810303,
      "learning_rate": 3.184959628721742e-05,
      "loss": 0.22,
      "num_input_tokens_seen": 11725824,
      "step": 17955
    },
    {
      "epoch": 9.41299790356394,
      "grad_norm": 4.735567092895508,
      "learning_rate": 3.183859882724443e-05,
      "loss": 0.1695,
      "num_input_tokens_seen": 11728480,
      "step": 17960
    },
    {
      "epoch": 9.415618448637316,
      "grad_norm": 1.9308109283447266,
      "learning_rate": 3.1827599936714584e-05,
      "loss": 0.1629,
      "num_input_tokens_seen": 11731040,
      "step": 17965
    },
    {
      "epoch": 9.418238993710691,
      "grad_norm": 2.5304665565490723,
      "learning_rate": 3.1816599617928736e-05,
      "loss": 0.1174,
      "num_input_tokens_seen": 11733664,
      "step": 17970
    },
    {
      "epoch": 9.420859538784066,
      "grad_norm": 2.2087533473968506,
      "learning_rate": 3.1805597873188e-05,
      "loss": 0.0983,
      "num_input_tokens_seen": 11737216,
      "step": 17975
    },
    {
      "epoch": 9.423480083857442,
      "grad_norm": 2.058854341506958,
      "learning_rate": 3.179459470479384e-05,
      "loss": 0.1264,
      "num_input_tokens_seen": 11740864,
      "step": 17980
    },
    {
      "epoch": 9.426100628930818,
      "grad_norm": 1.7453402280807495,
      "learning_rate": 3.1783590115047983e-05,
      "loss": 0.124,
      "num_input_tokens_seen": 11743776,
      "step": 17985
    },
    {
      "epoch": 9.428721174004194,
      "grad_norm": 3.3343207836151123,
      "learning_rate": 3.177258410625246e-05,
      "loss": 0.134,
      "num_input_tokens_seen": 11747552,
      "step": 17990
    },
    {
      "epoch": 9.431341719077569,
      "grad_norm": 2.3471455574035645,
      "learning_rate": 3.176157668070962e-05,
      "loss": 0.0918,
      "num_input_tokens_seen": 11750656,
      "step": 17995
    },
    {
      "epoch": 9.433962264150944,
      "grad_norm": 2.485933303833008,
      "learning_rate": 3.175056784072207e-05,
      "loss": 0.1406,
      "num_input_tokens_seen": 11753888,
      "step": 18000
    },
    {
      "epoch": 9.43658280922432,
      "grad_norm": 6.298439979553223,
      "learning_rate": 3.173955758859274e-05,
      "loss": 0.1653,
      "num_input_tokens_seen": 11757184,
      "step": 18005
    },
    {
      "epoch": 9.439203354297694,
      "grad_norm": 5.11331033706665,
      "learning_rate": 3.172854592662486e-05,
      "loss": 0.1178,
      "num_input_tokens_seen": 11759936,
      "step": 18010
    },
    {
      "epoch": 9.44182389937107,
      "grad_norm": 1.9694706201553345,
      "learning_rate": 3.1717532857121926e-05,
      "loss": 0.108,
      "num_input_tokens_seen": 11762816,
      "step": 18015
    },
    {
      "epoch": 9.444444444444445,
      "grad_norm": 9.546574592590332,
      "learning_rate": 3.170651838238776e-05,
      "loss": 0.1206,
      "num_input_tokens_seen": 11765824,
      "step": 18020
    },
    {
      "epoch": 9.44706498951782,
      "grad_norm": 2.3600339889526367,
      "learning_rate": 3.169550250472646e-05,
      "loss": 0.1198,
      "num_input_tokens_seen": 11769056,
      "step": 18025
    },
    {
      "epoch": 9.449685534591195,
      "grad_norm": 2.4890570640563965,
      "learning_rate": 3.168448522644243e-05,
      "loss": 0.0742,
      "num_input_tokens_seen": 11771808,
      "step": 18030
    },
    {
      "epoch": 9.45230607966457,
      "grad_norm": 1.7801152467727661,
      "learning_rate": 3.167346654984034e-05,
      "loss": 0.1097,
      "num_input_tokens_seen": 11774720,
      "step": 18035
    },
    {
      "epoch": 9.454926624737945,
      "grad_norm": 2.4018771648406982,
      "learning_rate": 3.166244647722519e-05,
      "loss": 0.1191,
      "num_input_tokens_seen": 11777664,
      "step": 18040
    },
    {
      "epoch": 9.45754716981132,
      "grad_norm": 2.6747868061065674,
      "learning_rate": 3.1651425010902244e-05,
      "loss": 0.1243,
      "num_input_tokens_seen": 11781696,
      "step": 18045
    },
    {
      "epoch": 9.460167714884696,
      "grad_norm": 1.8083683252334595,
      "learning_rate": 3.164040215317706e-05,
      "loss": 0.1343,
      "num_input_tokens_seen": 11784672,
      "step": 18050
    },
    {
      "epoch": 9.46278825995807,
      "grad_norm": 3.39559006690979,
      "learning_rate": 3.162937790635552e-05,
      "loss": 0.1508,
      "num_input_tokens_seen": 11787424,
      "step": 18055
    },
    {
      "epoch": 9.465408805031446,
      "grad_norm": 2.3885927200317383,
      "learning_rate": 3.1618352272743746e-05,
      "loss": 0.137,
      "num_input_tokens_seen": 11789824,
      "step": 18060
    },
    {
      "epoch": 9.468029350104821,
      "grad_norm": 0.877875804901123,
      "learning_rate": 3.160732525464818e-05,
      "loss": 0.0488,
      "num_input_tokens_seen": 11793120,
      "step": 18065
    },
    {
      "epoch": 9.470649895178196,
      "grad_norm": 2.7332510948181152,
      "learning_rate": 3.1596296854375554e-05,
      "loss": 0.1431,
      "num_input_tokens_seen": 11798720,
      "step": 18070
    },
    {
      "epoch": 9.473270440251572,
      "grad_norm": 3.0761237144470215,
      "learning_rate": 3.1585267074232884e-05,
      "loss": 0.1317,
      "num_input_tokens_seen": 11801600,
      "step": 18075
    },
    {
      "epoch": 9.475890985324948,
      "grad_norm": 3.991903781890869,
      "learning_rate": 3.157423591652746e-05,
      "loss": 0.1015,
      "num_input_tokens_seen": 11804640,
      "step": 18080
    },
    {
      "epoch": 9.478511530398324,
      "grad_norm": 2.1387410163879395,
      "learning_rate": 3.1563203383566894e-05,
      "loss": 0.1237,
      "num_input_tokens_seen": 11807296,
      "step": 18085
    },
    {
      "epoch": 9.481132075471699,
      "grad_norm": 3.07684588432312,
      "learning_rate": 3.155216947765906e-05,
      "loss": 0.1007,
      "num_input_tokens_seen": 11809920,
      "step": 18090
    },
    {
      "epoch": 9.483752620545074,
      "grad_norm": 2.11385440826416,
      "learning_rate": 3.154113420111211e-05,
      "loss": 0.0851,
      "num_input_tokens_seen": 11813440,
      "step": 18095
    },
    {
      "epoch": 9.48637316561845,
      "grad_norm": 1.9069437980651855,
      "learning_rate": 3.1530097556234526e-05,
      "loss": 0.1517,
      "num_input_tokens_seen": 11817440,
      "step": 18100
    },
    {
      "epoch": 9.488993710691824,
      "grad_norm": 4.7933030128479,
      "learning_rate": 3.151905954533502e-05,
      "loss": 0.1729,
      "num_input_tokens_seen": 11820192,
      "step": 18105
    },
    {
      "epoch": 9.4916142557652,
      "grad_norm": 3.301913261413574,
      "learning_rate": 3.150802017072263e-05,
      "loss": 0.2419,
      "num_input_tokens_seen": 11824192,
      "step": 18110
    },
    {
      "epoch": 9.494234800838575,
      "grad_norm": 1.8627917766571045,
      "learning_rate": 3.149697943470668e-05,
      "loss": 0.2214,
      "num_input_tokens_seen": 11826944,
      "step": 18115
    },
    {
      "epoch": 9.49685534591195,
      "grad_norm": 1.7056105136871338,
      "learning_rate": 3.1485937339596736e-05,
      "loss": 0.1525,
      "num_input_tokens_seen": 11830784,
      "step": 18120
    },
    {
      "epoch": 9.499475890985325,
      "grad_norm": 3.0216081142425537,
      "learning_rate": 3.147489388770271e-05,
      "loss": 0.1956,
      "num_input_tokens_seen": 11833632,
      "step": 18125
    },
    {
      "epoch": 9.5020964360587,
      "grad_norm": 2.15727162361145,
      "learning_rate": 3.146384908133475e-05,
      "loss": 0.1712,
      "num_input_tokens_seen": 11837056,
      "step": 18130
    },
    {
      "epoch": 9.504716981132075,
      "grad_norm": 7.412652969360352,
      "learning_rate": 3.14528029228033e-05,
      "loss": 0.109,
      "num_input_tokens_seen": 11840736,
      "step": 18135
    },
    {
      "epoch": 9.50733752620545,
      "grad_norm": 2.254871129989624,
      "learning_rate": 3.144175541441911e-05,
      "loss": 0.2012,
      "num_input_tokens_seen": 11843328,
      "step": 18140
    },
    {
      "epoch": 9.509958071278826,
      "grad_norm": 2.552013635635376,
      "learning_rate": 3.1430706558493166e-05,
      "loss": 0.167,
      "num_input_tokens_seen": 11846784,
      "step": 18145
    },
    {
      "epoch": 9.5125786163522,
      "grad_norm": 1.9040549993515015,
      "learning_rate": 3.141965635733677e-05,
      "loss": 0.1247,
      "num_input_tokens_seen": 11849984,
      "step": 18150
    },
    {
      "epoch": 9.515199161425576,
      "grad_norm": 2.412036895751953,
      "learning_rate": 3.140860481326152e-05,
      "loss": 0.094,
      "num_input_tokens_seen": 11852992,
      "step": 18155
    },
    {
      "epoch": 9.517819706498951,
      "grad_norm": 4.000802993774414,
      "learning_rate": 3.139755192857925e-05,
      "loss": 0.1039,
      "num_input_tokens_seen": 11855776,
      "step": 18160
    },
    {
      "epoch": 9.520440251572326,
      "grad_norm": 1.7347300052642822,
      "learning_rate": 3.1386497705602094e-05,
      "loss": 0.1093,
      "num_input_tokens_seen": 11859008,
      "step": 18165
    },
    {
      "epoch": 9.523060796645701,
      "grad_norm": 3.1610074043273926,
      "learning_rate": 3.137544214664249e-05,
      "loss": 0.0815,
      "num_input_tokens_seen": 11861632,
      "step": 18170
    },
    {
      "epoch": 9.525681341719078,
      "grad_norm": 1.5870718955993652,
      "learning_rate": 3.136438525401312e-05,
      "loss": 0.12,
      "num_input_tokens_seen": 11864864,
      "step": 18175
    },
    {
      "epoch": 9.528301886792454,
      "grad_norm": 1.8308842182159424,
      "learning_rate": 3.135332703002696e-05,
      "loss": 0.1989,
      "num_input_tokens_seen": 11868224,
      "step": 18180
    },
    {
      "epoch": 9.530922431865829,
      "grad_norm": 2.0612268447875977,
      "learning_rate": 3.1342267476997276e-05,
      "loss": 0.106,
      "num_input_tokens_seen": 11871072,
      "step": 18185
    },
    {
      "epoch": 9.533542976939204,
      "grad_norm": 3.167226791381836,
      "learning_rate": 3.1331206597237584e-05,
      "loss": 0.1375,
      "num_input_tokens_seen": 11873888,
      "step": 18190
    },
    {
      "epoch": 9.536163522012579,
      "grad_norm": 3.607041120529175,
      "learning_rate": 3.1320144393061715e-05,
      "loss": 0.1286,
      "num_input_tokens_seen": 11876608,
      "step": 18195
    },
    {
      "epoch": 9.538784067085954,
      "grad_norm": 1.3741086721420288,
      "learning_rate": 3.130908086678373e-05,
      "loss": 0.1178,
      "num_input_tokens_seen": 11880864,
      "step": 18200
    },
    {
      "epoch": 9.54140461215933,
      "grad_norm": 2.3035922050476074,
      "learning_rate": 3.1298016020718014e-05,
      "loss": 0.1514,
      "num_input_tokens_seen": 11883520,
      "step": 18205
    },
    {
      "epoch": 9.544025157232705,
      "grad_norm": 1.0371276140213013,
      "learning_rate": 3.128694985717919e-05,
      "loss": 0.095,
      "num_input_tokens_seen": 11887168,
      "step": 18210
    },
    {
      "epoch": 9.54664570230608,
      "grad_norm": 2.154703378677368,
      "learning_rate": 3.1275882378482185e-05,
      "loss": 0.2088,
      "num_input_tokens_seen": 11890976,
      "step": 18215
    },
    {
      "epoch": 9.549266247379455,
      "grad_norm": 2.074033260345459,
      "learning_rate": 3.1264813586942185e-05,
      "loss": 0.1229,
      "num_input_tokens_seen": 11893888,
      "step": 18220
    },
    {
      "epoch": 9.55188679245283,
      "grad_norm": 2.2876136302948,
      "learning_rate": 3.125374348487466e-05,
      "loss": 0.0952,
      "num_input_tokens_seen": 11896704,
      "step": 18225
    },
    {
      "epoch": 9.554507337526205,
      "grad_norm": 2.7084200382232666,
      "learning_rate": 3.1242672074595334e-05,
      "loss": 0.1644,
      "num_input_tokens_seen": 11900736,
      "step": 18230
    },
    {
      "epoch": 9.55712788259958,
      "grad_norm": 3.0116090774536133,
      "learning_rate": 3.123159935842024e-05,
      "loss": 0.208,
      "num_input_tokens_seen": 11904160,
      "step": 18235
    },
    {
      "epoch": 9.559748427672956,
      "grad_norm": 1.315717101097107,
      "learning_rate": 3.122052533866565e-05,
      "loss": 0.133,
      "num_input_tokens_seen": 11907200,
      "step": 18240
    },
    {
      "epoch": 9.56236897274633,
      "grad_norm": 1.1581127643585205,
      "learning_rate": 3.120945001764813e-05,
      "loss": 0.1089,
      "num_input_tokens_seen": 11910368,
      "step": 18245
    },
    {
      "epoch": 9.564989517819706,
      "grad_norm": 2.3482816219329834,
      "learning_rate": 3.11983733976845e-05,
      "loss": 0.15,
      "num_input_tokens_seen": 11913472,
      "step": 18250
    },
    {
      "epoch": 9.567610062893081,
      "grad_norm": 1.0448956489562988,
      "learning_rate": 3.1187295481091876e-05,
      "loss": 0.1053,
      "num_input_tokens_seen": 11917280,
      "step": 18255
    },
    {
      "epoch": 9.570230607966456,
      "grad_norm": 1.3403133153915405,
      "learning_rate": 3.1176216270187625e-05,
      "loss": 0.1093,
      "num_input_tokens_seen": 11919744,
      "step": 18260
    },
    {
      "epoch": 9.572851153039831,
      "grad_norm": 1.9417965412139893,
      "learning_rate": 3.116513576728939e-05,
      "loss": 0.1882,
      "num_input_tokens_seen": 11924096,
      "step": 18265
    },
    {
      "epoch": 9.575471698113208,
      "grad_norm": 4.597141265869141,
      "learning_rate": 3.1154053974715085e-05,
      "loss": 0.1048,
      "num_input_tokens_seen": 11927744,
      "step": 18270
    },
    {
      "epoch": 9.578092243186584,
      "grad_norm": 2.893819570541382,
      "learning_rate": 3.1142970894782906e-05,
      "loss": 0.1127,
      "num_input_tokens_seen": 11930368,
      "step": 18275
    },
    {
      "epoch": 9.580712788259959,
      "grad_norm": 3.7693331241607666,
      "learning_rate": 3.1131886529811296e-05,
      "loss": 0.1163,
      "num_input_tokens_seen": 11933344,
      "step": 18280
    },
    {
      "epoch": 9.583333333333334,
      "grad_norm": 2.7738966941833496,
      "learning_rate": 3.1120800882118985e-05,
      "loss": 0.114,
      "num_input_tokens_seen": 11936992,
      "step": 18285
    },
    {
      "epoch": 9.585953878406709,
      "grad_norm": 3.422551393508911,
      "learning_rate": 3.1109713954024945e-05,
      "loss": 0.1101,
      "num_input_tokens_seen": 11940288,
      "step": 18290
    },
    {
      "epoch": 9.588574423480084,
      "grad_norm": 4.431211471557617,
      "learning_rate": 3.109862574784846e-05,
      "loss": 0.1719,
      "num_input_tokens_seen": 11943072,
      "step": 18295
    },
    {
      "epoch": 9.59119496855346,
      "grad_norm": 4.098977088928223,
      "learning_rate": 3.108753626590905e-05,
      "loss": 0.086,
      "num_input_tokens_seen": 11946016,
      "step": 18300
    },
    {
      "epoch": 9.593815513626835,
      "grad_norm": 4.904597282409668,
      "learning_rate": 3.1076445510526486e-05,
      "loss": 0.1687,
      "num_input_tokens_seen": 11949280,
      "step": 18305
    },
    {
      "epoch": 9.59643605870021,
      "grad_norm": 3.4560916423797607,
      "learning_rate": 3.106535348402086e-05,
      "loss": 0.1763,
      "num_input_tokens_seen": 11953216,
      "step": 18310
    },
    {
      "epoch": 9.599056603773585,
      "grad_norm": 1.6955018043518066,
      "learning_rate": 3.105426018871247e-05,
      "loss": 0.0961,
      "num_input_tokens_seen": 11956320,
      "step": 18315
    },
    {
      "epoch": 9.60167714884696,
      "grad_norm": 1.7881721258163452,
      "learning_rate": 3.104316562692193e-05,
      "loss": 0.1533,
      "num_input_tokens_seen": 11959104,
      "step": 18320
    },
    {
      "epoch": 9.604297693920335,
      "grad_norm": 2.679929256439209,
      "learning_rate": 3.103206980097009e-05,
      "loss": 0.1382,
      "num_input_tokens_seen": 11962208,
      "step": 18325
    },
    {
      "epoch": 9.60691823899371,
      "grad_norm": 3.5387685298919678,
      "learning_rate": 3.102097271317805e-05,
      "loss": 0.1688,
      "num_input_tokens_seen": 11964832,
      "step": 18330
    },
    {
      "epoch": 9.609538784067086,
      "grad_norm": 2.583984851837158,
      "learning_rate": 3.100987436586721e-05,
      "loss": 0.1917,
      "num_input_tokens_seen": 11968000,
      "step": 18335
    },
    {
      "epoch": 9.61215932914046,
      "grad_norm": 3.409912347793579,
      "learning_rate": 3.099877476135923e-05,
      "loss": 0.1953,
      "num_input_tokens_seen": 11970912,
      "step": 18340
    },
    {
      "epoch": 9.614779874213836,
      "grad_norm": 1.4878007173538208,
      "learning_rate": 3.0987673901976e-05,
      "loss": 0.0829,
      "num_input_tokens_seen": 11973472,
      "step": 18345
    },
    {
      "epoch": 9.617400419287211,
      "grad_norm": 3.871152400970459,
      "learning_rate": 3.0976571790039697e-05,
      "loss": 0.1354,
      "num_input_tokens_seen": 11976864,
      "step": 18350
    },
    {
      "epoch": 9.620020964360586,
      "grad_norm": 6.990462303161621,
      "learning_rate": 3.096546842787277e-05,
      "loss": 0.1788,
      "num_input_tokens_seen": 11979872,
      "step": 18355
    },
    {
      "epoch": 9.622641509433961,
      "grad_norm": 2.104689598083496,
      "learning_rate": 3.095436381779789e-05,
      "loss": 0.1484,
      "num_input_tokens_seen": 11983168,
      "step": 18360
    },
    {
      "epoch": 9.625262054507338,
      "grad_norm": 2.261141538619995,
      "learning_rate": 3.0943257962138045e-05,
      "loss": 0.1282,
      "num_input_tokens_seen": 11986432,
      "step": 18365
    },
    {
      "epoch": 9.627882599580714,
      "grad_norm": 1.5327115058898926,
      "learning_rate": 3.093215086321644e-05,
      "loss": 0.0869,
      "num_input_tokens_seen": 11988512,
      "step": 18370
    },
    {
      "epoch": 9.630503144654089,
      "grad_norm": 5.413748264312744,
      "learning_rate": 3.092104252335654e-05,
      "loss": 0.3608,
      "num_input_tokens_seen": 11991424,
      "step": 18375
    },
    {
      "epoch": 9.633123689727464,
      "grad_norm": 1.700157642364502,
      "learning_rate": 3.09099329448821e-05,
      "loss": 0.1146,
      "num_input_tokens_seen": 11994208,
      "step": 18380
    },
    {
      "epoch": 9.635744234800839,
      "grad_norm": 2.264288902282715,
      "learning_rate": 3.089882213011711e-05,
      "loss": 0.1254,
      "num_input_tokens_seen": 11997312,
      "step": 18385
    },
    {
      "epoch": 9.638364779874214,
      "grad_norm": 3.39972186088562,
      "learning_rate": 3.088771008138583e-05,
      "loss": 0.1935,
      "num_input_tokens_seen": 12000288,
      "step": 18390
    },
    {
      "epoch": 9.64098532494759,
      "grad_norm": 2.9729442596435547,
      "learning_rate": 3.087659680101278e-05,
      "loss": 0.2187,
      "num_input_tokens_seen": 12003328,
      "step": 18395
    },
    {
      "epoch": 9.643605870020965,
      "grad_norm": 2.8323957920074463,
      "learning_rate": 3.086548229132271e-05,
      "loss": 0.1386,
      "num_input_tokens_seen": 12006176,
      "step": 18400
    },
    {
      "epoch": 9.64622641509434,
      "grad_norm": 2.2183456420898438,
      "learning_rate": 3.085436655464067e-05,
      "loss": 0.1686,
      "num_input_tokens_seen": 12009856,
      "step": 18405
    },
    {
      "epoch": 9.648846960167715,
      "grad_norm": 1.55134916305542,
      "learning_rate": 3.084324959329193e-05,
      "loss": 0.1133,
      "num_input_tokens_seen": 12013376,
      "step": 18410
    },
    {
      "epoch": 9.65146750524109,
      "grad_norm": 1.649340271949768,
      "learning_rate": 3.083213140960204e-05,
      "loss": 0.2647,
      "num_input_tokens_seen": 12016192,
      "step": 18415
    },
    {
      "epoch": 9.654088050314465,
      "grad_norm": 2.4331836700439453,
      "learning_rate": 3.08210120058968e-05,
      "loss": 0.1946,
      "num_input_tokens_seen": 12019680,
      "step": 18420
    },
    {
      "epoch": 9.65670859538784,
      "grad_norm": 1.425220012664795,
      "learning_rate": 3.080989138450225e-05,
      "loss": 0.1006,
      "num_input_tokens_seen": 12023200,
      "step": 18425
    },
    {
      "epoch": 9.659329140461216,
      "grad_norm": 2.3913004398345947,
      "learning_rate": 3.0798769547744713e-05,
      "loss": 0.177,
      "num_input_tokens_seen": 12026880,
      "step": 18430
    },
    {
      "epoch": 9.66194968553459,
      "grad_norm": 2.2977311611175537,
      "learning_rate": 3.0787646497950736e-05,
      "loss": 0.1003,
      "num_input_tokens_seen": 12029472,
      "step": 18435
    },
    {
      "epoch": 9.664570230607966,
      "grad_norm": 3.299078941345215,
      "learning_rate": 3.0776522237447146e-05,
      "loss": 0.1674,
      "num_input_tokens_seen": 12032128,
      "step": 18440
    },
    {
      "epoch": 9.667190775681341,
      "grad_norm": 1.6295584440231323,
      "learning_rate": 3.076539676856101e-05,
      "loss": 0.1188,
      "num_input_tokens_seen": 12034912,
      "step": 18445
    },
    {
      "epoch": 9.669811320754716,
      "grad_norm": 2.4261314868927,
      "learning_rate": 3.0754270093619634e-05,
      "loss": 0.1363,
      "num_input_tokens_seen": 12037888,
      "step": 18450
    },
    {
      "epoch": 9.672431865828091,
      "grad_norm": 3.107215404510498,
      "learning_rate": 3.074314221495061e-05,
      "loss": 0.1072,
      "num_input_tokens_seen": 12040576,
      "step": 18455
    },
    {
      "epoch": 9.675052410901468,
      "grad_norm": 2.378927707672119,
      "learning_rate": 3.0732013134881756e-05,
      "loss": 0.1811,
      "num_input_tokens_seen": 12044096,
      "step": 18460
    },
    {
      "epoch": 9.677672955974844,
      "grad_norm": 1.8097699880599976,
      "learning_rate": 3.072088285574114e-05,
      "loss": 0.0818,
      "num_input_tokens_seen": 12047424,
      "step": 18465
    },
    {
      "epoch": 9.680293501048219,
      "grad_norm": 2.103905200958252,
      "learning_rate": 3.070975137985711e-05,
      "loss": 0.1381,
      "num_input_tokens_seen": 12050560,
      "step": 18470
    },
    {
      "epoch": 9.682914046121594,
      "grad_norm": 1.8832138776779175,
      "learning_rate": 3.0698618709558226e-05,
      "loss": 0.1425,
      "num_input_tokens_seen": 12053952,
      "step": 18475
    },
    {
      "epoch": 9.685534591194969,
      "grad_norm": 2.8417575359344482,
      "learning_rate": 3.068748484717332e-05,
      "loss": 0.1036,
      "num_input_tokens_seen": 12056416,
      "step": 18480
    },
    {
      "epoch": 9.688155136268344,
      "grad_norm": 2.5575623512268066,
      "learning_rate": 3.067634979503147e-05,
      "loss": 0.1847,
      "num_input_tokens_seen": 12060448,
      "step": 18485
    },
    {
      "epoch": 9.69077568134172,
      "grad_norm": 1.572045922279358,
      "learning_rate": 3.0665213555462e-05,
      "loss": 0.1273,
      "num_input_tokens_seen": 12064960,
      "step": 18490
    },
    {
      "epoch": 9.693396226415095,
      "grad_norm": 1.6292834281921387,
      "learning_rate": 3.0654076130794494e-05,
      "loss": 0.1143,
      "num_input_tokens_seen": 12068192,
      "step": 18495
    },
    {
      "epoch": 9.69601677148847,
      "grad_norm": 3.151254892349243,
      "learning_rate": 3.064293752335876e-05,
      "loss": 0.1544,
      "num_input_tokens_seen": 12071200,
      "step": 18500
    },
    {
      "epoch": 9.698637316561845,
      "grad_norm": 8.60757827758789,
      "learning_rate": 3.063179773548487e-05,
      "loss": 0.1421,
      "num_input_tokens_seen": 12074112,
      "step": 18505
    },
    {
      "epoch": 9.70125786163522,
      "grad_norm": 8.561177253723145,
      "learning_rate": 3.0620656769503164e-05,
      "loss": 0.1408,
      "num_input_tokens_seen": 12078048,
      "step": 18510
    },
    {
      "epoch": 9.703878406708595,
      "grad_norm": 3.103412389755249,
      "learning_rate": 3.060951462774417e-05,
      "loss": 0.1733,
      "num_input_tokens_seen": 12080896,
      "step": 18515
    },
    {
      "epoch": 9.70649895178197,
      "grad_norm": 0.977677047252655,
      "learning_rate": 3.0598371312538714e-05,
      "loss": 0.1015,
      "num_input_tokens_seen": 12083744,
      "step": 18520
    },
    {
      "epoch": 9.709119496855346,
      "grad_norm": 1.981260061264038,
      "learning_rate": 3.058722682621785e-05,
      "loss": 0.173,
      "num_input_tokens_seen": 12086496,
      "step": 18525
    },
    {
      "epoch": 9.71174004192872,
      "grad_norm": 1.5202946662902832,
      "learning_rate": 3.0576081171112875e-05,
      "loss": 0.108,
      "num_input_tokens_seen": 12089728,
      "step": 18530
    },
    {
      "epoch": 9.714360587002096,
      "grad_norm": 7.700599670410156,
      "learning_rate": 3.056493434955534e-05,
      "loss": 0.1968,
      "num_input_tokens_seen": 12093120,
      "step": 18535
    },
    {
      "epoch": 9.716981132075471,
      "grad_norm": 2.039824962615967,
      "learning_rate": 3.0553786363877027e-05,
      "loss": 0.1556,
      "num_input_tokens_seen": 12095680,
      "step": 18540
    },
    {
      "epoch": 9.719601677148846,
      "grad_norm": 2.2491812705993652,
      "learning_rate": 3.054263721640996e-05,
      "loss": 0.16,
      "num_input_tokens_seen": 12099040,
      "step": 18545
    },
    {
      "epoch": 9.722222222222221,
      "grad_norm": 1.1762458086013794,
      "learning_rate": 3.053148690948642e-05,
      "loss": 0.1282,
      "num_input_tokens_seen": 12102112,
      "step": 18550
    },
    {
      "epoch": 9.724842767295598,
      "grad_norm": 1.83250093460083,
      "learning_rate": 3.0520335445438935e-05,
      "loss": 0.1237,
      "num_input_tokens_seen": 12104544,
      "step": 18555
    },
    {
      "epoch": 9.727463312368974,
      "grad_norm": 3.2827870845794678,
      "learning_rate": 3.0509182826600246e-05,
      "loss": 0.1399,
      "num_input_tokens_seen": 12107712,
      "step": 18560
    },
    {
      "epoch": 9.730083857442349,
      "grad_norm": 2.694486618041992,
      "learning_rate": 3.049802905530337e-05,
      "loss": 0.1319,
      "num_input_tokens_seen": 12110528,
      "step": 18565
    },
    {
      "epoch": 9.732704402515724,
      "grad_norm": 2.850738763809204,
      "learning_rate": 3.0486874133881533e-05,
      "loss": 0.1225,
      "num_input_tokens_seen": 12113600,
      "step": 18570
    },
    {
      "epoch": 9.735324947589099,
      "grad_norm": 1.907478928565979,
      "learning_rate": 3.047571806466823e-05,
      "loss": 0.1514,
      "num_input_tokens_seen": 12116896,
      "step": 18575
    },
    {
      "epoch": 9.737945492662474,
      "grad_norm": 1.3979626893997192,
      "learning_rate": 3.0464560849997177e-05,
      "loss": 0.1201,
      "num_input_tokens_seen": 12120832,
      "step": 18580
    },
    {
      "epoch": 9.74056603773585,
      "grad_norm": 1.7284998893737793,
      "learning_rate": 3.0453402492202325e-05,
      "loss": 0.1377,
      "num_input_tokens_seen": 12124704,
      "step": 18585
    },
    {
      "epoch": 9.743186582809225,
      "grad_norm": 3.321753978729248,
      "learning_rate": 3.0442242993617904e-05,
      "loss": 0.0974,
      "num_input_tokens_seen": 12129760,
      "step": 18590
    },
    {
      "epoch": 9.7458071278826,
      "grad_norm": 2.466897487640381,
      "learning_rate": 3.0431082356578334e-05,
      "loss": 0.0762,
      "num_input_tokens_seen": 12132096,
      "step": 18595
    },
    {
      "epoch": 9.748427672955975,
      "grad_norm": 1.703812599182129,
      "learning_rate": 3.0419920583418294e-05,
      "loss": 0.1709,
      "num_input_tokens_seen": 12135008,
      "step": 18600
    },
    {
      "epoch": 9.75104821802935,
      "grad_norm": 6.355600357055664,
      "learning_rate": 3.0408757676472694e-05,
      "loss": 0.2228,
      "num_input_tokens_seen": 12138144,
      "step": 18605
    },
    {
      "epoch": 9.753668763102725,
      "grad_norm": 4.060497283935547,
      "learning_rate": 3.0397593638076705e-05,
      "loss": 0.1329,
      "num_input_tokens_seen": 12141568,
      "step": 18610
    },
    {
      "epoch": 9.7562893081761,
      "grad_norm": 2.4409642219543457,
      "learning_rate": 3.0386428470565702e-05,
      "loss": 0.1718,
      "num_input_tokens_seen": 12145248,
      "step": 18615
    },
    {
      "epoch": 9.758909853249476,
      "grad_norm": 1.3459734916687012,
      "learning_rate": 3.0375262176275308e-05,
      "loss": 0.1266,
      "num_input_tokens_seen": 12149472,
      "step": 18620
    },
    {
      "epoch": 9.76153039832285,
      "grad_norm": 1.988484263420105,
      "learning_rate": 3.03640947575414e-05,
      "loss": 0.1857,
      "num_input_tokens_seen": 12152992,
      "step": 18625
    },
    {
      "epoch": 9.764150943396226,
      "grad_norm": 8.141427040100098,
      "learning_rate": 3.0352926216700067e-05,
      "loss": 0.1449,
      "num_input_tokens_seen": 12155712,
      "step": 18630
    },
    {
      "epoch": 9.766771488469601,
      "grad_norm": 2.540423631668091,
      "learning_rate": 3.034175655608763e-05,
      "loss": 0.1809,
      "num_input_tokens_seen": 12158688,
      "step": 18635
    },
    {
      "epoch": 9.769392033542976,
      "grad_norm": 3.744297981262207,
      "learning_rate": 3.0330585778040677e-05,
      "loss": 0.2323,
      "num_input_tokens_seen": 12161824,
      "step": 18640
    },
    {
      "epoch": 9.772012578616351,
      "grad_norm": 2.14129376411438,
      "learning_rate": 3.031941388489598e-05,
      "loss": 0.1273,
      "num_input_tokens_seen": 12164992,
      "step": 18645
    },
    {
      "epoch": 9.774633123689728,
      "grad_norm": 6.698681831359863,
      "learning_rate": 3.0308240878990584e-05,
      "loss": 0.1294,
      "num_input_tokens_seen": 12167808,
      "step": 18650
    },
    {
      "epoch": 9.777253668763104,
      "grad_norm": 1.9553992748260498,
      "learning_rate": 3.0297066762661773e-05,
      "loss": 0.1069,
      "num_input_tokens_seen": 12170752,
      "step": 18655
    },
    {
      "epoch": 9.779874213836479,
      "grad_norm": 3.748317003250122,
      "learning_rate": 3.0285891538247015e-05,
      "loss": 0.1034,
      "num_input_tokens_seen": 12173664,
      "step": 18660
    },
    {
      "epoch": 9.782494758909854,
      "grad_norm": 1.4237275123596191,
      "learning_rate": 3.0274715208084053e-05,
      "loss": 0.1311,
      "num_input_tokens_seen": 12176960,
      "step": 18665
    },
    {
      "epoch": 9.785115303983229,
      "grad_norm": 2.2459444999694824,
      "learning_rate": 3.0263537774510847e-05,
      "loss": 0.2255,
      "num_input_tokens_seen": 12180096,
      "step": 18670
    },
    {
      "epoch": 9.787735849056604,
      "grad_norm": 2.8950376510620117,
      "learning_rate": 3.0252359239865585e-05,
      "loss": 0.0857,
      "num_input_tokens_seen": 12183424,
      "step": 18675
    },
    {
      "epoch": 9.79035639412998,
      "grad_norm": 3.7420437335968018,
      "learning_rate": 3.0241179606486704e-05,
      "loss": 0.1193,
      "num_input_tokens_seen": 12186592,
      "step": 18680
    },
    {
      "epoch": 9.792976939203355,
      "grad_norm": 3.714282274246216,
      "learning_rate": 3.0229998876712827e-05,
      "loss": 0.1715,
      "num_input_tokens_seen": 12190240,
      "step": 18685
    },
    {
      "epoch": 9.79559748427673,
      "grad_norm": 3.294039249420166,
      "learning_rate": 3.0218817052882858e-05,
      "loss": 0.1621,
      "num_input_tokens_seen": 12193024,
      "step": 18690
    },
    {
      "epoch": 9.798218029350105,
      "grad_norm": 1.9128468036651611,
      "learning_rate": 3.020763413733591e-05,
      "loss": 0.1295,
      "num_input_tokens_seen": 12196128,
      "step": 18695
    },
    {
      "epoch": 9.80083857442348,
      "grad_norm": 2.3238420486450195,
      "learning_rate": 3.0196450132411302e-05,
      "loss": 0.166,
      "num_input_tokens_seen": 12198880,
      "step": 18700
    },
    {
      "epoch": 9.803459119496855,
      "grad_norm": 2.0688588619232178,
      "learning_rate": 3.0185265040448616e-05,
      "loss": 0.1199,
      "num_input_tokens_seen": 12201760,
      "step": 18705
    },
    {
      "epoch": 9.80607966457023,
      "grad_norm": 1.849652647972107,
      "learning_rate": 3.017407886378764e-05,
      "loss": 0.2199,
      "num_input_tokens_seen": 12206272,
      "step": 18710
    },
    {
      "epoch": 9.808700209643606,
      "grad_norm": 1.6057674884796143,
      "learning_rate": 3.016289160476839e-05,
      "loss": 0.1485,
      "num_input_tokens_seen": 12209472,
      "step": 18715
    },
    {
      "epoch": 9.81132075471698,
      "grad_norm": 3.360506057739258,
      "learning_rate": 3.0151703265731125e-05,
      "loss": 0.178,
      "num_input_tokens_seen": 12212256,
      "step": 18720
    },
    {
      "epoch": 9.813941299790356,
      "grad_norm": 1.863788366317749,
      "learning_rate": 3.0140513849016305e-05,
      "loss": 0.1792,
      "num_input_tokens_seen": 12215456,
      "step": 18725
    },
    {
      "epoch": 9.816561844863731,
      "grad_norm": 1.391972541809082,
      "learning_rate": 3.0129323356964627e-05,
      "loss": 0.1525,
      "num_input_tokens_seen": 12219936,
      "step": 18730
    },
    {
      "epoch": 9.819182389937106,
      "grad_norm": 2.5574114322662354,
      "learning_rate": 3.0118131791917032e-05,
      "loss": 0.1407,
      "num_input_tokens_seen": 12223200,
      "step": 18735
    },
    {
      "epoch": 9.821802935010481,
      "grad_norm": 2.620016574859619,
      "learning_rate": 3.010693915621466e-05,
      "loss": 0.117,
      "num_input_tokens_seen": 12226432,
      "step": 18740
    },
    {
      "epoch": 9.824423480083858,
      "grad_norm": 1.886383295059204,
      "learning_rate": 3.009574545219887e-05,
      "loss": 0.1099,
      "num_input_tokens_seen": 12229120,
      "step": 18745
    },
    {
      "epoch": 9.827044025157234,
      "grad_norm": 2.88826060295105,
      "learning_rate": 3.0084550682211272e-05,
      "loss": 0.1101,
      "num_input_tokens_seen": 12233248,
      "step": 18750
    },
    {
      "epoch": 9.829664570230609,
      "grad_norm": 0.9343876242637634,
      "learning_rate": 3.0073354848593677e-05,
      "loss": 0.185,
      "num_input_tokens_seen": 12237696,
      "step": 18755
    },
    {
      "epoch": 9.832285115303984,
      "grad_norm": 2.2541422843933105,
      "learning_rate": 3.0062157953688125e-05,
      "loss": 0.1015,
      "num_input_tokens_seen": 12241504,
      "step": 18760
    },
    {
      "epoch": 9.834905660377359,
      "grad_norm": 1.546067714691162,
      "learning_rate": 3.0050959999836887e-05,
      "loss": 0.1287,
      "num_input_tokens_seen": 12244832,
      "step": 18765
    },
    {
      "epoch": 9.837526205450734,
      "grad_norm": 4.926446437835693,
      "learning_rate": 3.003976098938244e-05,
      "loss": 0.1138,
      "num_input_tokens_seen": 12247776,
      "step": 18770
    },
    {
      "epoch": 9.84014675052411,
      "grad_norm": 3.2491090297698975,
      "learning_rate": 3.0028560924667487e-05,
      "loss": 0.1513,
      "num_input_tokens_seen": 12250560,
      "step": 18775
    },
    {
      "epoch": 9.842767295597485,
      "grad_norm": 5.148683547973633,
      "learning_rate": 3.001735980803496e-05,
      "loss": 0.1863,
      "num_input_tokens_seen": 12253504,
      "step": 18780
    },
    {
      "epoch": 9.84538784067086,
      "grad_norm": 4.114687919616699,
      "learning_rate": 3.0006157641828004e-05,
      "loss": 0.1272,
      "num_input_tokens_seen": 12256128,
      "step": 18785
    },
    {
      "epoch": 9.848008385744235,
      "grad_norm": 1.0791139602661133,
      "learning_rate": 2.9994954428389977e-05,
      "loss": 0.2572,
      "num_input_tokens_seen": 12260032,
      "step": 18790
    },
    {
      "epoch": 9.85062893081761,
      "grad_norm": 1.4457712173461914,
      "learning_rate": 2.9983750170064472e-05,
      "loss": 0.1306,
      "num_input_tokens_seen": 12263136,
      "step": 18795
    },
    {
      "epoch": 9.853249475890985,
      "grad_norm": 4.99285888671875,
      "learning_rate": 2.9972544869195292e-05,
      "loss": 0.1481,
      "num_input_tokens_seen": 12266496,
      "step": 18800
    },
    {
      "epoch": 9.85587002096436,
      "grad_norm": 2.04868745803833,
      "learning_rate": 2.9961338528126447e-05,
      "loss": 0.131,
      "num_input_tokens_seen": 12268768,
      "step": 18805
    },
    {
      "epoch": 9.858490566037736,
      "grad_norm": 1.4051166772842407,
      "learning_rate": 2.9950131149202192e-05,
      "loss": 0.1623,
      "num_input_tokens_seen": 12272160,
      "step": 18810
    },
    {
      "epoch": 9.86111111111111,
      "grad_norm": 2.5819952487945557,
      "learning_rate": 2.9938922734766973e-05,
      "loss": 0.1494,
      "num_input_tokens_seen": 12275776,
      "step": 18815
    },
    {
      "epoch": 9.863731656184486,
      "grad_norm": 1.1185588836669922,
      "learning_rate": 2.992771328716546e-05,
      "loss": 0.1033,
      "num_input_tokens_seen": 12278880,
      "step": 18820
    },
    {
      "epoch": 9.866352201257861,
      "grad_norm": 5.401435375213623,
      "learning_rate": 2.991650280874255e-05,
      "loss": 0.1092,
      "num_input_tokens_seen": 12281536,
      "step": 18825
    },
    {
      "epoch": 9.868972746331236,
      "grad_norm": 1.9791313409805298,
      "learning_rate": 2.990529130184334e-05,
      "loss": 0.1777,
      "num_input_tokens_seen": 12284416,
      "step": 18830
    },
    {
      "epoch": 9.871593291404611,
      "grad_norm": 1.212364912033081,
      "learning_rate": 2.989407876881316e-05,
      "loss": 0.141,
      "num_input_tokens_seen": 12288224,
      "step": 18835
    },
    {
      "epoch": 9.874213836477988,
      "grad_norm": 2.727854013442993,
      "learning_rate": 2.988286521199753e-05,
      "loss": 0.1183,
      "num_input_tokens_seen": 12291520,
      "step": 18840
    },
    {
      "epoch": 9.876834381551364,
      "grad_norm": 2.6136467456817627,
      "learning_rate": 2.9871650633742198e-05,
      "loss": 0.1613,
      "num_input_tokens_seen": 12295264,
      "step": 18845
    },
    {
      "epoch": 9.879454926624739,
      "grad_norm": 1.2435115575790405,
      "learning_rate": 2.9860435036393142e-05,
      "loss": 0.1255,
      "num_input_tokens_seen": 12297984,
      "step": 18850
    },
    {
      "epoch": 9.882075471698114,
      "grad_norm": 16.383853912353516,
      "learning_rate": 2.9849218422296527e-05,
      "loss": 0.1353,
      "num_input_tokens_seen": 12301184,
      "step": 18855
    },
    {
      "epoch": 9.884696016771489,
      "grad_norm": 4.6951494216918945,
      "learning_rate": 2.983800079379873e-05,
      "loss": 0.2143,
      "num_input_tokens_seen": 12303968,
      "step": 18860
    },
    {
      "epoch": 9.887316561844864,
      "grad_norm": 1.900771975517273,
      "learning_rate": 2.982678215324638e-05,
      "loss": 0.0895,
      "num_input_tokens_seen": 12308256,
      "step": 18865
    },
    {
      "epoch": 9.88993710691824,
      "grad_norm": 3.1792070865631104,
      "learning_rate": 2.9815562502986256e-05,
      "loss": 0.1646,
      "num_input_tokens_seen": 12310816,
      "step": 18870
    },
    {
      "epoch": 9.892557651991615,
      "grad_norm": 1.5040051937103271,
      "learning_rate": 2.9804341845365398e-05,
      "loss": 0.135,
      "num_input_tokens_seen": 12314752,
      "step": 18875
    },
    {
      "epoch": 9.89517819706499,
      "grad_norm": 3.8824353218078613,
      "learning_rate": 2.9793120182731037e-05,
      "loss": 0.1479,
      "num_input_tokens_seen": 12318272,
      "step": 18880
    },
    {
      "epoch": 9.897798742138365,
      "grad_norm": 1.8093374967575073,
      "learning_rate": 2.9781897517430617e-05,
      "loss": 0.1691,
      "num_input_tokens_seen": 12322336,
      "step": 18885
    },
    {
      "epoch": 9.90041928721174,
      "grad_norm": 1.9848707914352417,
      "learning_rate": 2.977067385181179e-05,
      "loss": 0.1367,
      "num_input_tokens_seen": 12325760,
      "step": 18890
    },
    {
      "epoch": 9.903039832285115,
      "grad_norm": 7.2974982261657715,
      "learning_rate": 2.9759449188222426e-05,
      "loss": 0.22,
      "num_input_tokens_seen": 12330304,
      "step": 18895
    },
    {
      "epoch": 9.90566037735849,
      "grad_norm": 3.8917882442474365,
      "learning_rate": 2.9748223529010582e-05,
      "loss": 0.1322,
      "num_input_tokens_seen": 12332960,
      "step": 18900
    },
    {
      "epoch": 9.908280922431866,
      "grad_norm": 2.656193256378174,
      "learning_rate": 2.9736996876524546e-05,
      "loss": 0.1926,
      "num_input_tokens_seen": 12337536,
      "step": 18905
    },
    {
      "epoch": 9.91090146750524,
      "grad_norm": 9.072586059570312,
      "learning_rate": 2.9725769233112816e-05,
      "loss": 0.1479,
      "num_input_tokens_seen": 12341120,
      "step": 18910
    },
    {
      "epoch": 9.913522012578616,
      "grad_norm": 3.1326510906219482,
      "learning_rate": 2.9714540601124074e-05,
      "loss": 0.1327,
      "num_input_tokens_seen": 12343904,
      "step": 18915
    },
    {
      "epoch": 9.916142557651991,
      "grad_norm": 2.8016250133514404,
      "learning_rate": 2.9703310982907218e-05,
      "loss": 0.1747,
      "num_input_tokens_seen": 12346912,
      "step": 18920
    },
    {
      "epoch": 9.918763102725366,
      "grad_norm": 3.269914388656616,
      "learning_rate": 2.9692080380811365e-05,
      "loss": 0.097,
      "num_input_tokens_seen": 12350272,
      "step": 18925
    },
    {
      "epoch": 9.921383647798741,
      "grad_norm": 3.5559418201446533,
      "learning_rate": 2.9680848797185828e-05,
      "loss": 0.2303,
      "num_input_tokens_seen": 12352800,
      "step": 18930
    },
    {
      "epoch": 9.924004192872118,
      "grad_norm": 1.6159560680389404,
      "learning_rate": 2.966961623438012e-05,
      "loss": 0.1149,
      "num_input_tokens_seen": 12355616,
      "step": 18935
    },
    {
      "epoch": 9.926624737945493,
      "grad_norm": 1.6791996955871582,
      "learning_rate": 2.9658382694743973e-05,
      "loss": 0.1426,
      "num_input_tokens_seen": 12358784,
      "step": 18940
    },
    {
      "epoch": 9.929245283018869,
      "grad_norm": 2.94187331199646,
      "learning_rate": 2.9647148180627317e-05,
      "loss": 0.1705,
      "num_input_tokens_seen": 12361920,
      "step": 18945
    },
    {
      "epoch": 9.931865828092244,
      "grad_norm": 3.7829484939575195,
      "learning_rate": 2.9635912694380275e-05,
      "loss": 0.1881,
      "num_input_tokens_seen": 12365600,
      "step": 18950
    },
    {
      "epoch": 9.934486373165619,
      "grad_norm": 3.0970563888549805,
      "learning_rate": 2.962467623835319e-05,
      "loss": 0.1838,
      "num_input_tokens_seen": 12368000,
      "step": 18955
    },
    {
      "epoch": 9.937106918238994,
      "grad_norm": 1.0232508182525635,
      "learning_rate": 2.961343881489659e-05,
      "loss": 0.1792,
      "num_input_tokens_seen": 12370592,
      "step": 18960
    },
    {
      "epoch": 9.93972746331237,
      "grad_norm": 1.7843968868255615,
      "learning_rate": 2.9602200426361226e-05,
      "loss": 0.1681,
      "num_input_tokens_seen": 12373760,
      "step": 18965
    },
    {
      "epoch": 9.942348008385745,
      "grad_norm": 2.6873223781585693,
      "learning_rate": 2.9590961075098038e-05,
      "loss": 0.1441,
      "num_input_tokens_seen": 12377600,
      "step": 18970
    },
    {
      "epoch": 9.94496855345912,
      "grad_norm": 8.4054594039917,
      "learning_rate": 2.957972076345817e-05,
      "loss": 0.1138,
      "num_input_tokens_seen": 12381152,
      "step": 18975
    },
    {
      "epoch": 9.947589098532495,
      "grad_norm": 2.8098597526550293,
      "learning_rate": 2.9568479493792967e-05,
      "loss": 0.1373,
      "num_input_tokens_seen": 12384096,
      "step": 18980
    },
    {
      "epoch": 9.95020964360587,
      "grad_norm": 3.103639841079712,
      "learning_rate": 2.955723726845397e-05,
      "loss": 0.1436,
      "num_input_tokens_seen": 12386848,
      "step": 18985
    },
    {
      "epoch": 9.952830188679245,
      "grad_norm": 3.1903815269470215,
      "learning_rate": 2.9545994089792928e-05,
      "loss": 0.1543,
      "num_input_tokens_seen": 12389344,
      "step": 18990
    },
    {
      "epoch": 9.95545073375262,
      "grad_norm": 2.5677080154418945,
      "learning_rate": 2.9534749960161795e-05,
      "loss": 0.1335,
      "num_input_tokens_seen": 12392640,
      "step": 18995
    },
    {
      "epoch": 9.958071278825996,
      "grad_norm": 2.751044988632202,
      "learning_rate": 2.9523504881912694e-05,
      "loss": 0.1483,
      "num_input_tokens_seen": 12397248,
      "step": 19000
    },
    {
      "epoch": 9.96069182389937,
      "grad_norm": 2.1231203079223633,
      "learning_rate": 2.9512258857397978e-05,
      "loss": 0.1415,
      "num_input_tokens_seen": 12400352,
      "step": 19005
    },
    {
      "epoch": 9.963312368972746,
      "grad_norm": 1.7048921585083008,
      "learning_rate": 2.9501011888970203e-05,
      "loss": 0.0988,
      "num_input_tokens_seen": 12402944,
      "step": 19010
    },
    {
      "epoch": 9.965932914046121,
      "grad_norm": 4.167205333709717,
      "learning_rate": 2.9489763978982075e-05,
      "loss": 0.1218,
      "num_input_tokens_seen": 12405952,
      "step": 19015
    },
    {
      "epoch": 9.968553459119496,
      "grad_norm": 1.8150699138641357,
      "learning_rate": 2.9478515129786556e-05,
      "loss": 0.2106,
      "num_input_tokens_seen": 12409376,
      "step": 19020
    },
    {
      "epoch": 9.971174004192871,
      "grad_norm": 2.6504998207092285,
      "learning_rate": 2.946726534373676e-05,
      "loss": 0.1459,
      "num_input_tokens_seen": 12413216,
      "step": 19025
    },
    {
      "epoch": 9.973794549266248,
      "grad_norm": 2.1173455715179443,
      "learning_rate": 2.945601462318601e-05,
      "loss": 0.1096,
      "num_input_tokens_seen": 12416512,
      "step": 19030
    },
    {
      "epoch": 9.976415094339622,
      "grad_norm": 2.09895396232605,
      "learning_rate": 2.944476297048786e-05,
      "loss": 0.1696,
      "num_input_tokens_seen": 12420192,
      "step": 19035
    },
    {
      "epoch": 9.979035639412999,
      "grad_norm": 7.111752033233643,
      "learning_rate": 2.9433510387995987e-05,
      "loss": 0.1519,
      "num_input_tokens_seen": 12422464,
      "step": 19040
    },
    {
      "epoch": 9.981656184486374,
      "grad_norm": 1.8234740495681763,
      "learning_rate": 2.9422256878064325e-05,
      "loss": 0.1711,
      "num_input_tokens_seen": 12425248,
      "step": 19045
    },
    {
      "epoch": 9.984276729559749,
      "grad_norm": 3.0055813789367676,
      "learning_rate": 2.9411002443046982e-05,
      "loss": 0.1318,
      "num_input_tokens_seen": 12428704,
      "step": 19050
    },
    {
      "epoch": 9.986897274633124,
      "grad_norm": 2.7304270267486572,
      "learning_rate": 2.9399747085298245e-05,
      "loss": 0.176,
      "num_input_tokens_seen": 12432480,
      "step": 19055
    },
    {
      "epoch": 9.9895178197065,
      "grad_norm": 3.267474889755249,
      "learning_rate": 2.9388490807172614e-05,
      "loss": 0.1645,
      "num_input_tokens_seen": 12434912,
      "step": 19060
    },
    {
      "epoch": 9.992138364779874,
      "grad_norm": 4.156236171722412,
      "learning_rate": 2.9377233611024775e-05,
      "loss": 0.117,
      "num_input_tokens_seen": 12438016,
      "step": 19065
    },
    {
      "epoch": 9.99475890985325,
      "grad_norm": 3.7960705757141113,
      "learning_rate": 2.9365975499209596e-05,
      "loss": 0.1569,
      "num_input_tokens_seen": 12440672,
      "step": 19070
    },
    {
      "epoch": 9.997379454926625,
      "grad_norm": 2.1026451587677,
      "learning_rate": 2.9354716474082154e-05,
      "loss": 0.1606,
      "num_input_tokens_seen": 12443840,
      "step": 19075
    },
    {
      "epoch": 10.0,
      "grad_norm": 2.9558794498443604,
      "learning_rate": 2.9343456537997714e-05,
      "loss": 0.1702,
      "num_input_tokens_seen": 12446408,
      "step": 19080
    },
    {
      "epoch": 10.0,
      "eval_loss": 0.7744925022125244,
      "eval_runtime": 11.5781,
      "eval_samples_per_second": 73.241,
      "eval_steps_per_second": 18.31,
      "num_input_tokens_seen": 12446408,
      "step": 19080
    },
    {
      "epoch": 10.002620545073375,
      "grad_norm": 1.3176683187484741,
      "learning_rate": 2.9332195693311714e-05,
      "loss": 0.08,
      "num_input_tokens_seen": 12450472,
      "step": 19085
    },
    {
      "epoch": 10.00524109014675,
      "grad_norm": 1.9946550130844116,
      "learning_rate": 2.9320933942379803e-05,
      "loss": 0.0703,
      "num_input_tokens_seen": 12454056,
      "step": 19090
    },
    {
      "epoch": 10.007861635220126,
      "grad_norm": 1.6485408544540405,
      "learning_rate": 2.9309671287557805e-05,
      "loss": 0.0978,
      "num_input_tokens_seen": 12457576,
      "step": 19095
    },
    {
      "epoch": 10.0104821802935,
      "grad_norm": 1.7243224382400513,
      "learning_rate": 2.9298407731201737e-05,
      "loss": 0.0901,
      "num_input_tokens_seen": 12461608,
      "step": 19100
    },
    {
      "epoch": 10.013102725366876,
      "grad_norm": 1.6120023727416992,
      "learning_rate": 2.928714327566781e-05,
      "loss": 0.0758,
      "num_input_tokens_seen": 12464968,
      "step": 19105
    },
    {
      "epoch": 10.015723270440251,
      "grad_norm": 3.675417900085449,
      "learning_rate": 2.9275877923312427e-05,
      "loss": 0.1086,
      "num_input_tokens_seen": 12468648,
      "step": 19110
    },
    {
      "epoch": 10.018343815513626,
      "grad_norm": 1.411941647529602,
      "learning_rate": 2.9264611676492165e-05,
      "loss": 0.1171,
      "num_input_tokens_seen": 12471784,
      "step": 19115
    },
    {
      "epoch": 10.020964360587001,
      "grad_norm": 1.3108415603637695,
      "learning_rate": 2.9253344537563782e-05,
      "loss": 0.15,
      "num_input_tokens_seen": 12474632,
      "step": 19120
    },
    {
      "epoch": 10.023584905660377,
      "grad_norm": 1.575670599937439,
      "learning_rate": 2.924207650888426e-05,
      "loss": 0.1782,
      "num_input_tokens_seen": 12477832,
      "step": 19125
    },
    {
      "epoch": 10.026205450733753,
      "grad_norm": 1.9763786792755127,
      "learning_rate": 2.9230807592810723e-05,
      "loss": 0.1367,
      "num_input_tokens_seen": 12482312,
      "step": 19130
    },
    {
      "epoch": 10.028825995807129,
      "grad_norm": 2.0625672340393066,
      "learning_rate": 2.92195377917005e-05,
      "loss": 0.1078,
      "num_input_tokens_seen": 12485768,
      "step": 19135
    },
    {
      "epoch": 10.031446540880504,
      "grad_norm": 1.6691006422042847,
      "learning_rate": 2.920826710791112e-05,
      "loss": 0.0923,
      "num_input_tokens_seen": 12488776,
      "step": 19140
    },
    {
      "epoch": 10.034067085953879,
      "grad_norm": 1.4594815969467163,
      "learning_rate": 2.919699554380026e-05,
      "loss": 0.0926,
      "num_input_tokens_seen": 12491464,
      "step": 19145
    },
    {
      "epoch": 10.036687631027254,
      "grad_norm": 1.7257260084152222,
      "learning_rate": 2.9185723101725825e-05,
      "loss": 0.0859,
      "num_input_tokens_seen": 12494600,
      "step": 19150
    },
    {
      "epoch": 10.03930817610063,
      "grad_norm": 5.908061504364014,
      "learning_rate": 2.9174449784045867e-05,
      "loss": 0.1513,
      "num_input_tokens_seen": 12498696,
      "step": 19155
    },
    {
      "epoch": 10.041928721174004,
      "grad_norm": 2.956681251525879,
      "learning_rate": 2.916317559311863e-05,
      "loss": 0.0886,
      "num_input_tokens_seen": 12501960,
      "step": 19160
    },
    {
      "epoch": 10.04454926624738,
      "grad_norm": 5.179711818695068,
      "learning_rate": 2.9151900531302566e-05,
      "loss": 0.169,
      "num_input_tokens_seen": 12505224,
      "step": 19165
    },
    {
      "epoch": 10.047169811320755,
      "grad_norm": 1.3111562728881836,
      "learning_rate": 2.9140624600956272e-05,
      "loss": 0.105,
      "num_input_tokens_seen": 12508744,
      "step": 19170
    },
    {
      "epoch": 10.04979035639413,
      "grad_norm": 2.4287588596343994,
      "learning_rate": 2.912934780443855e-05,
      "loss": 0.131,
      "num_input_tokens_seen": 12511464,
      "step": 19175
    },
    {
      "epoch": 10.052410901467505,
      "grad_norm": 1.9512068033218384,
      "learning_rate": 2.9118070144108388e-05,
      "loss": 0.1066,
      "num_input_tokens_seen": 12513736,
      "step": 19180
    },
    {
      "epoch": 10.05503144654088,
      "grad_norm": 4.176904201507568,
      "learning_rate": 2.9106791622324924e-05,
      "loss": 0.0814,
      "num_input_tokens_seen": 12516584,
      "step": 19185
    },
    {
      "epoch": 10.057651991614255,
      "grad_norm": 2.9277212619781494,
      "learning_rate": 2.909551224144751e-05,
      "loss": 0.0717,
      "num_input_tokens_seen": 12520680,
      "step": 19190
    },
    {
      "epoch": 10.06027253668763,
      "grad_norm": 2.0331149101257324,
      "learning_rate": 2.9084232003835653e-05,
      "loss": 0.0974,
      "num_input_tokens_seen": 12523624,
      "step": 19195
    },
    {
      "epoch": 10.062893081761006,
      "grad_norm": 5.41481351852417,
      "learning_rate": 2.9072950911849058e-05,
      "loss": 0.0861,
      "num_input_tokens_seen": 12526472,
      "step": 19200
    },
    {
      "epoch": 10.065513626834381,
      "grad_norm": 2.374115467071533,
      "learning_rate": 2.90616689678476e-05,
      "loss": 0.1083,
      "num_input_tokens_seen": 12529768,
      "step": 19205
    },
    {
      "epoch": 10.068134171907756,
      "grad_norm": 1.7005506753921509,
      "learning_rate": 2.9050386174191337e-05,
      "loss": 0.1231,
      "num_input_tokens_seen": 12532680,
      "step": 19210
    },
    {
      "epoch": 10.070754716981131,
      "grad_norm": 2.181854486465454,
      "learning_rate": 2.903910253324048e-05,
      "loss": 0.1299,
      "num_input_tokens_seen": 12535944,
      "step": 19215
    },
    {
      "epoch": 10.073375262054507,
      "grad_norm": 2.5591623783111572,
      "learning_rate": 2.9027818047355477e-05,
      "loss": 0.1027,
      "num_input_tokens_seen": 12540008,
      "step": 19220
    },
    {
      "epoch": 10.075995807127883,
      "grad_norm": 2.609504222869873,
      "learning_rate": 2.9016532718896873e-05,
      "loss": 0.0733,
      "num_input_tokens_seen": 12542952,
      "step": 19225
    },
    {
      "epoch": 10.078616352201259,
      "grad_norm": 5.903504848480225,
      "learning_rate": 2.900524655022545e-05,
      "loss": 0.1079,
      "num_input_tokens_seen": 12545672,
      "step": 19230
    },
    {
      "epoch": 10.081236897274634,
      "grad_norm": 1.9040652513504028,
      "learning_rate": 2.899395954370215e-05,
      "loss": 0.0796,
      "num_input_tokens_seen": 12548808,
      "step": 19235
    },
    {
      "epoch": 10.083857442348009,
      "grad_norm": 5.2340087890625,
      "learning_rate": 2.898267170168807e-05,
      "loss": 0.1039,
      "num_input_tokens_seen": 12551944,
      "step": 19240
    },
    {
      "epoch": 10.086477987421384,
      "grad_norm": 1.0645755529403687,
      "learning_rate": 2.8971383026544507e-05,
      "loss": 0.0944,
      "num_input_tokens_seen": 12555048,
      "step": 19245
    },
    {
      "epoch": 10.08909853249476,
      "grad_norm": 0.8089860081672668,
      "learning_rate": 2.896009352063293e-05,
      "loss": 0.0583,
      "num_input_tokens_seen": 12558312,
      "step": 19250
    },
    {
      "epoch": 10.091719077568134,
      "grad_norm": 1.8036081790924072,
      "learning_rate": 2.894880318631496e-05,
      "loss": 0.0992,
      "num_input_tokens_seen": 12561992,
      "step": 19255
    },
    {
      "epoch": 10.09433962264151,
      "grad_norm": 0.6833407282829285,
      "learning_rate": 2.8937512025952417e-05,
      "loss": 0.0998,
      "num_input_tokens_seen": 12565544,
      "step": 19260
    },
    {
      "epoch": 10.096960167714885,
      "grad_norm": 3.0588786602020264,
      "learning_rate": 2.8926220041907283e-05,
      "loss": 0.1456,
      "num_input_tokens_seen": 12568232,
      "step": 19265
    },
    {
      "epoch": 10.09958071278826,
      "grad_norm": 10.485493659973145,
      "learning_rate": 2.8914927236541706e-05,
      "loss": 0.0848,
      "num_input_tokens_seen": 12570568,
      "step": 19270
    },
    {
      "epoch": 10.102201257861635,
      "grad_norm": 0.9601419568061829,
      "learning_rate": 2.8903633612218012e-05,
      "loss": 0.0703,
      "num_input_tokens_seen": 12573448,
      "step": 19275
    },
    {
      "epoch": 10.10482180293501,
      "grad_norm": 1.8828089237213135,
      "learning_rate": 2.8892339171298703e-05,
      "loss": 0.0815,
      "num_input_tokens_seen": 12577256,
      "step": 19280
    },
    {
      "epoch": 10.107442348008385,
      "grad_norm": 1.4403151273727417,
      "learning_rate": 2.888104391614645e-05,
      "loss": 0.0793,
      "num_input_tokens_seen": 12580808,
      "step": 19285
    },
    {
      "epoch": 10.11006289308176,
      "grad_norm": 1.384382724761963,
      "learning_rate": 2.886974784912408e-05,
      "loss": 0.0984,
      "num_input_tokens_seen": 12585896,
      "step": 19290
    },
    {
      "epoch": 10.112683438155136,
      "grad_norm": 2.9278223514556885,
      "learning_rate": 2.8858450972594612e-05,
      "loss": 0.1014,
      "num_input_tokens_seen": 12589000,
      "step": 19295
    },
    {
      "epoch": 10.115303983228511,
      "grad_norm": 2.498228073120117,
      "learning_rate": 2.8847153288921226e-05,
      "loss": 0.117,
      "num_input_tokens_seen": 12592104,
      "step": 19300
    },
    {
      "epoch": 10.117924528301886,
      "grad_norm": 1.3207834959030151,
      "learning_rate": 2.883585480046725e-05,
      "loss": 0.1015,
      "num_input_tokens_seen": 12595240,
      "step": 19305
    },
    {
      "epoch": 10.120545073375261,
      "grad_norm": 2.025364637374878,
      "learning_rate": 2.8824555509596223e-05,
      "loss": 0.0682,
      "num_input_tokens_seen": 12597480,
      "step": 19310
    },
    {
      "epoch": 10.123165618448636,
      "grad_norm": 1.5529725551605225,
      "learning_rate": 2.8813255418671815e-05,
      "loss": 0.1242,
      "num_input_tokens_seen": 12600328,
      "step": 19315
    },
    {
      "epoch": 10.125786163522013,
      "grad_norm": 1.9658182859420776,
      "learning_rate": 2.8801954530057874e-05,
      "loss": 0.0881,
      "num_input_tokens_seen": 12602152,
      "step": 19320
    },
    {
      "epoch": 10.128406708595389,
      "grad_norm": 4.64812707901001,
      "learning_rate": 2.879065284611842e-05,
      "loss": 0.101,
      "num_input_tokens_seen": 12605352,
      "step": 19325
    },
    {
      "epoch": 10.131027253668764,
      "grad_norm": 1.5554955005645752,
      "learning_rate": 2.8779350369217634e-05,
      "loss": 0.0663,
      "num_input_tokens_seen": 12607784,
      "step": 19330
    },
    {
      "epoch": 10.133647798742139,
      "grad_norm": 3.634164571762085,
      "learning_rate": 2.876804710171987e-05,
      "loss": 0.1308,
      "num_input_tokens_seen": 12614056,
      "step": 19335
    },
    {
      "epoch": 10.136268343815514,
      "grad_norm": 2.6198930740356445,
      "learning_rate": 2.8756743045989643e-05,
      "loss": 0.1058,
      "num_input_tokens_seen": 12616552,
      "step": 19340
    },
    {
      "epoch": 10.13888888888889,
      "grad_norm": 2.325732469558716,
      "learning_rate": 2.8745438204391623e-05,
      "loss": 0.1027,
      "num_input_tokens_seen": 12619976,
      "step": 19345
    },
    {
      "epoch": 10.141509433962264,
      "grad_norm": 2.3005313873291016,
      "learning_rate": 2.8734132579290675e-05,
      "loss": 0.1005,
      "num_input_tokens_seen": 12623272,
      "step": 19350
    },
    {
      "epoch": 10.14412997903564,
      "grad_norm": 2.5846571922302246,
      "learning_rate": 2.8722826173051782e-05,
      "loss": 0.0912,
      "num_input_tokens_seen": 12625960,
      "step": 19355
    },
    {
      "epoch": 10.146750524109015,
      "grad_norm": 3.038029193878174,
      "learning_rate": 2.871151898804012e-05,
      "loss": 0.1133,
      "num_input_tokens_seen": 12630184,
      "step": 19360
    },
    {
      "epoch": 10.14937106918239,
      "grad_norm": 2.1382503509521484,
      "learning_rate": 2.8700211026621043e-05,
      "loss": 0.1294,
      "num_input_tokens_seen": 12633224,
      "step": 19365
    },
    {
      "epoch": 10.151991614255765,
      "grad_norm": 2.2935197353363037,
      "learning_rate": 2.8688902291160025e-05,
      "loss": 0.0939,
      "num_input_tokens_seen": 12636584,
      "step": 19370
    },
    {
      "epoch": 10.15461215932914,
      "grad_norm": 2.799645185470581,
      "learning_rate": 2.867759278402273e-05,
      "loss": 0.0976,
      "num_input_tokens_seen": 12639944,
      "step": 19375
    },
    {
      "epoch": 10.157232704402515,
      "grad_norm": 3.343681573867798,
      "learning_rate": 2.8666282507574994e-05,
      "loss": 0.1395,
      "num_input_tokens_seen": 12643624,
      "step": 19380
    },
    {
      "epoch": 10.15985324947589,
      "grad_norm": 1.3892109394073486,
      "learning_rate": 2.8654971464182773e-05,
      "loss": 0.0718,
      "num_input_tokens_seen": 12646792,
      "step": 19385
    },
    {
      "epoch": 10.162473794549266,
      "grad_norm": 2.284339189529419,
      "learning_rate": 2.8643659656212222e-05,
      "loss": 0.1489,
      "num_input_tokens_seen": 12650696,
      "step": 19390
    },
    {
      "epoch": 10.165094339622641,
      "grad_norm": 2.2342982292175293,
      "learning_rate": 2.8632347086029642e-05,
      "loss": 0.1127,
      "num_input_tokens_seen": 12653512,
      "step": 19395
    },
    {
      "epoch": 10.167714884696016,
      "grad_norm": 2.948185443878174,
      "learning_rate": 2.8621033756001486e-05,
      "loss": 0.106,
      "num_input_tokens_seen": 12657160,
      "step": 19400
    },
    {
      "epoch": 10.170335429769391,
      "grad_norm": 2.06465744972229,
      "learning_rate": 2.8609719668494394e-05,
      "loss": 0.1181,
      "num_input_tokens_seen": 12661096,
      "step": 19405
    },
    {
      "epoch": 10.172955974842766,
      "grad_norm": 1.8972054719924927,
      "learning_rate": 2.8598404825875113e-05,
      "loss": 0.0728,
      "num_input_tokens_seen": 12664168,
      "step": 19410
    },
    {
      "epoch": 10.175576519916143,
      "grad_norm": 2.030916213989258,
      "learning_rate": 2.8587089230510604e-05,
      "loss": 0.089,
      "num_input_tokens_seen": 12667304,
      "step": 19415
    },
    {
      "epoch": 10.178197064989519,
      "grad_norm": 2.034014940261841,
      "learning_rate": 2.857577288476795e-05,
      "loss": 0.0933,
      "num_input_tokens_seen": 12670760,
      "step": 19420
    },
    {
      "epoch": 10.180817610062894,
      "grad_norm": 2.5462183952331543,
      "learning_rate": 2.85644557910144e-05,
      "loss": 0.1347,
      "num_input_tokens_seen": 12673384,
      "step": 19425
    },
    {
      "epoch": 10.183438155136269,
      "grad_norm": 1.8496750593185425,
      "learning_rate": 2.8553137951617366e-05,
      "loss": 0.1389,
      "num_input_tokens_seen": 12676168,
      "step": 19430
    },
    {
      "epoch": 10.186058700209644,
      "grad_norm": 0.5345591902732849,
      "learning_rate": 2.854181936894441e-05,
      "loss": 0.0538,
      "num_input_tokens_seen": 12678984,
      "step": 19435
    },
    {
      "epoch": 10.18867924528302,
      "grad_norm": 2.852572441101074,
      "learning_rate": 2.8530500045363247e-05,
      "loss": 0.2011,
      "num_input_tokens_seen": 12682600,
      "step": 19440
    },
    {
      "epoch": 10.191299790356394,
      "grad_norm": 1.8644851446151733,
      "learning_rate": 2.851917998324175e-05,
      "loss": 0.0931,
      "num_input_tokens_seen": 12685704,
      "step": 19445
    },
    {
      "epoch": 10.19392033542977,
      "grad_norm": 2.8808696269989014,
      "learning_rate": 2.8507859184947957e-05,
      "loss": 0.081,
      "num_input_tokens_seen": 12689256,
      "step": 19450
    },
    {
      "epoch": 10.196540880503145,
      "grad_norm": 1.6990561485290527,
      "learning_rate": 2.8496537652850043e-05,
      "loss": 0.0941,
      "num_input_tokens_seen": 12692968,
      "step": 19455
    },
    {
      "epoch": 10.19916142557652,
      "grad_norm": 3.506253242492676,
      "learning_rate": 2.8485215389316333e-05,
      "loss": 0.0882,
      "num_input_tokens_seen": 12696712,
      "step": 19460
    },
    {
      "epoch": 10.201781970649895,
      "grad_norm": 3.489452362060547,
      "learning_rate": 2.8473892396715328e-05,
      "loss": 0.1104,
      "num_input_tokens_seen": 12699848,
      "step": 19465
    },
    {
      "epoch": 10.20440251572327,
      "grad_norm": 2.0825798511505127,
      "learning_rate": 2.846256867741567e-05,
      "loss": 0.0845,
      "num_input_tokens_seen": 12703176,
      "step": 19470
    },
    {
      "epoch": 10.207023060796645,
      "grad_norm": 1.53977370262146,
      "learning_rate": 2.845124423378614e-05,
      "loss": 0.1148,
      "num_input_tokens_seen": 12706280,
      "step": 19475
    },
    {
      "epoch": 10.20964360587002,
      "grad_norm": 10.485526084899902,
      "learning_rate": 2.8439919068195697e-05,
      "loss": 0.1614,
      "num_input_tokens_seen": 12709224,
      "step": 19480
    },
    {
      "epoch": 10.212264150943396,
      "grad_norm": 1.6391068696975708,
      "learning_rate": 2.842859318301343e-05,
      "loss": 0.102,
      "num_input_tokens_seen": 12713064,
      "step": 19485
    },
    {
      "epoch": 10.214884696016771,
      "grad_norm": 3.2058277130126953,
      "learning_rate": 2.841726658060858e-05,
      "loss": 0.1227,
      "num_input_tokens_seen": 12716424,
      "step": 19490
    },
    {
      "epoch": 10.217505241090146,
      "grad_norm": 3.1342854499816895,
      "learning_rate": 2.8405939263350555e-05,
      "loss": 0.0877,
      "num_input_tokens_seen": 12719816,
      "step": 19495
    },
    {
      "epoch": 10.220125786163521,
      "grad_norm": 1.998641014099121,
      "learning_rate": 2.8394611233608886e-05,
      "loss": 0.0996,
      "num_input_tokens_seen": 12723624,
      "step": 19500
    },
    {
      "epoch": 10.222746331236896,
      "grad_norm": 2.6808922290802,
      "learning_rate": 2.8383282493753283e-05,
      "loss": 0.1305,
      "num_input_tokens_seen": 12727240,
      "step": 19505
    },
    {
      "epoch": 10.225366876310272,
      "grad_norm": 2.336766004562378,
      "learning_rate": 2.8371953046153575e-05,
      "loss": 0.0927,
      "num_input_tokens_seen": 12730792,
      "step": 19510
    },
    {
      "epoch": 10.227987421383649,
      "grad_norm": 2.996880531311035,
      "learning_rate": 2.8360622893179757e-05,
      "loss": 0.1175,
      "num_input_tokens_seen": 12733928,
      "step": 19515
    },
    {
      "epoch": 10.230607966457024,
      "grad_norm": 1.1830430030822754,
      "learning_rate": 2.834929203720197e-05,
      "loss": 0.0818,
      "num_input_tokens_seen": 12737992,
      "step": 19520
    },
    {
      "epoch": 10.233228511530399,
      "grad_norm": 1.5850952863693237,
      "learning_rate": 2.83379604805905e-05,
      "loss": 0.1158,
      "num_input_tokens_seen": 12740456,
      "step": 19525
    },
    {
      "epoch": 10.235849056603774,
      "grad_norm": 6.695531845092773,
      "learning_rate": 2.832662822571578e-05,
      "loss": 0.0821,
      "num_input_tokens_seen": 12743496,
      "step": 19530
    },
    {
      "epoch": 10.23846960167715,
      "grad_norm": 2.375149726867676,
      "learning_rate": 2.8315295274948395e-05,
      "loss": 0.0913,
      "num_input_tokens_seen": 12746952,
      "step": 19535
    },
    {
      "epoch": 10.241090146750524,
      "grad_norm": 3.5244522094726562,
      "learning_rate": 2.8303961630659047e-05,
      "loss": 0.1481,
      "num_input_tokens_seen": 12749800,
      "step": 19540
    },
    {
      "epoch": 10.2437106918239,
      "grad_norm": 2.9615533351898193,
      "learning_rate": 2.8292627295218627e-05,
      "loss": 0.1537,
      "num_input_tokens_seen": 12752872,
      "step": 19545
    },
    {
      "epoch": 10.246331236897275,
      "grad_norm": 2.3490564823150635,
      "learning_rate": 2.828129227099815e-05,
      "loss": 0.0789,
      "num_input_tokens_seen": 12755944,
      "step": 19550
    },
    {
      "epoch": 10.24895178197065,
      "grad_norm": 1.4067695140838623,
      "learning_rate": 2.826995656036875e-05,
      "loss": 0.0775,
      "num_input_tokens_seen": 12758344,
      "step": 19555
    },
    {
      "epoch": 10.251572327044025,
      "grad_norm": 1.0955578088760376,
      "learning_rate": 2.8258620165701747e-05,
      "loss": 0.0482,
      "num_input_tokens_seen": 12762312,
      "step": 19560
    },
    {
      "epoch": 10.2541928721174,
      "grad_norm": 2.0779612064361572,
      "learning_rate": 2.8247283089368586e-05,
      "loss": 0.0993,
      "num_input_tokens_seen": 12765416,
      "step": 19565
    },
    {
      "epoch": 10.256813417190775,
      "grad_norm": 3.1882190704345703,
      "learning_rate": 2.8235945333740842e-05,
      "loss": 0.0492,
      "num_input_tokens_seen": 12768552,
      "step": 19570
    },
    {
      "epoch": 10.25943396226415,
      "grad_norm": 6.325082302093506,
      "learning_rate": 2.822460690119026e-05,
      "loss": 0.0872,
      "num_input_tokens_seen": 12770920,
      "step": 19575
    },
    {
      "epoch": 10.262054507337526,
      "grad_norm": 1.4026174545288086,
      "learning_rate": 2.82132677940887e-05,
      "loss": 0.0933,
      "num_input_tokens_seen": 12774184,
      "step": 19580
    },
    {
      "epoch": 10.264675052410901,
      "grad_norm": 2.00113844871521,
      "learning_rate": 2.820192801480817e-05,
      "loss": 0.1108,
      "num_input_tokens_seen": 12777288,
      "step": 19585
    },
    {
      "epoch": 10.267295597484276,
      "grad_norm": 1.2483059167861938,
      "learning_rate": 2.8190587565720842e-05,
      "loss": 0.0794,
      "num_input_tokens_seen": 12780136,
      "step": 19590
    },
    {
      "epoch": 10.269916142557651,
      "grad_norm": 2.5535356998443604,
      "learning_rate": 2.8179246449198986e-05,
      "loss": 0.139,
      "num_input_tokens_seen": 12783144,
      "step": 19595
    },
    {
      "epoch": 10.272536687631026,
      "grad_norm": 2.741699457168579,
      "learning_rate": 2.8167904667615048e-05,
      "loss": 0.0431,
      "num_input_tokens_seen": 12785992,
      "step": 19600
    },
    {
      "epoch": 10.275157232704402,
      "grad_norm": 1.2711280584335327,
      "learning_rate": 2.8156562223341592e-05,
      "loss": 0.142,
      "num_input_tokens_seen": 12789000,
      "step": 19605
    },
    {
      "epoch": 10.277777777777779,
      "grad_norm": 2.180676221847534,
      "learning_rate": 2.814521911875133e-05,
      "loss": 0.1136,
      "num_input_tokens_seen": 12791976,
      "step": 19610
    },
    {
      "epoch": 10.280398322851154,
      "grad_norm": 1.466849446296692,
      "learning_rate": 2.8133875356217114e-05,
      "loss": 0.112,
      "num_input_tokens_seen": 12795976,
      "step": 19615
    },
    {
      "epoch": 10.283018867924529,
      "grad_norm": 0.7989209890365601,
      "learning_rate": 2.8122530938111923e-05,
      "loss": 0.1294,
      "num_input_tokens_seen": 12798984,
      "step": 19620
    },
    {
      "epoch": 10.285639412997904,
      "grad_norm": 3.619252920150757,
      "learning_rate": 2.8111185866808888e-05,
      "loss": 0.1172,
      "num_input_tokens_seen": 12801864,
      "step": 19625
    },
    {
      "epoch": 10.28825995807128,
      "grad_norm": 1.5955090522766113,
      "learning_rate": 2.809984014468126e-05,
      "loss": 0.1091,
      "num_input_tokens_seen": 12804968,
      "step": 19630
    },
    {
      "epoch": 10.290880503144654,
      "grad_norm": 5.560205459594727,
      "learning_rate": 2.8088493774102443e-05,
      "loss": 0.1272,
      "num_input_tokens_seen": 12807848,
      "step": 19635
    },
    {
      "epoch": 10.29350104821803,
      "grad_norm": 5.058000087738037,
      "learning_rate": 2.807714675744596e-05,
      "loss": 0.0812,
      "num_input_tokens_seen": 12810632,
      "step": 19640
    },
    {
      "epoch": 10.296121593291405,
      "grad_norm": 1.9938620328903198,
      "learning_rate": 2.8065799097085475e-05,
      "loss": 0.0927,
      "num_input_tokens_seen": 12814024,
      "step": 19645
    },
    {
      "epoch": 10.29874213836478,
      "grad_norm": 1.340552806854248,
      "learning_rate": 2.8054450795394804e-05,
      "loss": 0.0844,
      "num_input_tokens_seen": 12817800,
      "step": 19650
    },
    {
      "epoch": 10.301362683438155,
      "grad_norm": 2.047598361968994,
      "learning_rate": 2.8043101854747878e-05,
      "loss": 0.0838,
      "num_input_tokens_seen": 12820456,
      "step": 19655
    },
    {
      "epoch": 10.30398322851153,
      "grad_norm": 1.7372334003448486,
      "learning_rate": 2.8031752277518748e-05,
      "loss": 0.1421,
      "num_input_tokens_seen": 12822792,
      "step": 19660
    },
    {
      "epoch": 10.306603773584905,
      "grad_norm": 3.2170543670654297,
      "learning_rate": 2.802040206608164e-05,
      "loss": 0.2489,
      "num_input_tokens_seen": 12826888,
      "step": 19665
    },
    {
      "epoch": 10.30922431865828,
      "grad_norm": 1.8216043710708618,
      "learning_rate": 2.8009051222810877e-05,
      "loss": 0.1152,
      "num_input_tokens_seen": 12829896,
      "step": 19670
    },
    {
      "epoch": 10.311844863731656,
      "grad_norm": 2.4708592891693115,
      "learning_rate": 2.7997699750080918e-05,
      "loss": 0.1563,
      "num_input_tokens_seen": 12833480,
      "step": 19675
    },
    {
      "epoch": 10.314465408805031,
      "grad_norm": 2.81097412109375,
      "learning_rate": 2.7986347650266386e-05,
      "loss": 0.1134,
      "num_input_tokens_seen": 12836424,
      "step": 19680
    },
    {
      "epoch": 10.317085953878406,
      "grad_norm": 4.16261100769043,
      "learning_rate": 2.797499492574199e-05,
      "loss": 0.2059,
      "num_input_tokens_seen": 12839464,
      "step": 19685
    },
    {
      "epoch": 10.319706498951781,
      "grad_norm": 1.6867332458496094,
      "learning_rate": 2.7963641578882594e-05,
      "loss": 0.081,
      "num_input_tokens_seen": 12841896,
      "step": 19690
    },
    {
      "epoch": 10.322327044025156,
      "grad_norm": 2.54995059967041,
      "learning_rate": 2.7952287612063193e-05,
      "loss": 0.0817,
      "num_input_tokens_seen": 12845768,
      "step": 19695
    },
    {
      "epoch": 10.324947589098532,
      "grad_norm": 1.5342118740081787,
      "learning_rate": 2.79409330276589e-05,
      "loss": 0.1185,
      "num_input_tokens_seen": 12848680,
      "step": 19700
    },
    {
      "epoch": 10.327568134171909,
      "grad_norm": 4.14107084274292,
      "learning_rate": 2.7929577828044974e-05,
      "loss": 0.1106,
      "num_input_tokens_seen": 12851464,
      "step": 19705
    },
    {
      "epoch": 10.330188679245284,
      "grad_norm": 2.238546371459961,
      "learning_rate": 2.7918222015596784e-05,
      "loss": 0.0599,
      "num_input_tokens_seen": 12854504,
      "step": 19710
    },
    {
      "epoch": 10.332809224318659,
      "grad_norm": 1.2064765691757202,
      "learning_rate": 2.7906865592689847e-05,
      "loss": 0.1004,
      "num_input_tokens_seen": 12857736,
      "step": 19715
    },
    {
      "epoch": 10.335429769392034,
      "grad_norm": 2.399038076400757,
      "learning_rate": 2.789550856169979e-05,
      "loss": 0.0808,
      "num_input_tokens_seen": 12860360,
      "step": 19720
    },
    {
      "epoch": 10.33805031446541,
      "grad_norm": 3.4700729846954346,
      "learning_rate": 2.7884150925002374e-05,
      "loss": 0.1094,
      "num_input_tokens_seen": 12863016,
      "step": 19725
    },
    {
      "epoch": 10.340670859538784,
      "grad_norm": 5.7241082191467285,
      "learning_rate": 2.787279268497349e-05,
      "loss": 0.1346,
      "num_input_tokens_seen": 12865768,
      "step": 19730
    },
    {
      "epoch": 10.34329140461216,
      "grad_norm": 1.9873915910720825,
      "learning_rate": 2.786143384398915e-05,
      "loss": 0.0969,
      "num_input_tokens_seen": 12869256,
      "step": 19735
    },
    {
      "epoch": 10.345911949685535,
      "grad_norm": 1.983534336090088,
      "learning_rate": 2.785007440442549e-05,
      "loss": 0.1035,
      "num_input_tokens_seen": 12872456,
      "step": 19740
    },
    {
      "epoch": 10.34853249475891,
      "grad_norm": 1.9750648736953735,
      "learning_rate": 2.7838714368658785e-05,
      "loss": 0.077,
      "num_input_tokens_seen": 12875720,
      "step": 19745
    },
    {
      "epoch": 10.351153039832285,
      "grad_norm": 2.3067657947540283,
      "learning_rate": 2.7827353739065427e-05,
      "loss": 0.0954,
      "num_input_tokens_seen": 12878088,
      "step": 19750
    },
    {
      "epoch": 10.35377358490566,
      "grad_norm": 3.70986270904541,
      "learning_rate": 2.781599251802191e-05,
      "loss": 0.0853,
      "num_input_tokens_seen": 12880616,
      "step": 19755
    },
    {
      "epoch": 10.356394129979035,
      "grad_norm": 2.437324285507202,
      "learning_rate": 2.780463070790489e-05,
      "loss": 0.1369,
      "num_input_tokens_seen": 12883144,
      "step": 19760
    },
    {
      "epoch": 10.35901467505241,
      "grad_norm": 1.8588149547576904,
      "learning_rate": 2.7793268311091125e-05,
      "loss": 0.0587,
      "num_input_tokens_seen": 12885832,
      "step": 19765
    },
    {
      "epoch": 10.361635220125786,
      "grad_norm": 1.704996109008789,
      "learning_rate": 2.7781905329957496e-05,
      "loss": 0.0994,
      "num_input_tokens_seen": 12889000,
      "step": 19770
    },
    {
      "epoch": 10.364255765199161,
      "grad_norm": 3.8646044731140137,
      "learning_rate": 2.7770541766881008e-05,
      "loss": 0.0965,
      "num_input_tokens_seen": 12892616,
      "step": 19775
    },
    {
      "epoch": 10.366876310272536,
      "grad_norm": 3.7231502532958984,
      "learning_rate": 2.7759177624238786e-05,
      "loss": 0.0874,
      "num_input_tokens_seen": 12895112,
      "step": 19780
    },
    {
      "epoch": 10.369496855345911,
      "grad_norm": 2.228919744491577,
      "learning_rate": 2.7747812904408094e-05,
      "loss": 0.131,
      "num_input_tokens_seen": 12899176,
      "step": 19785
    },
    {
      "epoch": 10.372117400419286,
      "grad_norm": 5.971856117248535,
      "learning_rate": 2.7736447609766286e-05,
      "loss": 0.1152,
      "num_input_tokens_seen": 12902184,
      "step": 19790
    },
    {
      "epoch": 10.374737945492662,
      "grad_norm": 2.4076437950134277,
      "learning_rate": 2.772508174269085e-05,
      "loss": 0.0692,
      "num_input_tokens_seen": 12904712,
      "step": 19795
    },
    {
      "epoch": 10.377358490566039,
      "grad_norm": 5.277968406677246,
      "learning_rate": 2.7713715305559414e-05,
      "loss": 0.1048,
      "num_input_tokens_seen": 12907560,
      "step": 19800
    },
    {
      "epoch": 10.379979035639414,
      "grad_norm": 5.090015411376953,
      "learning_rate": 2.7702348300749692e-05,
      "loss": 0.0845,
      "num_input_tokens_seen": 12910568,
      "step": 19805
    },
    {
      "epoch": 10.382599580712789,
      "grad_norm": 2.6604557037353516,
      "learning_rate": 2.7690980730639533e-05,
      "loss": 0.2163,
      "num_input_tokens_seen": 12913608,
      "step": 19810
    },
    {
      "epoch": 10.385220125786164,
      "grad_norm": 8.349055290222168,
      "learning_rate": 2.7679612597606902e-05,
      "loss": 0.1014,
      "num_input_tokens_seen": 12917288,
      "step": 19815
    },
    {
      "epoch": 10.38784067085954,
      "grad_norm": 2.348376512527466,
      "learning_rate": 2.7668243904029894e-05,
      "loss": 0.1107,
      "num_input_tokens_seen": 12920232,
      "step": 19820
    },
    {
      "epoch": 10.390461215932914,
      "grad_norm": 1.3617658615112305,
      "learning_rate": 2.76568746522867e-05,
      "loss": 0.1318,
      "num_input_tokens_seen": 12923432,
      "step": 19825
    },
    {
      "epoch": 10.39308176100629,
      "grad_norm": 1.6936231851577759,
      "learning_rate": 2.7645504844755637e-05,
      "loss": 0.1179,
      "num_input_tokens_seen": 12927912,
      "step": 19830
    },
    {
      "epoch": 10.395702306079665,
      "grad_norm": 2.1570723056793213,
      "learning_rate": 2.7634134483815143e-05,
      "loss": 0.1028,
      "num_input_tokens_seen": 12931304,
      "step": 19835
    },
    {
      "epoch": 10.39832285115304,
      "grad_norm": 5.261480808258057,
      "learning_rate": 2.762276357184377e-05,
      "loss": 0.0842,
      "num_input_tokens_seen": 12934248,
      "step": 19840
    },
    {
      "epoch": 10.400943396226415,
      "grad_norm": 2.036329984664917,
      "learning_rate": 2.7611392111220176e-05,
      "loss": 0.1086,
      "num_input_tokens_seen": 12937608,
      "step": 19845
    },
    {
      "epoch": 10.40356394129979,
      "grad_norm": 2.413281202316284,
      "learning_rate": 2.7600020104323154e-05,
      "loss": 0.0841,
      "num_input_tokens_seen": 12940360,
      "step": 19850
    },
    {
      "epoch": 10.406184486373165,
      "grad_norm": 2.4257614612579346,
      "learning_rate": 2.7588647553531576e-05,
      "loss": 0.0994,
      "num_input_tokens_seen": 12943048,
      "step": 19855
    },
    {
      "epoch": 10.40880503144654,
      "grad_norm": 0.4137200117111206,
      "learning_rate": 2.7577274461224463e-05,
      "loss": 0.052,
      "num_input_tokens_seen": 12945704,
      "step": 19860
    },
    {
      "epoch": 10.411425576519916,
      "grad_norm": 4.305471420288086,
      "learning_rate": 2.756590082978095e-05,
      "loss": 0.0839,
      "num_input_tokens_seen": 12948616,
      "step": 19865
    },
    {
      "epoch": 10.414046121593291,
      "grad_norm": 2.769439220428467,
      "learning_rate": 2.7554526661580243e-05,
      "loss": 0.0626,
      "num_input_tokens_seen": 12951880,
      "step": 19870
    },
    {
      "epoch": 10.416666666666666,
      "grad_norm": 1.8460227251052856,
      "learning_rate": 2.7543151959001708e-05,
      "loss": 0.148,
      "num_input_tokens_seen": 12954984,
      "step": 19875
    },
    {
      "epoch": 10.419287211740041,
      "grad_norm": 4.142850875854492,
      "learning_rate": 2.7531776724424797e-05,
      "loss": 0.1078,
      "num_input_tokens_seen": 12957640,
      "step": 19880
    },
    {
      "epoch": 10.421907756813416,
      "grad_norm": 1.5585830211639404,
      "learning_rate": 2.752040096022907e-05,
      "loss": 0.0642,
      "num_input_tokens_seen": 12960488,
      "step": 19885
    },
    {
      "epoch": 10.424528301886792,
      "grad_norm": 5.618468761444092,
      "learning_rate": 2.750902466879423e-05,
      "loss": 0.1736,
      "num_input_tokens_seen": 12963560,
      "step": 19890
    },
    {
      "epoch": 10.427148846960169,
      "grad_norm": 2.3066537380218506,
      "learning_rate": 2.7497647852500048e-05,
      "loss": 0.1003,
      "num_input_tokens_seen": 12967208,
      "step": 19895
    },
    {
      "epoch": 10.429769392033544,
      "grad_norm": 2.026777982711792,
      "learning_rate": 2.7486270513726424e-05,
      "loss": 0.0671,
      "num_input_tokens_seen": 12970248,
      "step": 19900
    },
    {
      "epoch": 10.432389937106919,
      "grad_norm": 3.088625907897949,
      "learning_rate": 2.7474892654853384e-05,
      "loss": 0.086,
      "num_input_tokens_seen": 12973480,
      "step": 19905
    },
    {
      "epoch": 10.435010482180294,
      "grad_norm": 1.181910753250122,
      "learning_rate": 2.7463514278261034e-05,
      "loss": 0.1034,
      "num_input_tokens_seen": 12976904,
      "step": 19910
    },
    {
      "epoch": 10.43763102725367,
      "grad_norm": 6.230340480804443,
      "learning_rate": 2.7452135386329596e-05,
      "loss": 0.1086,
      "num_input_tokens_seen": 12979912,
      "step": 19915
    },
    {
      "epoch": 10.440251572327044,
      "grad_norm": 1.5450035333633423,
      "learning_rate": 2.7440755981439414e-05,
      "loss": 0.0706,
      "num_input_tokens_seen": 12982728,
      "step": 19920
    },
    {
      "epoch": 10.44287211740042,
      "grad_norm": 4.861103534698486,
      "learning_rate": 2.7429376065970924e-05,
      "loss": 0.0631,
      "num_input_tokens_seen": 12986472,
      "step": 19925
    },
    {
      "epoch": 10.445492662473795,
      "grad_norm": 0.9489057660102844,
      "learning_rate": 2.7417995642304678e-05,
      "loss": 0.0991,
      "num_input_tokens_seen": 12990600,
      "step": 19930
    },
    {
      "epoch": 10.44811320754717,
      "grad_norm": 3.483156442642212,
      "learning_rate": 2.7406614712821333e-05,
      "loss": 0.114,
      "num_input_tokens_seen": 12994216,
      "step": 19935
    },
    {
      "epoch": 10.450733752620545,
      "grad_norm": 2.986994504928589,
      "learning_rate": 2.739523327990164e-05,
      "loss": 0.1317,
      "num_input_tokens_seen": 12997256,
      "step": 19940
    },
    {
      "epoch": 10.45335429769392,
      "grad_norm": 1.7904554605484009,
      "learning_rate": 2.7383851345926477e-05,
      "loss": 0.091,
      "num_input_tokens_seen": 13001384,
      "step": 19945
    },
    {
      "epoch": 10.455974842767295,
      "grad_norm": 2.756312131881714,
      "learning_rate": 2.7372468913276804e-05,
      "loss": 0.185,
      "num_input_tokens_seen": 13004232,
      "step": 19950
    },
    {
      "epoch": 10.45859538784067,
      "grad_norm": 2.3138885498046875,
      "learning_rate": 2.7361085984333705e-05,
      "loss": 0.1384,
      "num_input_tokens_seen": 13007208,
      "step": 19955
    },
    {
      "epoch": 10.461215932914046,
      "grad_norm": 2.6427605152130127,
      "learning_rate": 2.7349702561478356e-05,
      "loss": 0.1197,
      "num_input_tokens_seen": 13010696,
      "step": 19960
    },
    {
      "epoch": 10.463836477987421,
      "grad_norm": 2.072465181350708,
      "learning_rate": 2.733831864709203e-05,
      "loss": 0.1252,
      "num_input_tokens_seen": 13013416,
      "step": 19965
    },
    {
      "epoch": 10.466457023060796,
      "grad_norm": 2.6351122856140137,
      "learning_rate": 2.732693424355613e-05,
      "loss": 0.1084,
      "num_input_tokens_seen": 13016904,
      "step": 19970
    },
    {
      "epoch": 10.469077568134171,
      "grad_norm": 2.7218334674835205,
      "learning_rate": 2.731554935325213e-05,
      "loss": 0.1021,
      "num_input_tokens_seen": 13019496,
      "step": 19975
    },
    {
      "epoch": 10.471698113207546,
      "grad_norm": 6.992243766784668,
      "learning_rate": 2.7304163978561613e-05,
      "loss": 0.1021,
      "num_input_tokens_seen": 13022184,
      "step": 19980
    },
    {
      "epoch": 10.474318658280922,
      "grad_norm": 2.73201322555542,
      "learning_rate": 2.729277812186629e-05,
      "loss": 0.1012,
      "num_input_tokens_seen": 13024968,
      "step": 19985
    },
    {
      "epoch": 10.476939203354299,
      "grad_norm": 2.437460422515869,
      "learning_rate": 2.7281391785547945e-05,
      "loss": 0.1272,
      "num_input_tokens_seen": 13028328,
      "step": 19990
    },
    {
      "epoch": 10.479559748427674,
      "grad_norm": 2.401125907897949,
      "learning_rate": 2.7270004971988456e-05,
      "loss": 0.0812,
      "num_input_tokens_seen": 13032040,
      "step": 19995
    },
    {
      "epoch": 10.482180293501049,
      "grad_norm": 1.812726378440857,
      "learning_rate": 2.7258617683569825e-05,
      "loss": 0.0796,
      "num_input_tokens_seen": 13035944,
      "step": 20000
    },
    {
      "epoch": 10.484800838574424,
      "grad_norm": 2.4724934101104736,
      "learning_rate": 2.7247229922674143e-05,
      "loss": 0.1435,
      "num_input_tokens_seen": 13042856,
      "step": 20005
    },
    {
      "epoch": 10.4874213836478,
      "grad_norm": 1.6762713193893433,
      "learning_rate": 2.7235841691683605e-05,
      "loss": 0.1046,
      "num_input_tokens_seen": 13046792,
      "step": 20010
    },
    {
      "epoch": 10.490041928721174,
      "grad_norm": 1.4370746612548828,
      "learning_rate": 2.7224452992980487e-05,
      "loss": 0.1053,
      "num_input_tokens_seen": 13050920,
      "step": 20015
    },
    {
      "epoch": 10.49266247379455,
      "grad_norm": 1.9908233880996704,
      "learning_rate": 2.7213063828947182e-05,
      "loss": 0.1449,
      "num_input_tokens_seen": 13058664,
      "step": 20020
    },
    {
      "epoch": 10.495283018867925,
      "grad_norm": 1.4342433214187622,
      "learning_rate": 2.7201674201966176e-05,
      "loss": 0.1304,
      "num_input_tokens_seen": 13062952,
      "step": 20025
    },
    {
      "epoch": 10.4979035639413,
      "grad_norm": 1.8120405673980713,
      "learning_rate": 2.7190284114420035e-05,
      "loss": 0.0695,
      "num_input_tokens_seen": 13065768,
      "step": 20030
    },
    {
      "epoch": 10.500524109014675,
      "grad_norm": 1.160063624382019,
      "learning_rate": 2.717889356869146e-05,
      "loss": 0.1355,
      "num_input_tokens_seen": 13069160,
      "step": 20035
    },
    {
      "epoch": 10.50314465408805,
      "grad_norm": 1.5578539371490479,
      "learning_rate": 2.7167502567163194e-05,
      "loss": 0.0654,
      "num_input_tokens_seen": 13072168,
      "step": 20040
    },
    {
      "epoch": 10.505765199161425,
      "grad_norm": 2.254727602005005,
      "learning_rate": 2.715611111221813e-05,
      "loss": 0.0672,
      "num_input_tokens_seen": 13074888,
      "step": 20045
    },
    {
      "epoch": 10.5083857442348,
      "grad_norm": 4.346913814544678,
      "learning_rate": 2.7144719206239217e-05,
      "loss": 0.285,
      "num_input_tokens_seen": 13078280,
      "step": 20050
    },
    {
      "epoch": 10.511006289308176,
      "grad_norm": 1.9697308540344238,
      "learning_rate": 2.7133326851609508e-05,
      "loss": 0.129,
      "num_input_tokens_seen": 13081544,
      "step": 20055
    },
    {
      "epoch": 10.51362683438155,
      "grad_norm": 2.9599905014038086,
      "learning_rate": 2.7121934050712166e-05,
      "loss": 0.1086,
      "num_input_tokens_seen": 13084360,
      "step": 20060
    },
    {
      "epoch": 10.516247379454926,
      "grad_norm": 1.5562528371810913,
      "learning_rate": 2.711054080593043e-05,
      "loss": 0.1225,
      "num_input_tokens_seen": 13087464,
      "step": 20065
    },
    {
      "epoch": 10.518867924528301,
      "grad_norm": 1.5653517246246338,
      "learning_rate": 2.7099147119647627e-05,
      "loss": 0.0836,
      "num_input_tokens_seen": 13091112,
      "step": 20070
    },
    {
      "epoch": 10.521488469601676,
      "grad_norm": 2.6149861812591553,
      "learning_rate": 2.7087752994247206e-05,
      "loss": 0.1393,
      "num_input_tokens_seen": 13094824,
      "step": 20075
    },
    {
      "epoch": 10.524109014675052,
      "grad_norm": 1.930888056755066,
      "learning_rate": 2.707635843211267e-05,
      "loss": 0.0582,
      "num_input_tokens_seen": 13097768,
      "step": 20080
    },
    {
      "epoch": 10.526729559748428,
      "grad_norm": 2.288470506668091,
      "learning_rate": 2.7064963435627637e-05,
      "loss": 0.126,
      "num_input_tokens_seen": 13100712,
      "step": 20085
    },
    {
      "epoch": 10.529350104821804,
      "grad_norm": 3.035097122192383,
      "learning_rate": 2.7053568007175823e-05,
      "loss": 0.0816,
      "num_input_tokens_seen": 13103464,
      "step": 20090
    },
    {
      "epoch": 10.531970649895179,
      "grad_norm": 2.4759395122528076,
      "learning_rate": 2.7042172149140994e-05,
      "loss": 0.0501,
      "num_input_tokens_seen": 13106312,
      "step": 20095
    },
    {
      "epoch": 10.534591194968554,
      "grad_norm": 2.796252965927124,
      "learning_rate": 2.703077586390706e-05,
      "loss": 0.1177,
      "num_input_tokens_seen": 13108936,
      "step": 20100
    },
    {
      "epoch": 10.53721174004193,
      "grad_norm": 2.613978147506714,
      "learning_rate": 2.7019379153857982e-05,
      "loss": 0.0789,
      "num_input_tokens_seen": 13113544,
      "step": 20105
    },
    {
      "epoch": 10.539832285115304,
      "grad_norm": 3.110102415084839,
      "learning_rate": 2.700798202137782e-05,
      "loss": 0.1364,
      "num_input_tokens_seen": 13116648,
      "step": 20110
    },
    {
      "epoch": 10.54245283018868,
      "grad_norm": 1.9119975566864014,
      "learning_rate": 2.6996584468850732e-05,
      "loss": 0.0996,
      "num_input_tokens_seen": 13119816,
      "step": 20115
    },
    {
      "epoch": 10.545073375262055,
      "grad_norm": 1.2875372171401978,
      "learning_rate": 2.6985186498660946e-05,
      "loss": 0.2747,
      "num_input_tokens_seen": 13123656,
      "step": 20120
    },
    {
      "epoch": 10.54769392033543,
      "grad_norm": 2.075265407562256,
      "learning_rate": 2.69737881131928e-05,
      "loss": 0.1361,
      "num_input_tokens_seen": 13126120,
      "step": 20125
    },
    {
      "epoch": 10.550314465408805,
      "grad_norm": 2.3390512466430664,
      "learning_rate": 2.6962389314830692e-05,
      "loss": 0.1239,
      "num_input_tokens_seen": 13128968,
      "step": 20130
    },
    {
      "epoch": 10.55293501048218,
      "grad_norm": 5.359927177429199,
      "learning_rate": 2.6950990105959135e-05,
      "loss": 0.0736,
      "num_input_tokens_seen": 13132168,
      "step": 20135
    },
    {
      "epoch": 10.555555555555555,
      "grad_norm": 5.727435111999512,
      "learning_rate": 2.6939590488962703e-05,
      "loss": 0.1234,
      "num_input_tokens_seen": 13135560,
      "step": 20140
    },
    {
      "epoch": 10.55817610062893,
      "grad_norm": 2.446403741836548,
      "learning_rate": 2.6928190466226067e-05,
      "loss": 0.1062,
      "num_input_tokens_seen": 13139400,
      "step": 20145
    },
    {
      "epoch": 10.560796645702306,
      "grad_norm": 1.3936439752578735,
      "learning_rate": 2.691679004013398e-05,
      "loss": 0.0907,
      "num_input_tokens_seen": 13144232,
      "step": 20150
    },
    {
      "epoch": 10.56341719077568,
      "grad_norm": 2.3969337940216064,
      "learning_rate": 2.6905389213071285e-05,
      "loss": 0.0587,
      "num_input_tokens_seen": 13147432,
      "step": 20155
    },
    {
      "epoch": 10.566037735849056,
      "grad_norm": 3.143637180328369,
      "learning_rate": 2.6893987987422907e-05,
      "loss": 0.0664,
      "num_input_tokens_seen": 13150632,
      "step": 20160
    },
    {
      "epoch": 10.568658280922431,
      "grad_norm": 2.092334508895874,
      "learning_rate": 2.6882586365573854e-05,
      "loss": 0.0926,
      "num_input_tokens_seen": 13154152,
      "step": 20165
    },
    {
      "epoch": 10.571278825995806,
      "grad_norm": 1.309403896331787,
      "learning_rate": 2.6871184349909206e-05,
      "loss": 0.0837,
      "num_input_tokens_seen": 13160552,
      "step": 20170
    },
    {
      "epoch": 10.573899371069182,
      "grad_norm": 3.7574872970581055,
      "learning_rate": 2.685978194281414e-05,
      "loss": 0.0869,
      "num_input_tokens_seen": 13163208,
      "step": 20175
    },
    {
      "epoch": 10.576519916142558,
      "grad_norm": 1.4990161657333374,
      "learning_rate": 2.6848379146673914e-05,
      "loss": 0.0842,
      "num_input_tokens_seen": 13166056,
      "step": 20180
    },
    {
      "epoch": 10.579140461215934,
      "grad_norm": 1.3535724878311157,
      "learning_rate": 2.6836975963873843e-05,
      "loss": 0.1024,
      "num_input_tokens_seen": 13168712,
      "step": 20185
    },
    {
      "epoch": 10.581761006289309,
      "grad_norm": 3.7219078540802,
      "learning_rate": 2.6825572396799377e-05,
      "loss": 0.1275,
      "num_input_tokens_seen": 13171400,
      "step": 20190
    },
    {
      "epoch": 10.584381551362684,
      "grad_norm": 2.5211548805236816,
      "learning_rate": 2.6814168447835986e-05,
      "loss": 0.0923,
      "num_input_tokens_seen": 13174888,
      "step": 20195
    },
    {
      "epoch": 10.58700209643606,
      "grad_norm": 2.6369967460632324,
      "learning_rate": 2.6802764119369245e-05,
      "loss": 0.1243,
      "num_input_tokens_seen": 13178920,
      "step": 20200
    },
    {
      "epoch": 10.589622641509434,
      "grad_norm": 2.007788896560669,
      "learning_rate": 2.6791359413784828e-05,
      "loss": 0.1354,
      "num_input_tokens_seen": 13183176,
      "step": 20205
    },
    {
      "epoch": 10.59224318658281,
      "grad_norm": 2.4652154445648193,
      "learning_rate": 2.677995433346845e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 13186760,
      "step": 20210
    },
    {
      "epoch": 10.594863731656185,
      "grad_norm": 1.492073893547058,
      "learning_rate": 2.6768548880805934e-05,
      "loss": 0.0791,
      "num_input_tokens_seen": 13189544,
      "step": 20215
    },
    {
      "epoch": 10.59748427672956,
      "grad_norm": 1.7623831033706665,
      "learning_rate": 2.6757143058183177e-05,
      "loss": 0.0969,
      "num_input_tokens_seen": 13193096,
      "step": 20220
    },
    {
      "epoch": 10.600104821802935,
      "grad_norm": 3.4183058738708496,
      "learning_rate": 2.6745736867986132e-05,
      "loss": 0.0732,
      "num_input_tokens_seen": 13195944,
      "step": 20225
    },
    {
      "epoch": 10.60272536687631,
      "grad_norm": 2.147651433944702,
      "learning_rate": 2.673433031260085e-05,
      "loss": 0.0879,
      "num_input_tokens_seen": 13198760,
      "step": 20230
    },
    {
      "epoch": 10.605345911949685,
      "grad_norm": 2.896214008331299,
      "learning_rate": 2.6722923394413456e-05,
      "loss": 0.079,
      "num_input_tokens_seen": 13201704,
      "step": 20235
    },
    {
      "epoch": 10.60796645702306,
      "grad_norm": 2.70573353767395,
      "learning_rate": 2.6711516115810136e-05,
      "loss": 0.1442,
      "num_input_tokens_seen": 13204872,
      "step": 20240
    },
    {
      "epoch": 10.610587002096436,
      "grad_norm": 3.1529932022094727,
      "learning_rate": 2.6700108479177188e-05,
      "loss": 0.0966,
      "num_input_tokens_seen": 13208200,
      "step": 20245
    },
    {
      "epoch": 10.61320754716981,
      "grad_norm": 1.6826366186141968,
      "learning_rate": 2.6688700486900926e-05,
      "loss": 0.0688,
      "num_input_tokens_seen": 13211848,
      "step": 20250
    },
    {
      "epoch": 10.615828092243186,
      "grad_norm": 1.7683372497558594,
      "learning_rate": 2.6677292141367787e-05,
      "loss": 0.0904,
      "num_input_tokens_seen": 13214760,
      "step": 20255
    },
    {
      "epoch": 10.618448637316561,
      "grad_norm": 1.9969604015350342,
      "learning_rate": 2.6665883444964278e-05,
      "loss": 0.0886,
      "num_input_tokens_seen": 13219016,
      "step": 20260
    },
    {
      "epoch": 10.621069182389936,
      "grad_norm": 2.1220171451568604,
      "learning_rate": 2.665447440007695e-05,
      "loss": 0.0833,
      "num_input_tokens_seen": 13221800,
      "step": 20265
    },
    {
      "epoch": 10.623689727463312,
      "grad_norm": 3.3333446979522705,
      "learning_rate": 2.664306500909246e-05,
      "loss": 0.1313,
      "num_input_tokens_seen": 13224936,
      "step": 20270
    },
    {
      "epoch": 10.626310272536688,
      "grad_norm": 1.9565118551254272,
      "learning_rate": 2.663165527439751e-05,
      "loss": 0.08,
      "num_input_tokens_seen": 13227560,
      "step": 20275
    },
    {
      "epoch": 10.628930817610064,
      "grad_norm": 2.054436206817627,
      "learning_rate": 2.662024519837889e-05,
      "loss": 0.0701,
      "num_input_tokens_seen": 13230888,
      "step": 20280
    },
    {
      "epoch": 10.631551362683439,
      "grad_norm": 2.1336543560028076,
      "learning_rate": 2.6608834783423465e-05,
      "loss": 0.1485,
      "num_input_tokens_seen": 13234152,
      "step": 20285
    },
    {
      "epoch": 10.634171907756814,
      "grad_norm": 2.046337366104126,
      "learning_rate": 2.6597424031918156e-05,
      "loss": 0.0871,
      "num_input_tokens_seen": 13237352,
      "step": 20290
    },
    {
      "epoch": 10.63679245283019,
      "grad_norm": 2.253282070159912,
      "learning_rate": 2.6586012946249954e-05,
      "loss": 0.0782,
      "num_input_tokens_seen": 13240264,
      "step": 20295
    },
    {
      "epoch": 10.639412997903564,
      "grad_norm": 2.933203935623169,
      "learning_rate": 2.6574601528805953e-05,
      "loss": 0.1328,
      "num_input_tokens_seen": 13243528,
      "step": 20300
    },
    {
      "epoch": 10.64203354297694,
      "grad_norm": 4.814035892486572,
      "learning_rate": 2.6563189781973268e-05,
      "loss": 0.1591,
      "num_input_tokens_seen": 13248712,
      "step": 20305
    },
    {
      "epoch": 10.644654088050315,
      "grad_norm": 1.7740662097930908,
      "learning_rate": 2.655177770813912e-05,
      "loss": 0.1517,
      "num_input_tokens_seen": 13252712,
      "step": 20310
    },
    {
      "epoch": 10.64727463312369,
      "grad_norm": 3.9187986850738525,
      "learning_rate": 2.654036530969077e-05,
      "loss": 0.1065,
      "num_input_tokens_seen": 13256392,
      "step": 20315
    },
    {
      "epoch": 10.649895178197065,
      "grad_norm": 2.4537670612335205,
      "learning_rate": 2.6528952589015576e-05,
      "loss": 0.1753,
      "num_input_tokens_seen": 13259240,
      "step": 20320
    },
    {
      "epoch": 10.65251572327044,
      "grad_norm": 1.6023987531661987,
      "learning_rate": 2.6517539548500947e-05,
      "loss": 0.1077,
      "num_input_tokens_seen": 13263176,
      "step": 20325
    },
    {
      "epoch": 10.655136268343815,
      "grad_norm": 3.620389223098755,
      "learning_rate": 2.6506126190534354e-05,
      "loss": 0.126,
      "num_input_tokens_seen": 13266696,
      "step": 20330
    },
    {
      "epoch": 10.65775681341719,
      "grad_norm": 4.821348667144775,
      "learning_rate": 2.649471251750334e-05,
      "loss": 0.0753,
      "num_input_tokens_seen": 13269512,
      "step": 20335
    },
    {
      "epoch": 10.660377358490566,
      "grad_norm": 1.831709861755371,
      "learning_rate": 2.6483298531795527e-05,
      "loss": 0.0823,
      "num_input_tokens_seen": 13272072,
      "step": 20340
    },
    {
      "epoch": 10.66299790356394,
      "grad_norm": 3.6714580059051514,
      "learning_rate": 2.6471884235798583e-05,
      "loss": 0.1333,
      "num_input_tokens_seen": 13275848,
      "step": 20345
    },
    {
      "epoch": 10.665618448637316,
      "grad_norm": 1.5279293060302734,
      "learning_rate": 2.6460469631900247e-05,
      "loss": 0.1247,
      "num_input_tokens_seen": 13278440,
      "step": 20350
    },
    {
      "epoch": 10.668238993710691,
      "grad_norm": 2.986541509628296,
      "learning_rate": 2.644905472248832e-05,
      "loss": 0.0623,
      "num_input_tokens_seen": 13281576,
      "step": 20355
    },
    {
      "epoch": 10.670859538784066,
      "grad_norm": 2.0717105865478516,
      "learning_rate": 2.6437639509950686e-05,
      "loss": 0.0877,
      "num_input_tokens_seen": 13284168,
      "step": 20360
    },
    {
      "epoch": 10.673480083857442,
      "grad_norm": 2.3673481941223145,
      "learning_rate": 2.6426223996675264e-05,
      "loss": 0.1359,
      "num_input_tokens_seen": 13286760,
      "step": 20365
    },
    {
      "epoch": 10.676100628930818,
      "grad_norm": 1.555944800376892,
      "learning_rate": 2.641480818505005e-05,
      "loss": 0.1472,
      "num_input_tokens_seen": 13289672,
      "step": 20370
    },
    {
      "epoch": 10.678721174004194,
      "grad_norm": 2.241882562637329,
      "learning_rate": 2.6403392077463108e-05,
      "loss": 0.1062,
      "num_input_tokens_seen": 13292616,
      "step": 20375
    },
    {
      "epoch": 10.681341719077569,
      "grad_norm": 2.261789083480835,
      "learning_rate": 2.639197567630256e-05,
      "loss": 0.1078,
      "num_input_tokens_seen": 13296424,
      "step": 20380
    },
    {
      "epoch": 10.683962264150944,
      "grad_norm": 2.3531227111816406,
      "learning_rate": 2.638055898395657e-05,
      "loss": 0.0736,
      "num_input_tokens_seen": 13299048,
      "step": 20385
    },
    {
      "epoch": 10.68658280922432,
      "grad_norm": 1.984775424003601,
      "learning_rate": 2.6369142002813403e-05,
      "loss": 0.0687,
      "num_input_tokens_seen": 13302152,
      "step": 20390
    },
    {
      "epoch": 10.689203354297694,
      "grad_norm": 1.2591780424118042,
      "learning_rate": 2.6357724735261335e-05,
      "loss": 0.0421,
      "num_input_tokens_seen": 13304840,
      "step": 20395
    },
    {
      "epoch": 10.69182389937107,
      "grad_norm": 3.8822743892669678,
      "learning_rate": 2.6346307183688752e-05,
      "loss": 0.0902,
      "num_input_tokens_seen": 13307976,
      "step": 20400
    },
    {
      "epoch": 10.694444444444445,
      "grad_norm": 2.2757627964019775,
      "learning_rate": 2.6334889350484064e-05,
      "loss": 0.0937,
      "num_input_tokens_seen": 13310856,
      "step": 20405
    },
    {
      "epoch": 10.69706498951782,
      "grad_norm": 1.8571428060531616,
      "learning_rate": 2.632347123803575e-05,
      "loss": 0.0933,
      "num_input_tokens_seen": 13313448,
      "step": 20410
    },
    {
      "epoch": 10.699685534591195,
      "grad_norm": 4.739620208740234,
      "learning_rate": 2.6312052848732353e-05,
      "loss": 0.143,
      "num_input_tokens_seen": 13316776,
      "step": 20415
    },
    {
      "epoch": 10.70230607966457,
      "grad_norm": 1.7686635255813599,
      "learning_rate": 2.6300634184962475e-05,
      "loss": 0.1506,
      "num_input_tokens_seen": 13319560,
      "step": 20420
    },
    {
      "epoch": 10.704926624737945,
      "grad_norm": 3.2297139167785645,
      "learning_rate": 2.628921524911475e-05,
      "loss": 0.1069,
      "num_input_tokens_seen": 13322440,
      "step": 20425
    },
    {
      "epoch": 10.70754716981132,
      "grad_norm": 2.4384684562683105,
      "learning_rate": 2.6277796043577924e-05,
      "loss": 0.0714,
      "num_input_tokens_seen": 13325448,
      "step": 20430
    },
    {
      "epoch": 10.710167714884696,
      "grad_norm": 1.0524581670761108,
      "learning_rate": 2.6266376570740725e-05,
      "loss": 0.0752,
      "num_input_tokens_seen": 13328520,
      "step": 20435
    },
    {
      "epoch": 10.71278825995807,
      "grad_norm": 2.221824884414673,
      "learning_rate": 2.6254956832992005e-05,
      "loss": 0.0612,
      "num_input_tokens_seen": 13331560,
      "step": 20440
    },
    {
      "epoch": 10.715408805031446,
      "grad_norm": 6.2606401443481445,
      "learning_rate": 2.6243536832720627e-05,
      "loss": 0.1044,
      "num_input_tokens_seen": 13334216,
      "step": 20445
    },
    {
      "epoch": 10.718029350104821,
      "grad_norm": 3.4623584747314453,
      "learning_rate": 2.623211657231553e-05,
      "loss": 0.1221,
      "num_input_tokens_seen": 13337768,
      "step": 20450
    },
    {
      "epoch": 10.720649895178196,
      "grad_norm": 2.0815343856811523,
      "learning_rate": 2.6220696054165704e-05,
      "loss": 0.1306,
      "num_input_tokens_seen": 13341064,
      "step": 20455
    },
    {
      "epoch": 10.723270440251572,
      "grad_norm": 1.9260834455490112,
      "learning_rate": 2.6209275280660194e-05,
      "loss": 0.0658,
      "num_input_tokens_seen": 13343688,
      "step": 20460
    },
    {
      "epoch": 10.725890985324948,
      "grad_norm": 2.968374013900757,
      "learning_rate": 2.6197854254188085e-05,
      "loss": 0.0825,
      "num_input_tokens_seen": 13347048,
      "step": 20465
    },
    {
      "epoch": 10.728511530398324,
      "grad_norm": 1.2765486240386963,
      "learning_rate": 2.618643297713854e-05,
      "loss": 0.0819,
      "num_input_tokens_seen": 13350952,
      "step": 20470
    },
    {
      "epoch": 10.731132075471699,
      "grad_norm": 4.106606483459473,
      "learning_rate": 2.617501145190075e-05,
      "loss": 0.1295,
      "num_input_tokens_seen": 13353672,
      "step": 20475
    },
    {
      "epoch": 10.733752620545074,
      "grad_norm": 2.4234776496887207,
      "learning_rate": 2.616358968086397e-05,
      "loss": 0.1337,
      "num_input_tokens_seen": 13357544,
      "step": 20480
    },
    {
      "epoch": 10.73637316561845,
      "grad_norm": 2.0520312786102295,
      "learning_rate": 2.6152167666417493e-05,
      "loss": 0.0878,
      "num_input_tokens_seen": 13361544,
      "step": 20485
    },
    {
      "epoch": 10.738993710691824,
      "grad_norm": 1.6597446203231812,
      "learning_rate": 2.6140745410950694e-05,
      "loss": 0.114,
      "num_input_tokens_seen": 13365480,
      "step": 20490
    },
    {
      "epoch": 10.7416142557652,
      "grad_norm": 2.3244919776916504,
      "learning_rate": 2.6129322916852973e-05,
      "loss": 0.0795,
      "num_input_tokens_seen": 13368072,
      "step": 20495
    },
    {
      "epoch": 10.744234800838575,
      "grad_norm": 1.3962265253067017,
      "learning_rate": 2.611790018651377e-05,
      "loss": 0.0951,
      "num_input_tokens_seen": 13370920,
      "step": 20500
    },
    {
      "epoch": 10.74685534591195,
      "grad_norm": 10.0730619430542,
      "learning_rate": 2.6106477222322613e-05,
      "loss": 0.2093,
      "num_input_tokens_seen": 13373832,
      "step": 20505
    },
    {
      "epoch": 10.749475890985325,
      "grad_norm": 2.2330753803253174,
      "learning_rate": 2.6095054026669037e-05,
      "loss": 0.0832,
      "num_input_tokens_seen": 13376712,
      "step": 20510
    },
    {
      "epoch": 10.7520964360587,
      "grad_norm": 7.389220714569092,
      "learning_rate": 2.608363060194265e-05,
      "loss": 0.121,
      "num_input_tokens_seen": 13379496,
      "step": 20515
    },
    {
      "epoch": 10.754716981132075,
      "grad_norm": 1.5064780712127686,
      "learning_rate": 2.6072206950533117e-05,
      "loss": 0.073,
      "num_input_tokens_seen": 13383688,
      "step": 20520
    },
    {
      "epoch": 10.75733752620545,
      "grad_norm": 1.9558371305465698,
      "learning_rate": 2.606078307483011e-05,
      "loss": 0.0801,
      "num_input_tokens_seen": 13387112,
      "step": 20525
    },
    {
      "epoch": 10.759958071278826,
      "grad_norm": 1.5473560094833374,
      "learning_rate": 2.6049358977223392e-05,
      "loss": 0.1234,
      "num_input_tokens_seen": 13391048,
      "step": 20530
    },
    {
      "epoch": 10.7625786163522,
      "grad_norm": 1.3621323108673096,
      "learning_rate": 2.6037934660102747e-05,
      "loss": 0.0728,
      "num_input_tokens_seen": 13393448,
      "step": 20535
    },
    {
      "epoch": 10.765199161425576,
      "grad_norm": 5.3130927085876465,
      "learning_rate": 2.6026510125858006e-05,
      "loss": 0.1453,
      "num_input_tokens_seen": 13396040,
      "step": 20540
    },
    {
      "epoch": 10.767819706498951,
      "grad_norm": 5.019746780395508,
      "learning_rate": 2.6015085376879063e-05,
      "loss": 0.1269,
      "num_input_tokens_seen": 13398824,
      "step": 20545
    },
    {
      "epoch": 10.770440251572326,
      "grad_norm": 10.937410354614258,
      "learning_rate": 2.600366041555584e-05,
      "loss": 0.0952,
      "num_input_tokens_seen": 13403368,
      "step": 20550
    },
    {
      "epoch": 10.773060796645701,
      "grad_norm": 2.805412530899048,
      "learning_rate": 2.5992235244278302e-05,
      "loss": 0.1292,
      "num_input_tokens_seen": 13406376,
      "step": 20555
    },
    {
      "epoch": 10.775681341719078,
      "grad_norm": 3.0087361335754395,
      "learning_rate": 2.5980809865436485e-05,
      "loss": 0.164,
      "num_input_tokens_seen": 13411208,
      "step": 20560
    },
    {
      "epoch": 10.778301886792454,
      "grad_norm": 0.8639830350875854,
      "learning_rate": 2.5969384281420424e-05,
      "loss": 0.0973,
      "num_input_tokens_seen": 13413704,
      "step": 20565
    },
    {
      "epoch": 10.780922431865829,
      "grad_norm": 2.6329288482666016,
      "learning_rate": 2.5957958494620227e-05,
      "loss": 0.0918,
      "num_input_tokens_seen": 13417128,
      "step": 20570
    },
    {
      "epoch": 10.783542976939204,
      "grad_norm": 3.6492180824279785,
      "learning_rate": 2.594653250742605e-05,
      "loss": 0.1075,
      "num_input_tokens_seen": 13420136,
      "step": 20575
    },
    {
      "epoch": 10.786163522012579,
      "grad_norm": 3.6350717544555664,
      "learning_rate": 2.5935106322228065e-05,
      "loss": 0.1111,
      "num_input_tokens_seen": 13422472,
      "step": 20580
    },
    {
      "epoch": 10.788784067085954,
      "grad_norm": 3.1556363105773926,
      "learning_rate": 2.5923679941416514e-05,
      "loss": 0.092,
      "num_input_tokens_seen": 13426280,
      "step": 20585
    },
    {
      "epoch": 10.79140461215933,
      "grad_norm": 2.5030741691589355,
      "learning_rate": 2.591225336738165e-05,
      "loss": 0.1109,
      "num_input_tokens_seen": 13429416,
      "step": 20590
    },
    {
      "epoch": 10.794025157232705,
      "grad_norm": 2.5253043174743652,
      "learning_rate": 2.5900826602513782e-05,
      "loss": 0.1565,
      "num_input_tokens_seen": 13432776,
      "step": 20595
    },
    {
      "epoch": 10.79664570230608,
      "grad_norm": 2.104055166244507,
      "learning_rate": 2.5889399649203277e-05,
      "loss": 0.0689,
      "num_input_tokens_seen": 13435976,
      "step": 20600
    },
    {
      "epoch": 10.799266247379455,
      "grad_norm": 2.682565450668335,
      "learning_rate": 2.5877972509840514e-05,
      "loss": 0.1084,
      "num_input_tokens_seen": 13439208,
      "step": 20605
    },
    {
      "epoch": 10.80188679245283,
      "grad_norm": 2.50510573387146,
      "learning_rate": 2.5866545186815912e-05,
      "loss": 0.0722,
      "num_input_tokens_seen": 13441928,
      "step": 20610
    },
    {
      "epoch": 10.804507337526205,
      "grad_norm": 1.5191274881362915,
      "learning_rate": 2.5855117682519954e-05,
      "loss": 0.0948,
      "num_input_tokens_seen": 13445000,
      "step": 20615
    },
    {
      "epoch": 10.80712788259958,
      "grad_norm": 13.45120906829834,
      "learning_rate": 2.584368999934312e-05,
      "loss": 0.1597,
      "num_input_tokens_seen": 13449032,
      "step": 20620
    },
    {
      "epoch": 10.809748427672956,
      "grad_norm": 4.195951461791992,
      "learning_rate": 2.5832262139675966e-05,
      "loss": 0.088,
      "num_input_tokens_seen": 13452392,
      "step": 20625
    },
    {
      "epoch": 10.81236897274633,
      "grad_norm": 2.6429545879364014,
      "learning_rate": 2.582083410590907e-05,
      "loss": 0.2105,
      "num_input_tokens_seen": 13455784,
      "step": 20630
    },
    {
      "epoch": 10.814989517819706,
      "grad_norm": 1.3765777349472046,
      "learning_rate": 2.5809405900433036e-05,
      "loss": 0.06,
      "num_input_tokens_seen": 13459368,
      "step": 20635
    },
    {
      "epoch": 10.817610062893081,
      "grad_norm": 1.3907818794250488,
      "learning_rate": 2.579797752563853e-05,
      "loss": 0.0641,
      "num_input_tokens_seen": 13462664,
      "step": 20640
    },
    {
      "epoch": 10.820230607966456,
      "grad_norm": 2.4742777347564697,
      "learning_rate": 2.5786548983916232e-05,
      "loss": 0.0976,
      "num_input_tokens_seen": 13464872,
      "step": 20645
    },
    {
      "epoch": 10.822851153039831,
      "grad_norm": 1.705925703048706,
      "learning_rate": 2.577512027765685e-05,
      "loss": 0.0782,
      "num_input_tokens_seen": 13467624,
      "step": 20650
    },
    {
      "epoch": 10.825471698113208,
      "grad_norm": 2.3676745891571045,
      "learning_rate": 2.5763691409251155e-05,
      "loss": 0.1328,
      "num_input_tokens_seen": 13470856,
      "step": 20655
    },
    {
      "epoch": 10.828092243186584,
      "grad_norm": 4.09085750579834,
      "learning_rate": 2.5752262381089932e-05,
      "loss": 0.0917,
      "num_input_tokens_seen": 13474024,
      "step": 20660
    },
    {
      "epoch": 10.830712788259959,
      "grad_norm": 2.742647647857666,
      "learning_rate": 2.5740833195563996e-05,
      "loss": 0.0863,
      "num_input_tokens_seen": 13478376,
      "step": 20665
    },
    {
      "epoch": 10.833333333333334,
      "grad_norm": 2.6638667583465576,
      "learning_rate": 2.572940385506421e-05,
      "loss": 0.1375,
      "num_input_tokens_seen": 13481480,
      "step": 20670
    },
    {
      "epoch": 10.835953878406709,
      "grad_norm": 2.2329978942871094,
      "learning_rate": 2.5717974361981457e-05,
      "loss": 0.1079,
      "num_input_tokens_seen": 13484456,
      "step": 20675
    },
    {
      "epoch": 10.838574423480084,
      "grad_norm": 2.799858808517456,
      "learning_rate": 2.5706544718706665e-05,
      "loss": 0.1044,
      "num_input_tokens_seen": 13487560,
      "step": 20680
    },
    {
      "epoch": 10.84119496855346,
      "grad_norm": 0.8601784110069275,
      "learning_rate": 2.569511492763077e-05,
      "loss": 0.0852,
      "num_input_tokens_seen": 13491272,
      "step": 20685
    },
    {
      "epoch": 10.843815513626835,
      "grad_norm": 4.037777900695801,
      "learning_rate": 2.5683684991144764e-05,
      "loss": 0.0741,
      "num_input_tokens_seen": 13493992,
      "step": 20690
    },
    {
      "epoch": 10.84643605870021,
      "grad_norm": 1.808732032775879,
      "learning_rate": 2.5672254911639666e-05,
      "loss": 0.1191,
      "num_input_tokens_seen": 13496968,
      "step": 20695
    },
    {
      "epoch": 10.849056603773585,
      "grad_norm": 3.9869861602783203,
      "learning_rate": 2.5660824691506496e-05,
      "loss": 0.1371,
      "num_input_tokens_seen": 13501032,
      "step": 20700
    },
    {
      "epoch": 10.85167714884696,
      "grad_norm": 6.13425350189209,
      "learning_rate": 2.5649394333136356e-05,
      "loss": 0.1451,
      "num_input_tokens_seen": 13502984,
      "step": 20705
    },
    {
      "epoch": 10.854297693920335,
      "grad_norm": 1.1699224710464478,
      "learning_rate": 2.5637963838920316e-05,
      "loss": 0.1255,
      "num_input_tokens_seen": 13507816,
      "step": 20710
    },
    {
      "epoch": 10.85691823899371,
      "grad_norm": 2.1909255981445312,
      "learning_rate": 2.5626533211249525e-05,
      "loss": 0.1439,
      "num_input_tokens_seen": 13511368,
      "step": 20715
    },
    {
      "epoch": 10.859538784067086,
      "grad_norm": 3.063363790512085,
      "learning_rate": 2.5615102452515138e-05,
      "loss": 0.1237,
      "num_input_tokens_seen": 13515720,
      "step": 20720
    },
    {
      "epoch": 10.86215932914046,
      "grad_norm": 1.49215567111969,
      "learning_rate": 2.5603671565108328e-05,
      "loss": 0.1145,
      "num_input_tokens_seen": 13519784,
      "step": 20725
    },
    {
      "epoch": 10.864779874213836,
      "grad_norm": 3.0885753631591797,
      "learning_rate": 2.5592240551420317e-05,
      "loss": 0.0926,
      "num_input_tokens_seen": 13523304,
      "step": 20730
    },
    {
      "epoch": 10.867400419287211,
      "grad_norm": 3.4758071899414062,
      "learning_rate": 2.5580809413842343e-05,
      "loss": 0.1683,
      "num_input_tokens_seen": 13526120,
      "step": 20735
    },
    {
      "epoch": 10.870020964360586,
      "grad_norm": 4.661879062652588,
      "learning_rate": 2.5569378154765655e-05,
      "loss": 0.116,
      "num_input_tokens_seen": 13529128,
      "step": 20740
    },
    {
      "epoch": 10.872641509433961,
      "grad_norm": 1.3645473718643188,
      "learning_rate": 2.5557946776581572e-05,
      "loss": 0.1302,
      "num_input_tokens_seen": 13532584,
      "step": 20745
    },
    {
      "epoch": 10.875262054507338,
      "grad_norm": 1.8900079727172852,
      "learning_rate": 2.554651528168137e-05,
      "loss": 0.0648,
      "num_input_tokens_seen": 13536104,
      "step": 20750
    },
    {
      "epoch": 10.877882599580714,
      "grad_norm": 1.5265352725982666,
      "learning_rate": 2.5535083672456407e-05,
      "loss": 0.0721,
      "num_input_tokens_seen": 13539528,
      "step": 20755
    },
    {
      "epoch": 10.880503144654089,
      "grad_norm": 12.502249717712402,
      "learning_rate": 2.5523651951298055e-05,
      "loss": 0.1583,
      "num_input_tokens_seen": 13542600,
      "step": 20760
    },
    {
      "epoch": 10.883123689727464,
      "grad_norm": 2.797752857208252,
      "learning_rate": 2.5512220120597686e-05,
      "loss": 0.0794,
      "num_input_tokens_seen": 13545704,
      "step": 20765
    },
    {
      "epoch": 10.885744234800839,
      "grad_norm": 4.42088508605957,
      "learning_rate": 2.550078818274671e-05,
      "loss": 0.0832,
      "num_input_tokens_seen": 13549096,
      "step": 20770
    },
    {
      "epoch": 10.888364779874214,
      "grad_norm": 1.5138245820999146,
      "learning_rate": 2.548935614013656e-05,
      "loss": 0.0788,
      "num_input_tokens_seen": 13552232,
      "step": 20775
    },
    {
      "epoch": 10.89098532494759,
      "grad_norm": 3.3832294940948486,
      "learning_rate": 2.5477923995158683e-05,
      "loss": 0.1154,
      "num_input_tokens_seen": 13554984,
      "step": 20780
    },
    {
      "epoch": 10.893605870020965,
      "grad_norm": 2.7777345180511475,
      "learning_rate": 2.5466491750204562e-05,
      "loss": 0.0649,
      "num_input_tokens_seen": 13561288,
      "step": 20785
    },
    {
      "epoch": 10.89622641509434,
      "grad_norm": 2.444042682647705,
      "learning_rate": 2.5455059407665693e-05,
      "loss": 0.1158,
      "num_input_tokens_seen": 13564072,
      "step": 20790
    },
    {
      "epoch": 10.898846960167715,
      "grad_norm": 2.684536933898926,
      "learning_rate": 2.544362696993358e-05,
      "loss": 0.1069,
      "num_input_tokens_seen": 13566856,
      "step": 20795
    },
    {
      "epoch": 10.90146750524109,
      "grad_norm": 2.9352943897247314,
      "learning_rate": 2.5432194439399775e-05,
      "loss": 0.1148,
      "num_input_tokens_seen": 13570696,
      "step": 20800
    },
    {
      "epoch": 10.904088050314465,
      "grad_norm": 2.09372615814209,
      "learning_rate": 2.542076181845582e-05,
      "loss": 0.0712,
      "num_input_tokens_seen": 13573416,
      "step": 20805
    },
    {
      "epoch": 10.90670859538784,
      "grad_norm": 2.1111249923706055,
      "learning_rate": 2.540932910949329e-05,
      "loss": 0.0968,
      "num_input_tokens_seen": 13576392,
      "step": 20810
    },
    {
      "epoch": 10.909329140461216,
      "grad_norm": 4.205684185028076,
      "learning_rate": 2.5397896314903785e-05,
      "loss": 0.1334,
      "num_input_tokens_seen": 13579048,
      "step": 20815
    },
    {
      "epoch": 10.91194968553459,
      "grad_norm": 5.082858085632324,
      "learning_rate": 2.5386463437078905e-05,
      "loss": 0.1426,
      "num_input_tokens_seen": 13582920,
      "step": 20820
    },
    {
      "epoch": 10.914570230607966,
      "grad_norm": 1.8748292922973633,
      "learning_rate": 2.5375030478410282e-05,
      "loss": 0.074,
      "num_input_tokens_seen": 13585576,
      "step": 20825
    },
    {
      "epoch": 10.917190775681341,
      "grad_norm": 2.2349557876586914,
      "learning_rate": 2.536359744128957e-05,
      "loss": 0.1024,
      "num_input_tokens_seen": 13588360,
      "step": 20830
    },
    {
      "epoch": 10.919811320754716,
      "grad_norm": 3.1255781650543213,
      "learning_rate": 2.5352164328108413e-05,
      "loss": 0.1435,
      "num_input_tokens_seen": 13591272,
      "step": 20835
    },
    {
      "epoch": 10.922431865828091,
      "grad_norm": 3.3098785877227783,
      "learning_rate": 2.53407311412585e-05,
      "loss": 0.1174,
      "num_input_tokens_seen": 13594184,
      "step": 20840
    },
    {
      "epoch": 10.925052410901468,
      "grad_norm": 3.900192975997925,
      "learning_rate": 2.5329297883131526e-05,
      "loss": 0.1116,
      "num_input_tokens_seen": 13597352,
      "step": 20845
    },
    {
      "epoch": 10.927672955974844,
      "grad_norm": 2.200202703475952,
      "learning_rate": 2.5317864556119182e-05,
      "loss": 0.1164,
      "num_input_tokens_seen": 13600008,
      "step": 20850
    },
    {
      "epoch": 10.930293501048219,
      "grad_norm": 1.7538506984710693,
      "learning_rate": 2.53064311626132e-05,
      "loss": 0.1154,
      "num_input_tokens_seen": 13603656,
      "step": 20855
    },
    {
      "epoch": 10.932914046121594,
      "grad_norm": 2.9917104244232178,
      "learning_rate": 2.5294997705005318e-05,
      "loss": 0.145,
      "num_input_tokens_seen": 13607176,
      "step": 20860
    },
    {
      "epoch": 10.935534591194969,
      "grad_norm": 5.742295742034912,
      "learning_rate": 2.5283564185687282e-05,
      "loss": 0.2052,
      "num_input_tokens_seen": 13610600,
      "step": 20865
    },
    {
      "epoch": 10.938155136268344,
      "grad_norm": 3.0434484481811523,
      "learning_rate": 2.5272130607050848e-05,
      "loss": 0.1396,
      "num_input_tokens_seen": 13614472,
      "step": 20870
    },
    {
      "epoch": 10.94077568134172,
      "grad_norm": 2.3142263889312744,
      "learning_rate": 2.5260696971487802e-05,
      "loss": 0.0773,
      "num_input_tokens_seen": 13617384,
      "step": 20875
    },
    {
      "epoch": 10.943396226415095,
      "grad_norm": 2.6792187690734863,
      "learning_rate": 2.5249263281389923e-05,
      "loss": 0.1132,
      "num_input_tokens_seen": 13620968,
      "step": 20880
    },
    {
      "epoch": 10.94601677148847,
      "grad_norm": 2.0191702842712402,
      "learning_rate": 2.5237829539149004e-05,
      "loss": 0.108,
      "num_input_tokens_seen": 13623752,
      "step": 20885
    },
    {
      "epoch": 10.948637316561845,
      "grad_norm": 2.7401599884033203,
      "learning_rate": 2.5226395747156866e-05,
      "loss": 0.1115,
      "num_input_tokens_seen": 13629000,
      "step": 20890
    },
    {
      "epoch": 10.95125786163522,
      "grad_norm": 5.424095630645752,
      "learning_rate": 2.5214961907805317e-05,
      "loss": 0.0877,
      "num_input_tokens_seen": 13632200,
      "step": 20895
    },
    {
      "epoch": 10.953878406708595,
      "grad_norm": 2.9979400634765625,
      "learning_rate": 2.520352802348619e-05,
      "loss": 0.0963,
      "num_input_tokens_seen": 13634536,
      "step": 20900
    },
    {
      "epoch": 10.95649895178197,
      "grad_norm": 1.709136962890625,
      "learning_rate": 2.5192094096591323e-05,
      "loss": 0.1847,
      "num_input_tokens_seen": 13638760,
      "step": 20905
    },
    {
      "epoch": 10.959119496855346,
      "grad_norm": 2.5063157081604004,
      "learning_rate": 2.5180660129512556e-05,
      "loss": 0.0754,
      "num_input_tokens_seen": 13641256,
      "step": 20910
    },
    {
      "epoch": 10.96174004192872,
      "grad_norm": 2.6232709884643555,
      "learning_rate": 2.516922612464176e-05,
      "loss": 0.0829,
      "num_input_tokens_seen": 13644808,
      "step": 20915
    },
    {
      "epoch": 10.964360587002096,
      "grad_norm": 2.2516791820526123,
      "learning_rate": 2.5157792084370783e-05,
      "loss": 0.1219,
      "num_input_tokens_seen": 13648168,
      "step": 20920
    },
    {
      "epoch": 10.966981132075471,
      "grad_norm": 1.5616695880889893,
      "learning_rate": 2.5146358011091502e-05,
      "loss": 0.1551,
      "num_input_tokens_seen": 13652520,
      "step": 20925
    },
    {
      "epoch": 10.969601677148846,
      "grad_norm": 2.9731736183166504,
      "learning_rate": 2.5134923907195805e-05,
      "loss": 0.1073,
      "num_input_tokens_seen": 13655656,
      "step": 20930
    },
    {
      "epoch": 10.972222222222221,
      "grad_norm": 1.8826667070388794,
      "learning_rate": 2.5123489775075558e-05,
      "loss": 0.0893,
      "num_input_tokens_seen": 13658440,
      "step": 20935
    },
    {
      "epoch": 10.974842767295598,
      "grad_norm": 11.097185134887695,
      "learning_rate": 2.511205561712266e-05,
      "loss": 0.0828,
      "num_input_tokens_seen": 13661160,
      "step": 20940
    },
    {
      "epoch": 10.977463312368974,
      "grad_norm": 2.72943377494812,
      "learning_rate": 2.5100621435729004e-05,
      "loss": 0.0835,
      "num_input_tokens_seen": 13664936,
      "step": 20945
    },
    {
      "epoch": 10.980083857442349,
      "grad_norm": 1.7040470838546753,
      "learning_rate": 2.5089187233286492e-05,
      "loss": 0.1202,
      "num_input_tokens_seen": 13668072,
      "step": 20950
    },
    {
      "epoch": 10.982704402515724,
      "grad_norm": 1.4353361129760742,
      "learning_rate": 2.5077753012187026e-05,
      "loss": 0.076,
      "num_input_tokens_seen": 13670856,
      "step": 20955
    },
    {
      "epoch": 10.985324947589099,
      "grad_norm": 2.931037664413452,
      "learning_rate": 2.506631877482252e-05,
      "loss": 0.1192,
      "num_input_tokens_seen": 13674792,
      "step": 20960
    },
    {
      "epoch": 10.987945492662474,
      "grad_norm": 2.6746888160705566,
      "learning_rate": 2.505488452358488e-05,
      "loss": 0.1197,
      "num_input_tokens_seen": 13677544,
      "step": 20965
    },
    {
      "epoch": 10.99056603773585,
      "grad_norm": 1.7486103773117065,
      "learning_rate": 2.504345026086603e-05,
      "loss": 0.1014,
      "num_input_tokens_seen": 13681896,
      "step": 20970
    },
    {
      "epoch": 10.993186582809225,
      "grad_norm": 3.114417552947998,
      "learning_rate": 2.5032015989057877e-05,
      "loss": 0.0724,
      "num_input_tokens_seen": 13685256,
      "step": 20975
    },
    {
      "epoch": 10.9958071278826,
      "grad_norm": 6.015811920166016,
      "learning_rate": 2.5020581710552348e-05,
      "loss": 0.1448,
      "num_input_tokens_seen": 13688072,
      "step": 20980
    },
    {
      "epoch": 10.998427672955975,
      "grad_norm": 2.082754135131836,
      "learning_rate": 2.5009147427741358e-05,
      "loss": 0.0753,
      "num_input_tokens_seen": 13690824,
      "step": 20985
    },
    {
      "epoch": 11.0,
      "eval_loss": 0.9029877781867981,
      "eval_runtime": 11.9203,
      "eval_samples_per_second": 71.139,
      "eval_steps_per_second": 17.785,
      "num_input_tokens_seen": 13691904,
      "step": 20988
    },
    {
      "epoch": 11.00104821802935,
      "grad_norm": 1.5027446746826172,
      "learning_rate": 2.4997713143016822e-05,
      "loss": 0.0578,
      "num_input_tokens_seen": 13692896,
      "step": 20990
    },
    {
      "epoch": 11.003668763102725,
      "grad_norm": 2.3038840293884277,
      "learning_rate": 2.498627885877068e-05,
      "loss": 0.0591,
      "num_input_tokens_seen": 13696160,
      "step": 20995
    },
    {
      "epoch": 11.0062893081761,
      "grad_norm": 1.4186824560165405,
      "learning_rate": 2.4974844577394846e-05,
      "loss": 0.0962,
      "num_input_tokens_seen": 13699136,
      "step": 21000
    },
    {
      "epoch": 11.008909853249476,
      "grad_norm": 1.7453798055648804,
      "learning_rate": 2.496341030128123e-05,
      "loss": 0.0637,
      "num_input_tokens_seen": 13702496,
      "step": 21005
    },
    {
      "epoch": 11.01153039832285,
      "grad_norm": 3.7096850872039795,
      "learning_rate": 2.495197603282177e-05,
      "loss": 0.071,
      "num_input_tokens_seen": 13706048,
      "step": 21010
    },
    {
      "epoch": 11.014150943396226,
      "grad_norm": 2.843912124633789,
      "learning_rate": 2.4940541774408368e-05,
      "loss": 0.0548,
      "num_input_tokens_seen": 13708640,
      "step": 21015
    },
    {
      "epoch": 11.016771488469601,
      "grad_norm": 2.069382667541504,
      "learning_rate": 2.4929107528432953e-05,
      "loss": 0.0478,
      "num_input_tokens_seen": 13712000,
      "step": 21020
    },
    {
      "epoch": 11.019392033542976,
      "grad_norm": 2.1734445095062256,
      "learning_rate": 2.4917673297287446e-05,
      "loss": 0.0446,
      "num_input_tokens_seen": 13715232,
      "step": 21025
    },
    {
      "epoch": 11.022012578616351,
      "grad_norm": 2.052694797515869,
      "learning_rate": 2.4906239083363728e-05,
      "loss": 0.2169,
      "num_input_tokens_seen": 13718368,
      "step": 21030
    },
    {
      "epoch": 11.024633123689727,
      "grad_norm": 3.2636008262634277,
      "learning_rate": 2.4894804889053732e-05,
      "loss": 0.1241,
      "num_input_tokens_seen": 13721824,
      "step": 21035
    },
    {
      "epoch": 11.027253668763104,
      "grad_norm": 1.0204224586486816,
      "learning_rate": 2.4883370716749347e-05,
      "loss": 0.0662,
      "num_input_tokens_seen": 13724480,
      "step": 21040
    },
    {
      "epoch": 11.029874213836479,
      "grad_norm": 1.5919456481933594,
      "learning_rate": 2.4871936568842486e-05,
      "loss": 0.0274,
      "num_input_tokens_seen": 13726656,
      "step": 21045
    },
    {
      "epoch": 11.032494758909854,
      "grad_norm": 4.39977502822876,
      "learning_rate": 2.4860502447725033e-05,
      "loss": 0.0619,
      "num_input_tokens_seen": 13729184,
      "step": 21050
    },
    {
      "epoch": 11.035115303983229,
      "grad_norm": 1.7164030075073242,
      "learning_rate": 2.4849068355788873e-05,
      "loss": 0.073,
      "num_input_tokens_seen": 13733600,
      "step": 21055
    },
    {
      "epoch": 11.037735849056604,
      "grad_norm": 2.682537317276001,
      "learning_rate": 2.4837634295425892e-05,
      "loss": 0.1227,
      "num_input_tokens_seen": 13736832,
      "step": 21060
    },
    {
      "epoch": 11.04035639412998,
      "grad_norm": 3.3466956615448,
      "learning_rate": 2.4826200269027957e-05,
      "loss": 0.0552,
      "num_input_tokens_seen": 13739360,
      "step": 21065
    },
    {
      "epoch": 11.042976939203355,
      "grad_norm": 1.2664868831634521,
      "learning_rate": 2.481476627898695e-05,
      "loss": 0.0425,
      "num_input_tokens_seen": 13743296,
      "step": 21070
    },
    {
      "epoch": 11.04559748427673,
      "grad_norm": 6.151544570922852,
      "learning_rate": 2.4803332327694724e-05,
      "loss": 0.0542,
      "num_input_tokens_seen": 13747072,
      "step": 21075
    },
    {
      "epoch": 11.048218029350105,
      "grad_norm": 1.8646790981292725,
      "learning_rate": 2.4791898417543136e-05,
      "loss": 0.0774,
      "num_input_tokens_seen": 13750208,
      "step": 21080
    },
    {
      "epoch": 11.05083857442348,
      "grad_norm": 6.053249359130859,
      "learning_rate": 2.4780464550924014e-05,
      "loss": 0.0816,
      "num_input_tokens_seen": 13752928,
      "step": 21085
    },
    {
      "epoch": 11.053459119496855,
      "grad_norm": 1.5954794883728027,
      "learning_rate": 2.4769030730229213e-05,
      "loss": 0.0829,
      "num_input_tokens_seen": 13756736,
      "step": 21090
    },
    {
      "epoch": 11.05607966457023,
      "grad_norm": 5.50632381439209,
      "learning_rate": 2.475759695785054e-05,
      "loss": 0.0747,
      "num_input_tokens_seen": 13759520,
      "step": 21095
    },
    {
      "epoch": 11.058700209643606,
      "grad_norm": 1.7260191440582275,
      "learning_rate": 2.4746163236179822e-05,
      "loss": 0.0749,
      "num_input_tokens_seen": 13762592,
      "step": 21100
    },
    {
      "epoch": 11.06132075471698,
      "grad_norm": 1.7735029458999634,
      "learning_rate": 2.4734729567608858e-05,
      "loss": 0.09,
      "num_input_tokens_seen": 13766464,
      "step": 21105
    },
    {
      "epoch": 11.063941299790356,
      "grad_norm": 0.2832834720611572,
      "learning_rate": 2.472329595452945e-05,
      "loss": 0.0589,
      "num_input_tokens_seen": 13772672,
      "step": 21110
    },
    {
      "epoch": 11.066561844863731,
      "grad_norm": 1.783766508102417,
      "learning_rate": 2.4711862399333367e-05,
      "loss": 0.0696,
      "num_input_tokens_seen": 13776000,
      "step": 21115
    },
    {
      "epoch": 11.069182389937106,
      "grad_norm": 7.909259796142578,
      "learning_rate": 2.4700428904412377e-05,
      "loss": 0.1525,
      "num_input_tokens_seen": 13778464,
      "step": 21120
    },
    {
      "epoch": 11.071802935010481,
      "grad_norm": 2.092740535736084,
      "learning_rate": 2.4688995472158245e-05,
      "loss": 0.0554,
      "num_input_tokens_seen": 13781536,
      "step": 21125
    },
    {
      "epoch": 11.074423480083857,
      "grad_norm": 3.6644954681396484,
      "learning_rate": 2.4677562104962712e-05,
      "loss": 0.0533,
      "num_input_tokens_seen": 13784800,
      "step": 21130
    },
    {
      "epoch": 11.077044025157234,
      "grad_norm": 1.129807949066162,
      "learning_rate": 2.4666128805217513e-05,
      "loss": 0.0744,
      "num_input_tokens_seen": 13788384,
      "step": 21135
    },
    {
      "epoch": 11.079664570230609,
      "grad_norm": 2.113585948944092,
      "learning_rate": 2.465469557531436e-05,
      "loss": 0.1164,
      "num_input_tokens_seen": 13791744,
      "step": 21140
    },
    {
      "epoch": 11.082285115303984,
      "grad_norm": 2.042508840560913,
      "learning_rate": 2.4643262417644944e-05,
      "loss": 0.0664,
      "num_input_tokens_seen": 13798304,
      "step": 21145
    },
    {
      "epoch": 11.084905660377359,
      "grad_norm": 2.619443893432617,
      "learning_rate": 2.463182933460096e-05,
      "loss": 0.0689,
      "num_input_tokens_seen": 13801184,
      "step": 21150
    },
    {
      "epoch": 11.087526205450734,
      "grad_norm": 1.5908523797988892,
      "learning_rate": 2.4620396328574087e-05,
      "loss": 0.0845,
      "num_input_tokens_seen": 13805120,
      "step": 21155
    },
    {
      "epoch": 11.09014675052411,
      "grad_norm": 1.327642560005188,
      "learning_rate": 2.4608963401955965e-05,
      "loss": 0.0478,
      "num_input_tokens_seen": 13808448,
      "step": 21160
    },
    {
      "epoch": 11.092767295597485,
      "grad_norm": 2.752108097076416,
      "learning_rate": 2.4597530557138247e-05,
      "loss": 0.0486,
      "num_input_tokens_seen": 13812320,
      "step": 21165
    },
    {
      "epoch": 11.09538784067086,
      "grad_norm": 0.6602232456207275,
      "learning_rate": 2.4586097796512538e-05,
      "loss": 0.0423,
      "num_input_tokens_seen": 13815168,
      "step": 21170
    },
    {
      "epoch": 11.098008385744235,
      "grad_norm": 3.0462539196014404,
      "learning_rate": 2.457466512247045e-05,
      "loss": 0.0509,
      "num_input_tokens_seen": 13817696,
      "step": 21175
    },
    {
      "epoch": 11.10062893081761,
      "grad_norm": 1.3802967071533203,
      "learning_rate": 2.4563232537403563e-05,
      "loss": 0.0954,
      "num_input_tokens_seen": 13820320,
      "step": 21180
    },
    {
      "epoch": 11.103249475890985,
      "grad_norm": 8.771728515625,
      "learning_rate": 2.4551800043703442e-05,
      "loss": 0.0734,
      "num_input_tokens_seen": 13823680,
      "step": 21185
    },
    {
      "epoch": 11.10587002096436,
      "grad_norm": 2.044081926345825,
      "learning_rate": 2.454036764376165e-05,
      "loss": 0.0571,
      "num_input_tokens_seen": 13826016,
      "step": 21190
    },
    {
      "epoch": 11.108490566037736,
      "grad_norm": 3.738626003265381,
      "learning_rate": 2.4528935339969705e-05,
      "loss": 0.0517,
      "num_input_tokens_seen": 13828416,
      "step": 21195
    },
    {
      "epoch": 11.11111111111111,
      "grad_norm": 7.414496421813965,
      "learning_rate": 2.45175031347191e-05,
      "loss": 0.0875,
      "num_input_tokens_seen": 13830976,
      "step": 21200
    },
    {
      "epoch": 11.113731656184486,
      "grad_norm": 1.222886562347412,
      "learning_rate": 2.4506071030401344e-05,
      "loss": 0.0669,
      "num_input_tokens_seen": 13833952,
      "step": 21205
    },
    {
      "epoch": 11.116352201257861,
      "grad_norm": 3.230543613433838,
      "learning_rate": 2.4494639029407884e-05,
      "loss": 0.0414,
      "num_input_tokens_seen": 13836224,
      "step": 21210
    },
    {
      "epoch": 11.118972746331236,
      "grad_norm": 4.9857892990112305,
      "learning_rate": 2.448320713413018e-05,
      "loss": 0.0693,
      "num_input_tokens_seen": 13840416,
      "step": 21215
    },
    {
      "epoch": 11.121593291404611,
      "grad_norm": 3.4078028202056885,
      "learning_rate": 2.4471775346959654e-05,
      "loss": 0.0335,
      "num_input_tokens_seen": 13843072,
      "step": 21220
    },
    {
      "epoch": 11.124213836477987,
      "grad_norm": 2.064009666442871,
      "learning_rate": 2.446034367028769e-05,
      "loss": 0.0898,
      "num_input_tokens_seen": 13845728,
      "step": 21225
    },
    {
      "epoch": 11.126834381551364,
      "grad_norm": 1.860490083694458,
      "learning_rate": 2.444891210650567e-05,
      "loss": 0.0368,
      "num_input_tokens_seen": 13849472,
      "step": 21230
    },
    {
      "epoch": 11.129454926624739,
      "grad_norm": 2.0644192695617676,
      "learning_rate": 2.4437480658004958e-05,
      "loss": 0.0667,
      "num_input_tokens_seen": 13852704,
      "step": 21235
    },
    {
      "epoch": 11.132075471698114,
      "grad_norm": 1.423624038696289,
      "learning_rate": 2.4426049327176868e-05,
      "loss": 0.0926,
      "num_input_tokens_seen": 13856000,
      "step": 21240
    },
    {
      "epoch": 11.134696016771489,
      "grad_norm": 5.354130744934082,
      "learning_rate": 2.4414618116412716e-05,
      "loss": 0.0823,
      "num_input_tokens_seen": 13858944,
      "step": 21245
    },
    {
      "epoch": 11.137316561844864,
      "grad_norm": 1.3496263027191162,
      "learning_rate": 2.440318702810378e-05,
      "loss": 0.0845,
      "num_input_tokens_seen": 13861504,
      "step": 21250
    },
    {
      "epoch": 11.13993710691824,
      "grad_norm": 4.5537190437316895,
      "learning_rate": 2.43917560646413e-05,
      "loss": 0.0686,
      "num_input_tokens_seen": 13865088,
      "step": 21255
    },
    {
      "epoch": 11.142557651991615,
      "grad_norm": 1.3058509826660156,
      "learning_rate": 2.4380325228416518e-05,
      "loss": 0.0821,
      "num_input_tokens_seen": 13867488,
      "step": 21260
    },
    {
      "epoch": 11.14517819706499,
      "grad_norm": 1.5958778858184814,
      "learning_rate": 2.4368894521820624e-05,
      "loss": 0.0722,
      "num_input_tokens_seen": 13870912,
      "step": 21265
    },
    {
      "epoch": 11.147798742138365,
      "grad_norm": 3.1558237075805664,
      "learning_rate": 2.43574639472448e-05,
      "loss": 0.0427,
      "num_input_tokens_seen": 13874080,
      "step": 21270
    },
    {
      "epoch": 11.15041928721174,
      "grad_norm": 2.631577491760254,
      "learning_rate": 2.4346033507080178e-05,
      "loss": 0.1317,
      "num_input_tokens_seen": 13877920,
      "step": 21275
    },
    {
      "epoch": 11.153039832285115,
      "grad_norm": 1.9219034910202026,
      "learning_rate": 2.4334603203717903e-05,
      "loss": 0.0709,
      "num_input_tokens_seen": 13880608,
      "step": 21280
    },
    {
      "epoch": 11.15566037735849,
      "grad_norm": 1.269942045211792,
      "learning_rate": 2.432317303954904e-05,
      "loss": 0.2004,
      "num_input_tokens_seen": 13883936,
      "step": 21285
    },
    {
      "epoch": 11.158280922431866,
      "grad_norm": 1.9025956392288208,
      "learning_rate": 2.4311743016964648e-05,
      "loss": 0.1636,
      "num_input_tokens_seen": 13887680,
      "step": 21290
    },
    {
      "epoch": 11.16090146750524,
      "grad_norm": 1.291785478591919,
      "learning_rate": 2.4300313138355772e-05,
      "loss": 0.082,
      "num_input_tokens_seen": 13890592,
      "step": 21295
    },
    {
      "epoch": 11.163522012578616,
      "grad_norm": 2.319446086883545,
      "learning_rate": 2.4288883406113396e-05,
      "loss": 0.0653,
      "num_input_tokens_seen": 13892704,
      "step": 21300
    },
    {
      "epoch": 11.166142557651991,
      "grad_norm": 3.9596269130706787,
      "learning_rate": 2.4277453822628512e-05,
      "loss": 0.0461,
      "num_input_tokens_seen": 13895616,
      "step": 21305
    },
    {
      "epoch": 11.168763102725366,
      "grad_norm": 2.199262857437134,
      "learning_rate": 2.4266024390292037e-05,
      "loss": 0.0786,
      "num_input_tokens_seen": 13898688,
      "step": 21310
    },
    {
      "epoch": 11.171383647798741,
      "grad_norm": 3.914546251296997,
      "learning_rate": 2.425459511149489e-05,
      "loss": 0.0688,
      "num_input_tokens_seen": 13901760,
      "step": 21315
    },
    {
      "epoch": 11.174004192872117,
      "grad_norm": 3.6471140384674072,
      "learning_rate": 2.424316598862793e-05,
      "loss": 0.0715,
      "num_input_tokens_seen": 13904416,
      "step": 21320
    },
    {
      "epoch": 11.176624737945493,
      "grad_norm": 1.9130146503448486,
      "learning_rate": 2.4231737024082023e-05,
      "loss": 0.0576,
      "num_input_tokens_seen": 13908192,
      "step": 21325
    },
    {
      "epoch": 11.179245283018869,
      "grad_norm": 0.9427489638328552,
      "learning_rate": 2.4220308220247954e-05,
      "loss": 0.0799,
      "num_input_tokens_seen": 13911392,
      "step": 21330
    },
    {
      "epoch": 11.181865828092244,
      "grad_norm": 3.0920767784118652,
      "learning_rate": 2.4208879579516528e-05,
      "loss": 0.0676,
      "num_input_tokens_seen": 13914368,
      "step": 21335
    },
    {
      "epoch": 11.184486373165619,
      "grad_norm": 1.0019235610961914,
      "learning_rate": 2.419745110427846e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 13916544,
      "step": 21340
    },
    {
      "epoch": 11.187106918238994,
      "grad_norm": 1.8966742753982544,
      "learning_rate": 2.418602279692446e-05,
      "loss": 0.1144,
      "num_input_tokens_seen": 13919520,
      "step": 21345
    },
    {
      "epoch": 11.18972746331237,
      "grad_norm": 3.0654585361480713,
      "learning_rate": 2.417459465984521e-05,
      "loss": 0.0561,
      "num_input_tokens_seen": 13922496,
      "step": 21350
    },
    {
      "epoch": 11.192348008385745,
      "grad_norm": 1.7702330350875854,
      "learning_rate": 2.4163166695431336e-05,
      "loss": 0.057,
      "num_input_tokens_seen": 13925440,
      "step": 21355
    },
    {
      "epoch": 11.19496855345912,
      "grad_norm": 3.1203434467315674,
      "learning_rate": 2.4151738906073446e-05,
      "loss": 0.0867,
      "num_input_tokens_seen": 13928544,
      "step": 21360
    },
    {
      "epoch": 11.197589098532495,
      "grad_norm": 1.8195722103118896,
      "learning_rate": 2.4140311294162105e-05,
      "loss": 0.0403,
      "num_input_tokens_seen": 13931200,
      "step": 21365
    },
    {
      "epoch": 11.20020964360587,
      "grad_norm": 10.73656940460205,
      "learning_rate": 2.412888386208782e-05,
      "loss": 0.1003,
      "num_input_tokens_seen": 13933984,
      "step": 21370
    },
    {
      "epoch": 11.202830188679245,
      "grad_norm": 1.9712278842926025,
      "learning_rate": 2.4117456612241103e-05,
      "loss": 0.0773,
      "num_input_tokens_seen": 13937184,
      "step": 21375
    },
    {
      "epoch": 11.20545073375262,
      "grad_norm": 1.6552625894546509,
      "learning_rate": 2.410602954701239e-05,
      "loss": 0.0787,
      "num_input_tokens_seen": 13940864,
      "step": 21380
    },
    {
      "epoch": 11.208071278825996,
      "grad_norm": 0.9852792024612427,
      "learning_rate": 2.409460266879209e-05,
      "loss": 0.0644,
      "num_input_tokens_seen": 13943904,
      "step": 21385
    },
    {
      "epoch": 11.21069182389937,
      "grad_norm": 1.5681148767471313,
      "learning_rate": 2.4083175979970592e-05,
      "loss": 0.0827,
      "num_input_tokens_seen": 13948192,
      "step": 21390
    },
    {
      "epoch": 11.213312368972746,
      "grad_norm": 2.1081902980804443,
      "learning_rate": 2.4071749482938226e-05,
      "loss": 0.0393,
      "num_input_tokens_seen": 13950912,
      "step": 21395
    },
    {
      "epoch": 11.215932914046121,
      "grad_norm": 1.8915884494781494,
      "learning_rate": 2.4060323180085266e-05,
      "loss": 0.0549,
      "num_input_tokens_seen": 13953664,
      "step": 21400
    },
    {
      "epoch": 11.218553459119496,
      "grad_norm": 2.530839443206787,
      "learning_rate": 2.404889707380198e-05,
      "loss": 0.0675,
      "num_input_tokens_seen": 13956576,
      "step": 21405
    },
    {
      "epoch": 11.221174004192871,
      "grad_norm": 1.6953712701797485,
      "learning_rate": 2.403747116647857e-05,
      "loss": 0.0563,
      "num_input_tokens_seen": 13959520,
      "step": 21410
    },
    {
      "epoch": 11.223794549266247,
      "grad_norm": 1.652655839920044,
      "learning_rate": 2.4026045460505216e-05,
      "loss": 0.0941,
      "num_input_tokens_seen": 13963552,
      "step": 21415
    },
    {
      "epoch": 11.226415094339623,
      "grad_norm": 2.181793451309204,
      "learning_rate": 2.4014619958272047e-05,
      "loss": 0.0814,
      "num_input_tokens_seen": 13967072,
      "step": 21420
    },
    {
      "epoch": 11.229035639412999,
      "grad_norm": 1.4891053438186646,
      "learning_rate": 2.4003194662169132e-05,
      "loss": 0.0389,
      "num_input_tokens_seen": 13970592,
      "step": 21425
    },
    {
      "epoch": 11.231656184486374,
      "grad_norm": 2.1714582443237305,
      "learning_rate": 2.3991769574586522e-05,
      "loss": 0.072,
      "num_input_tokens_seen": 13973696,
      "step": 21430
    },
    {
      "epoch": 11.234276729559749,
      "grad_norm": 1.692988634109497,
      "learning_rate": 2.3980344697914215e-05,
      "loss": 0.1029,
      "num_input_tokens_seen": 13977248,
      "step": 21435
    },
    {
      "epoch": 11.236897274633124,
      "grad_norm": 3.300014019012451,
      "learning_rate": 2.3968920034542168e-05,
      "loss": 0.0955,
      "num_input_tokens_seen": 13980704,
      "step": 21440
    },
    {
      "epoch": 11.2395178197065,
      "grad_norm": 1.1932857036590576,
      "learning_rate": 2.3957495586860278e-05,
      "loss": 0.0561,
      "num_input_tokens_seen": 13982976,
      "step": 21445
    },
    {
      "epoch": 11.242138364779874,
      "grad_norm": 0.5658292770385742,
      "learning_rate": 2.394607135725843e-05,
      "loss": 0.0738,
      "num_input_tokens_seen": 13985888,
      "step": 21450
    },
    {
      "epoch": 11.24475890985325,
      "grad_norm": 1.498225450515747,
      "learning_rate": 2.3934647348126426e-05,
      "loss": 0.0671,
      "num_input_tokens_seen": 13990048,
      "step": 21455
    },
    {
      "epoch": 11.247379454926625,
      "grad_norm": 3.394566059112549,
      "learning_rate": 2.3923223561854035e-05,
      "loss": 0.0622,
      "num_input_tokens_seen": 13993152,
      "step": 21460
    },
    {
      "epoch": 11.25,
      "grad_norm": 1.6433675289154053,
      "learning_rate": 2.391180000083099e-05,
      "loss": 0.0544,
      "num_input_tokens_seen": 13995776,
      "step": 21465
    },
    {
      "epoch": 11.252620545073375,
      "grad_norm": 2.2992303371429443,
      "learning_rate": 2.390037666744698e-05,
      "loss": 0.0964,
      "num_input_tokens_seen": 13999168,
      "step": 21470
    },
    {
      "epoch": 11.25524109014675,
      "grad_norm": 1.2782062292099,
      "learning_rate": 2.388895356409162e-05,
      "loss": 0.0525,
      "num_input_tokens_seen": 14002048,
      "step": 21475
    },
    {
      "epoch": 11.257861635220126,
      "grad_norm": 1.6180765628814697,
      "learning_rate": 2.3877530693154506e-05,
      "loss": 0.0755,
      "num_input_tokens_seen": 14004864,
      "step": 21480
    },
    {
      "epoch": 11.2604821802935,
      "grad_norm": 4.870395660400391,
      "learning_rate": 2.3866108057025162e-05,
      "loss": 0.0674,
      "num_input_tokens_seen": 14008352,
      "step": 21485
    },
    {
      "epoch": 11.263102725366876,
      "grad_norm": 2.5199193954467773,
      "learning_rate": 2.385468565809307e-05,
      "loss": 0.0403,
      "num_input_tokens_seen": 14011232,
      "step": 21490
    },
    {
      "epoch": 11.265723270440251,
      "grad_norm": 1.4059765338897705,
      "learning_rate": 2.3843263498747675e-05,
      "loss": 0.0815,
      "num_input_tokens_seen": 14014848,
      "step": 21495
    },
    {
      "epoch": 11.268343815513626,
      "grad_norm": 2.236309766769409,
      "learning_rate": 2.383184158137836e-05,
      "loss": 0.0589,
      "num_input_tokens_seen": 14017536,
      "step": 21500
    },
    {
      "epoch": 11.270964360587001,
      "grad_norm": 2.285006284713745,
      "learning_rate": 2.3820419908374465e-05,
      "loss": 0.076,
      "num_input_tokens_seen": 14021376,
      "step": 21505
    },
    {
      "epoch": 11.273584905660377,
      "grad_norm": 1.2487822771072388,
      "learning_rate": 2.3808998482125265e-05,
      "loss": 0.0818,
      "num_input_tokens_seen": 14024224,
      "step": 21510
    },
    {
      "epoch": 11.276205450733753,
      "grad_norm": 2.1438889503479004,
      "learning_rate": 2.3797577305019987e-05,
      "loss": 0.0834,
      "num_input_tokens_seen": 14026848,
      "step": 21515
    },
    {
      "epoch": 11.278825995807129,
      "grad_norm": 1.4285345077514648,
      "learning_rate": 2.378615637944783e-05,
      "loss": 0.0616,
      "num_input_tokens_seen": 14030720,
      "step": 21520
    },
    {
      "epoch": 11.281446540880504,
      "grad_norm": 1.3746497631072998,
      "learning_rate": 2.3774735707797895e-05,
      "loss": 0.0748,
      "num_input_tokens_seen": 14033632,
      "step": 21525
    },
    {
      "epoch": 11.284067085953879,
      "grad_norm": 1.8842288255691528,
      "learning_rate": 2.376331529245928e-05,
      "loss": 0.0603,
      "num_input_tokens_seen": 14036960,
      "step": 21530
    },
    {
      "epoch": 11.286687631027254,
      "grad_norm": 1.6688518524169922,
      "learning_rate": 2.3751895135821006e-05,
      "loss": 0.0887,
      "num_input_tokens_seen": 14039648,
      "step": 21535
    },
    {
      "epoch": 11.28930817610063,
      "grad_norm": 0.8833516836166382,
      "learning_rate": 2.374047524027202e-05,
      "loss": 0.1396,
      "num_input_tokens_seen": 14043840,
      "step": 21540
    },
    {
      "epoch": 11.291928721174004,
      "grad_norm": 2.6627275943756104,
      "learning_rate": 2.3729055608201243e-05,
      "loss": 0.0653,
      "num_input_tokens_seen": 14047200,
      "step": 21545
    },
    {
      "epoch": 11.29454926624738,
      "grad_norm": 0.8029502034187317,
      "learning_rate": 2.3717636241997537e-05,
      "loss": 0.043,
      "num_input_tokens_seen": 14050336,
      "step": 21550
    },
    {
      "epoch": 11.297169811320755,
      "grad_norm": 2.3172831535339355,
      "learning_rate": 2.3706217144049696e-05,
      "loss": 0.096,
      "num_input_tokens_seen": 14053312,
      "step": 21555
    },
    {
      "epoch": 11.29979035639413,
      "grad_norm": 1.9385241270065308,
      "learning_rate": 2.369479831674647e-05,
      "loss": 0.0892,
      "num_input_tokens_seen": 14056256,
      "step": 21560
    },
    {
      "epoch": 11.302410901467505,
      "grad_norm": 6.340381622314453,
      "learning_rate": 2.3683379762476555e-05,
      "loss": 0.063,
      "num_input_tokens_seen": 14059168,
      "step": 21565
    },
    {
      "epoch": 11.30503144654088,
      "grad_norm": 1.5453360080718994,
      "learning_rate": 2.3671961483628562e-05,
      "loss": 0.0628,
      "num_input_tokens_seen": 14062720,
      "step": 21570
    },
    {
      "epoch": 11.307651991614255,
      "grad_norm": 2.19832444190979,
      "learning_rate": 2.3660543482591078e-05,
      "loss": 0.0467,
      "num_input_tokens_seen": 14065952,
      "step": 21575
    },
    {
      "epoch": 11.31027253668763,
      "grad_norm": 1.4740239381790161,
      "learning_rate": 2.364912576175261e-05,
      "loss": 0.0906,
      "num_input_tokens_seen": 14069664,
      "step": 21580
    },
    {
      "epoch": 11.312893081761006,
      "grad_norm": 2.792006254196167,
      "learning_rate": 2.363770832350163e-05,
      "loss": 0.0683,
      "num_input_tokens_seen": 14073024,
      "step": 21585
    },
    {
      "epoch": 11.315513626834381,
      "grad_norm": 1.578382134437561,
      "learning_rate": 2.362629117022653e-05,
      "loss": 0.0605,
      "num_input_tokens_seen": 14076576,
      "step": 21590
    },
    {
      "epoch": 11.318134171907756,
      "grad_norm": 4.456874370574951,
      "learning_rate": 2.3614874304315633e-05,
      "loss": 0.071,
      "num_input_tokens_seen": 14080128,
      "step": 21595
    },
    {
      "epoch": 11.320754716981131,
      "grad_norm": 2.1434030532836914,
      "learning_rate": 2.3603457728157237e-05,
      "loss": 0.0689,
      "num_input_tokens_seen": 14084320,
      "step": 21600
    },
    {
      "epoch": 11.323375262054507,
      "grad_norm": 3.459315299987793,
      "learning_rate": 2.3592041444139547e-05,
      "loss": 0.0671,
      "num_input_tokens_seen": 14087008,
      "step": 21605
    },
    {
      "epoch": 11.325995807127883,
      "grad_norm": 2.8173229694366455,
      "learning_rate": 2.358062545465073e-05,
      "loss": 0.0391,
      "num_input_tokens_seen": 14090208,
      "step": 21610
    },
    {
      "epoch": 11.328616352201259,
      "grad_norm": 2.2039761543273926,
      "learning_rate": 2.3569209762078866e-05,
      "loss": 0.059,
      "num_input_tokens_seen": 14093184,
      "step": 21615
    },
    {
      "epoch": 11.331236897274634,
      "grad_norm": 1.8912549018859863,
      "learning_rate": 2.355779436881201e-05,
      "loss": 0.102,
      "num_input_tokens_seen": 14100064,
      "step": 21620
    },
    {
      "epoch": 11.333857442348009,
      "grad_norm": 1.702404499053955,
      "learning_rate": 2.3546379277238107e-05,
      "loss": 0.0633,
      "num_input_tokens_seen": 14103104,
      "step": 21625
    },
    {
      "epoch": 11.336477987421384,
      "grad_norm": 2.736966609954834,
      "learning_rate": 2.353496448974508e-05,
      "loss": 0.0592,
      "num_input_tokens_seen": 14106496,
      "step": 21630
    },
    {
      "epoch": 11.33909853249476,
      "grad_norm": 1.4802379608154297,
      "learning_rate": 2.3523550008720763e-05,
      "loss": 0.114,
      "num_input_tokens_seen": 14110368,
      "step": 21635
    },
    {
      "epoch": 11.341719077568134,
      "grad_norm": 1.8761100769042969,
      "learning_rate": 2.3512135836552945e-05,
      "loss": 0.1618,
      "num_input_tokens_seen": 14114944,
      "step": 21640
    },
    {
      "epoch": 11.34433962264151,
      "grad_norm": 1.8369712829589844,
      "learning_rate": 2.350072197562933e-05,
      "loss": 0.0585,
      "num_input_tokens_seen": 14118272,
      "step": 21645
    },
    {
      "epoch": 11.346960167714885,
      "grad_norm": 2.125890016555786,
      "learning_rate": 2.3489308428337586e-05,
      "loss": 0.0771,
      "num_input_tokens_seen": 14121472,
      "step": 21650
    },
    {
      "epoch": 11.34958071278826,
      "grad_norm": 2.183425188064575,
      "learning_rate": 2.3477895197065275e-05,
      "loss": 0.07,
      "num_input_tokens_seen": 14123808,
      "step": 21655
    },
    {
      "epoch": 11.352201257861635,
      "grad_norm": 7.739718914031982,
      "learning_rate": 2.346648228419992e-05,
      "loss": 0.0664,
      "num_input_tokens_seen": 14126144,
      "step": 21660
    },
    {
      "epoch": 11.35482180293501,
      "grad_norm": 2.417710542678833,
      "learning_rate": 2.3455069692128982e-05,
      "loss": 0.0947,
      "num_input_tokens_seen": 14129888,
      "step": 21665
    },
    {
      "epoch": 11.357442348008385,
      "grad_norm": 6.306484699249268,
      "learning_rate": 2.3443657423239832e-05,
      "loss": 0.073,
      "num_input_tokens_seen": 14132448,
      "step": 21670
    },
    {
      "epoch": 11.36006289308176,
      "grad_norm": 3.083005428314209,
      "learning_rate": 2.34322454799198e-05,
      "loss": 0.0654,
      "num_input_tokens_seen": 14136736,
      "step": 21675
    },
    {
      "epoch": 11.362683438155136,
      "grad_norm": 1.0364341735839844,
      "learning_rate": 2.3420833864556133e-05,
      "loss": 0.0741,
      "num_input_tokens_seen": 14139840,
      "step": 21680
    },
    {
      "epoch": 11.365303983228511,
      "grad_norm": 1.5193250179290771,
      "learning_rate": 2.3409422579535995e-05,
      "loss": 0.058,
      "num_input_tokens_seen": 14142528,
      "step": 21685
    },
    {
      "epoch": 11.367924528301886,
      "grad_norm": 1.6575040817260742,
      "learning_rate": 2.339801162724651e-05,
      "loss": 0.0353,
      "num_input_tokens_seen": 14145984,
      "step": 21690
    },
    {
      "epoch": 11.370545073375261,
      "grad_norm": 1.4014216661453247,
      "learning_rate": 2.338660101007471e-05,
      "loss": 0.1205,
      "num_input_tokens_seen": 14149120,
      "step": 21695
    },
    {
      "epoch": 11.373165618448636,
      "grad_norm": 4.9356536865234375,
      "learning_rate": 2.3375190730407578e-05,
      "loss": 0.0419,
      "num_input_tokens_seen": 14151936,
      "step": 21700
    },
    {
      "epoch": 11.375786163522013,
      "grad_norm": 2.0204148292541504,
      "learning_rate": 2.336378079063202e-05,
      "loss": 0.0703,
      "num_input_tokens_seen": 14155040,
      "step": 21705
    },
    {
      "epoch": 11.378406708595389,
      "grad_norm": 2.511274814605713,
      "learning_rate": 2.335237119313484e-05,
      "loss": 0.0583,
      "num_input_tokens_seen": 14158336,
      "step": 21710
    },
    {
      "epoch": 11.381027253668764,
      "grad_norm": 18.398771286010742,
      "learning_rate": 2.3340961940302807e-05,
      "loss": 0.0674,
      "num_input_tokens_seen": 14161248,
      "step": 21715
    },
    {
      "epoch": 11.383647798742139,
      "grad_norm": 2.880824327468872,
      "learning_rate": 2.3329553034522616e-05,
      "loss": 0.0841,
      "num_input_tokens_seen": 14164960,
      "step": 21720
    },
    {
      "epoch": 11.386268343815514,
      "grad_norm": 2.3924572467803955,
      "learning_rate": 2.3318144478180866e-05,
      "loss": 0.0394,
      "num_input_tokens_seen": 14167584,
      "step": 21725
    },
    {
      "epoch": 11.38888888888889,
      "grad_norm": 2.5631303787231445,
      "learning_rate": 2.330673627366411e-05,
      "loss": 0.0456,
      "num_input_tokens_seen": 14169888,
      "step": 21730
    },
    {
      "epoch": 11.391509433962264,
      "grad_norm": 1.9010827541351318,
      "learning_rate": 2.329532842335881e-05,
      "loss": 0.0454,
      "num_input_tokens_seen": 14173024,
      "step": 21735
    },
    {
      "epoch": 11.39412997903564,
      "grad_norm": 3.13659930229187,
      "learning_rate": 2.328392092965135e-05,
      "loss": 0.0942,
      "num_input_tokens_seen": 14176576,
      "step": 21740
    },
    {
      "epoch": 11.396750524109015,
      "grad_norm": 6.0331807136535645,
      "learning_rate": 2.3272513794928054e-05,
      "loss": 0.0801,
      "num_input_tokens_seen": 14180064,
      "step": 21745
    },
    {
      "epoch": 11.39937106918239,
      "grad_norm": 6.196530818939209,
      "learning_rate": 2.326110702157516e-05,
      "loss": 0.0825,
      "num_input_tokens_seen": 14183456,
      "step": 21750
    },
    {
      "epoch": 11.401991614255765,
      "grad_norm": 0.8780474066734314,
      "learning_rate": 2.3249700611978847e-05,
      "loss": 0.0567,
      "num_input_tokens_seen": 14186752,
      "step": 21755
    },
    {
      "epoch": 11.40461215932914,
      "grad_norm": 3.6248602867126465,
      "learning_rate": 2.323829456852519e-05,
      "loss": 0.0365,
      "num_input_tokens_seen": 14189280,
      "step": 21760
    },
    {
      "epoch": 11.407232704402515,
      "grad_norm": 2.874077796936035,
      "learning_rate": 2.3226888893600223e-05,
      "loss": 0.0799,
      "num_input_tokens_seen": 14193120,
      "step": 21765
    },
    {
      "epoch": 11.40985324947589,
      "grad_norm": 3.0482370853424072,
      "learning_rate": 2.3215483589589865e-05,
      "loss": 0.096,
      "num_input_tokens_seen": 14196352,
      "step": 21770
    },
    {
      "epoch": 11.412473794549266,
      "grad_norm": 14.681803703308105,
      "learning_rate": 2.3204078658879974e-05,
      "loss": 0.1017,
      "num_input_tokens_seen": 14199456,
      "step": 21775
    },
    {
      "epoch": 11.415094339622641,
      "grad_norm": 1.4114214181900024,
      "learning_rate": 2.3192674103856337e-05,
      "loss": 0.0685,
      "num_input_tokens_seen": 14201600,
      "step": 21780
    },
    {
      "epoch": 11.417714884696016,
      "grad_norm": 0.8942345976829529,
      "learning_rate": 2.318126992690467e-05,
      "loss": 0.0983,
      "num_input_tokens_seen": 14204672,
      "step": 21785
    },
    {
      "epoch": 11.420335429769391,
      "grad_norm": 1.3328512907028198,
      "learning_rate": 2.3169866130410585e-05,
      "loss": 0.103,
      "num_input_tokens_seen": 14208928,
      "step": 21790
    },
    {
      "epoch": 11.422955974842766,
      "grad_norm": 1.005858302116394,
      "learning_rate": 2.315846271675962e-05,
      "loss": 0.049,
      "num_input_tokens_seen": 14213888,
      "step": 21795
    },
    {
      "epoch": 11.425576519916142,
      "grad_norm": 0.42186281085014343,
      "learning_rate": 2.3147059688337244e-05,
      "loss": 0.0363,
      "num_input_tokens_seen": 14216704,
      "step": 21800
    },
    {
      "epoch": 11.428197064989519,
      "grad_norm": 2.0913913249969482,
      "learning_rate": 2.313565704752884e-05,
      "loss": 0.0936,
      "num_input_tokens_seen": 14219616,
      "step": 21805
    },
    {
      "epoch": 11.430817610062894,
      "grad_norm": 3.0053741931915283,
      "learning_rate": 2.3124254796719718e-05,
      "loss": 0.0819,
      "num_input_tokens_seen": 14222368,
      "step": 21810
    },
    {
      "epoch": 11.433438155136269,
      "grad_norm": 1.4692319631576538,
      "learning_rate": 2.3112852938295088e-05,
      "loss": 0.0864,
      "num_input_tokens_seen": 14229888,
      "step": 21815
    },
    {
      "epoch": 11.436058700209644,
      "grad_norm": 1.9344491958618164,
      "learning_rate": 2.3101451474640104e-05,
      "loss": 0.0552,
      "num_input_tokens_seen": 14233440,
      "step": 21820
    },
    {
      "epoch": 11.43867924528302,
      "grad_norm": 2.8728818893432617,
      "learning_rate": 2.3090050408139808e-05,
      "loss": 0.0565,
      "num_input_tokens_seen": 14235488,
      "step": 21825
    },
    {
      "epoch": 11.441299790356394,
      "grad_norm": 5.254438400268555,
      "learning_rate": 2.307864974117917e-05,
      "loss": 0.0811,
      "num_input_tokens_seen": 14240128,
      "step": 21830
    },
    {
      "epoch": 11.44392033542977,
      "grad_norm": 2.6324994564056396,
      "learning_rate": 2.3067249476143087e-05,
      "loss": 0.0547,
      "num_input_tokens_seen": 14243264,
      "step": 21835
    },
    {
      "epoch": 11.446540880503145,
      "grad_norm": 2.38383412361145,
      "learning_rate": 2.3055849615416363e-05,
      "loss": 0.1735,
      "num_input_tokens_seen": 14247712,
      "step": 21840
    },
    {
      "epoch": 11.44916142557652,
      "grad_norm": 1.8171141147613525,
      "learning_rate": 2.304445016138372e-05,
      "loss": 0.0552,
      "num_input_tokens_seen": 14251360,
      "step": 21845
    },
    {
      "epoch": 11.451781970649895,
      "grad_norm": 2.9201455116271973,
      "learning_rate": 2.303305111642981e-05,
      "loss": 0.0687,
      "num_input_tokens_seen": 14254432,
      "step": 21850
    },
    {
      "epoch": 11.45440251572327,
      "grad_norm": 2.9346237182617188,
      "learning_rate": 2.3021652482939144e-05,
      "loss": 0.0612,
      "num_input_tokens_seen": 14257664,
      "step": 21855
    },
    {
      "epoch": 11.457023060796645,
      "grad_norm": 2.3052279949188232,
      "learning_rate": 2.3010254263296214e-05,
      "loss": 0.0671,
      "num_input_tokens_seen": 14260960,
      "step": 21860
    },
    {
      "epoch": 11.45964360587002,
      "grad_norm": 2.0384058952331543,
      "learning_rate": 2.2998856459885396e-05,
      "loss": 0.1231,
      "num_input_tokens_seen": 14264960,
      "step": 21865
    },
    {
      "epoch": 11.462264150943396,
      "grad_norm": 4.598583698272705,
      "learning_rate": 2.298745907509097e-05,
      "loss": 0.0704,
      "num_input_tokens_seen": 14267840,
      "step": 21870
    },
    {
      "epoch": 11.464884696016771,
      "grad_norm": 1.7523728609085083,
      "learning_rate": 2.2976062111297156e-05,
      "loss": 0.0822,
      "num_input_tokens_seen": 14270368,
      "step": 21875
    },
    {
      "epoch": 11.467505241090146,
      "grad_norm": 2.103588342666626,
      "learning_rate": 2.296466557088805e-05,
      "loss": 0.0566,
      "num_input_tokens_seen": 14273248,
      "step": 21880
    },
    {
      "epoch": 11.470125786163521,
      "grad_norm": 1.519741177558899,
      "learning_rate": 2.2953269456247688e-05,
      "loss": 0.0872,
      "num_input_tokens_seen": 14281472,
      "step": 21885
    },
    {
      "epoch": 11.472746331236896,
      "grad_norm": 3.047112464904785,
      "learning_rate": 2.2941873769760006e-05,
      "loss": 0.1069,
      "num_input_tokens_seen": 14284864,
      "step": 21890
    },
    {
      "epoch": 11.475366876310272,
      "grad_norm": 1.542165994644165,
      "learning_rate": 2.2930478513808844e-05,
      "loss": 0.0632,
      "num_input_tokens_seen": 14287968,
      "step": 21895
    },
    {
      "epoch": 11.477987421383649,
      "grad_norm": 1.0692861080169678,
      "learning_rate": 2.2919083690777976e-05,
      "loss": 0.0856,
      "num_input_tokens_seen": 14291392,
      "step": 21900
    },
    {
      "epoch": 11.480607966457024,
      "grad_norm": 2.220658302307129,
      "learning_rate": 2.2907689303051062e-05,
      "loss": 0.0696,
      "num_input_tokens_seen": 14294432,
      "step": 21905
    },
    {
      "epoch": 11.483228511530399,
      "grad_norm": 1.756933331489563,
      "learning_rate": 2.289629535301167e-05,
      "loss": 0.0546,
      "num_input_tokens_seen": 14299328,
      "step": 21910
    },
    {
      "epoch": 11.485849056603774,
      "grad_norm": 3.0920217037200928,
      "learning_rate": 2.288490184304329e-05,
      "loss": 0.0479,
      "num_input_tokens_seen": 14302272,
      "step": 21915
    },
    {
      "epoch": 11.48846960167715,
      "grad_norm": 2.4532840251922607,
      "learning_rate": 2.2873508775529308e-05,
      "loss": 0.0959,
      "num_input_tokens_seen": 14306432,
      "step": 21920
    },
    {
      "epoch": 11.491090146750524,
      "grad_norm": 2.0413365364074707,
      "learning_rate": 2.2862116152853034e-05,
      "loss": 0.1255,
      "num_input_tokens_seen": 14311680,
      "step": 21925
    },
    {
      "epoch": 11.4937106918239,
      "grad_norm": 2.2933287620544434,
      "learning_rate": 2.285072397739767e-05,
      "loss": 0.078,
      "num_input_tokens_seen": 14314208,
      "step": 21930
    },
    {
      "epoch": 11.496331236897275,
      "grad_norm": 2.1772773265838623,
      "learning_rate": 2.283933225154634e-05,
      "loss": 0.125,
      "num_input_tokens_seen": 14317696,
      "step": 21935
    },
    {
      "epoch": 11.49895178197065,
      "grad_norm": 4.621442794799805,
      "learning_rate": 2.2827940977682045e-05,
      "loss": 0.0758,
      "num_input_tokens_seen": 14320864,
      "step": 21940
    },
    {
      "epoch": 11.501572327044025,
      "grad_norm": 1.4697867631912231,
      "learning_rate": 2.2816550158187717e-05,
      "loss": 0.0643,
      "num_input_tokens_seen": 14323520,
      "step": 21945
    },
    {
      "epoch": 11.5041928721174,
      "grad_norm": 2.959469795227051,
      "learning_rate": 2.2805159795446182e-05,
      "loss": 0.0928,
      "num_input_tokens_seen": 14326400,
      "step": 21950
    },
    {
      "epoch": 11.506813417190775,
      "grad_norm": 2.7142691612243652,
      "learning_rate": 2.2793769891840184e-05,
      "loss": 0.0467,
      "num_input_tokens_seen": 14328992,
      "step": 21955
    },
    {
      "epoch": 11.50943396226415,
      "grad_norm": 6.735982894897461,
      "learning_rate": 2.278238044975236e-05,
      "loss": 0.0676,
      "num_input_tokens_seen": 14331584,
      "step": 21960
    },
    {
      "epoch": 11.512054507337526,
      "grad_norm": 0.6584152579307556,
      "learning_rate": 2.2770991471565236e-05,
      "loss": 0.071,
      "num_input_tokens_seen": 14335360,
      "step": 21965
    },
    {
      "epoch": 11.514675052410901,
      "grad_norm": 2.4622833728790283,
      "learning_rate": 2.2759602959661265e-05,
      "loss": 0.0941,
      "num_input_tokens_seen": 14339360,
      "step": 21970
    },
    {
      "epoch": 11.517295597484276,
      "grad_norm": 2.7136073112487793,
      "learning_rate": 2.2748214916422796e-05,
      "loss": 0.082,
      "num_input_tokens_seen": 14342112,
      "step": 21975
    },
    {
      "epoch": 11.519916142557651,
      "grad_norm": 1.6037120819091797,
      "learning_rate": 2.2736827344232082e-05,
      "loss": 0.0393,
      "num_input_tokens_seen": 14344960,
      "step": 21980
    },
    {
      "epoch": 11.522536687631026,
      "grad_norm": 6.290964603424072,
      "learning_rate": 2.272544024547126e-05,
      "loss": 0.098,
      "num_input_tokens_seen": 14348384,
      "step": 21985
    },
    {
      "epoch": 11.525157232704402,
      "grad_norm": 9.302305221557617,
      "learning_rate": 2.27140536225224e-05,
      "loss": 0.1309,
      "num_input_tokens_seen": 14351616,
      "step": 21990
    },
    {
      "epoch": 11.527777777777779,
      "grad_norm": 1.0041412115097046,
      "learning_rate": 2.270266747776744e-05,
      "loss": 0.1088,
      "num_input_tokens_seen": 14354848,
      "step": 21995
    },
    {
      "epoch": 11.530398322851154,
      "grad_norm": 1.5027570724487305,
      "learning_rate": 2.2691281813588226e-05,
      "loss": 0.0688,
      "num_input_tokens_seen": 14358496,
      "step": 22000
    },
    {
      "epoch": 11.533018867924529,
      "grad_norm": 4.184244155883789,
      "learning_rate": 2.267989663236652e-05,
      "loss": 0.1018,
      "num_input_tokens_seen": 14361728,
      "step": 22005
    },
    {
      "epoch": 11.535639412997904,
      "grad_norm": 1.8998123407363892,
      "learning_rate": 2.2668511936483975e-05,
      "loss": 0.046,
      "num_input_tokens_seen": 14365248,
      "step": 22010
    },
    {
      "epoch": 11.53825995807128,
      "grad_norm": 1.2217397689819336,
      "learning_rate": 2.2657127728322124e-05,
      "loss": 0.0911,
      "num_input_tokens_seen": 14368416,
      "step": 22015
    },
    {
      "epoch": 11.540880503144654,
      "grad_norm": 1.2637536525726318,
      "learning_rate": 2.264574401026244e-05,
      "loss": 0.1154,
      "num_input_tokens_seen": 14371424,
      "step": 22020
    },
    {
      "epoch": 11.54350104821803,
      "grad_norm": 2.6253015995025635,
      "learning_rate": 2.2634360784686242e-05,
      "loss": 0.0808,
      "num_input_tokens_seen": 14374592,
      "step": 22025
    },
    {
      "epoch": 11.546121593291405,
      "grad_norm": 2.6040868759155273,
      "learning_rate": 2.2622978053974776e-05,
      "loss": 0.0422,
      "num_input_tokens_seen": 14377504,
      "step": 22030
    },
    {
      "epoch": 11.54874213836478,
      "grad_norm": 2.672168731689453,
      "learning_rate": 2.2611595820509186e-05,
      "loss": 0.1014,
      "num_input_tokens_seen": 14380960,
      "step": 22035
    },
    {
      "epoch": 11.551362683438155,
      "grad_norm": 2.20640230178833,
      "learning_rate": 2.2600214086670494e-05,
      "loss": 0.0756,
      "num_input_tokens_seen": 14384608,
      "step": 22040
    },
    {
      "epoch": 11.55398322851153,
      "grad_norm": 1.7878389358520508,
      "learning_rate": 2.2588832854839644e-05,
      "loss": 0.0526,
      "num_input_tokens_seen": 14388160,
      "step": 22045
    },
    {
      "epoch": 11.556603773584905,
      "grad_norm": 2.1507585048675537,
      "learning_rate": 2.2577452127397463e-05,
      "loss": 0.0652,
      "num_input_tokens_seen": 14391296,
      "step": 22050
    },
    {
      "epoch": 11.55922431865828,
      "grad_norm": 0.9244423508644104,
      "learning_rate": 2.256607190672464e-05,
      "loss": 0.0385,
      "num_input_tokens_seen": 14394752,
      "step": 22055
    },
    {
      "epoch": 11.561844863731656,
      "grad_norm": 1.268524169921875,
      "learning_rate": 2.2554692195201817e-05,
      "loss": 0.0839,
      "num_input_tokens_seen": 14397376,
      "step": 22060
    },
    {
      "epoch": 11.564465408805031,
      "grad_norm": 2.5803933143615723,
      "learning_rate": 2.254331299520948e-05,
      "loss": 0.0762,
      "num_input_tokens_seen": 14399872,
      "step": 22065
    },
    {
      "epoch": 11.567085953878406,
      "grad_norm": 2.050302267074585,
      "learning_rate": 2.253193430912805e-05,
      "loss": 0.09,
      "num_input_tokens_seen": 14403552,
      "step": 22070
    },
    {
      "epoch": 11.569706498951781,
      "grad_norm": 2.4875240325927734,
      "learning_rate": 2.2520556139337803e-05,
      "loss": 0.0677,
      "num_input_tokens_seen": 14407552,
      "step": 22075
    },
    {
      "epoch": 11.572327044025156,
      "grad_norm": 2.940589189529419,
      "learning_rate": 2.250917848821892e-05,
      "loss": 0.0792,
      "num_input_tokens_seen": 14410912,
      "step": 22080
    },
    {
      "epoch": 11.574947589098532,
      "grad_norm": 8.108137130737305,
      "learning_rate": 2.2497801358151472e-05,
      "loss": 0.0684,
      "num_input_tokens_seen": 14413440,
      "step": 22085
    },
    {
      "epoch": 11.577568134171909,
      "grad_norm": 2.221998929977417,
      "learning_rate": 2.248642475151545e-05,
      "loss": 0.0736,
      "num_input_tokens_seen": 14417568,
      "step": 22090
    },
    {
      "epoch": 11.580188679245284,
      "grad_norm": 1.487188458442688,
      "learning_rate": 2.2475048670690678e-05,
      "loss": 0.044,
      "num_input_tokens_seen": 14420736,
      "step": 22095
    },
    {
      "epoch": 11.582809224318659,
      "grad_norm": 2.5005853176116943,
      "learning_rate": 2.2463673118056927e-05,
      "loss": 0.0685,
      "num_input_tokens_seen": 14424224,
      "step": 22100
    },
    {
      "epoch": 11.585429769392034,
      "grad_norm": 3.0310587882995605,
      "learning_rate": 2.245229809599383e-05,
      "loss": 0.0767,
      "num_input_tokens_seen": 14427264,
      "step": 22105
    },
    {
      "epoch": 11.58805031446541,
      "grad_norm": 2.2660350799560547,
      "learning_rate": 2.2440923606880896e-05,
      "loss": 0.0817,
      "num_input_tokens_seen": 14429504,
      "step": 22110
    },
    {
      "epoch": 11.590670859538784,
      "grad_norm": 2.122863531112671,
      "learning_rate": 2.242954965309755e-05,
      "loss": 0.0893,
      "num_input_tokens_seen": 14432672,
      "step": 22115
    },
    {
      "epoch": 11.59329140461216,
      "grad_norm": 2.042201280593872,
      "learning_rate": 2.2418176237023088e-05,
      "loss": 0.0449,
      "num_input_tokens_seen": 14435072,
      "step": 22120
    },
    {
      "epoch": 11.595911949685535,
      "grad_norm": 2.657618522644043,
      "learning_rate": 2.2406803361036706e-05,
      "loss": 0.0526,
      "num_input_tokens_seen": 14439424,
      "step": 22125
    },
    {
      "epoch": 11.59853249475891,
      "grad_norm": 1.0352575778961182,
      "learning_rate": 2.2395431027517475e-05,
      "loss": 0.0488,
      "num_input_tokens_seen": 14441888,
      "step": 22130
    },
    {
      "epoch": 11.601153039832285,
      "grad_norm": 1.3203833103179932,
      "learning_rate": 2.2384059238844364e-05,
      "loss": 0.0953,
      "num_input_tokens_seen": 14444896,
      "step": 22135
    },
    {
      "epoch": 11.60377358490566,
      "grad_norm": 1.1906989812850952,
      "learning_rate": 2.2372687997396215e-05,
      "loss": 0.0746,
      "num_input_tokens_seen": 14447904,
      "step": 22140
    },
    {
      "epoch": 11.606394129979035,
      "grad_norm": 7.720602989196777,
      "learning_rate": 2.2361317305551756e-05,
      "loss": 0.0752,
      "num_input_tokens_seen": 14450624,
      "step": 22145
    },
    {
      "epoch": 11.60901467505241,
      "grad_norm": 6.326379299163818,
      "learning_rate": 2.2349947165689615e-05,
      "loss": 0.0687,
      "num_input_tokens_seen": 14454240,
      "step": 22150
    },
    {
      "epoch": 11.611635220125786,
      "grad_norm": 4.040680408477783,
      "learning_rate": 2.2338577580188296e-05,
      "loss": 0.0599,
      "num_input_tokens_seen": 14457088,
      "step": 22155
    },
    {
      "epoch": 11.614255765199161,
      "grad_norm": 3.2019431591033936,
      "learning_rate": 2.2327208551426188e-05,
      "loss": 0.0542,
      "num_input_tokens_seen": 14460608,
      "step": 22160
    },
    {
      "epoch": 11.616876310272536,
      "grad_norm": 1.0873104333877563,
      "learning_rate": 2.2315840081781546e-05,
      "loss": 0.0987,
      "num_input_tokens_seen": 14464064,
      "step": 22165
    },
    {
      "epoch": 11.619496855345911,
      "grad_norm": 3.0455074310302734,
      "learning_rate": 2.230447217363255e-05,
      "loss": 0.0895,
      "num_input_tokens_seen": 14467584,
      "step": 22170
    },
    {
      "epoch": 11.622117400419286,
      "grad_norm": 2.452845811843872,
      "learning_rate": 2.2293104829357212e-05,
      "loss": 0.0597,
      "num_input_tokens_seen": 14471328,
      "step": 22175
    },
    {
      "epoch": 11.624737945492662,
      "grad_norm": 1.6111645698547363,
      "learning_rate": 2.2281738051333467e-05,
      "loss": 0.0664,
      "num_input_tokens_seen": 14475072,
      "step": 22180
    },
    {
      "epoch": 11.627358490566039,
      "grad_norm": 1.9536821842193604,
      "learning_rate": 2.2270371841939105e-05,
      "loss": 0.0393,
      "num_input_tokens_seen": 14477984,
      "step": 22185
    },
    {
      "epoch": 11.629979035639414,
      "grad_norm": 0.9773322343826294,
      "learning_rate": 2.2259006203551824e-05,
      "loss": 0.0502,
      "num_input_tokens_seen": 14481984,
      "step": 22190
    },
    {
      "epoch": 11.632599580712789,
      "grad_norm": 2.957465410232544,
      "learning_rate": 2.2247641138549168e-05,
      "loss": 0.0758,
      "num_input_tokens_seen": 14485760,
      "step": 22195
    },
    {
      "epoch": 11.635220125786164,
      "grad_norm": 2.556544065475464,
      "learning_rate": 2.2236276649308583e-05,
      "loss": 0.0842,
      "num_input_tokens_seen": 14489664,
      "step": 22200
    },
    {
      "epoch": 11.63784067085954,
      "grad_norm": 1.2761470079421997,
      "learning_rate": 2.2224912738207395e-05,
      "loss": 0.0423,
      "num_input_tokens_seen": 14493056,
      "step": 22205
    },
    {
      "epoch": 11.640461215932914,
      "grad_norm": 3.1488490104675293,
      "learning_rate": 2.2213549407622796e-05,
      "loss": 0.0442,
      "num_input_tokens_seen": 14495680,
      "step": 22210
    },
    {
      "epoch": 11.64308176100629,
      "grad_norm": 2.3937957286834717,
      "learning_rate": 2.220218665993188e-05,
      "loss": 0.0543,
      "num_input_tokens_seen": 14498592,
      "step": 22215
    },
    {
      "epoch": 11.645702306079665,
      "grad_norm": 0.5745162963867188,
      "learning_rate": 2.21908244975116e-05,
      "loss": 0.0611,
      "num_input_tokens_seen": 14501888,
      "step": 22220
    },
    {
      "epoch": 11.64832285115304,
      "grad_norm": 1.9362560510635376,
      "learning_rate": 2.2179462922738776e-05,
      "loss": 0.0539,
      "num_input_tokens_seen": 14505088,
      "step": 22225
    },
    {
      "epoch": 11.650943396226415,
      "grad_norm": 1.8023693561553955,
      "learning_rate": 2.2168101937990135e-05,
      "loss": 0.0982,
      "num_input_tokens_seen": 14507744,
      "step": 22230
    },
    {
      "epoch": 11.65356394129979,
      "grad_norm": 6.99685001373291,
      "learning_rate": 2.2156741545642253e-05,
      "loss": 0.0553,
      "num_input_tokens_seen": 14510560,
      "step": 22235
    },
    {
      "epoch": 11.656184486373165,
      "grad_norm": 2.595533847808838,
      "learning_rate": 2.2145381748071606e-05,
      "loss": 0.0775,
      "num_input_tokens_seen": 14514432,
      "step": 22240
    },
    {
      "epoch": 11.65880503144654,
      "grad_norm": 3.297585964202881,
      "learning_rate": 2.2134022547654538e-05,
      "loss": 0.092,
      "num_input_tokens_seen": 14517664,
      "step": 22245
    },
    {
      "epoch": 11.661425576519916,
      "grad_norm": 3.4316246509552,
      "learning_rate": 2.212266394676725e-05,
      "loss": 0.1105,
      "num_input_tokens_seen": 14520704,
      "step": 22250
    },
    {
      "epoch": 11.664046121593291,
      "grad_norm": 1.334931492805481,
      "learning_rate": 2.2111305947785836e-05,
      "loss": 0.055,
      "num_input_tokens_seen": 14525152,
      "step": 22255
    },
    {
      "epoch": 11.666666666666666,
      "grad_norm": 2.2153968811035156,
      "learning_rate": 2.2099948553086263e-05,
      "loss": 0.0659,
      "num_input_tokens_seen": 14529824,
      "step": 22260
    },
    {
      "epoch": 11.669287211740041,
      "grad_norm": 3.472399950027466,
      "learning_rate": 2.2088591765044365e-05,
      "loss": 0.0661,
      "num_input_tokens_seen": 14532416,
      "step": 22265
    },
    {
      "epoch": 11.671907756813416,
      "grad_norm": 1.8555833101272583,
      "learning_rate": 2.207723558603586e-05,
      "loss": 0.1005,
      "num_input_tokens_seen": 14535776,
      "step": 22270
    },
    {
      "epoch": 11.674528301886792,
      "grad_norm": 2.4784164428710938,
      "learning_rate": 2.206588001843633e-05,
      "loss": 0.1247,
      "num_input_tokens_seen": 14538624,
      "step": 22275
    },
    {
      "epoch": 11.677148846960169,
      "grad_norm": 1.4975731372833252,
      "learning_rate": 2.2054525064621216e-05,
      "loss": 0.0491,
      "num_input_tokens_seen": 14542112,
      "step": 22280
    },
    {
      "epoch": 11.679769392033544,
      "grad_norm": 4.694295406341553,
      "learning_rate": 2.2043170726965858e-05,
      "loss": 0.0644,
      "num_input_tokens_seen": 14545024,
      "step": 22285
    },
    {
      "epoch": 11.682389937106919,
      "grad_norm": 3.277996301651001,
      "learning_rate": 2.2031817007845447e-05,
      "loss": 0.0872,
      "num_input_tokens_seen": 14549440,
      "step": 22290
    },
    {
      "epoch": 11.685010482180294,
      "grad_norm": 1.6395775079727173,
      "learning_rate": 2.202046390963506e-05,
      "loss": 0.07,
      "num_input_tokens_seen": 14552224,
      "step": 22295
    },
    {
      "epoch": 11.68763102725367,
      "grad_norm": 2.0107972621917725,
      "learning_rate": 2.2009111434709627e-05,
      "loss": 0.0513,
      "num_input_tokens_seen": 14555040,
      "step": 22300
    },
    {
      "epoch": 11.690251572327044,
      "grad_norm": 2.506927967071533,
      "learning_rate": 2.1997759585443968e-05,
      "loss": 0.0655,
      "num_input_tokens_seen": 14558016,
      "step": 22305
    },
    {
      "epoch": 11.69287211740042,
      "grad_norm": 1.7746706008911133,
      "learning_rate": 2.198640836421275e-05,
      "loss": 0.07,
      "num_input_tokens_seen": 14560640,
      "step": 22310
    },
    {
      "epoch": 11.695492662473795,
      "grad_norm": 2.539863109588623,
      "learning_rate": 2.1975057773390515e-05,
      "loss": 0.0574,
      "num_input_tokens_seen": 14563264,
      "step": 22315
    },
    {
      "epoch": 11.69811320754717,
      "grad_norm": 4.220993518829346,
      "learning_rate": 2.1963707815351683e-05,
      "loss": 0.0598,
      "num_input_tokens_seen": 14566432,
      "step": 22320
    },
    {
      "epoch": 11.700733752620545,
      "grad_norm": 2.3315961360931396,
      "learning_rate": 2.1952358492470544e-05,
      "loss": 0.0759,
      "num_input_tokens_seen": 14570176,
      "step": 22325
    },
    {
      "epoch": 11.70335429769392,
      "grad_norm": 2.8169236183166504,
      "learning_rate": 2.194100980712124e-05,
      "loss": 0.1537,
      "num_input_tokens_seen": 14573536,
      "step": 22330
    },
    {
      "epoch": 11.705974842767295,
      "grad_norm": 2.0307869911193848,
      "learning_rate": 2.1929661761677774e-05,
      "loss": 0.0637,
      "num_input_tokens_seen": 14576672,
      "step": 22335
    },
    {
      "epoch": 11.70859538784067,
      "grad_norm": 1.4345394372940063,
      "learning_rate": 2.191831435851405e-05,
      "loss": 0.0554,
      "num_input_tokens_seen": 14579584,
      "step": 22340
    },
    {
      "epoch": 11.711215932914046,
      "grad_norm": 1.5411914587020874,
      "learning_rate": 2.1906967600003793e-05,
      "loss": 0.0753,
      "num_input_tokens_seen": 14582528,
      "step": 22345
    },
    {
      "epoch": 11.713836477987421,
      "grad_norm": 2.444655179977417,
      "learning_rate": 2.1895621488520636e-05,
      "loss": 0.0929,
      "num_input_tokens_seen": 14586464,
      "step": 22350
    },
    {
      "epoch": 11.716457023060796,
      "grad_norm": 1.4407517910003662,
      "learning_rate": 2.1884276026438044e-05,
      "loss": 0.0725,
      "num_input_tokens_seen": 14589440,
      "step": 22355
    },
    {
      "epoch": 11.719077568134171,
      "grad_norm": 2.6928465366363525,
      "learning_rate": 2.187293121612937e-05,
      "loss": 0.0797,
      "num_input_tokens_seen": 14593280,
      "step": 22360
    },
    {
      "epoch": 11.721698113207546,
      "grad_norm": 1.9232475757598877,
      "learning_rate": 2.1861587059967804e-05,
      "loss": 0.0615,
      "num_input_tokens_seen": 14596800,
      "step": 22365
    },
    {
      "epoch": 11.724318658280922,
      "grad_norm": 4.054222583770752,
      "learning_rate": 2.1850243560326414e-05,
      "loss": 0.1046,
      "num_input_tokens_seen": 14599552,
      "step": 22370
    },
    {
      "epoch": 11.726939203354299,
      "grad_norm": 4.9322404861450195,
      "learning_rate": 2.1838900719578143e-05,
      "loss": 0.0778,
      "num_input_tokens_seen": 14601952,
      "step": 22375
    },
    {
      "epoch": 11.729559748427674,
      "grad_norm": 2.5950868129730225,
      "learning_rate": 2.1827558540095772e-05,
      "loss": 0.066,
      "num_input_tokens_seen": 14606336,
      "step": 22380
    },
    {
      "epoch": 11.732180293501049,
      "grad_norm": 2.72826886177063,
      "learning_rate": 2.1816217024251966e-05,
      "loss": 0.0559,
      "num_input_tokens_seen": 14608864,
      "step": 22385
    },
    {
      "epoch": 11.734800838574424,
      "grad_norm": 1.9621756076812744,
      "learning_rate": 2.1804876174419242e-05,
      "loss": 0.044,
      "num_input_tokens_seen": 14611488,
      "step": 22390
    },
    {
      "epoch": 11.7374213836478,
      "grad_norm": 3.231922149658203,
      "learning_rate": 2.179353599296996e-05,
      "loss": 0.0605,
      "num_input_tokens_seen": 14615296,
      "step": 22395
    },
    {
      "epoch": 11.740041928721174,
      "grad_norm": 2.502261161804199,
      "learning_rate": 2.1782196482276368e-05,
      "loss": 0.0767,
      "num_input_tokens_seen": 14620096,
      "step": 22400
    },
    {
      "epoch": 11.74266247379455,
      "grad_norm": 2.8017117977142334,
      "learning_rate": 2.1770857644710567e-05,
      "loss": 0.1092,
      "num_input_tokens_seen": 14622912,
      "step": 22405
    },
    {
      "epoch": 11.745283018867925,
      "grad_norm": 1.293246865272522,
      "learning_rate": 2.1759519482644495e-05,
      "loss": 0.1066,
      "num_input_tokens_seen": 14625568,
      "step": 22410
    },
    {
      "epoch": 11.7479035639413,
      "grad_norm": 9.205803871154785,
      "learning_rate": 2.174818199844999e-05,
      "loss": 0.0755,
      "num_input_tokens_seen": 14628576,
      "step": 22415
    },
    {
      "epoch": 11.750524109014675,
      "grad_norm": 1.8530144691467285,
      "learning_rate": 2.173684519449872e-05,
      "loss": 0.0601,
      "num_input_tokens_seen": 14632096,
      "step": 22420
    },
    {
      "epoch": 11.75314465408805,
      "grad_norm": 3.992825746536255,
      "learning_rate": 2.1725509073162188e-05,
      "loss": 0.0381,
      "num_input_tokens_seen": 14636192,
      "step": 22425
    },
    {
      "epoch": 11.755765199161425,
      "grad_norm": 4.692019939422607,
      "learning_rate": 2.171417363681181e-05,
      "loss": 0.0832,
      "num_input_tokens_seen": 14638944,
      "step": 22430
    },
    {
      "epoch": 11.7583857442348,
      "grad_norm": 2.572242498397827,
      "learning_rate": 2.1702838887818816e-05,
      "loss": 0.081,
      "num_input_tokens_seen": 14641088,
      "step": 22435
    },
    {
      "epoch": 11.761006289308176,
      "grad_norm": 2.821650743484497,
      "learning_rate": 2.1691504828554315e-05,
      "loss": 0.0525,
      "num_input_tokens_seen": 14644992,
      "step": 22440
    },
    {
      "epoch": 11.76362683438155,
      "grad_norm": 3.827516555786133,
      "learning_rate": 2.1680171461389265e-05,
      "loss": 0.0944,
      "num_input_tokens_seen": 14648000,
      "step": 22445
    },
    {
      "epoch": 11.766247379454926,
      "grad_norm": 3.764730453491211,
      "learning_rate": 2.166883878869446e-05,
      "loss": 0.1043,
      "num_input_tokens_seen": 14651168,
      "step": 22450
    },
    {
      "epoch": 11.768867924528301,
      "grad_norm": 2.1954078674316406,
      "learning_rate": 2.1657506812840582e-05,
      "loss": 0.097,
      "num_input_tokens_seen": 14654528,
      "step": 22455
    },
    {
      "epoch": 11.771488469601676,
      "grad_norm": 3.0464727878570557,
      "learning_rate": 2.164617553619814e-05,
      "loss": 0.075,
      "num_input_tokens_seen": 14657088,
      "step": 22460
    },
    {
      "epoch": 11.774109014675052,
      "grad_norm": 2.1172354221343994,
      "learning_rate": 2.1634844961137517e-05,
      "loss": 0.0701,
      "num_input_tokens_seen": 14660352,
      "step": 22465
    },
    {
      "epoch": 11.776729559748428,
      "grad_norm": 2.027362108230591,
      "learning_rate": 2.162351509002893e-05,
      "loss": 0.1216,
      "num_input_tokens_seen": 14664288,
      "step": 22470
    },
    {
      "epoch": 11.779350104821804,
      "grad_norm": 4.709740161895752,
      "learning_rate": 2.161218592524248e-05,
      "loss": 0.2658,
      "num_input_tokens_seen": 14668000,
      "step": 22475
    },
    {
      "epoch": 11.781970649895179,
      "grad_norm": 4.074515342712402,
      "learning_rate": 2.160085746914807e-05,
      "loss": 0.0394,
      "num_input_tokens_seen": 14671328,
      "step": 22480
    },
    {
      "epoch": 11.784591194968554,
      "grad_norm": 4.023965358734131,
      "learning_rate": 2.1589529724115503e-05,
      "loss": 0.0746,
      "num_input_tokens_seen": 14674272,
      "step": 22485
    },
    {
      "epoch": 11.78721174004193,
      "grad_norm": 0.957377016544342,
      "learning_rate": 2.1578202692514407e-05,
      "loss": 0.0628,
      "num_input_tokens_seen": 14676896,
      "step": 22490
    },
    {
      "epoch": 11.789832285115304,
      "grad_norm": 6.914583683013916,
      "learning_rate": 2.1566876376714274e-05,
      "loss": 0.108,
      "num_input_tokens_seen": 14679840,
      "step": 22495
    },
    {
      "epoch": 11.79245283018868,
      "grad_norm": 1.7708346843719482,
      "learning_rate": 2.155555077908443e-05,
      "loss": 0.0925,
      "num_input_tokens_seen": 14683456,
      "step": 22500
    },
    {
      "epoch": 11.795073375262055,
      "grad_norm": 0.5207756161689758,
      "learning_rate": 2.1544225901994077e-05,
      "loss": 0.0942,
      "num_input_tokens_seen": 14686112,
      "step": 22505
    },
    {
      "epoch": 11.79769392033543,
      "grad_norm": 1.3632197380065918,
      "learning_rate": 2.1532901747812234e-05,
      "loss": 0.0271,
      "num_input_tokens_seen": 14689024,
      "step": 22510
    },
    {
      "epoch": 11.800314465408805,
      "grad_norm": 3.762749671936035,
      "learning_rate": 2.1521578318907787e-05,
      "loss": 0.0662,
      "num_input_tokens_seen": 14691712,
      "step": 22515
    },
    {
      "epoch": 11.80293501048218,
      "grad_norm": 2.918569803237915,
      "learning_rate": 2.1510255617649476e-05,
      "loss": 0.0899,
      "num_input_tokens_seen": 14695104,
      "step": 22520
    },
    {
      "epoch": 11.805555555555555,
      "grad_norm": 2.4222981929779053,
      "learning_rate": 2.149893364640587e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 14697952,
      "step": 22525
    },
    {
      "epoch": 11.80817610062893,
      "grad_norm": 1.6734334230422974,
      "learning_rate": 2.1487612407545417e-05,
      "loss": 0.0513,
      "num_input_tokens_seen": 14700704,
      "step": 22530
    },
    {
      "epoch": 11.810796645702306,
      "grad_norm": 5.196577072143555,
      "learning_rate": 2.147629190343637e-05,
      "loss": 0.0902,
      "num_input_tokens_seen": 14703488,
      "step": 22535
    },
    {
      "epoch": 11.81341719077568,
      "grad_norm": 9.2919282913208,
      "learning_rate": 2.1464972136446852e-05,
      "loss": 0.1014,
      "num_input_tokens_seen": 14706752,
      "step": 22540
    },
    {
      "epoch": 11.816037735849056,
      "grad_norm": 4.431067943572998,
      "learning_rate": 2.1453653108944838e-05,
      "loss": 0.074,
      "num_input_tokens_seen": 14710176,
      "step": 22545
    },
    {
      "epoch": 11.818658280922431,
      "grad_norm": 2.0898895263671875,
      "learning_rate": 2.1442334823298128e-05,
      "loss": 0.1313,
      "num_input_tokens_seen": 14713376,
      "step": 22550
    },
    {
      "epoch": 11.821278825995806,
      "grad_norm": 2.7057111263275146,
      "learning_rate": 2.1431017281874386e-05,
      "loss": 0.1039,
      "num_input_tokens_seen": 14716864,
      "step": 22555
    },
    {
      "epoch": 11.823899371069182,
      "grad_norm": 6.227921962738037,
      "learning_rate": 2.1419700487041123e-05,
      "loss": 0.0637,
      "num_input_tokens_seen": 14719776,
      "step": 22560
    },
    {
      "epoch": 11.826519916142558,
      "grad_norm": 2.092376708984375,
      "learning_rate": 2.140838444116566e-05,
      "loss": 0.1343,
      "num_input_tokens_seen": 14724512,
      "step": 22565
    },
    {
      "epoch": 11.829140461215934,
      "grad_norm": 4.315322399139404,
      "learning_rate": 2.13970691466152e-05,
      "loss": 0.0649,
      "num_input_tokens_seen": 14727168,
      "step": 22570
    },
    {
      "epoch": 11.831761006289309,
      "grad_norm": 1.388301134109497,
      "learning_rate": 2.1385754605756773e-05,
      "loss": 0.0599,
      "num_input_tokens_seen": 14730112,
      "step": 22575
    },
    {
      "epoch": 11.834381551362684,
      "grad_norm": 4.384705066680908,
      "learning_rate": 2.137444082095724e-05,
      "loss": 0.0336,
      "num_input_tokens_seen": 14733312,
      "step": 22580
    },
    {
      "epoch": 11.83700209643606,
      "grad_norm": 3.542975425720215,
      "learning_rate": 2.1363127794583335e-05,
      "loss": 0.1002,
      "num_input_tokens_seen": 14737440,
      "step": 22585
    },
    {
      "epoch": 11.839622641509434,
      "grad_norm": 3.4859328269958496,
      "learning_rate": 2.1351815529001605e-05,
      "loss": 0.0548,
      "num_input_tokens_seen": 14740672,
      "step": 22590
    },
    {
      "epoch": 11.84224318658281,
      "grad_norm": 1.9070159196853638,
      "learning_rate": 2.1340504026578438e-05,
      "loss": 0.0678,
      "num_input_tokens_seen": 14743552,
      "step": 22595
    },
    {
      "epoch": 11.844863731656185,
      "grad_norm": 8.430022239685059,
      "learning_rate": 2.1329193289680084e-05,
      "loss": 0.1058,
      "num_input_tokens_seen": 14746816,
      "step": 22600
    },
    {
      "epoch": 11.84748427672956,
      "grad_norm": 2.8647329807281494,
      "learning_rate": 2.1317883320672608e-05,
      "loss": 0.0868,
      "num_input_tokens_seen": 14749696,
      "step": 22605
    },
    {
      "epoch": 11.850104821802935,
      "grad_norm": 2.8278214931488037,
      "learning_rate": 2.1306574121921942e-05,
      "loss": 0.0949,
      "num_input_tokens_seen": 14753376,
      "step": 22610
    },
    {
      "epoch": 11.85272536687631,
      "grad_norm": 2.7608182430267334,
      "learning_rate": 2.1295265695793836e-05,
      "loss": 0.1215,
      "num_input_tokens_seen": 14756576,
      "step": 22615
    },
    {
      "epoch": 11.855345911949685,
      "grad_norm": 1.3389785289764404,
      "learning_rate": 2.128395804465387e-05,
      "loss": 0.0596,
      "num_input_tokens_seen": 14759552,
      "step": 22620
    },
    {
      "epoch": 11.85796645702306,
      "grad_norm": 1.3689894676208496,
      "learning_rate": 2.1272651170867487e-05,
      "loss": 0.0864,
      "num_input_tokens_seen": 14763424,
      "step": 22625
    },
    {
      "epoch": 11.860587002096436,
      "grad_norm": 8.703678131103516,
      "learning_rate": 2.1261345076799955e-05,
      "loss": 0.0807,
      "num_input_tokens_seen": 14767392,
      "step": 22630
    },
    {
      "epoch": 11.86320754716981,
      "grad_norm": 3.5909736156463623,
      "learning_rate": 2.125003976481638e-05,
      "loss": 0.1166,
      "num_input_tokens_seen": 14770272,
      "step": 22635
    },
    {
      "epoch": 11.865828092243186,
      "grad_norm": 3.452565908432007,
      "learning_rate": 2.123873523728171e-05,
      "loss": 0.0627,
      "num_input_tokens_seen": 14773664,
      "step": 22640
    },
    {
      "epoch": 11.868448637316561,
      "grad_norm": 4.038092613220215,
      "learning_rate": 2.122743149656072e-05,
      "loss": 0.0334,
      "num_input_tokens_seen": 14776640,
      "step": 22645
    },
    {
      "epoch": 11.871069182389936,
      "grad_norm": 1.7359052896499634,
      "learning_rate": 2.1216128545018012e-05,
      "loss": 0.0873,
      "num_input_tokens_seen": 14779584,
      "step": 22650
    },
    {
      "epoch": 11.873689727463312,
      "grad_norm": 1.945333480834961,
      "learning_rate": 2.1204826385018052e-05,
      "loss": 0.1279,
      "num_input_tokens_seen": 14782784,
      "step": 22655
    },
    {
      "epoch": 11.876310272536688,
      "grad_norm": 2.162353992462158,
      "learning_rate": 2.1193525018925106e-05,
      "loss": 0.1427,
      "num_input_tokens_seen": 14787040,
      "step": 22660
    },
    {
      "epoch": 11.878930817610064,
      "grad_norm": 1.3217856884002686,
      "learning_rate": 2.1182224449103304e-05,
      "loss": 0.0708,
      "num_input_tokens_seen": 14790560,
      "step": 22665
    },
    {
      "epoch": 11.881551362683439,
      "grad_norm": 2.678680181503296,
      "learning_rate": 2.117092467791659e-05,
      "loss": 0.1138,
      "num_input_tokens_seen": 14794976,
      "step": 22670
    },
    {
      "epoch": 11.884171907756814,
      "grad_norm": 1.9690847396850586,
      "learning_rate": 2.115962570772876e-05,
      "loss": 0.044,
      "num_input_tokens_seen": 14798368,
      "step": 22675
    },
    {
      "epoch": 11.88679245283019,
      "grad_norm": 1.9580293893814087,
      "learning_rate": 2.1148327540903408e-05,
      "loss": 0.0938,
      "num_input_tokens_seen": 14801536,
      "step": 22680
    },
    {
      "epoch": 11.889412997903564,
      "grad_norm": 3.538437604904175,
      "learning_rate": 2.113703017980399e-05,
      "loss": 0.0424,
      "num_input_tokens_seen": 14804544,
      "step": 22685
    },
    {
      "epoch": 11.89203354297694,
      "grad_norm": 2.4033350944519043,
      "learning_rate": 2.1125733626793793e-05,
      "loss": 0.0594,
      "num_input_tokens_seen": 14807328,
      "step": 22690
    },
    {
      "epoch": 11.894654088050315,
      "grad_norm": 3.8534016609191895,
      "learning_rate": 2.1114437884235915e-05,
      "loss": 0.0792,
      "num_input_tokens_seen": 14810464,
      "step": 22695
    },
    {
      "epoch": 11.89727463312369,
      "grad_norm": 2.1702539920806885,
      "learning_rate": 2.1103142954493313e-05,
      "loss": 0.0725,
      "num_input_tokens_seen": 14814080,
      "step": 22700
    },
    {
      "epoch": 11.899895178197065,
      "grad_norm": 1.5538909435272217,
      "learning_rate": 2.1091848839928745e-05,
      "loss": 0.1176,
      "num_input_tokens_seen": 14818368,
      "step": 22705
    },
    {
      "epoch": 11.90251572327044,
      "grad_norm": 1.540670394897461,
      "learning_rate": 2.1080555542904805e-05,
      "loss": 0.0568,
      "num_input_tokens_seen": 14821344,
      "step": 22710
    },
    {
      "epoch": 11.905136268343815,
      "grad_norm": 1.7249053716659546,
      "learning_rate": 2.106926306578393e-05,
      "loss": 0.0582,
      "num_input_tokens_seen": 14824352,
      "step": 22715
    },
    {
      "epoch": 11.90775681341719,
      "grad_norm": 0.8635146617889404,
      "learning_rate": 2.1057971410928383e-05,
      "loss": 0.0855,
      "num_input_tokens_seen": 14828704,
      "step": 22720
    },
    {
      "epoch": 11.910377358490566,
      "grad_norm": 3.1507883071899414,
      "learning_rate": 2.104668058070024e-05,
      "loss": 0.1021,
      "num_input_tokens_seen": 14831456,
      "step": 22725
    },
    {
      "epoch": 11.91299790356394,
      "grad_norm": 2.5505247116088867,
      "learning_rate": 2.1035390577461427e-05,
      "loss": 0.0662,
      "num_input_tokens_seen": 14833856,
      "step": 22730
    },
    {
      "epoch": 11.915618448637316,
      "grad_norm": 2.930844306945801,
      "learning_rate": 2.102410140357367e-05,
      "loss": 0.0589,
      "num_input_tokens_seen": 14837120,
      "step": 22735
    },
    {
      "epoch": 11.918238993710691,
      "grad_norm": 2.177499532699585,
      "learning_rate": 2.1012813061398534e-05,
      "loss": 0.0773,
      "num_input_tokens_seen": 14841056,
      "step": 22740
    },
    {
      "epoch": 11.920859538784066,
      "grad_norm": 0.9839150905609131,
      "learning_rate": 2.1001525553297422e-05,
      "loss": 0.1014,
      "num_input_tokens_seen": 14844064,
      "step": 22745
    },
    {
      "epoch": 11.923480083857442,
      "grad_norm": 5.130632400512695,
      "learning_rate": 2.099023888163154e-05,
      "loss": 0.0789,
      "num_input_tokens_seen": 14846656,
      "step": 22750
    },
    {
      "epoch": 11.926100628930818,
      "grad_norm": 3.3664116859436035,
      "learning_rate": 2.0978953048761947e-05,
      "loss": 0.0981,
      "num_input_tokens_seen": 14849248,
      "step": 22755
    },
    {
      "epoch": 11.928721174004194,
      "grad_norm": 1.029818058013916,
      "learning_rate": 2.0967668057049507e-05,
      "loss": 0.0789,
      "num_input_tokens_seen": 14852384,
      "step": 22760
    },
    {
      "epoch": 11.931341719077569,
      "grad_norm": 10.013434410095215,
      "learning_rate": 2.0956383908854895e-05,
      "loss": 0.0424,
      "num_input_tokens_seen": 14854720,
      "step": 22765
    },
    {
      "epoch": 11.933962264150944,
      "grad_norm": 3.0003011226654053,
      "learning_rate": 2.0945100606538644e-05,
      "loss": 0.0571,
      "num_input_tokens_seen": 14857696,
      "step": 22770
    },
    {
      "epoch": 11.93658280922432,
      "grad_norm": 3.242377758026123,
      "learning_rate": 2.0933818152461077e-05,
      "loss": 0.1074,
      "num_input_tokens_seen": 14861216,
      "step": 22775
    },
    {
      "epoch": 11.939203354297694,
      "grad_norm": 3.300795793533325,
      "learning_rate": 2.092253654898237e-05,
      "loss": 0.0861,
      "num_input_tokens_seen": 14864064,
      "step": 22780
    },
    {
      "epoch": 11.94182389937107,
      "grad_norm": 4.6036248207092285,
      "learning_rate": 2.0911255798462494e-05,
      "loss": 0.0908,
      "num_input_tokens_seen": 14866784,
      "step": 22785
    },
    {
      "epoch": 11.944444444444445,
      "grad_norm": 1.1363956928253174,
      "learning_rate": 2.0899975903261266e-05,
      "loss": 0.0374,
      "num_input_tokens_seen": 14869664,
      "step": 22790
    },
    {
      "epoch": 11.94706498951782,
      "grad_norm": 0.5255609750747681,
      "learning_rate": 2.0888696865738295e-05,
      "loss": 0.0919,
      "num_input_tokens_seen": 14872640,
      "step": 22795
    },
    {
      "epoch": 11.949685534591195,
      "grad_norm": 2.7078981399536133,
      "learning_rate": 2.087741868825304e-05,
      "loss": 0.1035,
      "num_input_tokens_seen": 14876512,
      "step": 22800
    },
    {
      "epoch": 11.95230607966457,
      "grad_norm": 2.5546109676361084,
      "learning_rate": 2.0866141373164758e-05,
      "loss": 0.0898,
      "num_input_tokens_seen": 14879776,
      "step": 22805
    },
    {
      "epoch": 11.954926624737945,
      "grad_norm": 1.8782544136047363,
      "learning_rate": 2.0854864922832544e-05,
      "loss": 0.0537,
      "num_input_tokens_seen": 14883552,
      "step": 22810
    },
    {
      "epoch": 11.95754716981132,
      "grad_norm": 3.8693082332611084,
      "learning_rate": 2.0843589339615304e-05,
      "loss": 0.0669,
      "num_input_tokens_seen": 14885984,
      "step": 22815
    },
    {
      "epoch": 11.960167714884696,
      "grad_norm": 3.238375425338745,
      "learning_rate": 2.0832314625871742e-05,
      "loss": 0.0931,
      "num_input_tokens_seen": 14888992,
      "step": 22820
    },
    {
      "epoch": 11.96278825995807,
      "grad_norm": 1.2434883117675781,
      "learning_rate": 2.0821040783960423e-05,
      "loss": 0.0596,
      "num_input_tokens_seen": 14891744,
      "step": 22825
    },
    {
      "epoch": 11.965408805031446,
      "grad_norm": 2.40242338180542,
      "learning_rate": 2.0809767816239686e-05,
      "loss": 0.1944,
      "num_input_tokens_seen": 14894880,
      "step": 22830
    },
    {
      "epoch": 11.968029350104821,
      "grad_norm": 2.607235908508301,
      "learning_rate": 2.0798495725067725e-05,
      "loss": 0.1102,
      "num_input_tokens_seen": 14897760,
      "step": 22835
    },
    {
      "epoch": 11.970649895178196,
      "grad_norm": 0.5919278860092163,
      "learning_rate": 2.078722451280252e-05,
      "loss": 0.0727,
      "num_input_tokens_seen": 14900736,
      "step": 22840
    },
    {
      "epoch": 11.973270440251572,
      "grad_norm": 1.8208988904953003,
      "learning_rate": 2.0775954181801893e-05,
      "loss": 0.0497,
      "num_input_tokens_seen": 14904544,
      "step": 22845
    },
    {
      "epoch": 11.975890985324948,
      "grad_norm": 2.6980903148651123,
      "learning_rate": 2.0764684734423455e-05,
      "loss": 0.1035,
      "num_input_tokens_seen": 14907648,
      "step": 22850
    },
    {
      "epoch": 11.978511530398324,
      "grad_norm": 1.271986484527588,
      "learning_rate": 2.075341617302465e-05,
      "loss": 0.0624,
      "num_input_tokens_seen": 14910816,
      "step": 22855
    },
    {
      "epoch": 11.981132075471699,
      "grad_norm": 1.7898235321044922,
      "learning_rate": 2.074214849996274e-05,
      "loss": 0.0466,
      "num_input_tokens_seen": 14913632,
      "step": 22860
    },
    {
      "epoch": 11.983752620545074,
      "grad_norm": 1.4303858280181885,
      "learning_rate": 2.0730881717594776e-05,
      "loss": 0.0466,
      "num_input_tokens_seen": 14916768,
      "step": 22865
    },
    {
      "epoch": 11.98637316561845,
      "grad_norm": 2.791747808456421,
      "learning_rate": 2.0719615828277657e-05,
      "loss": 0.0876,
      "num_input_tokens_seen": 14920896,
      "step": 22870
    },
    {
      "epoch": 11.988993710691824,
      "grad_norm": 2.2932071685791016,
      "learning_rate": 2.0708350834368083e-05,
      "loss": 0.0743,
      "num_input_tokens_seen": 14924192,
      "step": 22875
    },
    {
      "epoch": 11.9916142557652,
      "grad_norm": 1.4790012836456299,
      "learning_rate": 2.069708673822255e-05,
      "loss": 0.0845,
      "num_input_tokens_seen": 14926784,
      "step": 22880
    },
    {
      "epoch": 11.994234800838575,
      "grad_norm": 3.3081910610198975,
      "learning_rate": 2.068582354219737e-05,
      "loss": 0.1442,
      "num_input_tokens_seen": 14930432,
      "step": 22885
    },
    {
      "epoch": 11.99685534591195,
      "grad_norm": 1.755604863166809,
      "learning_rate": 2.0674561248648697e-05,
      "loss": 0.1,
      "num_input_tokens_seen": 14933664,
      "step": 22890
    },
    {
      "epoch": 11.999475890985325,
      "grad_norm": 1.6410818099975586,
      "learning_rate": 2.0663299859932456e-05,
      "loss": 0.0887,
      "num_input_tokens_seen": 14937088,
      "step": 22895
    },
    {
      "epoch": 12.0,
      "eval_loss": 1.008393406867981,
      "eval_runtime": 11.8965,
      "eval_samples_per_second": 71.281,
      "eval_steps_per_second": 17.82,
      "num_input_tokens_seen": 14937216,
      "step": 22896
    },
    {
      "epoch": 12.0020964360587,
      "grad_norm": 1.5891164541244507,
      "learning_rate": 2.065203937840442e-05,
      "loss": 0.0342,
      "num_input_tokens_seen": 14939712,
      "step": 22900
    },
    {
      "epoch": 12.004716981132075,
      "grad_norm": 4.4820332527160645,
      "learning_rate": 2.0640779806420136e-05,
      "loss": 0.0506,
      "num_input_tokens_seen": 14942272,
      "step": 22905
    },
    {
      "epoch": 12.00733752620545,
      "grad_norm": 1.3047118186950684,
      "learning_rate": 2.0629521146334978e-05,
      "loss": 0.0299,
      "num_input_tokens_seen": 14945632,
      "step": 22910
    },
    {
      "epoch": 12.009958071278826,
      "grad_norm": 1.2835240364074707,
      "learning_rate": 2.061826340050414e-05,
      "loss": 0.1155,
      "num_input_tokens_seen": 14949376,
      "step": 22915
    },
    {
      "epoch": 12.0125786163522,
      "grad_norm": 1.8773654699325562,
      "learning_rate": 2.0607006571282605e-05,
      "loss": 0.0622,
      "num_input_tokens_seen": 14951840,
      "step": 22920
    },
    {
      "epoch": 12.015199161425576,
      "grad_norm": 2.134777069091797,
      "learning_rate": 2.0595750661025183e-05,
      "loss": 0.0444,
      "num_input_tokens_seen": 14956768,
      "step": 22925
    },
    {
      "epoch": 12.017819706498951,
      "grad_norm": 1.3420261144638062,
      "learning_rate": 2.0584495672086483e-05,
      "loss": 0.0184,
      "num_input_tokens_seen": 14960384,
      "step": 22930
    },
    {
      "epoch": 12.020440251572326,
      "grad_norm": 2.0895328521728516,
      "learning_rate": 2.05732416068209e-05,
      "loss": 0.0423,
      "num_input_tokens_seen": 14963392,
      "step": 22935
    },
    {
      "epoch": 12.023060796645701,
      "grad_norm": 3.027205467224121,
      "learning_rate": 2.0561988467582676e-05,
      "loss": 0.058,
      "num_input_tokens_seen": 14967392,
      "step": 22940
    },
    {
      "epoch": 12.025681341719078,
      "grad_norm": 1.446829915046692,
      "learning_rate": 2.0550736256725828e-05,
      "loss": 0.1176,
      "num_input_tokens_seen": 14972704,
      "step": 22945
    },
    {
      "epoch": 12.028301886792454,
      "grad_norm": 1.7612450122833252,
      "learning_rate": 2.0539484976604196e-05,
      "loss": 0.049,
      "num_input_tokens_seen": 14975744,
      "step": 22950
    },
    {
      "epoch": 12.030922431865829,
      "grad_norm": 1.094801902770996,
      "learning_rate": 2.052823462957142e-05,
      "loss": 0.0358,
      "num_input_tokens_seen": 14979104,
      "step": 22955
    },
    {
      "epoch": 12.033542976939204,
      "grad_norm": 1.8562630414962769,
      "learning_rate": 2.0516985217980953e-05,
      "loss": 0.0491,
      "num_input_tokens_seen": 14981600,
      "step": 22960
    },
    {
      "epoch": 12.036163522012579,
      "grad_norm": 2.9605257511138916,
      "learning_rate": 2.050573674418602e-05,
      "loss": 0.0621,
      "num_input_tokens_seen": 14985152,
      "step": 22965
    },
    {
      "epoch": 12.038784067085954,
      "grad_norm": 1.8612940311431885,
      "learning_rate": 2.0494489210539693e-05,
      "loss": 0.0441,
      "num_input_tokens_seen": 14989568,
      "step": 22970
    },
    {
      "epoch": 12.04140461215933,
      "grad_norm": 1.0541702508926392,
      "learning_rate": 2.0483242619394817e-05,
      "loss": 0.0263,
      "num_input_tokens_seen": 14992352,
      "step": 22975
    },
    {
      "epoch": 12.044025157232705,
      "grad_norm": 1.5833027362823486,
      "learning_rate": 2.0471996973104063e-05,
      "loss": 0.0244,
      "num_input_tokens_seen": 14995168,
      "step": 22980
    },
    {
      "epoch": 12.04664570230608,
      "grad_norm": 5.010478973388672,
      "learning_rate": 2.0460752274019886e-05,
      "loss": 0.0695,
      "num_input_tokens_seen": 14998240,
      "step": 22985
    },
    {
      "epoch": 12.049266247379455,
      "grad_norm": 3.4963817596435547,
      "learning_rate": 2.0449508524494537e-05,
      "loss": 0.0384,
      "num_input_tokens_seen": 15001248,
      "step": 22990
    },
    {
      "epoch": 12.05188679245283,
      "grad_norm": 2.1382298469543457,
      "learning_rate": 2.0438265726880092e-05,
      "loss": 0.0507,
      "num_input_tokens_seen": 15004352,
      "step": 22995
    },
    {
      "epoch": 12.054507337526205,
      "grad_norm": 7.161627769470215,
      "learning_rate": 2.042702388352841e-05,
      "loss": 0.0281,
      "num_input_tokens_seen": 15007968,
      "step": 23000
    },
    {
      "epoch": 12.05712788259958,
      "grad_norm": 2.1651787757873535,
      "learning_rate": 2.041578299679117e-05,
      "loss": 0.0555,
      "num_input_tokens_seen": 15011648,
      "step": 23005
    },
    {
      "epoch": 12.059748427672956,
      "grad_norm": 4.592746257781982,
      "learning_rate": 2.0404543069019818e-05,
      "loss": 0.0379,
      "num_input_tokens_seen": 15014176,
      "step": 23010
    },
    {
      "epoch": 12.06236897274633,
      "grad_norm": 6.271389961242676,
      "learning_rate": 2.0393304102565642e-05,
      "loss": 0.0642,
      "num_input_tokens_seen": 15017184,
      "step": 23015
    },
    {
      "epoch": 12.064989517819706,
      "grad_norm": 1.7858694791793823,
      "learning_rate": 2.0382066099779685e-05,
      "loss": 0.0345,
      "num_input_tokens_seen": 15020416,
      "step": 23020
    },
    {
      "epoch": 12.067610062893081,
      "grad_norm": 3.0488059520721436,
      "learning_rate": 2.0370829063012808e-05,
      "loss": 0.0456,
      "num_input_tokens_seen": 15024800,
      "step": 23025
    },
    {
      "epoch": 12.070230607966456,
      "grad_norm": 4.454789161682129,
      "learning_rate": 2.035959299461568e-05,
      "loss": 0.0376,
      "num_input_tokens_seen": 15027456,
      "step": 23030
    },
    {
      "epoch": 12.072851153039831,
      "grad_norm": 1.6882842779159546,
      "learning_rate": 2.0348357896938762e-05,
      "loss": 0.08,
      "num_input_tokens_seen": 15030208,
      "step": 23035
    },
    {
      "epoch": 12.075471698113208,
      "grad_norm": 1.7377797365188599,
      "learning_rate": 2.03371237723323e-05,
      "loss": 0.0386,
      "num_input_tokens_seen": 15033120,
      "step": 23040
    },
    {
      "epoch": 12.078092243186584,
      "grad_norm": 1.721742033958435,
      "learning_rate": 2.032589062314636e-05,
      "loss": 0.0441,
      "num_input_tokens_seen": 15036960,
      "step": 23045
    },
    {
      "epoch": 12.080712788259959,
      "grad_norm": 0.9303075671195984,
      "learning_rate": 2.0314658451730767e-05,
      "loss": 0.0243,
      "num_input_tokens_seen": 15040992,
      "step": 23050
    },
    {
      "epoch": 12.083333333333334,
      "grad_norm": 0.6149488687515259,
      "learning_rate": 2.030342726043517e-05,
      "loss": 0.0423,
      "num_input_tokens_seen": 15044096,
      "step": 23055
    },
    {
      "epoch": 12.085953878406709,
      "grad_norm": 0.8560744524002075,
      "learning_rate": 2.0292197051609013e-05,
      "loss": 0.0583,
      "num_input_tokens_seen": 15047680,
      "step": 23060
    },
    {
      "epoch": 12.088574423480084,
      "grad_norm": 2.3256430625915527,
      "learning_rate": 2.0280967827601518e-05,
      "loss": 0.032,
      "num_input_tokens_seen": 15051328,
      "step": 23065
    },
    {
      "epoch": 12.09119496855346,
      "grad_norm": 2.8781938552856445,
      "learning_rate": 2.0269739590761723e-05,
      "loss": 0.025,
      "num_input_tokens_seen": 15054560,
      "step": 23070
    },
    {
      "epoch": 12.093815513626835,
      "grad_norm": 0.5171972513198853,
      "learning_rate": 2.0258512343438442e-05,
      "loss": 0.0409,
      "num_input_tokens_seen": 15057888,
      "step": 23075
    },
    {
      "epoch": 12.09643605870021,
      "grad_norm": 0.9584274291992188,
      "learning_rate": 2.0247286087980276e-05,
      "loss": 0.0472,
      "num_input_tokens_seen": 15061120,
      "step": 23080
    },
    {
      "epoch": 12.099056603773585,
      "grad_norm": 3.868999481201172,
      "learning_rate": 2.0236060826735644e-05,
      "loss": 0.0396,
      "num_input_tokens_seen": 15064224,
      "step": 23085
    },
    {
      "epoch": 12.10167714884696,
      "grad_norm": 1.3250383138656616,
      "learning_rate": 2.0224836562052732e-05,
      "loss": 0.0307,
      "num_input_tokens_seen": 15067360,
      "step": 23090
    },
    {
      "epoch": 12.104297693920335,
      "grad_norm": 0.862735390663147,
      "learning_rate": 2.0213613296279533e-05,
      "loss": 0.0264,
      "num_input_tokens_seen": 15070688,
      "step": 23095
    },
    {
      "epoch": 12.10691823899371,
      "grad_norm": 2.296875476837158,
      "learning_rate": 2.0202391031763844e-05,
      "loss": 0.0373,
      "num_input_tokens_seen": 15074304,
      "step": 23100
    },
    {
      "epoch": 12.109538784067086,
      "grad_norm": 1.171388030052185,
      "learning_rate": 2.0191169770853195e-05,
      "loss": 0.0335,
      "num_input_tokens_seen": 15076768,
      "step": 23105
    },
    {
      "epoch": 12.11215932914046,
      "grad_norm": 1.4175605773925781,
      "learning_rate": 2.0179949515894975e-05,
      "loss": 0.0412,
      "num_input_tokens_seen": 15079232,
      "step": 23110
    },
    {
      "epoch": 12.114779874213836,
      "grad_norm": 1.4695168733596802,
      "learning_rate": 2.016873026923633e-05,
      "loss": 0.0268,
      "num_input_tokens_seen": 15082688,
      "step": 23115
    },
    {
      "epoch": 12.117400419287211,
      "grad_norm": 2.7607991695404053,
      "learning_rate": 2.0157512033224187e-05,
      "loss": 0.0632,
      "num_input_tokens_seen": 15086400,
      "step": 23120
    },
    {
      "epoch": 12.120020964360586,
      "grad_norm": 2.4599356651306152,
      "learning_rate": 2.0146294810205287e-05,
      "loss": 0.0567,
      "num_input_tokens_seen": 15089184,
      "step": 23125
    },
    {
      "epoch": 12.122641509433961,
      "grad_norm": 0.8121739029884338,
      "learning_rate": 2.0135078602526146e-05,
      "loss": 0.0249,
      "num_input_tokens_seen": 15092384,
      "step": 23130
    },
    {
      "epoch": 12.125262054507338,
      "grad_norm": 2.238556385040283,
      "learning_rate": 2.0123863412533046e-05,
      "loss": 0.0494,
      "num_input_tokens_seen": 15095776,
      "step": 23135
    },
    {
      "epoch": 12.127882599580714,
      "grad_norm": 5.1331048011779785,
      "learning_rate": 2.0112649242572095e-05,
      "loss": 0.088,
      "num_input_tokens_seen": 15098624,
      "step": 23140
    },
    {
      "epoch": 12.130503144654089,
      "grad_norm": 1.489573359489441,
      "learning_rate": 2.0101436094989164e-05,
      "loss": 0.0619,
      "num_input_tokens_seen": 15100992,
      "step": 23145
    },
    {
      "epoch": 12.133123689727464,
      "grad_norm": 2.702333688735962,
      "learning_rate": 2.0090223972129918e-05,
      "loss": 0.0549,
      "num_input_tokens_seen": 15104352,
      "step": 23150
    },
    {
      "epoch": 12.135744234800839,
      "grad_norm": 1.8806854486465454,
      "learning_rate": 2.0079012876339797e-05,
      "loss": 0.0518,
      "num_input_tokens_seen": 15107200,
      "step": 23155
    },
    {
      "epoch": 12.138364779874214,
      "grad_norm": 1.4036577939987183,
      "learning_rate": 2.0067802809964055e-05,
      "loss": 0.0657,
      "num_input_tokens_seen": 15111008,
      "step": 23160
    },
    {
      "epoch": 12.14098532494759,
      "grad_norm": 3.964979887008667,
      "learning_rate": 2.0056593775347686e-05,
      "loss": 0.0428,
      "num_input_tokens_seen": 15113440,
      "step": 23165
    },
    {
      "epoch": 12.143605870020965,
      "grad_norm": 3.6355137825012207,
      "learning_rate": 2.00453857748355e-05,
      "loss": 0.0699,
      "num_input_tokens_seen": 15116256,
      "step": 23170
    },
    {
      "epoch": 12.14622641509434,
      "grad_norm": 1.1390252113342285,
      "learning_rate": 2.0034178810772082e-05,
      "loss": 0.0452,
      "num_input_tokens_seen": 15119296,
      "step": 23175
    },
    {
      "epoch": 12.148846960167715,
      "grad_norm": 1.5787153244018555,
      "learning_rate": 2.002297288550181e-05,
      "loss": 0.0391,
      "num_input_tokens_seen": 15123584,
      "step": 23180
    },
    {
      "epoch": 12.15146750524109,
      "grad_norm": 3.23732590675354,
      "learning_rate": 2.0011768001368835e-05,
      "loss": 0.029,
      "num_input_tokens_seen": 15127008,
      "step": 23185
    },
    {
      "epoch": 12.154088050314465,
      "grad_norm": 1.0648192167282104,
      "learning_rate": 2.0000564160717076e-05,
      "loss": 0.049,
      "num_input_tokens_seen": 15129728,
      "step": 23190
    },
    {
      "epoch": 12.15670859538784,
      "grad_norm": 0.9109271764755249,
      "learning_rate": 1.9989361365890263e-05,
      "loss": 0.043,
      "num_input_tokens_seen": 15133152,
      "step": 23195
    },
    {
      "epoch": 12.159329140461216,
      "grad_norm": 1.5615668296813965,
      "learning_rate": 1.997815961923188e-05,
      "loss": 0.0552,
      "num_input_tokens_seen": 15136960,
      "step": 23200
    },
    {
      "epoch": 12.16194968553459,
      "grad_norm": 2.162489175796509,
      "learning_rate": 1.996695892308522e-05,
      "loss": 0.0386,
      "num_input_tokens_seen": 15140128,
      "step": 23205
    },
    {
      "epoch": 12.164570230607966,
      "grad_norm": 2.3171913623809814,
      "learning_rate": 1.9955759279793327e-05,
      "loss": 0.0609,
      "num_input_tokens_seen": 15144128,
      "step": 23210
    },
    {
      "epoch": 12.167190775681341,
      "grad_norm": 2.0759236812591553,
      "learning_rate": 1.9944560691699057e-05,
      "loss": 0.0451,
      "num_input_tokens_seen": 15147968,
      "step": 23215
    },
    {
      "epoch": 12.169811320754716,
      "grad_norm": 1.1266456842422485,
      "learning_rate": 1.9933363161145006e-05,
      "loss": 0.0211,
      "num_input_tokens_seen": 15150976,
      "step": 23220
    },
    {
      "epoch": 12.172431865828091,
      "grad_norm": 1.6630487442016602,
      "learning_rate": 1.9922166690473575e-05,
      "loss": 0.0799,
      "num_input_tokens_seen": 15154368,
      "step": 23225
    },
    {
      "epoch": 12.175052410901468,
      "grad_norm": 1.4512466192245483,
      "learning_rate": 1.991097128202695e-05,
      "loss": 0.0478,
      "num_input_tokens_seen": 15157984,
      "step": 23230
    },
    {
      "epoch": 12.177672955974844,
      "grad_norm": 1.343061923980713,
      "learning_rate": 1.9899776938147065e-05,
      "loss": 0.1049,
      "num_input_tokens_seen": 15162208,
      "step": 23235
    },
    {
      "epoch": 12.180293501048219,
      "grad_norm": 1.5618456602096558,
      "learning_rate": 1.9888583661175665e-05,
      "loss": 0.0507,
      "num_input_tokens_seen": 15165088,
      "step": 23240
    },
    {
      "epoch": 12.182914046121594,
      "grad_norm": 2.398700714111328,
      "learning_rate": 1.9877391453454253e-05,
      "loss": 0.0378,
      "num_input_tokens_seen": 15167200,
      "step": 23245
    },
    {
      "epoch": 12.185534591194969,
      "grad_norm": 3.024097204208374,
      "learning_rate": 1.98662003173241e-05,
      "loss": 0.0572,
      "num_input_tokens_seen": 15170304,
      "step": 23250
    },
    {
      "epoch": 12.188155136268344,
      "grad_norm": 3.955712080001831,
      "learning_rate": 1.9855010255126266e-05,
      "loss": 0.0366,
      "num_input_tokens_seen": 15173024,
      "step": 23255
    },
    {
      "epoch": 12.19077568134172,
      "grad_norm": 2.9188942909240723,
      "learning_rate": 1.9843821269201598e-05,
      "loss": 0.0527,
      "num_input_tokens_seen": 15176960,
      "step": 23260
    },
    {
      "epoch": 12.193396226415095,
      "grad_norm": 1.4515939950942993,
      "learning_rate": 1.983263336189069e-05,
      "loss": 0.0638,
      "num_input_tokens_seen": 15184576,
      "step": 23265
    },
    {
      "epoch": 12.19601677148847,
      "grad_norm": 0.6088401079177856,
      "learning_rate": 1.9821446535533937e-05,
      "loss": 0.0723,
      "num_input_tokens_seen": 15187840,
      "step": 23270
    },
    {
      "epoch": 12.198637316561845,
      "grad_norm": 0.9187520146369934,
      "learning_rate": 1.9810260792471492e-05,
      "loss": 0.0438,
      "num_input_tokens_seen": 15191072,
      "step": 23275
    },
    {
      "epoch": 12.20125786163522,
      "grad_norm": 0.18483483791351318,
      "learning_rate": 1.979907613504327e-05,
      "loss": 0.0399,
      "num_input_tokens_seen": 15194592,
      "step": 23280
    },
    {
      "epoch": 12.203878406708595,
      "grad_norm": 2.283893346786499,
      "learning_rate": 1.978789256558899e-05,
      "loss": 0.0314,
      "num_input_tokens_seen": 15198240,
      "step": 23285
    },
    {
      "epoch": 12.20649895178197,
      "grad_norm": 1.490874171257019,
      "learning_rate": 1.977671008644812e-05,
      "loss": 0.0658,
      "num_input_tokens_seen": 15201056,
      "step": 23290
    },
    {
      "epoch": 12.209119496855346,
      "grad_norm": 2.560783624649048,
      "learning_rate": 1.9765528699959912e-05,
      "loss": 0.0503,
      "num_input_tokens_seen": 15204320,
      "step": 23295
    },
    {
      "epoch": 12.21174004192872,
      "grad_norm": 1.9961833953857422,
      "learning_rate": 1.975434840846339e-05,
      "loss": 0.0717,
      "num_input_tokens_seen": 15207264,
      "step": 23300
    },
    {
      "epoch": 12.214360587002096,
      "grad_norm": 0.8833478689193726,
      "learning_rate": 1.9743169214297326e-05,
      "loss": 0.0327,
      "num_input_tokens_seen": 15210112,
      "step": 23305
    },
    {
      "epoch": 12.216981132075471,
      "grad_norm": 2.149683713912964,
      "learning_rate": 1.973199111980029e-05,
      "loss": 0.0416,
      "num_input_tokens_seen": 15213504,
      "step": 23310
    },
    {
      "epoch": 12.219601677148846,
      "grad_norm": 4.467939853668213,
      "learning_rate": 1.9720814127310608e-05,
      "loss": 0.0476,
      "num_input_tokens_seen": 15215808,
      "step": 23315
    },
    {
      "epoch": 12.222222222222221,
      "grad_norm": 3.803388833999634,
      "learning_rate": 1.9709638239166386e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 15218304,
      "step": 23320
    },
    {
      "epoch": 12.224842767295598,
      "grad_norm": 2.4599905014038086,
      "learning_rate": 1.9698463457705485e-05,
      "loss": 0.1257,
      "num_input_tokens_seen": 15221024,
      "step": 23325
    },
    {
      "epoch": 12.227463312368974,
      "grad_norm": 0.6246377825737,
      "learning_rate": 1.968728978526556e-05,
      "loss": 0.0467,
      "num_input_tokens_seen": 15223872,
      "step": 23330
    },
    {
      "epoch": 12.230083857442349,
      "grad_norm": 0.6201596856117249,
      "learning_rate": 1.9676117224183986e-05,
      "loss": 0.0308,
      "num_input_tokens_seen": 15227168,
      "step": 23335
    },
    {
      "epoch": 12.232704402515724,
      "grad_norm": 1.3443471193313599,
      "learning_rate": 1.9664945776797956e-05,
      "loss": 0.0385,
      "num_input_tokens_seen": 15230816,
      "step": 23340
    },
    {
      "epoch": 12.235324947589099,
      "grad_norm": 3.3468146324157715,
      "learning_rate": 1.96537754454444e-05,
      "loss": 0.0447,
      "num_input_tokens_seen": 15233568,
      "step": 23345
    },
    {
      "epoch": 12.237945492662474,
      "grad_norm": 2.4988489151000977,
      "learning_rate": 1.964260623246003e-05,
      "loss": 0.0214,
      "num_input_tokens_seen": 15236064,
      "step": 23350
    },
    {
      "epoch": 12.24056603773585,
      "grad_norm": 1.4268535375595093,
      "learning_rate": 1.9631438140181324e-05,
      "loss": 0.0467,
      "num_input_tokens_seen": 15239168,
      "step": 23355
    },
    {
      "epoch": 12.243186582809225,
      "grad_norm": 2.511467695236206,
      "learning_rate": 1.9620271170944503e-05,
      "loss": 0.0455,
      "num_input_tokens_seen": 15241984,
      "step": 23360
    },
    {
      "epoch": 12.2458071278826,
      "grad_norm": 1.9577322006225586,
      "learning_rate": 1.960910532708558e-05,
      "loss": 0.0538,
      "num_input_tokens_seen": 15244800,
      "step": 23365
    },
    {
      "epoch": 12.248427672955975,
      "grad_norm": 1.9663578271865845,
      "learning_rate": 1.9597940610940315e-05,
      "loss": 0.0602,
      "num_input_tokens_seen": 15248416,
      "step": 23370
    },
    {
      "epoch": 12.25104821802935,
      "grad_norm": 2.1604743003845215,
      "learning_rate": 1.958677702484425e-05,
      "loss": 0.025,
      "num_input_tokens_seen": 15251008,
      "step": 23375
    },
    {
      "epoch": 12.253668763102725,
      "grad_norm": 1.5049406290054321,
      "learning_rate": 1.957561457113267e-05,
      "loss": 0.0648,
      "num_input_tokens_seen": 15256992,
      "step": 23380
    },
    {
      "epoch": 12.2562893081761,
      "grad_norm": 1.094089150428772,
      "learning_rate": 1.9564453252140648e-05,
      "loss": 0.0392,
      "num_input_tokens_seen": 15259712,
      "step": 23385
    },
    {
      "epoch": 12.258909853249476,
      "grad_norm": 3.72495436668396,
      "learning_rate": 1.9553293070202984e-05,
      "loss": 0.0894,
      "num_input_tokens_seen": 15262880,
      "step": 23390
    },
    {
      "epoch": 12.26153039832285,
      "grad_norm": 2.494004011154175,
      "learning_rate": 1.9542134027654267e-05,
      "loss": 0.0535,
      "num_input_tokens_seen": 15266048,
      "step": 23395
    },
    {
      "epoch": 12.264150943396226,
      "grad_norm": 3.467081308364868,
      "learning_rate": 1.953097612682885e-05,
      "loss": 0.0922,
      "num_input_tokens_seen": 15268928,
      "step": 23400
    },
    {
      "epoch": 12.266771488469601,
      "grad_norm": 1.493283987045288,
      "learning_rate": 1.9519819370060832e-05,
      "loss": 0.0986,
      "num_input_tokens_seen": 15271872,
      "step": 23405
    },
    {
      "epoch": 12.269392033542976,
      "grad_norm": 1.4935554265975952,
      "learning_rate": 1.9508663759684074e-05,
      "loss": 0.06,
      "num_input_tokens_seen": 15276704,
      "step": 23410
    },
    {
      "epoch": 12.272012578616351,
      "grad_norm": 1.5912994146347046,
      "learning_rate": 1.9497509298032224e-05,
      "loss": 0.0365,
      "num_input_tokens_seen": 15279648,
      "step": 23415
    },
    {
      "epoch": 12.274633123689728,
      "grad_norm": 2.3199360370635986,
      "learning_rate": 1.9486355987438647e-05,
      "loss": 0.0425,
      "num_input_tokens_seen": 15282176,
      "step": 23420
    },
    {
      "epoch": 12.277253668763104,
      "grad_norm": 1.6085611581802368,
      "learning_rate": 1.947520383023649e-05,
      "loss": 0.0475,
      "num_input_tokens_seen": 15284640,
      "step": 23425
    },
    {
      "epoch": 12.279874213836479,
      "grad_norm": 1.735613465309143,
      "learning_rate": 1.9464052828758666e-05,
      "loss": 0.0428,
      "num_input_tokens_seen": 15287712,
      "step": 23430
    },
    {
      "epoch": 12.282494758909854,
      "grad_norm": 2.3607935905456543,
      "learning_rate": 1.945290298533783e-05,
      "loss": 0.0474,
      "num_input_tokens_seen": 15291296,
      "step": 23435
    },
    {
      "epoch": 12.285115303983229,
      "grad_norm": 1.4369264841079712,
      "learning_rate": 1.9441754302306406e-05,
      "loss": 0.0644,
      "num_input_tokens_seen": 15294464,
      "step": 23440
    },
    {
      "epoch": 12.287735849056604,
      "grad_norm": 1.1887909173965454,
      "learning_rate": 1.9430606781996578e-05,
      "loss": 0.0193,
      "num_input_tokens_seen": 15299392,
      "step": 23445
    },
    {
      "epoch": 12.29035639412998,
      "grad_norm": 1.7493114471435547,
      "learning_rate": 1.941946042674026e-05,
      "loss": 0.0534,
      "num_input_tokens_seen": 15302496,
      "step": 23450
    },
    {
      "epoch": 12.292976939203355,
      "grad_norm": 1.5030969381332397,
      "learning_rate": 1.940831523886916e-05,
      "loss": 0.1005,
      "num_input_tokens_seen": 15306464,
      "step": 23455
    },
    {
      "epoch": 12.29559748427673,
      "grad_norm": 1.3286375999450684,
      "learning_rate": 1.9397171220714714e-05,
      "loss": 0.038,
      "num_input_tokens_seen": 15309792,
      "step": 23460
    },
    {
      "epoch": 12.298218029350105,
      "grad_norm": 2.6059319972991943,
      "learning_rate": 1.9386028374608134e-05,
      "loss": 0.1036,
      "num_input_tokens_seen": 15312576,
      "step": 23465
    },
    {
      "epoch": 12.30083857442348,
      "grad_norm": 3.2369749546051025,
      "learning_rate": 1.9374886702880372e-05,
      "loss": 0.0329,
      "num_input_tokens_seen": 15315296,
      "step": 23470
    },
    {
      "epoch": 12.303459119496855,
      "grad_norm": 1.6340727806091309,
      "learning_rate": 1.936374620786213e-05,
      "loss": 0.0511,
      "num_input_tokens_seen": 15318112,
      "step": 23475
    },
    {
      "epoch": 12.30607966457023,
      "grad_norm": 3.9422056674957275,
      "learning_rate": 1.9352606891883885e-05,
      "loss": 0.0454,
      "num_input_tokens_seen": 15320896,
      "step": 23480
    },
    {
      "epoch": 12.308700209643606,
      "grad_norm": 6.149346351623535,
      "learning_rate": 1.9341468757275844e-05,
      "loss": 0.0217,
      "num_input_tokens_seen": 15323904,
      "step": 23485
    },
    {
      "epoch": 12.31132075471698,
      "grad_norm": 3.1884751319885254,
      "learning_rate": 1.933033180636798e-05,
      "loss": 0.0461,
      "num_input_tokens_seen": 15326944,
      "step": 23490
    },
    {
      "epoch": 12.313941299790356,
      "grad_norm": 1.2243235111236572,
      "learning_rate": 1.9319196041490027e-05,
      "loss": 0.043,
      "num_input_tokens_seen": 15331776,
      "step": 23495
    },
    {
      "epoch": 12.316561844863731,
      "grad_norm": 1.8624602556228638,
      "learning_rate": 1.930806146497146e-05,
      "loss": 0.0474,
      "num_input_tokens_seen": 15334624,
      "step": 23500
    },
    {
      "epoch": 12.319182389937106,
      "grad_norm": 0.9603529572486877,
      "learning_rate": 1.9296928079141485e-05,
      "loss": 0.0568,
      "num_input_tokens_seen": 15339072,
      "step": 23505
    },
    {
      "epoch": 12.321802935010481,
      "grad_norm": 2.2838387489318848,
      "learning_rate": 1.9285795886329093e-05,
      "loss": 0.0572,
      "num_input_tokens_seen": 15342240,
      "step": 23510
    },
    {
      "epoch": 12.324423480083858,
      "grad_norm": 1.4229201078414917,
      "learning_rate": 1.927466488886301e-05,
      "loss": 0.0345,
      "num_input_tokens_seen": 15346528,
      "step": 23515
    },
    {
      "epoch": 12.327044025157234,
      "grad_norm": 1.2664192914962769,
      "learning_rate": 1.926353508907172e-05,
      "loss": 0.0839,
      "num_input_tokens_seen": 15350144,
      "step": 23520
    },
    {
      "epoch": 12.329664570230609,
      "grad_norm": 1.2547588348388672,
      "learning_rate": 1.9252406489283437e-05,
      "loss": 0.0194,
      "num_input_tokens_seen": 15352832,
      "step": 23525
    },
    {
      "epoch": 12.332285115303984,
      "grad_norm": 1.4271361827850342,
      "learning_rate": 1.9241279091826155e-05,
      "loss": 0.0427,
      "num_input_tokens_seen": 15355392,
      "step": 23530
    },
    {
      "epoch": 12.334905660377359,
      "grad_norm": 8.341511726379395,
      "learning_rate": 1.923015289902758e-05,
      "loss": 0.0374,
      "num_input_tokens_seen": 15357504,
      "step": 23535
    },
    {
      "epoch": 12.337526205450734,
      "grad_norm": 1.0933375358581543,
      "learning_rate": 1.9219027913215193e-05,
      "loss": 0.0276,
      "num_input_tokens_seen": 15360672,
      "step": 23540
    },
    {
      "epoch": 12.34014675052411,
      "grad_norm": 0.616122841835022,
      "learning_rate": 1.9207904136716214e-05,
      "loss": 0.1289,
      "num_input_tokens_seen": 15364128,
      "step": 23545
    },
    {
      "epoch": 12.342767295597485,
      "grad_norm": 2.900543451309204,
      "learning_rate": 1.9196781571857608e-05,
      "loss": 0.0251,
      "num_input_tokens_seen": 15367072,
      "step": 23550
    },
    {
      "epoch": 12.34538784067086,
      "grad_norm": 1.393324851989746,
      "learning_rate": 1.9185660220966097e-05,
      "loss": 0.0399,
      "num_input_tokens_seen": 15370848,
      "step": 23555
    },
    {
      "epoch": 12.348008385744235,
      "grad_norm": 2.603376626968384,
      "learning_rate": 1.9174540086368135e-05,
      "loss": 0.0538,
      "num_input_tokens_seen": 15375680,
      "step": 23560
    },
    {
      "epoch": 12.35062893081761,
      "grad_norm": 4.67948055267334,
      "learning_rate": 1.9163421170389918e-05,
      "loss": 0.0714,
      "num_input_tokens_seen": 15378784,
      "step": 23565
    },
    {
      "epoch": 12.353249475890985,
      "grad_norm": 2.604543685913086,
      "learning_rate": 1.9152303475357402e-05,
      "loss": 0.055,
      "num_input_tokens_seen": 15381856,
      "step": 23570
    },
    {
      "epoch": 12.35587002096436,
      "grad_norm": 2.2634613513946533,
      "learning_rate": 1.9141187003596295e-05,
      "loss": 0.058,
      "num_input_tokens_seen": 15385376,
      "step": 23575
    },
    {
      "epoch": 12.358490566037736,
      "grad_norm": 2.3618338108062744,
      "learning_rate": 1.913007175743202e-05,
      "loss": 0.0416,
      "num_input_tokens_seen": 15389568,
      "step": 23580
    },
    {
      "epoch": 12.36111111111111,
      "grad_norm": 1.1344000101089478,
      "learning_rate": 1.9118957739189773e-05,
      "loss": 0.0226,
      "num_input_tokens_seen": 15392160,
      "step": 23585
    },
    {
      "epoch": 12.363731656184486,
      "grad_norm": 2.2277941703796387,
      "learning_rate": 1.910784495119447e-05,
      "loss": 0.0319,
      "num_input_tokens_seen": 15395520,
      "step": 23590
    },
    {
      "epoch": 12.366352201257861,
      "grad_norm": 2.72904896736145,
      "learning_rate": 1.9096733395770778e-05,
      "loss": 0.0882,
      "num_input_tokens_seen": 15398592,
      "step": 23595
    },
    {
      "epoch": 12.368972746331236,
      "grad_norm": 2.5831105709075928,
      "learning_rate": 1.9085623075243114e-05,
      "loss": 0.0398,
      "num_input_tokens_seen": 15401824,
      "step": 23600
    },
    {
      "epoch": 12.371593291404611,
      "grad_norm": 0.4076848030090332,
      "learning_rate": 1.9074513991935623e-05,
      "loss": 0.0511,
      "num_input_tokens_seen": 15405536,
      "step": 23605
    },
    {
      "epoch": 12.374213836477987,
      "grad_norm": 2.379743814468384,
      "learning_rate": 1.906340614817221e-05,
      "loss": 0.0585,
      "num_input_tokens_seen": 15408800,
      "step": 23610
    },
    {
      "epoch": 12.376834381551364,
      "grad_norm": 1.966694712638855,
      "learning_rate": 1.9052299546276506e-05,
      "loss": 0.0774,
      "num_input_tokens_seen": 15411392,
      "step": 23615
    },
    {
      "epoch": 12.379454926624739,
      "grad_norm": 3.761150598526001,
      "learning_rate": 1.904119418857187e-05,
      "loss": 0.0606,
      "num_input_tokens_seen": 15414144,
      "step": 23620
    },
    {
      "epoch": 12.382075471698114,
      "grad_norm": 1.532896876335144,
      "learning_rate": 1.903009007738144e-05,
      "loss": 0.0843,
      "num_input_tokens_seen": 15417984,
      "step": 23625
    },
    {
      "epoch": 12.384696016771489,
      "grad_norm": 4.221816539764404,
      "learning_rate": 1.9018987215028045e-05,
      "loss": 0.0227,
      "num_input_tokens_seen": 15421280,
      "step": 23630
    },
    {
      "epoch": 12.387316561844864,
      "grad_norm": 1.9102933406829834,
      "learning_rate": 1.9007885603834298e-05,
      "loss": 0.0434,
      "num_input_tokens_seen": 15423872,
      "step": 23635
    },
    {
      "epoch": 12.38993710691824,
      "grad_norm": 1.3527566194534302,
      "learning_rate": 1.8996785246122526e-05,
      "loss": 0.0348,
      "num_input_tokens_seen": 15426976,
      "step": 23640
    },
    {
      "epoch": 12.392557651991615,
      "grad_norm": 1.3809731006622314,
      "learning_rate": 1.898568614421478e-05,
      "loss": 0.0384,
      "num_input_tokens_seen": 15430240,
      "step": 23645
    },
    {
      "epoch": 12.39517819706499,
      "grad_norm": 1.119235873222351,
      "learning_rate": 1.897458830043288e-05,
      "loss": 0.0401,
      "num_input_tokens_seen": 15433664,
      "step": 23650
    },
    {
      "epoch": 12.397798742138365,
      "grad_norm": 3.1429250240325928,
      "learning_rate": 1.8963491717098372e-05,
      "loss": 0.0633,
      "num_input_tokens_seen": 15437088,
      "step": 23655
    },
    {
      "epoch": 12.40041928721174,
      "grad_norm": 1.4521797895431519,
      "learning_rate": 1.8952396396532522e-05,
      "loss": 0.0304,
      "num_input_tokens_seen": 15440096,
      "step": 23660
    },
    {
      "epoch": 12.403039832285115,
      "grad_norm": 2.1337344646453857,
      "learning_rate": 1.8941302341056356e-05,
      "loss": 0.0666,
      "num_input_tokens_seen": 15444352,
      "step": 23665
    },
    {
      "epoch": 12.40566037735849,
      "grad_norm": 1.555101752281189,
      "learning_rate": 1.8930209552990625e-05,
      "loss": 0.0602,
      "num_input_tokens_seen": 15448736,
      "step": 23670
    },
    {
      "epoch": 12.408280922431866,
      "grad_norm": 0.7924271821975708,
      "learning_rate": 1.8919118034655796e-05,
      "loss": 0.023,
      "num_input_tokens_seen": 15451712,
      "step": 23675
    },
    {
      "epoch": 12.41090146750524,
      "grad_norm": 5.936469554901123,
      "learning_rate": 1.8908027788372107e-05,
      "loss": 0.05,
      "num_input_tokens_seen": 15454272,
      "step": 23680
    },
    {
      "epoch": 12.413522012578616,
      "grad_norm": 3.5538582801818848,
      "learning_rate": 1.88969388164595e-05,
      "loss": 0.072,
      "num_input_tokens_seen": 15457856,
      "step": 23685
    },
    {
      "epoch": 12.416142557651991,
      "grad_norm": 5.238059997558594,
      "learning_rate": 1.8885851121237666e-05,
      "loss": 0.0384,
      "num_input_tokens_seen": 15461408,
      "step": 23690
    },
    {
      "epoch": 12.418763102725366,
      "grad_norm": 6.79067850112915,
      "learning_rate": 1.8874764705026023e-05,
      "loss": 0.066,
      "num_input_tokens_seen": 15464992,
      "step": 23695
    },
    {
      "epoch": 12.421383647798741,
      "grad_norm": 1.475769281387329,
      "learning_rate": 1.8863679570143728e-05,
      "loss": 0.0503,
      "num_input_tokens_seen": 15468096,
      "step": 23700
    },
    {
      "epoch": 12.424004192872117,
      "grad_norm": 1.5157322883605957,
      "learning_rate": 1.885259571890966e-05,
      "loss": 0.0374,
      "num_input_tokens_seen": 15470752,
      "step": 23705
    },
    {
      "epoch": 12.426624737945493,
      "grad_norm": 0.7951902151107788,
      "learning_rate": 1.8841513153642427e-05,
      "loss": 0.0383,
      "num_input_tokens_seen": 15473952,
      "step": 23710
    },
    {
      "epoch": 12.429245283018869,
      "grad_norm": 1.9278556108474731,
      "learning_rate": 1.8830431876660388e-05,
      "loss": 0.0487,
      "num_input_tokens_seen": 15477760,
      "step": 23715
    },
    {
      "epoch": 12.431865828092244,
      "grad_norm": 2.5013272762298584,
      "learning_rate": 1.8819351890281607e-05,
      "loss": 0.0473,
      "num_input_tokens_seen": 15480384,
      "step": 23720
    },
    {
      "epoch": 12.434486373165619,
      "grad_norm": 4.862572193145752,
      "learning_rate": 1.880827319682391e-05,
      "loss": 0.0599,
      "num_input_tokens_seen": 15482976,
      "step": 23725
    },
    {
      "epoch": 12.437106918238994,
      "grad_norm": 5.775207042694092,
      "learning_rate": 1.879719579860481e-05,
      "loss": 0.0355,
      "num_input_tokens_seen": 15486016,
      "step": 23730
    },
    {
      "epoch": 12.43972746331237,
      "grad_norm": 2.9845781326293945,
      "learning_rate": 1.878611969794159e-05,
      "loss": 0.044,
      "num_input_tokens_seen": 15489472,
      "step": 23735
    },
    {
      "epoch": 12.442348008385745,
      "grad_norm": 3.5147361755371094,
      "learning_rate": 1.8775044897151232e-05,
      "loss": 0.0471,
      "num_input_tokens_seen": 15493120,
      "step": 23740
    },
    {
      "epoch": 12.44496855345912,
      "grad_norm": 1.7379909753799438,
      "learning_rate": 1.876397139855047e-05,
      "loss": 0.0561,
      "num_input_tokens_seen": 15495968,
      "step": 23745
    },
    {
      "epoch": 12.447589098532495,
      "grad_norm": 1.2742093801498413,
      "learning_rate": 1.875289920445574e-05,
      "loss": 0.0288,
      "num_input_tokens_seen": 15498784,
      "step": 23750
    },
    {
      "epoch": 12.45020964360587,
      "grad_norm": 1.4277106523513794,
      "learning_rate": 1.8741828317183234e-05,
      "loss": 0.0768,
      "num_input_tokens_seen": 15502784,
      "step": 23755
    },
    {
      "epoch": 12.452830188679245,
      "grad_norm": 1.1336331367492676,
      "learning_rate": 1.873075873904885e-05,
      "loss": 0.0291,
      "num_input_tokens_seen": 15505952,
      "step": 23760
    },
    {
      "epoch": 12.45545073375262,
      "grad_norm": 5.560193061828613,
      "learning_rate": 1.87196904723682e-05,
      "loss": 0.0324,
      "num_input_tokens_seen": 15509344,
      "step": 23765
    },
    {
      "epoch": 12.458071278825996,
      "grad_norm": 2.577239990234375,
      "learning_rate": 1.8708623519456665e-05,
      "loss": 0.0552,
      "num_input_tokens_seen": 15512544,
      "step": 23770
    },
    {
      "epoch": 12.46069182389937,
      "grad_norm": 1.6018413305282593,
      "learning_rate": 1.8697557882629308e-05,
      "loss": 0.0706,
      "num_input_tokens_seen": 15519104,
      "step": 23775
    },
    {
      "epoch": 12.463312368972746,
      "grad_norm": 2.8963840007781982,
      "learning_rate": 1.868649356420094e-05,
      "loss": 0.0454,
      "num_input_tokens_seen": 15522528,
      "step": 23780
    },
    {
      "epoch": 12.465932914046121,
      "grad_norm": 1.2811774015426636,
      "learning_rate": 1.86754305664861e-05,
      "loss": 0.0528,
      "num_input_tokens_seen": 15525440,
      "step": 23785
    },
    {
      "epoch": 12.468553459119496,
      "grad_norm": 3.972163200378418,
      "learning_rate": 1.866436889179902e-05,
      "loss": 0.0403,
      "num_input_tokens_seen": 15528448,
      "step": 23790
    },
    {
      "epoch": 12.471174004192871,
      "grad_norm": 0.7337716817855835,
      "learning_rate": 1.865330854245369e-05,
      "loss": 0.0198,
      "num_input_tokens_seen": 15531392,
      "step": 23795
    },
    {
      "epoch": 12.473794549266247,
      "grad_norm": 3.849102735519409,
      "learning_rate": 1.8642249520763804e-05,
      "loss": 0.0398,
      "num_input_tokens_seen": 15534784,
      "step": 23800
    },
    {
      "epoch": 12.476415094339623,
      "grad_norm": 1.3151403665542603,
      "learning_rate": 1.863119182904278e-05,
      "loss": 0.0463,
      "num_input_tokens_seen": 15538144,
      "step": 23805
    },
    {
      "epoch": 12.479035639412999,
      "grad_norm": 1.4970276355743408,
      "learning_rate": 1.8620135469603777e-05,
      "loss": 0.047,
      "num_input_tokens_seen": 15541600,
      "step": 23810
    },
    {
      "epoch": 12.481656184486374,
      "grad_norm": 1.9441750049591064,
      "learning_rate": 1.8609080444759653e-05,
      "loss": 0.0988,
      "num_input_tokens_seen": 15545728,
      "step": 23815
    },
    {
      "epoch": 12.484276729559749,
      "grad_norm": 1.7480846643447876,
      "learning_rate": 1.8598026756822978e-05,
      "loss": 0.056,
      "num_input_tokens_seen": 15548000,
      "step": 23820
    },
    {
      "epoch": 12.486897274633124,
      "grad_norm": 0.9821299314498901,
      "learning_rate": 1.8586974408106074e-05,
      "loss": 0.0457,
      "num_input_tokens_seen": 15550496,
      "step": 23825
    },
    {
      "epoch": 12.4895178197065,
      "grad_norm": 5.093685150146484,
      "learning_rate": 1.8575923400920954e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 15554304,
      "step": 23830
    },
    {
      "epoch": 12.492138364779874,
      "grad_norm": 4.149637222290039,
      "learning_rate": 1.856487373757937e-05,
      "loss": 0.0894,
      "num_input_tokens_seen": 15557824,
      "step": 23835
    },
    {
      "epoch": 12.49475890985325,
      "grad_norm": 3.987257242202759,
      "learning_rate": 1.8553825420392792e-05,
      "loss": 0.0566,
      "num_input_tokens_seen": 15560704,
      "step": 23840
    },
    {
      "epoch": 12.497379454926625,
      "grad_norm": 2.0088396072387695,
      "learning_rate": 1.8542778451672385e-05,
      "loss": 0.0616,
      "num_input_tokens_seen": 15564832,
      "step": 23845
    },
    {
      "epoch": 12.5,
      "grad_norm": 3.6258740425109863,
      "learning_rate": 1.8531732833729063e-05,
      "loss": 0.0495,
      "num_input_tokens_seen": 15567904,
      "step": 23850
    },
    {
      "epoch": 12.502620545073375,
      "grad_norm": 4.09792947769165,
      "learning_rate": 1.852068856887343e-05,
      "loss": 0.0767,
      "num_input_tokens_seen": 15571168,
      "step": 23855
    },
    {
      "epoch": 12.50524109014675,
      "grad_norm": 2.5482146739959717,
      "learning_rate": 1.8509645659415837e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 15574144,
      "step": 23860
    },
    {
      "epoch": 12.507861635220126,
      "grad_norm": 1.0915831327438354,
      "learning_rate": 1.849860410766632e-05,
      "loss": 0.0484,
      "num_input_tokens_seen": 15576160,
      "step": 23865
    },
    {
      "epoch": 12.5104821802935,
      "grad_norm": 1.5838028192520142,
      "learning_rate": 1.8487563915934664e-05,
      "loss": 0.0427,
      "num_input_tokens_seen": 15579424,
      "step": 23870
    },
    {
      "epoch": 12.513102725366876,
      "grad_norm": 1.6606312990188599,
      "learning_rate": 1.847652508653033e-05,
      "loss": 0.0337,
      "num_input_tokens_seen": 15584032,
      "step": 23875
    },
    {
      "epoch": 12.515723270440251,
      "grad_norm": 4.937249183654785,
      "learning_rate": 1.846548762176252e-05,
      "loss": 0.0609,
      "num_input_tokens_seen": 15588544,
      "step": 23880
    },
    {
      "epoch": 12.518343815513626,
      "grad_norm": 3.066725015640259,
      "learning_rate": 1.845445152394015e-05,
      "loss": 0.0333,
      "num_input_tokens_seen": 15594336,
      "step": 23885
    },
    {
      "epoch": 12.520964360587001,
      "grad_norm": 3.6009926795959473,
      "learning_rate": 1.8443416795371853e-05,
      "loss": 0.0557,
      "num_input_tokens_seen": 15597760,
      "step": 23890
    },
    {
      "epoch": 12.523584905660378,
      "grad_norm": 1.3796197175979614,
      "learning_rate": 1.8432383438365957e-05,
      "loss": 0.0362,
      "num_input_tokens_seen": 15601824,
      "step": 23895
    },
    {
      "epoch": 12.526205450733752,
      "grad_norm": 3.879650592803955,
      "learning_rate": 1.8421351455230528e-05,
      "loss": 0.1301,
      "num_input_tokens_seen": 15605824,
      "step": 23900
    },
    {
      "epoch": 12.528825995807129,
      "grad_norm": 1.524578332901001,
      "learning_rate": 1.8410320848273315e-05,
      "loss": 0.0478,
      "num_input_tokens_seen": 15609312,
      "step": 23905
    },
    {
      "epoch": 12.531446540880504,
      "grad_norm": 1.058725357055664,
      "learning_rate": 1.8399291619801794e-05,
      "loss": 0.0889,
      "num_input_tokens_seen": 15613216,
      "step": 23910
    },
    {
      "epoch": 12.534067085953879,
      "grad_norm": 0.6191256642341614,
      "learning_rate": 1.838826377212317e-05,
      "loss": 0.0375,
      "num_input_tokens_seen": 15616352,
      "step": 23915
    },
    {
      "epoch": 12.536687631027254,
      "grad_norm": 2.003742218017578,
      "learning_rate": 1.8377237307544326e-05,
      "loss": 0.0525,
      "num_input_tokens_seen": 15619072,
      "step": 23920
    },
    {
      "epoch": 12.53930817610063,
      "grad_norm": 0.9811654686927795,
      "learning_rate": 1.8366212228371894e-05,
      "loss": 0.061,
      "num_input_tokens_seen": 15622144,
      "step": 23925
    },
    {
      "epoch": 12.541928721174004,
      "grad_norm": 4.141577243804932,
      "learning_rate": 1.8355188536912173e-05,
      "loss": 0.0342,
      "num_input_tokens_seen": 15625152,
      "step": 23930
    },
    {
      "epoch": 12.54454926624738,
      "grad_norm": 3.447864532470703,
      "learning_rate": 1.83441662354712e-05,
      "loss": 0.0637,
      "num_input_tokens_seen": 15628128,
      "step": 23935
    },
    {
      "epoch": 12.547169811320755,
      "grad_norm": 4.815441608428955,
      "learning_rate": 1.8333145326354722e-05,
      "loss": 0.0653,
      "num_input_tokens_seen": 15631328,
      "step": 23940
    },
    {
      "epoch": 12.54979035639413,
      "grad_norm": 1.5844682455062866,
      "learning_rate": 1.8322125811868175e-05,
      "loss": 0.0898,
      "num_input_tokens_seen": 15634304,
      "step": 23945
    },
    {
      "epoch": 12.552410901467505,
      "grad_norm": 1.1307458877563477,
      "learning_rate": 1.831110769431673e-05,
      "loss": 0.03,
      "num_input_tokens_seen": 15637984,
      "step": 23950
    },
    {
      "epoch": 12.55503144654088,
      "grad_norm": 16.857358932495117,
      "learning_rate": 1.8300090976005247e-05,
      "loss": 0.0415,
      "num_input_tokens_seen": 15640672,
      "step": 23955
    },
    {
      "epoch": 12.557651991614255,
      "grad_norm": 0.9088637828826904,
      "learning_rate": 1.8289075659238285e-05,
      "loss": 0.0438,
      "num_input_tokens_seen": 15644096,
      "step": 23960
    },
    {
      "epoch": 12.56027253668763,
      "grad_norm": 1.9831628799438477,
      "learning_rate": 1.8278061746320134e-05,
      "loss": 0.071,
      "num_input_tokens_seen": 15647040,
      "step": 23965
    },
    {
      "epoch": 12.562893081761006,
      "grad_norm": 0.9504967927932739,
      "learning_rate": 1.8267049239554786e-05,
      "loss": 0.0326,
      "num_input_tokens_seen": 15649824,
      "step": 23970
    },
    {
      "epoch": 12.565513626834381,
      "grad_norm": 2.5113425254821777,
      "learning_rate": 1.8256038141245912e-05,
      "loss": 0.0548,
      "num_input_tokens_seen": 15652992,
      "step": 23975
    },
    {
      "epoch": 12.568134171907756,
      "grad_norm": 14.873490333557129,
      "learning_rate": 1.824502845369693e-05,
      "loss": 0.038,
      "num_input_tokens_seen": 15655680,
      "step": 23980
    },
    {
      "epoch": 12.570754716981131,
      "grad_norm": 1.8457932472229004,
      "learning_rate": 1.8234020179210938e-05,
      "loss": 0.025,
      "num_input_tokens_seen": 15658496,
      "step": 23985
    },
    {
      "epoch": 12.573375262054507,
      "grad_norm": 1.2240824699401855,
      "learning_rate": 1.8223013320090715e-05,
      "loss": 0.079,
      "num_input_tokens_seen": 15661248,
      "step": 23990
    },
    {
      "epoch": 12.575995807127882,
      "grad_norm": 2.824542284011841,
      "learning_rate": 1.82120078786388e-05,
      "loss": 0.0735,
      "num_input_tokens_seen": 15663840,
      "step": 23995
    },
    {
      "epoch": 12.578616352201259,
      "grad_norm": 2.8796305656433105,
      "learning_rate": 1.820100385715739e-05,
      "loss": 0.0687,
      "num_input_tokens_seen": 15667744,
      "step": 24000
    },
    {
      "epoch": 12.581236897274634,
      "grad_norm": 2.6365485191345215,
      "learning_rate": 1.8190001257948414e-05,
      "loss": 0.0475,
      "num_input_tokens_seen": 15670592,
      "step": 24005
    },
    {
      "epoch": 12.583857442348009,
      "grad_norm": 0.994749903678894,
      "learning_rate": 1.8179000083313483e-05,
      "loss": 0.0435,
      "num_input_tokens_seen": 15674688,
      "step": 24010
    },
    {
      "epoch": 12.586477987421384,
      "grad_norm": 0.5492402911186218,
      "learning_rate": 1.8168000335553907e-05,
      "loss": 0.044,
      "num_input_tokens_seen": 15677440,
      "step": 24015
    },
    {
      "epoch": 12.58909853249476,
      "grad_norm": 5.633612155914307,
      "learning_rate": 1.8157002016970723e-05,
      "loss": 0.0643,
      "num_input_tokens_seen": 15680416,
      "step": 24020
    },
    {
      "epoch": 12.591719077568134,
      "grad_norm": 1.162410020828247,
      "learning_rate": 1.8146005129864644e-05,
      "loss": 0.031,
      "num_input_tokens_seen": 15684800,
      "step": 24025
    },
    {
      "epoch": 12.59433962264151,
      "grad_norm": 1.8817524909973145,
      "learning_rate": 1.81350096765361e-05,
      "loss": 0.088,
      "num_input_tokens_seen": 15687936,
      "step": 24030
    },
    {
      "epoch": 12.596960167714885,
      "grad_norm": 4.091881275177002,
      "learning_rate": 1.812401565928521e-05,
      "loss": 0.0381,
      "num_input_tokens_seen": 15691104,
      "step": 24035
    },
    {
      "epoch": 12.59958071278826,
      "grad_norm": 3.3525006771087646,
      "learning_rate": 1.8113023080411806e-05,
      "loss": 0.0596,
      "num_input_tokens_seen": 15694432,
      "step": 24040
    },
    {
      "epoch": 12.602201257861635,
      "grad_norm": 1.9270914793014526,
      "learning_rate": 1.810203194221539e-05,
      "loss": 0.0786,
      "num_input_tokens_seen": 15697504,
      "step": 24045
    },
    {
      "epoch": 12.60482180293501,
      "grad_norm": 2.7288360595703125,
      "learning_rate": 1.8091042246995206e-05,
      "loss": 0.0943,
      "num_input_tokens_seen": 15701824,
      "step": 24050
    },
    {
      "epoch": 12.607442348008385,
      "grad_norm": 2.171001672744751,
      "learning_rate": 1.8080053997050154e-05,
      "loss": 0.0556,
      "num_input_tokens_seen": 15704736,
      "step": 24055
    },
    {
      "epoch": 12.61006289308176,
      "grad_norm": 2.771547794342041,
      "learning_rate": 1.8069067194678867e-05,
      "loss": 0.0666,
      "num_input_tokens_seen": 15707744,
      "step": 24060
    },
    {
      "epoch": 12.612683438155136,
      "grad_norm": 2.197376251220703,
      "learning_rate": 1.8058081842179642e-05,
      "loss": 0.1032,
      "num_input_tokens_seen": 15710208,
      "step": 24065
    },
    {
      "epoch": 12.615303983228511,
      "grad_norm": 0.31904032826423645,
      "learning_rate": 1.804709794185051e-05,
      "loss": 0.0499,
      "num_input_tokens_seen": 15713792,
      "step": 24070
    },
    {
      "epoch": 12.617924528301886,
      "grad_norm": 2.3994691371917725,
      "learning_rate": 1.8036115495989164e-05,
      "loss": 0.0403,
      "num_input_tokens_seen": 15717792,
      "step": 24075
    },
    {
      "epoch": 12.620545073375261,
      "grad_norm": 2.4305059909820557,
      "learning_rate": 1.8025134506892997e-05,
      "loss": 0.0644,
      "num_input_tokens_seen": 15720896,
      "step": 24080
    },
    {
      "epoch": 12.623165618448636,
      "grad_norm": 2.0353925228118896,
      "learning_rate": 1.8014154976859126e-05,
      "loss": 0.0418,
      "num_input_tokens_seen": 15723904,
      "step": 24085
    },
    {
      "epoch": 12.625786163522012,
      "grad_norm": 3.9105403423309326,
      "learning_rate": 1.800317690818433e-05,
      "loss": 0.0347,
      "num_input_tokens_seen": 15727168,
      "step": 24090
    },
    {
      "epoch": 12.628406708595389,
      "grad_norm": 2.0899956226348877,
      "learning_rate": 1.799220030316511e-05,
      "loss": 0.0771,
      "num_input_tokens_seen": 15730688,
      "step": 24095
    },
    {
      "epoch": 12.631027253668764,
      "grad_norm": 1.967990756034851,
      "learning_rate": 1.7981225164097633e-05,
      "loss": 0.108,
      "num_input_tokens_seen": 15733664,
      "step": 24100
    },
    {
      "epoch": 12.633647798742139,
      "grad_norm": 2.1246814727783203,
      "learning_rate": 1.7970251493277774e-05,
      "loss": 0.0679,
      "num_input_tokens_seen": 15736352,
      "step": 24105
    },
    {
      "epoch": 12.636268343815514,
      "grad_norm": 0.8021544218063354,
      "learning_rate": 1.7959279293001102e-05,
      "loss": 0.0595,
      "num_input_tokens_seen": 15739648,
      "step": 24110
    },
    {
      "epoch": 12.63888888888889,
      "grad_norm": 1.8959406614303589,
      "learning_rate": 1.7948308565562876e-05,
      "loss": 0.0485,
      "num_input_tokens_seen": 15742464,
      "step": 24115
    },
    {
      "epoch": 12.641509433962264,
      "grad_norm": 3.198558807373047,
      "learning_rate": 1.793733931325805e-05,
      "loss": 0.0403,
      "num_input_tokens_seen": 15745984,
      "step": 24120
    },
    {
      "epoch": 12.64412997903564,
      "grad_norm": 0.33584117889404297,
      "learning_rate": 1.792637153838127e-05,
      "loss": 0.0603,
      "num_input_tokens_seen": 15748608,
      "step": 24125
    },
    {
      "epoch": 12.646750524109015,
      "grad_norm": 2.5842325687408447,
      "learning_rate": 1.7915405243226864e-05,
      "loss": 0.0776,
      "num_input_tokens_seen": 15751520,
      "step": 24130
    },
    {
      "epoch": 12.64937106918239,
      "grad_norm": 1.9176685810089111,
      "learning_rate": 1.7904440430088847e-05,
      "loss": 0.0468,
      "num_input_tokens_seen": 15754912,
      "step": 24135
    },
    {
      "epoch": 12.651991614255765,
      "grad_norm": 1.425389051437378,
      "learning_rate": 1.7893477101260947e-05,
      "loss": 0.0614,
      "num_input_tokens_seen": 15758688,
      "step": 24140
    },
    {
      "epoch": 12.65461215932914,
      "grad_norm": 0.877512514591217,
      "learning_rate": 1.7882515259036558e-05,
      "loss": 0.0213,
      "num_input_tokens_seen": 15761248,
      "step": 24145
    },
    {
      "epoch": 12.657232704402515,
      "grad_norm": 0.9159378409385681,
      "learning_rate": 1.787155490570878e-05,
      "loss": 0.083,
      "num_input_tokens_seen": 15765088,
      "step": 24150
    },
    {
      "epoch": 12.65985324947589,
      "grad_norm": 2.662672758102417,
      "learning_rate": 1.7860596043570398e-05,
      "loss": 0.0716,
      "num_input_tokens_seen": 15768896,
      "step": 24155
    },
    {
      "epoch": 12.662473794549266,
      "grad_norm": 1.0961027145385742,
      "learning_rate": 1.7849638674913863e-05,
      "loss": 0.0324,
      "num_input_tokens_seen": 15771584,
      "step": 24160
    },
    {
      "epoch": 12.665094339622641,
      "grad_norm": 5.790891647338867,
      "learning_rate": 1.7838682802031344e-05,
      "loss": 0.0885,
      "num_input_tokens_seen": 15774080,
      "step": 24165
    },
    {
      "epoch": 12.667714884696016,
      "grad_norm": 11.881088256835938,
      "learning_rate": 1.782772842721468e-05,
      "loss": 0.0957,
      "num_input_tokens_seen": 15777280,
      "step": 24170
    },
    {
      "epoch": 12.670335429769391,
      "grad_norm": 0.38075020909309387,
      "learning_rate": 1.7816775552755406e-05,
      "loss": 0.0755,
      "num_input_tokens_seen": 15780544,
      "step": 24175
    },
    {
      "epoch": 12.672955974842766,
      "grad_norm": 1.4033849239349365,
      "learning_rate": 1.780582418094473e-05,
      "loss": 0.0968,
      "num_input_tokens_seen": 15783680,
      "step": 24180
    },
    {
      "epoch": 12.675576519916142,
      "grad_norm": 7.102364540100098,
      "learning_rate": 1.7794874314073573e-05,
      "loss": 0.0395,
      "num_input_tokens_seen": 15786752,
      "step": 24185
    },
    {
      "epoch": 12.678197064989519,
      "grad_norm": 4.416104316711426,
      "learning_rate": 1.7783925954432503e-05,
      "loss": 0.0274,
      "num_input_tokens_seen": 15790368,
      "step": 24190
    },
    {
      "epoch": 12.680817610062894,
      "grad_norm": 2.316952705383301,
      "learning_rate": 1.777297910431179e-05,
      "loss": 0.0453,
      "num_input_tokens_seen": 15793120,
      "step": 24195
    },
    {
      "epoch": 12.683438155136269,
      "grad_norm": 1.1597594022750854,
      "learning_rate": 1.7762033766001402e-05,
      "loss": 0.0265,
      "num_input_tokens_seen": 15797600,
      "step": 24200
    },
    {
      "epoch": 12.686058700209644,
      "grad_norm": 1.3675286769866943,
      "learning_rate": 1.7751089941790977e-05,
      "loss": 0.0607,
      "num_input_tokens_seen": 15802304,
      "step": 24205
    },
    {
      "epoch": 12.68867924528302,
      "grad_norm": 2.5448672771453857,
      "learning_rate": 1.7740147633969844e-05,
      "loss": 0.0559,
      "num_input_tokens_seen": 15804672,
      "step": 24210
    },
    {
      "epoch": 12.691299790356394,
      "grad_norm": 2.735692024230957,
      "learning_rate": 1.772920684482699e-05,
      "loss": 0.0456,
      "num_input_tokens_seen": 15807168,
      "step": 24215
    },
    {
      "epoch": 12.69392033542977,
      "grad_norm": 2.2147791385650635,
      "learning_rate": 1.771826757665112e-05,
      "loss": 0.0305,
      "num_input_tokens_seen": 15809824,
      "step": 24220
    },
    {
      "epoch": 12.696540880503145,
      "grad_norm": 2.3277947902679443,
      "learning_rate": 1.770732983173059e-05,
      "loss": 0.095,
      "num_input_tokens_seen": 15813408,
      "step": 24225
    },
    {
      "epoch": 12.69916142557652,
      "grad_norm": 3.6031336784362793,
      "learning_rate": 1.7696393612353467e-05,
      "loss": 0.0908,
      "num_input_tokens_seen": 15816096,
      "step": 24230
    },
    {
      "epoch": 12.701781970649895,
      "grad_norm": 1.473151683807373,
      "learning_rate": 1.7685458920807474e-05,
      "loss": 0.0206,
      "num_input_tokens_seen": 15819328,
      "step": 24235
    },
    {
      "epoch": 12.70440251572327,
      "grad_norm": 9.114688873291016,
      "learning_rate": 1.7674525759380033e-05,
      "loss": 0.0439,
      "num_input_tokens_seen": 15822176,
      "step": 24240
    },
    {
      "epoch": 12.707023060796645,
      "grad_norm": 2.5251543521881104,
      "learning_rate": 1.766359413035823e-05,
      "loss": 0.0801,
      "num_input_tokens_seen": 15824768,
      "step": 24245
    },
    {
      "epoch": 12.70964360587002,
      "grad_norm": 1.800036907196045,
      "learning_rate": 1.7652664036028827e-05,
      "loss": 0.087,
      "num_input_tokens_seen": 15828608,
      "step": 24250
    },
    {
      "epoch": 12.712264150943396,
      "grad_norm": 1.288957953453064,
      "learning_rate": 1.7641735478678288e-05,
      "loss": 0.0776,
      "num_input_tokens_seen": 15830880,
      "step": 24255
    },
    {
      "epoch": 12.714884696016771,
      "grad_norm": 1.8092336654663086,
      "learning_rate": 1.763080846059274e-05,
      "loss": 0.0451,
      "num_input_tokens_seen": 15834112,
      "step": 24260
    },
    {
      "epoch": 12.717505241090146,
      "grad_norm": 1.5083413124084473,
      "learning_rate": 1.7619882984057987e-05,
      "loss": 0.0148,
      "num_input_tokens_seen": 15836832,
      "step": 24265
    },
    {
      "epoch": 12.720125786163521,
      "grad_norm": 2.0469911098480225,
      "learning_rate": 1.760895905135953e-05,
      "loss": 0.036,
      "num_input_tokens_seen": 15839840,
      "step": 24270
    },
    {
      "epoch": 12.722746331236896,
      "grad_norm": 1.5868321657180786,
      "learning_rate": 1.7598036664782508e-05,
      "loss": 0.0635,
      "num_input_tokens_seen": 15842848,
      "step": 24275
    },
    {
      "epoch": 12.725366876310272,
      "grad_norm": 2.5962376594543457,
      "learning_rate": 1.7587115826611766e-05,
      "loss": 0.042,
      "num_input_tokens_seen": 15846496,
      "step": 24280
    },
    {
      "epoch": 12.727987421383649,
      "grad_norm": 7.514969825744629,
      "learning_rate": 1.7576196539131828e-05,
      "loss": 0.0573,
      "num_input_tokens_seen": 15852832,
      "step": 24285
    },
    {
      "epoch": 12.730607966457024,
      "grad_norm": 0.8362910151481628,
      "learning_rate": 1.7565278804626872e-05,
      "loss": 0.0405,
      "num_input_tokens_seen": 15856416,
      "step": 24290
    },
    {
      "epoch": 12.733228511530399,
      "grad_norm": 2.248537540435791,
      "learning_rate": 1.755436262538078e-05,
      "loss": 0.0566,
      "num_input_tokens_seen": 15859104,
      "step": 24295
    },
    {
      "epoch": 12.735849056603774,
      "grad_norm": 1.8909951448440552,
      "learning_rate": 1.7543448003677082e-05,
      "loss": 0.0481,
      "num_input_tokens_seen": 15861792,
      "step": 24300
    },
    {
      "epoch": 12.73846960167715,
      "grad_norm": 0.4033501446247101,
      "learning_rate": 1.7532534941798985e-05,
      "loss": 0.0352,
      "num_input_tokens_seen": 15864320,
      "step": 24305
    },
    {
      "epoch": 12.741090146750524,
      "grad_norm": 1.5084532499313354,
      "learning_rate": 1.752162344202939e-05,
      "loss": 0.0379,
      "num_input_tokens_seen": 15867168,
      "step": 24310
    },
    {
      "epoch": 12.7437106918239,
      "grad_norm": 4.2958292961120605,
      "learning_rate": 1.751071350665085e-05,
      "loss": 0.0403,
      "num_input_tokens_seen": 15870048,
      "step": 24315
    },
    {
      "epoch": 12.746331236897275,
      "grad_norm": 3.5135793685913086,
      "learning_rate": 1.7499805137945602e-05,
      "loss": 0.039,
      "num_input_tokens_seen": 15873024,
      "step": 24320
    },
    {
      "epoch": 12.74895178197065,
      "grad_norm": 3.040804862976074,
      "learning_rate": 1.7488898338195566e-05,
      "loss": 0.0421,
      "num_input_tokens_seen": 15875648,
      "step": 24325
    },
    {
      "epoch": 12.751572327044025,
      "grad_norm": 2.0898869037628174,
      "learning_rate": 1.7477993109682288e-05,
      "loss": 0.0307,
      "num_input_tokens_seen": 15878240,
      "step": 24330
    },
    {
      "epoch": 12.7541928721174,
      "grad_norm": 3.4645092487335205,
      "learning_rate": 1.7467089454687043e-05,
      "loss": 0.074,
      "num_input_tokens_seen": 15881152,
      "step": 24335
    },
    {
      "epoch": 12.756813417190775,
      "grad_norm": 6.796135902404785,
      "learning_rate": 1.745618737549074e-05,
      "loss": 0.0599,
      "num_input_tokens_seen": 15884384,
      "step": 24340
    },
    {
      "epoch": 12.75943396226415,
      "grad_norm": 2.654449701309204,
      "learning_rate": 1.7445286874373974e-05,
      "loss": 0.0557,
      "num_input_tokens_seen": 15887232,
      "step": 24345
    },
    {
      "epoch": 12.762054507337526,
      "grad_norm": 0.27786844968795776,
      "learning_rate": 1.743438795361701e-05,
      "loss": 0.0371,
      "num_input_tokens_seen": 15889344,
      "step": 24350
    },
    {
      "epoch": 12.764675052410901,
      "grad_norm": 2.385789394378662,
      "learning_rate": 1.742349061549978e-05,
      "loss": 0.0638,
      "num_input_tokens_seen": 15892704,
      "step": 24355
    },
    {
      "epoch": 12.767295597484276,
      "grad_norm": 2.3157880306243896,
      "learning_rate": 1.741259486230186e-05,
      "loss": 0.2528,
      "num_input_tokens_seen": 15895840,
      "step": 24360
    },
    {
      "epoch": 12.769916142557651,
      "grad_norm": 4.847455024719238,
      "learning_rate": 1.7401700696302546e-05,
      "loss": 0.0714,
      "num_input_tokens_seen": 15898720,
      "step": 24365
    },
    {
      "epoch": 12.772536687631026,
      "grad_norm": 1.154628038406372,
      "learning_rate": 1.739080811978075e-05,
      "loss": 0.0249,
      "num_input_tokens_seen": 15901408,
      "step": 24370
    },
    {
      "epoch": 12.775157232704402,
      "grad_norm": 2.963916778564453,
      "learning_rate": 1.7379917135015093e-05,
      "loss": 0.0384,
      "num_input_tokens_seen": 15904160,
      "step": 24375
    },
    {
      "epoch": 12.777777777777779,
      "grad_norm": 1.2744455337524414,
      "learning_rate": 1.736902774428384e-05,
      "loss": 0.0408,
      "num_input_tokens_seen": 15907488,
      "step": 24380
    },
    {
      "epoch": 12.780398322851154,
      "grad_norm": 1.104005217552185,
      "learning_rate": 1.735813994986491e-05,
      "loss": 0.0333,
      "num_input_tokens_seen": 15910784,
      "step": 24385
    },
    {
      "epoch": 12.783018867924529,
      "grad_norm": 1.9676319360733032,
      "learning_rate": 1.7347253754035933e-05,
      "loss": 0.0687,
      "num_input_tokens_seen": 15914016,
      "step": 24390
    },
    {
      "epoch": 12.785639412997904,
      "grad_norm": 1.794409990310669,
      "learning_rate": 1.7336369159074154e-05,
      "loss": 0.0569,
      "num_input_tokens_seen": 15916672,
      "step": 24395
    },
    {
      "epoch": 12.78825995807128,
      "grad_norm": 1.6381560564041138,
      "learning_rate": 1.7325486167256522e-05,
      "loss": 0.0508,
      "num_input_tokens_seen": 15920288,
      "step": 24400
    },
    {
      "epoch": 12.790880503144654,
      "grad_norm": 2.0917742252349854,
      "learning_rate": 1.7314604780859623e-05,
      "loss": 0.0792,
      "num_input_tokens_seen": 15924736,
      "step": 24405
    },
    {
      "epoch": 12.79350104821803,
      "grad_norm": 1.5265761613845825,
      "learning_rate": 1.730372500215973e-05,
      "loss": 0.0239,
      "num_input_tokens_seen": 15928544,
      "step": 24410
    },
    {
      "epoch": 12.796121593291405,
      "grad_norm": 1.1187617778778076,
      "learning_rate": 1.7292846833432763e-05,
      "loss": 0.0325,
      "num_input_tokens_seen": 15932864,
      "step": 24415
    },
    {
      "epoch": 12.79874213836478,
      "grad_norm": 1.2674742937088013,
      "learning_rate": 1.7281970276954305e-05,
      "loss": 0.0351,
      "num_input_tokens_seen": 15936512,
      "step": 24420
    },
    {
      "epoch": 12.801362683438155,
      "grad_norm": 2.6697206497192383,
      "learning_rate": 1.727109533499961e-05,
      "loss": 0.0419,
      "num_input_tokens_seen": 15939392,
      "step": 24425
    },
    {
      "epoch": 12.80398322851153,
      "grad_norm": 3.0973737239837646,
      "learning_rate": 1.72602220098436e-05,
      "loss": 0.0529,
      "num_input_tokens_seen": 15942272,
      "step": 24430
    },
    {
      "epoch": 12.806603773584905,
      "grad_norm": 5.719843864440918,
      "learning_rate": 1.7249350303760846e-05,
      "loss": 0.0412,
      "num_input_tokens_seen": 15945280,
      "step": 24435
    },
    {
      "epoch": 12.80922431865828,
      "grad_norm": 1.562102198600769,
      "learning_rate": 1.7238480219025593e-05,
      "loss": 0.0633,
      "num_input_tokens_seen": 15948704,
      "step": 24440
    },
    {
      "epoch": 12.811844863731656,
      "grad_norm": 1.8730601072311401,
      "learning_rate": 1.7227611757911723e-05,
      "loss": 0.0551,
      "num_input_tokens_seen": 15952320,
      "step": 24445
    },
    {
      "epoch": 12.814465408805031,
      "grad_norm": 1.8310829401016235,
      "learning_rate": 1.7216744922692794e-05,
      "loss": 0.0288,
      "num_input_tokens_seen": 15956576,
      "step": 24450
    },
    {
      "epoch": 12.817085953878406,
      "grad_norm": 2.1208715438842773,
      "learning_rate": 1.7205879715642037e-05,
      "loss": 0.0493,
      "num_input_tokens_seen": 15959232,
      "step": 24455
    },
    {
      "epoch": 12.819706498951781,
      "grad_norm": 1.411087155342102,
      "learning_rate": 1.719501613903232e-05,
      "loss": 0.0416,
      "num_input_tokens_seen": 15962720,
      "step": 24460
    },
    {
      "epoch": 12.822327044025156,
      "grad_norm": 0.9193647503852844,
      "learning_rate": 1.7184154195136192e-05,
      "loss": 0.0623,
      "num_input_tokens_seen": 15966240,
      "step": 24465
    },
    {
      "epoch": 12.824947589098532,
      "grad_norm": 1.8756825923919678,
      "learning_rate": 1.7173293886225837e-05,
      "loss": 0.0366,
      "num_input_tokens_seen": 15968800,
      "step": 24470
    },
    {
      "epoch": 12.827568134171909,
      "grad_norm": 3.5461738109588623,
      "learning_rate": 1.71624352145731e-05,
      "loss": 0.063,
      "num_input_tokens_seen": 15971680,
      "step": 24475
    },
    {
      "epoch": 12.830188679245284,
      "grad_norm": 2.4237637519836426,
      "learning_rate": 1.7151578182449508e-05,
      "loss": 0.081,
      "num_input_tokens_seen": 15975552,
      "step": 24480
    },
    {
      "epoch": 12.832809224318659,
      "grad_norm": 5.613265514373779,
      "learning_rate": 1.7140722792126212e-05,
      "loss": 0.032,
      "num_input_tokens_seen": 15978656,
      "step": 24485
    },
    {
      "epoch": 12.835429769392034,
      "grad_norm": 2.9653406143188477,
      "learning_rate": 1.7129869045874053e-05,
      "loss": 0.0452,
      "num_input_tokens_seen": 15981856,
      "step": 24490
    },
    {
      "epoch": 12.83805031446541,
      "grad_norm": 2.4325873851776123,
      "learning_rate": 1.7119016945963506e-05,
      "loss": 0.0622,
      "num_input_tokens_seen": 15984736,
      "step": 24495
    },
    {
      "epoch": 12.840670859538784,
      "grad_norm": 1.8342853784561157,
      "learning_rate": 1.7108166494664694e-05,
      "loss": 0.0444,
      "num_input_tokens_seen": 15987424,
      "step": 24500
    },
    {
      "epoch": 12.84329140461216,
      "grad_norm": 5.292817115783691,
      "learning_rate": 1.7097317694247417e-05,
      "loss": 0.1763,
      "num_input_tokens_seen": 15991296,
      "step": 24505
    },
    {
      "epoch": 12.845911949685535,
      "grad_norm": 2.258321523666382,
      "learning_rate": 1.7086470546981122e-05,
      "loss": 0.1095,
      "num_input_tokens_seen": 15993760,
      "step": 24510
    },
    {
      "epoch": 12.84853249475891,
      "grad_norm": 2.067310333251953,
      "learning_rate": 1.7075625055134904e-05,
      "loss": 0.0194,
      "num_input_tokens_seen": 15996288,
      "step": 24515
    },
    {
      "epoch": 12.851153039832285,
      "grad_norm": 3.350414276123047,
      "learning_rate": 1.706478122097752e-05,
      "loss": 0.0739,
      "num_input_tokens_seen": 15999328,
      "step": 24520
    },
    {
      "epoch": 12.85377358490566,
      "grad_norm": 4.046437740325928,
      "learning_rate": 1.7053939046777384e-05,
      "loss": 0.0468,
      "num_input_tokens_seen": 16002336,
      "step": 24525
    },
    {
      "epoch": 12.856394129979035,
      "grad_norm": 1.6501861810684204,
      "learning_rate": 1.7043098534802533e-05,
      "loss": 0.0417,
      "num_input_tokens_seen": 16005440,
      "step": 24530
    },
    {
      "epoch": 12.85901467505241,
      "grad_norm": 6.914071559906006,
      "learning_rate": 1.7032259687320695e-05,
      "loss": 0.0718,
      "num_input_tokens_seen": 16009376,
      "step": 24535
    },
    {
      "epoch": 12.861635220125786,
      "grad_norm": 2.7966055870056152,
      "learning_rate": 1.7021422506599228e-05,
      "loss": 0.0462,
      "num_input_tokens_seen": 16012768,
      "step": 24540
    },
    {
      "epoch": 12.864255765199161,
      "grad_norm": 2.186540126800537,
      "learning_rate": 1.7010586994905155e-05,
      "loss": 0.0582,
      "num_input_tokens_seen": 16015136,
      "step": 24545
    },
    {
      "epoch": 12.866876310272536,
      "grad_norm": 1.786733627319336,
      "learning_rate": 1.6999753154505123e-05,
      "loss": 0.0779,
      "num_input_tokens_seen": 16018688,
      "step": 24550
    },
    {
      "epoch": 12.869496855345911,
      "grad_norm": 1.068780541419983,
      "learning_rate": 1.6988920987665476e-05,
      "loss": 0.08,
      "num_input_tokens_seen": 16021504,
      "step": 24555
    },
    {
      "epoch": 12.872117400419286,
      "grad_norm": 2.0662457942962646,
      "learning_rate": 1.6978090496652156e-05,
      "loss": 0.0596,
      "num_input_tokens_seen": 16024736,
      "step": 24560
    },
    {
      "epoch": 12.874737945492662,
      "grad_norm": 1.8330496549606323,
      "learning_rate": 1.6967261683730778e-05,
      "loss": 0.0375,
      "num_input_tokens_seen": 16027296,
      "step": 24565
    },
    {
      "epoch": 12.877358490566039,
      "grad_norm": 1.230633020401001,
      "learning_rate": 1.695643455116662e-05,
      "loss": 0.024,
      "num_input_tokens_seen": 16029920,
      "step": 24570
    },
    {
      "epoch": 12.879979035639414,
      "grad_norm": 2.2939224243164062,
      "learning_rate": 1.694560910122458e-05,
      "loss": 0.0411,
      "num_input_tokens_seen": 16032672,
      "step": 24575
    },
    {
      "epoch": 12.882599580712789,
      "grad_norm": 5.497593879699707,
      "learning_rate": 1.693478533616924e-05,
      "loss": 0.0242,
      "num_input_tokens_seen": 16035040,
      "step": 24580
    },
    {
      "epoch": 12.885220125786164,
      "grad_norm": 0.838363766670227,
      "learning_rate": 1.692396325826479e-05,
      "loss": 0.0451,
      "num_input_tokens_seen": 16040064,
      "step": 24585
    },
    {
      "epoch": 12.88784067085954,
      "grad_norm": 2.4818191528320312,
      "learning_rate": 1.6913142869775088e-05,
      "loss": 0.0513,
      "num_input_tokens_seen": 16042784,
      "step": 24590
    },
    {
      "epoch": 12.890461215932914,
      "grad_norm": 1.8766090869903564,
      "learning_rate": 1.6902324172963635e-05,
      "loss": 0.054,
      "num_input_tokens_seen": 16045952,
      "step": 24595
    },
    {
      "epoch": 12.89308176100629,
      "grad_norm": 4.418149471282959,
      "learning_rate": 1.6891507170093586e-05,
      "loss": 0.0522,
      "num_input_tokens_seen": 16049248,
      "step": 24600
    },
    {
      "epoch": 12.895702306079665,
      "grad_norm": 1.3879903554916382,
      "learning_rate": 1.6880691863427725e-05,
      "loss": 0.0628,
      "num_input_tokens_seen": 16052576,
      "step": 24605
    },
    {
      "epoch": 12.89832285115304,
      "grad_norm": 1.167816400527954,
      "learning_rate": 1.686987825522851e-05,
      "loss": 0.2235,
      "num_input_tokens_seen": 16055936,
      "step": 24610
    },
    {
      "epoch": 12.900943396226415,
      "grad_norm": 6.002771377563477,
      "learning_rate": 1.6859066347758002e-05,
      "loss": 0.0715,
      "num_input_tokens_seen": 16058176,
      "step": 24615
    },
    {
      "epoch": 12.90356394129979,
      "grad_norm": 1.6286605596542358,
      "learning_rate": 1.6848256143277933e-05,
      "loss": 0.0719,
      "num_input_tokens_seen": 16061568,
      "step": 24620
    },
    {
      "epoch": 12.906184486373165,
      "grad_norm": 2.0254459381103516,
      "learning_rate": 1.683744764404968e-05,
      "loss": 0.0843,
      "num_input_tokens_seen": 16065536,
      "step": 24625
    },
    {
      "epoch": 12.90880503144654,
      "grad_norm": 3.688796043395996,
      "learning_rate": 1.6826640852334257e-05,
      "loss": 0.0429,
      "num_input_tokens_seen": 16068192,
      "step": 24630
    },
    {
      "epoch": 12.911425576519916,
      "grad_norm": 1.6965994834899902,
      "learning_rate": 1.6815835770392324e-05,
      "loss": 0.0635,
      "num_input_tokens_seen": 16071008,
      "step": 24635
    },
    {
      "epoch": 12.914046121593291,
      "grad_norm": 3.699657917022705,
      "learning_rate": 1.6805032400484184e-05,
      "loss": 0.0725,
      "num_input_tokens_seen": 16074688,
      "step": 24640
    },
    {
      "epoch": 12.916666666666666,
      "grad_norm": 2.5085978507995605,
      "learning_rate": 1.6794230744869767e-05,
      "loss": 0.0404,
      "num_input_tokens_seen": 16076928,
      "step": 24645
    },
    {
      "epoch": 12.919287211740041,
      "grad_norm": 1.1539255380630493,
      "learning_rate": 1.6783430805808658e-05,
      "loss": 0.1176,
      "num_input_tokens_seen": 16079904,
      "step": 24650
    },
    {
      "epoch": 12.921907756813416,
      "grad_norm": 2.0363481044769287,
      "learning_rate": 1.6772632585560084e-05,
      "loss": 0.0298,
      "num_input_tokens_seen": 16083168,
      "step": 24655
    },
    {
      "epoch": 12.924528301886792,
      "grad_norm": 1.914652705192566,
      "learning_rate": 1.6761836086382908e-05,
      "loss": 0.0673,
      "num_input_tokens_seen": 16086592,
      "step": 24660
    },
    {
      "epoch": 12.927148846960169,
      "grad_norm": 2.7534866333007812,
      "learning_rate": 1.6751041310535653e-05,
      "loss": 0.0704,
      "num_input_tokens_seen": 16089088,
      "step": 24665
    },
    {
      "epoch": 12.929769392033544,
      "grad_norm": 2.0080792903900146,
      "learning_rate": 1.6740248260276437e-05,
      "loss": 0.1476,
      "num_input_tokens_seen": 16091936,
      "step": 24670
    },
    {
      "epoch": 12.932389937106919,
      "grad_norm": 5.200592994689941,
      "learning_rate": 1.6729456937863047e-05,
      "loss": 0.0567,
      "num_input_tokens_seen": 16094336,
      "step": 24675
    },
    {
      "epoch": 12.935010482180294,
      "grad_norm": 4.31965970993042,
      "learning_rate": 1.6718667345552918e-05,
      "loss": 0.0574,
      "num_input_tokens_seen": 16098304,
      "step": 24680
    },
    {
      "epoch": 12.93763102725367,
      "grad_norm": 2.3914666175842285,
      "learning_rate": 1.6707879485603097e-05,
      "loss": 0.0468,
      "num_input_tokens_seen": 16101632,
      "step": 24685
    },
    {
      "epoch": 12.940251572327044,
      "grad_norm": 2.820580244064331,
      "learning_rate": 1.669709336027029e-05,
      "loss": 0.0218,
      "num_input_tokens_seen": 16105024,
      "step": 24690
    },
    {
      "epoch": 12.94287211740042,
      "grad_norm": 1.6586987972259521,
      "learning_rate": 1.6686308971810833e-05,
      "loss": 0.0714,
      "num_input_tokens_seen": 16108096,
      "step": 24695
    },
    {
      "epoch": 12.945492662473795,
      "grad_norm": 2.58920955657959,
      "learning_rate": 1.6675526322480677e-05,
      "loss": 0.0487,
      "num_input_tokens_seen": 16111648,
      "step": 24700
    },
    {
      "epoch": 12.94811320754717,
      "grad_norm": 2.9761555194854736,
      "learning_rate": 1.6664745414535454e-05,
      "loss": 0.086,
      "num_input_tokens_seen": 16114752,
      "step": 24705
    },
    {
      "epoch": 12.950733752620545,
      "grad_norm": 0.613891065120697,
      "learning_rate": 1.665396625023039e-05,
      "loss": 0.0946,
      "num_input_tokens_seen": 16119616,
      "step": 24710
    },
    {
      "epoch": 12.95335429769392,
      "grad_norm": 2.1948468685150146,
      "learning_rate": 1.6643188831820375e-05,
      "loss": 0.0854,
      "num_input_tokens_seen": 16123232,
      "step": 24715
    },
    {
      "epoch": 12.955974842767295,
      "grad_norm": 3.2086539268493652,
      "learning_rate": 1.6632413161559912e-05,
      "loss": 0.0695,
      "num_input_tokens_seen": 16127552,
      "step": 24720
    },
    {
      "epoch": 12.95859538784067,
      "grad_norm": 1.9747101068496704,
      "learning_rate": 1.6621639241703162e-05,
      "loss": 0.0539,
      "num_input_tokens_seen": 16130368,
      "step": 24725
    },
    {
      "epoch": 12.961215932914046,
      "grad_norm": 0.8997452855110168,
      "learning_rate": 1.6610867074503893e-05,
      "loss": 0.0421,
      "num_input_tokens_seen": 16134208,
      "step": 24730
    },
    {
      "epoch": 12.963836477987421,
      "grad_norm": 1.5187002420425415,
      "learning_rate": 1.660009666221552e-05,
      "loss": 0.0613,
      "num_input_tokens_seen": 16137120,
      "step": 24735
    },
    {
      "epoch": 12.966457023060796,
      "grad_norm": 1.5053592920303345,
      "learning_rate": 1.6589328007091093e-05,
      "loss": 0.0818,
      "num_input_tokens_seen": 16140576,
      "step": 24740
    },
    {
      "epoch": 12.969077568134171,
      "grad_norm": 2.6702730655670166,
      "learning_rate": 1.65785611113833e-05,
      "loss": 0.0605,
      "num_input_tokens_seen": 16143808,
      "step": 24745
    },
    {
      "epoch": 12.971698113207546,
      "grad_norm": 2.421015977859497,
      "learning_rate": 1.6567795977344447e-05,
      "loss": 0.0444,
      "num_input_tokens_seen": 16147136,
      "step": 24750
    },
    {
      "epoch": 12.974318658280922,
      "grad_norm": 1.5306161642074585,
      "learning_rate": 1.6557032607226466e-05,
      "loss": 0.0421,
      "num_input_tokens_seen": 16150272,
      "step": 24755
    },
    {
      "epoch": 12.976939203354299,
      "grad_norm": 2.678931951522827,
      "learning_rate": 1.654627100328095e-05,
      "loss": 0.0623,
      "num_input_tokens_seen": 16153920,
      "step": 24760
    },
    {
      "epoch": 12.979559748427674,
      "grad_norm": 1.2012640237808228,
      "learning_rate": 1.6535511167759085e-05,
      "loss": 0.0305,
      "num_input_tokens_seen": 16157280,
      "step": 24765
    },
    {
      "epoch": 12.982180293501049,
      "grad_norm": 3.1400146484375,
      "learning_rate": 1.652475310291172e-05,
      "loss": 0.0651,
      "num_input_tokens_seen": 16161056,
      "step": 24770
    },
    {
      "epoch": 12.984800838574424,
      "grad_norm": 2.1816248893737793,
      "learning_rate": 1.6513996810989317e-05,
      "loss": 0.0355,
      "num_input_tokens_seen": 16163776,
      "step": 24775
    },
    {
      "epoch": 12.9874213836478,
      "grad_norm": 4.243335723876953,
      "learning_rate": 1.6503242294241973e-05,
      "loss": 0.048,
      "num_input_tokens_seen": 16166176,
      "step": 24780
    },
    {
      "epoch": 12.990041928721174,
      "grad_norm": 6.321218490600586,
      "learning_rate": 1.6492489554919404e-05,
      "loss": 0.037,
      "num_input_tokens_seen": 16168704,
      "step": 24785
    },
    {
      "epoch": 12.99266247379455,
      "grad_norm": 1.0203107595443726,
      "learning_rate": 1.6481738595270953e-05,
      "loss": 0.0448,
      "num_input_tokens_seen": 16171584,
      "step": 24790
    },
    {
      "epoch": 12.995283018867925,
      "grad_norm": 9.053261756896973,
      "learning_rate": 1.6470989417545608e-05,
      "loss": 0.0615,
      "num_input_tokens_seen": 16174656,
      "step": 24795
    },
    {
      "epoch": 12.9979035639413,
      "grad_norm": 2.9405741691589355,
      "learning_rate": 1.646024202399197e-05,
      "loss": 0.0529,
      "num_input_tokens_seen": 16177344,
      "step": 24800
    },
    {
      "epoch": 13.0,
      "eval_loss": 1.1340941190719604,
      "eval_runtime": 12.0173,
      "eval_samples_per_second": 70.565,
      "eval_steps_per_second": 17.641,
      "num_input_tokens_seen": 16179624,
      "step": 24804
    },
    {
      "epoch": 13.000524109014675,
      "grad_norm": 0.7598346471786499,
      "learning_rate": 1.6449496416858284e-05,
      "loss": 0.0625,
      "num_input_tokens_seen": 16180040,
      "step": 24805
    },
    {
      "epoch": 13.00314465408805,
      "grad_norm": 1.903404712677002,
      "learning_rate": 1.64387525983924e-05,
      "loss": 0.0468,
      "num_input_tokens_seen": 16183176,
      "step": 24810
    },
    {
      "epoch": 13.005765199161425,
      "grad_norm": 1.2867375612258911,
      "learning_rate": 1.6428010570841788e-05,
      "loss": 0.0183,
      "num_input_tokens_seen": 16185928,
      "step": 24815
    },
    {
      "epoch": 13.0083857442348,
      "grad_norm": 0.8603997230529785,
      "learning_rate": 1.6417270336453573e-05,
      "loss": 0.0548,
      "num_input_tokens_seen": 16189160,
      "step": 24820
    },
    {
      "epoch": 13.011006289308176,
      "grad_norm": 1.3666590452194214,
      "learning_rate": 1.6406531897474492e-05,
      "loss": 0.0211,
      "num_input_tokens_seen": 16192392,
      "step": 24825
    },
    {
      "epoch": 13.01362683438155,
      "grad_norm": 0.6413553357124329,
      "learning_rate": 1.639579525615089e-05,
      "loss": 0.0443,
      "num_input_tokens_seen": 16195752,
      "step": 24830
    },
    {
      "epoch": 13.016247379454926,
      "grad_norm": 1.8485468626022339,
      "learning_rate": 1.638506041472877e-05,
      "loss": 0.0238,
      "num_input_tokens_seen": 16199176,
      "step": 24835
    },
    {
      "epoch": 13.018867924528301,
      "grad_norm": 3.0590434074401855,
      "learning_rate": 1.6374327375453724e-05,
      "loss": 0.0163,
      "num_input_tokens_seen": 16201768,
      "step": 24840
    },
    {
      "epoch": 13.021488469601676,
      "grad_norm": 0.9982655048370361,
      "learning_rate": 1.6363596140570982e-05,
      "loss": 0.014,
      "num_input_tokens_seen": 16204680,
      "step": 24845
    },
    {
      "epoch": 13.024109014675052,
      "grad_norm": 2.206482410430908,
      "learning_rate": 1.6352866712325395e-05,
      "loss": 0.0201,
      "num_input_tokens_seen": 16208264,
      "step": 24850
    },
    {
      "epoch": 13.026729559748428,
      "grad_norm": 2.1365530490875244,
      "learning_rate": 1.634213909296144e-05,
      "loss": 0.0164,
      "num_input_tokens_seen": 16210536,
      "step": 24855
    },
    {
      "epoch": 13.029350104821804,
      "grad_norm": 3.3933966159820557,
      "learning_rate": 1.6331413284723217e-05,
      "loss": 0.1001,
      "num_input_tokens_seen": 16214376,
      "step": 24860
    },
    {
      "epoch": 13.031970649895179,
      "grad_norm": 0.7685306668281555,
      "learning_rate": 1.6320689289854442e-05,
      "loss": 0.0274,
      "num_input_tokens_seen": 16217416,
      "step": 24865
    },
    {
      "epoch": 13.034591194968554,
      "grad_norm": 2.6227219104766846,
      "learning_rate": 1.6309967110598438e-05,
      "loss": 0.0342,
      "num_input_tokens_seen": 16220584,
      "step": 24870
    },
    {
      "epoch": 13.03721174004193,
      "grad_norm": 1.7104334831237793,
      "learning_rate": 1.6299246749198183e-05,
      "loss": 0.0224,
      "num_input_tokens_seen": 16224040,
      "step": 24875
    },
    {
      "epoch": 13.039832285115304,
      "grad_norm": 6.270849704742432,
      "learning_rate": 1.6288528207896236e-05,
      "loss": 0.0457,
      "num_input_tokens_seen": 16227016,
      "step": 24880
    },
    {
      "epoch": 13.04245283018868,
      "grad_norm": 6.053859710693359,
      "learning_rate": 1.627781148893481e-05,
      "loss": 0.0413,
      "num_input_tokens_seen": 16229768,
      "step": 24885
    },
    {
      "epoch": 13.045073375262055,
      "grad_norm": 0.8719395995140076,
      "learning_rate": 1.6267096594555703e-05,
      "loss": 0.0165,
      "num_input_tokens_seen": 16233224,
      "step": 24890
    },
    {
      "epoch": 13.04769392033543,
      "grad_norm": 1.923399806022644,
      "learning_rate": 1.6256383527000372e-05,
      "loss": 0.0236,
      "num_input_tokens_seen": 16236264,
      "step": 24895
    },
    {
      "epoch": 13.050314465408805,
      "grad_norm": 0.8466050028800964,
      "learning_rate": 1.6245672288509843e-05,
      "loss": 0.0132,
      "num_input_tokens_seen": 16238984,
      "step": 24900
    },
    {
      "epoch": 13.05293501048218,
      "grad_norm": 1.6091065406799316,
      "learning_rate": 1.62349628813248e-05,
      "loss": 0.0151,
      "num_input_tokens_seen": 16241608,
      "step": 24905
    },
    {
      "epoch": 13.055555555555555,
      "grad_norm": 0.806632936000824,
      "learning_rate": 1.6224255307685525e-05,
      "loss": 0.0174,
      "num_input_tokens_seen": 16244328,
      "step": 24910
    },
    {
      "epoch": 13.05817610062893,
      "grad_norm": 1.16786527633667,
      "learning_rate": 1.6213549569831922e-05,
      "loss": 0.0468,
      "num_input_tokens_seen": 16247912,
      "step": 24915
    },
    {
      "epoch": 13.060796645702306,
      "grad_norm": 1.5025227069854736,
      "learning_rate": 1.62028456700035e-05,
      "loss": 0.0509,
      "num_input_tokens_seen": 16250920,
      "step": 24920
    },
    {
      "epoch": 13.06341719077568,
      "grad_norm": 2.1814284324645996,
      "learning_rate": 1.619214361043942e-05,
      "loss": 0.0518,
      "num_input_tokens_seen": 16254728,
      "step": 24925
    },
    {
      "epoch": 13.066037735849056,
      "grad_norm": 3.7287204265594482,
      "learning_rate": 1.61814433933784e-05,
      "loss": 0.0221,
      "num_input_tokens_seen": 16257640,
      "step": 24930
    },
    {
      "epoch": 13.068658280922431,
      "grad_norm": 1.4551249742507935,
      "learning_rate": 1.617074502105881e-05,
      "loss": 0.0285,
      "num_input_tokens_seen": 16260808,
      "step": 24935
    },
    {
      "epoch": 13.071278825995806,
      "grad_norm": 1.2675420045852661,
      "learning_rate": 1.616004849571864e-05,
      "loss": 0.0206,
      "num_input_tokens_seen": 16264264,
      "step": 24940
    },
    {
      "epoch": 13.073899371069182,
      "grad_norm": 1.539682388305664,
      "learning_rate": 1.6149353819595463e-05,
      "loss": 0.0267,
      "num_input_tokens_seen": 16267528,
      "step": 24945
    },
    {
      "epoch": 13.076519916142558,
      "grad_norm": 3.0428857803344727,
      "learning_rate": 1.6138660994926508e-05,
      "loss": 0.0424,
      "num_input_tokens_seen": 16270088,
      "step": 24950
    },
    {
      "epoch": 13.079140461215934,
      "grad_norm": 4.345203399658203,
      "learning_rate": 1.612797002394857e-05,
      "loss": 0.0688,
      "num_input_tokens_seen": 16273320,
      "step": 24955
    },
    {
      "epoch": 13.081761006289309,
      "grad_norm": 0.14923667907714844,
      "learning_rate": 1.611728090889808e-05,
      "loss": 0.0118,
      "num_input_tokens_seen": 16276232,
      "step": 24960
    },
    {
      "epoch": 13.084381551362684,
      "grad_norm": 1.0449597835540771,
      "learning_rate": 1.610659365201109e-05,
      "loss": 0.0364,
      "num_input_tokens_seen": 16279272,
      "step": 24965
    },
    {
      "epoch": 13.08700209643606,
      "grad_norm": 1.5970213413238525,
      "learning_rate": 1.609590825552324e-05,
      "loss": 0.0111,
      "num_input_tokens_seen": 16282024,
      "step": 24970
    },
    {
      "epoch": 13.089622641509434,
      "grad_norm": 4.544630527496338,
      "learning_rate": 1.6085224721669807e-05,
      "loss": 0.0324,
      "num_input_tokens_seen": 16284808,
      "step": 24975
    },
    {
      "epoch": 13.09224318658281,
      "grad_norm": 1.261298656463623,
      "learning_rate": 1.607454305268566e-05,
      "loss": 0.0667,
      "num_input_tokens_seen": 16288200,
      "step": 24980
    },
    {
      "epoch": 13.094863731656185,
      "grad_norm": 1.1755309104919434,
      "learning_rate": 1.606386325080528e-05,
      "loss": 0.029,
      "num_input_tokens_seen": 16291176,
      "step": 24985
    },
    {
      "epoch": 13.09748427672956,
      "grad_norm": 13.660431861877441,
      "learning_rate": 1.6053185318262746e-05,
      "loss": 0.0311,
      "num_input_tokens_seen": 16293896,
      "step": 24990
    },
    {
      "epoch": 13.100104821802935,
      "grad_norm": 5.477999687194824,
      "learning_rate": 1.6042509257291783e-05,
      "loss": 0.0358,
      "num_input_tokens_seen": 16297192,
      "step": 24995
    },
    {
      "epoch": 13.10272536687631,
      "grad_norm": 0.9410223960876465,
      "learning_rate": 1.6031835070125685e-05,
      "loss": 0.0661,
      "num_input_tokens_seen": 16300680,
      "step": 25000
    },
    {
      "epoch": 13.105345911949685,
      "grad_norm": 0.2272917926311493,
      "learning_rate": 1.6021162758997382e-05,
      "loss": 0.0147,
      "num_input_tokens_seen": 16303080,
      "step": 25005
    },
    {
      "epoch": 13.10796645702306,
      "grad_norm": 1.8232743740081787,
      "learning_rate": 1.6010492326139398e-05,
      "loss": 0.0339,
      "num_input_tokens_seen": 16305224,
      "step": 25010
    },
    {
      "epoch": 13.110587002096436,
      "grad_norm": 2.4745993614196777,
      "learning_rate": 1.599982377378385e-05,
      "loss": 0.0544,
      "num_input_tokens_seen": 16308168,
      "step": 25015
    },
    {
      "epoch": 13.11320754716981,
      "grad_norm": 1.1027562618255615,
      "learning_rate": 1.5989157104162496e-05,
      "loss": 0.0338,
      "num_input_tokens_seen": 16312072,
      "step": 25020
    },
    {
      "epoch": 13.115828092243186,
      "grad_norm": 1.3006291389465332,
      "learning_rate": 1.5978492319506673e-05,
      "loss": 0.0369,
      "num_input_tokens_seen": 16315112,
      "step": 25025
    },
    {
      "epoch": 13.118448637316561,
      "grad_norm": 1.562187671661377,
      "learning_rate": 1.596782942204733e-05,
      "loss": 0.0283,
      "num_input_tokens_seen": 16317704,
      "step": 25030
    },
    {
      "epoch": 13.121069182389936,
      "grad_norm": 1.1057289838790894,
      "learning_rate": 1.595716841401504e-05,
      "loss": 0.0545,
      "num_input_tokens_seen": 16321288,
      "step": 25035
    },
    {
      "epoch": 13.123689727463312,
      "grad_norm": 2.411456346511841,
      "learning_rate": 1.594650929763994e-05,
      "loss": 0.0214,
      "num_input_tokens_seen": 16324360,
      "step": 25040
    },
    {
      "epoch": 13.126310272536688,
      "grad_norm": 3.1133406162261963,
      "learning_rate": 1.5935852075151812e-05,
      "loss": 0.0343,
      "num_input_tokens_seen": 16327784,
      "step": 25045
    },
    {
      "epoch": 13.128930817610064,
      "grad_norm": 0.6825483441352844,
      "learning_rate": 1.5925196748780016e-05,
      "loss": 0.0699,
      "num_input_tokens_seen": 16331848,
      "step": 25050
    },
    {
      "epoch": 13.131551362683439,
      "grad_norm": 3.8135452270507812,
      "learning_rate": 1.5914543320753528e-05,
      "loss": 0.0377,
      "num_input_tokens_seen": 16335592,
      "step": 25055
    },
    {
      "epoch": 13.134171907756814,
      "grad_norm": 1.8007125854492188,
      "learning_rate": 1.5903891793300932e-05,
      "loss": 0.0569,
      "num_input_tokens_seen": 16340200,
      "step": 25060
    },
    {
      "epoch": 13.13679245283019,
      "grad_norm": 0.8432220816612244,
      "learning_rate": 1.5893242168650403e-05,
      "loss": 0.025,
      "num_input_tokens_seen": 16343208,
      "step": 25065
    },
    {
      "epoch": 13.139412997903564,
      "grad_norm": 2.6342270374298096,
      "learning_rate": 1.5882594449029705e-05,
      "loss": 0.0317,
      "num_input_tokens_seen": 16346088,
      "step": 25070
    },
    {
      "epoch": 13.14203354297694,
      "grad_norm": 1.8349508047103882,
      "learning_rate": 1.587194863666624e-05,
      "loss": 0.0435,
      "num_input_tokens_seen": 16349768,
      "step": 25075
    },
    {
      "epoch": 13.144654088050315,
      "grad_norm": 3.2394890785217285,
      "learning_rate": 1.5861304733786974e-05,
      "loss": 0.0542,
      "num_input_tokens_seen": 16353960,
      "step": 25080
    },
    {
      "epoch": 13.14727463312369,
      "grad_norm": 3.8690476417541504,
      "learning_rate": 1.5850662742618505e-05,
      "loss": 0.0384,
      "num_input_tokens_seen": 16356904,
      "step": 25085
    },
    {
      "epoch": 13.149895178197065,
      "grad_norm": 0.6094176173210144,
      "learning_rate": 1.5840022665387004e-05,
      "loss": 0.0275,
      "num_input_tokens_seen": 16360040,
      "step": 25090
    },
    {
      "epoch": 13.15251572327044,
      "grad_norm": 0.20499497652053833,
      "learning_rate": 1.5829384504318268e-05,
      "loss": 0.0375,
      "num_input_tokens_seen": 16363496,
      "step": 25095
    },
    {
      "epoch": 13.155136268343815,
      "grad_norm": 2.3321399688720703,
      "learning_rate": 1.5818748261637662e-05,
      "loss": 0.0568,
      "num_input_tokens_seen": 16366376,
      "step": 25100
    },
    {
      "epoch": 13.15775681341719,
      "grad_norm": 1.9374651908874512,
      "learning_rate": 1.5808113939570173e-05,
      "loss": 0.0477,
      "num_input_tokens_seen": 16370184,
      "step": 25105
    },
    {
      "epoch": 13.160377358490566,
      "grad_norm": 1.9085791110992432,
      "learning_rate": 1.5797481540340386e-05,
      "loss": 0.0248,
      "num_input_tokens_seen": 16374152,
      "step": 25110
    },
    {
      "epoch": 13.16299790356394,
      "grad_norm": 1.9459229707717896,
      "learning_rate": 1.5786851066172466e-05,
      "loss": 0.0292,
      "num_input_tokens_seen": 16376904,
      "step": 25115
    },
    {
      "epoch": 13.165618448637316,
      "grad_norm": 1.1857717037200928,
      "learning_rate": 1.5776222519290208e-05,
      "loss": 0.0309,
      "num_input_tokens_seen": 16379784,
      "step": 25120
    },
    {
      "epoch": 13.168238993710691,
      "grad_norm": 1.0824190378189087,
      "learning_rate": 1.576559590191696e-05,
      "loss": 0.037,
      "num_input_tokens_seen": 16382888,
      "step": 25125
    },
    {
      "epoch": 13.170859538784066,
      "grad_norm": 0.4554799497127533,
      "learning_rate": 1.5754971216275692e-05,
      "loss": 0.0234,
      "num_input_tokens_seen": 16386472,
      "step": 25130
    },
    {
      "epoch": 13.173480083857442,
      "grad_norm": 0.5044985413551331,
      "learning_rate": 1.5744348464588975e-05,
      "loss": 0.0643,
      "num_input_tokens_seen": 16389608,
      "step": 25135
    },
    {
      "epoch": 13.176100628930818,
      "grad_norm": 3.522279739379883,
      "learning_rate": 1.5733727649078968e-05,
      "loss": 0.0319,
      "num_input_tokens_seen": 16393032,
      "step": 25140
    },
    {
      "epoch": 13.178721174004194,
      "grad_norm": 1.2224416732788086,
      "learning_rate": 1.5723108771967414e-05,
      "loss": 0.0253,
      "num_input_tokens_seen": 16396712,
      "step": 25145
    },
    {
      "epoch": 13.181341719077569,
      "grad_norm": 2.7157297134399414,
      "learning_rate": 1.5712491835475685e-05,
      "loss": 0.0267,
      "num_input_tokens_seen": 16399272,
      "step": 25150
    },
    {
      "epoch": 13.183962264150944,
      "grad_norm": 3.2016875743865967,
      "learning_rate": 1.5701876841824698e-05,
      "loss": 0.0252,
      "num_input_tokens_seen": 16402088,
      "step": 25155
    },
    {
      "epoch": 13.18658280922432,
      "grad_norm": 0.8241831064224243,
      "learning_rate": 1.5691263793234993e-05,
      "loss": 0.0457,
      "num_input_tokens_seen": 16406600,
      "step": 25160
    },
    {
      "epoch": 13.189203354297694,
      "grad_norm": 1.1790591478347778,
      "learning_rate": 1.5680652691926706e-05,
      "loss": 0.0202,
      "num_input_tokens_seen": 16409224,
      "step": 25165
    },
    {
      "epoch": 13.19182389937107,
      "grad_norm": 1.5729055404663086,
      "learning_rate": 1.5670043540119553e-05,
      "loss": 0.1344,
      "num_input_tokens_seen": 16412136,
      "step": 25170
    },
    {
      "epoch": 13.194444444444445,
      "grad_norm": 1.176261067390442,
      "learning_rate": 1.5659436340032855e-05,
      "loss": 0.019,
      "num_input_tokens_seen": 16416392,
      "step": 25175
    },
    {
      "epoch": 13.19706498951782,
      "grad_norm": 2.9187231063842773,
      "learning_rate": 1.5648831093885515e-05,
      "loss": 0.0302,
      "num_input_tokens_seen": 16419688,
      "step": 25180
    },
    {
      "epoch": 13.199685534591195,
      "grad_norm": 2.665745735168457,
      "learning_rate": 1.5638227803896015e-05,
      "loss": 0.0332,
      "num_input_tokens_seen": 16422856,
      "step": 25185
    },
    {
      "epoch": 13.20230607966457,
      "grad_norm": 0.470244437456131,
      "learning_rate": 1.562762647228246e-05,
      "loss": 0.0487,
      "num_input_tokens_seen": 16425608,
      "step": 25190
    },
    {
      "epoch": 13.204926624737945,
      "grad_norm": 1.8969111442565918,
      "learning_rate": 1.5617027101262515e-05,
      "loss": 0.0349,
      "num_input_tokens_seen": 16429288,
      "step": 25195
    },
    {
      "epoch": 13.20754716981132,
      "grad_norm": 1.038817286491394,
      "learning_rate": 1.5606429693053463e-05,
      "loss": 0.0223,
      "num_input_tokens_seen": 16432744,
      "step": 25200
    },
    {
      "epoch": 13.210167714884696,
      "grad_norm": 1.2638226747512817,
      "learning_rate": 1.5595834249872145e-05,
      "loss": 0.021,
      "num_input_tokens_seen": 16435944,
      "step": 25205
    },
    {
      "epoch": 13.21278825995807,
      "grad_norm": 1.3865435123443604,
      "learning_rate": 1.558524077393502e-05,
      "loss": 0.0153,
      "num_input_tokens_seen": 16438760,
      "step": 25210
    },
    {
      "epoch": 13.215408805031446,
      "grad_norm": 1.7061959505081177,
      "learning_rate": 1.557464926745811e-05,
      "loss": 0.0193,
      "num_input_tokens_seen": 16442280,
      "step": 25215
    },
    {
      "epoch": 13.218029350104821,
      "grad_norm": 1.527209758758545,
      "learning_rate": 1.5564059732657045e-05,
      "loss": 0.0195,
      "num_input_tokens_seen": 16445384,
      "step": 25220
    },
    {
      "epoch": 13.220649895178196,
      "grad_norm": 0.3494674861431122,
      "learning_rate": 1.555347217174703e-05,
      "loss": 0.0225,
      "num_input_tokens_seen": 16448040,
      "step": 25225
    },
    {
      "epoch": 13.223270440251572,
      "grad_norm": 1.905441164970398,
      "learning_rate": 1.554288658694287e-05,
      "loss": 0.02,
      "num_input_tokens_seen": 16451432,
      "step": 25230
    },
    {
      "epoch": 13.225890985324948,
      "grad_norm": 2.6024677753448486,
      "learning_rate": 1.553230298045895e-05,
      "loss": 0.0238,
      "num_input_tokens_seen": 16455624,
      "step": 25235
    },
    {
      "epoch": 13.228511530398324,
      "grad_norm": 0.46675026416778564,
      "learning_rate": 1.5521721354509223e-05,
      "loss": 0.025,
      "num_input_tokens_seen": 16459112,
      "step": 25240
    },
    {
      "epoch": 13.231132075471699,
      "grad_norm": 2.946664333343506,
      "learning_rate": 1.5511141711307258e-05,
      "loss": 0.0366,
      "num_input_tokens_seen": 16462312,
      "step": 25245
    },
    {
      "epoch": 13.233752620545074,
      "grad_norm": 1.5822782516479492,
      "learning_rate": 1.5500564053066184e-05,
      "loss": 0.0806,
      "num_input_tokens_seen": 16465128,
      "step": 25250
    },
    {
      "epoch": 13.23637316561845,
      "grad_norm": 1.177115559577942,
      "learning_rate": 1.5489988381998743e-05,
      "loss": 0.0337,
      "num_input_tokens_seen": 16469320,
      "step": 25255
    },
    {
      "epoch": 13.238993710691824,
      "grad_norm": 2.3580026626586914,
      "learning_rate": 1.5479414700317237e-05,
      "loss": 0.1117,
      "num_input_tokens_seen": 16472296,
      "step": 25260
    },
    {
      "epoch": 13.2416142557652,
      "grad_norm": 2.22310209274292,
      "learning_rate": 1.546884301023356e-05,
      "loss": 0.0328,
      "num_input_tokens_seen": 16475240,
      "step": 25265
    },
    {
      "epoch": 13.244234800838575,
      "grad_norm": 0.601416826248169,
      "learning_rate": 1.5458273313959186e-05,
      "loss": 0.0426,
      "num_input_tokens_seen": 16477736,
      "step": 25270
    },
    {
      "epoch": 13.24685534591195,
      "grad_norm": 2.8030624389648438,
      "learning_rate": 1.5447705613705177e-05,
      "loss": 0.0466,
      "num_input_tokens_seen": 16480776,
      "step": 25275
    },
    {
      "epoch": 13.249475890985325,
      "grad_norm": 0.8869103789329529,
      "learning_rate": 1.5437139911682176e-05,
      "loss": 0.0443,
      "num_input_tokens_seen": 16484584,
      "step": 25280
    },
    {
      "epoch": 13.2520964360587,
      "grad_norm": 1.8396215438842773,
      "learning_rate": 1.54265762101004e-05,
      "loss": 0.0438,
      "num_input_tokens_seen": 16487400,
      "step": 25285
    },
    {
      "epoch": 13.254716981132075,
      "grad_norm": 5.9761881828308105,
      "learning_rate": 1.5416014511169663e-05,
      "loss": 0.0166,
      "num_input_tokens_seen": 16490024,
      "step": 25290
    },
    {
      "epoch": 13.25733752620545,
      "grad_norm": 0.9678974747657776,
      "learning_rate": 1.5405454817099363e-05,
      "loss": 0.0259,
      "num_input_tokens_seen": 16493064,
      "step": 25295
    },
    {
      "epoch": 13.259958071278826,
      "grad_norm": 13.449429512023926,
      "learning_rate": 1.5394897130098446e-05,
      "loss": 0.0228,
      "num_input_tokens_seen": 16496136,
      "step": 25300
    },
    {
      "epoch": 13.2625786163522,
      "grad_norm": 0.9523223042488098,
      "learning_rate": 1.5384341452375465e-05,
      "loss": 0.0494,
      "num_input_tokens_seen": 16500328,
      "step": 25305
    },
    {
      "epoch": 13.265199161425576,
      "grad_norm": 2.1529359817504883,
      "learning_rate": 1.5373787786138556e-05,
      "loss": 0.0519,
      "num_input_tokens_seen": 16503688,
      "step": 25310
    },
    {
      "epoch": 13.267819706498951,
      "grad_norm": 0.8045190572738647,
      "learning_rate": 1.5363236133595416e-05,
      "loss": 0.0329,
      "num_input_tokens_seen": 16507112,
      "step": 25315
    },
    {
      "epoch": 13.270440251572326,
      "grad_norm": 1.4934308528900146,
      "learning_rate": 1.5352686496953344e-05,
      "loss": 0.075,
      "num_input_tokens_seen": 16510920,
      "step": 25320
    },
    {
      "epoch": 13.273060796645701,
      "grad_norm": 1.7804259061813354,
      "learning_rate": 1.534213887841919e-05,
      "loss": 0.045,
      "num_input_tokens_seen": 16513736,
      "step": 25325
    },
    {
      "epoch": 13.275681341719078,
      "grad_norm": 1.8790868520736694,
      "learning_rate": 1.5331593280199395e-05,
      "loss": 0.0304,
      "num_input_tokens_seen": 16516744,
      "step": 25330
    },
    {
      "epoch": 13.278301886792454,
      "grad_norm": 1.743255853652954,
      "learning_rate": 1.532104970449999e-05,
      "loss": 0.0655,
      "num_input_tokens_seen": 16519976,
      "step": 25335
    },
    {
      "epoch": 13.280922431865829,
      "grad_norm": 1.3424230813980103,
      "learning_rate": 1.5310508153526555e-05,
      "loss": 0.027,
      "num_input_tokens_seen": 16523112,
      "step": 25340
    },
    {
      "epoch": 13.283542976939204,
      "grad_norm": 1.3093018531799316,
      "learning_rate": 1.5299968629484276e-05,
      "loss": 0.0273,
      "num_input_tokens_seen": 16526376,
      "step": 25345
    },
    {
      "epoch": 13.286163522012579,
      "grad_norm": 1.0605000257492065,
      "learning_rate": 1.5289431134577904e-05,
      "loss": 0.0453,
      "num_input_tokens_seen": 16533160,
      "step": 25350
    },
    {
      "epoch": 13.288784067085954,
      "grad_norm": 2.4988174438476562,
      "learning_rate": 1.527889567101174e-05,
      "loss": 0.0405,
      "num_input_tokens_seen": 16536680,
      "step": 25355
    },
    {
      "epoch": 13.29140461215933,
      "grad_norm": 9.96928882598877,
      "learning_rate": 1.5268362240989707e-05,
      "loss": 0.0289,
      "num_input_tokens_seen": 16539656,
      "step": 25360
    },
    {
      "epoch": 13.294025157232705,
      "grad_norm": 1.4081991910934448,
      "learning_rate": 1.525783084671526e-05,
      "loss": 0.0423,
      "num_input_tokens_seen": 16542376,
      "step": 25365
    },
    {
      "epoch": 13.29664570230608,
      "grad_norm": 0.7206712961196899,
      "learning_rate": 1.5247301490391458e-05,
      "loss": 0.0297,
      "num_input_tokens_seen": 16545352,
      "step": 25370
    },
    {
      "epoch": 13.299266247379455,
      "grad_norm": 2.1898305416107178,
      "learning_rate": 1.5236774174220922e-05,
      "loss": 0.0434,
      "num_input_tokens_seen": 16548424,
      "step": 25375
    },
    {
      "epoch": 13.30188679245283,
      "grad_norm": 1.9951153993606567,
      "learning_rate": 1.5226248900405852e-05,
      "loss": 0.0321,
      "num_input_tokens_seen": 16551592,
      "step": 25380
    },
    {
      "epoch": 13.304507337526205,
      "grad_norm": 0.8882479071617126,
      "learning_rate": 1.5215725671147996e-05,
      "loss": 0.0166,
      "num_input_tokens_seen": 16554568,
      "step": 25385
    },
    {
      "epoch": 13.30712788259958,
      "grad_norm": 1.279465913772583,
      "learning_rate": 1.5205204488648709e-05,
      "loss": 0.0206,
      "num_input_tokens_seen": 16557416,
      "step": 25390
    },
    {
      "epoch": 13.309748427672956,
      "grad_norm": 1.6463310718536377,
      "learning_rate": 1.5194685355108893e-05,
      "loss": 0.0236,
      "num_input_tokens_seen": 16560424,
      "step": 25395
    },
    {
      "epoch": 13.31236897274633,
      "grad_norm": 1.4091523885726929,
      "learning_rate": 1.518416827272904e-05,
      "loss": 0.0295,
      "num_input_tokens_seen": 16563880,
      "step": 25400
    },
    {
      "epoch": 13.314989517819706,
      "grad_norm": 8.98974323272705,
      "learning_rate": 1.5173653243709207e-05,
      "loss": 0.0206,
      "num_input_tokens_seen": 16567432,
      "step": 25405
    },
    {
      "epoch": 13.317610062893081,
      "grad_norm": 3.104326009750366,
      "learning_rate": 1.5163140270249e-05,
      "loss": 0.0438,
      "num_input_tokens_seen": 16570888,
      "step": 25410
    },
    {
      "epoch": 13.320230607966456,
      "grad_norm": 1.4785972833633423,
      "learning_rate": 1.5152629354547626e-05,
      "loss": 0.0258,
      "num_input_tokens_seen": 16573384,
      "step": 25415
    },
    {
      "epoch": 13.322851153039831,
      "grad_norm": 2.0610060691833496,
      "learning_rate": 1.5142120498803842e-05,
      "loss": 0.0338,
      "num_input_tokens_seen": 16576584,
      "step": 25420
    },
    {
      "epoch": 13.325471698113208,
      "grad_norm": 1.814790964126587,
      "learning_rate": 1.5131613705215991e-05,
      "loss": 0.0303,
      "num_input_tokens_seen": 16580584,
      "step": 25425
    },
    {
      "epoch": 13.328092243186584,
      "grad_norm": 3.4023725986480713,
      "learning_rate": 1.5121108975981965e-05,
      "loss": 0.0489,
      "num_input_tokens_seen": 16583848,
      "step": 25430
    },
    {
      "epoch": 13.330712788259959,
      "grad_norm": 0.47741881012916565,
      "learning_rate": 1.5110606313299248e-05,
      "loss": 0.0123,
      "num_input_tokens_seen": 16587272,
      "step": 25435
    },
    {
      "epoch": 13.333333333333334,
      "grad_norm": 2.101491928100586,
      "learning_rate": 1.5100105719364855e-05,
      "loss": 0.024,
      "num_input_tokens_seen": 16590088,
      "step": 25440
    },
    {
      "epoch": 13.335953878406709,
      "grad_norm": 1.702028751373291,
      "learning_rate": 1.50896071963754e-05,
      "loss": 0.0426,
      "num_input_tokens_seen": 16593320,
      "step": 25445
    },
    {
      "epoch": 13.338574423480084,
      "grad_norm": 1.7944238185882568,
      "learning_rate": 1.5079110746527055e-05,
      "loss": 0.04,
      "num_input_tokens_seen": 16596328,
      "step": 25450
    },
    {
      "epoch": 13.34119496855346,
      "grad_norm": 3.2517082691192627,
      "learning_rate": 1.5068616372015565e-05,
      "loss": 0.0799,
      "num_input_tokens_seen": 16599080,
      "step": 25455
    },
    {
      "epoch": 13.343815513626835,
      "grad_norm": 3.9015448093414307,
      "learning_rate": 1.5058124075036223e-05,
      "loss": 0.022,
      "num_input_tokens_seen": 16602056,
      "step": 25460
    },
    {
      "epoch": 13.34643605870021,
      "grad_norm": 2.9970130920410156,
      "learning_rate": 1.5047633857783908e-05,
      "loss": 0.0379,
      "num_input_tokens_seen": 16606216,
      "step": 25465
    },
    {
      "epoch": 13.349056603773585,
      "grad_norm": 0.4527445435523987,
      "learning_rate": 1.5037145722453044e-05,
      "loss": 0.0247,
      "num_input_tokens_seen": 16609064,
      "step": 25470
    },
    {
      "epoch": 13.35167714884696,
      "grad_norm": 1.7430131435394287,
      "learning_rate": 1.5026659671237624e-05,
      "loss": 0.0284,
      "num_input_tokens_seen": 16611560,
      "step": 25475
    },
    {
      "epoch": 13.354297693920335,
      "grad_norm": 1.7983179092407227,
      "learning_rate": 1.501617570633123e-05,
      "loss": 0.031,
      "num_input_tokens_seen": 16614408,
      "step": 25480
    },
    {
      "epoch": 13.35691823899371,
      "grad_norm": 2.081955671310425,
      "learning_rate": 1.500569382992697e-05,
      "loss": 0.0251,
      "num_input_tokens_seen": 16617256,
      "step": 25485
    },
    {
      "epoch": 13.359538784067086,
      "grad_norm": 2.9785397052764893,
      "learning_rate": 1.4995214044217547e-05,
      "loss": 0.0342,
      "num_input_tokens_seen": 16620232,
      "step": 25490
    },
    {
      "epoch": 13.36215932914046,
      "grad_norm": 5.377990245819092,
      "learning_rate": 1.49847363513952e-05,
      "loss": 0.0479,
      "num_input_tokens_seen": 16623912,
      "step": 25495
    },
    {
      "epoch": 13.364779874213836,
      "grad_norm": 1.042116403579712,
      "learning_rate": 1.4974260753651748e-05,
      "loss": 0.0394,
      "num_input_tokens_seen": 16626440,
      "step": 25500
    },
    {
      "epoch": 13.367400419287211,
      "grad_norm": 1.2376922369003296,
      "learning_rate": 1.4963787253178568e-05,
      "loss": 0.0773,
      "num_input_tokens_seen": 16630440,
      "step": 25505
    },
    {
      "epoch": 13.370020964360586,
      "grad_norm": 3.4759814739227295,
      "learning_rate": 1.4953315852166594e-05,
      "loss": 0.0362,
      "num_input_tokens_seen": 16633352,
      "step": 25510
    },
    {
      "epoch": 13.372641509433961,
      "grad_norm": 0.8308486938476562,
      "learning_rate": 1.4942846552806322e-05,
      "loss": 0.0183,
      "num_input_tokens_seen": 16636392,
      "step": 25515
    },
    {
      "epoch": 13.375262054507338,
      "grad_norm": 3.4033384323120117,
      "learning_rate": 1.4932379357287829e-05,
      "loss": 0.0253,
      "num_input_tokens_seen": 16639688,
      "step": 25520
    },
    {
      "epoch": 13.377882599580714,
      "grad_norm": 1.6206347942352295,
      "learning_rate": 1.49219142678007e-05,
      "loss": 0.0223,
      "num_input_tokens_seen": 16643176,
      "step": 25525
    },
    {
      "epoch": 13.380503144654089,
      "grad_norm": 1.9371089935302734,
      "learning_rate": 1.4911451286534128e-05,
      "loss": 0.0483,
      "num_input_tokens_seen": 16646952,
      "step": 25530
    },
    {
      "epoch": 13.383123689727464,
      "grad_norm": 2.1355273723602295,
      "learning_rate": 1.490099041567686e-05,
      "loss": 0.0325,
      "num_input_tokens_seen": 16650792,
      "step": 25535
    },
    {
      "epoch": 13.385744234800839,
      "grad_norm": 0.6771548986434937,
      "learning_rate": 1.4890531657417176e-05,
      "loss": 0.0234,
      "num_input_tokens_seen": 16654568,
      "step": 25540
    },
    {
      "epoch": 13.388364779874214,
      "grad_norm": 4.0402021408081055,
      "learning_rate": 1.4880075013942938e-05,
      "loss": 0.0287,
      "num_input_tokens_seen": 16657608,
      "step": 25545
    },
    {
      "epoch": 13.39098532494759,
      "grad_norm": 2.589852809906006,
      "learning_rate": 1.4869620487441558e-05,
      "loss": 0.0314,
      "num_input_tokens_seen": 16660360,
      "step": 25550
    },
    {
      "epoch": 13.393605870020965,
      "grad_norm": 1.469387412071228,
      "learning_rate": 1.4859168080099988e-05,
      "loss": 0.0359,
      "num_input_tokens_seen": 16663048,
      "step": 25555
    },
    {
      "epoch": 13.39622641509434,
      "grad_norm": 1.9181679487228394,
      "learning_rate": 1.484871779410477e-05,
      "loss": 0.0319,
      "num_input_tokens_seen": 16667080,
      "step": 25560
    },
    {
      "epoch": 13.398846960167715,
      "grad_norm": 2.7096664905548096,
      "learning_rate": 1.4838269631641971e-05,
      "loss": 0.0361,
      "num_input_tokens_seen": 16669608,
      "step": 25565
    },
    {
      "epoch": 13.40146750524109,
      "grad_norm": 2.310617208480835,
      "learning_rate": 1.482782359489724e-05,
      "loss": 0.0389,
      "num_input_tokens_seen": 16676968,
      "step": 25570
    },
    {
      "epoch": 13.404088050314465,
      "grad_norm": 2.362976551055908,
      "learning_rate": 1.4817379686055754e-05,
      "loss": 0.0366,
      "num_input_tokens_seen": 16680552,
      "step": 25575
    },
    {
      "epoch": 13.40670859538784,
      "grad_norm": 1.387063980102539,
      "learning_rate": 1.4806937907302282e-05,
      "loss": 0.0502,
      "num_input_tokens_seen": 16683880,
      "step": 25580
    },
    {
      "epoch": 13.409329140461216,
      "grad_norm": 0.6573610901832581,
      "learning_rate": 1.4796498260821103e-05,
      "loss": 0.0358,
      "num_input_tokens_seen": 16686312,
      "step": 25585
    },
    {
      "epoch": 13.41194968553459,
      "grad_norm": 0.6064860224723816,
      "learning_rate": 1.4786060748796076e-05,
      "loss": 0.0413,
      "num_input_tokens_seen": 16688776,
      "step": 25590
    },
    {
      "epoch": 13.414570230607966,
      "grad_norm": 29.545028686523438,
      "learning_rate": 1.4775625373410617e-05,
      "loss": 0.0282,
      "num_input_tokens_seen": 16692104,
      "step": 25595
    },
    {
      "epoch": 13.417190775681341,
      "grad_norm": 1.5932762622833252,
      "learning_rate": 1.4765192136847685e-05,
      "loss": 0.0314,
      "num_input_tokens_seen": 16694600,
      "step": 25600
    },
    {
      "epoch": 13.419811320754716,
      "grad_norm": 6.931950569152832,
      "learning_rate": 1.47547610412898e-05,
      "loss": 0.017,
      "num_input_tokens_seen": 16697384,
      "step": 25605
    },
    {
      "epoch": 13.422431865828091,
      "grad_norm": 1.277968406677246,
      "learning_rate": 1.4744332088919011e-05,
      "loss": 0.0384,
      "num_input_tokens_seen": 16700776,
      "step": 25610
    },
    {
      "epoch": 13.425052410901468,
      "grad_norm": 2.7172038555145264,
      "learning_rate": 1.4733905281916954e-05,
      "loss": 0.0285,
      "num_input_tokens_seen": 16703272,
      "step": 25615
    },
    {
      "epoch": 13.427672955974844,
      "grad_norm": 1.0716701745986938,
      "learning_rate": 1.472348062246478e-05,
      "loss": 0.0214,
      "num_input_tokens_seen": 16706440,
      "step": 25620
    },
    {
      "epoch": 13.430293501048219,
      "grad_norm": 2.607462167739868,
      "learning_rate": 1.4713058112743228e-05,
      "loss": 0.0258,
      "num_input_tokens_seen": 16709256,
      "step": 25625
    },
    {
      "epoch": 13.432914046121594,
      "grad_norm": 3.1475539207458496,
      "learning_rate": 1.4702637754932553e-05,
      "loss": 0.0389,
      "num_input_tokens_seen": 16712712,
      "step": 25630
    },
    {
      "epoch": 13.435534591194969,
      "grad_norm": 1.8263050317764282,
      "learning_rate": 1.4692219551212587e-05,
      "loss": 0.0341,
      "num_input_tokens_seen": 16715464,
      "step": 25635
    },
    {
      "epoch": 13.438155136268344,
      "grad_norm": 3.8822460174560547,
      "learning_rate": 1.4681803503762695e-05,
      "loss": 0.0784,
      "num_input_tokens_seen": 16719304,
      "step": 25640
    },
    {
      "epoch": 13.44077568134172,
      "grad_norm": 1.9001425504684448,
      "learning_rate": 1.4671389614761788e-05,
      "loss": 0.0244,
      "num_input_tokens_seen": 16722760,
      "step": 25645
    },
    {
      "epoch": 13.443396226415095,
      "grad_norm": 1.638866662979126,
      "learning_rate": 1.4660977886388339e-05,
      "loss": 0.0182,
      "num_input_tokens_seen": 16726152,
      "step": 25650
    },
    {
      "epoch": 13.44601677148847,
      "grad_norm": 1.8090757131576538,
      "learning_rate": 1.4650568320820362e-05,
      "loss": 0.0201,
      "num_input_tokens_seen": 16728808,
      "step": 25655
    },
    {
      "epoch": 13.448637316561845,
      "grad_norm": 2.377183437347412,
      "learning_rate": 1.4640160920235423e-05,
      "loss": 0.0211,
      "num_input_tokens_seen": 16731240,
      "step": 25660
    },
    {
      "epoch": 13.45125786163522,
      "grad_norm": 2.3049604892730713,
      "learning_rate": 1.4629755686810631e-05,
      "loss": 0.0648,
      "num_input_tokens_seen": 16735112,
      "step": 25665
    },
    {
      "epoch": 13.453878406708595,
      "grad_norm": 1.3790712356567383,
      "learning_rate": 1.4619352622722632e-05,
      "loss": 0.0663,
      "num_input_tokens_seen": 16738568,
      "step": 25670
    },
    {
      "epoch": 13.45649895178197,
      "grad_norm": 1.3793138265609741,
      "learning_rate": 1.4608951730147635e-05,
      "loss": 0.0414,
      "num_input_tokens_seen": 16742056,
      "step": 25675
    },
    {
      "epoch": 13.459119496855346,
      "grad_norm": 2.9984335899353027,
      "learning_rate": 1.4598553011261395e-05,
      "loss": 0.0257,
      "num_input_tokens_seen": 16744616,
      "step": 25680
    },
    {
      "epoch": 13.46174004192872,
      "grad_norm": 0.9027290940284729,
      "learning_rate": 1.4588156468239192e-05,
      "loss": 0.0217,
      "num_input_tokens_seen": 16747464,
      "step": 25685
    },
    {
      "epoch": 13.464360587002096,
      "grad_norm": 1.2690061330795288,
      "learning_rate": 1.4577762103255882e-05,
      "loss": 0.0394,
      "num_input_tokens_seen": 16752104,
      "step": 25690
    },
    {
      "epoch": 13.466981132075471,
      "grad_norm": 0.5061679482460022,
      "learning_rate": 1.4567369918485828e-05,
      "loss": 0.0267,
      "num_input_tokens_seen": 16755720,
      "step": 25695
    },
    {
      "epoch": 13.469601677148846,
      "grad_norm": 1.260847806930542,
      "learning_rate": 1.4556979916102976e-05,
      "loss": 0.0418,
      "num_input_tokens_seen": 16758600,
      "step": 25700
    },
    {
      "epoch": 13.472222222222221,
      "grad_norm": 1.7091368436813354,
      "learning_rate": 1.454659209828077e-05,
      "loss": 0.0203,
      "num_input_tokens_seen": 16761352,
      "step": 25705
    },
    {
      "epoch": 13.474842767295598,
      "grad_norm": 1.0042979717254639,
      "learning_rate": 1.4536206467192243e-05,
      "loss": 0.1073,
      "num_input_tokens_seen": 16765512,
      "step": 25710
    },
    {
      "epoch": 13.477463312368974,
      "grad_norm": 2.1752138137817383,
      "learning_rate": 1.4525823025009944e-05,
      "loss": 0.0586,
      "num_input_tokens_seen": 16769064,
      "step": 25715
    },
    {
      "epoch": 13.480083857442349,
      "grad_norm": 2.499318838119507,
      "learning_rate": 1.4515441773905983e-05,
      "loss": 0.0408,
      "num_input_tokens_seen": 16772232,
      "step": 25720
    },
    {
      "epoch": 13.482704402515724,
      "grad_norm": 0.4497223496437073,
      "learning_rate": 1.4505062716051988e-05,
      "loss": 0.0139,
      "num_input_tokens_seen": 16776616,
      "step": 25725
    },
    {
      "epoch": 13.485324947589099,
      "grad_norm": 0.5025182366371155,
      "learning_rate": 1.4494685853619133e-05,
      "loss": 0.0256,
      "num_input_tokens_seen": 16779496,
      "step": 25730
    },
    {
      "epoch": 13.487945492662474,
      "grad_norm": 1.050767183303833,
      "learning_rate": 1.4484311188778143e-05,
      "loss": 0.0338,
      "num_input_tokens_seen": 16783368,
      "step": 25735
    },
    {
      "epoch": 13.49056603773585,
      "grad_norm": 0.6992195844650269,
      "learning_rate": 1.4473938723699285e-05,
      "loss": 0.026,
      "num_input_tokens_seen": 16785928,
      "step": 25740
    },
    {
      "epoch": 13.493186582809225,
      "grad_norm": 3.7271223068237305,
      "learning_rate": 1.4463568460552368e-05,
      "loss": 0.043,
      "num_input_tokens_seen": 16789320,
      "step": 25745
    },
    {
      "epoch": 13.4958071278826,
      "grad_norm": 4.3586955070495605,
      "learning_rate": 1.4453200401506723e-05,
      "loss": 0.0533,
      "num_input_tokens_seen": 16793928,
      "step": 25750
    },
    {
      "epoch": 13.498427672955975,
      "grad_norm": 0.24212917685508728,
      "learning_rate": 1.4442834548731218e-05,
      "loss": 0.039,
      "num_input_tokens_seen": 16796872,
      "step": 25755
    },
    {
      "epoch": 13.50104821802935,
      "grad_norm": 1.669560432434082,
      "learning_rate": 1.4432470904394285e-05,
      "loss": 0.0269,
      "num_input_tokens_seen": 16800552,
      "step": 25760
    },
    {
      "epoch": 13.503668763102725,
      "grad_norm": 2.5371670722961426,
      "learning_rate": 1.4422109470663878e-05,
      "loss": 0.0538,
      "num_input_tokens_seen": 16803880,
      "step": 25765
    },
    {
      "epoch": 13.5062893081761,
      "grad_norm": 1.6673004627227783,
      "learning_rate": 1.441175024970749e-05,
      "loss": 0.0241,
      "num_input_tokens_seen": 16806600,
      "step": 25770
    },
    {
      "epoch": 13.508909853249476,
      "grad_norm": 0.9125850200653076,
      "learning_rate": 1.4401393243692163e-05,
      "loss": 0.0215,
      "num_input_tokens_seen": 16809576,
      "step": 25775
    },
    {
      "epoch": 13.51153039832285,
      "grad_norm": 1.4588751792907715,
      "learning_rate": 1.4391038454784455e-05,
      "loss": 0.0883,
      "num_input_tokens_seen": 16812776,
      "step": 25780
    },
    {
      "epoch": 13.514150943396226,
      "grad_norm": 18.27788543701172,
      "learning_rate": 1.4380685885150458e-05,
      "loss": 0.0251,
      "num_input_tokens_seen": 16815496,
      "step": 25785
    },
    {
      "epoch": 13.516771488469601,
      "grad_norm": 1.5087782144546509,
      "learning_rate": 1.437033553695582e-05,
      "loss": 0.0589,
      "num_input_tokens_seen": 16818984,
      "step": 25790
    },
    {
      "epoch": 13.519392033542976,
      "grad_norm": 1.6825478076934814,
      "learning_rate": 1.4359987412365721e-05,
      "loss": 0.0364,
      "num_input_tokens_seen": 16825096,
      "step": 25795
    },
    {
      "epoch": 13.522012578616351,
      "grad_norm": 2.8491084575653076,
      "learning_rate": 1.4349641513544876e-05,
      "loss": 0.0273,
      "num_input_tokens_seen": 16827880,
      "step": 25800
    },
    {
      "epoch": 13.524633123689728,
      "grad_norm": 3.3475680351257324,
      "learning_rate": 1.4339297842657523e-05,
      "loss": 0.0505,
      "num_input_tokens_seen": 16830504,
      "step": 25805
    },
    {
      "epoch": 13.527253668763104,
      "grad_norm": 2.865718126296997,
      "learning_rate": 1.4328956401867427e-05,
      "loss": 0.0298,
      "num_input_tokens_seen": 16833128,
      "step": 25810
    },
    {
      "epoch": 13.529874213836479,
      "grad_norm": 14.063119888305664,
      "learning_rate": 1.4318617193337908e-05,
      "loss": 0.0387,
      "num_input_tokens_seen": 16835592,
      "step": 25815
    },
    {
      "epoch": 13.532494758909854,
      "grad_norm": 2.8643252849578857,
      "learning_rate": 1.4308280219231812e-05,
      "loss": 0.0623,
      "num_input_tokens_seen": 16839752,
      "step": 25820
    },
    {
      "epoch": 13.535115303983229,
      "grad_norm": 1.6023608446121216,
      "learning_rate": 1.4297945481711527e-05,
      "loss": 0.035,
      "num_input_tokens_seen": 16842568,
      "step": 25825
    },
    {
      "epoch": 13.537735849056604,
      "grad_norm": 2.8209805488586426,
      "learning_rate": 1.4287612982938946e-05,
      "loss": 0.0286,
      "num_input_tokens_seen": 16845320,
      "step": 25830
    },
    {
      "epoch": 13.54035639412998,
      "grad_norm": 2.159853935241699,
      "learning_rate": 1.4277282725075523e-05,
      "loss": 0.0293,
      "num_input_tokens_seen": 16848552,
      "step": 25835
    },
    {
      "epoch": 13.542976939203355,
      "grad_norm": 0.4480031728744507,
      "learning_rate": 1.426695471028221e-05,
      "loss": 0.062,
      "num_input_tokens_seen": 16853608,
      "step": 25840
    },
    {
      "epoch": 13.54559748427673,
      "grad_norm": 1.72585129737854,
      "learning_rate": 1.4256628940719524e-05,
      "loss": 0.0377,
      "num_input_tokens_seen": 16856648,
      "step": 25845
    },
    {
      "epoch": 13.548218029350105,
      "grad_norm": 1.02649986743927,
      "learning_rate": 1.42463054185475e-05,
      "loss": 0.043,
      "num_input_tokens_seen": 16861864,
      "step": 25850
    },
    {
      "epoch": 13.55083857442348,
      "grad_norm": 2.0441977977752686,
      "learning_rate": 1.4235984145925705e-05,
      "loss": 0.0131,
      "num_input_tokens_seen": 16865256,
      "step": 25855
    },
    {
      "epoch": 13.553459119496855,
      "grad_norm": 2.0544486045837402,
      "learning_rate": 1.4225665125013224e-05,
      "loss": 0.023,
      "num_input_tokens_seen": 16868936,
      "step": 25860
    },
    {
      "epoch": 13.55607966457023,
      "grad_norm": 0.5696252584457397,
      "learning_rate": 1.4215348357968669e-05,
      "loss": 0.0162,
      "num_input_tokens_seen": 16872200,
      "step": 25865
    },
    {
      "epoch": 13.558700209643606,
      "grad_norm": 2.3434324264526367,
      "learning_rate": 1.42050338469502e-05,
      "loss": 0.0308,
      "num_input_tokens_seen": 16878824,
      "step": 25870
    },
    {
      "epoch": 13.56132075471698,
      "grad_norm": 2.6128928661346436,
      "learning_rate": 1.4194721594115495e-05,
      "loss": 0.03,
      "num_input_tokens_seen": 16883080,
      "step": 25875
    },
    {
      "epoch": 13.563941299790356,
      "grad_norm": 2.5373528003692627,
      "learning_rate": 1.4184411601621769e-05,
      "loss": 0.1258,
      "num_input_tokens_seen": 16887336,
      "step": 25880
    },
    {
      "epoch": 13.566561844863731,
      "grad_norm": 2.1336097717285156,
      "learning_rate": 1.4174103871625738e-05,
      "loss": 0.0275,
      "num_input_tokens_seen": 16890760,
      "step": 25885
    },
    {
      "epoch": 13.569182389937106,
      "grad_norm": 1.467437505722046,
      "learning_rate": 1.4163798406283674e-05,
      "loss": 0.0188,
      "num_input_tokens_seen": 16893928,
      "step": 25890
    },
    {
      "epoch": 13.571802935010481,
      "grad_norm": 4.19868803024292,
      "learning_rate": 1.415349520775135e-05,
      "loss": 0.0325,
      "num_input_tokens_seen": 16897032,
      "step": 25895
    },
    {
      "epoch": 13.574423480083858,
      "grad_norm": 0.6989461183547974,
      "learning_rate": 1.4143194278184086e-05,
      "loss": 0.0173,
      "num_input_tokens_seen": 16900648,
      "step": 25900
    },
    {
      "epoch": 13.577044025157234,
      "grad_norm": 1.748942494392395,
      "learning_rate": 1.4132895619736724e-05,
      "loss": 0.0496,
      "num_input_tokens_seen": 16903880,
      "step": 25905
    },
    {
      "epoch": 13.579664570230609,
      "grad_norm": 1.1901981830596924,
      "learning_rate": 1.4122599234563613e-05,
      "loss": 0.0261,
      "num_input_tokens_seen": 16907816,
      "step": 25910
    },
    {
      "epoch": 13.582285115303984,
      "grad_norm": 1.150957465171814,
      "learning_rate": 1.4112305124818647e-05,
      "loss": 0.0257,
      "num_input_tokens_seen": 16910664,
      "step": 25915
    },
    {
      "epoch": 13.584905660377359,
      "grad_norm": 1.770677089691162,
      "learning_rate": 1.4102013292655242e-05,
      "loss": 0.0344,
      "num_input_tokens_seen": 16914408,
      "step": 25920
    },
    {
      "epoch": 13.587526205450734,
      "grad_norm": 4.172482013702393,
      "learning_rate": 1.409172374022632e-05,
      "loss": 0.0359,
      "num_input_tokens_seen": 16918120,
      "step": 25925
    },
    {
      "epoch": 13.59014675052411,
      "grad_norm": 4.182407379150391,
      "learning_rate": 1.4081436469684339e-05,
      "loss": 0.0393,
      "num_input_tokens_seen": 16920680,
      "step": 25930
    },
    {
      "epoch": 13.592767295597485,
      "grad_norm": 4.526569843292236,
      "learning_rate": 1.4071151483181294e-05,
      "loss": 0.0293,
      "num_input_tokens_seen": 16923592,
      "step": 25935
    },
    {
      "epoch": 13.59538784067086,
      "grad_norm": 1.9405065774917603,
      "learning_rate": 1.4060868782868669e-05,
      "loss": 0.0376,
      "num_input_tokens_seen": 16926920,
      "step": 25940
    },
    {
      "epoch": 13.598008385744235,
      "grad_norm": 0.9917536973953247,
      "learning_rate": 1.405058837089749e-05,
      "loss": 0.0231,
      "num_input_tokens_seen": 16930056,
      "step": 25945
    },
    {
      "epoch": 13.60062893081761,
      "grad_norm": 0.25327837467193604,
      "learning_rate": 1.404031024941832e-05,
      "loss": 0.0259,
      "num_input_tokens_seen": 16932424,
      "step": 25950
    },
    {
      "epoch": 13.603249475890985,
      "grad_norm": 1.0628656148910522,
      "learning_rate": 1.40300344205812e-05,
      "loss": 0.02,
      "num_input_tokens_seen": 16936200,
      "step": 25955
    },
    {
      "epoch": 13.60587002096436,
      "grad_norm": 1.6821290254592896,
      "learning_rate": 1.4019760886535738e-05,
      "loss": 0.0244,
      "num_input_tokens_seen": 16939176,
      "step": 25960
    },
    {
      "epoch": 13.608490566037736,
      "grad_norm": 1.0462003946304321,
      "learning_rate": 1.4009489649431018e-05,
      "loss": 0.0446,
      "num_input_tokens_seen": 16943048,
      "step": 25965
    },
    {
      "epoch": 13.61111111111111,
      "grad_norm": 1.5472139120101929,
      "learning_rate": 1.3999220711415675e-05,
      "loss": 0.0261,
      "num_input_tokens_seen": 16946056,
      "step": 25970
    },
    {
      "epoch": 13.613731656184486,
      "grad_norm": 2.402069330215454,
      "learning_rate": 1.3988954074637867e-05,
      "loss": 0.0558,
      "num_input_tokens_seen": 16948712,
      "step": 25975
    },
    {
      "epoch": 13.616352201257861,
      "grad_norm": 1.736923098564148,
      "learning_rate": 1.3978689741245238e-05,
      "loss": 0.0445,
      "num_input_tokens_seen": 16952168,
      "step": 25980
    },
    {
      "epoch": 13.618972746331236,
      "grad_norm": 2.887091636657715,
      "learning_rate": 1.3968427713384986e-05,
      "loss": 0.0312,
      "num_input_tokens_seen": 16955496,
      "step": 25985
    },
    {
      "epoch": 13.621593291404611,
      "grad_norm": 1.0761409997940063,
      "learning_rate": 1.3958167993203791e-05,
      "loss": 0.0329,
      "num_input_tokens_seen": 16958472,
      "step": 25990
    },
    {
      "epoch": 13.624213836477988,
      "grad_norm": 2.2221601009368896,
      "learning_rate": 1.3947910582847879e-05,
      "loss": 0.0383,
      "num_input_tokens_seen": 16961832,
      "step": 25995
    },
    {
      "epoch": 13.626834381551364,
      "grad_norm": 1.9318865537643433,
      "learning_rate": 1.3937655484462985e-05,
      "loss": 0.0157,
      "num_input_tokens_seen": 16964552,
      "step": 26000
    },
    {
      "epoch": 13.629454926624739,
      "grad_norm": 2.41764235496521,
      "learning_rate": 1.392740270019437e-05,
      "loss": 0.0245,
      "num_input_tokens_seen": 16967688,
      "step": 26005
    },
    {
      "epoch": 13.632075471698114,
      "grad_norm": 1.4931209087371826,
      "learning_rate": 1.3917152232186776e-05,
      "loss": 0.0427,
      "num_input_tokens_seen": 16971176,
      "step": 26010
    },
    {
      "epoch": 13.634696016771489,
      "grad_norm": 0.3408767282962799,
      "learning_rate": 1.3906904082584504e-05,
      "loss": 0.0327,
      "num_input_tokens_seen": 16974568,
      "step": 26015
    },
    {
      "epoch": 13.637316561844864,
      "grad_norm": 1.947601079940796,
      "learning_rate": 1.3896658253531334e-05,
      "loss": 0.0297,
      "num_input_tokens_seen": 16978728,
      "step": 26020
    },
    {
      "epoch": 13.63993710691824,
      "grad_norm": 1.465227484703064,
      "learning_rate": 1.3886414747170581e-05,
      "loss": 0.0116,
      "num_input_tokens_seen": 16981928,
      "step": 26025
    },
    {
      "epoch": 13.642557651991615,
      "grad_norm": 3.6056361198425293,
      "learning_rate": 1.3876173565645072e-05,
      "loss": 0.0199,
      "num_input_tokens_seen": 16984104,
      "step": 26030
    },
    {
      "epoch": 13.64517819706499,
      "grad_norm": 3.0021812915802,
      "learning_rate": 1.3865934711097156e-05,
      "loss": 0.0856,
      "num_input_tokens_seen": 16986952,
      "step": 26035
    },
    {
      "epoch": 13.647798742138365,
      "grad_norm": 2.0295937061309814,
      "learning_rate": 1.3855698185668675e-05,
      "loss": 0.048,
      "num_input_tokens_seen": 16993160,
      "step": 26040
    },
    {
      "epoch": 13.65041928721174,
      "grad_norm": 1.9738497734069824,
      "learning_rate": 1.3845463991500985e-05,
      "loss": 0.0417,
      "num_input_tokens_seen": 16995560,
      "step": 26045
    },
    {
      "epoch": 13.653039832285115,
      "grad_norm": 8.648484230041504,
      "learning_rate": 1.3835232130734968e-05,
      "loss": 0.0668,
      "num_input_tokens_seen": 16999016,
      "step": 26050
    },
    {
      "epoch": 13.65566037735849,
      "grad_norm": 1.675159215927124,
      "learning_rate": 1.3825002605511017e-05,
      "loss": 0.0309,
      "num_input_tokens_seen": 17001960,
      "step": 26055
    },
    {
      "epoch": 13.658280922431866,
      "grad_norm": 2.170511484146118,
      "learning_rate": 1.3814775417969037e-05,
      "loss": 0.0736,
      "num_input_tokens_seen": 17004936,
      "step": 26060
    },
    {
      "epoch": 13.66090146750524,
      "grad_norm": 7.779881000518799,
      "learning_rate": 1.3804550570248431e-05,
      "loss": 0.0339,
      "num_input_tokens_seen": 17008584,
      "step": 26065
    },
    {
      "epoch": 13.663522012578616,
      "grad_norm": 5.314223766326904,
      "learning_rate": 1.3794328064488116e-05,
      "loss": 0.0332,
      "num_input_tokens_seen": 17011048,
      "step": 26070
    },
    {
      "epoch": 13.666142557651991,
      "grad_norm": 6.564414978027344,
      "learning_rate": 1.3784107902826523e-05,
      "loss": 0.0312,
      "num_input_tokens_seen": 17013384,
      "step": 26075
    },
    {
      "epoch": 13.668763102725366,
      "grad_norm": 0.22391439974308014,
      "learning_rate": 1.3773890087401603e-05,
      "loss": 0.0262,
      "num_input_tokens_seen": 17016584,
      "step": 26080
    },
    {
      "epoch": 13.671383647798741,
      "grad_norm": 2.8764586448669434,
      "learning_rate": 1.3763674620350802e-05,
      "loss": 0.0264,
      "num_input_tokens_seen": 17019368,
      "step": 26085
    },
    {
      "epoch": 13.674004192872118,
      "grad_norm": 4.163600921630859,
      "learning_rate": 1.3753461503811089e-05,
      "loss": 0.0236,
      "num_input_tokens_seen": 17022184,
      "step": 26090
    },
    {
      "epoch": 13.676624737945493,
      "grad_norm": 1.5201746225357056,
      "learning_rate": 1.3743250739918922e-05,
      "loss": 0.0401,
      "num_input_tokens_seen": 17025288,
      "step": 26095
    },
    {
      "epoch": 13.679245283018869,
      "grad_norm": 2.8698537349700928,
      "learning_rate": 1.3733042330810265e-05,
      "loss": 0.016,
      "num_input_tokens_seen": 17028424,
      "step": 26100
    },
    {
      "epoch": 13.681865828092244,
      "grad_norm": 1.696495771408081,
      "learning_rate": 1.3722836278620616e-05,
      "loss": 0.0342,
      "num_input_tokens_seen": 17031720,
      "step": 26105
    },
    {
      "epoch": 13.684486373165619,
      "grad_norm": 1.1367286443710327,
      "learning_rate": 1.3712632585484957e-05,
      "loss": 0.0323,
      "num_input_tokens_seen": 17034888,
      "step": 26110
    },
    {
      "epoch": 13.687106918238994,
      "grad_norm": 2.2651078701019287,
      "learning_rate": 1.37024312535378e-05,
      "loss": 0.0462,
      "num_input_tokens_seen": 17041576,
      "step": 26115
    },
    {
      "epoch": 13.68972746331237,
      "grad_norm": 4.064920902252197,
      "learning_rate": 1.3692232284913133e-05,
      "loss": 0.0417,
      "num_input_tokens_seen": 17044424,
      "step": 26120
    },
    {
      "epoch": 13.692348008385745,
      "grad_norm": 2.4736812114715576,
      "learning_rate": 1.3682035681744459e-05,
      "loss": 0.0383,
      "num_input_tokens_seen": 17047976,
      "step": 26125
    },
    {
      "epoch": 13.69496855345912,
      "grad_norm": 2.70747971534729,
      "learning_rate": 1.3671841446164795e-05,
      "loss": 0.0666,
      "num_input_tokens_seen": 17051208,
      "step": 26130
    },
    {
      "epoch": 13.697589098532495,
      "grad_norm": 1.8868824243545532,
      "learning_rate": 1.3661649580306669e-05,
      "loss": 0.0318,
      "num_input_tokens_seen": 17053640,
      "step": 26135
    },
    {
      "epoch": 13.70020964360587,
      "grad_norm": 1.05879545211792,
      "learning_rate": 1.3651460086302087e-05,
      "loss": 0.0296,
      "num_input_tokens_seen": 17056968,
      "step": 26140
    },
    {
      "epoch": 13.702830188679245,
      "grad_norm": 7.247443199157715,
      "learning_rate": 1.3641272966282609e-05,
      "loss": 0.0258,
      "num_input_tokens_seen": 17059688,
      "step": 26145
    },
    {
      "epoch": 13.70545073375262,
      "grad_norm": 2.3426764011383057,
      "learning_rate": 1.3631088222379218e-05,
      "loss": 0.0422,
      "num_input_tokens_seen": 17064712,
      "step": 26150
    },
    {
      "epoch": 13.708071278825996,
      "grad_norm": 1.4573755264282227,
      "learning_rate": 1.3620905856722468e-05,
      "loss": 0.025,
      "num_input_tokens_seen": 17067528,
      "step": 26155
    },
    {
      "epoch": 13.71069182389937,
      "grad_norm": 3.1176319122314453,
      "learning_rate": 1.3610725871442392e-05,
      "loss": 0.0294,
      "num_input_tokens_seen": 17070952,
      "step": 26160
    },
    {
      "epoch": 13.713312368972746,
      "grad_norm": 2.314612627029419,
      "learning_rate": 1.3600548268668532e-05,
      "loss": 0.0406,
      "num_input_tokens_seen": 17073992,
      "step": 26165
    },
    {
      "epoch": 13.715932914046121,
      "grad_norm": 1.4214715957641602,
      "learning_rate": 1.3590373050529927e-05,
      "loss": 0.0313,
      "num_input_tokens_seen": 17078376,
      "step": 26170
    },
    {
      "epoch": 13.718553459119496,
      "grad_norm": 7.338004112243652,
      "learning_rate": 1.3580200219155116e-05,
      "loss": 0.0551,
      "num_input_tokens_seen": 17082952,
      "step": 26175
    },
    {
      "epoch": 13.721174004192871,
      "grad_norm": 1.301661729812622,
      "learning_rate": 1.3570029776672122e-05,
      "loss": 0.0423,
      "num_input_tokens_seen": 17085640,
      "step": 26180
    },
    {
      "epoch": 13.723794549266248,
      "grad_norm": 0.7495059370994568,
      "learning_rate": 1.35598617252085e-05,
      "loss": 0.0121,
      "num_input_tokens_seen": 17089096,
      "step": 26185
    },
    {
      "epoch": 13.726415094339622,
      "grad_norm": 1.7152961492538452,
      "learning_rate": 1.354969606689129e-05,
      "loss": 0.0721,
      "num_input_tokens_seen": 17091752,
      "step": 26190
    },
    {
      "epoch": 13.729035639412999,
      "grad_norm": 1.981466293334961,
      "learning_rate": 1.353953280384704e-05,
      "loss": 0.0214,
      "num_input_tokens_seen": 17094824,
      "step": 26195
    },
    {
      "epoch": 13.731656184486374,
      "grad_norm": 2.4352519512176514,
      "learning_rate": 1.352937193820177e-05,
      "loss": 0.0579,
      "num_input_tokens_seen": 17097992,
      "step": 26200
    },
    {
      "epoch": 13.734276729559749,
      "grad_norm": 0.6442468166351318,
      "learning_rate": 1.3519213472081038e-05,
      "loss": 0.027,
      "num_input_tokens_seen": 17100968,
      "step": 26205
    },
    {
      "epoch": 13.736897274633124,
      "grad_norm": 2.279310464859009,
      "learning_rate": 1.3509057407609854e-05,
      "loss": 0.028,
      "num_input_tokens_seen": 17103496,
      "step": 26210
    },
    {
      "epoch": 13.7395178197065,
      "grad_norm": 5.996795654296875,
      "learning_rate": 1.3498903746912769e-05,
      "loss": 0.0459,
      "num_input_tokens_seen": 17106920,
      "step": 26215
    },
    {
      "epoch": 13.742138364779874,
      "grad_norm": 7.121149063110352,
      "learning_rate": 1.3488752492113807e-05,
      "loss": 0.0403,
      "num_input_tokens_seen": 17110472,
      "step": 26220
    },
    {
      "epoch": 13.74475890985325,
      "grad_norm": 1.787037968635559,
      "learning_rate": 1.3478603645336508e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 17113640,
      "step": 26225
    },
    {
      "epoch": 13.747379454926625,
      "grad_norm": 2.5377304553985596,
      "learning_rate": 1.3468457208703877e-05,
      "loss": 0.1106,
      "num_input_tokens_seen": 17116808,
      "step": 26230
    },
    {
      "epoch": 13.75,
      "grad_norm": 1.3687015771865845,
      "learning_rate": 1.3458313184338437e-05,
      "loss": 0.0202,
      "num_input_tokens_seen": 17119144,
      "step": 26235
    },
    {
      "epoch": 13.752620545073375,
      "grad_norm": 1.6871086359024048,
      "learning_rate": 1.3448171574362201e-05,
      "loss": 0.0224,
      "num_input_tokens_seen": 17122056,
      "step": 26240
    },
    {
      "epoch": 13.75524109014675,
      "grad_norm": 2.5893452167510986,
      "learning_rate": 1.3438032380896681e-05,
      "loss": 0.0259,
      "num_input_tokens_seen": 17125352,
      "step": 26245
    },
    {
      "epoch": 13.757861635220126,
      "grad_norm": 10.669879913330078,
      "learning_rate": 1.342789560606289e-05,
      "loss": 0.0828,
      "num_input_tokens_seen": 17128456,
      "step": 26250
    },
    {
      "epoch": 13.7604821802935,
      "grad_norm": 3.6378841400146484,
      "learning_rate": 1.341776125198131e-05,
      "loss": 0.034,
      "num_input_tokens_seen": 17131496,
      "step": 26255
    },
    {
      "epoch": 13.763102725366876,
      "grad_norm": 1.2072253227233887,
      "learning_rate": 1.3407629320771948e-05,
      "loss": 0.0287,
      "num_input_tokens_seen": 17135528,
      "step": 26260
    },
    {
      "epoch": 13.765723270440251,
      "grad_norm": 2.6702423095703125,
      "learning_rate": 1.3397499814554266e-05,
      "loss": 0.0386,
      "num_input_tokens_seen": 17138728,
      "step": 26265
    },
    {
      "epoch": 13.768343815513626,
      "grad_norm": 2.258636713027954,
      "learning_rate": 1.3387372735447256e-05,
      "loss": 0.0378,
      "num_input_tokens_seen": 17142600,
      "step": 26270
    },
    {
      "epoch": 13.770964360587001,
      "grad_norm": 0.26588621735572815,
      "learning_rate": 1.3377248085569399e-05,
      "loss": 0.0453,
      "num_input_tokens_seen": 17145448,
      "step": 26275
    },
    {
      "epoch": 13.773584905660378,
      "grad_norm": 1.0802931785583496,
      "learning_rate": 1.336712586703863e-05,
      "loss": 0.0296,
      "num_input_tokens_seen": 17148712,
      "step": 26280
    },
    {
      "epoch": 13.776205450733752,
      "grad_norm": 2.7426605224609375,
      "learning_rate": 1.3357006081972414e-05,
      "loss": 0.0256,
      "num_input_tokens_seen": 17151144,
      "step": 26285
    },
    {
      "epoch": 13.778825995807129,
      "grad_norm": 2.4646596908569336,
      "learning_rate": 1.3346888732487706e-05,
      "loss": 0.027,
      "num_input_tokens_seen": 17153864,
      "step": 26290
    },
    {
      "epoch": 13.781446540880504,
      "grad_norm": 1.375509262084961,
      "learning_rate": 1.3336773820700921e-05,
      "loss": 0.0405,
      "num_input_tokens_seen": 17156968,
      "step": 26295
    },
    {
      "epoch": 13.784067085953879,
      "grad_norm": 1.2852448225021362,
      "learning_rate": 1.3326661348727992e-05,
      "loss": 0.064,
      "num_input_tokens_seen": 17160520,
      "step": 26300
    },
    {
      "epoch": 13.786687631027254,
      "grad_norm": 1.6069732904434204,
      "learning_rate": 1.3316551318684342e-05,
      "loss": 0.0262,
      "num_input_tokens_seen": 17163368,
      "step": 26305
    },
    {
      "epoch": 13.78930817610063,
      "grad_norm": 2.1096785068511963,
      "learning_rate": 1.3306443732684854e-05,
      "loss": 0.067,
      "num_input_tokens_seen": 17166472,
      "step": 26310
    },
    {
      "epoch": 13.791928721174004,
      "grad_norm": 2.6141090393066406,
      "learning_rate": 1.3296338592843929e-05,
      "loss": 0.0349,
      "num_input_tokens_seen": 17170088,
      "step": 26315
    },
    {
      "epoch": 13.79454926624738,
      "grad_norm": 0.840100884437561,
      "learning_rate": 1.3286235901275457e-05,
      "loss": 0.014,
      "num_input_tokens_seen": 17172840,
      "step": 26320
    },
    {
      "epoch": 13.797169811320755,
      "grad_norm": 1.4979032278060913,
      "learning_rate": 1.327613566009279e-05,
      "loss": 0.0165,
      "num_input_tokens_seen": 17175752,
      "step": 26325
    },
    {
      "epoch": 13.79979035639413,
      "grad_norm": 2.716517686843872,
      "learning_rate": 1.3266037871408798e-05,
      "loss": 0.0625,
      "num_input_tokens_seen": 17178504,
      "step": 26330
    },
    {
      "epoch": 13.802410901467505,
      "grad_norm": 2.204336166381836,
      "learning_rate": 1.3255942537335805e-05,
      "loss": 0.029,
      "num_input_tokens_seen": 17180648,
      "step": 26335
    },
    {
      "epoch": 13.80503144654088,
      "grad_norm": 2.06851863861084,
      "learning_rate": 1.3245849659985652e-05,
      "loss": 0.0684,
      "num_input_tokens_seen": 17184648,
      "step": 26340
    },
    {
      "epoch": 13.807651991614255,
      "grad_norm": 3.664202928543091,
      "learning_rate": 1.3235759241469658e-05,
      "loss": 0.0539,
      "num_input_tokens_seen": 17188264,
      "step": 26345
    },
    {
      "epoch": 13.81027253668763,
      "grad_norm": 1.7352885007858276,
      "learning_rate": 1.322567128389861e-05,
      "loss": 0.0215,
      "num_input_tokens_seen": 17191464,
      "step": 26350
    },
    {
      "epoch": 13.812893081761006,
      "grad_norm": 3.8574345111846924,
      "learning_rate": 1.3215585789382812e-05,
      "loss": 0.0354,
      "num_input_tokens_seen": 17194792,
      "step": 26355
    },
    {
      "epoch": 13.815513626834381,
      "grad_norm": 1.6803336143493652,
      "learning_rate": 1.320550276003201e-05,
      "loss": 0.0269,
      "num_input_tokens_seen": 17197864,
      "step": 26360
    },
    {
      "epoch": 13.818134171907756,
      "grad_norm": 1.7470371723175049,
      "learning_rate": 1.3195422197955473e-05,
      "loss": 0.0369,
      "num_input_tokens_seen": 17200840,
      "step": 26365
    },
    {
      "epoch": 13.820754716981131,
      "grad_norm": 2.0785841941833496,
      "learning_rate": 1.3185344105261937e-05,
      "loss": 0.038,
      "num_input_tokens_seen": 17204296,
      "step": 26370
    },
    {
      "epoch": 13.823375262054507,
      "grad_norm": 1.4956274032592773,
      "learning_rate": 1.3175268484059639e-05,
      "loss": 0.0218,
      "num_input_tokens_seen": 17206920,
      "step": 26375
    },
    {
      "epoch": 13.825995807127882,
      "grad_norm": 1.45342218875885,
      "learning_rate": 1.316519533645626e-05,
      "loss": 0.0278,
      "num_input_tokens_seen": 17209672,
      "step": 26380
    },
    {
      "epoch": 13.828616352201259,
      "grad_norm": 3.1361324787139893,
      "learning_rate": 1.3155124664559004e-05,
      "loss": 0.0542,
      "num_input_tokens_seen": 17212296,
      "step": 26385
    },
    {
      "epoch": 13.831236897274634,
      "grad_norm": 1.9838836193084717,
      "learning_rate": 1.3145056470474525e-05,
      "loss": 0.138,
      "num_input_tokens_seen": 17215880,
      "step": 26390
    },
    {
      "epoch": 13.833857442348009,
      "grad_norm": 3.004363775253296,
      "learning_rate": 1.313499075630899e-05,
      "loss": 0.031,
      "num_input_tokens_seen": 17218760,
      "step": 26395
    },
    {
      "epoch": 13.836477987421384,
      "grad_norm": 3.006329298019409,
      "learning_rate": 1.312492752416802e-05,
      "loss": 0.0672,
      "num_input_tokens_seen": 17222408,
      "step": 26400
    },
    {
      "epoch": 13.83909853249476,
      "grad_norm": 2.829622268676758,
      "learning_rate": 1.3114866776156748e-05,
      "loss": 0.0419,
      "num_input_tokens_seen": 17225800,
      "step": 26405
    },
    {
      "epoch": 13.841719077568134,
      "grad_norm": 2.4829981327056885,
      "learning_rate": 1.3104808514379748e-05,
      "loss": 0.0185,
      "num_input_tokens_seen": 17229256,
      "step": 26410
    },
    {
      "epoch": 13.84433962264151,
      "grad_norm": 7.308896541595459,
      "learning_rate": 1.3094752740941094e-05,
      "loss": 0.0829,
      "num_input_tokens_seen": 17232104,
      "step": 26415
    },
    {
      "epoch": 13.846960167714885,
      "grad_norm": 0.9149249792098999,
      "learning_rate": 1.3084699457944344e-05,
      "loss": 0.211,
      "num_input_tokens_seen": 17235528,
      "step": 26420
    },
    {
      "epoch": 13.84958071278826,
      "grad_norm": 2.2824699878692627,
      "learning_rate": 1.3074648667492528e-05,
      "loss": 0.0204,
      "num_input_tokens_seen": 17238088,
      "step": 26425
    },
    {
      "epoch": 13.852201257861635,
      "grad_norm": 2.224431037902832,
      "learning_rate": 1.3064600371688171e-05,
      "loss": 0.0518,
      "num_input_tokens_seen": 17241288,
      "step": 26430
    },
    {
      "epoch": 13.85482180293501,
      "grad_norm": 0.8538007736206055,
      "learning_rate": 1.3054554572633249e-05,
      "loss": 0.0312,
      "num_input_tokens_seen": 17244840,
      "step": 26435
    },
    {
      "epoch": 13.857442348008385,
      "grad_norm": 3.083048105239868,
      "learning_rate": 1.3044511272429222e-05,
      "loss": 0.0344,
      "num_input_tokens_seen": 17247688,
      "step": 26440
    },
    {
      "epoch": 13.86006289308176,
      "grad_norm": 1.059241771697998,
      "learning_rate": 1.3034470473177038e-05,
      "loss": 0.0255,
      "num_input_tokens_seen": 17252360,
      "step": 26445
    },
    {
      "epoch": 13.862683438155136,
      "grad_norm": 4.255823135375977,
      "learning_rate": 1.3024432176977125e-05,
      "loss": 0.0346,
      "num_input_tokens_seen": 17255368,
      "step": 26450
    },
    {
      "epoch": 13.865303983228511,
      "grad_norm": 3.7358622550964355,
      "learning_rate": 1.3014396385929372e-05,
      "loss": 0.0732,
      "num_input_tokens_seen": 17257832,
      "step": 26455
    },
    {
      "epoch": 13.867924528301886,
      "grad_norm": 1.6070592403411865,
      "learning_rate": 1.3004363102133165e-05,
      "loss": 0.0163,
      "num_input_tokens_seen": 17261480,
      "step": 26460
    },
    {
      "epoch": 13.870545073375261,
      "grad_norm": 2.1868011951446533,
      "learning_rate": 1.2994332327687348e-05,
      "loss": 0.0503,
      "num_input_tokens_seen": 17264360,
      "step": 26465
    },
    {
      "epoch": 13.873165618448636,
      "grad_norm": 2.2277960777282715,
      "learning_rate": 1.2984304064690228e-05,
      "loss": 0.0471,
      "num_input_tokens_seen": 17268104,
      "step": 26470
    },
    {
      "epoch": 13.875786163522012,
      "grad_norm": 2.5162696838378906,
      "learning_rate": 1.2974278315239619e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 17271048,
      "step": 26475
    },
    {
      "epoch": 13.878406708595389,
      "grad_norm": 2.061375617980957,
      "learning_rate": 1.2964255081432786e-05,
      "loss": 0.025,
      "num_input_tokens_seen": 17274056,
      "step": 26480
    },
    {
      "epoch": 13.881027253668764,
      "grad_norm": 1.392879605293274,
      "learning_rate": 1.2954234365366488e-05,
      "loss": 0.0276,
      "num_input_tokens_seen": 17277192,
      "step": 26485
    },
    {
      "epoch": 13.883647798742139,
      "grad_norm": 1.6994402408599854,
      "learning_rate": 1.2944216169136936e-05,
      "loss": 0.0267,
      "num_input_tokens_seen": 17281128,
      "step": 26490
    },
    {
      "epoch": 13.886268343815514,
      "grad_norm": 1.239956021308899,
      "learning_rate": 1.2934200494839815e-05,
      "loss": 0.0355,
      "num_input_tokens_seen": 17283848,
      "step": 26495
    },
    {
      "epoch": 13.88888888888889,
      "grad_norm": 4.044978618621826,
      "learning_rate": 1.2924187344570293e-05,
      "loss": 0.0448,
      "num_input_tokens_seen": 17287208,
      "step": 26500
    },
    {
      "epoch": 13.891509433962264,
      "grad_norm": 0.4555167555809021,
      "learning_rate": 1.2914176720423011e-05,
      "loss": 0.0267,
      "num_input_tokens_seen": 17290696,
      "step": 26505
    },
    {
      "epoch": 13.89412997903564,
      "grad_norm": 1.5446271896362305,
      "learning_rate": 1.2904168624492088e-05,
      "loss": 0.0216,
      "num_input_tokens_seen": 17294600,
      "step": 26510
    },
    {
      "epoch": 13.896750524109015,
      "grad_norm": 1.5860241651535034,
      "learning_rate": 1.2894163058871092e-05,
      "loss": 0.0462,
      "num_input_tokens_seen": 17297768,
      "step": 26515
    },
    {
      "epoch": 13.89937106918239,
      "grad_norm": 3.253612518310547,
      "learning_rate": 1.2884160025653064e-05,
      "loss": 0.0391,
      "num_input_tokens_seen": 17300808,
      "step": 26520
    },
    {
      "epoch": 13.901991614255765,
      "grad_norm": 2.240976095199585,
      "learning_rate": 1.2874159526930532e-05,
      "loss": 0.0433,
      "num_input_tokens_seen": 17303432,
      "step": 26525
    },
    {
      "epoch": 13.90461215932914,
      "grad_norm": 2.8699333667755127,
      "learning_rate": 1.286416156479549e-05,
      "loss": 0.0197,
      "num_input_tokens_seen": 17306152,
      "step": 26530
    },
    {
      "epoch": 13.907232704402515,
      "grad_norm": 1.1138877868652344,
      "learning_rate": 1.2854166141339394e-05,
      "loss": 0.0322,
      "num_input_tokens_seen": 17308840,
      "step": 26535
    },
    {
      "epoch": 13.90985324947589,
      "grad_norm": 2.682570457458496,
      "learning_rate": 1.2844173258653183e-05,
      "loss": 0.0516,
      "num_input_tokens_seen": 17311944,
      "step": 26540
    },
    {
      "epoch": 13.912473794549266,
      "grad_norm": 1.3768019676208496,
      "learning_rate": 1.2834182918827247e-05,
      "loss": 0.0269,
      "num_input_tokens_seen": 17314888,
      "step": 26545
    },
    {
      "epoch": 13.915094339622641,
      "grad_norm": 3.877530574798584,
      "learning_rate": 1.282419512395144e-05,
      "loss": 0.0408,
      "num_input_tokens_seen": 17318248,
      "step": 26550
    },
    {
      "epoch": 13.917714884696016,
      "grad_norm": 2.473621129989624,
      "learning_rate": 1.2814209876115102e-05,
      "loss": 0.0541,
      "num_input_tokens_seen": 17321320,
      "step": 26555
    },
    {
      "epoch": 13.920335429769391,
      "grad_norm": 1.8944398164749146,
      "learning_rate": 1.2804227177407038e-05,
      "loss": 0.045,
      "num_input_tokens_seen": 17324680,
      "step": 26560
    },
    {
      "epoch": 13.922955974842766,
      "grad_norm": 1.9160014390945435,
      "learning_rate": 1.279424702991552e-05,
      "loss": 0.0444,
      "num_input_tokens_seen": 17328072,
      "step": 26565
    },
    {
      "epoch": 13.925576519916142,
      "grad_norm": 2.1731553077697754,
      "learning_rate": 1.2784269435728264e-05,
      "loss": 0.0314,
      "num_input_tokens_seen": 17331272,
      "step": 26570
    },
    {
      "epoch": 13.928197064989519,
      "grad_norm": 1.292556643486023,
      "learning_rate": 1.2774294396932488e-05,
      "loss": 0.0217,
      "num_input_tokens_seen": 17334920,
      "step": 26575
    },
    {
      "epoch": 13.930817610062894,
      "grad_norm": 2.0725677013397217,
      "learning_rate": 1.2764321915614835e-05,
      "loss": 0.0294,
      "num_input_tokens_seen": 17338664,
      "step": 26580
    },
    {
      "epoch": 13.933438155136269,
      "grad_norm": 2.343698024749756,
      "learning_rate": 1.2754351993861444e-05,
      "loss": 0.0298,
      "num_input_tokens_seen": 17341000,
      "step": 26585
    },
    {
      "epoch": 13.936058700209644,
      "grad_norm": 1.3452532291412354,
      "learning_rate": 1.2744384633757922e-05,
      "loss": 0.0225,
      "num_input_tokens_seen": 17344456,
      "step": 26590
    },
    {
      "epoch": 13.93867924528302,
      "grad_norm": 8.435612678527832,
      "learning_rate": 1.2734419837389306e-05,
      "loss": 0.0648,
      "num_input_tokens_seen": 17347976,
      "step": 26595
    },
    {
      "epoch": 13.941299790356394,
      "grad_norm": 0.7590645551681519,
      "learning_rate": 1.2724457606840123e-05,
      "loss": 0.0644,
      "num_input_tokens_seen": 17351336,
      "step": 26600
    },
    {
      "epoch": 13.94392033542977,
      "grad_norm": 1.4588102102279663,
      "learning_rate": 1.2714497944194376e-05,
      "loss": 0.0254,
      "num_input_tokens_seen": 17356744,
      "step": 26605
    },
    {
      "epoch": 13.946540880503145,
      "grad_norm": 1.149521827697754,
      "learning_rate": 1.2704540851535484e-05,
      "loss": 0.0256,
      "num_input_tokens_seen": 17359240,
      "step": 26610
    },
    {
      "epoch": 13.94916142557652,
      "grad_norm": 4.367123603820801,
      "learning_rate": 1.2694586330946376e-05,
      "loss": 0.04,
      "num_input_tokens_seen": 17363592,
      "step": 26615
    },
    {
      "epoch": 13.951781970649895,
      "grad_norm": 2.9736227989196777,
      "learning_rate": 1.2684634384509427e-05,
      "loss": 0.0425,
      "num_input_tokens_seen": 17365992,
      "step": 26620
    },
    {
      "epoch": 13.95440251572327,
      "grad_norm": 2.2979001998901367,
      "learning_rate": 1.2674685014306456e-05,
      "loss": 0.0364,
      "num_input_tokens_seen": 17368520,
      "step": 26625
    },
    {
      "epoch": 13.957023060796645,
      "grad_norm": 2.3034770488739014,
      "learning_rate": 1.2664738222418776e-05,
      "loss": 0.0271,
      "num_input_tokens_seen": 17371880,
      "step": 26630
    },
    {
      "epoch": 13.95964360587002,
      "grad_norm": 2.451181650161743,
      "learning_rate": 1.2654794010927118e-05,
      "loss": 0.0144,
      "num_input_tokens_seen": 17374600,
      "step": 26635
    },
    {
      "epoch": 13.962264150943396,
      "grad_norm": 9.201007843017578,
      "learning_rate": 1.2644852381911715e-05,
      "loss": 0.0427,
      "num_input_tokens_seen": 17378600,
      "step": 26640
    },
    {
      "epoch": 13.964884696016771,
      "grad_norm": 1.556721568107605,
      "learning_rate": 1.2634913337452248e-05,
      "loss": 0.0281,
      "num_input_tokens_seen": 17381416,
      "step": 26645
    },
    {
      "epoch": 13.967505241090146,
      "grad_norm": 4.132486820220947,
      "learning_rate": 1.2624976879627832e-05,
      "loss": 0.0299,
      "num_input_tokens_seen": 17384264,
      "step": 26650
    },
    {
      "epoch": 13.970125786163521,
      "grad_norm": 0.449497789144516,
      "learning_rate": 1.2615043010517069e-05,
      "loss": 0.0704,
      "num_input_tokens_seen": 17387912,
      "step": 26655
    },
    {
      "epoch": 13.972746331236896,
      "grad_norm": 0.30466338992118835,
      "learning_rate": 1.2605111732198027e-05,
      "loss": 0.0174,
      "num_input_tokens_seen": 17390600,
      "step": 26660
    },
    {
      "epoch": 13.975366876310272,
      "grad_norm": 2.059965133666992,
      "learning_rate": 1.2595183046748188e-05,
      "loss": 0.0389,
      "num_input_tokens_seen": 17393704,
      "step": 26665
    },
    {
      "epoch": 13.977987421383649,
      "grad_norm": 2.360886812210083,
      "learning_rate": 1.2585256956244545e-05,
      "loss": 0.0526,
      "num_input_tokens_seen": 17396808,
      "step": 26670
    },
    {
      "epoch": 13.980607966457024,
      "grad_norm": 3.1605703830718994,
      "learning_rate": 1.2575333462763506e-05,
      "loss": 0.0419,
      "num_input_tokens_seen": 17400040,
      "step": 26675
    },
    {
      "epoch": 13.983228511530399,
      "grad_norm": 1.9220483303070068,
      "learning_rate": 1.2565412568380957e-05,
      "loss": 0.0323,
      "num_input_tokens_seen": 17403624,
      "step": 26680
    },
    {
      "epoch": 13.985849056603774,
      "grad_norm": 2.0590057373046875,
      "learning_rate": 1.2555494275172237e-05,
      "loss": 0.0436,
      "num_input_tokens_seen": 17406760,
      "step": 26685
    },
    {
      "epoch": 13.98846960167715,
      "grad_norm": 1.131144642829895,
      "learning_rate": 1.2545578585212148e-05,
      "loss": 0.0225,
      "num_input_tokens_seen": 17409832,
      "step": 26690
    },
    {
      "epoch": 13.991090146750524,
      "grad_norm": 1.7532923221588135,
      "learning_rate": 1.2535665500574922e-05,
      "loss": 0.0219,
      "num_input_tokens_seen": 17413672,
      "step": 26695
    },
    {
      "epoch": 13.9937106918239,
      "grad_norm": 5.717623710632324,
      "learning_rate": 1.2525755023334285e-05,
      "loss": 0.0531,
      "num_input_tokens_seen": 17416936,
      "step": 26700
    },
    {
      "epoch": 13.996331236897275,
      "grad_norm": 1.6007622480392456,
      "learning_rate": 1.251584715556337e-05,
      "loss": 0.0386,
      "num_input_tokens_seen": 17420520,
      "step": 26705
    },
    {
      "epoch": 13.99895178197065,
      "grad_norm": 2.2562527656555176,
      "learning_rate": 1.2505941899334805e-05,
      "loss": 0.0328,
      "num_input_tokens_seen": 17424840,
      "step": 26710
    },
    {
      "epoch": 14.0,
      "eval_loss": 1.2236897945404053,
      "eval_runtime": 11.9137,
      "eval_samples_per_second": 71.179,
      "eval_steps_per_second": 17.795,
      "num_input_tokens_seen": 17425368,
      "step": 26712
    },
    {
      "epoch": 14.001572327044025,
      "grad_norm": 0.4262751340866089,
      "learning_rate": 1.2496039256720665e-05,
      "loss": 0.0277,
      "num_input_tokens_seen": 17427448,
      "step": 26715
    },
    {
      "epoch": 14.0041928721174,
      "grad_norm": 0.5197495818138123,
      "learning_rate": 1.2486139229792452e-05,
      "loss": 0.0304,
      "num_input_tokens_seen": 17430008,
      "step": 26720
    },
    {
      "epoch": 14.006813417190775,
      "grad_norm": 1.7785847187042236,
      "learning_rate": 1.2476241820621152e-05,
      "loss": 0.0249,
      "num_input_tokens_seen": 17433080,
      "step": 26725
    },
    {
      "epoch": 14.00943396226415,
      "grad_norm": 1.061754822731018,
      "learning_rate": 1.2466347031277176e-05,
      "loss": 0.0214,
      "num_input_tokens_seen": 17436088,
      "step": 26730
    },
    {
      "epoch": 14.012054507337526,
      "grad_norm": 1.1856355667114258,
      "learning_rate": 1.2456454863830411e-05,
      "loss": 0.0204,
      "num_input_tokens_seen": 17438680,
      "step": 26735
    },
    {
      "epoch": 14.014675052410901,
      "grad_norm": 1.996592402458191,
      "learning_rate": 1.2446565320350185e-05,
      "loss": 0.0189,
      "num_input_tokens_seen": 17442424,
      "step": 26740
    },
    {
      "epoch": 14.017295597484276,
      "grad_norm": 2.7339107990264893,
      "learning_rate": 1.2436678402905283e-05,
      "loss": 0.0196,
      "num_input_tokens_seen": 17445784,
      "step": 26745
    },
    {
      "epoch": 14.019916142557651,
      "grad_norm": 0.3858039677143097,
      "learning_rate": 1.2426794113563933e-05,
      "loss": 0.0269,
      "num_input_tokens_seen": 17448280,
      "step": 26750
    },
    {
      "epoch": 14.022536687631026,
      "grad_norm": 4.8802995681762695,
      "learning_rate": 1.24169124543938e-05,
      "loss": 0.0324,
      "num_input_tokens_seen": 17452248,
      "step": 26755
    },
    {
      "epoch": 14.025157232704403,
      "grad_norm": 2.7421414852142334,
      "learning_rate": 1.2407033427462026e-05,
      "loss": 0.0211,
      "num_input_tokens_seen": 17455672,
      "step": 26760
    },
    {
      "epoch": 14.027777777777779,
      "grad_norm": 0.2686551809310913,
      "learning_rate": 1.2397157034835192e-05,
      "loss": 0.0081,
      "num_input_tokens_seen": 17458328,
      "step": 26765
    },
    {
      "epoch": 14.030398322851154,
      "grad_norm": 2.8449740409851074,
      "learning_rate": 1.2387283278579325e-05,
      "loss": 0.0252,
      "num_input_tokens_seen": 17460856,
      "step": 26770
    },
    {
      "epoch": 14.033018867924529,
      "grad_norm": 0.844871461391449,
      "learning_rate": 1.237741216075991e-05,
      "loss": 0.0112,
      "num_input_tokens_seen": 17464440,
      "step": 26775
    },
    {
      "epoch": 14.035639412997904,
      "grad_norm": 1.4527645111083984,
      "learning_rate": 1.2367543683441868e-05,
      "loss": 0.0151,
      "num_input_tokens_seen": 17467512,
      "step": 26780
    },
    {
      "epoch": 14.03825995807128,
      "grad_norm": 0.1196390688419342,
      "learning_rate": 1.2357677848689558e-05,
      "loss": 0.0275,
      "num_input_tokens_seen": 17471224,
      "step": 26785
    },
    {
      "epoch": 14.040880503144654,
      "grad_norm": 1.168246865272522,
      "learning_rate": 1.2347814658566811e-05,
      "loss": 0.0102,
      "num_input_tokens_seen": 17474616,
      "step": 26790
    },
    {
      "epoch": 14.04350104821803,
      "grad_norm": 0.6908698081970215,
      "learning_rate": 1.2337954115136894e-05,
      "loss": 0.0208,
      "num_input_tokens_seen": 17477880,
      "step": 26795
    },
    {
      "epoch": 14.046121593291405,
      "grad_norm": 0.8145492672920227,
      "learning_rate": 1.2328096220462529e-05,
      "loss": 0.0927,
      "num_input_tokens_seen": 17481016,
      "step": 26800
    },
    {
      "epoch": 14.04874213836478,
      "grad_norm": 5.025386810302734,
      "learning_rate": 1.2318240976605865e-05,
      "loss": 0.0146,
      "num_input_tokens_seen": 17484568,
      "step": 26805
    },
    {
      "epoch": 14.051362683438155,
      "grad_norm": 1.5107330083847046,
      "learning_rate": 1.23083883856285e-05,
      "loss": 0.0284,
      "num_input_tokens_seen": 17491032,
      "step": 26810
    },
    {
      "epoch": 14.05398322851153,
      "grad_norm": 1.9169410467147827,
      "learning_rate": 1.229853844959149e-05,
      "loss": 0.0152,
      "num_input_tokens_seen": 17494136,
      "step": 26815
    },
    {
      "epoch": 14.056603773584905,
      "grad_norm": 3.033569097518921,
      "learning_rate": 1.228869117055533e-05,
      "loss": 0.0283,
      "num_input_tokens_seen": 17496920,
      "step": 26820
    },
    {
      "epoch": 14.05922431865828,
      "grad_norm": 2.134632110595703,
      "learning_rate": 1.227884655057997e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 17500024,
      "step": 26825
    },
    {
      "epoch": 14.061844863731656,
      "grad_norm": 4.25255823135376,
      "learning_rate": 1.2269004591724777e-05,
      "loss": 0.0184,
      "num_input_tokens_seen": 17503416,
      "step": 26830
    },
    {
      "epoch": 14.064465408805031,
      "grad_norm": 1.4663476943969727,
      "learning_rate": 1.2259165296048575e-05,
      "loss": 0.0132,
      "num_input_tokens_seen": 17506648,
      "step": 26835
    },
    {
      "epoch": 14.067085953878406,
      "grad_norm": 2.2059359550476074,
      "learning_rate": 1.2249328665609638e-05,
      "loss": 0.0176,
      "num_input_tokens_seen": 17509496,
      "step": 26840
    },
    {
      "epoch": 14.069706498951781,
      "grad_norm": 3.790532112121582,
      "learning_rate": 1.2239494702465676e-05,
      "loss": 0.0702,
      "num_input_tokens_seen": 17513464,
      "step": 26845
    },
    {
      "epoch": 14.072327044025156,
      "grad_norm": 0.9320041537284851,
      "learning_rate": 1.2229663408673842e-05,
      "loss": 0.0154,
      "num_input_tokens_seen": 17516568,
      "step": 26850
    },
    {
      "epoch": 14.074947589098532,
      "grad_norm": 0.3184835612773895,
      "learning_rate": 1.221983478629074e-05,
      "loss": 0.0194,
      "num_input_tokens_seen": 17520440,
      "step": 26855
    },
    {
      "epoch": 14.077568134171909,
      "grad_norm": 1.2519352436065674,
      "learning_rate": 1.22100088373724e-05,
      "loss": 0.0204,
      "num_input_tokens_seen": 17523224,
      "step": 26860
    },
    {
      "epoch": 14.080188679245284,
      "grad_norm": 0.7380864024162292,
      "learning_rate": 1.2200185563974283e-05,
      "loss": 0.0559,
      "num_input_tokens_seen": 17530008,
      "step": 26865
    },
    {
      "epoch": 14.082809224318659,
      "grad_norm": 2.4094345569610596,
      "learning_rate": 1.219036496815132e-05,
      "loss": 0.0183,
      "num_input_tokens_seen": 17533048,
      "step": 26870
    },
    {
      "epoch": 14.085429769392034,
      "grad_norm": 0.4803023636341095,
      "learning_rate": 1.2180547051957863e-05,
      "loss": 0.0123,
      "num_input_tokens_seen": 17535608,
      "step": 26875
    },
    {
      "epoch": 14.08805031446541,
      "grad_norm": 1.3407543897628784,
      "learning_rate": 1.2170731817447723e-05,
      "loss": 0.0141,
      "num_input_tokens_seen": 17539096,
      "step": 26880
    },
    {
      "epoch": 14.090670859538784,
      "grad_norm": 1.0847245454788208,
      "learning_rate": 1.216091926667412e-05,
      "loss": 0.0346,
      "num_input_tokens_seen": 17542936,
      "step": 26885
    },
    {
      "epoch": 14.09329140461216,
      "grad_norm": 5.664044380187988,
      "learning_rate": 1.2151109401689723e-05,
      "loss": 0.0245,
      "num_input_tokens_seen": 17545688,
      "step": 26890
    },
    {
      "epoch": 14.095911949685535,
      "grad_norm": 0.6322547793388367,
      "learning_rate": 1.214130222454665e-05,
      "loss": 0.0184,
      "num_input_tokens_seen": 17548536,
      "step": 26895
    },
    {
      "epoch": 14.09853249475891,
      "grad_norm": 1.4486498832702637,
      "learning_rate": 1.2131497737296452e-05,
      "loss": 0.0153,
      "num_input_tokens_seen": 17552056,
      "step": 26900
    },
    {
      "epoch": 14.101153039832285,
      "grad_norm": 1.9280232191085815,
      "learning_rate": 1.2121695941990124e-05,
      "loss": 0.0203,
      "num_input_tokens_seen": 17555512,
      "step": 26905
    },
    {
      "epoch": 14.10377358490566,
      "grad_norm": 2.7557590007781982,
      "learning_rate": 1.2111896840678075e-05,
      "loss": 0.0177,
      "num_input_tokens_seen": 17559320,
      "step": 26910
    },
    {
      "epoch": 14.106394129979035,
      "grad_norm": 2.2389814853668213,
      "learning_rate": 1.210210043541018e-05,
      "loss": 0.011,
      "num_input_tokens_seen": 17562616,
      "step": 26915
    },
    {
      "epoch": 14.10901467505241,
      "grad_norm": 1.1332120895385742,
      "learning_rate": 1.2092306728235717e-05,
      "loss": 0.0239,
      "num_input_tokens_seen": 17566200,
      "step": 26920
    },
    {
      "epoch": 14.111635220125786,
      "grad_norm": 0.6133902668952942,
      "learning_rate": 1.2082515721203427e-05,
      "loss": 0.0205,
      "num_input_tokens_seen": 17569496,
      "step": 26925
    },
    {
      "epoch": 14.114255765199161,
      "grad_norm": 3.877614974975586,
      "learning_rate": 1.2072727416361479e-05,
      "loss": 0.0166,
      "num_input_tokens_seen": 17571832,
      "step": 26930
    },
    {
      "epoch": 14.116876310272536,
      "grad_norm": 4.961614608764648,
      "learning_rate": 1.2062941815757484e-05,
      "loss": 0.0249,
      "num_input_tokens_seen": 17574744,
      "step": 26935
    },
    {
      "epoch": 14.119496855345911,
      "grad_norm": 1.2995470762252808,
      "learning_rate": 1.2053158921438454e-05,
      "loss": 0.0372,
      "num_input_tokens_seen": 17578232,
      "step": 26940
    },
    {
      "epoch": 14.122117400419286,
      "grad_norm": 2.788172721862793,
      "learning_rate": 1.2043378735450885e-05,
      "loss": 0.0151,
      "num_input_tokens_seen": 17581688,
      "step": 26945
    },
    {
      "epoch": 14.124737945492662,
      "grad_norm": 1.9441455602645874,
      "learning_rate": 1.2033601259840657e-05,
      "loss": 0.0353,
      "num_input_tokens_seen": 17585336,
      "step": 26950
    },
    {
      "epoch": 14.127358490566039,
      "grad_norm": 1.0263170003890991,
      "learning_rate": 1.2023826496653118e-05,
      "loss": 0.0108,
      "num_input_tokens_seen": 17588120,
      "step": 26955
    },
    {
      "epoch": 14.129979035639414,
      "grad_norm": 0.8978033065795898,
      "learning_rate": 1.2014054447933044e-05,
      "loss": 0.0341,
      "num_input_tokens_seen": 17591928,
      "step": 26960
    },
    {
      "epoch": 14.132599580712789,
      "grad_norm": 1.251246690750122,
      "learning_rate": 1.200428511572462e-05,
      "loss": 0.0212,
      "num_input_tokens_seen": 17596120,
      "step": 26965
    },
    {
      "epoch": 14.135220125786164,
      "grad_norm": 2.1124114990234375,
      "learning_rate": 1.1994518502071484e-05,
      "loss": 0.0381,
      "num_input_tokens_seen": 17599416,
      "step": 26970
    },
    {
      "epoch": 14.13784067085954,
      "grad_norm": 1.6051076650619507,
      "learning_rate": 1.1984754609016716e-05,
      "loss": 0.0162,
      "num_input_tokens_seen": 17603192,
      "step": 26975
    },
    {
      "epoch": 14.140461215932914,
      "grad_norm": 1.8219337463378906,
      "learning_rate": 1.1974993438602785e-05,
      "loss": 0.0221,
      "num_input_tokens_seen": 17606520,
      "step": 26980
    },
    {
      "epoch": 14.14308176100629,
      "grad_norm": 3.0235846042633057,
      "learning_rate": 1.1965234992871641e-05,
      "loss": 0.0264,
      "num_input_tokens_seen": 17609112,
      "step": 26985
    },
    {
      "epoch": 14.145702306079665,
      "grad_norm": 2.843628168106079,
      "learning_rate": 1.195547927386462e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 17612440,
      "step": 26990
    },
    {
      "epoch": 14.14832285115304,
      "grad_norm": 3.2911691665649414,
      "learning_rate": 1.1945726283622514e-05,
      "loss": 0.0314,
      "num_input_tokens_seen": 17615384,
      "step": 26995
    },
    {
      "epoch": 14.150943396226415,
      "grad_norm": 0.6744157671928406,
      "learning_rate": 1.1935976024185547e-05,
      "loss": 0.0121,
      "num_input_tokens_seen": 17618456,
      "step": 27000
    },
    {
      "epoch": 14.15356394129979,
      "grad_norm": 2.9726011753082275,
      "learning_rate": 1.1926228497593341e-05,
      "loss": 0.081,
      "num_input_tokens_seen": 17621720,
      "step": 27005
    },
    {
      "epoch": 14.156184486373165,
      "grad_norm": 2.952303409576416,
      "learning_rate": 1.1916483705884985e-05,
      "loss": 0.0338,
      "num_input_tokens_seen": 17625112,
      "step": 27010
    },
    {
      "epoch": 14.15880503144654,
      "grad_norm": 1.053398847579956,
      "learning_rate": 1.1906741651098983e-05,
      "loss": 0.0364,
      "num_input_tokens_seen": 17628824,
      "step": 27015
    },
    {
      "epoch": 14.161425576519916,
      "grad_norm": 0.35252994298934937,
      "learning_rate": 1.1897002335273239e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 17632568,
      "step": 27020
    },
    {
      "epoch": 14.164046121593291,
      "grad_norm": 1.7066487073898315,
      "learning_rate": 1.1887265760445121e-05,
      "loss": 0.0246,
      "num_input_tokens_seen": 17635928,
      "step": 27025
    },
    {
      "epoch": 14.166666666666666,
      "grad_norm": 0.8771137595176697,
      "learning_rate": 1.1877531928651417e-05,
      "loss": 0.0296,
      "num_input_tokens_seen": 17639832,
      "step": 27030
    },
    {
      "epoch": 14.169287211740041,
      "grad_norm": 3.088376760482788,
      "learning_rate": 1.1867800841928313e-05,
      "loss": 0.0172,
      "num_input_tokens_seen": 17643160,
      "step": 27035
    },
    {
      "epoch": 14.171907756813416,
      "grad_norm": 0.9540116786956787,
      "learning_rate": 1.1858072502311465e-05,
      "loss": 0.0516,
      "num_input_tokens_seen": 17646040,
      "step": 27040
    },
    {
      "epoch": 14.174528301886792,
      "grad_norm": 1.318388819694519,
      "learning_rate": 1.1848346911835908e-05,
      "loss": 0.0148,
      "num_input_tokens_seen": 17648600,
      "step": 27045
    },
    {
      "epoch": 14.177148846960169,
      "grad_norm": 0.4658229351043701,
      "learning_rate": 1.1838624072536137e-05,
      "loss": 0.0176,
      "num_input_tokens_seen": 17652024,
      "step": 27050
    },
    {
      "epoch": 14.179769392033544,
      "grad_norm": 0.7263198494911194,
      "learning_rate": 1.1828903986446055e-05,
      "loss": 0.0156,
      "num_input_tokens_seen": 17655160,
      "step": 27055
    },
    {
      "epoch": 14.182389937106919,
      "grad_norm": 1.7073795795440674,
      "learning_rate": 1.1819186655599007e-05,
      "loss": 0.0194,
      "num_input_tokens_seen": 17658712,
      "step": 27060
    },
    {
      "epoch": 14.185010482180294,
      "grad_norm": 0.9275301694869995,
      "learning_rate": 1.1809472082027734e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 17661432,
      "step": 27065
    },
    {
      "epoch": 14.18763102725367,
      "grad_norm": 0.47678253054618835,
      "learning_rate": 1.1799760267764407e-05,
      "loss": 0.0119,
      "num_input_tokens_seen": 17664408,
      "step": 27070
    },
    {
      "epoch": 14.190251572327044,
      "grad_norm": 2.604835271835327,
      "learning_rate": 1.1790051214840634e-05,
      "loss": 0.0101,
      "num_input_tokens_seen": 17667608,
      "step": 27075
    },
    {
      "epoch": 14.19287211740042,
      "grad_norm": 0.9087868332862854,
      "learning_rate": 1.1780344925287443e-05,
      "loss": 0.0306,
      "num_input_tokens_seen": 17670872,
      "step": 27080
    },
    {
      "epoch": 14.195492662473795,
      "grad_norm": 3.3238167762756348,
      "learning_rate": 1.1770641401135282e-05,
      "loss": 0.0329,
      "num_input_tokens_seen": 17673368,
      "step": 27085
    },
    {
      "epoch": 14.19811320754717,
      "grad_norm": 0.8283311724662781,
      "learning_rate": 1.1760940644414003e-05,
      "loss": 0.0168,
      "num_input_tokens_seen": 17675800,
      "step": 27090
    },
    {
      "epoch": 14.200733752620545,
      "grad_norm": 1.2778326272964478,
      "learning_rate": 1.1751242657152914e-05,
      "loss": 0.0193,
      "num_input_tokens_seen": 17679288,
      "step": 27095
    },
    {
      "epoch": 14.20335429769392,
      "grad_norm": 0.46825867891311646,
      "learning_rate": 1.1741547441380698e-05,
      "loss": 0.0201,
      "num_input_tokens_seen": 17682392,
      "step": 27100
    },
    {
      "epoch": 14.205974842767295,
      "grad_norm": 0.8811206817626953,
      "learning_rate": 1.17318549991255e-05,
      "loss": 0.0075,
      "num_input_tokens_seen": 17685176,
      "step": 27105
    },
    {
      "epoch": 14.20859538784067,
      "grad_norm": 1.1181883811950684,
      "learning_rate": 1.1722165332414863e-05,
      "loss": 0.0194,
      "num_input_tokens_seen": 17687768,
      "step": 27110
    },
    {
      "epoch": 14.211215932914046,
      "grad_norm": 1.7437502145767212,
      "learning_rate": 1.1712478443275768e-05,
      "loss": 0.0176,
      "num_input_tokens_seen": 17690712,
      "step": 27115
    },
    {
      "epoch": 14.213836477987421,
      "grad_norm": 1.0805941820144653,
      "learning_rate": 1.170279433373459e-05,
      "loss": 0.0434,
      "num_input_tokens_seen": 17693944,
      "step": 27120
    },
    {
      "epoch": 14.216457023060796,
      "grad_norm": 4.7460856437683105,
      "learning_rate": 1.1693113005817125e-05,
      "loss": 0.0151,
      "num_input_tokens_seen": 17697528,
      "step": 27125
    },
    {
      "epoch": 14.219077568134171,
      "grad_norm": 0.29924482107162476,
      "learning_rate": 1.1683434461548606e-05,
      "loss": 0.0183,
      "num_input_tokens_seen": 17700472,
      "step": 27130
    },
    {
      "epoch": 14.221698113207546,
      "grad_norm": 4.227030277252197,
      "learning_rate": 1.1673758702953675e-05,
      "loss": 0.0211,
      "num_input_tokens_seen": 17704024,
      "step": 27135
    },
    {
      "epoch": 14.224318658280922,
      "grad_norm": 5.5345587730407715,
      "learning_rate": 1.1664085732056398e-05,
      "loss": 0.0159,
      "num_input_tokens_seen": 17706840,
      "step": 27140
    },
    {
      "epoch": 14.226939203354299,
      "grad_norm": 2.814542055130005,
      "learning_rate": 1.1654415550880243e-05,
      "loss": 0.0204,
      "num_input_tokens_seen": 17709592,
      "step": 27145
    },
    {
      "epoch": 14.229559748427674,
      "grad_norm": 0.6831146478652954,
      "learning_rate": 1.1644748161448088e-05,
      "loss": 0.02,
      "num_input_tokens_seen": 17713112,
      "step": 27150
    },
    {
      "epoch": 14.232180293501049,
      "grad_norm": 2.2501540184020996,
      "learning_rate": 1.1635083565782257e-05,
      "loss": 0.0482,
      "num_input_tokens_seen": 17716056,
      "step": 27155
    },
    {
      "epoch": 14.234800838574424,
      "grad_norm": 0.6734315156936646,
      "learning_rate": 1.1625421765904465e-05,
      "loss": 0.0143,
      "num_input_tokens_seen": 17719704,
      "step": 27160
    },
    {
      "epoch": 14.2374213836478,
      "grad_norm": 0.8729897141456604,
      "learning_rate": 1.1615762763835858e-05,
      "loss": 0.0099,
      "num_input_tokens_seen": 17722456,
      "step": 27165
    },
    {
      "epoch": 14.240041928721174,
      "grad_norm": 1.3238526582717896,
      "learning_rate": 1.1606106561596993e-05,
      "loss": 0.0409,
      "num_input_tokens_seen": 17725400,
      "step": 27170
    },
    {
      "epoch": 14.24266247379455,
      "grad_norm": 1.7056773900985718,
      "learning_rate": 1.1596453161207829e-05,
      "loss": 0.0578,
      "num_input_tokens_seen": 17728440,
      "step": 27175
    },
    {
      "epoch": 14.245283018867925,
      "grad_norm": 0.27879711985588074,
      "learning_rate": 1.1586802564687741e-05,
      "loss": 0.0147,
      "num_input_tokens_seen": 17731192,
      "step": 27180
    },
    {
      "epoch": 14.2479035639413,
      "grad_norm": 1.0062317848205566,
      "learning_rate": 1.1577154774055528e-05,
      "loss": 0.0102,
      "num_input_tokens_seen": 17735096,
      "step": 27185
    },
    {
      "epoch": 14.250524109014675,
      "grad_norm": 1.270738124847412,
      "learning_rate": 1.1567509791329401e-05,
      "loss": 0.0238,
      "num_input_tokens_seen": 17739000,
      "step": 27190
    },
    {
      "epoch": 14.25314465408805,
      "grad_norm": 6.15314245223999,
      "learning_rate": 1.155786761852699e-05,
      "loss": 0.0143,
      "num_input_tokens_seen": 17741784,
      "step": 27195
    },
    {
      "epoch": 14.255765199161425,
      "grad_norm": 2.1841328144073486,
      "learning_rate": 1.1548228257665317e-05,
      "loss": 0.0218,
      "num_input_tokens_seen": 17744440,
      "step": 27200
    },
    {
      "epoch": 14.2583857442348,
      "grad_norm": 2.227651834487915,
      "learning_rate": 1.1538591710760815e-05,
      "loss": 0.0909,
      "num_input_tokens_seen": 17747352,
      "step": 27205
    },
    {
      "epoch": 14.261006289308176,
      "grad_norm": 9.962973594665527,
      "learning_rate": 1.1528957979829355e-05,
      "loss": 0.022,
      "num_input_tokens_seen": 17750104,
      "step": 27210
    },
    {
      "epoch": 14.26362683438155,
      "grad_norm": 2.929755449295044,
      "learning_rate": 1.1519327066886198e-05,
      "loss": 0.0112,
      "num_input_tokens_seen": 17752728,
      "step": 27215
    },
    {
      "epoch": 14.266247379454926,
      "grad_norm": 1.0745000839233398,
      "learning_rate": 1.1509698973946032e-05,
      "loss": 0.0108,
      "num_input_tokens_seen": 17756248,
      "step": 27220
    },
    {
      "epoch": 14.268867924528301,
      "grad_norm": 2.1622679233551025,
      "learning_rate": 1.1500073703022926e-05,
      "loss": 0.0205,
      "num_input_tokens_seen": 17759704,
      "step": 27225
    },
    {
      "epoch": 14.271488469601676,
      "grad_norm": 1.0456639528274536,
      "learning_rate": 1.1490451256130394e-05,
      "loss": 0.0174,
      "num_input_tokens_seen": 17763064,
      "step": 27230
    },
    {
      "epoch": 14.274109014675052,
      "grad_norm": 1.7179075479507446,
      "learning_rate": 1.1480831635281328e-05,
      "loss": 0.0358,
      "num_input_tokens_seen": 17767384,
      "step": 27235
    },
    {
      "epoch": 14.276729559748428,
      "grad_norm": 1.168582797050476,
      "learning_rate": 1.1471214842488046e-05,
      "loss": 0.0431,
      "num_input_tokens_seen": 17770328,
      "step": 27240
    },
    {
      "epoch": 14.279350104821804,
      "grad_norm": 0.3024444282054901,
      "learning_rate": 1.1461600879762277e-05,
      "loss": 0.0127,
      "num_input_tokens_seen": 17774072,
      "step": 27245
    },
    {
      "epoch": 14.281970649895179,
      "grad_norm": 0.3360760509967804,
      "learning_rate": 1.1451989749115157e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 17776344,
      "step": 27250
    },
    {
      "epoch": 14.284591194968554,
      "grad_norm": 1.8393678665161133,
      "learning_rate": 1.1442381452557219e-05,
      "loss": 0.0273,
      "num_input_tokens_seen": 17780536,
      "step": 27255
    },
    {
      "epoch": 14.28721174004193,
      "grad_norm": 1.2446191310882568,
      "learning_rate": 1.1432775992098397e-05,
      "loss": 0.0437,
      "num_input_tokens_seen": 17784024,
      "step": 27260
    },
    {
      "epoch": 14.289832285115304,
      "grad_norm": 1.6001527309417725,
      "learning_rate": 1.1423173369748056e-05,
      "loss": 0.0534,
      "num_input_tokens_seen": 17788248,
      "step": 27265
    },
    {
      "epoch": 14.29245283018868,
      "grad_norm": 0.12211021035909653,
      "learning_rate": 1.1413573587514952e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 17790872,
      "step": 27270
    },
    {
      "epoch": 14.295073375262055,
      "grad_norm": 8.129863739013672,
      "learning_rate": 1.140397664740726e-05,
      "loss": 0.03,
      "num_input_tokens_seen": 17793912,
      "step": 27275
    },
    {
      "epoch": 14.29769392033543,
      "grad_norm": 2.96158504486084,
      "learning_rate": 1.1394382551432534e-05,
      "loss": 0.0223,
      "num_input_tokens_seen": 17797336,
      "step": 27280
    },
    {
      "epoch": 14.300314465408805,
      "grad_norm": 0.21423113346099854,
      "learning_rate": 1.1384791301597767e-05,
      "loss": 0.0204,
      "num_input_tokens_seen": 17800728,
      "step": 27285
    },
    {
      "epoch": 14.30293501048218,
      "grad_norm": 0.9901149272918701,
      "learning_rate": 1.1375202899909317e-05,
      "loss": 0.0376,
      "num_input_tokens_seen": 17803448,
      "step": 27290
    },
    {
      "epoch": 14.305555555555555,
      "grad_norm": 1.1598659753799438,
      "learning_rate": 1.1365617348372984e-05,
      "loss": 0.0165,
      "num_input_tokens_seen": 17807896,
      "step": 27295
    },
    {
      "epoch": 14.30817610062893,
      "grad_norm": 1.2024952173233032,
      "learning_rate": 1.135603464899396e-05,
      "loss": 0.0125,
      "num_input_tokens_seen": 17811032,
      "step": 27300
    },
    {
      "epoch": 14.310796645702306,
      "grad_norm": 1.1948177814483643,
      "learning_rate": 1.1346454803776821e-05,
      "loss": 0.0249,
      "num_input_tokens_seen": 17815640,
      "step": 27305
    },
    {
      "epoch": 14.31341719077568,
      "grad_norm": 1.0253020524978638,
      "learning_rate": 1.1336877814725566e-05,
      "loss": 0.0165,
      "num_input_tokens_seen": 17818744,
      "step": 27310
    },
    {
      "epoch": 14.316037735849056,
      "grad_norm": 9.877100944519043,
      "learning_rate": 1.1327303683843607e-05,
      "loss": 0.02,
      "num_input_tokens_seen": 17821432,
      "step": 27315
    },
    {
      "epoch": 14.318658280922431,
      "grad_norm": 1.1692105531692505,
      "learning_rate": 1.1317732413133723e-05,
      "loss": 0.0282,
      "num_input_tokens_seen": 17824312,
      "step": 27320
    },
    {
      "epoch": 14.321278825995806,
      "grad_norm": 0.37494543194770813,
      "learning_rate": 1.1308164004598118e-05,
      "loss": 0.0106,
      "num_input_tokens_seen": 17827288,
      "step": 27325
    },
    {
      "epoch": 14.323899371069182,
      "grad_norm": 1.1950328350067139,
      "learning_rate": 1.129859846023841e-05,
      "loss": 0.0186,
      "num_input_tokens_seen": 17830392,
      "step": 27330
    },
    {
      "epoch": 14.326519916142558,
      "grad_norm": 0.8023034930229187,
      "learning_rate": 1.1289035782055582e-05,
      "loss": 0.0137,
      "num_input_tokens_seen": 17833432,
      "step": 27335
    },
    {
      "epoch": 14.329140461215934,
      "grad_norm": 2.391554832458496,
      "learning_rate": 1.1279475972050044e-05,
      "loss": 0.02,
      "num_input_tokens_seen": 17837112,
      "step": 27340
    },
    {
      "epoch": 14.331761006289309,
      "grad_norm": 1.7782859802246094,
      "learning_rate": 1.1269919032221612e-05,
      "loss": 0.0101,
      "num_input_tokens_seen": 17840088,
      "step": 27345
    },
    {
      "epoch": 14.334381551362684,
      "grad_norm": 0.6574931740760803,
      "learning_rate": 1.1260364964569466e-05,
      "loss": 0.0125,
      "num_input_tokens_seen": 17843000,
      "step": 27350
    },
    {
      "epoch": 14.33700209643606,
      "grad_norm": 0.7587230205535889,
      "learning_rate": 1.125081377109223e-05,
      "loss": 0.0308,
      "num_input_tokens_seen": 17846872,
      "step": 27355
    },
    {
      "epoch": 14.339622641509434,
      "grad_norm": 1.7941786050796509,
      "learning_rate": 1.1241265453787882e-05,
      "loss": 0.032,
      "num_input_tokens_seen": 17850232,
      "step": 27360
    },
    {
      "epoch": 14.34224318658281,
      "grad_norm": 1.9852933883666992,
      "learning_rate": 1.1231720014653837e-05,
      "loss": 0.0973,
      "num_input_tokens_seen": 17854360,
      "step": 27365
    },
    {
      "epoch": 14.344863731656185,
      "grad_norm": 1.3056186437606812,
      "learning_rate": 1.1222177455686896e-05,
      "loss": 0.0219,
      "num_input_tokens_seen": 17858040,
      "step": 27370
    },
    {
      "epoch": 14.34748427672956,
      "grad_norm": 1.0927456617355347,
      "learning_rate": 1.1212637778883237e-05,
      "loss": 0.0131,
      "num_input_tokens_seen": 17861400,
      "step": 27375
    },
    {
      "epoch": 14.350104821802935,
      "grad_norm": 1.9620983600616455,
      "learning_rate": 1.1203100986238468e-05,
      "loss": 0.1199,
      "num_input_tokens_seen": 17865176,
      "step": 27380
    },
    {
      "epoch": 14.35272536687631,
      "grad_norm": 2.2370693683624268,
      "learning_rate": 1.1193567079747561e-05,
      "loss": 0.0194,
      "num_input_tokens_seen": 17867544,
      "step": 27385
    },
    {
      "epoch": 14.355345911949685,
      "grad_norm": 1.0261180400848389,
      "learning_rate": 1.1184036061404907e-05,
      "loss": 0.019,
      "num_input_tokens_seen": 17870584,
      "step": 27390
    },
    {
      "epoch": 14.35796645702306,
      "grad_norm": 0.11974847316741943,
      "learning_rate": 1.117450793320429e-05,
      "loss": 0.0128,
      "num_input_tokens_seen": 17873560,
      "step": 27395
    },
    {
      "epoch": 14.360587002096436,
      "grad_norm": 0.20178401470184326,
      "learning_rate": 1.1164982697138893e-05,
      "loss": 0.0311,
      "num_input_tokens_seen": 17876824,
      "step": 27400
    },
    {
      "epoch": 14.36320754716981,
      "grad_norm": 3.4733822345733643,
      "learning_rate": 1.1155460355201267e-05,
      "loss": 0.0224,
      "num_input_tokens_seen": 17879928,
      "step": 27405
    },
    {
      "epoch": 14.365828092243186,
      "grad_norm": 1.9183067083358765,
      "learning_rate": 1.1145940909383399e-05,
      "loss": 0.0328,
      "num_input_tokens_seen": 17883672,
      "step": 27410
    },
    {
      "epoch": 14.368448637316561,
      "grad_norm": 0.5597550272941589,
      "learning_rate": 1.1136424361676626e-05,
      "loss": 0.018,
      "num_input_tokens_seen": 17887032,
      "step": 27415
    },
    {
      "epoch": 14.371069182389936,
      "grad_norm": 0.372020959854126,
      "learning_rate": 1.1126910714071711e-05,
      "loss": 0.014,
      "num_input_tokens_seen": 17889816,
      "step": 27420
    },
    {
      "epoch": 14.373689727463312,
      "grad_norm": 1.0792667865753174,
      "learning_rate": 1.1117399968558806e-05,
      "loss": 0.0311,
      "num_input_tokens_seen": 17893016,
      "step": 27425
    },
    {
      "epoch": 14.376310272536688,
      "grad_norm": 2.588740110397339,
      "learning_rate": 1.1107892127127454e-05,
      "loss": 0.0188,
      "num_input_tokens_seen": 17896056,
      "step": 27430
    },
    {
      "epoch": 14.378930817610064,
      "grad_norm": 1.2901204824447632,
      "learning_rate": 1.109838719176658e-05,
      "loss": 0.0299,
      "num_input_tokens_seen": 17899224,
      "step": 27435
    },
    {
      "epoch": 14.381551362683439,
      "grad_norm": 3.727252244949341,
      "learning_rate": 1.1088885164464497e-05,
      "loss": 0.0324,
      "num_input_tokens_seen": 17902840,
      "step": 27440
    },
    {
      "epoch": 14.384171907756814,
      "grad_norm": 2.190955638885498,
      "learning_rate": 1.1079386047208932e-05,
      "loss": 0.0173,
      "num_input_tokens_seen": 17905272,
      "step": 27445
    },
    {
      "epoch": 14.38679245283019,
      "grad_norm": 1.2636809349060059,
      "learning_rate": 1.1069889841986991e-05,
      "loss": 0.0258,
      "num_input_tokens_seen": 17907992,
      "step": 27450
    },
    {
      "epoch": 14.389412997903564,
      "grad_norm": 2.23705792427063,
      "learning_rate": 1.1060396550785182e-05,
      "loss": 0.0139,
      "num_input_tokens_seen": 17910904,
      "step": 27455
    },
    {
      "epoch": 14.39203354297694,
      "grad_norm": 0.5826036930084229,
      "learning_rate": 1.1050906175589382e-05,
      "loss": 0.0233,
      "num_input_tokens_seen": 17914904,
      "step": 27460
    },
    {
      "epoch": 14.394654088050315,
      "grad_norm": 5.5793070793151855,
      "learning_rate": 1.1041418718384863e-05,
      "loss": 0.0194,
      "num_input_tokens_seen": 17918104,
      "step": 27465
    },
    {
      "epoch": 14.39727463312369,
      "grad_norm": 0.7730926275253296,
      "learning_rate": 1.1031934181156298e-05,
      "loss": 0.024,
      "num_input_tokens_seen": 17921720,
      "step": 27470
    },
    {
      "epoch": 14.399895178197065,
      "grad_norm": 0.5185240507125854,
      "learning_rate": 1.1022452565887747e-05,
      "loss": 0.037,
      "num_input_tokens_seen": 17924984,
      "step": 27475
    },
    {
      "epoch": 14.40251572327044,
      "grad_norm": 8.940163612365723,
      "learning_rate": 1.1012973874562657e-05,
      "loss": 0.0257,
      "num_input_tokens_seen": 17927672,
      "step": 27480
    },
    {
      "epoch": 14.405136268343815,
      "grad_norm": 0.925689160823822,
      "learning_rate": 1.1003498109163865e-05,
      "loss": 0.0533,
      "num_input_tokens_seen": 17931128,
      "step": 27485
    },
    {
      "epoch": 14.40775681341719,
      "grad_norm": 2.5385055541992188,
      "learning_rate": 1.0994025271673592e-05,
      "loss": 0.0198,
      "num_input_tokens_seen": 17934936,
      "step": 27490
    },
    {
      "epoch": 14.410377358490566,
      "grad_norm": 0.03166229650378227,
      "learning_rate": 1.0984555364073431e-05,
      "loss": 0.0094,
      "num_input_tokens_seen": 17937432,
      "step": 27495
    },
    {
      "epoch": 14.41299790356394,
      "grad_norm": 0.7979415655136108,
      "learning_rate": 1.0975088388344393e-05,
      "loss": 0.0118,
      "num_input_tokens_seen": 17939800,
      "step": 27500
    },
    {
      "epoch": 14.415618448637316,
      "grad_norm": 1.0502289533615112,
      "learning_rate": 1.096562434646686e-05,
      "loss": 0.0248,
      "num_input_tokens_seen": 17942808,
      "step": 27505
    },
    {
      "epoch": 14.418238993710691,
      "grad_norm": 0.37259602546691895,
      "learning_rate": 1.0956163240420611e-05,
      "loss": 0.0133,
      "num_input_tokens_seen": 17945592,
      "step": 27510
    },
    {
      "epoch": 14.420859538784066,
      "grad_norm": 0.22426919639110565,
      "learning_rate": 1.094670507218479e-05,
      "loss": 0.02,
      "num_input_tokens_seen": 17948536,
      "step": 27515
    },
    {
      "epoch": 14.423480083857442,
      "grad_norm": 1.4555414915084839,
      "learning_rate": 1.093724984373793e-05,
      "loss": 0.0152,
      "num_input_tokens_seen": 17952408,
      "step": 27520
    },
    {
      "epoch": 14.426100628930818,
      "grad_norm": 1.2339136600494385,
      "learning_rate": 1.0927797557057968e-05,
      "loss": 0.0423,
      "num_input_tokens_seen": 17955896,
      "step": 27525
    },
    {
      "epoch": 14.428721174004194,
      "grad_norm": 1.6913737058639526,
      "learning_rate": 1.0918348214122209e-05,
      "loss": 0.0431,
      "num_input_tokens_seen": 17963128,
      "step": 27530
    },
    {
      "epoch": 14.431341719077569,
      "grad_norm": 1.7053558826446533,
      "learning_rate": 1.0908901816907366e-05,
      "loss": 0.0164,
      "num_input_tokens_seen": 17965688,
      "step": 27535
    },
    {
      "epoch": 14.433962264150944,
      "grad_norm": 1.8420445919036865,
      "learning_rate": 1.08994583673895e-05,
      "loss": 0.0151,
      "num_input_tokens_seen": 17968888,
      "step": 27540
    },
    {
      "epoch": 14.43658280922432,
      "grad_norm": 0.6116586327552795,
      "learning_rate": 1.0890017867544068e-05,
      "loss": 0.0157,
      "num_input_tokens_seen": 17972120,
      "step": 27545
    },
    {
      "epoch": 14.439203354297694,
      "grad_norm": 4.4134063720703125,
      "learning_rate": 1.088058031934592e-05,
      "loss": 0.0278,
      "num_input_tokens_seen": 17975608,
      "step": 27550
    },
    {
      "epoch": 14.44182389937107,
      "grad_norm": 1.6900243759155273,
      "learning_rate": 1.087114572476929e-05,
      "loss": 0.0375,
      "num_input_tokens_seen": 17979128,
      "step": 27555
    },
    {
      "epoch": 14.444444444444445,
      "grad_norm": 6.009767055511475,
      "learning_rate": 1.0861714085787783e-05,
      "loss": 0.0466,
      "num_input_tokens_seen": 17982456,
      "step": 27560
    },
    {
      "epoch": 14.44706498951782,
      "grad_norm": 2.025533437728882,
      "learning_rate": 1.0852285404374402e-05,
      "loss": 0.0274,
      "num_input_tokens_seen": 17985720,
      "step": 27565
    },
    {
      "epoch": 14.449685534591195,
      "grad_norm": 1.0022780895233154,
      "learning_rate": 1.0842859682501505e-05,
      "loss": 0.0243,
      "num_input_tokens_seen": 17989272,
      "step": 27570
    },
    {
      "epoch": 14.45230607966457,
      "grad_norm": 0.8613561987876892,
      "learning_rate": 1.0833436922140841e-05,
      "loss": 0.0329,
      "num_input_tokens_seen": 17992408,
      "step": 27575
    },
    {
      "epoch": 14.454926624737945,
      "grad_norm": 0.6024644374847412,
      "learning_rate": 1.0824017125263555e-05,
      "loss": 0.0243,
      "num_input_tokens_seen": 17996568,
      "step": 27580
    },
    {
      "epoch": 14.45754716981132,
      "grad_norm": 1.6387990713119507,
      "learning_rate": 1.0814600293840154e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 17999832,
      "step": 27585
    },
    {
      "epoch": 14.460167714884696,
      "grad_norm": 2.519949436187744,
      "learning_rate": 1.080518642984055e-05,
      "loss": 0.0302,
      "num_input_tokens_seen": 18003704,
      "step": 27590
    },
    {
      "epoch": 14.46278825995807,
      "grad_norm": 1.5236082077026367,
      "learning_rate": 1.0795775535233988e-05,
      "loss": 0.0167,
      "num_input_tokens_seen": 18006328,
      "step": 27595
    },
    {
      "epoch": 14.465408805031446,
      "grad_norm": 2.262179136276245,
      "learning_rate": 1.0786367611989145e-05,
      "loss": 0.0197,
      "num_input_tokens_seen": 18010008,
      "step": 27600
    },
    {
      "epoch": 14.468029350104821,
      "grad_norm": 0.2763521075248718,
      "learning_rate": 1.0776962662074031e-05,
      "loss": 0.0341,
      "num_input_tokens_seen": 18013880,
      "step": 27605
    },
    {
      "epoch": 14.470649895178196,
      "grad_norm": 1.2433637380599976,
      "learning_rate": 1.0767560687456062e-05,
      "loss": 0.0189,
      "num_input_tokens_seen": 18016824,
      "step": 27610
    },
    {
      "epoch": 14.473270440251572,
      "grad_norm": 9.657166481018066,
      "learning_rate": 1.0758161690102036e-05,
      "loss": 0.0131,
      "num_input_tokens_seen": 18020472,
      "step": 27615
    },
    {
      "epoch": 14.475890985324948,
      "grad_norm": 0.14666646718978882,
      "learning_rate": 1.0748765671978095e-05,
      "loss": 0.0112,
      "num_input_tokens_seen": 18023128,
      "step": 27620
    },
    {
      "epoch": 14.478511530398324,
      "grad_norm": 0.7061927914619446,
      "learning_rate": 1.0739372635049797e-05,
      "loss": 0.0377,
      "num_input_tokens_seen": 18026872,
      "step": 27625
    },
    {
      "epoch": 14.481132075471699,
      "grad_norm": 1.389777421951294,
      "learning_rate": 1.0729982581282042e-05,
      "loss": 0.0318,
      "num_input_tokens_seen": 18030200,
      "step": 27630
    },
    {
      "epoch": 14.483752620545074,
      "grad_norm": 2.545889377593994,
      "learning_rate": 1.0720595512639129e-05,
      "loss": 0.0156,
      "num_input_tokens_seen": 18033528,
      "step": 27635
    },
    {
      "epoch": 14.48637316561845,
      "grad_norm": 0.8524379730224609,
      "learning_rate": 1.0711211431084728e-05,
      "loss": 0.0203,
      "num_input_tokens_seen": 18036088,
      "step": 27640
    },
    {
      "epoch": 14.488993710691824,
      "grad_norm": 0.5316661596298218,
      "learning_rate": 1.0701830338581887e-05,
      "loss": 0.0143,
      "num_input_tokens_seen": 18040408,
      "step": 27645
    },
    {
      "epoch": 14.4916142557652,
      "grad_norm": 3.7487142086029053,
      "learning_rate": 1.069245223709301e-05,
      "loss": 0.0129,
      "num_input_tokens_seen": 18043096,
      "step": 27650
    },
    {
      "epoch": 14.494234800838575,
      "grad_norm": 2.3164966106414795,
      "learning_rate": 1.0683077128579904e-05,
      "loss": 0.0257,
      "num_input_tokens_seen": 18046232,
      "step": 27655
    },
    {
      "epoch": 14.49685534591195,
      "grad_norm": 2.698127031326294,
      "learning_rate": 1.0673705015003718e-05,
      "loss": 0.0675,
      "num_input_tokens_seen": 18048792,
      "step": 27660
    },
    {
      "epoch": 14.499475890985325,
      "grad_norm": 1.4490184783935547,
      "learning_rate": 1.0664335898325002e-05,
      "loss": 0.1265,
      "num_input_tokens_seen": 18052696,
      "step": 27665
    },
    {
      "epoch": 14.5020964360587,
      "grad_norm": 0.4232819676399231,
      "learning_rate": 1.065496978050367e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 18055256,
      "step": 27670
    },
    {
      "epoch": 14.504716981132075,
      "grad_norm": 2.9337072372436523,
      "learning_rate": 1.0645606663499e-05,
      "loss": 0.0149,
      "num_input_tokens_seen": 18057784,
      "step": 27675
    },
    {
      "epoch": 14.50733752620545,
      "grad_norm": 1.582358956336975,
      "learning_rate": 1.0636246549269649e-05,
      "loss": 0.0245,
      "num_input_tokens_seen": 18060920,
      "step": 27680
    },
    {
      "epoch": 14.509958071278826,
      "grad_norm": 1.6294118165969849,
      "learning_rate": 1.062688943977366e-05,
      "loss": 0.0152,
      "num_input_tokens_seen": 18064184,
      "step": 27685
    },
    {
      "epoch": 14.5125786163522,
      "grad_norm": 1.2895948886871338,
      "learning_rate": 1.0617535336968413e-05,
      "loss": 0.0422,
      "num_input_tokens_seen": 18067672,
      "step": 27690
    },
    {
      "epoch": 14.515199161425576,
      "grad_norm": 0.6804348230361938,
      "learning_rate": 1.0608184242810698e-05,
      "loss": 0.0386,
      "num_input_tokens_seen": 18071032,
      "step": 27695
    },
    {
      "epoch": 14.517819706498951,
      "grad_norm": 1.3767642974853516,
      "learning_rate": 1.0598836159256642e-05,
      "loss": 0.0369,
      "num_input_tokens_seen": 18074200,
      "step": 27700
    },
    {
      "epoch": 14.520440251572326,
      "grad_norm": 1.7261306047439575,
      "learning_rate": 1.0589491088261763e-05,
      "loss": 0.0155,
      "num_input_tokens_seen": 18076824,
      "step": 27705
    },
    {
      "epoch": 14.523060796645701,
      "grad_norm": 0.42320188879966736,
      "learning_rate": 1.0580149031780945e-05,
      "loss": 0.0312,
      "num_input_tokens_seen": 18079512,
      "step": 27710
    },
    {
      "epoch": 14.525681341719078,
      "grad_norm": 0.6634547710418701,
      "learning_rate": 1.0570809991768447e-05,
      "loss": 0.0225,
      "num_input_tokens_seen": 18082424,
      "step": 27715
    },
    {
      "epoch": 14.528301886792454,
      "grad_norm": 0.9681453704833984,
      "learning_rate": 1.0561473970177873e-05,
      "loss": 0.0162,
      "num_input_tokens_seen": 18085400,
      "step": 27720
    },
    {
      "epoch": 14.530922431865829,
      "grad_norm": 3.81667160987854,
      "learning_rate": 1.0552140968962232e-05,
      "loss": 0.1109,
      "num_input_tokens_seen": 18088664,
      "step": 27725
    },
    {
      "epoch": 14.533542976939204,
      "grad_norm": 5.608269691467285,
      "learning_rate": 1.054281099007386e-05,
      "loss": 0.0326,
      "num_input_tokens_seen": 18091512,
      "step": 27730
    },
    {
      "epoch": 14.536163522012579,
      "grad_norm": 2.739961624145508,
      "learning_rate": 1.0533484035464492e-05,
      "loss": 0.0275,
      "num_input_tokens_seen": 18093752,
      "step": 27735
    },
    {
      "epoch": 14.538784067085954,
      "grad_norm": 1.603783369064331,
      "learning_rate": 1.0524160107085231e-05,
      "loss": 0.0113,
      "num_input_tokens_seen": 18096440,
      "step": 27740
    },
    {
      "epoch": 14.54140461215933,
      "grad_norm": 1.6871227025985718,
      "learning_rate": 1.0514839206886518e-05,
      "loss": 0.0146,
      "num_input_tokens_seen": 18099384,
      "step": 27745
    },
    {
      "epoch": 14.544025157232705,
      "grad_norm": 11.770313262939453,
      "learning_rate": 1.0505521336818195e-05,
      "loss": 0.0278,
      "num_input_tokens_seen": 18102744,
      "step": 27750
    },
    {
      "epoch": 14.54664570230608,
      "grad_norm": 1.026793360710144,
      "learning_rate": 1.0496206498829438e-05,
      "loss": 0.0141,
      "num_input_tokens_seen": 18105720,
      "step": 27755
    },
    {
      "epoch": 14.549266247379455,
      "grad_norm": 1.0639699697494507,
      "learning_rate": 1.0486894694868814e-05,
      "loss": 0.0126,
      "num_input_tokens_seen": 18108536,
      "step": 27760
    },
    {
      "epoch": 14.55188679245283,
      "grad_norm": 5.300747394561768,
      "learning_rate": 1.0477585926884246e-05,
      "loss": 0.0237,
      "num_input_tokens_seen": 18110904,
      "step": 27765
    },
    {
      "epoch": 14.554507337526205,
      "grad_norm": 1.0457435846328735,
      "learning_rate": 1.046828019682303e-05,
      "loss": 0.0176,
      "num_input_tokens_seen": 18113944,
      "step": 27770
    },
    {
      "epoch": 14.55712788259958,
      "grad_norm": 1.5992213487625122,
      "learning_rate": 1.0458977506631808e-05,
      "loss": 0.0378,
      "num_input_tokens_seen": 18117304,
      "step": 27775
    },
    {
      "epoch": 14.559748427672956,
      "grad_norm": 2.1068167686462402,
      "learning_rate": 1.0449677858256593e-05,
      "loss": 0.0317,
      "num_input_tokens_seen": 18119704,
      "step": 27780
    },
    {
      "epoch": 14.56236897274633,
      "grad_norm": 0.9547370672225952,
      "learning_rate": 1.0440381253642773e-05,
      "loss": 0.0608,
      "num_input_tokens_seen": 18124312,
      "step": 27785
    },
    {
      "epoch": 14.564989517819706,
      "grad_norm": 1.1822608709335327,
      "learning_rate": 1.0431087694735087e-05,
      "loss": 0.0203,
      "num_input_tokens_seen": 18128408,
      "step": 27790
    },
    {
      "epoch": 14.567610062893081,
      "grad_norm": 0.3676619529724121,
      "learning_rate": 1.0421797183477644e-05,
      "loss": 0.019,
      "num_input_tokens_seen": 18131416,
      "step": 27795
    },
    {
      "epoch": 14.570230607966456,
      "grad_norm": 1.0448894500732422,
      "learning_rate": 1.041250972181392e-05,
      "loss": 0.016,
      "num_input_tokens_seen": 18134872,
      "step": 27800
    },
    {
      "epoch": 14.572851153039831,
      "grad_norm": 0.5662185549736023,
      "learning_rate": 1.0403225311686743e-05,
      "loss": 0.0106,
      "num_input_tokens_seen": 18138776,
      "step": 27805
    },
    {
      "epoch": 14.575471698113208,
      "grad_norm": 3.7168941497802734,
      "learning_rate": 1.0393943955038293e-05,
      "loss": 0.0243,
      "num_input_tokens_seen": 18141752,
      "step": 27810
    },
    {
      "epoch": 14.578092243186584,
      "grad_norm": 1.434875726699829,
      "learning_rate": 1.038466565381013e-05,
      "loss": 0.0146,
      "num_input_tokens_seen": 18144152,
      "step": 27815
    },
    {
      "epoch": 14.580712788259959,
      "grad_norm": 1.1184277534484863,
      "learning_rate": 1.0375390409943175e-05,
      "loss": 0.0281,
      "num_input_tokens_seen": 18147192,
      "step": 27820
    },
    {
      "epoch": 14.583333333333334,
      "grad_norm": 0.6043213605880737,
      "learning_rate": 1.0366118225377708e-05,
      "loss": 0.0172,
      "num_input_tokens_seen": 18150456,
      "step": 27825
    },
    {
      "epoch": 14.585953878406709,
      "grad_norm": 0.5908741354942322,
      "learning_rate": 1.0356849102053353e-05,
      "loss": 0.0136,
      "num_input_tokens_seen": 18153464,
      "step": 27830
    },
    {
      "epoch": 14.588574423480084,
      "grad_norm": 2.4351704120635986,
      "learning_rate": 1.0347583041909104e-05,
      "loss": 0.0218,
      "num_input_tokens_seen": 18157336,
      "step": 27835
    },
    {
      "epoch": 14.59119496855346,
      "grad_norm": 0.5027487277984619,
      "learning_rate": 1.0338320046883315e-05,
      "loss": 0.0145,
      "num_input_tokens_seen": 18161016,
      "step": 27840
    },
    {
      "epoch": 14.593815513626835,
      "grad_norm": 1.6808381080627441,
      "learning_rate": 1.0329060118913705e-05,
      "loss": 0.0098,
      "num_input_tokens_seen": 18163480,
      "step": 27845
    },
    {
      "epoch": 14.59643605870021,
      "grad_norm": 1.6451201438903809,
      "learning_rate": 1.0319803259937353e-05,
      "loss": 0.0159,
      "num_input_tokens_seen": 18166584,
      "step": 27850
    },
    {
      "epoch": 14.599056603773585,
      "grad_norm": 7.596397399902344,
      "learning_rate": 1.0310549471890676e-05,
      "loss": 0.0257,
      "num_input_tokens_seen": 18169144,
      "step": 27855
    },
    {
      "epoch": 14.60167714884696,
      "grad_norm": 2.7386627197265625,
      "learning_rate": 1.0301298756709454e-05,
      "loss": 0.0282,
      "num_input_tokens_seen": 18172184,
      "step": 27860
    },
    {
      "epoch": 14.604297693920335,
      "grad_norm": 1.664441704750061,
      "learning_rate": 1.0292051116328844e-05,
      "loss": 0.0144,
      "num_input_tokens_seen": 18174840,
      "step": 27865
    },
    {
      "epoch": 14.60691823899371,
      "grad_norm": 1.057076334953308,
      "learning_rate": 1.028280655268334e-05,
      "loss": 0.0352,
      "num_input_tokens_seen": 18178584,
      "step": 27870
    },
    {
      "epoch": 14.609538784067086,
      "grad_norm": 1.3292350769042969,
      "learning_rate": 1.0273565067706806e-05,
      "loss": 0.0296,
      "num_input_tokens_seen": 18184792,
      "step": 27875
    },
    {
      "epoch": 14.61215932914046,
      "grad_norm": 1.0617148876190186,
      "learning_rate": 1.0264326663332457e-05,
      "loss": 0.0129,
      "num_input_tokens_seen": 18188248,
      "step": 27880
    },
    {
      "epoch": 14.614779874213836,
      "grad_norm": 1.2616889476776123,
      "learning_rate": 1.0255091341492859e-05,
      "loss": 0.0098,
      "num_input_tokens_seen": 18191160,
      "step": 27885
    },
    {
      "epoch": 14.617400419287211,
      "grad_norm": 1.5450024604797363,
      "learning_rate": 1.0245859104119923e-05,
      "loss": 0.0261,
      "num_input_tokens_seen": 18195000,
      "step": 27890
    },
    {
      "epoch": 14.620020964360586,
      "grad_norm": 1.9881815910339355,
      "learning_rate": 1.0236629953144938e-05,
      "loss": 0.011,
      "num_input_tokens_seen": 18198232,
      "step": 27895
    },
    {
      "epoch": 14.622641509433961,
      "grad_norm": 3.0363738536834717,
      "learning_rate": 1.0227403890498537e-05,
      "loss": 0.0302,
      "num_input_tokens_seen": 18200856,
      "step": 27900
    },
    {
      "epoch": 14.625262054507338,
      "grad_norm": 0.55893474817276,
      "learning_rate": 1.021818091811072e-05,
      "loss": 0.0167,
      "num_input_tokens_seen": 18203256,
      "step": 27905
    },
    {
      "epoch": 14.627882599580714,
      "grad_norm": 3.1429107189178467,
      "learning_rate": 1.0208961037910813e-05,
      "loss": 0.0448,
      "num_input_tokens_seen": 18206008,
      "step": 27910
    },
    {
      "epoch": 14.630503144654089,
      "grad_norm": 2.4467968940734863,
      "learning_rate": 1.0199744251827503e-05,
      "loss": 0.0222,
      "num_input_tokens_seen": 18208984,
      "step": 27915
    },
    {
      "epoch": 14.633123689727464,
      "grad_norm": 1.3891220092773438,
      "learning_rate": 1.0190530561788845e-05,
      "loss": 0.0101,
      "num_input_tokens_seen": 18213848,
      "step": 27920
    },
    {
      "epoch": 14.635744234800839,
      "grad_norm": 2.0301802158355713,
      "learning_rate": 1.0181319969722236e-05,
      "loss": 0.0307,
      "num_input_tokens_seen": 18217496,
      "step": 27925
    },
    {
      "epoch": 14.638364779874214,
      "grad_norm": 2.8422224521636963,
      "learning_rate": 1.0172112477554441e-05,
      "loss": 0.0171,
      "num_input_tokens_seen": 18220536,
      "step": 27930
    },
    {
      "epoch": 14.64098532494759,
      "grad_norm": 2.4089391231536865,
      "learning_rate": 1.016290808721154e-05,
      "loss": 0.0321,
      "num_input_tokens_seen": 18224888,
      "step": 27935
    },
    {
      "epoch": 14.643605870020965,
      "grad_norm": 3.663832187652588,
      "learning_rate": 1.0153706800619007e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 18229368,
      "step": 27940
    },
    {
      "epoch": 14.64622641509434,
      "grad_norm": 3.268555164337158,
      "learning_rate": 1.0144508619701623e-05,
      "loss": 0.056,
      "num_input_tokens_seen": 18232760,
      "step": 27945
    },
    {
      "epoch": 14.648846960167715,
      "grad_norm": 1.7134757041931152,
      "learning_rate": 1.013531354638356e-05,
      "loss": 0.0261,
      "num_input_tokens_seen": 18235576,
      "step": 27950
    },
    {
      "epoch": 14.65146750524109,
      "grad_norm": 0.8048228621482849,
      "learning_rate": 1.0126121582588316e-05,
      "loss": 0.013,
      "num_input_tokens_seen": 18239224,
      "step": 27955
    },
    {
      "epoch": 14.654088050314465,
      "grad_norm": 1.4218847751617432,
      "learning_rate": 1.0116932730238754e-05,
      "loss": 0.0274,
      "num_input_tokens_seen": 18241912,
      "step": 27960
    },
    {
      "epoch": 14.65670859538784,
      "grad_norm": 0.8134477138519287,
      "learning_rate": 1.0107746991257063e-05,
      "loss": 0.0138,
      "num_input_tokens_seen": 18244632,
      "step": 27965
    },
    {
      "epoch": 14.659329140461216,
      "grad_norm": 5.975481986999512,
      "learning_rate": 1.0098564367564811e-05,
      "loss": 0.0134,
      "num_input_tokens_seen": 18247768,
      "step": 27970
    },
    {
      "epoch": 14.66194968553459,
      "grad_norm": 0.05016527697443962,
      "learning_rate": 1.0089384861082876e-05,
      "loss": 0.0229,
      "num_input_tokens_seen": 18250456,
      "step": 27975
    },
    {
      "epoch": 14.664570230607966,
      "grad_norm": 1.3868519067764282,
      "learning_rate": 1.0080208473731523e-05,
      "loss": 0.0135,
      "num_input_tokens_seen": 18253688,
      "step": 27980
    },
    {
      "epoch": 14.667190775681341,
      "grad_norm": 0.8073798418045044,
      "learning_rate": 1.0071035207430352e-05,
      "loss": 0.0233,
      "num_input_tokens_seen": 18256952,
      "step": 27985
    },
    {
      "epoch": 14.669811320754716,
      "grad_norm": 1.006683349609375,
      "learning_rate": 1.0061865064098285e-05,
      "loss": 0.016,
      "num_input_tokens_seen": 18263064,
      "step": 27990
    },
    {
      "epoch": 14.672431865828091,
      "grad_norm": 1.702843189239502,
      "learning_rate": 1.0052698045653625e-05,
      "loss": 0.0147,
      "num_input_tokens_seen": 18265848,
      "step": 27995
    },
    {
      "epoch": 14.675052410901468,
      "grad_norm": 1.051212191581726,
      "learning_rate": 1.0043534154014015e-05,
      "loss": 0.0441,
      "num_input_tokens_seen": 18269208,
      "step": 28000
    },
    {
      "epoch": 14.677672955974844,
      "grad_norm": 1.1436634063720703,
      "learning_rate": 1.003437339109642e-05,
      "loss": 0.0285,
      "num_input_tokens_seen": 18274136,
      "step": 28005
    },
    {
      "epoch": 14.680293501048219,
      "grad_norm": 1.1823686361312866,
      "learning_rate": 1.0025215758817183e-05,
      "loss": 0.0295,
      "num_input_tokens_seen": 18277016,
      "step": 28010
    },
    {
      "epoch": 14.682914046121594,
      "grad_norm": 1.351428508758545,
      "learning_rate": 1.0016061259091958e-05,
      "loss": 0.017,
      "num_input_tokens_seen": 18279640,
      "step": 28015
    },
    {
      "epoch": 14.685534591194969,
      "grad_norm": 0.4405134618282318,
      "learning_rate": 1.000690989383577e-05,
      "loss": 0.0128,
      "num_input_tokens_seen": 18281816,
      "step": 28020
    },
    {
      "epoch": 14.688155136268344,
      "grad_norm": 0.4994393289089203,
      "learning_rate": 9.997761664962993e-06,
      "loss": 0.0473,
      "num_input_tokens_seen": 18284568,
      "step": 28025
    },
    {
      "epoch": 14.69077568134172,
      "grad_norm": 1.1089156866073608,
      "learning_rate": 9.988616574387314e-06,
      "loss": 0.0293,
      "num_input_tokens_seen": 18288792,
      "step": 28030
    },
    {
      "epoch": 14.693396226415095,
      "grad_norm": 0.8943035006523132,
      "learning_rate": 9.979474624021787e-06,
      "loss": 0.0343,
      "num_input_tokens_seen": 18292344,
      "step": 28035
    },
    {
      "epoch": 14.69601677148847,
      "grad_norm": 1.1961811780929565,
      "learning_rate": 9.970335815778815e-06,
      "loss": 0.0233,
      "num_input_tokens_seen": 18296184,
      "step": 28040
    },
    {
      "epoch": 14.698637316561845,
      "grad_norm": 0.3692907989025116,
      "learning_rate": 9.96120015157011e-06,
      "loss": 0.0359,
      "num_input_tokens_seen": 18299960,
      "step": 28045
    },
    {
      "epoch": 14.70125786163522,
      "grad_norm": 1.17764413356781,
      "learning_rate": 9.952067633306767e-06,
      "loss": 0.0155,
      "num_input_tokens_seen": 18304024,
      "step": 28050
    },
    {
      "epoch": 14.703878406708595,
      "grad_norm": 1.7127330303192139,
      "learning_rate": 9.942938262899204e-06,
      "loss": 0.0351,
      "num_input_tokens_seen": 18306808,
      "step": 28055
    },
    {
      "epoch": 14.70649895178197,
      "grad_norm": 0.9429686665534973,
      "learning_rate": 9.93381204225717e-06,
      "loss": 0.0232,
      "num_input_tokens_seen": 18308984,
      "step": 28060
    },
    {
      "epoch": 14.709119496855346,
      "grad_norm": 0.5683498978614807,
      "learning_rate": 9.924688973289782e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 18311416,
      "step": 28065
    },
    {
      "epoch": 14.71174004192872,
      "grad_norm": 0.40348154306411743,
      "learning_rate": 9.915569057905463e-06,
      "loss": 0.0435,
      "num_input_tokens_seen": 18314552,
      "step": 28070
    },
    {
      "epoch": 14.714360587002096,
      "grad_norm": 0.8719242811203003,
      "learning_rate": 9.906452298012003e-06,
      "loss": 0.021,
      "num_input_tokens_seen": 18317752,
      "step": 28075
    },
    {
      "epoch": 14.716981132075471,
      "grad_norm": 1.376915454864502,
      "learning_rate": 9.897338695516529e-06,
      "loss": 0.0405,
      "num_input_tokens_seen": 18320888,
      "step": 28080
    },
    {
      "epoch": 14.719601677148846,
      "grad_norm": 1.8359633684158325,
      "learning_rate": 9.888228252325507e-06,
      "loss": 0.0343,
      "num_input_tokens_seen": 18324600,
      "step": 28085
    },
    {
      "epoch": 14.722222222222221,
      "grad_norm": 10.793830871582031,
      "learning_rate": 9.87912097034473e-06,
      "loss": 0.0194,
      "num_input_tokens_seen": 18328184,
      "step": 28090
    },
    {
      "epoch": 14.724842767295598,
      "grad_norm": 1.3505613803863525,
      "learning_rate": 9.870016851479327e-06,
      "loss": 0.0211,
      "num_input_tokens_seen": 18330680,
      "step": 28095
    },
    {
      "epoch": 14.727463312368974,
      "grad_norm": 0.9174051284790039,
      "learning_rate": 9.860915897633791e-06,
      "loss": 0.0592,
      "num_input_tokens_seen": 18334552,
      "step": 28100
    },
    {
      "epoch": 14.730083857442349,
      "grad_norm": 1.6718217134475708,
      "learning_rate": 9.85181811071193e-06,
      "loss": 0.0206,
      "num_input_tokens_seen": 18338168,
      "step": 28105
    },
    {
      "epoch": 14.732704402515724,
      "grad_norm": 1.2035094499588013,
      "learning_rate": 9.842723492616913e-06,
      "loss": 0.014,
      "num_input_tokens_seen": 18341528,
      "step": 28110
    },
    {
      "epoch": 14.735324947589099,
      "grad_norm": 0.7795192003250122,
      "learning_rate": 9.833632045251207e-06,
      "loss": 0.0372,
      "num_input_tokens_seen": 18344920,
      "step": 28115
    },
    {
      "epoch": 14.737945492662474,
      "grad_norm": 0.7961195707321167,
      "learning_rate": 9.824543770516656e-06,
      "loss": 0.0089,
      "num_input_tokens_seen": 18348312,
      "step": 28120
    },
    {
      "epoch": 14.74056603773585,
      "grad_norm": 0.6068192720413208,
      "learning_rate": 9.815458670314412e-06,
      "loss": 0.0105,
      "num_input_tokens_seen": 18351160,
      "step": 28125
    },
    {
      "epoch": 14.743186582809225,
      "grad_norm": 7.311459541320801,
      "learning_rate": 9.80637674654498e-06,
      "loss": 0.0682,
      "num_input_tokens_seen": 18354104,
      "step": 28130
    },
    {
      "epoch": 14.7458071278826,
      "grad_norm": 0.7868527173995972,
      "learning_rate": 9.797298001108193e-06,
      "loss": 0.0306,
      "num_input_tokens_seen": 18357784,
      "step": 28135
    },
    {
      "epoch": 14.748427672955975,
      "grad_norm": 1.7144620418548584,
      "learning_rate": 9.788222435903232e-06,
      "loss": 0.0241,
      "num_input_tokens_seen": 18360984,
      "step": 28140
    },
    {
      "epoch": 14.75104821802935,
      "grad_norm": 1.3880261182785034,
      "learning_rate": 9.779150052828596e-06,
      "loss": 0.0217,
      "num_input_tokens_seen": 18365080,
      "step": 28145
    },
    {
      "epoch": 14.753668763102725,
      "grad_norm": 0.3821888566017151,
      "learning_rate": 9.770080853782108e-06,
      "loss": 0.028,
      "num_input_tokens_seen": 18367800,
      "step": 28150
    },
    {
      "epoch": 14.7562893081761,
      "grad_norm": 0.7214897871017456,
      "learning_rate": 9.76101484066095e-06,
      "loss": 0.0517,
      "num_input_tokens_seen": 18370392,
      "step": 28155
    },
    {
      "epoch": 14.758909853249476,
      "grad_norm": 0.5659173727035522,
      "learning_rate": 9.751952015361637e-06,
      "loss": 0.0215,
      "num_input_tokens_seen": 18373560,
      "step": 28160
    },
    {
      "epoch": 14.76153039832285,
      "grad_norm": 1.014453649520874,
      "learning_rate": 9.742892379780003e-06,
      "loss": 0.0169,
      "num_input_tokens_seen": 18378296,
      "step": 28165
    },
    {
      "epoch": 14.764150943396226,
      "grad_norm": 4.535385608673096,
      "learning_rate": 9.733835935811237e-06,
      "loss": 0.0155,
      "num_input_tokens_seen": 18381016,
      "step": 28170
    },
    {
      "epoch": 14.766771488469601,
      "grad_norm": 1.2587834596633911,
      "learning_rate": 9.724782685349809e-06,
      "loss": 0.0242,
      "num_input_tokens_seen": 18384248,
      "step": 28175
    },
    {
      "epoch": 14.769392033542976,
      "grad_norm": 4.010356426239014,
      "learning_rate": 9.715732630289576e-06,
      "loss": 0.0285,
      "num_input_tokens_seen": 18386648,
      "step": 28180
    },
    {
      "epoch": 14.772012578616351,
      "grad_norm": 1.0697240829467773,
      "learning_rate": 9.706685772523701e-06,
      "loss": 0.0211,
      "num_input_tokens_seen": 18390232,
      "step": 28185
    },
    {
      "epoch": 14.774633123689728,
      "grad_norm": 1.4800918102264404,
      "learning_rate": 9.697642113944692e-06,
      "loss": 0.0402,
      "num_input_tokens_seen": 18393624,
      "step": 28190
    },
    {
      "epoch": 14.777253668763104,
      "grad_norm": 1.400994896888733,
      "learning_rate": 9.688601656444381e-06,
      "loss": 0.0148,
      "num_input_tokens_seen": 18397176,
      "step": 28195
    },
    {
      "epoch": 14.779874213836479,
      "grad_norm": 0.28658464550971985,
      "learning_rate": 9.679564401913921e-06,
      "loss": 0.0289,
      "num_input_tokens_seen": 18400472,
      "step": 28200
    },
    {
      "epoch": 14.782494758909854,
      "grad_norm": 1.0839579105377197,
      "learning_rate": 9.670530352243798e-06,
      "loss": 0.0133,
      "num_input_tokens_seen": 18403224,
      "step": 28205
    },
    {
      "epoch": 14.785115303983229,
      "grad_norm": 1.1618297100067139,
      "learning_rate": 9.661499509323835e-06,
      "loss": 0.0133,
      "num_input_tokens_seen": 18406200,
      "step": 28210
    },
    {
      "epoch": 14.787735849056604,
      "grad_norm": 0.6445226669311523,
      "learning_rate": 9.652471875043182e-06,
      "loss": 0.0189,
      "num_input_tokens_seen": 18409432,
      "step": 28215
    },
    {
      "epoch": 14.79035639412998,
      "grad_norm": 0.30029964447021484,
      "learning_rate": 9.64344745129033e-06,
      "loss": 0.0171,
      "num_input_tokens_seen": 18411736,
      "step": 28220
    },
    {
      "epoch": 14.792976939203355,
      "grad_norm": 1.5935585498809814,
      "learning_rate": 9.634426239953073e-06,
      "loss": 0.06,
      "num_input_tokens_seen": 18415448,
      "step": 28225
    },
    {
      "epoch": 14.79559748427673,
      "grad_norm": 1.7062060832977295,
      "learning_rate": 9.62540824291854e-06,
      "loss": 0.0365,
      "num_input_tokens_seen": 18419032,
      "step": 28230
    },
    {
      "epoch": 14.798218029350105,
      "grad_norm": 2.584503412246704,
      "learning_rate": 9.616393462073197e-06,
      "loss": 0.0218,
      "num_input_tokens_seen": 18422136,
      "step": 28235
    },
    {
      "epoch": 14.80083857442348,
      "grad_norm": 0.9077717065811157,
      "learning_rate": 9.607381899302839e-06,
      "loss": 0.0842,
      "num_input_tokens_seen": 18424696,
      "step": 28240
    },
    {
      "epoch": 14.803459119496855,
      "grad_norm": 0.6788820624351501,
      "learning_rate": 9.59837355649258e-06,
      "loss": 0.0332,
      "num_input_tokens_seen": 18427704,
      "step": 28245
    },
    {
      "epoch": 14.80607966457023,
      "grad_norm": 1.5029211044311523,
      "learning_rate": 9.589368435526868e-06,
      "loss": 0.0128,
      "num_input_tokens_seen": 18430328,
      "step": 28250
    },
    {
      "epoch": 14.808700209643606,
      "grad_norm": 1.6385908126831055,
      "learning_rate": 9.580366538289468e-06,
      "loss": 0.0261,
      "num_input_tokens_seen": 18432920,
      "step": 28255
    },
    {
      "epoch": 14.81132075471698,
      "grad_norm": 0.5199474096298218,
      "learning_rate": 9.57136786666346e-06,
      "loss": 0.0703,
      "num_input_tokens_seen": 18435896,
      "step": 28260
    },
    {
      "epoch": 14.813941299790356,
      "grad_norm": 0.2178129255771637,
      "learning_rate": 9.562372422531276e-06,
      "loss": 0.0646,
      "num_input_tokens_seen": 18438712,
      "step": 28265
    },
    {
      "epoch": 14.816561844863731,
      "grad_norm": 1.9086846113204956,
      "learning_rate": 9.553380207774662e-06,
      "loss": 0.0189,
      "num_input_tokens_seen": 18441592,
      "step": 28270
    },
    {
      "epoch": 14.819182389937106,
      "grad_norm": 8.220511436462402,
      "learning_rate": 9.54439122427469e-06,
      "loss": 0.0094,
      "num_input_tokens_seen": 18444120,
      "step": 28275
    },
    {
      "epoch": 14.821802935010481,
      "grad_norm": 0.539993166923523,
      "learning_rate": 9.535405473911748e-06,
      "loss": 0.0167,
      "num_input_tokens_seen": 18447640,
      "step": 28280
    },
    {
      "epoch": 14.824423480083858,
      "grad_norm": 0.8184908628463745,
      "learning_rate": 9.526422958565546e-06,
      "loss": 0.0357,
      "num_input_tokens_seen": 18452312,
      "step": 28285
    },
    {
      "epoch": 14.827044025157234,
      "grad_norm": 1.3695992231369019,
      "learning_rate": 9.517443680115124e-06,
      "loss": 0.0298,
      "num_input_tokens_seen": 18455704,
      "step": 28290
    },
    {
      "epoch": 14.829664570230609,
      "grad_norm": 3.01346755027771,
      "learning_rate": 9.50846764043885e-06,
      "loss": 0.0204,
      "num_input_tokens_seen": 18459384,
      "step": 28295
    },
    {
      "epoch": 14.832285115303984,
      "grad_norm": 0.8550519347190857,
      "learning_rate": 9.499494841414422e-06,
      "loss": 0.0863,
      "num_input_tokens_seen": 18463512,
      "step": 28300
    },
    {
      "epoch": 14.834905660377359,
      "grad_norm": 1.6545058488845825,
      "learning_rate": 9.490525284918822e-06,
      "loss": 0.0127,
      "num_input_tokens_seen": 18466520,
      "step": 28305
    },
    {
      "epoch": 14.837526205450734,
      "grad_norm": 1.1254527568817139,
      "learning_rate": 9.4815589728284e-06,
      "loss": 0.0259,
      "num_input_tokens_seen": 18469240,
      "step": 28310
    },
    {
      "epoch": 14.84014675052411,
      "grad_norm": 0.7224273085594177,
      "learning_rate": 9.472595907018788e-06,
      "loss": 0.0404,
      "num_input_tokens_seen": 18473272,
      "step": 28315
    },
    {
      "epoch": 14.842767295597485,
      "grad_norm": 0.9648395776748657,
      "learning_rate": 9.463636089364968e-06,
      "loss": 0.0405,
      "num_input_tokens_seen": 18478392,
      "step": 28320
    },
    {
      "epoch": 14.84538784067086,
      "grad_norm": 3.6033225059509277,
      "learning_rate": 9.454679521741239e-06,
      "loss": 0.0269,
      "num_input_tokens_seen": 18481048,
      "step": 28325
    },
    {
      "epoch": 14.848008385744235,
      "grad_norm": 0.9709126949310303,
      "learning_rate": 9.445726206021194e-06,
      "loss": 0.0193,
      "num_input_tokens_seen": 18484376,
      "step": 28330
    },
    {
      "epoch": 14.85062893081761,
      "grad_norm": 1.1313740015029907,
      "learning_rate": 9.436776144077772e-06,
      "loss": 0.0115,
      "num_input_tokens_seen": 18487928,
      "step": 28335
    },
    {
      "epoch": 14.853249475890985,
      "grad_norm": 1.768547773361206,
      "learning_rate": 9.427829337783237e-06,
      "loss": 0.0525,
      "num_input_tokens_seen": 18491192,
      "step": 28340
    },
    {
      "epoch": 14.85587002096436,
      "grad_norm": 0.30685973167419434,
      "learning_rate": 9.418885789009138e-06,
      "loss": 0.0152,
      "num_input_tokens_seen": 18494744,
      "step": 28345
    },
    {
      "epoch": 14.858490566037736,
      "grad_norm": 0.8720101714134216,
      "learning_rate": 9.409945499626374e-06,
      "loss": 0.0272,
      "num_input_tokens_seen": 18498584,
      "step": 28350
    },
    {
      "epoch": 14.86111111111111,
      "grad_norm": 2.0115199089050293,
      "learning_rate": 9.40100847150516e-06,
      "loss": 0.0402,
      "num_input_tokens_seen": 18501400,
      "step": 28355
    },
    {
      "epoch": 14.863731656184486,
      "grad_norm": 1.8285832405090332,
      "learning_rate": 9.392074706515003e-06,
      "loss": 0.0339,
      "num_input_tokens_seen": 18503864,
      "step": 28360
    },
    {
      "epoch": 14.866352201257861,
      "grad_norm": 0.757003903388977,
      "learning_rate": 9.383144206524753e-06,
      "loss": 0.0192,
      "num_input_tokens_seen": 18506712,
      "step": 28365
    },
    {
      "epoch": 14.868972746331236,
      "grad_norm": 2.2024667263031006,
      "learning_rate": 9.374216973402575e-06,
      "loss": 0.0148,
      "num_input_tokens_seen": 18510360,
      "step": 28370
    },
    {
      "epoch": 14.871593291404611,
      "grad_norm": 3.1046392917633057,
      "learning_rate": 9.365293009015926e-06,
      "loss": 0.0342,
      "num_input_tokens_seen": 18513432,
      "step": 28375
    },
    {
      "epoch": 14.874213836477988,
      "grad_norm": 2.240967035293579,
      "learning_rate": 9.356372315231623e-06,
      "loss": 0.015,
      "num_input_tokens_seen": 18516632,
      "step": 28380
    },
    {
      "epoch": 14.876834381551364,
      "grad_norm": 1.903018832206726,
      "learning_rate": 9.347454893915749e-06,
      "loss": 0.0383,
      "num_input_tokens_seen": 18521624,
      "step": 28385
    },
    {
      "epoch": 14.879454926624739,
      "grad_norm": 2.565521240234375,
      "learning_rate": 9.338540746933741e-06,
      "loss": 0.0379,
      "num_input_tokens_seen": 18524760,
      "step": 28390
    },
    {
      "epoch": 14.882075471698114,
      "grad_norm": 1.5653598308563232,
      "learning_rate": 9.329629876150339e-06,
      "loss": 0.0169,
      "num_input_tokens_seen": 18527416,
      "step": 28395
    },
    {
      "epoch": 14.884696016771489,
      "grad_norm": 2.1940085887908936,
      "learning_rate": 9.320722283429581e-06,
      "loss": 0.0383,
      "num_input_tokens_seen": 18531032,
      "step": 28400
    },
    {
      "epoch": 14.887316561844864,
      "grad_norm": 1.0786057710647583,
      "learning_rate": 9.311817970634854e-06,
      "loss": 0.0174,
      "num_input_tokens_seen": 18534744,
      "step": 28405
    },
    {
      "epoch": 14.88993710691824,
      "grad_norm": 0.81184321641922,
      "learning_rate": 9.302916939628814e-06,
      "loss": 0.0277,
      "num_input_tokens_seen": 18538392,
      "step": 28410
    },
    {
      "epoch": 14.892557651991615,
      "grad_norm": 1.2120332717895508,
      "learning_rate": 9.29401919227347e-06,
      "loss": 0.0138,
      "num_input_tokens_seen": 18541400,
      "step": 28415
    },
    {
      "epoch": 14.89517819706499,
      "grad_norm": 0.13479354977607727,
      "learning_rate": 9.28512473043013e-06,
      "loss": 0.0343,
      "num_input_tokens_seen": 18544696,
      "step": 28420
    },
    {
      "epoch": 14.897798742138365,
      "grad_norm": 1.1987253427505493,
      "learning_rate": 9.276233555959419e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 18547864,
      "step": 28425
    },
    {
      "epoch": 14.90041928721174,
      "grad_norm": 1.4099390506744385,
      "learning_rate": 9.267345670721254e-06,
      "loss": 0.0262,
      "num_input_tokens_seen": 18550584,
      "step": 28430
    },
    {
      "epoch": 14.903039832285115,
      "grad_norm": 1.1763958930969238,
      "learning_rate": 9.258461076574895e-06,
      "loss": 0.0287,
      "num_input_tokens_seen": 18553624,
      "step": 28435
    },
    {
      "epoch": 14.90566037735849,
      "grad_norm": 1.1052600145339966,
      "learning_rate": 9.24957977537888e-06,
      "loss": 0.038,
      "num_input_tokens_seen": 18556472,
      "step": 28440
    },
    {
      "epoch": 14.908280922431866,
      "grad_norm": 1.056907296180725,
      "learning_rate": 9.240701768991086e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 18558840,
      "step": 28445
    },
    {
      "epoch": 14.91090146750524,
      "grad_norm": 0.5595157742500305,
      "learning_rate": 9.231827059268692e-06,
      "loss": 0.0271,
      "num_input_tokens_seen": 18562008,
      "step": 28450
    },
    {
      "epoch": 14.913522012578616,
      "grad_norm": 1.7096686363220215,
      "learning_rate": 9.222955648068193e-06,
      "loss": 0.0354,
      "num_input_tokens_seen": 18564952,
      "step": 28455
    },
    {
      "epoch": 14.916142557651991,
      "grad_norm": 2.0417869091033936,
      "learning_rate": 9.214087537245378e-06,
      "loss": 0.0288,
      "num_input_tokens_seen": 18567800,
      "step": 28460
    },
    {
      "epoch": 14.918763102725366,
      "grad_norm": 3.5166056156158447,
      "learning_rate": 9.205222728655344e-06,
      "loss": 0.0221,
      "num_input_tokens_seen": 18571096,
      "step": 28465
    },
    {
      "epoch": 14.921383647798741,
      "grad_norm": 0.2778494954109192,
      "learning_rate": 9.196361224152522e-06,
      "loss": 0.0153,
      "num_input_tokens_seen": 18574136,
      "step": 28470
    },
    {
      "epoch": 14.924004192872118,
      "grad_norm": 2.172600030899048,
      "learning_rate": 9.187503025590632e-06,
      "loss": 0.019,
      "num_input_tokens_seen": 18576664,
      "step": 28475
    },
    {
      "epoch": 14.926624737945493,
      "grad_norm": 5.997636795043945,
      "learning_rate": 9.17864813482272e-06,
      "loss": 0.0165,
      "num_input_tokens_seen": 18580824,
      "step": 28480
    },
    {
      "epoch": 14.929245283018869,
      "grad_norm": 3.022094488143921,
      "learning_rate": 9.169796553701122e-06,
      "loss": 0.027,
      "num_input_tokens_seen": 18583864,
      "step": 28485
    },
    {
      "epoch": 14.931865828092244,
      "grad_norm": 0.9176557660102844,
      "learning_rate": 9.160948284077473e-06,
      "loss": 0.0111,
      "num_input_tokens_seen": 18586680,
      "step": 28490
    },
    {
      "epoch": 14.934486373165619,
      "grad_norm": 1.090126872062683,
      "learning_rate": 9.152103327802738e-06,
      "loss": 0.0166,
      "num_input_tokens_seen": 18590776,
      "step": 28495
    },
    {
      "epoch": 14.937106918238994,
      "grad_norm": 1.5598143339157104,
      "learning_rate": 9.143261686727189e-06,
      "loss": 0.0129,
      "num_input_tokens_seen": 18594616,
      "step": 28500
    },
    {
      "epoch": 14.93972746331237,
      "grad_norm": 1.7252957820892334,
      "learning_rate": 9.134423362700392e-06,
      "loss": 0.036,
      "num_input_tokens_seen": 18597528,
      "step": 28505
    },
    {
      "epoch": 14.942348008385745,
      "grad_norm": 0.09738565236330032,
      "learning_rate": 9.12558835757123e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 18599992,
      "step": 28510
    },
    {
      "epoch": 14.94496855345912,
      "grad_norm": 2.608419179916382,
      "learning_rate": 9.116756673187878e-06,
      "loss": 0.0147,
      "num_input_tokens_seen": 18602872,
      "step": 28515
    },
    {
      "epoch": 14.947589098532495,
      "grad_norm": 0.5227094292640686,
      "learning_rate": 9.107928311397817e-06,
      "loss": 0.0144,
      "num_input_tokens_seen": 18606360,
      "step": 28520
    },
    {
      "epoch": 14.95020964360587,
      "grad_norm": 1.1929004192352295,
      "learning_rate": 9.09910327404785e-06,
      "loss": 0.0179,
      "num_input_tokens_seen": 18608824,
      "step": 28525
    },
    {
      "epoch": 14.952830188679245,
      "grad_norm": 1.776100754737854,
      "learning_rate": 9.090281562984068e-06,
      "loss": 0.0164,
      "num_input_tokens_seen": 18611864,
      "step": 28530
    },
    {
      "epoch": 14.95545073375262,
      "grad_norm": 0.9739536046981812,
      "learning_rate": 9.081463180051883e-06,
      "loss": 0.0314,
      "num_input_tokens_seen": 18615480,
      "step": 28535
    },
    {
      "epoch": 14.958071278825996,
      "grad_norm": 0.6449061036109924,
      "learning_rate": 9.072648127095996e-06,
      "loss": 0.0432,
      "num_input_tokens_seen": 18618392,
      "step": 28540
    },
    {
      "epoch": 14.96069182389937,
      "grad_norm": 5.675971984863281,
      "learning_rate": 9.0638364059604e-06,
      "loss": 0.0685,
      "num_input_tokens_seen": 18622200,
      "step": 28545
    },
    {
      "epoch": 14.963312368972746,
      "grad_norm": 1.8347281217575073,
      "learning_rate": 9.055028018488417e-06,
      "loss": 0.0293,
      "num_input_tokens_seen": 18625080,
      "step": 28550
    },
    {
      "epoch": 14.965932914046121,
      "grad_norm": 2.744197368621826,
      "learning_rate": 9.046222966522665e-06,
      "loss": 0.0198,
      "num_input_tokens_seen": 18627992,
      "step": 28555
    },
    {
      "epoch": 14.968553459119496,
      "grad_norm": 2.8561761379241943,
      "learning_rate": 9.037421251905059e-06,
      "loss": 0.0256,
      "num_input_tokens_seen": 18631192,
      "step": 28560
    },
    {
      "epoch": 14.971174004192871,
      "grad_norm": 1.471012830734253,
      "learning_rate": 9.028622876476833e-06,
      "loss": 0.0131,
      "num_input_tokens_seen": 18634904,
      "step": 28565
    },
    {
      "epoch": 14.973794549266248,
      "grad_norm": 0.3495420217514038,
      "learning_rate": 9.019827842078474e-06,
      "loss": 0.0276,
      "num_input_tokens_seen": 18637784,
      "step": 28570
    },
    {
      "epoch": 14.976415094339622,
      "grad_norm": 1.9663704633712769,
      "learning_rate": 9.011036150549817e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 18640120,
      "step": 28575
    },
    {
      "epoch": 14.979035639412999,
      "grad_norm": 1.1609609127044678,
      "learning_rate": 9.002247803729987e-06,
      "loss": 0.0553,
      "num_input_tokens_seen": 18643672,
      "step": 28580
    },
    {
      "epoch": 14.981656184486374,
      "grad_norm": 1.3317408561706543,
      "learning_rate": 8.993462803457404e-06,
      "loss": 0.0261,
      "num_input_tokens_seen": 18646872,
      "step": 28585
    },
    {
      "epoch": 14.984276729559749,
      "grad_norm": 8.171697616577148,
      "learning_rate": 8.984681151569799e-06,
      "loss": 0.0127,
      "num_input_tokens_seen": 18650744,
      "step": 28590
    },
    {
      "epoch": 14.986897274633124,
      "grad_norm": 5.849010467529297,
      "learning_rate": 8.975902849904183e-06,
      "loss": 0.023,
      "num_input_tokens_seen": 18654296,
      "step": 28595
    },
    {
      "epoch": 14.9895178197065,
      "grad_norm": 1.8317625522613525,
      "learning_rate": 8.967127900296874e-06,
      "loss": 0.0201,
      "num_input_tokens_seen": 18657560,
      "step": 28600
    },
    {
      "epoch": 14.992138364779874,
      "grad_norm": 1.600345253944397,
      "learning_rate": 8.95835630458349e-06,
      "loss": 0.0159,
      "num_input_tokens_seen": 18660728,
      "step": 28605
    },
    {
      "epoch": 14.99475890985325,
      "grad_norm": 0.45539623498916626,
      "learning_rate": 8.949588064598958e-06,
      "loss": 0.0096,
      "num_input_tokens_seen": 18663352,
      "step": 28610
    },
    {
      "epoch": 14.997379454926625,
      "grad_norm": 0.5915318131446838,
      "learning_rate": 8.940823182177499e-06,
      "loss": 0.0261,
      "num_input_tokens_seen": 18665816,
      "step": 28615
    },
    {
      "epoch": 15.0,
      "grad_norm": 1.1945768594741821,
      "learning_rate": 8.932061659152607e-06,
      "loss": 0.0106,
      "num_input_tokens_seen": 18668536,
      "step": 28620
    },
    {
      "epoch": 15.0,
      "eval_loss": 1.3183939456939697,
      "eval_runtime": 11.8432,
      "eval_samples_per_second": 71.602,
      "eval_steps_per_second": 17.9,
      "num_input_tokens_seen": 18668536,
      "step": 28620
    },
    {
      "epoch": 15.002620545073375,
      "grad_norm": 2.903273105621338,
      "learning_rate": 8.92330349735711e-06,
      "loss": 0.0581,
      "num_input_tokens_seen": 18672024,
      "step": 28625
    },
    {
      "epoch": 15.00524109014675,
      "grad_norm": 1.0009626150131226,
      "learning_rate": 8.914548698623104e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 18674712,
      "step": 28630
    },
    {
      "epoch": 15.007861635220126,
      "grad_norm": 0.25319698452949524,
      "learning_rate": 8.905797264781995e-06,
      "loss": 0.0456,
      "num_input_tokens_seen": 18682488,
      "step": 28635
    },
    {
      "epoch": 15.0104821802935,
      "grad_norm": 1.740484356880188,
      "learning_rate": 8.89704919766449e-06,
      "loss": 0.0157,
      "num_input_tokens_seen": 18685816,
      "step": 28640
    },
    {
      "epoch": 15.013102725366876,
      "grad_norm": 2.1003482341766357,
      "learning_rate": 8.888304499100586e-06,
      "loss": 0.0223,
      "num_input_tokens_seen": 18688760,
      "step": 28645
    },
    {
      "epoch": 15.015723270440251,
      "grad_norm": 1.2842174768447876,
      "learning_rate": 8.879563170919566e-06,
      "loss": 0.0079,
      "num_input_tokens_seen": 18691160,
      "step": 28650
    },
    {
      "epoch": 15.018343815513626,
      "grad_norm": 0.6592991948127747,
      "learning_rate": 8.870825214950016e-06,
      "loss": 0.0158,
      "num_input_tokens_seen": 18694616,
      "step": 28655
    },
    {
      "epoch": 15.020964360587001,
      "grad_norm": 0.5806623697280884,
      "learning_rate": 8.862090633019818e-06,
      "loss": 0.0134,
      "num_input_tokens_seen": 18697656,
      "step": 28660
    },
    {
      "epoch": 15.023584905660377,
      "grad_norm": 0.6077752709388733,
      "learning_rate": 8.853359426956146e-06,
      "loss": 0.0138,
      "num_input_tokens_seen": 18700536,
      "step": 28665
    },
    {
      "epoch": 15.026205450733753,
      "grad_norm": 0.46704286336898804,
      "learning_rate": 8.844631598585482e-06,
      "loss": 0.0132,
      "num_input_tokens_seen": 18703960,
      "step": 28670
    },
    {
      "epoch": 15.028825995807129,
      "grad_norm": 1.5051900148391724,
      "learning_rate": 8.835907149733569e-06,
      "loss": 0.0118,
      "num_input_tokens_seen": 18706744,
      "step": 28675
    },
    {
      "epoch": 15.031446540880504,
      "grad_norm": 0.7483696937561035,
      "learning_rate": 8.827186082225477e-06,
      "loss": 0.0208,
      "num_input_tokens_seen": 18710424,
      "step": 28680
    },
    {
      "epoch": 15.034067085953879,
      "grad_norm": 1.951737403869629,
      "learning_rate": 8.81846839788554e-06,
      "loss": 0.0359,
      "num_input_tokens_seen": 18715640,
      "step": 28685
    },
    {
      "epoch": 15.036687631027254,
      "grad_norm": 0.9412052631378174,
      "learning_rate": 8.809754098537407e-06,
      "loss": 0.0231,
      "num_input_tokens_seen": 18718680,
      "step": 28690
    },
    {
      "epoch": 15.03930817610063,
      "grad_norm": 0.5555939078330994,
      "learning_rate": 8.801043186004016e-06,
      "loss": 0.0158,
      "num_input_tokens_seen": 18721688,
      "step": 28695
    },
    {
      "epoch": 15.041928721174004,
      "grad_norm": 0.80327969789505,
      "learning_rate": 8.792335662107575e-06,
      "loss": 0.0097,
      "num_input_tokens_seen": 18724920,
      "step": 28700
    },
    {
      "epoch": 15.04454926624738,
      "grad_norm": 1.0941545963287354,
      "learning_rate": 8.783631528669605e-06,
      "loss": 0.0676,
      "num_input_tokens_seen": 18728376,
      "step": 28705
    },
    {
      "epoch": 15.047169811320755,
      "grad_norm": 0.8405272960662842,
      "learning_rate": 8.774930787510924e-06,
      "loss": 0.0091,
      "num_input_tokens_seen": 18731704,
      "step": 28710
    },
    {
      "epoch": 15.04979035639413,
      "grad_norm": 0.43727776408195496,
      "learning_rate": 8.766233440451608e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 18734968,
      "step": 28715
    },
    {
      "epoch": 15.052410901467505,
      "grad_norm": 1.9271520376205444,
      "learning_rate": 8.75753948931105e-06,
      "loss": 0.0153,
      "num_input_tokens_seen": 18738680,
      "step": 28720
    },
    {
      "epoch": 15.05503144654088,
      "grad_norm": 1.1831696033477783,
      "learning_rate": 8.748848935907936e-06,
      "loss": 0.0122,
      "num_input_tokens_seen": 18742136,
      "step": 28725
    },
    {
      "epoch": 15.057651991614255,
      "grad_norm": 1.4025064706802368,
      "learning_rate": 8.740161782060214e-06,
      "loss": 0.0215,
      "num_input_tokens_seen": 18745016,
      "step": 28730
    },
    {
      "epoch": 15.06027253668763,
      "grad_norm": 1.9359933137893677,
      "learning_rate": 8.731478029585144e-06,
      "loss": 0.0113,
      "num_input_tokens_seen": 18748920,
      "step": 28735
    },
    {
      "epoch": 15.062893081761006,
      "grad_norm": 1.0782978534698486,
      "learning_rate": 8.722797680299278e-06,
      "loss": 0.0209,
      "num_input_tokens_seen": 18753560,
      "step": 28740
    },
    {
      "epoch": 15.065513626834381,
      "grad_norm": 0.5312563180923462,
      "learning_rate": 8.71412073601843e-06,
      "loss": 0.0107,
      "num_input_tokens_seen": 18756952,
      "step": 28745
    },
    {
      "epoch": 15.068134171907756,
      "grad_norm": 6.403900623321533,
      "learning_rate": 8.705447198557731e-06,
      "loss": 0.0325,
      "num_input_tokens_seen": 18760472,
      "step": 28750
    },
    {
      "epoch": 15.070754716981131,
      "grad_norm": 0.520379900932312,
      "learning_rate": 8.696777069731574e-06,
      "loss": 0.0155,
      "num_input_tokens_seen": 18763928,
      "step": 28755
    },
    {
      "epoch": 15.073375262054507,
      "grad_norm": 0.696018397808075,
      "learning_rate": 8.688110351353654e-06,
      "loss": 0.0116,
      "num_input_tokens_seen": 18767512,
      "step": 28760
    },
    {
      "epoch": 15.075995807127883,
      "grad_norm": 1.7257916927337646,
      "learning_rate": 8.679447045236962e-06,
      "loss": 0.0327,
      "num_input_tokens_seen": 18774040,
      "step": 28765
    },
    {
      "epoch": 15.078616352201259,
      "grad_norm": 0.47730931639671326,
      "learning_rate": 8.670787153193746e-06,
      "loss": 0.0128,
      "num_input_tokens_seen": 18777336,
      "step": 28770
    },
    {
      "epoch": 15.081236897274634,
      "grad_norm": 0.9520760774612427,
      "learning_rate": 8.662130677035574e-06,
      "loss": 0.0123,
      "num_input_tokens_seen": 18780504,
      "step": 28775
    },
    {
      "epoch": 15.083857442348009,
      "grad_norm": 0.8928792476654053,
      "learning_rate": 8.65347761857326e-06,
      "loss": 0.0159,
      "num_input_tokens_seen": 18783800,
      "step": 28780
    },
    {
      "epoch": 15.086477987421384,
      "grad_norm": 0.11111858487129211,
      "learning_rate": 8.64482797961694e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 18786968,
      "step": 28785
    },
    {
      "epoch": 15.08909853249476,
      "grad_norm": 0.4339675009250641,
      "learning_rate": 8.636181761976016e-06,
      "loss": 0.0128,
      "num_input_tokens_seen": 18789720,
      "step": 28790
    },
    {
      "epoch": 15.091719077568134,
      "grad_norm": 0.8623266816139221,
      "learning_rate": 8.62753896745919e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 18792952,
      "step": 28795
    },
    {
      "epoch": 15.09433962264151,
      "grad_norm": 2.6346967220306396,
      "learning_rate": 8.618899597874413e-06,
      "loss": 0.0156,
      "num_input_tokens_seen": 18795480,
      "step": 28800
    },
    {
      "epoch": 15.096960167714885,
      "grad_norm": 12.582884788513184,
      "learning_rate": 8.610263655028964e-06,
      "loss": 0.0166,
      "num_input_tokens_seen": 18798648,
      "step": 28805
    },
    {
      "epoch": 15.09958071278826,
      "grad_norm": 2.6681299209594727,
      "learning_rate": 8.601631140729366e-06,
      "loss": 0.0097,
      "num_input_tokens_seen": 18802680,
      "step": 28810
    },
    {
      "epoch": 15.102201257861635,
      "grad_norm": 0.5695400238037109,
      "learning_rate": 8.593002056781451e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 18805112,
      "step": 28815
    },
    {
      "epoch": 15.10482180293501,
      "grad_norm": 3.0569252967834473,
      "learning_rate": 8.584376404990326e-06,
      "loss": 0.0129,
      "num_input_tokens_seen": 18809272,
      "step": 28820
    },
    {
      "epoch": 15.107442348008385,
      "grad_norm": 1.149688482284546,
      "learning_rate": 8.575754187160384e-06,
      "loss": 0.0542,
      "num_input_tokens_seen": 18812440,
      "step": 28825
    },
    {
      "epoch": 15.11006289308176,
      "grad_norm": 1.4211583137512207,
      "learning_rate": 8.567135405095291e-06,
      "loss": 0.0198,
      "num_input_tokens_seen": 18815288,
      "step": 28830
    },
    {
      "epoch": 15.112683438155136,
      "grad_norm": 4.930824279785156,
      "learning_rate": 8.558520060597985e-06,
      "loss": 0.0447,
      "num_input_tokens_seen": 18818424,
      "step": 28835
    },
    {
      "epoch": 15.115303983228511,
      "grad_norm": 3.892298698425293,
      "learning_rate": 8.54990815547071e-06,
      "loss": 0.0241,
      "num_input_tokens_seen": 18821144,
      "step": 28840
    },
    {
      "epoch": 15.117924528301886,
      "grad_norm": 2.9270451068878174,
      "learning_rate": 8.541299691514974e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 18824632,
      "step": 28845
    },
    {
      "epoch": 15.120545073375261,
      "grad_norm": 0.6566236615180969,
      "learning_rate": 8.53269467053158e-06,
      "loss": 0.0121,
      "num_input_tokens_seen": 18827224,
      "step": 28850
    },
    {
      "epoch": 15.123165618448636,
      "grad_norm": 0.9563750624656677,
      "learning_rate": 8.524093094320593e-06,
      "loss": 0.0164,
      "num_input_tokens_seen": 18830072,
      "step": 28855
    },
    {
      "epoch": 15.125786163522013,
      "grad_norm": 1.1510494947433472,
      "learning_rate": 8.515494964681357e-06,
      "loss": 0.0112,
      "num_input_tokens_seen": 18832856,
      "step": 28860
    },
    {
      "epoch": 15.128406708595389,
      "grad_norm": 0.4226421117782593,
      "learning_rate": 8.506900283412506e-06,
      "loss": 0.0424,
      "num_input_tokens_seen": 18835544,
      "step": 28865
    },
    {
      "epoch": 15.131027253668764,
      "grad_norm": 1.111057996749878,
      "learning_rate": 8.498309052311953e-06,
      "loss": 0.0164,
      "num_input_tokens_seen": 18838360,
      "step": 28870
    },
    {
      "epoch": 15.133647798742139,
      "grad_norm": 1.5340782403945923,
      "learning_rate": 8.489721273176887e-06,
      "loss": 0.0382,
      "num_input_tokens_seen": 18842168,
      "step": 28875
    },
    {
      "epoch": 15.136268343815514,
      "grad_norm": 0.8376998901367188,
      "learning_rate": 8.481136947803777e-06,
      "loss": 0.0223,
      "num_input_tokens_seen": 18845496,
      "step": 28880
    },
    {
      "epoch": 15.13888888888889,
      "grad_norm": 0.7799738645553589,
      "learning_rate": 8.472556077988362e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 18850232,
      "step": 28885
    },
    {
      "epoch": 15.141509433962264,
      "grad_norm": 6.583114147186279,
      "learning_rate": 8.46397866552565e-06,
      "loss": 0.0104,
      "num_input_tokens_seen": 18852536,
      "step": 28890
    },
    {
      "epoch": 15.14412997903564,
      "grad_norm": 0.3495711386203766,
      "learning_rate": 8.455404712209952e-06,
      "loss": 0.014,
      "num_input_tokens_seen": 18855544,
      "step": 28895
    },
    {
      "epoch": 15.146750524109015,
      "grad_norm": 1.950939655303955,
      "learning_rate": 8.446834219834836e-06,
      "loss": 0.0335,
      "num_input_tokens_seen": 18858584,
      "step": 28900
    },
    {
      "epoch": 15.14937106918239,
      "grad_norm": 5.687892436981201,
      "learning_rate": 8.43826719019316e-06,
      "loss": 0.0158,
      "num_input_tokens_seen": 18861688,
      "step": 28905
    },
    {
      "epoch": 15.151991614255765,
      "grad_norm": 3.4362008571624756,
      "learning_rate": 8.429703625077043e-06,
      "loss": 0.0241,
      "num_input_tokens_seen": 18864056,
      "step": 28910
    },
    {
      "epoch": 15.15461215932914,
      "grad_norm": 0.8720967173576355,
      "learning_rate": 8.421143526277875e-06,
      "loss": 0.0136,
      "num_input_tokens_seen": 18867960,
      "step": 28915
    },
    {
      "epoch": 15.157232704402515,
      "grad_norm": 0.5040133595466614,
      "learning_rate": 8.41258689558634e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 18870328,
      "step": 28920
    },
    {
      "epoch": 15.15985324947589,
      "grad_norm": 0.7953920364379883,
      "learning_rate": 8.404033734792386e-06,
      "loss": 0.0087,
      "num_input_tokens_seen": 18873976,
      "step": 28925
    },
    {
      "epoch": 15.162473794549266,
      "grad_norm": 1.2674672603607178,
      "learning_rate": 8.395484045685248e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 18877112,
      "step": 28930
    },
    {
      "epoch": 15.165094339622641,
      "grad_norm": 1.9871689081192017,
      "learning_rate": 8.386937830053412e-06,
      "loss": 0.0097,
      "num_input_tokens_seen": 18879480,
      "step": 28935
    },
    {
      "epoch": 15.167714884696016,
      "grad_norm": 0.290153443813324,
      "learning_rate": 8.37839508968464e-06,
      "loss": 0.0091,
      "num_input_tokens_seen": 18882072,
      "step": 28940
    },
    {
      "epoch": 15.170335429769391,
      "grad_norm": 3.0847251415252686,
      "learning_rate": 8.369855826365988e-06,
      "loss": 0.0732,
      "num_input_tokens_seen": 18884760,
      "step": 28945
    },
    {
      "epoch": 15.172955974842766,
      "grad_norm": 0.7473499774932861,
      "learning_rate": 8.361320041883772e-06,
      "loss": 0.0158,
      "num_input_tokens_seen": 18887736,
      "step": 28950
    },
    {
      "epoch": 15.175576519916143,
      "grad_norm": 1.3865774869918823,
      "learning_rate": 8.352787738023576e-06,
      "loss": 0.0147,
      "num_input_tokens_seen": 18890584,
      "step": 28955
    },
    {
      "epoch": 15.178197064989519,
      "grad_norm": 1.4132678508758545,
      "learning_rate": 8.344258916570274e-06,
      "loss": 0.0165,
      "num_input_tokens_seen": 18893208,
      "step": 28960
    },
    {
      "epoch": 15.180817610062894,
      "grad_norm": 1.5668050050735474,
      "learning_rate": 8.335733579307988e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 18895864,
      "step": 28965
    },
    {
      "epoch": 15.183438155136269,
      "grad_norm": 1.3085530996322632,
      "learning_rate": 8.327211728020113e-06,
      "loss": 0.05,
      "num_input_tokens_seen": 18898712,
      "step": 28970
    },
    {
      "epoch": 15.186058700209644,
      "grad_norm": 1.6474080085754395,
      "learning_rate": 8.31869336448933e-06,
      "loss": 0.0181,
      "num_input_tokens_seen": 18901624,
      "step": 28975
    },
    {
      "epoch": 15.18867924528302,
      "grad_norm": 0.46331286430358887,
      "learning_rate": 8.310178490497586e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 18905176,
      "step": 28980
    },
    {
      "epoch": 15.191299790356394,
      "grad_norm": 2.288252830505371,
      "learning_rate": 8.301667107826103e-06,
      "loss": 0.0118,
      "num_input_tokens_seen": 18907864,
      "step": 28985
    },
    {
      "epoch": 15.19392033542977,
      "grad_norm": 0.6361531615257263,
      "learning_rate": 8.293159218255345e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 18911064,
      "step": 28990
    },
    {
      "epoch": 15.196540880503145,
      "grad_norm": 0.5988888740539551,
      "learning_rate": 8.284654823565088e-06,
      "loss": 0.0257,
      "num_input_tokens_seen": 18914136,
      "step": 28995
    },
    {
      "epoch": 15.19916142557652,
      "grad_norm": 0.3592267334461212,
      "learning_rate": 8.276153925534333e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 18917048,
      "step": 29000
    },
    {
      "epoch": 15.201781970649895,
      "grad_norm": 1.6872751712799072,
      "learning_rate": 8.267656525941383e-06,
      "loss": 0.0932,
      "num_input_tokens_seen": 18919864,
      "step": 29005
    },
    {
      "epoch": 15.20440251572327,
      "grad_norm": 1.6937816143035889,
      "learning_rate": 8.259162626563801e-06,
      "loss": 0.0145,
      "num_input_tokens_seen": 18922968,
      "step": 29010
    },
    {
      "epoch": 15.207023060796645,
      "grad_norm": 1.1747242212295532,
      "learning_rate": 8.250672229178402e-06,
      "loss": 0.0145,
      "num_input_tokens_seen": 18926328,
      "step": 29015
    },
    {
      "epoch": 15.20964360587002,
      "grad_norm": 2.498093843460083,
      "learning_rate": 8.242185335561295e-06,
      "loss": 0.0106,
      "num_input_tokens_seen": 18929368,
      "step": 29020
    },
    {
      "epoch": 15.212264150943396,
      "grad_norm": 1.3990484476089478,
      "learning_rate": 8.23370194748782e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 18932056,
      "step": 29025
    },
    {
      "epoch": 15.214884696016771,
      "grad_norm": 0.6063945889472961,
      "learning_rate": 8.22522206673262e-06,
      "loss": 0.0114,
      "num_input_tokens_seen": 18935896,
      "step": 29030
    },
    {
      "epoch": 15.217505241090146,
      "grad_norm": 0.43707627058029175,
      "learning_rate": 8.216745695069589e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 18938968,
      "step": 29035
    },
    {
      "epoch": 15.220125786163521,
      "grad_norm": 1.5546252727508545,
      "learning_rate": 8.208272834271894e-06,
      "loss": 0.0843,
      "num_input_tokens_seen": 18943000,
      "step": 29040
    },
    {
      "epoch": 15.222746331236896,
      "grad_norm": 1.0729135274887085,
      "learning_rate": 8.19980348611194e-06,
      "loss": 0.0203,
      "num_input_tokens_seen": 18946104,
      "step": 29045
    },
    {
      "epoch": 15.225366876310272,
      "grad_norm": 1.0661183595657349,
      "learning_rate": 8.191337652361439e-06,
      "loss": 0.0197,
      "num_input_tokens_seen": 18950328,
      "step": 29050
    },
    {
      "epoch": 15.227987421383649,
      "grad_norm": 3.3163034915924072,
      "learning_rate": 8.182875334791332e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 18953144,
      "step": 29055
    },
    {
      "epoch": 15.230607966457024,
      "grad_norm": 2.8245368003845215,
      "learning_rate": 8.174416535171841e-06,
      "loss": 0.059,
      "num_input_tokens_seen": 18960248,
      "step": 29060
    },
    {
      "epoch": 15.233228511530399,
      "grad_norm": 0.5017924308776855,
      "learning_rate": 8.165961255272467e-06,
      "loss": 0.0129,
      "num_input_tokens_seen": 18962744,
      "step": 29065
    },
    {
      "epoch": 15.235849056603774,
      "grad_norm": 0.3244820237159729,
      "learning_rate": 8.157509496861935e-06,
      "loss": 0.0104,
      "num_input_tokens_seen": 18965816,
      "step": 29070
    },
    {
      "epoch": 15.23846960167715,
      "grad_norm": 0.19462719559669495,
      "learning_rate": 8.149061261708266e-06,
      "loss": 0.0108,
      "num_input_tokens_seen": 18969560,
      "step": 29075
    },
    {
      "epoch": 15.241090146750524,
      "grad_norm": 0.0232393778860569,
      "learning_rate": 8.140616551578745e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 18972792,
      "step": 29080
    },
    {
      "epoch": 15.2437106918239,
      "grad_norm": 0.46421384811401367,
      "learning_rate": 8.13217536823989e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 18975640,
      "step": 29085
    },
    {
      "epoch": 15.246331236897275,
      "grad_norm": 5.026428699493408,
      "learning_rate": 8.12373771345752e-06,
      "loss": 0.0099,
      "num_input_tokens_seen": 18978840,
      "step": 29090
    },
    {
      "epoch": 15.24895178197065,
      "grad_norm": 0.18660050630569458,
      "learning_rate": 8.115303588996676e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 18981464,
      "step": 29095
    },
    {
      "epoch": 15.251572327044025,
      "grad_norm": 1.6285384893417358,
      "learning_rate": 8.10687299662169e-06,
      "loss": 0.0139,
      "num_input_tokens_seen": 18985368,
      "step": 29100
    },
    {
      "epoch": 15.2541928721174,
      "grad_norm": 4.570073127746582,
      "learning_rate": 8.098445938096147e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 18988120,
      "step": 29105
    },
    {
      "epoch": 15.256813417190775,
      "grad_norm": 0.627843976020813,
      "learning_rate": 8.090022415182898e-06,
      "loss": 0.0179,
      "num_input_tokens_seen": 18991640,
      "step": 29110
    },
    {
      "epoch": 15.25943396226415,
      "grad_norm": 0.24971304833889008,
      "learning_rate": 8.081602429644033e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 18994552,
      "step": 29115
    },
    {
      "epoch": 15.262054507337526,
      "grad_norm": 1.8813836574554443,
      "learning_rate": 8.073185983240932e-06,
      "loss": 0.0679,
      "num_input_tokens_seen": 18999224,
      "step": 29120
    },
    {
      "epoch": 15.264675052410901,
      "grad_norm": 0.7103570699691772,
      "learning_rate": 8.064773077734206e-06,
      "loss": 0.0389,
      "num_input_tokens_seen": 19002648,
      "step": 29125
    },
    {
      "epoch": 15.267295597484276,
      "grad_norm": 0.7070103287696838,
      "learning_rate": 8.05636371488374e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 19005976,
      "step": 29130
    },
    {
      "epoch": 15.269916142557651,
      "grad_norm": 0.13274425268173218,
      "learning_rate": 8.047957896448696e-06,
      "loss": 0.0111,
      "num_input_tokens_seen": 19009144,
      "step": 29135
    },
    {
      "epoch": 15.272536687631026,
      "grad_norm": 0.6058032512664795,
      "learning_rate": 8.039555624187451e-06,
      "loss": 0.0099,
      "num_input_tokens_seen": 19011896,
      "step": 29140
    },
    {
      "epoch": 15.275157232704402,
      "grad_norm": 0.720441460609436,
      "learning_rate": 8.031156899857681e-06,
      "loss": 0.0289,
      "num_input_tokens_seen": 19015256,
      "step": 29145
    },
    {
      "epoch": 15.277777777777779,
      "grad_norm": 1.0394648313522339,
      "learning_rate": 8.022761725216288e-06,
      "loss": 0.0214,
      "num_input_tokens_seen": 19018360,
      "step": 29150
    },
    {
      "epoch": 15.280398322851154,
      "grad_norm": 1.1464051008224487,
      "learning_rate": 8.014370102019456e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 19022136,
      "step": 29155
    },
    {
      "epoch": 15.283018867924529,
      "grad_norm": 0.7884352803230286,
      "learning_rate": 8.005982032022616e-06,
      "loss": 0.0302,
      "num_input_tokens_seen": 19024920,
      "step": 29160
    },
    {
      "epoch": 15.285639412997904,
      "grad_norm": 12.31545639038086,
      "learning_rate": 7.997597516980467e-06,
      "loss": 0.0309,
      "num_input_tokens_seen": 19029464,
      "step": 29165
    },
    {
      "epoch": 15.28825995807128,
      "grad_norm": 0.8375712633132935,
      "learning_rate": 7.989216558646942e-06,
      "loss": 0.0134,
      "num_input_tokens_seen": 19033272,
      "step": 29170
    },
    {
      "epoch": 15.290880503144654,
      "grad_norm": 1.7169628143310547,
      "learning_rate": 7.980839158775235e-06,
      "loss": 0.0174,
      "num_input_tokens_seen": 19035736,
      "step": 29175
    },
    {
      "epoch": 15.29350104821803,
      "grad_norm": 0.42722588777542114,
      "learning_rate": 7.972465319117814e-06,
      "loss": 0.0135,
      "num_input_tokens_seen": 19038904,
      "step": 29180
    },
    {
      "epoch": 15.296121593291405,
      "grad_norm": 0.4951830506324768,
      "learning_rate": 7.964095041426386e-06,
      "loss": 0.0141,
      "num_input_tokens_seen": 19042424,
      "step": 29185
    },
    {
      "epoch": 15.29874213836478,
      "grad_norm": 0.461463063955307,
      "learning_rate": 7.95572832745192e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 19045784,
      "step": 29190
    },
    {
      "epoch": 15.301362683438155,
      "grad_norm": 0.5137214064598083,
      "learning_rate": 7.947365178944643e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 19049016,
      "step": 29195
    },
    {
      "epoch": 15.30398322851153,
      "grad_norm": 0.6121444702148438,
      "learning_rate": 7.939005597654025e-06,
      "loss": 0.0095,
      "num_input_tokens_seen": 19052536,
      "step": 29200
    },
    {
      "epoch": 15.306603773584905,
      "grad_norm": 1.6867741346359253,
      "learning_rate": 7.930649585328787e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 19055480,
      "step": 29205
    },
    {
      "epoch": 15.30922431865828,
      "grad_norm": 1.463381290435791,
      "learning_rate": 7.922297143716918e-06,
      "loss": 0.0156,
      "num_input_tokens_seen": 19059416,
      "step": 29210
    },
    {
      "epoch": 15.311844863731656,
      "grad_norm": 0.27514907717704773,
      "learning_rate": 7.913948274565652e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 19062360,
      "step": 29215
    },
    {
      "epoch": 15.314465408805031,
      "grad_norm": 0.34350264072418213,
      "learning_rate": 7.905602979621491e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 19065400,
      "step": 29220
    },
    {
      "epoch": 15.317085953878406,
      "grad_norm": 1.6153029203414917,
      "learning_rate": 7.897261260630159e-06,
      "loss": 0.0117,
      "num_input_tokens_seen": 19068376,
      "step": 29225
    },
    {
      "epoch": 15.319706498951781,
      "grad_norm": 0.20281651616096497,
      "learning_rate": 7.888923119336647e-06,
      "loss": 0.0115,
      "num_input_tokens_seen": 19071032,
      "step": 29230
    },
    {
      "epoch": 15.322327044025156,
      "grad_norm": 0.4626905918121338,
      "learning_rate": 7.880588557485203e-06,
      "loss": 0.02,
      "num_input_tokens_seen": 19073944,
      "step": 29235
    },
    {
      "epoch": 15.324947589098532,
      "grad_norm": 1.510143518447876,
      "learning_rate": 7.872257576819325e-06,
      "loss": 0.0361,
      "num_input_tokens_seen": 19076760,
      "step": 29240
    },
    {
      "epoch": 15.327568134171909,
      "grad_norm": 1.1796048879623413,
      "learning_rate": 7.863930179081769e-06,
      "loss": 0.0286,
      "num_input_tokens_seen": 19079480,
      "step": 29245
    },
    {
      "epoch": 15.330188679245284,
      "grad_norm": 1.4963984489440918,
      "learning_rate": 7.855606366014517e-06,
      "loss": 0.0236,
      "num_input_tokens_seen": 19081944,
      "step": 29250
    },
    {
      "epoch": 15.332809224318659,
      "grad_norm": 1.434722661972046,
      "learning_rate": 7.847286139358814e-06,
      "loss": 0.0116,
      "num_input_tokens_seen": 19085272,
      "step": 29255
    },
    {
      "epoch": 15.335429769392034,
      "grad_norm": 0.8183779716491699,
      "learning_rate": 7.838969500855162e-06,
      "loss": 0.0145,
      "num_input_tokens_seen": 19088152,
      "step": 29260
    },
    {
      "epoch": 15.33805031446541,
      "grad_norm": 0.7341460585594177,
      "learning_rate": 7.830656452243307e-06,
      "loss": 0.0381,
      "num_input_tokens_seen": 19091576,
      "step": 29265
    },
    {
      "epoch": 15.340670859538784,
      "grad_norm": 1.731497883796692,
      "learning_rate": 7.822346995262241e-06,
      "loss": 0.0278,
      "num_input_tokens_seen": 19095672,
      "step": 29270
    },
    {
      "epoch": 15.34329140461216,
      "grad_norm": 1.375065565109253,
      "learning_rate": 7.814041131650223e-06,
      "loss": 0.0224,
      "num_input_tokens_seen": 19099416,
      "step": 29275
    },
    {
      "epoch": 15.345911949685535,
      "grad_norm": 1.160731554031372,
      "learning_rate": 7.805738863144731e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 19102296,
      "step": 29280
    },
    {
      "epoch": 15.34853249475891,
      "grad_norm": 0.12598510086536407,
      "learning_rate": 7.797440191482496e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 19105432,
      "step": 29285
    },
    {
      "epoch": 15.351153039832285,
      "grad_norm": 0.3217839300632477,
      "learning_rate": 7.789145118399518e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 19108632,
      "step": 29290
    },
    {
      "epoch": 15.35377358490566,
      "grad_norm": 1.5694729089736938,
      "learning_rate": 7.780853645631032e-06,
      "loss": 0.0358,
      "num_input_tokens_seen": 19111832,
      "step": 29295
    },
    {
      "epoch": 15.356394129979035,
      "grad_norm": 0.4138725996017456,
      "learning_rate": 7.772565774911522e-06,
      "loss": 0.0081,
      "num_input_tokens_seen": 19115288,
      "step": 29300
    },
    {
      "epoch": 15.35901467505241,
      "grad_norm": 0.5549290776252747,
      "learning_rate": 7.76428150797471e-06,
      "loss": 0.0076,
      "num_input_tokens_seen": 19118456,
      "step": 29305
    },
    {
      "epoch": 15.361635220125786,
      "grad_norm": 2.709746837615967,
      "learning_rate": 7.756000846553562e-06,
      "loss": 0.0197,
      "num_input_tokens_seen": 19121272,
      "step": 29310
    },
    {
      "epoch": 15.364255765199161,
      "grad_norm": 1.0860228538513184,
      "learning_rate": 7.74772379238031e-06,
      "loss": 0.0081,
      "num_input_tokens_seen": 19124440,
      "step": 29315
    },
    {
      "epoch": 15.366876310272536,
      "grad_norm": 1.6905102729797363,
      "learning_rate": 7.739450347186417e-06,
      "loss": 0.0183,
      "num_input_tokens_seen": 19127672,
      "step": 29320
    },
    {
      "epoch": 15.369496855345911,
      "grad_norm": 0.713699996471405,
      "learning_rate": 7.731180512702599e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 19130776,
      "step": 29325
    },
    {
      "epoch": 15.372117400419286,
      "grad_norm": 0.39815738797187805,
      "learning_rate": 7.722914290658795e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 19133720,
      "step": 29330
    },
    {
      "epoch": 15.374737945492662,
      "grad_norm": 0.24201948940753937,
      "learning_rate": 7.714651682784224e-06,
      "loss": 0.0142,
      "num_input_tokens_seen": 19137112,
      "step": 29335
    },
    {
      "epoch": 15.377358490566039,
      "grad_norm": 0.611834704875946,
      "learning_rate": 7.706392690807313e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 19140408,
      "step": 29340
    },
    {
      "epoch": 15.379979035639414,
      "grad_norm": 2.63128662109375,
      "learning_rate": 7.69813731645575e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 19143672,
      "step": 29345
    },
    {
      "epoch": 15.382599580712789,
      "grad_norm": 2.457023859024048,
      "learning_rate": 7.689885561456475e-06,
      "loss": 0.014,
      "num_input_tokens_seen": 19147288,
      "step": 29350
    },
    {
      "epoch": 15.385220125786164,
      "grad_norm": 1.1188758611679077,
      "learning_rate": 7.681637427535663e-06,
      "loss": 0.0185,
      "num_input_tokens_seen": 19149688,
      "step": 29355
    },
    {
      "epoch": 15.38784067085954,
      "grad_norm": 1.0760329961776733,
      "learning_rate": 7.673392916418715e-06,
      "loss": 0.0331,
      "num_input_tokens_seen": 19153752,
      "step": 29360
    },
    {
      "epoch": 15.390461215932914,
      "grad_norm": 1.3546955585479736,
      "learning_rate": 7.665152029830303e-06,
      "loss": 0.0161,
      "num_input_tokens_seen": 19157208,
      "step": 29365
    },
    {
      "epoch": 15.39308176100629,
      "grad_norm": 0.28667235374450684,
      "learning_rate": 7.656914769494314e-06,
      "loss": 0.0177,
      "num_input_tokens_seen": 19161656,
      "step": 29370
    },
    {
      "epoch": 15.395702306079665,
      "grad_norm": 0.9909310936927795,
      "learning_rate": 7.648681137133892e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 19164824,
      "step": 29375
    },
    {
      "epoch": 15.39832285115304,
      "grad_norm": 2.5390355587005615,
      "learning_rate": 7.640451134471432e-06,
      "loss": 0.0189,
      "num_input_tokens_seen": 19167480,
      "step": 29380
    },
    {
      "epoch": 15.400943396226415,
      "grad_norm": 0.27574247121810913,
      "learning_rate": 7.632224763228538e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 19169752,
      "step": 29385
    },
    {
      "epoch": 15.40356394129979,
      "grad_norm": 0.639470636844635,
      "learning_rate": 7.6240020251260896e-06,
      "loss": 0.0187,
      "num_input_tokens_seen": 19172696,
      "step": 29390
    },
    {
      "epoch": 15.406184486373165,
      "grad_norm": 0.6363092064857483,
      "learning_rate": 7.615782921884174e-06,
      "loss": 0.0209,
      "num_input_tokens_seen": 19176600,
      "step": 29395
    },
    {
      "epoch": 15.40880503144654,
      "grad_norm": 0.3122979998588562,
      "learning_rate": 7.607567455222142e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 19179000,
      "step": 29400
    },
    {
      "epoch": 15.411425576519916,
      "grad_norm": 1.1845002174377441,
      "learning_rate": 7.599355626858582e-06,
      "loss": 0.0258,
      "num_input_tokens_seen": 19182264,
      "step": 29405
    },
    {
      "epoch": 15.414046121593291,
      "grad_norm": 0.6197790503501892,
      "learning_rate": 7.591147438511298e-06,
      "loss": 0.0093,
      "num_input_tokens_seen": 19185240,
      "step": 29410
    },
    {
      "epoch": 15.416666666666666,
      "grad_norm": 0.6896967887878418,
      "learning_rate": 7.5829428918973625e-06,
      "loss": 0.0095,
      "num_input_tokens_seen": 19188920,
      "step": 29415
    },
    {
      "epoch": 15.419287211740041,
      "grad_norm": 3.3137543201446533,
      "learning_rate": 7.574741988733075e-06,
      "loss": 0.0162,
      "num_input_tokens_seen": 19191672,
      "step": 29420
    },
    {
      "epoch": 15.421907756813416,
      "grad_norm": 3.4136486053466797,
      "learning_rate": 7.56654473073396e-06,
      "loss": 0.024,
      "num_input_tokens_seen": 19194744,
      "step": 29425
    },
    {
      "epoch": 15.424528301886792,
      "grad_norm": 1.4167801141738892,
      "learning_rate": 7.558351119614796e-06,
      "loss": 0.0149,
      "num_input_tokens_seen": 19197848,
      "step": 29430
    },
    {
      "epoch": 15.427148846960169,
      "grad_norm": 0.6769042015075684,
      "learning_rate": 7.5501611570896e-06,
      "loss": 0.0104,
      "num_input_tokens_seen": 19201432,
      "step": 29435
    },
    {
      "epoch": 15.429769392033544,
      "grad_norm": 0.38565194606781006,
      "learning_rate": 7.541974844871602e-06,
      "loss": 0.0105,
      "num_input_tokens_seen": 19205304,
      "step": 29440
    },
    {
      "epoch": 15.432389937106919,
      "grad_norm": 0.6011795401573181,
      "learning_rate": 7.5337921846732965e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 19208888,
      "step": 29445
    },
    {
      "epoch": 15.435010482180294,
      "grad_norm": 0.5976409316062927,
      "learning_rate": 7.525613178206409e-06,
      "loss": 0.0136,
      "num_input_tokens_seen": 19212216,
      "step": 29450
    },
    {
      "epoch": 15.43763102725367,
      "grad_norm": 2.8244786262512207,
      "learning_rate": 7.517437827181878e-06,
      "loss": 0.0362,
      "num_input_tokens_seen": 19215736,
      "step": 29455
    },
    {
      "epoch": 15.440251572327044,
      "grad_norm": 1.6305640935897827,
      "learning_rate": 7.509266133309908e-06,
      "loss": 0.0188,
      "num_input_tokens_seen": 19219224,
      "step": 29460
    },
    {
      "epoch": 15.44287211740042,
      "grad_norm": 0.7816621661186218,
      "learning_rate": 7.501098098299914e-06,
      "loss": 0.011,
      "num_input_tokens_seen": 19222424,
      "step": 29465
    },
    {
      "epoch": 15.445492662473795,
      "grad_norm": 0.7516186833381653,
      "learning_rate": 7.492933723860557e-06,
      "loss": 0.0119,
      "num_input_tokens_seen": 19225784,
      "step": 29470
    },
    {
      "epoch": 15.44811320754717,
      "grad_norm": 1.5169107913970947,
      "learning_rate": 7.4847730116997334e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 19228600,
      "step": 29475
    },
    {
      "epoch": 15.450733752620545,
      "grad_norm": 3.930760145187378,
      "learning_rate": 7.476615963524583e-06,
      "loss": 0.0498,
      "num_input_tokens_seen": 19231864,
      "step": 29480
    },
    {
      "epoch": 15.45335429769392,
      "grad_norm": 1.9598371982574463,
      "learning_rate": 7.468462581041452e-06,
      "loss": 0.0186,
      "num_input_tokens_seen": 19235480,
      "step": 29485
    },
    {
      "epoch": 15.455974842767295,
      "grad_norm": 0.6859087347984314,
      "learning_rate": 7.460312865955932e-06,
      "loss": 0.0095,
      "num_input_tokens_seen": 19238904,
      "step": 29490
    },
    {
      "epoch": 15.45859538784067,
      "grad_norm": 0.5808824300765991,
      "learning_rate": 7.4521668199728584e-06,
      "loss": 0.0231,
      "num_input_tokens_seen": 19242040,
      "step": 29495
    },
    {
      "epoch": 15.461215932914046,
      "grad_norm": 2.072489023208618,
      "learning_rate": 7.4440244447962884e-06,
      "loss": 0.0287,
      "num_input_tokens_seen": 19245336,
      "step": 29500
    },
    {
      "epoch": 15.463836477987421,
      "grad_norm": 1.7833330631256104,
      "learning_rate": 7.435885742129523e-06,
      "loss": 0.0492,
      "num_input_tokens_seen": 19250264,
      "step": 29505
    },
    {
      "epoch": 15.466457023060796,
      "grad_norm": 1.0321829319000244,
      "learning_rate": 7.427750713675071e-06,
      "loss": 0.045,
      "num_input_tokens_seen": 19253592,
      "step": 29510
    },
    {
      "epoch": 15.469077568134171,
      "grad_norm": 1.653998613357544,
      "learning_rate": 7.419619361134702e-06,
      "loss": 0.0144,
      "num_input_tokens_seen": 19256728,
      "step": 29515
    },
    {
      "epoch": 15.471698113207546,
      "grad_norm": 0.5973173975944519,
      "learning_rate": 7.411491686209387e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 19259640,
      "step": 29520
    },
    {
      "epoch": 15.474318658280922,
      "grad_norm": 0.2582685947418213,
      "learning_rate": 7.4033676905993525e-06,
      "loss": 0.009,
      "num_input_tokens_seen": 19263128,
      "step": 29525
    },
    {
      "epoch": 15.476939203354299,
      "grad_norm": 0.5622972846031189,
      "learning_rate": 7.3952473760040444e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 19267928,
      "step": 29530
    },
    {
      "epoch": 15.479559748427674,
      "grad_norm": 1.0745402574539185,
      "learning_rate": 7.387130744122148e-06,
      "loss": 0.0235,
      "num_input_tokens_seen": 19270552,
      "step": 29535
    },
    {
      "epoch": 15.482180293501049,
      "grad_norm": 1.3485400676727295,
      "learning_rate": 7.379017796651558e-06,
      "loss": 0.0105,
      "num_input_tokens_seen": 19274072,
      "step": 29540
    },
    {
      "epoch": 15.484800838574424,
      "grad_norm": 8.838066101074219,
      "learning_rate": 7.3709085352894085e-06,
      "loss": 0.0139,
      "num_input_tokens_seen": 19277048,
      "step": 29545
    },
    {
      "epoch": 15.4874213836478,
      "grad_norm": 0.8644327521324158,
      "learning_rate": 7.362802961732071e-06,
      "loss": 0.0096,
      "num_input_tokens_seen": 19279576,
      "step": 29550
    },
    {
      "epoch": 15.490041928721174,
      "grad_norm": 0.7816793322563171,
      "learning_rate": 7.354701077675136e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 19282200,
      "step": 29555
    },
    {
      "epoch": 15.49266247379455,
      "grad_norm": 0.50281822681427,
      "learning_rate": 7.346602884813439e-06,
      "loss": 0.0195,
      "num_input_tokens_seen": 19285336,
      "step": 29560
    },
    {
      "epoch": 15.495283018867925,
      "grad_norm": 0.5497046709060669,
      "learning_rate": 7.338508384841014e-06,
      "loss": 0.0142,
      "num_input_tokens_seen": 19288344,
      "step": 29565
    },
    {
      "epoch": 15.4979035639413,
      "grad_norm": 0.9082513451576233,
      "learning_rate": 7.3304175794511345e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 19291928,
      "step": 29570
    },
    {
      "epoch": 15.500524109014675,
      "grad_norm": 0.159030944108963,
      "learning_rate": 7.3223304703363135e-06,
      "loss": 0.0115,
      "num_input_tokens_seen": 19295736,
      "step": 29575
    },
    {
      "epoch": 15.50314465408805,
      "grad_norm": 0.13328564167022705,
      "learning_rate": 7.314247059188281e-06,
      "loss": 0.0162,
      "num_input_tokens_seen": 19298744,
      "step": 29580
    },
    {
      "epoch": 15.505765199161425,
      "grad_norm": 2.947488784790039,
      "learning_rate": 7.306167347697992e-06,
      "loss": 0.0205,
      "num_input_tokens_seen": 19301368,
      "step": 29585
    },
    {
      "epoch": 15.5083857442348,
      "grad_norm": 1.1974211931228638,
      "learning_rate": 7.29809133755564e-06,
      "loss": 0.0126,
      "num_input_tokens_seen": 19304952,
      "step": 29590
    },
    {
      "epoch": 15.511006289308176,
      "grad_norm": 0.866715669631958,
      "learning_rate": 7.290019030450629e-06,
      "loss": 0.026,
      "num_input_tokens_seen": 19308088,
      "step": 29595
    },
    {
      "epoch": 15.51362683438155,
      "grad_norm": 0.9124360680580139,
      "learning_rate": 7.2819504280715794e-06,
      "loss": 0.0146,
      "num_input_tokens_seen": 19310840,
      "step": 29600
    },
    {
      "epoch": 15.516247379454926,
      "grad_norm": 2.3318591117858887,
      "learning_rate": 7.273885532106364e-06,
      "loss": 0.01,
      "num_input_tokens_seen": 19313720,
      "step": 29605
    },
    {
      "epoch": 15.518867924528301,
      "grad_norm": 0.8105962872505188,
      "learning_rate": 7.265824344242064e-06,
      "loss": 0.0179,
      "num_input_tokens_seen": 19318424,
      "step": 29610
    },
    {
      "epoch": 15.521488469601676,
      "grad_norm": 0.46435433626174927,
      "learning_rate": 7.257766866165e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 19322488,
      "step": 29615
    },
    {
      "epoch": 15.524109014675052,
      "grad_norm": 1.389363408088684,
      "learning_rate": 7.24971309956069e-06,
      "loss": 0.0305,
      "num_input_tokens_seen": 19325592,
      "step": 29620
    },
    {
      "epoch": 15.526729559748428,
      "grad_norm": 1.730988621711731,
      "learning_rate": 7.241663046113887e-06,
      "loss": 0.0228,
      "num_input_tokens_seen": 19328408,
      "step": 29625
    },
    {
      "epoch": 15.529350104821804,
      "grad_norm": 0.9496517181396484,
      "learning_rate": 7.233616707508576e-06,
      "loss": 0.0605,
      "num_input_tokens_seen": 19330776,
      "step": 29630
    },
    {
      "epoch": 15.531970649895179,
      "grad_norm": 0.757592499256134,
      "learning_rate": 7.225574085427961e-06,
      "loss": 0.0149,
      "num_input_tokens_seen": 19334360,
      "step": 29635
    },
    {
      "epoch": 15.534591194968554,
      "grad_norm": 0.6412945985794067,
      "learning_rate": 7.217535181554474e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 19336664,
      "step": 29640
    },
    {
      "epoch": 15.53721174004193,
      "grad_norm": 0.47448182106018066,
      "learning_rate": 7.209499997569747e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 19339608,
      "step": 29645
    },
    {
      "epoch": 15.539832285115304,
      "grad_norm": 0.13927288353443146,
      "learning_rate": 7.201468535154663e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 19342712,
      "step": 29650
    },
    {
      "epoch": 15.54245283018868,
      "grad_norm": 1.3093864917755127,
      "learning_rate": 7.193440795989295e-06,
      "loss": 0.0117,
      "num_input_tokens_seen": 19346296,
      "step": 29655
    },
    {
      "epoch": 15.545073375262055,
      "grad_norm": 0.34233975410461426,
      "learning_rate": 7.185416781752965e-06,
      "loss": 0.0188,
      "num_input_tokens_seen": 19349656,
      "step": 29660
    },
    {
      "epoch": 15.54769392033543,
      "grad_norm": 1.5683186054229736,
      "learning_rate": 7.177396494124206e-06,
      "loss": 0.0136,
      "num_input_tokens_seen": 19352280,
      "step": 29665
    },
    {
      "epoch": 15.550314465408805,
      "grad_norm": 0.3635195195674896,
      "learning_rate": 7.169379934780779e-06,
      "loss": 0.0131,
      "num_input_tokens_seen": 19355128,
      "step": 29670
    },
    {
      "epoch": 15.55293501048218,
      "grad_norm": 0.15890270471572876,
      "learning_rate": 7.161367105399644e-06,
      "loss": 0.031,
      "num_input_tokens_seen": 19359096,
      "step": 29675
    },
    {
      "epoch": 15.555555555555555,
      "grad_norm": 0.4685521721839905,
      "learning_rate": 7.153358007656991e-06,
      "loss": 0.0119,
      "num_input_tokens_seen": 19361784,
      "step": 29680
    },
    {
      "epoch": 15.55817610062893,
      "grad_norm": 4.784372806549072,
      "learning_rate": 7.145352643228237e-06,
      "loss": 0.0128,
      "num_input_tokens_seen": 19364824,
      "step": 29685
    },
    {
      "epoch": 15.560796645702306,
      "grad_norm": 0.35899972915649414,
      "learning_rate": 7.137351013788013e-06,
      "loss": 0.0408,
      "num_input_tokens_seen": 19368056,
      "step": 29690
    },
    {
      "epoch": 15.56341719077568,
      "grad_norm": 0.7840330004692078,
      "learning_rate": 7.129353121010177e-06,
      "loss": 0.0106,
      "num_input_tokens_seen": 19371448,
      "step": 29695
    },
    {
      "epoch": 15.566037735849056,
      "grad_norm": 2.3991968631744385,
      "learning_rate": 7.121358966567779e-06,
      "loss": 0.0541,
      "num_input_tokens_seen": 19375448,
      "step": 29700
    },
    {
      "epoch": 15.568658280922431,
      "grad_norm": 1.2447997331619263,
      "learning_rate": 7.113368552133126e-06,
      "loss": 0.0149,
      "num_input_tokens_seen": 19379256,
      "step": 29705
    },
    {
      "epoch": 15.571278825995806,
      "grad_norm": 1.5164936780929565,
      "learning_rate": 7.105381879377701e-06,
      "loss": 0.0253,
      "num_input_tokens_seen": 19383032,
      "step": 29710
    },
    {
      "epoch": 15.573899371069182,
      "grad_norm": 0.23429206013679504,
      "learning_rate": 7.097398949972234e-06,
      "loss": 0.0146,
      "num_input_tokens_seen": 19387640,
      "step": 29715
    },
    {
      "epoch": 15.576519916142558,
      "grad_norm": 13.820521354675293,
      "learning_rate": 7.089419765586672e-06,
      "loss": 0.0274,
      "num_input_tokens_seen": 19389944,
      "step": 29720
    },
    {
      "epoch": 15.579140461215934,
      "grad_norm": 0.19657094776630402,
      "learning_rate": 7.081444327890152e-06,
      "loss": 0.0104,
      "num_input_tokens_seen": 19393464,
      "step": 29725
    },
    {
      "epoch": 15.581761006289309,
      "grad_norm": 0.9522466063499451,
      "learning_rate": 7.073472638551054e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 19395864,
      "step": 29730
    },
    {
      "epoch": 15.584381551362684,
      "grad_norm": 1.6365476846694946,
      "learning_rate": 7.065504699236969e-06,
      "loss": 0.0148,
      "num_input_tokens_seen": 19398552,
      "step": 29735
    },
    {
      "epoch": 15.58700209643606,
      "grad_norm": 0.6555074453353882,
      "learning_rate": 7.057540511614688e-06,
      "loss": 0.0119,
      "num_input_tokens_seen": 19401688,
      "step": 29740
    },
    {
      "epoch": 15.589622641509434,
      "grad_norm": 1.7628810405731201,
      "learning_rate": 7.049580077350232e-06,
      "loss": 0.0124,
      "num_input_tokens_seen": 19404472,
      "step": 29745
    },
    {
      "epoch": 15.59224318658281,
      "grad_norm": 0.5370182394981384,
      "learning_rate": 7.041623398108843e-06,
      "loss": 0.0122,
      "num_input_tokens_seen": 19407544,
      "step": 29750
    },
    {
      "epoch": 15.594863731656185,
      "grad_norm": 1.041642427444458,
      "learning_rate": 7.033670475554949e-06,
      "loss": 0.0166,
      "num_input_tokens_seen": 19410488,
      "step": 29755
    },
    {
      "epoch": 15.59748427672956,
      "grad_norm": 1.611209511756897,
      "learning_rate": 7.025721311352224e-06,
      "loss": 0.0342,
      "num_input_tokens_seen": 19414360,
      "step": 29760
    },
    {
      "epoch": 15.600104821802935,
      "grad_norm": 0.5352613925933838,
      "learning_rate": 7.017775907163546e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 19418072,
      "step": 29765
    },
    {
      "epoch": 15.60272536687631,
      "grad_norm": 1.3048139810562134,
      "learning_rate": 7.009834264650989e-06,
      "loss": 0.014,
      "num_input_tokens_seen": 19421016,
      "step": 29770
    },
    {
      "epoch": 15.605345911949685,
      "grad_norm": 1.1008777618408203,
      "learning_rate": 7.001896385475867e-06,
      "loss": 0.0231,
      "num_input_tokens_seen": 19424568,
      "step": 29775
    },
    {
      "epoch": 15.60796645702306,
      "grad_norm": 0.34472814202308655,
      "learning_rate": 6.993962271298682e-06,
      "loss": 0.0152,
      "num_input_tokens_seen": 19427352,
      "step": 29780
    },
    {
      "epoch": 15.610587002096436,
      "grad_norm": 0.6866914629936218,
      "learning_rate": 6.986031923779166e-06,
      "loss": 0.01,
      "num_input_tokens_seen": 19430328,
      "step": 29785
    },
    {
      "epoch": 15.61320754716981,
      "grad_norm": 0.24841803312301636,
      "learning_rate": 6.978105344576264e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 19433848,
      "step": 29790
    },
    {
      "epoch": 15.615828092243186,
      "grad_norm": 3.966459274291992,
      "learning_rate": 6.970182535348111e-06,
      "loss": 0.0219,
      "num_input_tokens_seen": 19436888,
      "step": 29795
    },
    {
      "epoch": 15.618448637316561,
      "grad_norm": 1.0262062549591064,
      "learning_rate": 6.962263497752086e-06,
      "loss": 0.0428,
      "num_input_tokens_seen": 19441176,
      "step": 29800
    },
    {
      "epoch": 15.621069182389936,
      "grad_norm": 18.50726318359375,
      "learning_rate": 6.9543482334447436e-06,
      "loss": 0.0267,
      "num_input_tokens_seen": 19444664,
      "step": 29805
    },
    {
      "epoch": 15.623689727463312,
      "grad_norm": 0.7837218046188354,
      "learning_rate": 6.946436744081875e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 19447736,
      "step": 29810
    },
    {
      "epoch": 15.626310272536688,
      "grad_norm": 0.692810595035553,
      "learning_rate": 6.938529031318472e-06,
      "loss": 0.0149,
      "num_input_tokens_seen": 19451384,
      "step": 29815
    },
    {
      "epoch": 15.628930817610064,
      "grad_norm": 4.33381462097168,
      "learning_rate": 6.9306250968087485e-06,
      "loss": 0.012,
      "num_input_tokens_seen": 19454552,
      "step": 29820
    },
    {
      "epoch": 15.631551362683439,
      "grad_norm": 0.1570337861776352,
      "learning_rate": 6.922724942206102e-06,
      "loss": 0.0162,
      "num_input_tokens_seen": 19457336,
      "step": 29825
    },
    {
      "epoch": 15.634171907756814,
      "grad_norm": 1.2736910581588745,
      "learning_rate": 6.914828569163167e-06,
      "loss": 0.0094,
      "num_input_tokens_seen": 19460344,
      "step": 29830
    },
    {
      "epoch": 15.63679245283019,
      "grad_norm": 1.6681309938430786,
      "learning_rate": 6.906935979331763e-06,
      "loss": 0.0145,
      "num_input_tokens_seen": 19462744,
      "step": 29835
    },
    {
      "epoch": 15.639412997903564,
      "grad_norm": 0.761781632900238,
      "learning_rate": 6.8990471743629356e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 19465880,
      "step": 29840
    },
    {
      "epoch": 15.64203354297694,
      "grad_norm": 1.123420238494873,
      "learning_rate": 6.89116215590693e-06,
      "loss": 0.0147,
      "num_input_tokens_seen": 19468632,
      "step": 29845
    },
    {
      "epoch": 15.644654088050315,
      "grad_norm": 4.143928527832031,
      "learning_rate": 6.8832809256132146e-06,
      "loss": 0.0338,
      "num_input_tokens_seen": 19471736,
      "step": 29850
    },
    {
      "epoch": 15.64727463312369,
      "grad_norm": 0.8680365085601807,
      "learning_rate": 6.875403485130444e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 19474424,
      "step": 29855
    },
    {
      "epoch": 15.649895178197065,
      "grad_norm": 0.01658419705927372,
      "learning_rate": 6.867529836106479e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 19477400,
      "step": 29860
    },
    {
      "epoch": 15.65251572327044,
      "grad_norm": 0.5500559210777283,
      "learning_rate": 6.8596599801884045e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 19480024,
      "step": 29865
    },
    {
      "epoch": 15.655136268343815,
      "grad_norm": 0.2800769805908203,
      "learning_rate": 6.851793919022509e-06,
      "loss": 0.0209,
      "num_input_tokens_seen": 19483256,
      "step": 29870
    },
    {
      "epoch": 15.65775681341719,
      "grad_norm": 0.3011908233165741,
      "learning_rate": 6.843931654254285e-06,
      "loss": 0.0141,
      "num_input_tokens_seen": 19486392,
      "step": 29875
    },
    {
      "epoch": 15.660377358490566,
      "grad_norm": 0.20673272013664246,
      "learning_rate": 6.836073187528425e-06,
      "loss": 0.0199,
      "num_input_tokens_seen": 19489464,
      "step": 29880
    },
    {
      "epoch": 15.66299790356394,
      "grad_norm": 0.5011268854141235,
      "learning_rate": 6.828218520488821e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 19492952,
      "step": 29885
    },
    {
      "epoch": 15.665618448637316,
      "grad_norm": 2.53262996673584,
      "learning_rate": 6.820367654778589e-06,
      "loss": 0.0257,
      "num_input_tokens_seen": 19495928,
      "step": 29890
    },
    {
      "epoch": 15.668238993710691,
      "grad_norm": 1.418296217918396,
      "learning_rate": 6.812520592040039e-06,
      "loss": 0.0106,
      "num_input_tokens_seen": 19499800,
      "step": 29895
    },
    {
      "epoch": 15.670859538784066,
      "grad_norm": 0.7058477401733398,
      "learning_rate": 6.804677333914689e-06,
      "loss": 0.0136,
      "num_input_tokens_seen": 19504408,
      "step": 29900
    },
    {
      "epoch": 15.673480083857442,
      "grad_norm": 0.8028666973114014,
      "learning_rate": 6.796837882043261e-06,
      "loss": 0.0091,
      "num_input_tokens_seen": 19507576,
      "step": 29905
    },
    {
      "epoch": 15.676100628930818,
      "grad_norm": 2.5593836307525635,
      "learning_rate": 6.7890022380656785e-06,
      "loss": 0.0786,
      "num_input_tokens_seen": 19510712,
      "step": 29910
    },
    {
      "epoch": 15.678721174004194,
      "grad_norm": 0.37010860443115234,
      "learning_rate": 6.781170403621056e-06,
      "loss": 0.0124,
      "num_input_tokens_seen": 19514168,
      "step": 29915
    },
    {
      "epoch": 15.681341719077569,
      "grad_norm": 7.754566192626953,
      "learning_rate": 6.773342380347736e-06,
      "loss": 0.0187,
      "num_input_tokens_seen": 19517304,
      "step": 29920
    },
    {
      "epoch": 15.683962264150944,
      "grad_norm": 0.6967717409133911,
      "learning_rate": 6.765518169883248e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 19520504,
      "step": 29925
    },
    {
      "epoch": 15.68658280922432,
      "grad_norm": 0.3486068844795227,
      "learning_rate": 6.757697773864338e-06,
      "loss": 0.0357,
      "num_input_tokens_seen": 19527000,
      "step": 29930
    },
    {
      "epoch": 15.689203354297694,
      "grad_norm": 1.671297311782837,
      "learning_rate": 6.749881193926932e-06,
      "loss": 0.0156,
      "num_input_tokens_seen": 19530008,
      "step": 29935
    },
    {
      "epoch": 15.69182389937107,
      "grad_norm": 1.731716513633728,
      "learning_rate": 6.742068431706167e-06,
      "loss": 0.0118,
      "num_input_tokens_seen": 19533208,
      "step": 29940
    },
    {
      "epoch": 15.694444444444445,
      "grad_norm": 1.2079068422317505,
      "learning_rate": 6.734259488836386e-06,
      "loss": 0.1045,
      "num_input_tokens_seen": 19536248,
      "step": 29945
    },
    {
      "epoch": 15.69706498951782,
      "grad_norm": 0.7562603950500488,
      "learning_rate": 6.726454366951135e-06,
      "loss": 0.0128,
      "num_input_tokens_seen": 19539480,
      "step": 29950
    },
    {
      "epoch": 15.699685534591195,
      "grad_norm": 2.351795196533203,
      "learning_rate": 6.7186530676831616e-06,
      "loss": 0.0211,
      "num_input_tokens_seen": 19542584,
      "step": 29955
    },
    {
      "epoch": 15.70230607966457,
      "grad_norm": 3.9911129474639893,
      "learning_rate": 6.710855592664403e-06,
      "loss": 0.024,
      "num_input_tokens_seen": 19546296,
      "step": 29960
    },
    {
      "epoch": 15.704926624737945,
      "grad_norm": 0.1826147735118866,
      "learning_rate": 6.703061943525993e-06,
      "loss": 0.0142,
      "num_input_tokens_seen": 19549400,
      "step": 29965
    },
    {
      "epoch": 15.70754716981132,
      "grad_norm": 0.5518291592597961,
      "learning_rate": 6.69527212189828e-06,
      "loss": 0.0311,
      "num_input_tokens_seen": 19552984,
      "step": 29970
    },
    {
      "epoch": 15.710167714884696,
      "grad_norm": 1.2945505380630493,
      "learning_rate": 6.687486129410811e-06,
      "loss": 0.0122,
      "num_input_tokens_seen": 19555448,
      "step": 29975
    },
    {
      "epoch": 15.71278825995807,
      "grad_norm": 0.732708215713501,
      "learning_rate": 6.679703967692322e-06,
      "loss": 0.0127,
      "num_input_tokens_seen": 19558136,
      "step": 29980
    },
    {
      "epoch": 15.715408805031446,
      "grad_norm": 1.2850173711776733,
      "learning_rate": 6.67192563837076e-06,
      "loss": 0.0076,
      "num_input_tokens_seen": 19560824,
      "step": 29985
    },
    {
      "epoch": 15.718029350104821,
      "grad_norm": 1.289280891418457,
      "learning_rate": 6.664151143073258e-06,
      "loss": 0.0182,
      "num_input_tokens_seen": 19564472,
      "step": 29990
    },
    {
      "epoch": 15.720649895178196,
      "grad_norm": 0.17729245126247406,
      "learning_rate": 6.656380483426141e-06,
      "loss": 0.0254,
      "num_input_tokens_seen": 19567960,
      "step": 29995
    },
    {
      "epoch": 15.723270440251572,
      "grad_norm": 0.4629959166049957,
      "learning_rate": 6.648613661054956e-06,
      "loss": 0.0159,
      "num_input_tokens_seen": 19570616,
      "step": 30000
    },
    {
      "epoch": 15.725890985324948,
      "grad_norm": 0.4217473566532135,
      "learning_rate": 6.6408506775844256e-06,
      "loss": 0.0265,
      "num_input_tokens_seen": 19573816,
      "step": 30005
    },
    {
      "epoch": 15.728511530398324,
      "grad_norm": 0.4564160704612732,
      "learning_rate": 6.633091534638492e-06,
      "loss": 0.0137,
      "num_input_tokens_seen": 19577272,
      "step": 30010
    },
    {
      "epoch": 15.731132075471699,
      "grad_norm": 0.22300033271312714,
      "learning_rate": 6.625336233840257e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 19580408,
      "step": 30015
    },
    {
      "epoch": 15.733752620545074,
      "grad_norm": 1.0399399995803833,
      "learning_rate": 6.617584776812064e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 19583640,
      "step": 30020
    },
    {
      "epoch": 15.73637316561845,
      "grad_norm": 1.1357100009918213,
      "learning_rate": 6.6098371651754085e-06,
      "loss": 0.0254,
      "num_input_tokens_seen": 19587128,
      "step": 30025
    },
    {
      "epoch": 15.738993710691824,
      "grad_norm": 0.733355700969696,
      "learning_rate": 6.602093400551012e-06,
      "loss": 0.0141,
      "num_input_tokens_seen": 19590584,
      "step": 30030
    },
    {
      "epoch": 15.7416142557652,
      "grad_norm": 0.13249970972537994,
      "learning_rate": 6.59435348455879e-06,
      "loss": 0.0303,
      "num_input_tokens_seen": 19594264,
      "step": 30035
    },
    {
      "epoch": 15.744234800838575,
      "grad_norm": 0.3069004714488983,
      "learning_rate": 6.586617418817828e-06,
      "loss": 0.016,
      "num_input_tokens_seen": 19597176,
      "step": 30040
    },
    {
      "epoch": 15.74685534591195,
      "grad_norm": 0.4236159920692444,
      "learning_rate": 6.578885204946439e-06,
      "loss": 0.0268,
      "num_input_tokens_seen": 19599896,
      "step": 30045
    },
    {
      "epoch": 15.749475890985325,
      "grad_norm": 0.22829319536685944,
      "learning_rate": 6.571156844562098e-06,
      "loss": 0.0084,
      "num_input_tokens_seen": 19602904,
      "step": 30050
    },
    {
      "epoch": 15.7520964360587,
      "grad_norm": 0.27890118956565857,
      "learning_rate": 6.563432339281497e-06,
      "loss": 0.0163,
      "num_input_tokens_seen": 19606456,
      "step": 30055
    },
    {
      "epoch": 15.754716981132075,
      "grad_norm": 0.22995053231716156,
      "learning_rate": 6.555711690720517e-06,
      "loss": 0.0063,
      "num_input_tokens_seen": 19609304,
      "step": 30060
    },
    {
      "epoch": 15.75733752620545,
      "grad_norm": 0.2773144245147705,
      "learning_rate": 6.547994900494234e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 19613304,
      "step": 30065
    },
    {
      "epoch": 15.759958071278826,
      "grad_norm": 0.9388023614883423,
      "learning_rate": 6.540281970216899e-06,
      "loss": 0.0255,
      "num_input_tokens_seen": 19616952,
      "step": 30070
    },
    {
      "epoch": 15.7625786163522,
      "grad_norm": 0.9773328900337219,
      "learning_rate": 6.5325729015019845e-06,
      "loss": 0.0122,
      "num_input_tokens_seen": 19620056,
      "step": 30075
    },
    {
      "epoch": 15.765199161425576,
      "grad_norm": 2.0716824531555176,
      "learning_rate": 6.524867695962122e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 19623096,
      "step": 30080
    },
    {
      "epoch": 15.767819706498951,
      "grad_norm": 2.4352147579193115,
      "learning_rate": 6.517166355209165e-06,
      "loss": 0.0381,
      "num_input_tokens_seen": 19625624,
      "step": 30085
    },
    {
      "epoch": 15.770440251572326,
      "grad_norm": 2.717266082763672,
      "learning_rate": 6.50946888085415e-06,
      "loss": 0.0319,
      "num_input_tokens_seen": 19628312,
      "step": 30090
    },
    {
      "epoch": 15.773060796645701,
      "grad_norm": 0.6667969822883606,
      "learning_rate": 6.5017752745072855e-06,
      "loss": 0.0084,
      "num_input_tokens_seen": 19632056,
      "step": 30095
    },
    {
      "epoch": 15.775681341719078,
      "grad_norm": 0.8021436929702759,
      "learning_rate": 6.494085537777994e-06,
      "loss": 0.0197,
      "num_input_tokens_seen": 19637560,
      "step": 30100
    },
    {
      "epoch": 15.778301886792454,
      "grad_norm": 1.5233381986618042,
      "learning_rate": 6.48639967227489e-06,
      "loss": 0.0167,
      "num_input_tokens_seen": 19640600,
      "step": 30105
    },
    {
      "epoch": 15.780922431865829,
      "grad_norm": 0.2134053260087967,
      "learning_rate": 6.47871767960575e-06,
      "loss": 0.0172,
      "num_input_tokens_seen": 19643608,
      "step": 30110
    },
    {
      "epoch": 15.783542976939204,
      "grad_norm": 12.632755279541016,
      "learning_rate": 6.471039561377581e-06,
      "loss": 0.0177,
      "num_input_tokens_seen": 19646040,
      "step": 30115
    },
    {
      "epoch": 15.786163522012579,
      "grad_norm": 0.6254862546920776,
      "learning_rate": 6.463365319196538e-06,
      "loss": 0.0427,
      "num_input_tokens_seen": 19648600,
      "step": 30120
    },
    {
      "epoch": 15.788784067085954,
      "grad_norm": 0.9162606596946716,
      "learning_rate": 6.4556949546679905e-06,
      "loss": 0.0141,
      "num_input_tokens_seen": 19652280,
      "step": 30125
    },
    {
      "epoch": 15.79140461215933,
      "grad_norm": 1.5116982460021973,
      "learning_rate": 6.448028469396497e-06,
      "loss": 0.0339,
      "num_input_tokens_seen": 19655736,
      "step": 30130
    },
    {
      "epoch": 15.794025157232705,
      "grad_norm": 0.9483078122138977,
      "learning_rate": 6.440365864985801e-06,
      "loss": 0.0092,
      "num_input_tokens_seen": 19658456,
      "step": 30135
    },
    {
      "epoch": 15.79664570230608,
      "grad_norm": 0.5739696025848389,
      "learning_rate": 6.432707143038818e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 19661688,
      "step": 30140
    },
    {
      "epoch": 15.799266247379455,
      "grad_norm": 0.7007136940956116,
      "learning_rate": 6.4250523051576834e-06,
      "loss": 0.0158,
      "num_input_tokens_seen": 19665400,
      "step": 30145
    },
    {
      "epoch": 15.80188679245283,
      "grad_norm": 0.8811801075935364,
      "learning_rate": 6.417401352943686e-06,
      "loss": 0.0234,
      "num_input_tokens_seen": 19668472,
      "step": 30150
    },
    {
      "epoch": 15.804507337526205,
      "grad_norm": 0.4256729185581207,
      "learning_rate": 6.409754287997322e-06,
      "loss": 0.0099,
      "num_input_tokens_seen": 19672088,
      "step": 30155
    },
    {
      "epoch": 15.80712788259958,
      "grad_norm": 0.40929317474365234,
      "learning_rate": 6.40211111191828e-06,
      "loss": 0.0197,
      "num_input_tokens_seen": 19675864,
      "step": 30160
    },
    {
      "epoch": 15.809748427672956,
      "grad_norm": 1.0548646450042725,
      "learning_rate": 6.394471826305409e-06,
      "loss": 0.0234,
      "num_input_tokens_seen": 19680216,
      "step": 30165
    },
    {
      "epoch": 15.81236897274633,
      "grad_norm": 1.9652701616287231,
      "learning_rate": 6.3868364327567795e-06,
      "loss": 0.0251,
      "num_input_tokens_seen": 19683480,
      "step": 30170
    },
    {
      "epoch": 15.814989517819706,
      "grad_norm": 1.8166369199752808,
      "learning_rate": 6.379204932869606e-06,
      "loss": 0.0317,
      "num_input_tokens_seen": 19687160,
      "step": 30175
    },
    {
      "epoch": 15.817610062893081,
      "grad_norm": 0.3277996778488159,
      "learning_rate": 6.371577328240327e-06,
      "loss": 0.0301,
      "num_input_tokens_seen": 19690424,
      "step": 30180
    },
    {
      "epoch": 15.820230607966456,
      "grad_norm": 0.7668513059616089,
      "learning_rate": 6.363953620464547e-06,
      "loss": 0.0228,
      "num_input_tokens_seen": 19693656,
      "step": 30185
    },
    {
      "epoch": 15.822851153039831,
      "grad_norm": 5.5576629638671875,
      "learning_rate": 6.356333811137064e-06,
      "loss": 0.0223,
      "num_input_tokens_seen": 19696920,
      "step": 30190
    },
    {
      "epoch": 15.825471698113208,
      "grad_norm": 0.4974193572998047,
      "learning_rate": 6.34871790185185e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 19700088,
      "step": 30195
    },
    {
      "epoch": 15.828092243186584,
      "grad_norm": 1.2817901372909546,
      "learning_rate": 6.341105894202057e-06,
      "loss": 0.0112,
      "num_input_tokens_seen": 19703320,
      "step": 30200
    },
    {
      "epoch": 15.830712788259959,
      "grad_norm": 4.105336666107178,
      "learning_rate": 6.333497789780041e-06,
      "loss": 0.0147,
      "num_input_tokens_seen": 19706168,
      "step": 30205
    },
    {
      "epoch": 15.833333333333334,
      "grad_norm": 0.6410620808601379,
      "learning_rate": 6.325893590177329e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 19708984,
      "step": 30210
    },
    {
      "epoch": 15.835953878406709,
      "grad_norm": 0.2816322147846222,
      "learning_rate": 6.318293296984631e-06,
      "loss": 0.0122,
      "num_input_tokens_seen": 19711928,
      "step": 30215
    },
    {
      "epoch": 15.838574423480084,
      "grad_norm": 1.1676092147827148,
      "learning_rate": 6.3106969117918495e-06,
      "loss": 0.015,
      "num_input_tokens_seen": 19715640,
      "step": 30220
    },
    {
      "epoch": 15.84119496855346,
      "grad_norm": 7.203997611999512,
      "learning_rate": 6.303104436188057e-06,
      "loss": 0.0188,
      "num_input_tokens_seen": 19718616,
      "step": 30225
    },
    {
      "epoch": 15.843815513626835,
      "grad_norm": 0.911602795124054,
      "learning_rate": 6.2955158717615036e-06,
      "loss": 0.025,
      "num_input_tokens_seen": 19721688,
      "step": 30230
    },
    {
      "epoch": 15.84643605870021,
      "grad_norm": 0.4443688690662384,
      "learning_rate": 6.287931220099638e-06,
      "loss": 0.0174,
      "num_input_tokens_seen": 19725272,
      "step": 30235
    },
    {
      "epoch": 15.849056603773585,
      "grad_norm": 0.42832833528518677,
      "learning_rate": 6.280350482789082e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 19727960,
      "step": 30240
    },
    {
      "epoch": 15.85167714884696,
      "grad_norm": 0.5211641192436218,
      "learning_rate": 6.272773661415645e-06,
      "loss": 0.0316,
      "num_input_tokens_seen": 19730904,
      "step": 30245
    },
    {
      "epoch": 15.854297693920335,
      "grad_norm": 0.31435292959213257,
      "learning_rate": 6.26520075756431e-06,
      "loss": 0.0208,
      "num_input_tokens_seen": 19734552,
      "step": 30250
    },
    {
      "epoch": 15.85691823899371,
      "grad_norm": 0.3425162434577942,
      "learning_rate": 6.2576317728192304e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 19737400,
      "step": 30255
    },
    {
      "epoch": 15.859538784067086,
      "grad_norm": 2.1611621379852295,
      "learning_rate": 6.250066708763761e-06,
      "loss": 0.0114,
      "num_input_tokens_seen": 19740792,
      "step": 30260
    },
    {
      "epoch": 15.86215932914046,
      "grad_norm": 4.276920318603516,
      "learning_rate": 6.242505566980422e-06,
      "loss": 0.016,
      "num_input_tokens_seen": 19744248,
      "step": 30265
    },
    {
      "epoch": 15.864779874213836,
      "grad_norm": 1.0383199453353882,
      "learning_rate": 6.234948349050931e-06,
      "loss": 0.0084,
      "num_input_tokens_seen": 19746744,
      "step": 30270
    },
    {
      "epoch": 15.867400419287211,
      "grad_norm": 0.517645537853241,
      "learning_rate": 6.227395056556162e-06,
      "loss": 0.01,
      "num_input_tokens_seen": 19749688,
      "step": 30275
    },
    {
      "epoch": 15.870020964360586,
      "grad_norm": 0.7000155448913574,
      "learning_rate": 6.219845691076173e-06,
      "loss": 0.0145,
      "num_input_tokens_seen": 19753208,
      "step": 30280
    },
    {
      "epoch": 15.872641509433961,
      "grad_norm": 0.6976566910743713,
      "learning_rate": 6.212300254190206e-06,
      "loss": 0.0238,
      "num_input_tokens_seen": 19756568,
      "step": 30285
    },
    {
      "epoch": 15.875262054507338,
      "grad_norm": 3.0948452949523926,
      "learning_rate": 6.204758747476688e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 19759576,
      "step": 30290
    },
    {
      "epoch": 15.877882599580714,
      "grad_norm": 0.5730569958686829,
      "learning_rate": 6.1972211725132095e-06,
      "loss": 0.0094,
      "num_input_tokens_seen": 19763608,
      "step": 30295
    },
    {
      "epoch": 15.880503144654089,
      "grad_norm": 1.2307788133621216,
      "learning_rate": 6.189687530876559e-06,
      "loss": 0.0233,
      "num_input_tokens_seen": 19766808,
      "step": 30300
    },
    {
      "epoch": 15.883123689727464,
      "grad_norm": 0.7329351902008057,
      "learning_rate": 6.182157824142676e-06,
      "loss": 0.01,
      "num_input_tokens_seen": 19769752,
      "step": 30305
    },
    {
      "epoch": 15.885744234800839,
      "grad_norm": 1.933807373046875,
      "learning_rate": 6.174632053886681e-06,
      "loss": 0.0184,
      "num_input_tokens_seen": 19772920,
      "step": 30310
    },
    {
      "epoch": 15.888364779874214,
      "grad_norm": 0.35336971282958984,
      "learning_rate": 6.167110221682893e-06,
      "loss": 0.0096,
      "num_input_tokens_seen": 19775512,
      "step": 30315
    },
    {
      "epoch": 15.89098532494759,
      "grad_norm": 0.25916916131973267,
      "learning_rate": 6.159592329104788e-06,
      "loss": 0.0457,
      "num_input_tokens_seen": 19778968,
      "step": 30320
    },
    {
      "epoch": 15.893605870020965,
      "grad_norm": 0.1842079907655716,
      "learning_rate": 6.152078377725032e-06,
      "loss": 0.0154,
      "num_input_tokens_seen": 19781656,
      "step": 30325
    },
    {
      "epoch": 15.89622641509434,
      "grad_norm": 2.6323485374450684,
      "learning_rate": 6.144568369115453e-06,
      "loss": 0.0266,
      "num_input_tokens_seen": 19785112,
      "step": 30330
    },
    {
      "epoch": 15.898846960167715,
      "grad_norm": 0.35014569759368896,
      "learning_rate": 6.137062304847046e-06,
      "loss": 0.0147,
      "num_input_tokens_seen": 19788088,
      "step": 30335
    },
    {
      "epoch": 15.90146750524109,
      "grad_norm": 0.41011670231819153,
      "learning_rate": 6.129560186490008e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 19791352,
      "step": 30340
    },
    {
      "epoch": 15.904088050314465,
      "grad_norm": 0.6718912124633789,
      "learning_rate": 6.122062015613694e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 19795288,
      "step": 30345
    },
    {
      "epoch": 15.90670859538784,
      "grad_norm": 2.1097960472106934,
      "learning_rate": 6.114567793786641e-06,
      "loss": 0.0239,
      "num_input_tokens_seen": 19798328,
      "step": 30350
    },
    {
      "epoch": 15.909329140461216,
      "grad_norm": 0.3541700839996338,
      "learning_rate": 6.107077522576543e-06,
      "loss": 0.0124,
      "num_input_tokens_seen": 19801432,
      "step": 30355
    },
    {
      "epoch": 15.91194968553459,
      "grad_norm": 1.4202417135238647,
      "learning_rate": 6.099591203550292e-06,
      "loss": 0.0144,
      "num_input_tokens_seen": 19804280,
      "step": 30360
    },
    {
      "epoch": 15.914570230607966,
      "grad_norm": 0.6070963740348816,
      "learning_rate": 6.092108838273927e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 19807768,
      "step": 30365
    },
    {
      "epoch": 15.917190775681341,
      "grad_norm": 1.6114380359649658,
      "learning_rate": 6.08463042831268e-06,
      "loss": 0.0115,
      "num_input_tokens_seen": 19810616,
      "step": 30370
    },
    {
      "epoch": 15.919811320754716,
      "grad_norm": 0.8191472887992859,
      "learning_rate": 6.0771559752309496e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 19816472,
      "step": 30375
    },
    {
      "epoch": 15.922431865828091,
      "grad_norm": 1.145078420639038,
      "learning_rate": 6.069685480592313e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 19819928,
      "step": 30380
    },
    {
      "epoch": 15.925052410901468,
      "grad_norm": 0.7951226830482483,
      "learning_rate": 6.062218945959497e-06,
      "loss": 0.0089,
      "num_input_tokens_seen": 19823704,
      "step": 30385
    },
    {
      "epoch": 15.927672955974844,
      "grad_norm": 1.1106464862823486,
      "learning_rate": 6.054756372894435e-06,
      "loss": 0.0167,
      "num_input_tokens_seen": 19826456,
      "step": 30390
    },
    {
      "epoch": 15.930293501048219,
      "grad_norm": 0.9727358818054199,
      "learning_rate": 6.04729776295819e-06,
      "loss": 0.0291,
      "num_input_tokens_seen": 19829848,
      "step": 30395
    },
    {
      "epoch": 15.932914046121594,
      "grad_norm": 0.2916163504123688,
      "learning_rate": 6.0398431177110306e-06,
      "loss": 0.0094,
      "num_input_tokens_seen": 19832856,
      "step": 30400
    },
    {
      "epoch": 15.935534591194969,
      "grad_norm": 4.522012233734131,
      "learning_rate": 6.032392438712389e-06,
      "loss": 0.0124,
      "num_input_tokens_seen": 19835704,
      "step": 30405
    },
    {
      "epoch": 15.938155136268344,
      "grad_norm": 0.638934850692749,
      "learning_rate": 6.024945727520847e-06,
      "loss": 0.0091,
      "num_input_tokens_seen": 19839352,
      "step": 30410
    },
    {
      "epoch": 15.94077568134172,
      "grad_norm": 0.7943176627159119,
      "learning_rate": 6.01750298569419e-06,
      "loss": 0.0367,
      "num_input_tokens_seen": 19842840,
      "step": 30415
    },
    {
      "epoch": 15.943396226415095,
      "grad_norm": 1.610430121421814,
      "learning_rate": 6.010064214789335e-06,
      "loss": 0.0223,
      "num_input_tokens_seen": 19846808,
      "step": 30420
    },
    {
      "epoch": 15.94601677148847,
      "grad_norm": 2.6526341438293457,
      "learning_rate": 6.002629416362399e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 19849816,
      "step": 30425
    },
    {
      "epoch": 15.948637316561845,
      "grad_norm": 1.2571210861206055,
      "learning_rate": 5.995198591968662e-06,
      "loss": 0.0114,
      "num_input_tokens_seen": 19852568,
      "step": 30430
    },
    {
      "epoch": 15.95125786163522,
      "grad_norm": 1.749250888824463,
      "learning_rate": 5.987771743162554e-06,
      "loss": 0.01,
      "num_input_tokens_seen": 19857976,
      "step": 30435
    },
    {
      "epoch": 15.953878406708595,
      "grad_norm": 1.5910183191299438,
      "learning_rate": 5.9803488714976955e-06,
      "loss": 0.0099,
      "num_input_tokens_seen": 19861432,
      "step": 30440
    },
    {
      "epoch": 15.95649895178197,
      "grad_norm": 1.8771045207977295,
      "learning_rate": 5.972929978526872e-06,
      "loss": 0.0471,
      "num_input_tokens_seen": 19865144,
      "step": 30445
    },
    {
      "epoch": 15.959119496855346,
      "grad_norm": 0.1569216400384903,
      "learning_rate": 5.965515065802019e-06,
      "loss": 0.0371,
      "num_input_tokens_seen": 19867928,
      "step": 30450
    },
    {
      "epoch": 15.96174004192872,
      "grad_norm": 0.619284451007843,
      "learning_rate": 5.958104134874254e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 19870456,
      "step": 30455
    },
    {
      "epoch": 15.964360587002096,
      "grad_norm": 0.2001328468322754,
      "learning_rate": 5.950697187293872e-06,
      "loss": 0.0023,
      "num_input_tokens_seen": 19872824,
      "step": 30460
    },
    {
      "epoch": 15.966981132075471,
      "grad_norm": 0.6296656727790833,
      "learning_rate": 5.943294224610305e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 19875256,
      "step": 30465
    },
    {
      "epoch": 15.969601677148846,
      "grad_norm": 0.055828966200351715,
      "learning_rate": 5.935895248372175e-06,
      "loss": 0.0248,
      "num_input_tokens_seen": 19878328,
      "step": 30470
    },
    {
      "epoch": 15.972222222222221,
      "grad_norm": 0.428238183259964,
      "learning_rate": 5.928500260127273e-06,
      "loss": 0.0136,
      "num_input_tokens_seen": 19881784,
      "step": 30475
    },
    {
      "epoch": 15.974842767295598,
      "grad_norm": 1.2003017663955688,
      "learning_rate": 5.921109261422531e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 19885144,
      "step": 30480
    },
    {
      "epoch": 15.977463312368974,
      "grad_norm": 0.4014683663845062,
      "learning_rate": 5.913722253804071e-06,
      "loss": 0.0114,
      "num_input_tokens_seen": 19888120,
      "step": 30485
    },
    {
      "epoch": 15.980083857442349,
      "grad_norm": 0.5032795667648315,
      "learning_rate": 5.906339238817163e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 19891352,
      "step": 30490
    },
    {
      "epoch": 15.982704402515724,
      "grad_norm": 0.9084998965263367,
      "learning_rate": 5.8989602180062554e-06,
      "loss": 0.0215,
      "num_input_tokens_seen": 19894872,
      "step": 30495
    },
    {
      "epoch": 15.985324947589099,
      "grad_norm": 0.17991453409194946,
      "learning_rate": 5.891585192914953e-06,
      "loss": 0.0166,
      "num_input_tokens_seen": 19898904,
      "step": 30500
    },
    {
      "epoch": 15.987945492662474,
      "grad_norm": 0.6352884769439697,
      "learning_rate": 5.884214165086036e-06,
      "loss": 0.012,
      "num_input_tokens_seen": 19902840,
      "step": 30505
    },
    {
      "epoch": 15.99056603773585,
      "grad_norm": 1.3728225231170654,
      "learning_rate": 5.876847136061428e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 19906168,
      "step": 30510
    },
    {
      "epoch": 15.993186582809225,
      "grad_norm": 0.262046217918396,
      "learning_rate": 5.869484107382228e-06,
      "loss": 0.0177,
      "num_input_tokens_seen": 19908920,
      "step": 30515
    },
    {
      "epoch": 15.9958071278826,
      "grad_norm": 1.1174263954162598,
      "learning_rate": 5.862125080588696e-06,
      "loss": 0.0164,
      "num_input_tokens_seen": 19912216,
      "step": 30520
    },
    {
      "epoch": 15.998427672955975,
      "grad_norm": 0.7220062017440796,
      "learning_rate": 5.854770057220263e-06,
      "loss": 0.0104,
      "num_input_tokens_seen": 19914904,
      "step": 30525
    },
    {
      "epoch": 16.0,
      "eval_loss": 1.435498833656311,
      "eval_runtime": 11.6862,
      "eval_samples_per_second": 72.564,
      "eval_steps_per_second": 18.141,
      "num_input_tokens_seen": 19916008,
      "step": 30528
    },
    {
      "epoch": 16.00104821802935,
      "grad_norm": 0.5477792620658875,
      "learning_rate": 5.8474190388155216e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 19917032,
      "step": 30530
    },
    {
      "epoch": 16.003668763102727,
      "grad_norm": 0.5400522947311401,
      "learning_rate": 5.840072026912205e-06,
      "loss": 0.0628,
      "num_input_tokens_seen": 19919976,
      "step": 30535
    },
    {
      "epoch": 16.0062893081761,
      "grad_norm": 0.4817967414855957,
      "learning_rate": 5.83272902304724e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 19922312,
      "step": 30540
    },
    {
      "epoch": 16.008909853249477,
      "grad_norm": 0.5775009989738464,
      "learning_rate": 5.825390028756683e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 19924392,
      "step": 30545
    },
    {
      "epoch": 16.01153039832285,
      "grad_norm": 0.5514383912086487,
      "learning_rate": 5.818055045575776e-06,
      "loss": 0.0136,
      "num_input_tokens_seen": 19927528,
      "step": 30550
    },
    {
      "epoch": 16.014150943396228,
      "grad_norm": 0.5219689011573792,
      "learning_rate": 5.810724075038917e-06,
      "loss": 0.0131,
      "num_input_tokens_seen": 19930920,
      "step": 30555
    },
    {
      "epoch": 16.0167714884696,
      "grad_norm": 2.1276679039001465,
      "learning_rate": 5.803397118679665e-06,
      "loss": 0.0111,
      "num_input_tokens_seen": 19934152,
      "step": 30560
    },
    {
      "epoch": 16.019392033542978,
      "grad_norm": 0.30083537101745605,
      "learning_rate": 5.796074178030727e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 19938216,
      "step": 30565
    },
    {
      "epoch": 16.02201257861635,
      "grad_norm": 0.854387104511261,
      "learning_rate": 5.788755254623976e-06,
      "loss": 0.0126,
      "num_input_tokens_seen": 19942088,
      "step": 30570
    },
    {
      "epoch": 16.02463312368973,
      "grad_norm": 0.6483595371246338,
      "learning_rate": 5.7814403499904525e-06,
      "loss": 0.1253,
      "num_input_tokens_seen": 19944744,
      "step": 30575
    },
    {
      "epoch": 16.0272536687631,
      "grad_norm": 1.3598747253417969,
      "learning_rate": 5.774129465660352e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 19948360,
      "step": 30580
    },
    {
      "epoch": 16.02987421383648,
      "grad_norm": 0.49785542488098145,
      "learning_rate": 5.7668226031630246e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 19951016,
      "step": 30585
    },
    {
      "epoch": 16.032494758909852,
      "grad_norm": 0.9399791359901428,
      "learning_rate": 5.759519764026999e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 19954376,
      "step": 30590
    },
    {
      "epoch": 16.03511530398323,
      "grad_norm": 0.4043161869049072,
      "learning_rate": 5.7522209497799185e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 19957864,
      "step": 30595
    },
    {
      "epoch": 16.037735849056602,
      "grad_norm": 0.19324487447738647,
      "learning_rate": 5.7449261619486246e-06,
      "loss": 0.0139,
      "num_input_tokens_seen": 19961256,
      "step": 30600
    },
    {
      "epoch": 16.04035639412998,
      "grad_norm": 0.2920653522014618,
      "learning_rate": 5.737635402059105e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 19964744,
      "step": 30605
    },
    {
      "epoch": 16.042976939203353,
      "grad_norm": 1.000065803527832,
      "learning_rate": 5.730348671636501e-06,
      "loss": 0.0094,
      "num_input_tokens_seen": 19968232,
      "step": 30610
    },
    {
      "epoch": 16.04559748427673,
      "grad_norm": 0.2156754434108734,
      "learning_rate": 5.723065972205124e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 19971176,
      "step": 30615
    },
    {
      "epoch": 16.048218029350103,
      "grad_norm": 1.3682634830474854,
      "learning_rate": 5.715787305288423e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 19974888,
      "step": 30620
    },
    {
      "epoch": 16.05083857442348,
      "grad_norm": 0.43960076570510864,
      "learning_rate": 5.7085126724090024e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 19977864,
      "step": 30625
    },
    {
      "epoch": 16.053459119496857,
      "grad_norm": 0.5911868214607239,
      "learning_rate": 5.7012420750886445e-06,
      "loss": 0.0108,
      "num_input_tokens_seen": 19981640,
      "step": 30630
    },
    {
      "epoch": 16.05607966457023,
      "grad_norm": 2.3839476108551025,
      "learning_rate": 5.693975514848271e-06,
      "loss": 0.0194,
      "num_input_tokens_seen": 19985640,
      "step": 30635
    },
    {
      "epoch": 16.058700209643607,
      "grad_norm": 0.18420197069644928,
      "learning_rate": 5.686712993207971e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 19988776,
      "step": 30640
    },
    {
      "epoch": 16.06132075471698,
      "grad_norm": 1.2364325523376465,
      "learning_rate": 5.679454511686977e-06,
      "loss": 0.0137,
      "num_input_tokens_seen": 19991976,
      "step": 30645
    },
    {
      "epoch": 16.063941299790358,
      "grad_norm": 0.04868265241384506,
      "learning_rate": 5.672200071803671e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 19995048,
      "step": 30650
    },
    {
      "epoch": 16.06656184486373,
      "grad_norm": 0.28956338763237,
      "learning_rate": 5.664949675075607e-06,
      "loss": 0.0198,
      "num_input_tokens_seen": 19997992,
      "step": 30655
    },
    {
      "epoch": 16.069182389937108,
      "grad_norm": 0.26350051164627075,
      "learning_rate": 5.657703323019486e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 20000520,
      "step": 30660
    },
    {
      "epoch": 16.07180293501048,
      "grad_norm": 0.32135504484176636,
      "learning_rate": 5.65046101715116e-06,
      "loss": 0.0147,
      "num_input_tokens_seen": 20003176,
      "step": 30665
    },
    {
      "epoch": 16.07442348008386,
      "grad_norm": 1.9168561697006226,
      "learning_rate": 5.643222758985644e-06,
      "loss": 0.0181,
      "num_input_tokens_seen": 20006664,
      "step": 30670
    },
    {
      "epoch": 16.07704402515723,
      "grad_norm": 0.5510475635528564,
      "learning_rate": 5.635988550037094e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 20009320,
      "step": 30675
    },
    {
      "epoch": 16.07966457023061,
      "grad_norm": 0.6006308794021606,
      "learning_rate": 5.628758391818817e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 20011656,
      "step": 30680
    },
    {
      "epoch": 16.082285115303982,
      "grad_norm": 0.5061732530593872,
      "learning_rate": 5.621532285843287e-06,
      "loss": 0.0101,
      "num_input_tokens_seen": 20015304,
      "step": 30685
    },
    {
      "epoch": 16.08490566037736,
      "grad_norm": 1.1198434829711914,
      "learning_rate": 5.614310233622119e-06,
      "loss": 0.01,
      "num_input_tokens_seen": 20018504,
      "step": 30690
    },
    {
      "epoch": 16.087526205450732,
      "grad_norm": 1.7454876899719238,
      "learning_rate": 5.607092236666093e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 20021096,
      "step": 30695
    },
    {
      "epoch": 16.09014675052411,
      "grad_norm": 0.20404867827892303,
      "learning_rate": 5.599878296485125e-06,
      "loss": 0.0348,
      "num_input_tokens_seen": 20023912,
      "step": 30700
    },
    {
      "epoch": 16.092767295597483,
      "grad_norm": 0.17273996770381927,
      "learning_rate": 5.592668414588284e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 20026792,
      "step": 30705
    },
    {
      "epoch": 16.09538784067086,
      "grad_norm": 0.2609710991382599,
      "learning_rate": 5.585462592483801e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 20028840,
      "step": 30710
    },
    {
      "epoch": 16.098008385744233,
      "grad_norm": 1.1093231439590454,
      "learning_rate": 5.578260831679047e-06,
      "loss": 0.0251,
      "num_input_tokens_seen": 20032072,
      "step": 30715
    },
    {
      "epoch": 16.10062893081761,
      "grad_norm": 1.0344507694244385,
      "learning_rate": 5.571063133680562e-06,
      "loss": 0.0116,
      "num_input_tokens_seen": 20034824,
      "step": 30720
    },
    {
      "epoch": 16.103249475890987,
      "grad_norm": 0.263896644115448,
      "learning_rate": 5.563869499994004e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 20038792,
      "step": 30725
    },
    {
      "epoch": 16.10587002096436,
      "grad_norm": 0.8825254440307617,
      "learning_rate": 5.556679932124215e-06,
      "loss": 0.0116,
      "num_input_tokens_seen": 20042024,
      "step": 30730
    },
    {
      "epoch": 16.108490566037737,
      "grad_norm": 0.5707991719245911,
      "learning_rate": 5.549494431575159e-06,
      "loss": 0.0096,
      "num_input_tokens_seen": 20045576,
      "step": 30735
    },
    {
      "epoch": 16.11111111111111,
      "grad_norm": 0.7333518862724304,
      "learning_rate": 5.542312999849961e-06,
      "loss": 0.01,
      "num_input_tokens_seen": 20048872,
      "step": 30740
    },
    {
      "epoch": 16.113731656184488,
      "grad_norm": 0.3269088566303253,
      "learning_rate": 5.5351356384509e-06,
      "loss": 0.0158,
      "num_input_tokens_seen": 20053096,
      "step": 30745
    },
    {
      "epoch": 16.11635220125786,
      "grad_norm": 0.3557480275630951,
      "learning_rate": 5.527962348879401e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 20056296,
      "step": 30750
    },
    {
      "epoch": 16.118972746331238,
      "grad_norm": 0.536072313785553,
      "learning_rate": 5.520793132636027e-06,
      "loss": 0.0111,
      "num_input_tokens_seen": 20059368,
      "step": 30755
    },
    {
      "epoch": 16.12159329140461,
      "grad_norm": 0.7253335118293762,
      "learning_rate": 5.513627991220502e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 20062472,
      "step": 30760
    },
    {
      "epoch": 16.12421383647799,
      "grad_norm": 0.3576550781726837,
      "learning_rate": 5.506466926131684e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 20066024,
      "step": 30765
    },
    {
      "epoch": 16.12683438155136,
      "grad_norm": 0.29994577169418335,
      "learning_rate": 5.4993099388675916e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 20068968,
      "step": 30770
    },
    {
      "epoch": 16.12945492662474,
      "grad_norm": 0.6192778944969177,
      "learning_rate": 5.49215703092539e-06,
      "loss": 0.0119,
      "num_input_tokens_seen": 20071720,
      "step": 30775
    },
    {
      "epoch": 16.132075471698112,
      "grad_norm": 1.1009365320205688,
      "learning_rate": 5.4850082038013755e-06,
      "loss": 0.0864,
      "num_input_tokens_seen": 20074984,
      "step": 30780
    },
    {
      "epoch": 16.13469601677149,
      "grad_norm": 0.5644765496253967,
      "learning_rate": 5.4778634589910115e-06,
      "loss": 0.0129,
      "num_input_tokens_seen": 20078760,
      "step": 30785
    },
    {
      "epoch": 16.137316561844862,
      "grad_norm": 1.034560203552246,
      "learning_rate": 5.4707227979888834e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 20081480,
      "step": 30790
    },
    {
      "epoch": 16.13993710691824,
      "grad_norm": 0.3073098659515381,
      "learning_rate": 5.463586222288747e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 20084072,
      "step": 30795
    },
    {
      "epoch": 16.142557651991613,
      "grad_norm": 0.22696088254451752,
      "learning_rate": 5.456453733383493e-06,
      "loss": 0.0346,
      "num_input_tokens_seen": 20088488,
      "step": 30800
    },
    {
      "epoch": 16.14517819706499,
      "grad_norm": 1.5286335945129395,
      "learning_rate": 5.449325332765151e-06,
      "loss": 0.0224,
      "num_input_tokens_seen": 20092008,
      "step": 30805
    },
    {
      "epoch": 16.147798742138363,
      "grad_norm": 0.5985147356987,
      "learning_rate": 5.4422010219249e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 20094664,
      "step": 30810
    },
    {
      "epoch": 16.15041928721174,
      "grad_norm": 0.2535502314567566,
      "learning_rate": 5.435080802353077e-06,
      "loss": 0.0369,
      "num_input_tokens_seen": 20101640,
      "step": 30815
    },
    {
      "epoch": 16.153039832285117,
      "grad_norm": 0.7219299077987671,
      "learning_rate": 5.4279646755391325e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 20104840,
      "step": 30820
    },
    {
      "epoch": 16.15566037735849,
      "grad_norm": 1.3921045064926147,
      "learning_rate": 5.420852642971691e-06,
      "loss": 0.0179,
      "num_input_tokens_seen": 20108456,
      "step": 30825
    },
    {
      "epoch": 16.158280922431867,
      "grad_norm": 0.14962878823280334,
      "learning_rate": 5.41374470613851e-06,
      "loss": 0.0128,
      "num_input_tokens_seen": 20110792,
      "step": 30830
    },
    {
      "epoch": 16.16090146750524,
      "grad_norm": 0.3600738048553467,
      "learning_rate": 5.40664086652648e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 20113256,
      "step": 30835
    },
    {
      "epoch": 16.163522012578618,
      "grad_norm": 1.418216347694397,
      "learning_rate": 5.399541125621646e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 20116424,
      "step": 30840
    },
    {
      "epoch": 16.16614255765199,
      "grad_norm": 0.7959379553794861,
      "learning_rate": 5.392445484909203e-06,
      "loss": 0.0172,
      "num_input_tokens_seen": 20120200,
      "step": 30845
    },
    {
      "epoch": 16.168763102725368,
      "grad_norm": 0.25227805972099304,
      "learning_rate": 5.385353945873464e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 20122664,
      "step": 30850
    },
    {
      "epoch": 16.17138364779874,
      "grad_norm": 6.588363170623779,
      "learning_rate": 5.378266509997909e-06,
      "loss": 0.0107,
      "num_input_tokens_seen": 20125224,
      "step": 30855
    },
    {
      "epoch": 16.17400419287212,
      "grad_norm": 0.7807950973510742,
      "learning_rate": 5.371183178765135e-06,
      "loss": 0.0122,
      "num_input_tokens_seen": 20128968,
      "step": 30860
    },
    {
      "epoch": 16.17662473794549,
      "grad_norm": 0.5847586393356323,
      "learning_rate": 5.364103953656904e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 20131656,
      "step": 30865
    },
    {
      "epoch": 16.17924528301887,
      "grad_norm": 0.36685827374458313,
      "learning_rate": 5.35702883615411e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 20134696,
      "step": 30870
    },
    {
      "epoch": 16.181865828092242,
      "grad_norm": 0.5758365988731384,
      "learning_rate": 5.34995782773679e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 20137192,
      "step": 30875
    },
    {
      "epoch": 16.18448637316562,
      "grad_norm": 0.21131712198257446,
      "learning_rate": 5.3428909298841115e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 20140712,
      "step": 30880
    },
    {
      "epoch": 16.187106918238992,
      "grad_norm": 4.96359395980835,
      "learning_rate": 5.335828144074384e-06,
      "loss": 0.0235,
      "num_input_tokens_seen": 20143496,
      "step": 30885
    },
    {
      "epoch": 16.18972746331237,
      "grad_norm": 2.5074756145477295,
      "learning_rate": 5.3287694717850695e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 20148296,
      "step": 30890
    },
    {
      "epoch": 16.192348008385743,
      "grad_norm": 2.339343786239624,
      "learning_rate": 5.321714914492759e-06,
      "loss": 0.0154,
      "num_input_tokens_seen": 20152168,
      "step": 30895
    },
    {
      "epoch": 16.19496855345912,
      "grad_norm": 0.20593011379241943,
      "learning_rate": 5.314664473673195e-06,
      "loss": 0.0189,
      "num_input_tokens_seen": 20155784,
      "step": 30900
    },
    {
      "epoch": 16.197589098532493,
      "grad_norm": 0.543411910533905,
      "learning_rate": 5.307618150801239e-06,
      "loss": 0.0362,
      "num_input_tokens_seen": 20159112,
      "step": 30905
    },
    {
      "epoch": 16.20020964360587,
      "grad_norm": 1.0383780002593994,
      "learning_rate": 5.300575947350897e-06,
      "loss": 0.0102,
      "num_input_tokens_seen": 20162152,
      "step": 30910
    },
    {
      "epoch": 16.202830188679247,
      "grad_norm": 0.8407219052314758,
      "learning_rate": 5.293537864795325e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 20165032,
      "step": 30915
    },
    {
      "epoch": 16.20545073375262,
      "grad_norm": 4.915651798248291,
      "learning_rate": 5.2865039046068075e-06,
      "loss": 0.0099,
      "num_input_tokens_seen": 20168136,
      "step": 30920
    },
    {
      "epoch": 16.208071278825997,
      "grad_norm": 1.2845054864883423,
      "learning_rate": 5.279474068256767e-06,
      "loss": 0.0076,
      "num_input_tokens_seen": 20171368,
      "step": 30925
    },
    {
      "epoch": 16.21069182389937,
      "grad_norm": 0.02293407917022705,
      "learning_rate": 5.272448357215779e-06,
      "loss": 0.0022,
      "num_input_tokens_seen": 20174664,
      "step": 30930
    },
    {
      "epoch": 16.213312368972748,
      "grad_norm": 0.22139237821102142,
      "learning_rate": 5.26542677295353e-06,
      "loss": 0.0157,
      "num_input_tokens_seen": 20177224,
      "step": 30935
    },
    {
      "epoch": 16.21593291404612,
      "grad_norm": 1.1868330240249634,
      "learning_rate": 5.258409316938848e-06,
      "loss": 0.02,
      "num_input_tokens_seen": 20180200,
      "step": 30940
    },
    {
      "epoch": 16.218553459119498,
      "grad_norm": 0.886307418346405,
      "learning_rate": 5.251395990639713e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 20183592,
      "step": 30945
    },
    {
      "epoch": 16.22117400419287,
      "grad_norm": 2.006237268447876,
      "learning_rate": 5.244386795523235e-06,
      "loss": 0.0116,
      "num_input_tokens_seen": 20186760,
      "step": 30950
    },
    {
      "epoch": 16.22379454926625,
      "grad_norm": 0.44948506355285645,
      "learning_rate": 5.23738173305566e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 20189864,
      "step": 30955
    },
    {
      "epoch": 16.22641509433962,
      "grad_norm": 1.3794103860855103,
      "learning_rate": 5.230380804702362e-06,
      "loss": 0.0174,
      "num_input_tokens_seen": 20193096,
      "step": 30960
    },
    {
      "epoch": 16.229035639413,
      "grad_norm": 0.748615026473999,
      "learning_rate": 5.223384011927851e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 20196136,
      "step": 30965
    },
    {
      "epoch": 16.231656184486372,
      "grad_norm": 0.2594362199306488,
      "learning_rate": 5.216391356195779e-06,
      "loss": 0.0101,
      "num_input_tokens_seen": 20199880,
      "step": 30970
    },
    {
      "epoch": 16.23427672955975,
      "grad_norm": 1.3163120746612549,
      "learning_rate": 5.209402838968935e-06,
      "loss": 0.0118,
      "num_input_tokens_seen": 20203752,
      "step": 30975
    },
    {
      "epoch": 16.236897274633122,
      "grad_norm": 0.11356808245182037,
      "learning_rate": 5.202418461709233e-06,
      "loss": 0.007,
      "num_input_tokens_seen": 20206760,
      "step": 30980
    },
    {
      "epoch": 16.2395178197065,
      "grad_norm": 0.3076604902744293,
      "learning_rate": 5.195438225877738e-06,
      "loss": 0.0129,
      "num_input_tokens_seen": 20209096,
      "step": 30985
    },
    {
      "epoch": 16.242138364779873,
      "grad_norm": 0.8483437299728394,
      "learning_rate": 5.18846213293461e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 20212520,
      "step": 30990
    },
    {
      "epoch": 16.24475890985325,
      "grad_norm": 0.5415068864822388,
      "learning_rate": 5.181490184339185e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 20215720,
      "step": 30995
    },
    {
      "epoch": 16.247379454926623,
      "grad_norm": 0.5661665201187134,
      "learning_rate": 5.17452238154991e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 20219048,
      "step": 31000
    },
    {
      "epoch": 16.25,
      "grad_norm": 2.205169200897217,
      "learning_rate": 5.1675587260243715e-06,
      "loss": 0.0196,
      "num_input_tokens_seen": 20222440,
      "step": 31005
    },
    {
      "epoch": 16.252620545073377,
      "grad_norm": 1.028670310974121,
      "learning_rate": 5.160599219219295e-06,
      "loss": 0.0118,
      "num_input_tokens_seen": 20225224,
      "step": 31010
    },
    {
      "epoch": 16.25524109014675,
      "grad_norm": 4.08309268951416,
      "learning_rate": 5.1536438625905185e-06,
      "loss": 0.0172,
      "num_input_tokens_seen": 20228680,
      "step": 31015
    },
    {
      "epoch": 16.257861635220127,
      "grad_norm": 2.357529401779175,
      "learning_rate": 5.146692657593025e-06,
      "loss": 0.0203,
      "num_input_tokens_seen": 20231176,
      "step": 31020
    },
    {
      "epoch": 16.2604821802935,
      "grad_norm": 0.436156690120697,
      "learning_rate": 5.139745605680926e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 20233672,
      "step": 31025
    },
    {
      "epoch": 16.263102725366878,
      "grad_norm": 0.39846834540367126,
      "learning_rate": 5.132802708307469e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 20236424,
      "step": 31030
    },
    {
      "epoch": 16.26572327044025,
      "grad_norm": 0.48189669847488403,
      "learning_rate": 5.125863966925035e-06,
      "loss": 0.0099,
      "num_input_tokens_seen": 20239336,
      "step": 31035
    },
    {
      "epoch": 16.268343815513628,
      "grad_norm": 2.536848783493042,
      "learning_rate": 5.118929382985116e-06,
      "loss": 0.0136,
      "num_input_tokens_seen": 20242056,
      "step": 31040
    },
    {
      "epoch": 16.270964360587,
      "grad_norm": 0.233001708984375,
      "learning_rate": 5.111998957938363e-06,
      "loss": 0.0076,
      "num_input_tokens_seen": 20245384,
      "step": 31045
    },
    {
      "epoch": 16.27358490566038,
      "grad_norm": 1.0688034296035767,
      "learning_rate": 5.105072693234525e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 20248936,
      "step": 31050
    },
    {
      "epoch": 16.27620545073375,
      "grad_norm": 0.5766226649284363,
      "learning_rate": 5.098150590322506e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 20253416,
      "step": 31055
    },
    {
      "epoch": 16.27882599580713,
      "grad_norm": 0.9543234705924988,
      "learning_rate": 5.091232650650335e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 20256968,
      "step": 31060
    },
    {
      "epoch": 16.281446540880502,
      "grad_norm": 0.670295238494873,
      "learning_rate": 5.0843188756651645e-06,
      "loss": 0.0222,
      "num_input_tokens_seen": 20259752,
      "step": 31065
    },
    {
      "epoch": 16.28406708595388,
      "grad_norm": 0.4377047121524811,
      "learning_rate": 5.077409266813276e-06,
      "loss": 0.0122,
      "num_input_tokens_seen": 20262600,
      "step": 31070
    },
    {
      "epoch": 16.286687631027252,
      "grad_norm": 0.2430819720029831,
      "learning_rate": 5.0705038255400765e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 20265896,
      "step": 31075
    },
    {
      "epoch": 16.28930817610063,
      "grad_norm": 2.5168912410736084,
      "learning_rate": 5.063602553290106e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 20269224,
      "step": 31080
    },
    {
      "epoch": 16.291928721174003,
      "grad_norm": 1.505314588546753,
      "learning_rate": 5.056705451507035e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 20272072,
      "step": 31085
    },
    {
      "epoch": 16.29454926624738,
      "grad_norm": 0.42103642225265503,
      "learning_rate": 5.049812521633668e-06,
      "loss": 0.0172,
      "num_input_tokens_seen": 20275400,
      "step": 31090
    },
    {
      "epoch": 16.297169811320753,
      "grad_norm": 0.7070803642272949,
      "learning_rate": 5.042923765111907e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 20279560,
      "step": 31095
    },
    {
      "epoch": 16.29979035639413,
      "grad_norm": 1.1817501783370972,
      "learning_rate": 5.036039183382821e-06,
      "loss": 0.0198,
      "num_input_tokens_seen": 20282728,
      "step": 31100
    },
    {
      "epoch": 16.302410901467507,
      "grad_norm": 1.9262986183166504,
      "learning_rate": 5.02915877788657e-06,
      "loss": 0.0093,
      "num_input_tokens_seen": 20285224,
      "step": 31105
    },
    {
      "epoch": 16.30503144654088,
      "grad_norm": 0.9886387586593628,
      "learning_rate": 5.0222825500624625e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 20288552,
      "step": 31110
    },
    {
      "epoch": 16.307651991614257,
      "grad_norm": 0.4120577573776245,
      "learning_rate": 5.015410501348935e-06,
      "loss": 0.0259,
      "num_input_tokens_seen": 20292776,
      "step": 31115
    },
    {
      "epoch": 16.31027253668763,
      "grad_norm": 0.30827102065086365,
      "learning_rate": 5.008542633183527e-06,
      "loss": 0.0335,
      "num_input_tokens_seen": 20295208,
      "step": 31120
    },
    {
      "epoch": 16.312893081761008,
      "grad_norm": 0.6136900186538696,
      "learning_rate": 5.001678947002927e-06,
      "loss": 0.0116,
      "num_input_tokens_seen": 20297960,
      "step": 31125
    },
    {
      "epoch": 16.31551362683438,
      "grad_norm": 0.6810111403465271,
      "learning_rate": 4.994819444242943e-06,
      "loss": 0.0104,
      "num_input_tokens_seen": 20300968,
      "step": 31130
    },
    {
      "epoch": 16.318134171907758,
      "grad_norm": 0.7951546907424927,
      "learning_rate": 4.9879641263384956e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 20304648,
      "step": 31135
    },
    {
      "epoch": 16.32075471698113,
      "grad_norm": 2.6693756580352783,
      "learning_rate": 4.981112994723644e-06,
      "loss": 0.0256,
      "num_input_tokens_seen": 20308488,
      "step": 31140
    },
    {
      "epoch": 16.32337526205451,
      "grad_norm": 0.6570689678192139,
      "learning_rate": 4.974266050831572e-06,
      "loss": 0.0091,
      "num_input_tokens_seen": 20311560,
      "step": 31145
    },
    {
      "epoch": 16.32599580712788,
      "grad_norm": 0.3655424416065216,
      "learning_rate": 4.967423296094567e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 20315592,
      "step": 31150
    },
    {
      "epoch": 16.32861635220126,
      "grad_norm": 1.7834395170211792,
      "learning_rate": 4.960584731944076e-06,
      "loss": 0.0103,
      "num_input_tokens_seen": 20319304,
      "step": 31155
    },
    {
      "epoch": 16.331236897274632,
      "grad_norm": 0.6995689868927002,
      "learning_rate": 4.953750359810627e-06,
      "loss": 0.0164,
      "num_input_tokens_seen": 20324264,
      "step": 31160
    },
    {
      "epoch": 16.33385744234801,
      "grad_norm": 0.9216568470001221,
      "learning_rate": 4.946920181123904e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 20326696,
      "step": 31165
    },
    {
      "epoch": 16.336477987421382,
      "grad_norm": 1.00605309009552,
      "learning_rate": 4.940094197312709e-06,
      "loss": 0.0129,
      "num_input_tokens_seen": 20330248,
      "step": 31170
    },
    {
      "epoch": 16.33909853249476,
      "grad_norm": 0.19245880842208862,
      "learning_rate": 4.933272409804946e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 20332552,
      "step": 31175
    },
    {
      "epoch": 16.341719077568133,
      "grad_norm": 0.8432872295379639,
      "learning_rate": 4.926454820027663e-06,
      "loss": 0.007,
      "num_input_tokens_seen": 20335304,
      "step": 31180
    },
    {
      "epoch": 16.34433962264151,
      "grad_norm": 1.211719036102295,
      "learning_rate": 4.919641429407024e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 20337992,
      "step": 31185
    },
    {
      "epoch": 16.346960167714883,
      "grad_norm": 1.6243999004364014,
      "learning_rate": 4.912832239368304e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 20340680,
      "step": 31190
    },
    {
      "epoch": 16.34958071278826,
      "grad_norm": 0.3961341381072998,
      "learning_rate": 4.906027251335918e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 20343848,
      "step": 31195
    },
    {
      "epoch": 16.352201257861637,
      "grad_norm": 0.34568294882774353,
      "learning_rate": 4.899226466733381e-06,
      "loss": 0.0018,
      "num_input_tokens_seen": 20346792,
      "step": 31200
    },
    {
      "epoch": 16.35482180293501,
      "grad_norm": 0.7668083310127258,
      "learning_rate": 4.892429886983346e-06,
      "loss": 0.012,
      "num_input_tokens_seen": 20349544,
      "step": 31205
    },
    {
      "epoch": 16.357442348008387,
      "grad_norm": 0.5268957018852234,
      "learning_rate": 4.88563751350758e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 20352904,
      "step": 31210
    },
    {
      "epoch": 16.36006289308176,
      "grad_norm": 1.3636215925216675,
      "learning_rate": 4.878849347726977e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 20356424,
      "step": 31215
    },
    {
      "epoch": 16.362683438155138,
      "grad_norm": 0.4083479046821594,
      "learning_rate": 4.872065391061526e-06,
      "loss": 0.0089,
      "num_input_tokens_seen": 20360264,
      "step": 31220
    },
    {
      "epoch": 16.36530398322851,
      "grad_norm": 0.7980655431747437,
      "learning_rate": 4.865285644930373e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 20363848,
      "step": 31225
    },
    {
      "epoch": 16.367924528301888,
      "grad_norm": 0.31300339102745056,
      "learning_rate": 4.858510110751746e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 20366280,
      "step": 31230
    },
    {
      "epoch": 16.37054507337526,
      "grad_norm": 0.15330298244953156,
      "learning_rate": 4.851738789943019e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 20370568,
      "step": 31235
    },
    {
      "epoch": 16.37316561844864,
      "grad_norm": 0.14784030616283417,
      "learning_rate": 4.844971683920674e-06,
      "loss": 0.0131,
      "num_input_tokens_seen": 20373928,
      "step": 31240
    },
    {
      "epoch": 16.37578616352201,
      "grad_norm": 0.1191398873925209,
      "learning_rate": 4.838208794100316e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 20376488,
      "step": 31245
    },
    {
      "epoch": 16.37840670859539,
      "grad_norm": 2.5254299640655518,
      "learning_rate": 4.831450121896661e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 20379176,
      "step": 31250
    },
    {
      "epoch": 16.381027253668762,
      "grad_norm": 1.0433505773544312,
      "learning_rate": 4.824695668723542e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 20381736,
      "step": 31255
    },
    {
      "epoch": 16.38364779874214,
      "grad_norm": 0.2984147071838379,
      "learning_rate": 4.817945435993915e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 20385064,
      "step": 31260
    },
    {
      "epoch": 16.386268343815512,
      "grad_norm": 1.3702656030654907,
      "learning_rate": 4.811199425119853e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 20388072,
      "step": 31265
    },
    {
      "epoch": 16.38888888888889,
      "grad_norm": 1.9702976942062378,
      "learning_rate": 4.804457637512552e-06,
      "loss": 0.0109,
      "num_input_tokens_seen": 20391080,
      "step": 31270
    },
    {
      "epoch": 16.391509433962263,
      "grad_norm": 0.44444525241851807,
      "learning_rate": 4.79772007458231e-06,
      "loss": 0.0185,
      "num_input_tokens_seen": 20394024,
      "step": 31275
    },
    {
      "epoch": 16.39412997903564,
      "grad_norm": 0.3661399781703949,
      "learning_rate": 4.7909867377385424e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 20397672,
      "step": 31280
    },
    {
      "epoch": 16.396750524109013,
      "grad_norm": 0.933451771736145,
      "learning_rate": 4.784257628389794e-06,
      "loss": 0.03,
      "num_input_tokens_seen": 20401352,
      "step": 31285
    },
    {
      "epoch": 16.39937106918239,
      "grad_norm": 0.6328639984130859,
      "learning_rate": 4.7775327479437166e-06,
      "loss": 0.037,
      "num_input_tokens_seen": 20404008,
      "step": 31290
    },
    {
      "epoch": 16.401991614255767,
      "grad_norm": 0.41190335154533386,
      "learning_rate": 4.770812097807076e-06,
      "loss": 0.0327,
      "num_input_tokens_seen": 20407496,
      "step": 31295
    },
    {
      "epoch": 16.40461215932914,
      "grad_norm": 0.08901432901620865,
      "learning_rate": 4.764095679385766e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 20410696,
      "step": 31300
    },
    {
      "epoch": 16.407232704402517,
      "grad_norm": 0.0719812884926796,
      "learning_rate": 4.7573834940847745e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 20413736,
      "step": 31305
    },
    {
      "epoch": 16.40985324947589,
      "grad_norm": 0.09337153285741806,
      "learning_rate": 4.750675543308212e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 20417160,
      "step": 31310
    },
    {
      "epoch": 16.412473794549268,
      "grad_norm": 4.135255336761475,
      "learning_rate": 4.743971828459307e-06,
      "loss": 0.0117,
      "num_input_tokens_seen": 20420520,
      "step": 31315
    },
    {
      "epoch": 16.41509433962264,
      "grad_norm": 1.9968879222869873,
      "learning_rate": 4.7372723509404025e-06,
      "loss": 0.0169,
      "num_input_tokens_seen": 20423272,
      "step": 31320
    },
    {
      "epoch": 16.417714884696018,
      "grad_norm": 0.3007817268371582,
      "learning_rate": 4.7305771121529615e-06,
      "loss": 0.0101,
      "num_input_tokens_seen": 20426216,
      "step": 31325
    },
    {
      "epoch": 16.42033542976939,
      "grad_norm": 0.22605346143245697,
      "learning_rate": 4.723886113497542e-06,
      "loss": 0.016,
      "num_input_tokens_seen": 20429512,
      "step": 31330
    },
    {
      "epoch": 16.42295597484277,
      "grad_norm": 2.059013605117798,
      "learning_rate": 4.7171993563738195e-06,
      "loss": 0.025,
      "num_input_tokens_seen": 20433064,
      "step": 31335
    },
    {
      "epoch": 16.42557651991614,
      "grad_norm": 13.597480773925781,
      "learning_rate": 4.7105168421805924e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 20436232,
      "step": 31340
    },
    {
      "epoch": 16.42819706498952,
      "grad_norm": 1.817907452583313,
      "learning_rate": 4.703838572315767e-06,
      "loss": 0.0471,
      "num_input_tokens_seen": 20439688,
      "step": 31345
    },
    {
      "epoch": 16.430817610062892,
      "grad_norm": 3.984135866165161,
      "learning_rate": 4.697164548176367e-06,
      "loss": 0.0787,
      "num_input_tokens_seen": 20442888,
      "step": 31350
    },
    {
      "epoch": 16.43343815513627,
      "grad_norm": 0.49323606491088867,
      "learning_rate": 4.690494771158519e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 20445928,
      "step": 31355
    },
    {
      "epoch": 16.436058700209642,
      "grad_norm": 0.8720424175262451,
      "learning_rate": 4.683829242657453e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 20448552,
      "step": 31360
    },
    {
      "epoch": 16.43867924528302,
      "grad_norm": 0.21740032732486725,
      "learning_rate": 4.677167964067533e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 20451304,
      "step": 31365
    },
    {
      "epoch": 16.441299790356393,
      "grad_norm": 0.7473642230033875,
      "learning_rate": 4.670510936782216e-06,
      "loss": 0.0147,
      "num_input_tokens_seen": 20454408,
      "step": 31370
    },
    {
      "epoch": 16.44392033542977,
      "grad_norm": 0.2068573236465454,
      "learning_rate": 4.6638581621940815e-06,
      "loss": 0.0179,
      "num_input_tokens_seen": 20457384,
      "step": 31375
    },
    {
      "epoch": 16.446540880503143,
      "grad_norm": 0.6172621846199036,
      "learning_rate": 4.657209641694821e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 20460456,
      "step": 31380
    },
    {
      "epoch": 16.44916142557652,
      "grad_norm": 1.7729580402374268,
      "learning_rate": 4.65056537667522e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 20463336,
      "step": 31385
    },
    {
      "epoch": 16.451781970649897,
      "grad_norm": 0.250288188457489,
      "learning_rate": 4.643925368525176e-06,
      "loss": 0.0159,
      "num_input_tokens_seen": 20465928,
      "step": 31390
    },
    {
      "epoch": 16.45440251572327,
      "grad_norm": 0.30496254563331604,
      "learning_rate": 4.637289618633711e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 20468584,
      "step": 31395
    },
    {
      "epoch": 16.457023060796647,
      "grad_norm": 0.5215209126472473,
      "learning_rate": 4.630658128388948e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 20471336,
      "step": 31400
    },
    {
      "epoch": 16.45964360587002,
      "grad_norm": 0.3513084650039673,
      "learning_rate": 4.624030899178125e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 20474696,
      "step": 31405
    },
    {
      "epoch": 16.462264150943398,
      "grad_norm": 0.14585822820663452,
      "learning_rate": 4.617407932387569e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 20477832,
      "step": 31410
    },
    {
      "epoch": 16.46488469601677,
      "grad_norm": 0.2967078983783722,
      "learning_rate": 4.610789229402743e-06,
      "loss": 0.0185,
      "num_input_tokens_seen": 20481160,
      "step": 31415
    },
    {
      "epoch": 16.467505241090148,
      "grad_norm": 0.6581713557243347,
      "learning_rate": 4.604174791608193e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 20484648,
      "step": 31420
    },
    {
      "epoch": 16.47012578616352,
      "grad_norm": 2.082451820373535,
      "learning_rate": 4.597564620387587e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 20486696,
      "step": 31425
    },
    {
      "epoch": 16.4727463312369,
      "grad_norm": 1.1319204568862915,
      "learning_rate": 4.590958717123708e-06,
      "loss": 0.0209,
      "num_input_tokens_seen": 20489640,
      "step": 31430
    },
    {
      "epoch": 16.47536687631027,
      "grad_norm": 0.6411850452423096,
      "learning_rate": 4.5843570831984174e-06,
      "loss": 0.0084,
      "num_input_tokens_seen": 20492392,
      "step": 31435
    },
    {
      "epoch": 16.47798742138365,
      "grad_norm": 0.2075870931148529,
      "learning_rate": 4.577759719992719e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 20494984,
      "step": 31440
    },
    {
      "epoch": 16.480607966457022,
      "grad_norm": 0.2979000508785248,
      "learning_rate": 4.571166628886692e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 20497832,
      "step": 31445
    },
    {
      "epoch": 16.4832285115304,
      "grad_norm": 1.033849835395813,
      "learning_rate": 4.564577811259541e-06,
      "loss": 0.0099,
      "num_input_tokens_seen": 20500488,
      "step": 31450
    },
    {
      "epoch": 16.485849056603772,
      "grad_norm": 0.4460292458534241,
      "learning_rate": 4.557993268489574e-06,
      "loss": 0.0221,
      "num_input_tokens_seen": 20503688,
      "step": 31455
    },
    {
      "epoch": 16.48846960167715,
      "grad_norm": 1.0316718816757202,
      "learning_rate": 4.551413001954205e-06,
      "loss": 0.0161,
      "num_input_tokens_seen": 20507816,
      "step": 31460
    },
    {
      "epoch": 16.491090146750523,
      "grad_norm": 4.083456516265869,
      "learning_rate": 4.54483701302994e-06,
      "loss": 0.0288,
      "num_input_tokens_seen": 20510920,
      "step": 31465
    },
    {
      "epoch": 16.4937106918239,
      "grad_norm": 0.7330138683319092,
      "learning_rate": 4.538265303092415e-06,
      "loss": 0.0107,
      "num_input_tokens_seen": 20514408,
      "step": 31470
    },
    {
      "epoch": 16.496331236897273,
      "grad_norm": 2.418144464492798,
      "learning_rate": 4.531697873516338e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 20518472,
      "step": 31475
    },
    {
      "epoch": 16.49895178197065,
      "grad_norm": 0.9588221311569214,
      "learning_rate": 4.525134725675556e-06,
      "loss": 0.0201,
      "num_input_tokens_seen": 20521064,
      "step": 31480
    },
    {
      "epoch": 16.501572327044027,
      "grad_norm": 3.58870530128479,
      "learning_rate": 4.518575860943003e-06,
      "loss": 0.0099,
      "num_input_tokens_seen": 20523784,
      "step": 31485
    },
    {
      "epoch": 16.5041928721174,
      "grad_norm": 1.2817691564559937,
      "learning_rate": 4.512021280690712e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 20527976,
      "step": 31490
    },
    {
      "epoch": 16.506813417190777,
      "grad_norm": 0.41619089245796204,
      "learning_rate": 4.505470986289828e-06,
      "loss": 0.0244,
      "num_input_tokens_seen": 20531784,
      "step": 31495
    },
    {
      "epoch": 16.50943396226415,
      "grad_norm": 0.43829238414764404,
      "learning_rate": 4.498924979110605e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 20535048,
      "step": 31500
    },
    {
      "epoch": 16.512054507337528,
      "grad_norm": 2.256054401397705,
      "learning_rate": 4.492383260522384e-06,
      "loss": 0.025,
      "num_input_tokens_seen": 20537672,
      "step": 31505
    },
    {
      "epoch": 16.5146750524109,
      "grad_norm": 0.7338768839836121,
      "learning_rate": 4.485845831893626e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 20540264,
      "step": 31510
    },
    {
      "epoch": 16.517295597484278,
      "grad_norm": 1.1316536664962769,
      "learning_rate": 4.479312694591878e-06,
      "loss": 0.0288,
      "num_input_tokens_seen": 20543208,
      "step": 31515
    },
    {
      "epoch": 16.51991614255765,
      "grad_norm": 0.5796536207199097,
      "learning_rate": 4.4727838499838e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 20546632,
      "step": 31520
    },
    {
      "epoch": 16.52253668763103,
      "grad_norm": 1.655537724494934,
      "learning_rate": 4.4662592994351545e-06,
      "loss": 0.0133,
      "num_input_tokens_seen": 20549256,
      "step": 31525
    },
    {
      "epoch": 16.5251572327044,
      "grad_norm": 0.3814236521720886,
      "learning_rate": 4.4597390443108065e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 20551880,
      "step": 31530
    },
    {
      "epoch": 16.52777777777778,
      "grad_norm": 2.196300506591797,
      "learning_rate": 4.453223085974712e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 20555368,
      "step": 31535
    },
    {
      "epoch": 16.530398322851152,
      "grad_norm": 0.34657829999923706,
      "learning_rate": 4.4467114257899425e-06,
      "loss": 0.0021,
      "num_input_tokens_seen": 20557864,
      "step": 31540
    },
    {
      "epoch": 16.53301886792453,
      "grad_norm": 1.0744200944900513,
      "learning_rate": 4.44020406511865e-06,
      "loss": 0.0079,
      "num_input_tokens_seen": 20561800,
      "step": 31545
    },
    {
      "epoch": 16.535639412997902,
      "grad_norm": 0.37403514981269836,
      "learning_rate": 4.43370100532211e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 20565128,
      "step": 31550
    },
    {
      "epoch": 16.53825995807128,
      "grad_norm": 0.8701136708259583,
      "learning_rate": 4.427202247760695e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 20568648,
      "step": 31555
    },
    {
      "epoch": 16.540880503144653,
      "grad_norm": 0.6420125961303711,
      "learning_rate": 4.420707793793854e-06,
      "loss": 0.01,
      "num_input_tokens_seen": 20571816,
      "step": 31560
    },
    {
      "epoch": 16.54350104821803,
      "grad_norm": 0.5315579771995544,
      "learning_rate": 4.414217644780172e-06,
      "loss": 0.019,
      "num_input_tokens_seen": 20576136,
      "step": 31565
    },
    {
      "epoch": 16.546121593291403,
      "grad_norm": 0.20076225697994232,
      "learning_rate": 4.407731802077294e-06,
      "loss": 0.0136,
      "num_input_tokens_seen": 20578472,
      "step": 31570
    },
    {
      "epoch": 16.54874213836478,
      "grad_norm": 0.5573241710662842,
      "learning_rate": 4.401250267041996e-06,
      "loss": 0.0237,
      "num_input_tokens_seen": 20581800,
      "step": 31575
    },
    {
      "epoch": 16.551362683438157,
      "grad_norm": 0.8747309446334839,
      "learning_rate": 4.394773041030137e-06,
      "loss": 0.0136,
      "num_input_tokens_seen": 20584296,
      "step": 31580
    },
    {
      "epoch": 16.55398322851153,
      "grad_norm": 1.2434709072113037,
      "learning_rate": 4.388300125396685e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 20587368,
      "step": 31585
    },
    {
      "epoch": 16.556603773584907,
      "grad_norm": 0.7782166004180908,
      "learning_rate": 4.381831521495699e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 20590504,
      "step": 31590
    },
    {
      "epoch": 16.55922431865828,
      "grad_norm": 0.7778090238571167,
      "learning_rate": 4.37536723068033e-06,
      "loss": 0.0121,
      "num_input_tokens_seen": 20593448,
      "step": 31595
    },
    {
      "epoch": 16.561844863731658,
      "grad_norm": 0.5676372051239014,
      "learning_rate": 4.368907254302837e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 20596904,
      "step": 31600
    },
    {
      "epoch": 16.56446540880503,
      "grad_norm": 0.4990226626396179,
      "learning_rate": 4.3624515937145725e-06,
      "loss": 0.0095,
      "num_input_tokens_seen": 20600040,
      "step": 31605
    },
    {
      "epoch": 16.567085953878408,
      "grad_norm": 0.43288546800613403,
      "learning_rate": 4.3560002502659905e-06,
      "loss": 0.0103,
      "num_input_tokens_seen": 20603112,
      "step": 31610
    },
    {
      "epoch": 16.56970649895178,
      "grad_norm": 0.6901464462280273,
      "learning_rate": 4.349553225306644e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 20606760,
      "step": 31615
    },
    {
      "epoch": 16.572327044025158,
      "grad_norm": 0.7277502417564392,
      "learning_rate": 4.343110520185167e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 20609896,
      "step": 31620
    },
    {
      "epoch": 16.57494758909853,
      "grad_norm": 0.7263473272323608,
      "learning_rate": 4.336672136249298e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 20612712,
      "step": 31625
    },
    {
      "epoch": 16.57756813417191,
      "grad_norm": 1.807384729385376,
      "learning_rate": 4.33023807484588e-06,
      "loss": 0.0545,
      "num_input_tokens_seen": 20616488,
      "step": 31630
    },
    {
      "epoch": 16.580188679245282,
      "grad_norm": 2.4772868156433105,
      "learning_rate": 4.323808337320842e-06,
      "loss": 0.0564,
      "num_input_tokens_seen": 20619944,
      "step": 31635
    },
    {
      "epoch": 16.58280922431866,
      "grad_norm": 0.5477007031440735,
      "learning_rate": 4.31738292501922e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 20622952,
      "step": 31640
    },
    {
      "epoch": 16.585429769392032,
      "grad_norm": 1.8910479545593262,
      "learning_rate": 4.31096183928513e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 20626248,
      "step": 31645
    },
    {
      "epoch": 16.58805031446541,
      "grad_norm": 1.283941388130188,
      "learning_rate": 4.304545081461783e-06,
      "loss": 0.0094,
      "num_input_tokens_seen": 20630088,
      "step": 31650
    },
    {
      "epoch": 16.590670859538783,
      "grad_norm": 0.8707976341247559,
      "learning_rate": 4.298132652891501e-06,
      "loss": 0.035,
      "num_input_tokens_seen": 20632840,
      "step": 31655
    },
    {
      "epoch": 16.59329140461216,
      "grad_norm": 4.0288190841674805,
      "learning_rate": 4.291724554915688e-06,
      "loss": 0.0142,
      "num_input_tokens_seen": 20635720,
      "step": 31660
    },
    {
      "epoch": 16.595911949685533,
      "grad_norm": 0.19630827009677887,
      "learning_rate": 4.285320788874853e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 20639400,
      "step": 31665
    },
    {
      "epoch": 16.59853249475891,
      "grad_norm": 0.753302812576294,
      "learning_rate": 4.278921356108581e-06,
      "loss": 0.0115,
      "num_input_tokens_seen": 20643656,
      "step": 31670
    },
    {
      "epoch": 16.601153039832283,
      "grad_norm": 0.37127864360809326,
      "learning_rate": 4.272526257955556e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 20646504,
      "step": 31675
    },
    {
      "epoch": 16.60377358490566,
      "grad_norm": 3.2474396228790283,
      "learning_rate": 4.266135495753567e-06,
      "loss": 0.0127,
      "num_input_tokens_seen": 20650120,
      "step": 31680
    },
    {
      "epoch": 16.606394129979037,
      "grad_norm": 0.709563136100769,
      "learning_rate": 4.259749070839491e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 20653064,
      "step": 31685
    },
    {
      "epoch": 16.60901467505241,
      "grad_norm": 0.5835222601890564,
      "learning_rate": 4.253366984549289e-06,
      "loss": 0.0264,
      "num_input_tokens_seen": 20657160,
      "step": 31690
    },
    {
      "epoch": 16.611635220125788,
      "grad_norm": 0.17140313982963562,
      "learning_rate": 4.24698923821803e-06,
      "loss": 0.0108,
      "num_input_tokens_seen": 20659912,
      "step": 31695
    },
    {
      "epoch": 16.61425576519916,
      "grad_norm": 0.4987437427043915,
      "learning_rate": 4.240615833179861e-06,
      "loss": 0.0142,
      "num_input_tokens_seen": 20662312,
      "step": 31700
    },
    {
      "epoch": 16.616876310272538,
      "grad_norm": 0.9433786869049072,
      "learning_rate": 4.234246770768016e-06,
      "loss": 0.0272,
      "num_input_tokens_seen": 20666248,
      "step": 31705
    },
    {
      "epoch": 16.61949685534591,
      "grad_norm": 0.4341430068016052,
      "learning_rate": 4.227882052314841e-06,
      "loss": 0.0127,
      "num_input_tokens_seen": 20670888,
      "step": 31710
    },
    {
      "epoch": 16.622117400419288,
      "grad_norm": 0.5642845630645752,
      "learning_rate": 4.2215216791517575e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 20673320,
      "step": 31715
    },
    {
      "epoch": 16.62473794549266,
      "grad_norm": 0.41952118277549744,
      "learning_rate": 4.215165652609293e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 20676552,
      "step": 31720
    },
    {
      "epoch": 16.62735849056604,
      "grad_norm": 0.5346153974533081,
      "learning_rate": 4.208813974017048e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 20678792,
      "step": 31725
    },
    {
      "epoch": 16.629979035639412,
      "grad_norm": 0.21000336110591888,
      "learning_rate": 4.202466644703712e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 20683240,
      "step": 31730
    },
    {
      "epoch": 16.63259958071279,
      "grad_norm": 1.2346223592758179,
      "learning_rate": 4.196123665997087e-06,
      "loss": 0.0284,
      "num_input_tokens_seen": 20686536,
      "step": 31735
    },
    {
      "epoch": 16.635220125786162,
      "grad_norm": 0.39739394187927246,
      "learning_rate": 4.189785039224045e-06,
      "loss": 0.0112,
      "num_input_tokens_seen": 20689832,
      "step": 31740
    },
    {
      "epoch": 16.63784067085954,
      "grad_norm": 0.3190076947212219,
      "learning_rate": 4.183450765710567e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 20692648,
      "step": 31745
    },
    {
      "epoch": 16.640461215932913,
      "grad_norm": 0.2589654326438904,
      "learning_rate": 4.177120846781693e-06,
      "loss": 0.0196,
      "num_input_tokens_seen": 20695048,
      "step": 31750
    },
    {
      "epoch": 16.64308176100629,
      "grad_norm": 0.35218340158462524,
      "learning_rate": 4.170795283761583e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 20698216,
      "step": 31755
    },
    {
      "epoch": 16.645702306079663,
      "grad_norm": 0.690047025680542,
      "learning_rate": 4.164474077973462e-06,
      "loss": 0.0444,
      "num_input_tokens_seen": 20702024,
      "step": 31760
    },
    {
      "epoch": 16.64832285115304,
      "grad_norm": 0.4971773624420166,
      "learning_rate": 4.158157230739659e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 20705032,
      "step": 31765
    },
    {
      "epoch": 16.650943396226417,
      "grad_norm": 0.26167967915534973,
      "learning_rate": 4.151844743381589e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 20708072,
      "step": 31770
    },
    {
      "epoch": 16.65356394129979,
      "grad_norm": 0.4503209590911865,
      "learning_rate": 4.145536617219756e-06,
      "loss": 0.0091,
      "num_input_tokens_seen": 20711432,
      "step": 31775
    },
    {
      "epoch": 16.656184486373167,
      "grad_norm": 0.2802673876285553,
      "learning_rate": 4.139232853573735e-06,
      "loss": 0.0177,
      "num_input_tokens_seen": 20715112,
      "step": 31780
    },
    {
      "epoch": 16.65880503144654,
      "grad_norm": 0.5542395710945129,
      "learning_rate": 4.132933453762214e-06,
      "loss": 0.0093,
      "num_input_tokens_seen": 20718024,
      "step": 31785
    },
    {
      "epoch": 16.661425576519918,
      "grad_norm": 0.9590911865234375,
      "learning_rate": 4.126638419102946e-06,
      "loss": 0.0094,
      "num_input_tokens_seen": 20721192,
      "step": 31790
    },
    {
      "epoch": 16.66404612159329,
      "grad_norm": 1.598754644393921,
      "learning_rate": 4.120347750912784e-06,
      "loss": 0.0097,
      "num_input_tokens_seen": 20723656,
      "step": 31795
    },
    {
      "epoch": 16.666666666666668,
      "grad_norm": 0.43592071533203125,
      "learning_rate": 4.1140614505076724e-06,
      "loss": 0.0149,
      "num_input_tokens_seen": 20726088,
      "step": 31800
    },
    {
      "epoch": 16.66928721174004,
      "grad_norm": 0.36974504590034485,
      "learning_rate": 4.10777951920262e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 20728840,
      "step": 31805
    },
    {
      "epoch": 16.671907756813418,
      "grad_norm": 0.21766556799411774,
      "learning_rate": 4.1015019583117455e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 20732424,
      "step": 31810
    },
    {
      "epoch": 16.67452830188679,
      "grad_norm": 0.34313350915908813,
      "learning_rate": 4.095228769148232e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 20738184,
      "step": 31815
    },
    {
      "epoch": 16.67714884696017,
      "grad_norm": 0.42678338289260864,
      "learning_rate": 4.088959953024368e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 20743464,
      "step": 31820
    },
    {
      "epoch": 16.679769392033542,
      "grad_norm": 0.9138143658638,
      "learning_rate": 4.082695511251522e-06,
      "loss": 0.0081,
      "num_input_tokens_seen": 20746536,
      "step": 31825
    },
    {
      "epoch": 16.68238993710692,
      "grad_norm": 2.0379531383514404,
      "learning_rate": 4.076435445140131e-06,
      "loss": 0.0199,
      "num_input_tokens_seen": 20749896,
      "step": 31830
    },
    {
      "epoch": 16.685010482180292,
      "grad_norm": 2.095736503601074,
      "learning_rate": 4.070179755999737e-06,
      "loss": 0.0162,
      "num_input_tokens_seen": 20753448,
      "step": 31835
    },
    {
      "epoch": 16.68763102725367,
      "grad_norm": 2.3444128036499023,
      "learning_rate": 4.063928445138962e-06,
      "loss": 0.0158,
      "num_input_tokens_seen": 20757032,
      "step": 31840
    },
    {
      "epoch": 16.690251572327043,
      "grad_norm": 4.196317672729492,
      "learning_rate": 4.057681513865499e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 20760136,
      "step": 31845
    },
    {
      "epoch": 16.69287211740042,
      "grad_norm": 1.3781400918960571,
      "learning_rate": 4.0514389634861426e-06,
      "loss": 0.0256,
      "num_input_tokens_seen": 20763144,
      "step": 31850
    },
    {
      "epoch": 16.695492662473793,
      "grad_norm": 0.1714288294315338,
      "learning_rate": 4.045200795306767e-06,
      "loss": 0.0103,
      "num_input_tokens_seen": 20766248,
      "step": 31855
    },
    {
      "epoch": 16.69811320754717,
      "grad_norm": 5.065093994140625,
      "learning_rate": 4.038967010632314e-06,
      "loss": 0.0144,
      "num_input_tokens_seen": 20769160,
      "step": 31860
    },
    {
      "epoch": 16.700733752620543,
      "grad_norm": 0.5738794803619385,
      "learning_rate": 4.032737610766826e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 20773384,
      "step": 31865
    },
    {
      "epoch": 16.70335429769392,
      "grad_norm": 0.7153509259223938,
      "learning_rate": 4.026512597013429e-06,
      "loss": 0.0158,
      "num_input_tokens_seen": 20777416,
      "step": 31870
    },
    {
      "epoch": 16.705974842767297,
      "grad_norm": 3.7541563510894775,
      "learning_rate": 4.020291970674312e-06,
      "loss": 0.0328,
      "num_input_tokens_seen": 20780808,
      "step": 31875
    },
    {
      "epoch": 16.70859538784067,
      "grad_norm": 0.2020443081855774,
      "learning_rate": 4.0140757330507705e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 20784040,
      "step": 31880
    },
    {
      "epoch": 16.711215932914047,
      "grad_norm": 0.721714437007904,
      "learning_rate": 4.007863885443161e-06,
      "loss": 0.0081,
      "num_input_tokens_seen": 20787144,
      "step": 31885
    },
    {
      "epoch": 16.71383647798742,
      "grad_norm": 0.7526418566703796,
      "learning_rate": 4.001656429150935e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 20790408,
      "step": 31890
    },
    {
      "epoch": 16.716457023060798,
      "grad_norm": 0.21520042419433594,
      "learning_rate": 3.9954533654726235e-06,
      "loss": 0.0678,
      "num_input_tokens_seen": 20793768,
      "step": 31895
    },
    {
      "epoch": 16.71907756813417,
      "grad_norm": 0.24763146042823792,
      "learning_rate": 3.989254695705841e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 20799624,
      "step": 31900
    },
    {
      "epoch": 16.721698113207548,
      "grad_norm": 1.2592394351959229,
      "learning_rate": 3.983060421147272e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 20802152,
      "step": 31905
    },
    {
      "epoch": 16.72431865828092,
      "grad_norm": 1.343449354171753,
      "learning_rate": 3.976870543092684e-06,
      "loss": 0.0104,
      "num_input_tokens_seen": 20804776,
      "step": 31910
    },
    {
      "epoch": 16.7269392033543,
      "grad_norm": 1.0522931814193726,
      "learning_rate": 3.970685062836932e-06,
      "loss": 0.015,
      "num_input_tokens_seen": 20807880,
      "step": 31915
    },
    {
      "epoch": 16.729559748427672,
      "grad_norm": 1.0726780891418457,
      "learning_rate": 3.96450398167395e-06,
      "loss": 0.0285,
      "num_input_tokens_seen": 20810568,
      "step": 31920
    },
    {
      "epoch": 16.73218029350105,
      "grad_norm": 0.17994509637355804,
      "learning_rate": 3.958327300896758e-06,
      "loss": 0.0162,
      "num_input_tokens_seen": 20815368,
      "step": 31925
    },
    {
      "epoch": 16.734800838574422,
      "grad_norm": 0.44691038131713867,
      "learning_rate": 3.952155021797429e-06,
      "loss": 0.0113,
      "num_input_tokens_seen": 20818472,
      "step": 31930
    },
    {
      "epoch": 16.7374213836478,
      "grad_norm": 0.21836401522159576,
      "learning_rate": 3.945987145667151e-06,
      "loss": 0.0102,
      "num_input_tokens_seen": 20822120,
      "step": 31935
    },
    {
      "epoch": 16.740041928721173,
      "grad_norm": 0.5884442329406738,
      "learning_rate": 3.939823673796162e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 20825160,
      "step": 31940
    },
    {
      "epoch": 16.74266247379455,
      "grad_norm": 1.4185588359832764,
      "learning_rate": 3.933664607473789e-06,
      "loss": 0.0179,
      "num_input_tokens_seen": 20830056,
      "step": 31945
    },
    {
      "epoch": 16.745283018867923,
      "grad_norm": 1.1929017305374146,
      "learning_rate": 3.927509947988445e-06,
      "loss": 0.0239,
      "num_input_tokens_seen": 20833032,
      "step": 31950
    },
    {
      "epoch": 16.7479035639413,
      "grad_norm": 0.29756221175193787,
      "learning_rate": 3.9213596966276164e-06,
      "loss": 0.0222,
      "num_input_tokens_seen": 20836200,
      "step": 31955
    },
    {
      "epoch": 16.750524109014677,
      "grad_norm": 0.8175833821296692,
      "learning_rate": 3.9152138546778625e-06,
      "loss": 0.0107,
      "num_input_tokens_seen": 20839784,
      "step": 31960
    },
    {
      "epoch": 16.75314465408805,
      "grad_norm": 3.4147069454193115,
      "learning_rate": 3.9090724234248175e-06,
      "loss": 0.0063,
      "num_input_tokens_seen": 20842952,
      "step": 31965
    },
    {
      "epoch": 16.755765199161427,
      "grad_norm": 0.7607231736183167,
      "learning_rate": 3.902935404153202e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 20846344,
      "step": 31970
    },
    {
      "epoch": 16.7583857442348,
      "grad_norm": 0.4074370861053467,
      "learning_rate": 3.896802798146809e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 20849512,
      "step": 31975
    },
    {
      "epoch": 16.761006289308177,
      "grad_norm": 1.8688098192214966,
      "learning_rate": 3.890674606688524e-06,
      "loss": 0.0087,
      "num_input_tokens_seen": 20852744,
      "step": 31980
    },
    {
      "epoch": 16.76362683438155,
      "grad_norm": 1.331151008605957,
      "learning_rate": 3.88455083106028e-06,
      "loss": 0.0238,
      "num_input_tokens_seen": 20856392,
      "step": 31985
    },
    {
      "epoch": 16.766247379454928,
      "grad_norm": 1.8472710847854614,
      "learning_rate": 3.878431472543098e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 20859048,
      "step": 31990
    },
    {
      "epoch": 16.7688679245283,
      "grad_norm": 0.1616421341896057,
      "learning_rate": 3.872316532417086e-06,
      "loss": 0.0099,
      "num_input_tokens_seen": 20861992,
      "step": 31995
    },
    {
      "epoch": 16.771488469601678,
      "grad_norm": 0.12520760297775269,
      "learning_rate": 3.866206011961418e-06,
      "loss": 0.01,
      "num_input_tokens_seen": 20864904,
      "step": 32000
    },
    {
      "epoch": 16.77410901467505,
      "grad_norm": 6.491955757141113,
      "learning_rate": 3.860099912454346e-06,
      "loss": 0.0105,
      "num_input_tokens_seen": 20867976,
      "step": 32005
    },
    {
      "epoch": 16.77672955974843,
      "grad_norm": 5.198907852172852,
      "learning_rate": 3.853998235173201e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 20871464,
      "step": 32010
    },
    {
      "epoch": 16.779350104821802,
      "grad_norm": 0.5573555827140808,
      "learning_rate": 3.847900981394379e-06,
      "loss": 0.0086,
      "num_input_tokens_seen": 20874984,
      "step": 32015
    },
    {
      "epoch": 16.78197064989518,
      "grad_norm": 0.8550716638565063,
      "learning_rate": 3.841808152393348e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 20877832,
      "step": 32020
    },
    {
      "epoch": 16.784591194968552,
      "grad_norm": 0.9855374693870544,
      "learning_rate": 3.835719749444666e-06,
      "loss": 0.0323,
      "num_input_tokens_seen": 20881096,
      "step": 32025
    },
    {
      "epoch": 16.78721174004193,
      "grad_norm": 0.7573882937431335,
      "learning_rate": 3.8296357738219606e-06,
      "loss": 0.0181,
      "num_input_tokens_seen": 20884680,
      "step": 32030
    },
    {
      "epoch": 16.789832285115303,
      "grad_norm": 1.1191214323043823,
      "learning_rate": 3.823556226797931e-06,
      "loss": 0.0503,
      "num_input_tokens_seen": 20888008,
      "step": 32035
    },
    {
      "epoch": 16.79245283018868,
      "grad_norm": 0.5954616665840149,
      "learning_rate": 3.817481109644344e-06,
      "loss": 0.0108,
      "num_input_tokens_seen": 20891240,
      "step": 32040
    },
    {
      "epoch": 16.795073375262053,
      "grad_norm": 0.5275477766990662,
      "learning_rate": 3.811410423632042e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 20894664,
      "step": 32045
    },
    {
      "epoch": 16.79769392033543,
      "grad_norm": 0.27433475852012634,
      "learning_rate": 3.8053441700309473e-06,
      "loss": 0.0116,
      "num_input_tokens_seen": 20897608,
      "step": 32050
    },
    {
      "epoch": 16.800314465408803,
      "grad_norm": 0.5194885730743408,
      "learning_rate": 3.7992823501100507e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 20901096,
      "step": 32055
    },
    {
      "epoch": 16.80293501048218,
      "grad_norm": 2.521355390548706,
      "learning_rate": 3.7932249651374207e-06,
      "loss": 0.0115,
      "num_input_tokens_seen": 20904520,
      "step": 32060
    },
    {
      "epoch": 16.805555555555557,
      "grad_norm": 2.999119758605957,
      "learning_rate": 3.787172016380183e-06,
      "loss": 0.0092,
      "num_input_tokens_seen": 20907592,
      "step": 32065
    },
    {
      "epoch": 16.80817610062893,
      "grad_norm": 1.283959150314331,
      "learning_rate": 3.7811235051045558e-06,
      "loss": 0.0133,
      "num_input_tokens_seen": 20911752,
      "step": 32070
    },
    {
      "epoch": 16.810796645702307,
      "grad_norm": 0.6939139366149902,
      "learning_rate": 3.775079432575809e-06,
      "loss": 0.0123,
      "num_input_tokens_seen": 20914728,
      "step": 32075
    },
    {
      "epoch": 16.81341719077568,
      "grad_norm": 1.5652790069580078,
      "learning_rate": 3.7690398000582964e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 20917704,
      "step": 32080
    },
    {
      "epoch": 16.816037735849058,
      "grad_norm": 1.7636741399765015,
      "learning_rate": 3.763004608815443e-06,
      "loss": 0.0119,
      "num_input_tokens_seen": 20920808,
      "step": 32085
    },
    {
      "epoch": 16.81865828092243,
      "grad_norm": 9.38960075378418,
      "learning_rate": 3.756973860109747e-06,
      "loss": 0.0092,
      "num_input_tokens_seen": 20923528,
      "step": 32090
    },
    {
      "epoch": 16.821278825995808,
      "grad_norm": 0.7498302459716797,
      "learning_rate": 3.7509475552027663e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 20926376,
      "step": 32095
    },
    {
      "epoch": 16.82389937106918,
      "grad_norm": 0.39610448479652405,
      "learning_rate": 3.744925695355131e-06,
      "loss": 0.0093,
      "num_input_tokens_seen": 20929672,
      "step": 32100
    },
    {
      "epoch": 16.82651991614256,
      "grad_norm": 0.16451583802700043,
      "learning_rate": 3.7389082818265476e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 20932616,
      "step": 32105
    },
    {
      "epoch": 16.829140461215932,
      "grad_norm": 0.12307649105787277,
      "learning_rate": 3.732895315875792e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 20935400,
      "step": 32110
    },
    {
      "epoch": 16.83176100628931,
      "grad_norm": 0.2478790134191513,
      "learning_rate": 3.726886798760715e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 20938824,
      "step": 32115
    },
    {
      "epoch": 16.834381551362682,
      "grad_norm": 3.7604637145996094,
      "learning_rate": 3.720882731738215e-06,
      "loss": 0.0863,
      "num_input_tokens_seen": 20941960,
      "step": 32120
    },
    {
      "epoch": 16.83700209643606,
      "grad_norm": 0.7341609597206116,
      "learning_rate": 3.714883116064291e-06,
      "loss": 0.0303,
      "num_input_tokens_seen": 20948712,
      "step": 32125
    },
    {
      "epoch": 16.839622641509433,
      "grad_norm": 0.26157593727111816,
      "learning_rate": 3.708887952993978e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 20951592,
      "step": 32130
    },
    {
      "epoch": 16.84224318658281,
      "grad_norm": 0.30742353200912476,
      "learning_rate": 3.7028972437814057e-06,
      "loss": 0.0315,
      "num_input_tokens_seen": 20955880,
      "step": 32135
    },
    {
      "epoch": 16.844863731656183,
      "grad_norm": 1.6140036582946777,
      "learning_rate": 3.696910989679764e-06,
      "loss": 0.0101,
      "num_input_tokens_seen": 20959112,
      "step": 32140
    },
    {
      "epoch": 16.84748427672956,
      "grad_norm": 0.5055554509162903,
      "learning_rate": 3.6909291919413003e-06,
      "loss": 0.0022,
      "num_input_tokens_seen": 20961992,
      "step": 32145
    },
    {
      "epoch": 16.850104821802937,
      "grad_norm": 1.5692137479782104,
      "learning_rate": 3.684951851817342e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 20964776,
      "step": 32150
    },
    {
      "epoch": 16.85272536687631,
      "grad_norm": 0.31262561678886414,
      "learning_rate": 3.678978970558286e-06,
      "loss": 0.0063,
      "num_input_tokens_seen": 20967944,
      "step": 32155
    },
    {
      "epoch": 16.855345911949687,
      "grad_norm": 0.9211285710334778,
      "learning_rate": 3.673010549413583e-06,
      "loss": 0.0162,
      "num_input_tokens_seen": 20971976,
      "step": 32160
    },
    {
      "epoch": 16.85796645702306,
      "grad_norm": 1.0867059230804443,
      "learning_rate": 3.6670465896317592e-06,
      "loss": 0.015,
      "num_input_tokens_seen": 20975208,
      "step": 32165
    },
    {
      "epoch": 16.860587002096437,
      "grad_norm": 2.1209769248962402,
      "learning_rate": 3.661087092460419e-06,
      "loss": 0.0101,
      "num_input_tokens_seen": 20977736,
      "step": 32170
    },
    {
      "epoch": 16.86320754716981,
      "grad_norm": 1.5633426904678345,
      "learning_rate": 3.655132059146202e-06,
      "loss": 0.0232,
      "num_input_tokens_seen": 20981704,
      "step": 32175
    },
    {
      "epoch": 16.865828092243188,
      "grad_norm": 3.7573184967041016,
      "learning_rate": 3.6491814909348526e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 20985992,
      "step": 32180
    },
    {
      "epoch": 16.86844863731656,
      "grad_norm": 0.554051399230957,
      "learning_rate": 3.6432353890711424e-06,
      "loss": 0.0272,
      "num_input_tokens_seen": 20989544,
      "step": 32185
    },
    {
      "epoch": 16.871069182389938,
      "grad_norm": 1.5431188344955444,
      "learning_rate": 3.6372937547989425e-06,
      "loss": 0.0306,
      "num_input_tokens_seen": 20996392,
      "step": 32190
    },
    {
      "epoch": 16.87368972746331,
      "grad_norm": 0.6087685823440552,
      "learning_rate": 3.631356589361173e-06,
      "loss": 0.0149,
      "num_input_tokens_seen": 20999400,
      "step": 32195
    },
    {
      "epoch": 16.87631027253669,
      "grad_norm": 0.6828840374946594,
      "learning_rate": 3.625423893999813e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 21001704,
      "step": 32200
    },
    {
      "epoch": 16.878930817610062,
      "grad_norm": 1.543589472770691,
      "learning_rate": 3.6194956699559217e-06,
      "loss": 0.0076,
      "num_input_tokens_seen": 21004872,
      "step": 32205
    },
    {
      "epoch": 16.88155136268344,
      "grad_norm": 0.47859299182891846,
      "learning_rate": 3.6135719184696188e-06,
      "loss": 0.0313,
      "num_input_tokens_seen": 21011656,
      "step": 32210
    },
    {
      "epoch": 16.884171907756812,
      "grad_norm": 0.2833123207092285,
      "learning_rate": 3.6076526407800766e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 21014472,
      "step": 32215
    },
    {
      "epoch": 16.88679245283019,
      "grad_norm": 0.43101629614830017,
      "learning_rate": 3.601737838125552e-06,
      "loss": 0.0025,
      "num_input_tokens_seen": 21017960,
      "step": 32220
    },
    {
      "epoch": 16.889412997903563,
      "grad_norm": 1.1586098670959473,
      "learning_rate": 3.595827511743341e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 21021288,
      "step": 32225
    },
    {
      "epoch": 16.89203354297694,
      "grad_norm": 0.9264497756958008,
      "learning_rate": 3.5899216628698217e-06,
      "loss": 0.0162,
      "num_input_tokens_seen": 21024872,
      "step": 32230
    },
    {
      "epoch": 16.894654088050313,
      "grad_norm": 1.4217265844345093,
      "learning_rate": 3.584020292740431e-06,
      "loss": 0.016,
      "num_input_tokens_seen": 21028008,
      "step": 32235
    },
    {
      "epoch": 16.89727463312369,
      "grad_norm": 0.18516337871551514,
      "learning_rate": 3.578123402589675e-06,
      "loss": 0.0023,
      "num_input_tokens_seen": 21030920,
      "step": 32240
    },
    {
      "epoch": 16.899895178197063,
      "grad_norm": 0.20835009217262268,
      "learning_rate": 3.5722309936511044e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 21034248,
      "step": 32245
    },
    {
      "epoch": 16.90251572327044,
      "grad_norm": 0.4869280159473419,
      "learning_rate": 3.5663430671573572e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 21037672,
      "step": 32250
    },
    {
      "epoch": 16.905136268343817,
      "grad_norm": 0.32982727885246277,
      "learning_rate": 3.5604596243401038e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 21041192,
      "step": 32255
    },
    {
      "epoch": 16.90775681341719,
      "grad_norm": 0.3641188144683838,
      "learning_rate": 3.554580666430102e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 21044488,
      "step": 32260
    },
    {
      "epoch": 16.910377358490567,
      "grad_norm": 0.5227034687995911,
      "learning_rate": 3.5487061946571614e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 21047752,
      "step": 32265
    },
    {
      "epoch": 16.91299790356394,
      "grad_norm": 0.07603361457586288,
      "learning_rate": 3.542836210250161e-06,
      "loss": 0.0141,
      "num_input_tokens_seen": 21050344,
      "step": 32270
    },
    {
      "epoch": 16.915618448637318,
      "grad_norm": 2.1435656547546387,
      "learning_rate": 3.536970714437032e-06,
      "loss": 0.0089,
      "num_input_tokens_seen": 21053288,
      "step": 32275
    },
    {
      "epoch": 16.91823899371069,
      "grad_norm": 0.25465282797813416,
      "learning_rate": 3.5311097084447575e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 21056904,
      "step": 32280
    },
    {
      "epoch": 16.920859538784068,
      "grad_norm": 0.7157108187675476,
      "learning_rate": 3.5252531934994033e-06,
      "loss": 0.0091,
      "num_input_tokens_seen": 21059848,
      "step": 32285
    },
    {
      "epoch": 16.92348008385744,
      "grad_norm": 0.361804723739624,
      "learning_rate": 3.5194011708260866e-06,
      "loss": 0.0116,
      "num_input_tokens_seen": 21062472,
      "step": 32290
    },
    {
      "epoch": 16.92610062893082,
      "grad_norm": 0.848329484462738,
      "learning_rate": 3.513553641648984e-06,
      "loss": 0.0256,
      "num_input_tokens_seen": 21067752,
      "step": 32295
    },
    {
      "epoch": 16.928721174004192,
      "grad_norm": 0.5778276920318604,
      "learning_rate": 3.5077106071913307e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 21070632,
      "step": 32300
    },
    {
      "epoch": 16.93134171907757,
      "grad_norm": 0.550251841545105,
      "learning_rate": 3.5018720686754133e-06,
      "loss": 0.0097,
      "num_input_tokens_seen": 21074568,
      "step": 32305
    },
    {
      "epoch": 16.933962264150942,
      "grad_norm": 0.5732455849647522,
      "learning_rate": 3.4960380273225985e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 21077288,
      "step": 32310
    },
    {
      "epoch": 16.93658280922432,
      "grad_norm": 1.8973678350448608,
      "learning_rate": 3.490208484353297e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 21079912,
      "step": 32315
    },
    {
      "epoch": 16.939203354297693,
      "grad_norm": 0.486153781414032,
      "learning_rate": 3.4843834409869824e-06,
      "loss": 0.0149,
      "num_input_tokens_seen": 21084040,
      "step": 32320
    },
    {
      "epoch": 16.94182389937107,
      "grad_norm": 0.2896505296230316,
      "learning_rate": 3.4785628984421977e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 21087528,
      "step": 32325
    },
    {
      "epoch": 16.944444444444443,
      "grad_norm": 0.42570897936820984,
      "learning_rate": 3.4727468579365235e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 21090888,
      "step": 32330
    },
    {
      "epoch": 16.94706498951782,
      "grad_norm": 0.1768069863319397,
      "learning_rate": 3.466935320686604e-06,
      "loss": 0.016,
      "num_input_tokens_seen": 21094792,
      "step": 32335
    },
    {
      "epoch": 16.949685534591197,
      "grad_norm": 0.1367226392030716,
      "learning_rate": 3.4611282879081553e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 21097736,
      "step": 32340
    },
    {
      "epoch": 16.95230607966457,
      "grad_norm": 0.4401592016220093,
      "learning_rate": 3.455325760815939e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 21100776,
      "step": 32345
    },
    {
      "epoch": 16.954926624737947,
      "grad_norm": 0.33492350578308105,
      "learning_rate": 3.4495277406237843e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 21104040,
      "step": 32350
    },
    {
      "epoch": 16.95754716981132,
      "grad_norm": 0.5282681584358215,
      "learning_rate": 3.443734228544565e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 21106856,
      "step": 32355
    },
    {
      "epoch": 16.960167714884697,
      "grad_norm": 0.17220629751682281,
      "learning_rate": 3.437945225790215e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 21109768,
      "step": 32360
    },
    {
      "epoch": 16.96278825995807,
      "grad_norm": 0.5646587014198303,
      "learning_rate": 3.432160733571729e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 21113512,
      "step": 32365
    },
    {
      "epoch": 16.965408805031448,
      "grad_norm": 1.087955117225647,
      "learning_rate": 3.4263807530991595e-06,
      "loss": 0.0117,
      "num_input_tokens_seen": 21117544,
      "step": 32370
    },
    {
      "epoch": 16.96802935010482,
      "grad_norm": 1.2263907194137573,
      "learning_rate": 3.4206052855816155e-06,
      "loss": 0.0084,
      "num_input_tokens_seen": 21121864,
      "step": 32375
    },
    {
      "epoch": 16.970649895178198,
      "grad_norm": 0.7975999712944031,
      "learning_rate": 3.4148343322272567e-06,
      "loss": 0.0174,
      "num_input_tokens_seen": 21124328,
      "step": 32380
    },
    {
      "epoch": 16.97327044025157,
      "grad_norm": 0.6218838095664978,
      "learning_rate": 3.4090678942432923e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 21127336,
      "step": 32385
    },
    {
      "epoch": 16.97589098532495,
      "grad_norm": 0.619422197341919,
      "learning_rate": 3.4033059728360018e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 21129896,
      "step": 32390
    },
    {
      "epoch": 16.978511530398322,
      "grad_norm": 1.342516541481018,
      "learning_rate": 3.397548569210715e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 21133512,
      "step": 32395
    },
    {
      "epoch": 16.9811320754717,
      "grad_norm": 1.2485376596450806,
      "learning_rate": 3.3917956845718136e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 21135816,
      "step": 32400
    },
    {
      "epoch": 16.983752620545072,
      "grad_norm": 0.2115437239408493,
      "learning_rate": 3.38604732012274e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 21138536,
      "step": 32405
    },
    {
      "epoch": 16.98637316561845,
      "grad_norm": 0.24402695894241333,
      "learning_rate": 3.3803034770659825e-06,
      "loss": 0.0023,
      "num_input_tokens_seen": 21141160,
      "step": 32410
    },
    {
      "epoch": 16.988993710691823,
      "grad_norm": 4.603599548339844,
      "learning_rate": 3.3745641566030826e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 21143624,
      "step": 32415
    },
    {
      "epoch": 16.9916142557652,
      "grad_norm": 1.7874257564544678,
      "learning_rate": 3.3688293599346442e-06,
      "loss": 0.0177,
      "num_input_tokens_seen": 21147528,
      "step": 32420
    },
    {
      "epoch": 16.994234800838573,
      "grad_norm": 1.0570741891860962,
      "learning_rate": 3.3630990882603213e-06,
      "loss": 0.0115,
      "num_input_tokens_seen": 21150376,
      "step": 32425
    },
    {
      "epoch": 16.99685534591195,
      "grad_norm": 1.3858152627944946,
      "learning_rate": 3.35737334277883e-06,
      "loss": 0.0321,
      "num_input_tokens_seen": 21155016,
      "step": 32430
    },
    {
      "epoch": 16.999475890985323,
      "grad_norm": 0.19848765432834625,
      "learning_rate": 3.351652124687918e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 21157992,
      "step": 32435
    },
    {
      "epoch": 17.0,
      "eval_loss": 1.5084550380706787,
      "eval_runtime": 11.968,
      "eval_samples_per_second": 70.856,
      "eval_steps_per_second": 17.714,
      "num_input_tokens_seen": 21158120,
      "step": 32436
    },
    {
      "epoch": 17.0020964360587,
      "grad_norm": 0.5448803901672363,
      "learning_rate": 3.345935435184408e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 21160712,
      "step": 32440
    },
    {
      "epoch": 17.004716981132077,
      "grad_norm": 0.6190202236175537,
      "learning_rate": 3.3402232754641576e-06,
      "loss": 0.0029,
      "num_input_tokens_seen": 21163528,
      "step": 32445
    },
    {
      "epoch": 17.00733752620545,
      "grad_norm": 0.10061965882778168,
      "learning_rate": 3.334515646722089e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 21167304,
      "step": 32450
    },
    {
      "epoch": 17.009958071278827,
      "grad_norm": 0.4238622486591339,
      "learning_rate": 3.32881255015218e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 21170792,
      "step": 32455
    },
    {
      "epoch": 17.0125786163522,
      "grad_norm": 0.11556193977594376,
      "learning_rate": 3.3231139869474433e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 21174408,
      "step": 32460
    },
    {
      "epoch": 17.015199161425578,
      "grad_norm": 1.0339423418045044,
      "learning_rate": 3.317419958299961e-06,
      "loss": 0.0095,
      "num_input_tokens_seen": 21178248,
      "step": 32465
    },
    {
      "epoch": 17.01781970649895,
      "grad_norm": 0.1425914466381073,
      "learning_rate": 3.31173046540085e-06,
      "loss": 0.0025,
      "num_input_tokens_seen": 21181064,
      "step": 32470
    },
    {
      "epoch": 17.020440251572328,
      "grad_norm": 0.44976580142974854,
      "learning_rate": 3.306045509440295e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 21184296,
      "step": 32475
    },
    {
      "epoch": 17.0230607966457,
      "grad_norm": 0.6114928722381592,
      "learning_rate": 3.300365091607521e-06,
      "loss": 0.0115,
      "num_input_tokens_seen": 21187176,
      "step": 32480
    },
    {
      "epoch": 17.02568134171908,
      "grad_norm": 0.15030749142169952,
      "learning_rate": 3.2946892130908127e-06,
      "loss": 0.0023,
      "num_input_tokens_seen": 21189800,
      "step": 32485
    },
    {
      "epoch": 17.028301886792452,
      "grad_norm": 0.2492496818304062,
      "learning_rate": 3.289017875077488e-06,
      "loss": 0.0689,
      "num_input_tokens_seen": 21196328,
      "step": 32490
    },
    {
      "epoch": 17.03092243186583,
      "grad_norm": 0.22735068202018738,
      "learning_rate": 3.283351078753938e-06,
      "loss": 0.0105,
      "num_input_tokens_seen": 21199400,
      "step": 32495
    },
    {
      "epoch": 17.033542976939202,
      "grad_norm": 1.2235853672027588,
      "learning_rate": 3.2776888253055777e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 21202248,
      "step": 32500
    },
    {
      "epoch": 17.03616352201258,
      "grad_norm": 0.6400109529495239,
      "learning_rate": 3.272031115916896e-06,
      "loss": 0.0093,
      "num_input_tokens_seen": 21205448,
      "step": 32505
    },
    {
      "epoch": 17.038784067085953,
      "grad_norm": 0.4713119864463806,
      "learning_rate": 3.2663779517714238e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 21209576,
      "step": 32510
    },
    {
      "epoch": 17.04140461215933,
      "grad_norm": 0.36211681365966797,
      "learning_rate": 3.260729334051729e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 21212904,
      "step": 32515
    },
    {
      "epoch": 17.044025157232703,
      "grad_norm": 0.3500851094722748,
      "learning_rate": 3.2550852639394446e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 21216104,
      "step": 32520
    },
    {
      "epoch": 17.04664570230608,
      "grad_norm": 0.756913423538208,
      "learning_rate": 3.2494457426152452e-06,
      "loss": 0.0285,
      "num_input_tokens_seen": 21220104,
      "step": 32525
    },
    {
      "epoch": 17.049266247379453,
      "grad_norm": 0.9226870536804199,
      "learning_rate": 3.243810771258851e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 21222920,
      "step": 32530
    },
    {
      "epoch": 17.05188679245283,
      "grad_norm": 0.5488624572753906,
      "learning_rate": 3.2381803510490387e-06,
      "loss": 0.0102,
      "num_input_tokens_seen": 21226408,
      "step": 32535
    },
    {
      "epoch": 17.054507337526207,
      "grad_norm": 0.47933462262153625,
      "learning_rate": 3.2325544831636217e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 21230184,
      "step": 32540
    },
    {
      "epoch": 17.05712788259958,
      "grad_norm": 0.6194645762443542,
      "learning_rate": 3.22693316877947e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 21233288,
      "step": 32545
    },
    {
      "epoch": 17.059748427672957,
      "grad_norm": 1.278383493423462,
      "learning_rate": 3.2213164090725057e-06,
      "loss": 0.0212,
      "num_input_tokens_seen": 21235720,
      "step": 32550
    },
    {
      "epoch": 17.06236897274633,
      "grad_norm": 0.2585483491420746,
      "learning_rate": 3.2157042052176763e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 21238408,
      "step": 32555
    },
    {
      "epoch": 17.064989517819708,
      "grad_norm": 0.17393560707569122,
      "learning_rate": 3.2100965583890003e-06,
      "loss": 0.007,
      "num_input_tokens_seen": 21242632,
      "step": 32560
    },
    {
      "epoch": 17.06761006289308,
      "grad_norm": 1.700261116027832,
      "learning_rate": 3.204493469759537e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 21246056,
      "step": 32565
    },
    {
      "epoch": 17.070230607966458,
      "grad_norm": 0.4184681475162506,
      "learning_rate": 3.1988949405013764e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 21249032,
      "step": 32570
    },
    {
      "epoch": 17.07285115303983,
      "grad_norm": 1.4921303987503052,
      "learning_rate": 3.193300971785676e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 21251816,
      "step": 32575
    },
    {
      "epoch": 17.07547169811321,
      "grad_norm": 0.08214873820543289,
      "learning_rate": 3.1877115647826335e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 21256200,
      "step": 32580
    },
    {
      "epoch": 17.078092243186582,
      "grad_norm": 0.7460284233093262,
      "learning_rate": 3.1821267206614797e-06,
      "loss": 0.0087,
      "num_input_tokens_seen": 21260392,
      "step": 32585
    },
    {
      "epoch": 17.08071278825996,
      "grad_norm": 0.03914866968989372,
      "learning_rate": 3.176546440590508e-06,
      "loss": 0.0114,
      "num_input_tokens_seen": 21264296,
      "step": 32590
    },
    {
      "epoch": 17.083333333333332,
      "grad_norm": 0.09022851288318634,
      "learning_rate": 3.170970725737041e-06,
      "loss": 0.0314,
      "num_input_tokens_seen": 21267848,
      "step": 32595
    },
    {
      "epoch": 17.08595387840671,
      "grad_norm": 0.25571170449256897,
      "learning_rate": 3.165399577267461e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 21271560,
      "step": 32600
    },
    {
      "epoch": 17.088574423480082,
      "grad_norm": 0.25932756066322327,
      "learning_rate": 3.159832996347184e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 21274760,
      "step": 32605
    },
    {
      "epoch": 17.09119496855346,
      "grad_norm": 0.15337501466274261,
      "learning_rate": 3.1542709841406865e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 21277352,
      "step": 32610
    },
    {
      "epoch": 17.093815513626833,
      "grad_norm": 0.6086652278900146,
      "learning_rate": 3.148713541811468e-06,
      "loss": 0.0155,
      "num_input_tokens_seen": 21280168,
      "step": 32615
    },
    {
      "epoch": 17.09643605870021,
      "grad_norm": 0.1435161530971527,
      "learning_rate": 3.1431606705220803e-06,
      "loss": 0.0076,
      "num_input_tokens_seen": 21283688,
      "step": 32620
    },
    {
      "epoch": 17.099056603773583,
      "grad_norm": 0.32606711983680725,
      "learning_rate": 3.1376123714341234e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 21286920,
      "step": 32625
    },
    {
      "epoch": 17.10167714884696,
      "grad_norm": 0.9746379256248474,
      "learning_rate": 3.1320686457082387e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 21290600,
      "step": 32630
    },
    {
      "epoch": 17.104297693920337,
      "grad_norm": 0.5441438555717468,
      "learning_rate": 3.126529494504113e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 21293032,
      "step": 32635
    },
    {
      "epoch": 17.10691823899371,
      "grad_norm": 0.35713431239128113,
      "learning_rate": 3.1209949189804744e-06,
      "loss": 0.0214,
      "num_input_tokens_seen": 21295976,
      "step": 32640
    },
    {
      "epoch": 17.109538784067087,
      "grad_norm": 0.8876765966415405,
      "learning_rate": 3.115464920295089e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 21299816,
      "step": 32645
    },
    {
      "epoch": 17.11215932914046,
      "grad_norm": 0.19433113932609558,
      "learning_rate": 3.1099394996047666e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 21302696,
      "step": 32650
    },
    {
      "epoch": 17.114779874213838,
      "grad_norm": 0.3881048858165741,
      "learning_rate": 3.104418658065367e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 21305288,
      "step": 32655
    },
    {
      "epoch": 17.11740041928721,
      "grad_norm": 0.20899398624897003,
      "learning_rate": 3.098902396831782e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 21308808,
      "step": 32660
    },
    {
      "epoch": 17.120020964360588,
      "grad_norm": 0.10355743765830994,
      "learning_rate": 3.0933907170579646e-06,
      "loss": 0.013,
      "num_input_tokens_seen": 21312008,
      "step": 32665
    },
    {
      "epoch": 17.12264150943396,
      "grad_norm": 0.8517723083496094,
      "learning_rate": 3.0878836198968835e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 21314280,
      "step": 32670
    },
    {
      "epoch": 17.12526205450734,
      "grad_norm": 0.9055953025817871,
      "learning_rate": 3.0823811065005564e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 21316808,
      "step": 32675
    },
    {
      "epoch": 17.127882599580712,
      "grad_norm": 0.7339469790458679,
      "learning_rate": 3.0768831780200542e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 21320712,
      "step": 32680
    },
    {
      "epoch": 17.13050314465409,
      "grad_norm": 0.15372300148010254,
      "learning_rate": 3.071389835605476e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 21324968,
      "step": 32685
    },
    {
      "epoch": 17.133123689727462,
      "grad_norm": 1.560091257095337,
      "learning_rate": 3.065901080405978e-06,
      "loss": 0.0089,
      "num_input_tokens_seen": 21328360,
      "step": 32690
    },
    {
      "epoch": 17.13574423480084,
      "grad_norm": 0.3974732458591461,
      "learning_rate": 3.060416913569733e-06,
      "loss": 0.0097,
      "num_input_tokens_seen": 21331336,
      "step": 32695
    },
    {
      "epoch": 17.138364779874212,
      "grad_norm": 1.5538630485534668,
      "learning_rate": 3.054937336243968e-06,
      "loss": 0.0531,
      "num_input_tokens_seen": 21334344,
      "step": 32700
    },
    {
      "epoch": 17.14098532494759,
      "grad_norm": 0.23417015373706818,
      "learning_rate": 3.049462349574947e-06,
      "loss": 0.0185,
      "num_input_tokens_seen": 21340360,
      "step": 32705
    },
    {
      "epoch": 17.143605870020963,
      "grad_norm": 0.1456325501203537,
      "learning_rate": 3.043991954707981e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 21343176,
      "step": 32710
    },
    {
      "epoch": 17.14622641509434,
      "grad_norm": 0.2659904956817627,
      "learning_rate": 3.0385261527874075e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 21346024,
      "step": 32715
    },
    {
      "epoch": 17.148846960167713,
      "grad_norm": 1.0112380981445312,
      "learning_rate": 3.0330649449566205e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 21350120,
      "step": 32720
    },
    {
      "epoch": 17.15146750524109,
      "grad_norm": 2.271172523498535,
      "learning_rate": 3.027608332358034e-06,
      "loss": 0.0781,
      "num_input_tokens_seen": 21353608,
      "step": 32725
    },
    {
      "epoch": 17.154088050314467,
      "grad_norm": 0.4004916846752167,
      "learning_rate": 3.022156316133104e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 21358376,
      "step": 32730
    },
    {
      "epoch": 17.15670859538784,
      "grad_norm": 0.2518470287322998,
      "learning_rate": 3.016708897422335e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 21361128,
      "step": 32735
    },
    {
      "epoch": 17.159329140461217,
      "grad_norm": 0.16718697547912598,
      "learning_rate": 3.0112660773652685e-06,
      "loss": 0.0087,
      "num_input_tokens_seen": 21364584,
      "step": 32740
    },
    {
      "epoch": 17.16194968553459,
      "grad_norm": 0.2974741756916046,
      "learning_rate": 3.005827857100482e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 21368168,
      "step": 32745
    },
    {
      "epoch": 17.164570230607968,
      "grad_norm": 0.6038219332695007,
      "learning_rate": 3.000394237765586e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 21371752,
      "step": 32750
    },
    {
      "epoch": 17.16719077568134,
      "grad_norm": 0.30255115032196045,
      "learning_rate": 2.9949652204972254e-06,
      "loss": 0.0126,
      "num_input_tokens_seen": 21374632,
      "step": 32755
    },
    {
      "epoch": 17.169811320754718,
      "grad_norm": 0.04602810740470886,
      "learning_rate": 2.989540806431093e-06,
      "loss": 0.007,
      "num_input_tokens_seen": 21377608,
      "step": 32760
    },
    {
      "epoch": 17.17243186582809,
      "grad_norm": 0.516758918762207,
      "learning_rate": 2.984120996701917e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 21381192,
      "step": 32765
    },
    {
      "epoch": 17.17505241090147,
      "grad_norm": 0.895058274269104,
      "learning_rate": 2.9787057924434618e-06,
      "loss": 0.0111,
      "num_input_tokens_seen": 21384168,
      "step": 32770
    },
    {
      "epoch": 17.177672955974842,
      "grad_norm": 0.6156255006790161,
      "learning_rate": 2.9732951947885166e-06,
      "loss": 0.0026,
      "num_input_tokens_seen": 21387240,
      "step": 32775
    },
    {
      "epoch": 17.18029350104822,
      "grad_norm": 0.11394346505403519,
      "learning_rate": 2.967889204868929e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 21390216,
      "step": 32780
    },
    {
      "epoch": 17.182914046121592,
      "grad_norm": 0.7683333158493042,
      "learning_rate": 2.96248782381556e-06,
      "loss": 0.0195,
      "num_input_tokens_seen": 21393544,
      "step": 32785
    },
    {
      "epoch": 17.18553459119497,
      "grad_norm": 0.598351240158081,
      "learning_rate": 2.957091052758321e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 21397288,
      "step": 32790
    },
    {
      "epoch": 17.188155136268342,
      "grad_norm": 0.18426157534122467,
      "learning_rate": 2.9516988928261517e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 21399912,
      "step": 32795
    },
    {
      "epoch": 17.19077568134172,
      "grad_norm": 0.6662938594818115,
      "learning_rate": 2.9463113451470403e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 21402472,
      "step": 32800
    },
    {
      "epoch": 17.193396226415093,
      "grad_norm": 0.25318461656570435,
      "learning_rate": 2.94092841084799e-06,
      "loss": 0.0171,
      "num_input_tokens_seen": 21405352,
      "step": 32805
    },
    {
      "epoch": 17.19601677148847,
      "grad_norm": 0.17791491746902466,
      "learning_rate": 2.9355500910550586e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 21408008,
      "step": 32810
    },
    {
      "epoch": 17.198637316561843,
      "grad_norm": 0.20727258920669556,
      "learning_rate": 2.9301763868933157e-06,
      "loss": 0.0247,
      "num_input_tokens_seen": 21411880,
      "step": 32815
    },
    {
      "epoch": 17.20125786163522,
      "grad_norm": 0.28140193223953247,
      "learning_rate": 2.9248072994868896e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 21414504,
      "step": 32820
    },
    {
      "epoch": 17.203878406708597,
      "grad_norm": 0.276796817779541,
      "learning_rate": 2.919442829958932e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 21417608,
      "step": 32825
    },
    {
      "epoch": 17.20649895178197,
      "grad_norm": 2.2334775924682617,
      "learning_rate": 2.9140829794316234e-06,
      "loss": 0.0106,
      "num_input_tokens_seen": 21420712,
      "step": 32830
    },
    {
      "epoch": 17.209119496855347,
      "grad_norm": 0.29340073466300964,
      "learning_rate": 2.908727749026191e-06,
      "loss": 0.0132,
      "num_input_tokens_seen": 21423400,
      "step": 32835
    },
    {
      "epoch": 17.21174004192872,
      "grad_norm": 0.3101610243320465,
      "learning_rate": 2.9033771398628756e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 21426888,
      "step": 32840
    },
    {
      "epoch": 17.214360587002098,
      "grad_norm": 0.19981130957603455,
      "learning_rate": 2.898031153060973e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 21429736,
      "step": 32845
    },
    {
      "epoch": 17.21698113207547,
      "grad_norm": 0.5619325041770935,
      "learning_rate": 2.8926897897388033e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 21433320,
      "step": 32850
    },
    {
      "epoch": 17.219601677148848,
      "grad_norm": 0.135417640209198,
      "learning_rate": 2.8873530510137136e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 21436584,
      "step": 32855
    },
    {
      "epoch": 17.22222222222222,
      "grad_norm": 1.6153870820999146,
      "learning_rate": 2.8820209380020892e-06,
      "loss": 0.0217,
      "num_input_tokens_seen": 21439528,
      "step": 32860
    },
    {
      "epoch": 17.2248427672956,
      "grad_norm": 0.12790784239768982,
      "learning_rate": 2.8766934518193566e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 21442824,
      "step": 32865
    },
    {
      "epoch": 17.22746331236897,
      "grad_norm": 0.3039587736129761,
      "learning_rate": 2.87137059357995e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 21446024,
      "step": 32870
    },
    {
      "epoch": 17.23008385744235,
      "grad_norm": 0.45617103576660156,
      "learning_rate": 2.8660523643973618e-06,
      "loss": 0.0022,
      "num_input_tokens_seen": 21448744,
      "step": 32875
    },
    {
      "epoch": 17.232704402515722,
      "grad_norm": 0.12881053984165192,
      "learning_rate": 2.860738765384105e-06,
      "loss": 0.0122,
      "num_input_tokens_seen": 21451688,
      "step": 32880
    },
    {
      "epoch": 17.2353249475891,
      "grad_norm": 0.8477606773376465,
      "learning_rate": 2.8554297976517154e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 21453864,
      "step": 32885
    },
    {
      "epoch": 17.237945492662472,
      "grad_norm": 0.2901606559753418,
      "learning_rate": 2.8501254623107707e-06,
      "loss": 0.0155,
      "num_input_tokens_seen": 21456936,
      "step": 32890
    },
    {
      "epoch": 17.24056603773585,
      "grad_norm": 0.5059356093406677,
      "learning_rate": 2.84482576047089e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 21460424,
      "step": 32895
    },
    {
      "epoch": 17.243186582809223,
      "grad_norm": 1.3139156103134155,
      "learning_rate": 2.839530693240694e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 21463016,
      "step": 32900
    },
    {
      "epoch": 17.2458071278826,
      "grad_norm": 1.842179536819458,
      "learning_rate": 2.8342402617278606e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 21466632,
      "step": 32905
    },
    {
      "epoch": 17.248427672955973,
      "grad_norm": 0.0938156396150589,
      "learning_rate": 2.82895446703908e-06,
      "loss": 0.0021,
      "num_input_tokens_seen": 21469544,
      "step": 32910
    },
    {
      "epoch": 17.25104821802935,
      "grad_norm": 0.17599345743656158,
      "learning_rate": 2.823673310280084e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 21472040,
      "step": 32915
    },
    {
      "epoch": 17.253668763102727,
      "grad_norm": 0.17746397852897644,
      "learning_rate": 2.8183967925556305e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 21475784,
      "step": 32920
    },
    {
      "epoch": 17.2562893081761,
      "grad_norm": 1.5758014917373657,
      "learning_rate": 2.8131249149695115e-06,
      "loss": 0.0118,
      "num_input_tokens_seen": 21479656,
      "step": 32925
    },
    {
      "epoch": 17.258909853249477,
      "grad_norm": 0.5146558284759521,
      "learning_rate": 2.8078576786245397e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 21482824,
      "step": 32930
    },
    {
      "epoch": 17.26153039832285,
      "grad_norm": 0.3087303340435028,
      "learning_rate": 2.8025950846225557e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 21485672,
      "step": 32935
    },
    {
      "epoch": 17.264150943396228,
      "grad_norm": 0.2959838807582855,
      "learning_rate": 2.7973371340644373e-06,
      "loss": 0.0026,
      "num_input_tokens_seen": 21488616,
      "step": 32940
    },
    {
      "epoch": 17.2667714884696,
      "grad_norm": 0.30447137355804443,
      "learning_rate": 2.7920838280500907e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 21490952,
      "step": 32945
    },
    {
      "epoch": 17.269392033542978,
      "grad_norm": 1.1541979312896729,
      "learning_rate": 2.7868351676784504e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 21493832,
      "step": 32950
    },
    {
      "epoch": 17.27201257861635,
      "grad_norm": 0.27999740839004517,
      "learning_rate": 2.7815911540474665e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 21496136,
      "step": 32955
    },
    {
      "epoch": 17.27463312368973,
      "grad_norm": 0.4296635389328003,
      "learning_rate": 2.7763517882541364e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 21499432,
      "step": 32960
    },
    {
      "epoch": 17.2772536687631,
      "grad_norm": 0.39539822936058044,
      "learning_rate": 2.771117071394466e-06,
      "loss": 0.0354,
      "num_input_tokens_seen": 21502472,
      "step": 32965
    },
    {
      "epoch": 17.27987421383648,
      "grad_norm": 0.7823672890663147,
      "learning_rate": 2.7658870045635055e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 21506088,
      "step": 32970
    },
    {
      "epoch": 17.282494758909852,
      "grad_norm": 1.1701098680496216,
      "learning_rate": 2.7606615888553234e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 21508904,
      "step": 32975
    },
    {
      "epoch": 17.28511530398323,
      "grad_norm": 0.13219860196113586,
      "learning_rate": 2.755440825363023e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 21511624,
      "step": 32980
    },
    {
      "epoch": 17.287735849056602,
      "grad_norm": 0.4857844114303589,
      "learning_rate": 2.75022471517872e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 21514504,
      "step": 32985
    },
    {
      "epoch": 17.29035639412998,
      "grad_norm": 0.1472516655921936,
      "learning_rate": 2.745013259393564e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 21517256,
      "step": 32990
    },
    {
      "epoch": 17.292976939203353,
      "grad_norm": 0.23347438871860504,
      "learning_rate": 2.739806459097735e-06,
      "loss": 0.0063,
      "num_input_tokens_seen": 21520648,
      "step": 32995
    },
    {
      "epoch": 17.29559748427673,
      "grad_norm": 0.6743999719619751,
      "learning_rate": 2.7346043153804403e-06,
      "loss": 0.0097,
      "num_input_tokens_seen": 21523688,
      "step": 33000
    },
    {
      "epoch": 17.298218029350103,
      "grad_norm": 1.1976673603057861,
      "learning_rate": 2.729406829329903e-06,
      "loss": 0.0076,
      "num_input_tokens_seen": 21527016,
      "step": 33005
    },
    {
      "epoch": 17.30083857442348,
      "grad_norm": 2.6219139099121094,
      "learning_rate": 2.72421400203339e-06,
      "loss": 0.0096,
      "num_input_tokens_seen": 21530408,
      "step": 33010
    },
    {
      "epoch": 17.303459119496857,
      "grad_norm": 0.49481910467147827,
      "learning_rate": 2.719025834577163e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 21533288,
      "step": 33015
    },
    {
      "epoch": 17.30607966457023,
      "grad_norm": 0.269280344247818,
      "learning_rate": 2.713842328046537e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 21536520,
      "step": 33020
    },
    {
      "epoch": 17.308700209643607,
      "grad_norm": 0.9373739957809448,
      "learning_rate": 2.7086634835258416e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 21539304,
      "step": 33025
    },
    {
      "epoch": 17.31132075471698,
      "grad_norm": 0.5195123553276062,
      "learning_rate": 2.7034893020984333e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 21542472,
      "step": 33030
    },
    {
      "epoch": 17.313941299790358,
      "grad_norm": 0.1399776190519333,
      "learning_rate": 2.6983197848466952e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 21545736,
      "step": 33035
    },
    {
      "epoch": 17.31656184486373,
      "grad_norm": 0.3188340663909912,
      "learning_rate": 2.6931549328520265e-06,
      "loss": 0.0105,
      "num_input_tokens_seen": 21548744,
      "step": 33040
    },
    {
      "epoch": 17.319182389937108,
      "grad_norm": 0.9101284742355347,
      "learning_rate": 2.6879947471948523e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 21551592,
      "step": 33045
    },
    {
      "epoch": 17.32180293501048,
      "grad_norm": 0.41371703147888184,
      "learning_rate": 2.6828392289546285e-06,
      "loss": 0.0091,
      "num_input_tokens_seen": 21554216,
      "step": 33050
    },
    {
      "epoch": 17.32442348008386,
      "grad_norm": 0.1251818835735321,
      "learning_rate": 2.677688379209828e-06,
      "loss": 0.012,
      "num_input_tokens_seen": 21556776,
      "step": 33055
    },
    {
      "epoch": 17.32704402515723,
      "grad_norm": 0.6948235034942627,
      "learning_rate": 2.6725421990379586e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 21559976,
      "step": 33060
    },
    {
      "epoch": 17.32966457023061,
      "grad_norm": 0.3442188799381256,
      "learning_rate": 2.667400689515537e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 21562792,
      "step": 33065
    },
    {
      "epoch": 17.332285115303982,
      "grad_norm": 0.43584775924682617,
      "learning_rate": 2.6622638517181e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 21565736,
      "step": 33070
    },
    {
      "epoch": 17.33490566037736,
      "grad_norm": 0.06992492079734802,
      "learning_rate": 2.657131686720224e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 21569864,
      "step": 33075
    },
    {
      "epoch": 17.337526205450732,
      "grad_norm": 0.12504057586193085,
      "learning_rate": 2.652004195595498e-06,
      "loss": 0.0014,
      "num_input_tokens_seen": 21572776,
      "step": 33080
    },
    {
      "epoch": 17.34014675052411,
      "grad_norm": 1.0110503435134888,
      "learning_rate": 2.6468813794165356e-06,
      "loss": 0.0308,
      "num_input_tokens_seen": 21576584,
      "step": 33085
    },
    {
      "epoch": 17.342767295597483,
      "grad_norm": 0.29702189564704895,
      "learning_rate": 2.641763239254977e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 21579496,
      "step": 33090
    },
    {
      "epoch": 17.34538784067086,
      "grad_norm": 0.5304747223854065,
      "learning_rate": 2.6366497761814713e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 21582120,
      "step": 33095
    },
    {
      "epoch": 17.348008385744233,
      "grad_norm": 0.29316720366477966,
      "learning_rate": 2.631540991265691e-06,
      "loss": 0.0147,
      "num_input_tokens_seen": 21585224,
      "step": 33100
    },
    {
      "epoch": 17.35062893081761,
      "grad_norm": 0.7369664907455444,
      "learning_rate": 2.6264368855763443e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 21588264,
      "step": 33105
    },
    {
      "epoch": 17.353249475890987,
      "grad_norm": 1.5138568878173828,
      "learning_rate": 2.6213374601811504e-06,
      "loss": 0.0154,
      "num_input_tokens_seen": 21591464,
      "step": 33110
    },
    {
      "epoch": 17.35587002096436,
      "grad_norm": 1.6029707193374634,
      "learning_rate": 2.616242716146855e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 21594024,
      "step": 33115
    },
    {
      "epoch": 17.358490566037737,
      "grad_norm": 0.23679998517036438,
      "learning_rate": 2.6111526545392145e-06,
      "loss": 0.0223,
      "num_input_tokens_seen": 21596776,
      "step": 33120
    },
    {
      "epoch": 17.36111111111111,
      "grad_norm": 0.16498902440071106,
      "learning_rate": 2.6060672764230106e-06,
      "loss": 0.0022,
      "num_input_tokens_seen": 21600392,
      "step": 33125
    },
    {
      "epoch": 17.363731656184488,
      "grad_norm": 0.1946028769016266,
      "learning_rate": 2.600986582862047e-06,
      "loss": 0.0519,
      "num_input_tokens_seen": 21603880,
      "step": 33130
    },
    {
      "epoch": 17.36635220125786,
      "grad_norm": 0.20909956097602844,
      "learning_rate": 2.5959105749191498e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 21606600,
      "step": 33135
    },
    {
      "epoch": 17.368972746331238,
      "grad_norm": 0.7354848980903625,
      "learning_rate": 2.590839253656166e-06,
      "loss": 0.0136,
      "num_input_tokens_seen": 21610856,
      "step": 33140
    },
    {
      "epoch": 17.37159329140461,
      "grad_norm": 0.6037650108337402,
      "learning_rate": 2.5857726201339473e-06,
      "loss": 0.0116,
      "num_input_tokens_seen": 21615016,
      "step": 33145
    },
    {
      "epoch": 17.37421383647799,
      "grad_norm": 0.4516383707523346,
      "learning_rate": 2.5807106754123873e-06,
      "loss": 0.0166,
      "num_input_tokens_seen": 21618920,
      "step": 33150
    },
    {
      "epoch": 17.37683438155136,
      "grad_norm": 0.2920919358730316,
      "learning_rate": 2.575653420550378e-06,
      "loss": 0.0023,
      "num_input_tokens_seen": 21621960,
      "step": 33155
    },
    {
      "epoch": 17.37945492662474,
      "grad_norm": 0.1753271222114563,
      "learning_rate": 2.57060085660584e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 21624584,
      "step": 33160
    },
    {
      "epoch": 17.382075471698112,
      "grad_norm": 1.377813458442688,
      "learning_rate": 2.565552984635716e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 21628200,
      "step": 33165
    },
    {
      "epoch": 17.38469601677149,
      "grad_norm": 0.44959747791290283,
      "learning_rate": 2.5605098056959652e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 21630792,
      "step": 33170
    },
    {
      "epoch": 17.387316561844862,
      "grad_norm": 0.3879428207874298,
      "learning_rate": 2.555471320841557e-06,
      "loss": 0.018,
      "num_input_tokens_seen": 21633480,
      "step": 33175
    },
    {
      "epoch": 17.38993710691824,
      "grad_norm": 0.2918028235435486,
      "learning_rate": 2.55043753112649e-06,
      "loss": 0.0208,
      "num_input_tokens_seen": 21636584,
      "step": 33180
    },
    {
      "epoch": 17.392557651991613,
      "grad_norm": 0.17004796862602234,
      "learning_rate": 2.5454084376037717e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 21639624,
      "step": 33185
    },
    {
      "epoch": 17.39517819706499,
      "grad_norm": 1.153369665145874,
      "learning_rate": 2.5403840413254304e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 21642248,
      "step": 33190
    },
    {
      "epoch": 17.397798742138363,
      "grad_norm": 0.4491409361362457,
      "learning_rate": 2.5353643433425196e-06,
      "loss": 0.0491,
      "num_input_tokens_seen": 21645576,
      "step": 33195
    },
    {
      "epoch": 17.40041928721174,
      "grad_norm": 0.31811070442199707,
      "learning_rate": 2.5303493447050965e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 21648712,
      "step": 33200
    },
    {
      "epoch": 17.403039832285117,
      "grad_norm": 1.6781976222991943,
      "learning_rate": 2.525339046462244e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 21651624,
      "step": 33205
    },
    {
      "epoch": 17.40566037735849,
      "grad_norm": 1.328167200088501,
      "learning_rate": 2.5203334496620544e-06,
      "loss": 0.0101,
      "num_input_tokens_seen": 21655624,
      "step": 33210
    },
    {
      "epoch": 17.408280922431867,
      "grad_norm": 0.8768362402915955,
      "learning_rate": 2.515332555351646e-06,
      "loss": 0.0087,
      "num_input_tokens_seen": 21658632,
      "step": 33215
    },
    {
      "epoch": 17.41090146750524,
      "grad_norm": 0.2367629110813141,
      "learning_rate": 2.510336364577154e-06,
      "loss": 0.0019,
      "num_input_tokens_seen": 21661736,
      "step": 33220
    },
    {
      "epoch": 17.413522012578618,
      "grad_norm": 1.092977523803711,
      "learning_rate": 2.5053448783837143e-06,
      "loss": 0.0096,
      "num_input_tokens_seen": 21664776,
      "step": 33225
    },
    {
      "epoch": 17.41614255765199,
      "grad_norm": 0.09068329632282257,
      "learning_rate": 2.5003580978154917e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 21667752,
      "step": 33230
    },
    {
      "epoch": 17.418763102725368,
      "grad_norm": 1.3779175281524658,
      "learning_rate": 2.4953760239156716e-06,
      "loss": 0.0705,
      "num_input_tokens_seen": 21672168,
      "step": 33235
    },
    {
      "epoch": 17.42138364779874,
      "grad_norm": 0.5740678310394287,
      "learning_rate": 2.4903986577264366e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 21674472,
      "step": 33240
    },
    {
      "epoch": 17.42400419287212,
      "grad_norm": 0.5436230301856995,
      "learning_rate": 2.4854260002889986e-06,
      "loss": 0.0183,
      "num_input_tokens_seen": 21677608,
      "step": 33245
    },
    {
      "epoch": 17.42662473794549,
      "grad_norm": 1.0593334436416626,
      "learning_rate": 2.480458052643586e-06,
      "loss": 0.03,
      "num_input_tokens_seen": 21680552,
      "step": 33250
    },
    {
      "epoch": 17.42924528301887,
      "grad_norm": 1.1535135507583618,
      "learning_rate": 2.4754948158294293e-06,
      "loss": 0.0164,
      "num_input_tokens_seen": 21682696,
      "step": 33255
    },
    {
      "epoch": 17.431865828092242,
      "grad_norm": 0.9866966009140015,
      "learning_rate": 2.4705362908847814e-06,
      "loss": 0.0141,
      "num_input_tokens_seen": 21686920,
      "step": 33260
    },
    {
      "epoch": 17.43448637316562,
      "grad_norm": 0.21012674272060394,
      "learning_rate": 2.4655824788469172e-06,
      "loss": 0.0019,
      "num_input_tokens_seen": 21689416,
      "step": 33265
    },
    {
      "epoch": 17.437106918238992,
      "grad_norm": 0.31474125385284424,
      "learning_rate": 2.4606333807521087e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 21693480,
      "step": 33270
    },
    {
      "epoch": 17.43972746331237,
      "grad_norm": 0.5393490195274353,
      "learning_rate": 2.455688997635658e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 21696296,
      "step": 33275
    },
    {
      "epoch": 17.442348008385743,
      "grad_norm": 1.492989420890808,
      "learning_rate": 2.4507493305318652e-06,
      "loss": 0.0232,
      "num_input_tokens_seen": 21702088,
      "step": 33280
    },
    {
      "epoch": 17.44496855345912,
      "grad_norm": 0.8613982796669006,
      "learning_rate": 2.445814380474057e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 21705160,
      "step": 33285
    },
    {
      "epoch": 17.447589098532493,
      "grad_norm": 0.3704718053340912,
      "learning_rate": 2.4408841484945684e-06,
      "loss": 0.0108,
      "num_input_tokens_seen": 21708328,
      "step": 33290
    },
    {
      "epoch": 17.45020964360587,
      "grad_norm": 0.851575493812561,
      "learning_rate": 2.435958635624755e-06,
      "loss": 0.0163,
      "num_input_tokens_seen": 21713768,
      "step": 33295
    },
    {
      "epoch": 17.452830188679247,
      "grad_norm": 0.6608280539512634,
      "learning_rate": 2.4310378428949678e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 21717192,
      "step": 33300
    },
    {
      "epoch": 17.45545073375262,
      "grad_norm": 0.47299379110336304,
      "learning_rate": 2.4261217713345804e-06,
      "loss": 0.0177,
      "num_input_tokens_seen": 21721384,
      "step": 33305
    },
    {
      "epoch": 17.458071278825997,
      "grad_norm": 0.18517054617404938,
      "learning_rate": 2.421210421971981e-06,
      "loss": 0.0093,
      "num_input_tokens_seen": 21724392,
      "step": 33310
    },
    {
      "epoch": 17.46069182389937,
      "grad_norm": 0.3534640669822693,
      "learning_rate": 2.416303795834571e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 21727208,
      "step": 33315
    },
    {
      "epoch": 17.463312368972748,
      "grad_norm": 0.11692266911268234,
      "learning_rate": 2.4114018939487615e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 21729960,
      "step": 33320
    },
    {
      "epoch": 17.46593291404612,
      "grad_norm": 1.3142033815383911,
      "learning_rate": 2.4065047173399662e-06,
      "loss": 0.0247,
      "num_input_tokens_seen": 21734216,
      "step": 33325
    },
    {
      "epoch": 17.468553459119498,
      "grad_norm": 1.5068742036819458,
      "learning_rate": 2.4016122670326295e-06,
      "loss": 0.0112,
      "num_input_tokens_seen": 21738152,
      "step": 33330
    },
    {
      "epoch": 17.47117400419287,
      "grad_norm": 0.28144291043281555,
      "learning_rate": 2.396724544050186e-06,
      "loss": 0.045,
      "num_input_tokens_seen": 21741576,
      "step": 33335
    },
    {
      "epoch": 17.47379454926625,
      "grad_norm": 1.5477254390716553,
      "learning_rate": 2.3918415494150944e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 21745224,
      "step": 33340
    },
    {
      "epoch": 17.47641509433962,
      "grad_norm": 4.287103176116943,
      "learning_rate": 2.3869632841488245e-06,
      "loss": 0.0145,
      "num_input_tokens_seen": 21748424,
      "step": 33345
    },
    {
      "epoch": 17.479035639413,
      "grad_norm": 0.25029483437538147,
      "learning_rate": 2.382089749271857e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 21751336,
      "step": 33350
    },
    {
      "epoch": 17.481656184486372,
      "grad_norm": 0.1649494767189026,
      "learning_rate": 2.3772209458036737e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 21754568,
      "step": 33355
    },
    {
      "epoch": 17.48427672955975,
      "grad_norm": 0.4135783910751343,
      "learning_rate": 2.372356874762771e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 21757832,
      "step": 33360
    },
    {
      "epoch": 17.486897274633122,
      "grad_norm": 0.18501891195774078,
      "learning_rate": 2.36749753716666e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 21760968,
      "step": 33365
    },
    {
      "epoch": 17.4895178197065,
      "grad_norm": 1.4372137784957886,
      "learning_rate": 2.3626429340318613e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 21765160,
      "step": 33370
    },
    {
      "epoch": 17.492138364779873,
      "grad_norm": 0.18515172600746155,
      "learning_rate": 2.357793066373901e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 21768584,
      "step": 33375
    },
    {
      "epoch": 17.49475890985325,
      "grad_norm": 9.070976257324219,
      "learning_rate": 2.352947935207317e-06,
      "loss": 0.0092,
      "num_input_tokens_seen": 21771304,
      "step": 33380
    },
    {
      "epoch": 17.497379454926623,
      "grad_norm": 1.1300419569015503,
      "learning_rate": 2.3481075415456495e-06,
      "loss": 0.0126,
      "num_input_tokens_seen": 21774568,
      "step": 33385
    },
    {
      "epoch": 17.5,
      "grad_norm": 0.31416773796081543,
      "learning_rate": 2.3432718864014576e-06,
      "loss": 0.026,
      "num_input_tokens_seen": 21777608,
      "step": 33390
    },
    {
      "epoch": 17.502620545073377,
      "grad_norm": 0.08859699219465256,
      "learning_rate": 2.338440970786304e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 21781288,
      "step": 33395
    },
    {
      "epoch": 17.50524109014675,
      "grad_norm": 0.45408979058265686,
      "learning_rate": 2.3336147957107644e-06,
      "loss": 0.0096,
      "num_input_tokens_seen": 21784168,
      "step": 33400
    },
    {
      "epoch": 17.507861635220127,
      "grad_norm": 0.061936646699905396,
      "learning_rate": 2.3287933621844227e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 21786760,
      "step": 33405
    },
    {
      "epoch": 17.5104821802935,
      "grad_norm": 0.6141833066940308,
      "learning_rate": 2.3239766712158633e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 21789352,
      "step": 33410
    },
    {
      "epoch": 17.513102725366878,
      "grad_norm": 0.23580962419509888,
      "learning_rate": 2.319164723812675e-06,
      "loss": 0.0081,
      "num_input_tokens_seen": 21792296,
      "step": 33415
    },
    {
      "epoch": 17.51572327044025,
      "grad_norm": 0.14057126641273499,
      "learning_rate": 2.314357520981472e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 21794824,
      "step": 33420
    },
    {
      "epoch": 17.518343815513628,
      "grad_norm": 0.7509210705757141,
      "learning_rate": 2.309555063727864e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 21797928,
      "step": 33425
    },
    {
      "epoch": 17.520964360587,
      "grad_norm": 0.15318378806114197,
      "learning_rate": 2.304757353056475e-06,
      "loss": 0.1046,
      "num_input_tokens_seen": 21801352,
      "step": 33430
    },
    {
      "epoch": 17.52358490566038,
      "grad_norm": 0.5747050642967224,
      "learning_rate": 2.2999643899709267e-06,
      "loss": 0.0081,
      "num_input_tokens_seen": 21805768,
      "step": 33435
    },
    {
      "epoch": 17.52620545073375,
      "grad_norm": 0.119967520236969,
      "learning_rate": 2.2951761754738456e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 21808648,
      "step": 33440
    },
    {
      "epoch": 17.52882599580713,
      "grad_norm": 0.4391154646873474,
      "learning_rate": 2.29039271056688e-06,
      "loss": 0.0024,
      "num_input_tokens_seen": 21811080,
      "step": 33445
    },
    {
      "epoch": 17.531446540880502,
      "grad_norm": 0.34433260560035706,
      "learning_rate": 2.285613996250674e-06,
      "loss": 0.0025,
      "num_input_tokens_seen": 21814568,
      "step": 33450
    },
    {
      "epoch": 17.53406708595388,
      "grad_norm": 0.6530725955963135,
      "learning_rate": 2.2808400335248865e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 21817416,
      "step": 33455
    },
    {
      "epoch": 17.536687631027252,
      "grad_norm": 0.2562796473503113,
      "learning_rate": 2.2760708233881654e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 21820648,
      "step": 33460
    },
    {
      "epoch": 17.53930817610063,
      "grad_norm": 0.38872116804122925,
      "learning_rate": 2.271306366838183e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 21823560,
      "step": 33465
    },
    {
      "epoch": 17.541928721174003,
      "grad_norm": 0.5862330794334412,
      "learning_rate": 2.266546664871602e-06,
      "loss": 0.0162,
      "num_input_tokens_seen": 21827208,
      "step": 33470
    },
    {
      "epoch": 17.54454926624738,
      "grad_norm": 0.6092174053192139,
      "learning_rate": 2.261791718484102e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 21830120,
      "step": 33475
    },
    {
      "epoch": 17.547169811320753,
      "grad_norm": 0.7964821457862854,
      "learning_rate": 2.257041528670362e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 21833064,
      "step": 33480
    },
    {
      "epoch": 17.54979035639413,
      "grad_norm": 0.3722606301307678,
      "learning_rate": 2.2522960964240753e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 21835752,
      "step": 33485
    },
    {
      "epoch": 17.552410901467507,
      "grad_norm": 1.2588634490966797,
      "learning_rate": 2.247555422737924e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 21838664,
      "step": 33490
    },
    {
      "epoch": 17.55503144654088,
      "grad_norm": 0.26473331451416016,
      "learning_rate": 2.2428195086036024e-06,
      "loss": 0.0187,
      "num_input_tokens_seen": 21842184,
      "step": 33495
    },
    {
      "epoch": 17.557651991614257,
      "grad_norm": 0.6392934918403625,
      "learning_rate": 2.238088355011811e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 21845224,
      "step": 33500
    },
    {
      "epoch": 17.56027253668763,
      "grad_norm": 2.322519302368164,
      "learning_rate": 2.2333619629522518e-06,
      "loss": 0.0526,
      "num_input_tokens_seen": 21848552,
      "step": 33505
    },
    {
      "epoch": 17.562893081761008,
      "grad_norm": 0.20402157306671143,
      "learning_rate": 2.2286403334136424e-06,
      "loss": 0.0099,
      "num_input_tokens_seen": 21850856,
      "step": 33510
    },
    {
      "epoch": 17.56551362683438,
      "grad_norm": 0.15759187936782837,
      "learning_rate": 2.223923467383679e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 21856104,
      "step": 33515
    },
    {
      "epoch": 17.568134171907758,
      "grad_norm": 0.355281263589859,
      "learning_rate": 2.2192113658490886e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 21858440,
      "step": 33520
    },
    {
      "epoch": 17.57075471698113,
      "grad_norm": 0.22642239928245544,
      "learning_rate": 2.2145040297955776e-06,
      "loss": 0.0327,
      "num_input_tokens_seen": 21861064,
      "step": 33525
    },
    {
      "epoch": 17.57337526205451,
      "grad_norm": 0.04495874047279358,
      "learning_rate": 2.209801460207875e-06,
      "loss": 0.0017,
      "num_input_tokens_seen": 21863848,
      "step": 33530
    },
    {
      "epoch": 17.57599580712788,
      "grad_norm": 0.17422278225421906,
      "learning_rate": 2.2051036580697042e-06,
      "loss": 0.0121,
      "num_input_tokens_seen": 21867208,
      "step": 33535
    },
    {
      "epoch": 17.57861635220126,
      "grad_norm": 3.3637545108795166,
      "learning_rate": 2.2004106243637885e-06,
      "loss": 0.0097,
      "num_input_tokens_seen": 21871944,
      "step": 33540
    },
    {
      "epoch": 17.581236897274632,
      "grad_norm": 0.1958916038274765,
      "learning_rate": 2.1957223600718587e-06,
      "loss": 0.0017,
      "num_input_tokens_seen": 21874728,
      "step": 33545
    },
    {
      "epoch": 17.58385744234801,
      "grad_norm": 0.8553177714347839,
      "learning_rate": 2.1910388661746493e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 21877320,
      "step": 33550
    },
    {
      "epoch": 17.586477987421382,
      "grad_norm": 0.3160340189933777,
      "learning_rate": 2.186360143651886e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 21880488,
      "step": 33555
    },
    {
      "epoch": 17.58909853249476,
      "grad_norm": 4.618019104003906,
      "learning_rate": 2.1816861934823103e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 21883080,
      "step": 33560
    },
    {
      "epoch": 17.591719077568133,
      "grad_norm": 0.6970313191413879,
      "learning_rate": 2.1770170166436595e-06,
      "loss": 0.0087,
      "num_input_tokens_seen": 21886536,
      "step": 33565
    },
    {
      "epoch": 17.59433962264151,
      "grad_norm": 0.37431538105010986,
      "learning_rate": 2.172352614112669e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 21889896,
      "step": 33570
    },
    {
      "epoch": 17.596960167714883,
      "grad_norm": 1.2848490476608276,
      "learning_rate": 2.167692986865083e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 21893256,
      "step": 33575
    },
    {
      "epoch": 17.59958071278826,
      "grad_norm": 0.2379029244184494,
      "learning_rate": 2.1630381358756362e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 21896328,
      "step": 33580
    },
    {
      "epoch": 17.602201257861637,
      "grad_norm": 0.3602483868598938,
      "learning_rate": 2.158388062118072e-06,
      "loss": 0.0215,
      "num_input_tokens_seen": 21898856,
      "step": 33585
    },
    {
      "epoch": 17.60482180293501,
      "grad_norm": 0.2994285523891449,
      "learning_rate": 2.1537427665651393e-06,
      "loss": 0.0138,
      "num_input_tokens_seen": 21901992,
      "step": 33590
    },
    {
      "epoch": 17.607442348008387,
      "grad_norm": 0.3010431230068207,
      "learning_rate": 2.1491022501885726e-06,
      "loss": 0.0021,
      "num_input_tokens_seen": 21904904,
      "step": 33595
    },
    {
      "epoch": 17.61006289308176,
      "grad_norm": 0.6028243899345398,
      "learning_rate": 2.1444665139591174e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 21908808,
      "step": 33600
    },
    {
      "epoch": 17.612683438155138,
      "grad_norm": 0.49938079714775085,
      "learning_rate": 2.139835558846523e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 21911720,
      "step": 33605
    },
    {
      "epoch": 17.61530398322851,
      "grad_norm": 0.4597567915916443,
      "learning_rate": 2.1352093858195225e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 21914024,
      "step": 33610
    },
    {
      "epoch": 17.617924528301888,
      "grad_norm": 1.800225853919983,
      "learning_rate": 2.1305879958458676e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 21916648,
      "step": 33615
    },
    {
      "epoch": 17.62054507337526,
      "grad_norm": 5.611691951751709,
      "learning_rate": 2.1259713898922894e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 21920136,
      "step": 33620
    },
    {
      "epoch": 17.62316561844864,
      "grad_norm": 1.0120257139205933,
      "learning_rate": 2.1213595689245386e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 21924168,
      "step": 33625
    },
    {
      "epoch": 17.62578616352201,
      "grad_norm": 0.13077932596206665,
      "learning_rate": 2.1167525339073513e-06,
      "loss": 0.0283,
      "num_input_tokens_seen": 21931656,
      "step": 33630
    },
    {
      "epoch": 17.62840670859539,
      "grad_norm": 0.4235876500606537,
      "learning_rate": 2.112150285804473e-06,
      "loss": 0.0063,
      "num_input_tokens_seen": 21934184,
      "step": 33635
    },
    {
      "epoch": 17.631027253668762,
      "grad_norm": 0.24156980216503143,
      "learning_rate": 2.107552825578632e-06,
      "loss": 0.0142,
      "num_input_tokens_seen": 21938856,
      "step": 33640
    },
    {
      "epoch": 17.63364779874214,
      "grad_norm": 0.20727480947971344,
      "learning_rate": 2.102960154191572e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 21941736,
      "step": 33645
    },
    {
      "epoch": 17.636268343815512,
      "grad_norm": 1.0998852252960205,
      "learning_rate": 2.098372272604021e-06,
      "loss": 0.0254,
      "num_input_tokens_seen": 21945256,
      "step": 33650
    },
    {
      "epoch": 17.63888888888889,
      "grad_norm": 0.6520236730575562,
      "learning_rate": 2.0937891817757155e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 21947432,
      "step": 33655
    },
    {
      "epoch": 17.641509433962263,
      "grad_norm": 0.1602245569229126,
      "learning_rate": 2.089210882665385e-06,
      "loss": 0.0097,
      "num_input_tokens_seen": 21950664,
      "step": 33660
    },
    {
      "epoch": 17.64412997903564,
      "grad_norm": 0.6894542574882507,
      "learning_rate": 2.0846373762307635e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 21953704,
      "step": 33665
    },
    {
      "epoch": 17.646750524109013,
      "grad_norm": 1.2029997110366821,
      "learning_rate": 2.080068663428569e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 21956936,
      "step": 33670
    },
    {
      "epoch": 17.64937106918239,
      "grad_norm": 0.7853367328643799,
      "learning_rate": 2.075504745214521e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 21960200,
      "step": 33675
    },
    {
      "epoch": 17.651991614255767,
      "grad_norm": 0.5480356812477112,
      "learning_rate": 2.0709456225433437e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 21962824,
      "step": 33680
    },
    {
      "epoch": 17.65461215932914,
      "grad_norm": 0.13921472430229187,
      "learning_rate": 2.0663912963687554e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 21966504,
      "step": 33685
    },
    {
      "epoch": 17.657232704402517,
      "grad_norm": 0.276241272687912,
      "learning_rate": 2.0618417676434713e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 21970664,
      "step": 33690
    },
    {
      "epoch": 17.65985324947589,
      "grad_norm": 1.0093121528625488,
      "learning_rate": 2.0572970373191975e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 21974312,
      "step": 33695
    },
    {
      "epoch": 17.662473794549268,
      "grad_norm": 0.9065364599227905,
      "learning_rate": 2.052757106346634e-06,
      "loss": 0.0589,
      "num_input_tokens_seen": 21977288,
      "step": 33700
    },
    {
      "epoch": 17.66509433962264,
      "grad_norm": 0.5786466002464294,
      "learning_rate": 2.048221975675488e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 21979720,
      "step": 33705
    },
    {
      "epoch": 17.667714884696018,
      "grad_norm": 0.1749134361743927,
      "learning_rate": 2.043691646254456e-06,
      "loss": 0.0024,
      "num_input_tokens_seen": 21982600,
      "step": 33710
    },
    {
      "epoch": 17.67033542976939,
      "grad_norm": 0.19804885983467102,
      "learning_rate": 2.039166119031233e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 21986184,
      "step": 33715
    },
    {
      "epoch": 17.67295597484277,
      "grad_norm": 0.43651625514030457,
      "learning_rate": 2.034645394952511e-06,
      "loss": 0.0151,
      "num_input_tokens_seen": 21989352,
      "step": 33720
    },
    {
      "epoch": 17.67557651991614,
      "grad_norm": 0.44183287024497986,
      "learning_rate": 2.030129474963971e-06,
      "loss": 0.0122,
      "num_input_tokens_seen": 21992072,
      "step": 33725
    },
    {
      "epoch": 17.67819706498952,
      "grad_norm": 0.10405300557613373,
      "learning_rate": 2.025618360010284e-06,
      "loss": 0.0111,
      "num_input_tokens_seen": 21996200,
      "step": 33730
    },
    {
      "epoch": 17.680817610062892,
      "grad_norm": 0.4257368743419647,
      "learning_rate": 2.0211120510351317e-06,
      "loss": 0.0026,
      "num_input_tokens_seen": 21998728,
      "step": 33735
    },
    {
      "epoch": 17.68343815513627,
      "grad_norm": 0.5531609058380127,
      "learning_rate": 2.0166105489811822e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 22001480,
      "step": 33740
    },
    {
      "epoch": 17.686058700209642,
      "grad_norm": 0.19033269584178925,
      "learning_rate": 2.0121138547901016e-06,
      "loss": 0.0203,
      "num_input_tokens_seen": 22005800,
      "step": 33745
    },
    {
      "epoch": 17.68867924528302,
      "grad_norm": 0.11395059525966644,
      "learning_rate": 2.0076219694025427e-06,
      "loss": 0.0023,
      "num_input_tokens_seen": 22009096,
      "step": 33750
    },
    {
      "epoch": 17.691299790356393,
      "grad_norm": 0.2249070405960083,
      "learning_rate": 2.003134893758152e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 22012072,
      "step": 33755
    },
    {
      "epoch": 17.69392033542977,
      "grad_norm": 0.49793022871017456,
      "learning_rate": 1.9986526287955805e-06,
      "loss": 0.0157,
      "num_input_tokens_seen": 22015304,
      "step": 33760
    },
    {
      "epoch": 17.696540880503143,
      "grad_norm": 0.6281898021697998,
      "learning_rate": 1.9941751754524644e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 22019016,
      "step": 33765
    },
    {
      "epoch": 17.69916142557652,
      "grad_norm": 1.6937439441680908,
      "learning_rate": 1.9897025346654412e-06,
      "loss": 0.0278,
      "num_input_tokens_seen": 22021992,
      "step": 33770
    },
    {
      "epoch": 17.701781970649897,
      "grad_norm": 0.419238418340683,
      "learning_rate": 1.985234707370134e-06,
      "loss": 0.0026,
      "num_input_tokens_seen": 22024808,
      "step": 33775
    },
    {
      "epoch": 17.70440251572327,
      "grad_norm": 0.11057634651660919,
      "learning_rate": 1.9807716945011544e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 22027336,
      "step": 33780
    },
    {
      "epoch": 17.707023060796647,
      "grad_norm": 0.3077448308467865,
      "learning_rate": 1.9763134969921215e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 22030440,
      "step": 33785
    },
    {
      "epoch": 17.70964360587002,
      "grad_norm": 0.6522386074066162,
      "learning_rate": 1.971860115775634e-06,
      "loss": 0.0221,
      "num_input_tokens_seen": 22033160,
      "step": 33790
    },
    {
      "epoch": 17.712264150943398,
      "grad_norm": 0.23328499495983124,
      "learning_rate": 1.967411551783294e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 22035720,
      "step": 33795
    },
    {
      "epoch": 17.71488469601677,
      "grad_norm": 0.24811919033527374,
      "learning_rate": 1.9629678059456895e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 22038664,
      "step": 33800
    },
    {
      "epoch": 17.717505241090148,
      "grad_norm": 0.36983150243759155,
      "learning_rate": 1.9585288791924004e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 22041512,
      "step": 33805
    },
    {
      "epoch": 17.72012578616352,
      "grad_norm": 0.5139662623405457,
      "learning_rate": 1.9540947724519947e-06,
      "loss": 0.0063,
      "num_input_tokens_seen": 22045320,
      "step": 33810
    },
    {
      "epoch": 17.7227463312369,
      "grad_norm": 0.15824870765209198,
      "learning_rate": 1.9496654866520414e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 22048232,
      "step": 33815
    },
    {
      "epoch": 17.72536687631027,
      "grad_norm": 1.1099311113357544,
      "learning_rate": 1.9452410227190937e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 22051528,
      "step": 33820
    },
    {
      "epoch": 17.72798742138365,
      "grad_norm": 0.9817964434623718,
      "learning_rate": 1.9408213815787087e-06,
      "loss": 0.0079,
      "num_input_tokens_seen": 22054248,
      "step": 33825
    },
    {
      "epoch": 17.730607966457022,
      "grad_norm": 0.6715094447135925,
      "learning_rate": 1.936406564155413e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 22058600,
      "step": 33830
    },
    {
      "epoch": 17.7332285115304,
      "grad_norm": 2.40167236328125,
      "learning_rate": 1.9319965713727435e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 22061896,
      "step": 33835
    },
    {
      "epoch": 17.735849056603772,
      "grad_norm": 0.6046209335327148,
      "learning_rate": 1.9275914041532144e-06,
      "loss": 0.0245,
      "num_input_tokens_seen": 22065192,
      "step": 33840
    },
    {
      "epoch": 17.73846960167715,
      "grad_norm": 0.19295015931129456,
      "learning_rate": 1.9231910634183394e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 22068296,
      "step": 33845
    },
    {
      "epoch": 17.741090146750523,
      "grad_norm": 0.30538463592529297,
      "learning_rate": 1.918795550088623e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 22071976,
      "step": 33850
    },
    {
      "epoch": 17.7437106918239,
      "grad_norm": 0.9151598215103149,
      "learning_rate": 1.9144048650835496e-06,
      "loss": 0.0137,
      "num_input_tokens_seen": 22075656,
      "step": 33855
    },
    {
      "epoch": 17.746331236897273,
      "grad_norm": 1.1378501653671265,
      "learning_rate": 1.9100190093216093e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 22079816,
      "step": 33860
    },
    {
      "epoch": 17.74895178197065,
      "grad_norm": 0.43634364008903503,
      "learning_rate": 1.9056379837202627e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 22082472,
      "step": 33865
    },
    {
      "epoch": 17.751572327044027,
      "grad_norm": 0.1450774073600769,
      "learning_rate": 1.9012617891959766e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 22086024,
      "step": 33870
    },
    {
      "epoch": 17.7541928721174,
      "grad_norm": 0.7938163876533508,
      "learning_rate": 1.8968904266641996e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 22090024,
      "step": 33875
    },
    {
      "epoch": 17.756813417190777,
      "grad_norm": 0.4663667678833008,
      "learning_rate": 1.892523897039372e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 22092680,
      "step": 33880
    },
    {
      "epoch": 17.75943396226415,
      "grad_norm": 1.6047269105911255,
      "learning_rate": 1.888162201234922e-06,
      "loss": 0.0119,
      "num_input_tokens_seen": 22095112,
      "step": 33885
    },
    {
      "epoch": 17.762054507337528,
      "grad_norm": 0.49989742040634155,
      "learning_rate": 1.8838053401632666e-06,
      "loss": 0.02,
      "num_input_tokens_seen": 22098696,
      "step": 33890
    },
    {
      "epoch": 17.7646750524109,
      "grad_norm": 0.3626878261566162,
      "learning_rate": 1.8794533147358074e-06,
      "loss": 0.0105,
      "num_input_tokens_seen": 22102024,
      "step": 33895
    },
    {
      "epoch": 17.767295597484278,
      "grad_norm": 0.44925183057785034,
      "learning_rate": 1.875106125862941e-06,
      "loss": 0.0109,
      "num_input_tokens_seen": 22105064,
      "step": 33900
    },
    {
      "epoch": 17.76991614255765,
      "grad_norm": 0.7028487920761108,
      "learning_rate": 1.870763774454057e-06,
      "loss": 0.0245,
      "num_input_tokens_seen": 22108968,
      "step": 33905
    },
    {
      "epoch": 17.77253668763103,
      "grad_norm": 0.46664169430732727,
      "learning_rate": 1.8664262614175115e-06,
      "loss": 0.0093,
      "num_input_tokens_seen": 22113288,
      "step": 33910
    },
    {
      "epoch": 17.7751572327044,
      "grad_norm": 0.8587592840194702,
      "learning_rate": 1.862093587660671e-06,
      "loss": 0.0183,
      "num_input_tokens_seen": 22117768,
      "step": 33915
    },
    {
      "epoch": 17.77777777777778,
      "grad_norm": 0.7082538604736328,
      "learning_rate": 1.8577657540898851e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 22120936,
      "step": 33920
    },
    {
      "epoch": 17.780398322851152,
      "grad_norm": 0.08796797692775726,
      "learning_rate": 1.8534427616104798e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 22124616,
      "step": 33925
    },
    {
      "epoch": 17.78301886792453,
      "grad_norm": 0.5916779041290283,
      "learning_rate": 1.8491246111267846e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 22128072,
      "step": 33930
    },
    {
      "epoch": 17.785639412997902,
      "grad_norm": 0.21410241723060608,
      "learning_rate": 1.8448113035420934e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 22131272,
      "step": 33935
    },
    {
      "epoch": 17.78825995807128,
      "grad_norm": 0.4399666488170624,
      "learning_rate": 1.840502839758712e-06,
      "loss": 0.0089,
      "num_input_tokens_seen": 22134888,
      "step": 33940
    },
    {
      "epoch": 17.790880503144653,
      "grad_norm": 0.4558431804180145,
      "learning_rate": 1.8361992206779226e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 22137480,
      "step": 33945
    },
    {
      "epoch": 17.79350104821803,
      "grad_norm": 0.23536376655101776,
      "learning_rate": 1.8319004471999856e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 22141000,
      "step": 33950
    },
    {
      "epoch": 17.796121593291403,
      "grad_norm": 0.05073879659175873,
      "learning_rate": 1.8276065202241592e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 22143752,
      "step": 33955
    },
    {
      "epoch": 17.79874213836478,
      "grad_norm": 0.4485999643802643,
      "learning_rate": 1.8233174406486863e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 22147272,
      "step": 33960
    },
    {
      "epoch": 17.801362683438157,
      "grad_norm": 0.06269992142915726,
      "learning_rate": 1.8190332093707907e-06,
      "loss": 0.0019,
      "num_input_tokens_seen": 22151080,
      "step": 33965
    },
    {
      "epoch": 17.80398322851153,
      "grad_norm": 0.09578154981136322,
      "learning_rate": 1.8147538272866833e-06,
      "loss": 0.0029,
      "num_input_tokens_seen": 22154664,
      "step": 33970
    },
    {
      "epoch": 17.806603773584907,
      "grad_norm": 1.3300824165344238,
      "learning_rate": 1.8104792952915672e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 22157608,
      "step": 33975
    },
    {
      "epoch": 17.80922431865828,
      "grad_norm": 1.001826286315918,
      "learning_rate": 1.8062096142796192e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 22160136,
      "step": 33980
    },
    {
      "epoch": 17.811844863731658,
      "grad_norm": 0.08702949434518814,
      "learning_rate": 1.8019447851440163e-06,
      "loss": 0.0027,
      "num_input_tokens_seen": 22162376,
      "step": 33985
    },
    {
      "epoch": 17.81446540880503,
      "grad_norm": 0.5060375928878784,
      "learning_rate": 1.7976848087769032e-06,
      "loss": 0.0416,
      "num_input_tokens_seen": 22166376,
      "step": 33990
    },
    {
      "epoch": 17.817085953878408,
      "grad_norm": 0.1488654464483261,
      "learning_rate": 1.79342968606942e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 22169128,
      "step": 33995
    },
    {
      "epoch": 17.81970649895178,
      "grad_norm": 3.2146618366241455,
      "learning_rate": 1.7891794179116933e-06,
      "loss": 0.0162,
      "num_input_tokens_seen": 22173544,
      "step": 34000
    },
    {
      "epoch": 17.822327044025158,
      "grad_norm": 0.279445081949234,
      "learning_rate": 1.7849340051928316e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 22176712,
      "step": 34005
    },
    {
      "epoch": 17.82494758909853,
      "grad_norm": 0.20360544323921204,
      "learning_rate": 1.780693448800927e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 22179816,
      "step": 34010
    },
    {
      "epoch": 17.82756813417191,
      "grad_norm": 0.23009106516838074,
      "learning_rate": 1.7764577496230478e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 22182888,
      "step": 34015
    },
    {
      "epoch": 17.830188679245282,
      "grad_norm": 0.9493369460105896,
      "learning_rate": 1.7722269085452602e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 22186472,
      "step": 34020
    },
    {
      "epoch": 17.83280922431866,
      "grad_norm": 0.8256745934486389,
      "learning_rate": 1.7680009264526038e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 22189320,
      "step": 34025
    },
    {
      "epoch": 17.835429769392032,
      "grad_norm": 0.12553909420967102,
      "learning_rate": 1.7637798042291126e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 22192200,
      "step": 34030
    },
    {
      "epoch": 17.83805031446541,
      "grad_norm": 0.1621808111667633,
      "learning_rate": 1.7595635427577944e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 22194696,
      "step": 34035
    },
    {
      "epoch": 17.840670859538783,
      "grad_norm": 0.4850284457206726,
      "learning_rate": 1.7553521429206465e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 22197192,
      "step": 34040
    },
    {
      "epoch": 17.84329140461216,
      "grad_norm": 1.1204547882080078,
      "learning_rate": 1.7511456055986335e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 22199976,
      "step": 34045
    },
    {
      "epoch": 17.845911949685533,
      "grad_norm": 1.5999424457550049,
      "learning_rate": 1.746943931671724e-06,
      "loss": 0.01,
      "num_input_tokens_seen": 22203176,
      "step": 34050
    },
    {
      "epoch": 17.84853249475891,
      "grad_norm": 0.45547324419021606,
      "learning_rate": 1.742747122018859e-06,
      "loss": 0.0099,
      "num_input_tokens_seen": 22207368,
      "step": 34055
    },
    {
      "epoch": 17.851153039832283,
      "grad_norm": 0.4065294563770294,
      "learning_rate": 1.738555177517967e-06,
      "loss": 0.0079,
      "num_input_tokens_seen": 22210376,
      "step": 34060
    },
    {
      "epoch": 17.85377358490566,
      "grad_norm": 0.5373441576957703,
      "learning_rate": 1.734368099045952e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 22214856,
      "step": 34065
    },
    {
      "epoch": 17.856394129979037,
      "grad_norm": 1.0346076488494873,
      "learning_rate": 1.730185887478697e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 22218504,
      "step": 34070
    },
    {
      "epoch": 17.85901467505241,
      "grad_norm": 0.21556827425956726,
      "learning_rate": 1.7260085436910794e-06,
      "loss": 0.0189,
      "num_input_tokens_seen": 22221384,
      "step": 34075
    },
    {
      "epoch": 17.861635220125788,
      "grad_norm": 0.25744426250457764,
      "learning_rate": 1.7218360685569506e-06,
      "loss": 0.0134,
      "num_input_tokens_seen": 22225480,
      "step": 34080
    },
    {
      "epoch": 17.86425576519916,
      "grad_norm": 0.4512694180011749,
      "learning_rate": 1.7176684629491513e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 22228488,
      "step": 34085
    },
    {
      "epoch": 17.866876310272538,
      "grad_norm": 0.42068469524383545,
      "learning_rate": 1.7135057277394868e-06,
      "loss": 0.018,
      "num_input_tokens_seen": 22231912,
      "step": 34090
    },
    {
      "epoch": 17.86949685534591,
      "grad_norm": 0.48326706886291504,
      "learning_rate": 1.7093478637987548e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 22235144,
      "step": 34095
    },
    {
      "epoch": 17.872117400419288,
      "grad_norm": 2.650491714477539,
      "learning_rate": 1.7051948719967375e-06,
      "loss": 0.0248,
      "num_input_tokens_seen": 22238088,
      "step": 34100
    },
    {
      "epoch": 17.87473794549266,
      "grad_norm": 0.9445568919181824,
      "learning_rate": 1.7010467532021901e-06,
      "loss": 0.0287,
      "num_input_tokens_seen": 22241800,
      "step": 34105
    },
    {
      "epoch": 17.87735849056604,
      "grad_norm": 0.3361772894859314,
      "learning_rate": 1.6969035082828544e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 22245512,
      "step": 34110
    },
    {
      "epoch": 17.879979035639412,
      "grad_norm": 0.8762651681900024,
      "learning_rate": 1.692765138105451e-06,
      "loss": 0.009,
      "num_input_tokens_seen": 22248328,
      "step": 34115
    },
    {
      "epoch": 17.88259958071279,
      "grad_norm": 0.44955575466156006,
      "learning_rate": 1.6886316435356792e-06,
      "loss": 0.0124,
      "num_input_tokens_seen": 22252744,
      "step": 34120
    },
    {
      "epoch": 17.885220125786162,
      "grad_norm": 0.3234851658344269,
      "learning_rate": 1.6845030254382138e-06,
      "loss": 0.0092,
      "num_input_tokens_seen": 22257544,
      "step": 34125
    },
    {
      "epoch": 17.88784067085954,
      "grad_norm": 0.4268438518047333,
      "learning_rate": 1.6803792846767197e-06,
      "loss": 0.009,
      "num_input_tokens_seen": 22260456,
      "step": 34130
    },
    {
      "epoch": 17.890461215932913,
      "grad_norm": 1.5896973609924316,
      "learning_rate": 1.6762604221138316e-06,
      "loss": 0.0116,
      "num_input_tokens_seen": 22264328,
      "step": 34135
    },
    {
      "epoch": 17.89308176100629,
      "grad_norm": 0.18672776222229004,
      "learning_rate": 1.6721464386111774e-06,
      "loss": 0.0093,
      "num_input_tokens_seen": 22267080,
      "step": 34140
    },
    {
      "epoch": 17.895702306079663,
      "grad_norm": 0.3978385627269745,
      "learning_rate": 1.6680373350293516e-06,
      "loss": 0.0019,
      "num_input_tokens_seen": 22270248,
      "step": 34145
    },
    {
      "epoch": 17.89832285115304,
      "grad_norm": 0.6797674894332886,
      "learning_rate": 1.6639331122279223e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 22273960,
      "step": 34150
    },
    {
      "epoch": 17.900943396226417,
      "grad_norm": 1.136540174484253,
      "learning_rate": 1.6598337710654583e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 22277256,
      "step": 34155
    },
    {
      "epoch": 17.90356394129979,
      "grad_norm": 1.3170791864395142,
      "learning_rate": 1.6557393123994874e-06,
      "loss": 0.0173,
      "num_input_tokens_seen": 22280424,
      "step": 34160
    },
    {
      "epoch": 17.906184486373167,
      "grad_norm": 0.7536428570747375,
      "learning_rate": 1.651649737086533e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 22284808,
      "step": 34165
    },
    {
      "epoch": 17.90880503144654,
      "grad_norm": 1.7247669696807861,
      "learning_rate": 1.6475650459820774e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 22287912,
      "step": 34170
    },
    {
      "epoch": 17.911425576519918,
      "grad_norm": 0.5828481316566467,
      "learning_rate": 1.643485239940598e-06,
      "loss": 0.0079,
      "num_input_tokens_seen": 22291112,
      "step": 34175
    },
    {
      "epoch": 17.91404612159329,
      "grad_norm": 0.4691927134990692,
      "learning_rate": 1.6394103198155375e-06,
      "loss": 0.0113,
      "num_input_tokens_seen": 22294280,
      "step": 34180
    },
    {
      "epoch": 17.916666666666668,
      "grad_norm": 0.03706243261694908,
      "learning_rate": 1.6353402864593276e-06,
      "loss": 0.0027,
      "num_input_tokens_seen": 22297640,
      "step": 34185
    },
    {
      "epoch": 17.91928721174004,
      "grad_norm": 0.5419816970825195,
      "learning_rate": 1.6312751407233706e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 22300712,
      "step": 34190
    },
    {
      "epoch": 17.921907756813418,
      "grad_norm": 0.5582758188247681,
      "learning_rate": 1.6272148834580559e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 22304072,
      "step": 34195
    },
    {
      "epoch": 17.92452830188679,
      "grad_norm": 0.42027679085731506,
      "learning_rate": 1.6231595155127317e-06,
      "loss": 0.0494,
      "num_input_tokens_seen": 22307336,
      "step": 34200
    },
    {
      "epoch": 17.92714884696017,
      "grad_norm": 0.33613499999046326,
      "learning_rate": 1.6191090377357414e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 22310216,
      "step": 34205
    },
    {
      "epoch": 17.929769392033542,
      "grad_norm": 0.7170149683952332,
      "learning_rate": 1.6150634509743939e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 22314216,
      "step": 34210
    },
    {
      "epoch": 17.93238993710692,
      "grad_norm": 0.6987794041633606,
      "learning_rate": 1.611022756074984e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 22317416,
      "step": 34215
    },
    {
      "epoch": 17.935010482180292,
      "grad_norm": 0.1366363912820816,
      "learning_rate": 1.606986953882783e-06,
      "loss": 0.0128,
      "num_input_tokens_seen": 22320456,
      "step": 34220
    },
    {
      "epoch": 17.93763102725367,
      "grad_norm": 1.1576389074325562,
      "learning_rate": 1.6029560452420238e-06,
      "loss": 0.0101,
      "num_input_tokens_seen": 22323048,
      "step": 34225
    },
    {
      "epoch": 17.940251572327043,
      "grad_norm": 0.2302757352590561,
      "learning_rate": 1.5989300309959377e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 22326568,
      "step": 34230
    },
    {
      "epoch": 17.94287211740042,
      "grad_norm": 0.3614889979362488,
      "learning_rate": 1.5949089119867116e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 22329864,
      "step": 34235
    },
    {
      "epoch": 17.945492662473793,
      "grad_norm": 1.023543357849121,
      "learning_rate": 1.5908926890555204e-06,
      "loss": 0.0079,
      "num_input_tokens_seen": 22333000,
      "step": 34240
    },
    {
      "epoch": 17.94811320754717,
      "grad_norm": 0.3363635540008545,
      "learning_rate": 1.5868813630425167e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 22336328,
      "step": 34245
    },
    {
      "epoch": 17.950733752620543,
      "grad_norm": 0.15913750231266022,
      "learning_rate": 1.5828749347868183e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 22338920,
      "step": 34250
    },
    {
      "epoch": 17.95335429769392,
      "grad_norm": 0.2852732539176941,
      "learning_rate": 1.5788734051265268e-06,
      "loss": 0.0467,
      "num_input_tokens_seen": 22342568,
      "step": 34255
    },
    {
      "epoch": 17.955974842767297,
      "grad_norm": 0.05063238739967346,
      "learning_rate": 1.5748767748987231e-06,
      "loss": 0.0192,
      "num_input_tokens_seen": 22345512,
      "step": 34260
    },
    {
      "epoch": 17.95859538784067,
      "grad_norm": 1.4061158895492554,
      "learning_rate": 1.5708850449394436e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 22348488,
      "step": 34265
    },
    {
      "epoch": 17.961215932914047,
      "grad_norm": 0.9642172455787659,
      "learning_rate": 1.5668982160837176e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 22351848,
      "step": 34270
    },
    {
      "epoch": 17.96383647798742,
      "grad_norm": 0.4702301025390625,
      "learning_rate": 1.562916289165553e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 22354856,
      "step": 34275
    },
    {
      "epoch": 17.966457023060798,
      "grad_norm": 0.3116982877254486,
      "learning_rate": 1.5589392650179114e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 22358152,
      "step": 34280
    },
    {
      "epoch": 17.96907756813417,
      "grad_norm": 1.0717085599899292,
      "learning_rate": 1.5549671444727437e-06,
      "loss": 0.0106,
      "num_input_tokens_seen": 22361704,
      "step": 34285
    },
    {
      "epoch": 17.971698113207548,
      "grad_norm": 0.9256333112716675,
      "learning_rate": 1.55099992836098e-06,
      "loss": 0.0164,
      "num_input_tokens_seen": 22364680,
      "step": 34290
    },
    {
      "epoch": 17.97431865828092,
      "grad_norm": 2.38000226020813,
      "learning_rate": 1.5470376175125061e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 22367752,
      "step": 34295
    },
    {
      "epoch": 17.9769392033543,
      "grad_norm": 0.13630153238773346,
      "learning_rate": 1.5430802127562038e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 22370632,
      "step": 34300
    },
    {
      "epoch": 17.979559748427672,
      "grad_norm": 2.5033092498779297,
      "learning_rate": 1.5391277149199023e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 22373224,
      "step": 34305
    },
    {
      "epoch": 17.98218029350105,
      "grad_norm": 0.11300334334373474,
      "learning_rate": 1.5351801248304292e-06,
      "loss": 0.0197,
      "num_input_tokens_seen": 22379816,
      "step": 34310
    },
    {
      "epoch": 17.984800838574422,
      "grad_norm": 0.646631121635437,
      "learning_rate": 1.5312374433135767e-06,
      "loss": 0.0027,
      "num_input_tokens_seen": 22383112,
      "step": 34315
    },
    {
      "epoch": 17.9874213836478,
      "grad_norm": 0.3077564835548401,
      "learning_rate": 1.5272996711941018e-06,
      "loss": 0.0115,
      "num_input_tokens_seen": 22386472,
      "step": 34320
    },
    {
      "epoch": 17.990041928721173,
      "grad_norm": 0.2111087292432785,
      "learning_rate": 1.5233668092957482e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 22388968,
      "step": 34325
    },
    {
      "epoch": 17.99266247379455,
      "grad_norm": 0.11174899339675903,
      "learning_rate": 1.5194388584412217e-06,
      "loss": 0.0027,
      "num_input_tokens_seen": 22392008,
      "step": 34330
    },
    {
      "epoch": 17.995283018867923,
      "grad_norm": 0.7447190880775452,
      "learning_rate": 1.5155158194522013e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 22394984,
      "step": 34335
    },
    {
      "epoch": 17.9979035639413,
      "grad_norm": 0.18005065619945526,
      "learning_rate": 1.5115976931493498e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 22398376,
      "step": 34340
    },
    {
      "epoch": 18.0,
      "eval_loss": 1.551309585571289,
      "eval_runtime": 11.8161,
      "eval_samples_per_second": 71.766,
      "eval_steps_per_second": 17.942,
      "num_input_tokens_seen": 22400368,
      "step": 34344
    },
    {
      "epoch": 18.000524109014677,
      "grad_norm": 0.3139896094799042,
      "learning_rate": 1.5076844803522922e-06,
      "loss": 0.0123,
      "num_input_tokens_seen": 22400816,
      "step": 34345
    },
    {
      "epoch": 18.00314465408805,
      "grad_norm": 0.5004844069480896,
      "learning_rate": 1.5037761818796236e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 22403888,
      "step": 34350
    },
    {
      "epoch": 18.005765199161427,
      "grad_norm": 0.9630014896392822,
      "learning_rate": 1.4998727985489235e-06,
      "loss": 0.0153,
      "num_input_tokens_seen": 22408208,
      "step": 34355
    },
    {
      "epoch": 18.0083857442348,
      "grad_norm": 0.29016658663749695,
      "learning_rate": 1.4959743311767249e-06,
      "loss": 0.0076,
      "num_input_tokens_seen": 22411856,
      "step": 34360
    },
    {
      "epoch": 18.011006289308177,
      "grad_norm": 0.3149995505809784,
      "learning_rate": 1.4920807805785502e-06,
      "loss": 0.0118,
      "num_input_tokens_seen": 22415248,
      "step": 34365
    },
    {
      "epoch": 18.01362683438155,
      "grad_norm": 0.15176182985305786,
      "learning_rate": 1.4881921475688815e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 22418416,
      "step": 34370
    },
    {
      "epoch": 18.016247379454928,
      "grad_norm": 1.7492330074310303,
      "learning_rate": 1.4843084329611844e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 22421456,
      "step": 34375
    },
    {
      "epoch": 18.0188679245283,
      "grad_norm": 0.05384744703769684,
      "learning_rate": 1.480429637567879e-06,
      "loss": 0.0024,
      "num_input_tokens_seen": 22424464,
      "step": 34380
    },
    {
      "epoch": 18.021488469601678,
      "grad_norm": 0.10477714985609055,
      "learning_rate": 1.4765557622003684e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 22427152,
      "step": 34385
    },
    {
      "epoch": 18.02410901467505,
      "grad_norm": 0.5062572360038757,
      "learning_rate": 1.4726868076690186e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 22431696,
      "step": 34390
    },
    {
      "epoch": 18.02672955974843,
      "grad_norm": 0.14888429641723633,
      "learning_rate": 1.4688227747831767e-06,
      "loss": 0.039,
      "num_input_tokens_seen": 22434704,
      "step": 34395
    },
    {
      "epoch": 18.029350104821802,
      "grad_norm": 0.3360211253166199,
      "learning_rate": 1.4649636643511573e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 22437776,
      "step": 34400
    },
    {
      "epoch": 18.03197064989518,
      "grad_norm": 0.26662394404411316,
      "learning_rate": 1.4611094771802364e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 22440912,
      "step": 34405
    },
    {
      "epoch": 18.034591194968552,
      "grad_norm": 0.4800783097743988,
      "learning_rate": 1.457260214076664e-06,
      "loss": 0.0129,
      "num_input_tokens_seen": 22444112,
      "step": 34410
    },
    {
      "epoch": 18.03721174004193,
      "grad_norm": 0.688169002532959,
      "learning_rate": 1.4534158758456623e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 22448080,
      "step": 34415
    },
    {
      "epoch": 18.039832285115303,
      "grad_norm": 1.0196620225906372,
      "learning_rate": 1.4495764632914271e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 22450800,
      "step": 34420
    },
    {
      "epoch": 18.04245283018868,
      "grad_norm": 0.14827735722064972,
      "learning_rate": 1.4457419772171215e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 22453648,
      "step": 34425
    },
    {
      "epoch": 18.045073375262053,
      "grad_norm": 0.9770633578300476,
      "learning_rate": 1.4419124184248733e-06,
      "loss": 0.0029,
      "num_input_tokens_seen": 22458480,
      "step": 34430
    },
    {
      "epoch": 18.04769392033543,
      "grad_norm": 0.9777001738548279,
      "learning_rate": 1.4380877877157834e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 22461840,
      "step": 34435
    },
    {
      "epoch": 18.050314465408807,
      "grad_norm": 0.6410273909568787,
      "learning_rate": 1.4342680858899143e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 22465104,
      "step": 34440
    },
    {
      "epoch": 18.05293501048218,
      "grad_norm": 0.18136417865753174,
      "learning_rate": 1.4304533137463134e-06,
      "loss": 0.0115,
      "num_input_tokens_seen": 22468976,
      "step": 34445
    },
    {
      "epoch": 18.055555555555557,
      "grad_norm": 0.28504428267478943,
      "learning_rate": 1.4266434720829808e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 22471888,
      "step": 34450
    },
    {
      "epoch": 18.05817610062893,
      "grad_norm": 1.3523701429367065,
      "learning_rate": 1.4228385616968987e-06,
      "loss": 0.0314,
      "num_input_tokens_seen": 22474640,
      "step": 34455
    },
    {
      "epoch": 18.060796645702307,
      "grad_norm": 1.5612207651138306,
      "learning_rate": 1.419038583384008e-06,
      "loss": 0.0099,
      "num_input_tokens_seen": 22477168,
      "step": 34460
    },
    {
      "epoch": 18.06341719077568,
      "grad_norm": 1.3353685140609741,
      "learning_rate": 1.4152435379392148e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 22480240,
      "step": 34465
    },
    {
      "epoch": 18.066037735849058,
      "grad_norm": 0.4205595552921295,
      "learning_rate": 1.411453426156406e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 22483344,
      "step": 34470
    },
    {
      "epoch": 18.06865828092243,
      "grad_norm": 0.5009838342666626,
      "learning_rate": 1.4076682488284282e-06,
      "loss": 0.0027,
      "num_input_tokens_seen": 22485840,
      "step": 34475
    },
    {
      "epoch": 18.071278825995808,
      "grad_norm": 0.9746367931365967,
      "learning_rate": 1.4038880067470978e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 22489392,
      "step": 34480
    },
    {
      "epoch": 18.07389937106918,
      "grad_norm": 0.11353243142366409,
      "learning_rate": 1.400112700703196e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 22491696,
      "step": 34485
    },
    {
      "epoch": 18.07651991614256,
      "grad_norm": 0.18711906671524048,
      "learning_rate": 1.3963423314864805e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 22494448,
      "step": 34490
    },
    {
      "epoch": 18.079140461215932,
      "grad_norm": 0.6360182762145996,
      "learning_rate": 1.3925768998856587e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 22497264,
      "step": 34495
    },
    {
      "epoch": 18.08176100628931,
      "grad_norm": 0.19708798825740814,
      "learning_rate": 1.388816406688423e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 22500080,
      "step": 34500
    },
    {
      "epoch": 18.084381551362682,
      "grad_norm": 0.2798057198524475,
      "learning_rate": 1.3850608526814246e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 22503152,
      "step": 34505
    },
    {
      "epoch": 18.08700209643606,
      "grad_norm": 0.3390401601791382,
      "learning_rate": 1.3813102386502853e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 22506288,
      "step": 34510
    },
    {
      "epoch": 18.089622641509433,
      "grad_norm": 0.7069648504257202,
      "learning_rate": 1.3775645653795881e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 22508496,
      "step": 34515
    },
    {
      "epoch": 18.09224318658281,
      "grad_norm": 1.2353324890136719,
      "learning_rate": 1.373823833652882e-06,
      "loss": 0.0187,
      "num_input_tokens_seen": 22512496,
      "step": 34520
    },
    {
      "epoch": 18.094863731656183,
      "grad_norm": 0.8220340013504028,
      "learning_rate": 1.3700880442526876e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 22515888,
      "step": 34525
    },
    {
      "epoch": 18.09748427672956,
      "grad_norm": 0.1870029717683792,
      "learning_rate": 1.3663571979604884e-06,
      "loss": 0.0029,
      "num_input_tokens_seen": 22518928,
      "step": 34530
    },
    {
      "epoch": 18.100104821802937,
      "grad_norm": 1.5434991121292114,
      "learning_rate": 1.3626312955567406e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 22523120,
      "step": 34535
    },
    {
      "epoch": 18.10272536687631,
      "grad_norm": 0.193222314119339,
      "learning_rate": 1.358910337820854e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 22525392,
      "step": 34540
    },
    {
      "epoch": 18.105345911949687,
      "grad_norm": 0.36823999881744385,
      "learning_rate": 1.3551943255312172e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 22527920,
      "step": 34545
    },
    {
      "epoch": 18.10796645702306,
      "grad_norm": 0.4599471092224121,
      "learning_rate": 1.3514832594651694e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 22531472,
      "step": 34550
    },
    {
      "epoch": 18.110587002096437,
      "grad_norm": 0.22845028340816498,
      "learning_rate": 1.347777140399026e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 22534672,
      "step": 34555
    },
    {
      "epoch": 18.11320754716981,
      "grad_norm": 0.17232076823711395,
      "learning_rate": 1.344075969108069e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 22537424,
      "step": 34560
    },
    {
      "epoch": 18.115828092243188,
      "grad_norm": 1.1920158863067627,
      "learning_rate": 1.3403797463665352e-06,
      "loss": 0.0094,
      "num_input_tokens_seen": 22540944,
      "step": 34565
    },
    {
      "epoch": 18.11844863731656,
      "grad_norm": 0.3548569977283478,
      "learning_rate": 1.3366884729476365e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 22544368,
      "step": 34570
    },
    {
      "epoch": 18.121069182389938,
      "grad_norm": 0.2101547122001648,
      "learning_rate": 1.3330021496235467e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 22547568,
      "step": 34575
    },
    {
      "epoch": 18.12368972746331,
      "grad_norm": 0.329196035861969,
      "learning_rate": 1.3293207771653965e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 22551504,
      "step": 34580
    },
    {
      "epoch": 18.12631027253669,
      "grad_norm": 0.2985304594039917,
      "learning_rate": 1.3256443563432891e-06,
      "loss": 0.0022,
      "num_input_tokens_seen": 22554416,
      "step": 34585
    },
    {
      "epoch": 18.128930817610062,
      "grad_norm": 0.14823344349861145,
      "learning_rate": 1.3219728879262956e-06,
      "loss": 0.0013,
      "num_input_tokens_seen": 22557232,
      "step": 34590
    },
    {
      "epoch": 18.13155136268344,
      "grad_norm": 0.3414294719696045,
      "learning_rate": 1.3183063726824375e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 22560976,
      "step": 34595
    },
    {
      "epoch": 18.134171907756812,
      "grad_norm": 0.7781268358230591,
      "learning_rate": 1.3146448113787151e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 22564272,
      "step": 34600
    },
    {
      "epoch": 18.13679245283019,
      "grad_norm": 2.6251683235168457,
      "learning_rate": 1.3109882047810773e-06,
      "loss": 0.097,
      "num_input_tokens_seen": 22567280,
      "step": 34605
    },
    {
      "epoch": 18.139412997903563,
      "grad_norm": 0.6335629224777222,
      "learning_rate": 1.307336553654448e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 22571120,
      "step": 34610
    },
    {
      "epoch": 18.14203354297694,
      "grad_norm": 0.58203125,
      "learning_rate": 1.303689858762716e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 22574672,
      "step": 34615
    },
    {
      "epoch": 18.144654088050313,
      "grad_norm": 0.35775408148765564,
      "learning_rate": 1.3000481208687188e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 22577616,
      "step": 34620
    },
    {
      "epoch": 18.14727463312369,
      "grad_norm": 0.48837119340896606,
      "learning_rate": 1.2964113407342743e-06,
      "loss": 0.0176,
      "num_input_tokens_seen": 22579888,
      "step": 34625
    },
    {
      "epoch": 18.149895178197063,
      "grad_norm": 0.20790739357471466,
      "learning_rate": 1.292779519120152e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 22582896,
      "step": 34630
    },
    {
      "epoch": 18.15251572327044,
      "grad_norm": 0.15081964433193207,
      "learning_rate": 1.2891526567860856e-06,
      "loss": 0.0024,
      "num_input_tokens_seen": 22585616,
      "step": 34635
    },
    {
      "epoch": 18.155136268343817,
      "grad_norm": 0.2273818850517273,
      "learning_rate": 1.2855307544907825e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 22588496,
      "step": 34640
    },
    {
      "epoch": 18.15775681341719,
      "grad_norm": 0.6343610882759094,
      "learning_rate": 1.281913812991889e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 22590896,
      "step": 34645
    },
    {
      "epoch": 18.160377358490567,
      "grad_norm": 0.23999032378196716,
      "learning_rate": 1.2783018330460389e-06,
      "loss": 0.1172,
      "num_input_tokens_seen": 22593104,
      "step": 34650
    },
    {
      "epoch": 18.16299790356394,
      "grad_norm": 0.3285539448261261,
      "learning_rate": 1.2746948154088111e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 22596656,
      "step": 34655
    },
    {
      "epoch": 18.165618448637318,
      "grad_norm": 0.15351954102516174,
      "learning_rate": 1.2710927608347605e-06,
      "loss": 0.0113,
      "num_input_tokens_seen": 22599312,
      "step": 34660
    },
    {
      "epoch": 18.16823899371069,
      "grad_norm": 0.5283377170562744,
      "learning_rate": 1.2674956700773872e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 22603440,
      "step": 34665
    },
    {
      "epoch": 18.170859538784068,
      "grad_norm": 0.48126208782196045,
      "learning_rate": 1.2639035438891666e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 22606512,
      "step": 34670
    },
    {
      "epoch": 18.17348008385744,
      "grad_norm": 1.3576443195343018,
      "learning_rate": 1.260316383021526e-06,
      "loss": 0.0145,
      "num_input_tokens_seen": 22609584,
      "step": 34675
    },
    {
      "epoch": 18.17610062893082,
      "grad_norm": 0.8491919040679932,
      "learning_rate": 1.2567341882248645e-06,
      "loss": 0.0176,
      "num_input_tokens_seen": 22613712,
      "step": 34680
    },
    {
      "epoch": 18.178721174004192,
      "grad_norm": 0.2896917462348938,
      "learning_rate": 1.2531569602485305e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 22617008,
      "step": 34685
    },
    {
      "epoch": 18.18134171907757,
      "grad_norm": 0.07221490144729614,
      "learning_rate": 1.2495846998408445e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 22620112,
      "step": 34690
    },
    {
      "epoch": 18.183962264150942,
      "grad_norm": 0.3660907447338104,
      "learning_rate": 1.246017407749084e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 22622608,
      "step": 34695
    },
    {
      "epoch": 18.18658280922432,
      "grad_norm": 0.5590631365776062,
      "learning_rate": 1.2424550847194739e-06,
      "loss": 0.0225,
      "num_input_tokens_seen": 22626416,
      "step": 34700
    },
    {
      "epoch": 18.189203354297693,
      "grad_norm": 0.42574480175971985,
      "learning_rate": 1.2388977314972238e-06,
      "loss": 0.0022,
      "num_input_tokens_seen": 22629328,
      "step": 34705
    },
    {
      "epoch": 18.19182389937107,
      "grad_norm": 0.16041059792041779,
      "learning_rate": 1.235345348826486e-06,
      "loss": 0.0517,
      "num_input_tokens_seen": 22632144,
      "step": 34710
    },
    {
      "epoch": 18.194444444444443,
      "grad_norm": 2.9442739486694336,
      "learning_rate": 1.2317979374503825e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 22634992,
      "step": 34715
    },
    {
      "epoch": 18.19706498951782,
      "grad_norm": 0.8454090356826782,
      "learning_rate": 1.228255498110989e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 22638256,
      "step": 34720
    },
    {
      "epoch": 18.199685534591197,
      "grad_norm": 1.0081558227539062,
      "learning_rate": 1.224718031549338e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 22641712,
      "step": 34725
    },
    {
      "epoch": 18.20230607966457,
      "grad_norm": 0.49881479144096375,
      "learning_rate": 1.2211855385054344e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 22645904,
      "step": 34730
    },
    {
      "epoch": 18.204926624737947,
      "grad_norm": 0.5074368119239807,
      "learning_rate": 1.217658019718232e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 22648944,
      "step": 34735
    },
    {
      "epoch": 18.20754716981132,
      "grad_norm": 0.17503401637077332,
      "learning_rate": 1.2141354759256484e-06,
      "loss": 0.0091,
      "num_input_tokens_seen": 22651888,
      "step": 34740
    },
    {
      "epoch": 18.210167714884697,
      "grad_norm": 0.15079191327095032,
      "learning_rate": 1.2106179078645641e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 22655120,
      "step": 34745
    },
    {
      "epoch": 18.21278825995807,
      "grad_norm": 0.5276674032211304,
      "learning_rate": 1.2071053162708096e-06,
      "loss": 0.046,
      "num_input_tokens_seen": 22658928,
      "step": 34750
    },
    {
      "epoch": 18.215408805031448,
      "grad_norm": 0.4475869834423065,
      "learning_rate": 1.203597701879175e-06,
      "loss": 0.0312,
      "num_input_tokens_seen": 22662288,
      "step": 34755
    },
    {
      "epoch": 18.21802935010482,
      "grad_norm": 0.44797977805137634,
      "learning_rate": 1.2000950654234177e-06,
      "loss": 0.0024,
      "num_input_tokens_seen": 22664848,
      "step": 34760
    },
    {
      "epoch": 18.220649895178198,
      "grad_norm": 0.2787492871284485,
      "learning_rate": 1.1965974076362491e-06,
      "loss": 0.0018,
      "num_input_tokens_seen": 22667376,
      "step": 34765
    },
    {
      "epoch": 18.22327044025157,
      "grad_norm": 0.8850340247154236,
      "learning_rate": 1.193104729249342e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 22671440,
      "step": 34770
    },
    {
      "epoch": 18.22589098532495,
      "grad_norm": 0.10346444696187973,
      "learning_rate": 1.189617030993323e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 22675408,
      "step": 34775
    },
    {
      "epoch": 18.228511530398322,
      "grad_norm": 0.9364968538284302,
      "learning_rate": 1.1861343135977753e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 22678224,
      "step": 34780
    },
    {
      "epoch": 18.2311320754717,
      "grad_norm": 0.3903258442878723,
      "learning_rate": 1.182656577791244e-06,
      "loss": 0.0118,
      "num_input_tokens_seen": 22681712,
      "step": 34785
    },
    {
      "epoch": 18.233752620545072,
      "grad_norm": 0.13868562877178192,
      "learning_rate": 1.1791838243012359e-06,
      "loss": 0.001,
      "num_input_tokens_seen": 22685008,
      "step": 34790
    },
    {
      "epoch": 18.23637316561845,
      "grad_norm": 0.5423795580863953,
      "learning_rate": 1.1757160538542117e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 22688368,
      "step": 34795
    },
    {
      "epoch": 18.238993710691823,
      "grad_norm": 0.5097355246543884,
      "learning_rate": 1.1722532671755882e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 22691184,
      "step": 34800
    },
    {
      "epoch": 18.2416142557652,
      "grad_norm": 0.5627036690711975,
      "learning_rate": 1.168795464989736e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 22695024,
      "step": 34805
    },
    {
      "epoch": 18.244234800838573,
      "grad_norm": 0.862285315990448,
      "learning_rate": 1.165342648019993e-06,
      "loss": 0.0105,
      "num_input_tokens_seen": 22699088,
      "step": 34810
    },
    {
      "epoch": 18.24685534591195,
      "grad_norm": 0.3720567226409912,
      "learning_rate": 1.1618948169886457e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 22702480,
      "step": 34815
    },
    {
      "epoch": 18.249475890985323,
      "grad_norm": 1.8588134050369263,
      "learning_rate": 1.1584519726169446e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 22705840,
      "step": 34820
    },
    {
      "epoch": 18.2520964360587,
      "grad_norm": 0.5388926863670349,
      "learning_rate": 1.1550141156250942e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 22708752,
      "step": 34825
    },
    {
      "epoch": 18.254716981132077,
      "grad_norm": 3.0489113330841064,
      "learning_rate": 1.1515812467322556e-06,
      "loss": 0.0196,
      "num_input_tokens_seen": 22711504,
      "step": 34830
    },
    {
      "epoch": 18.25733752620545,
      "grad_norm": 0.9564632177352905,
      "learning_rate": 1.1481533666565375e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 22714384,
      "step": 34835
    },
    {
      "epoch": 18.259958071278827,
      "grad_norm": 0.8699585795402527,
      "learning_rate": 1.144730476115019e-06,
      "loss": 0.016,
      "num_input_tokens_seen": 22718192,
      "step": 34840
    },
    {
      "epoch": 18.2625786163522,
      "grad_norm": 0.3945006728172302,
      "learning_rate": 1.1413125758237304e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 22721616,
      "step": 34845
    },
    {
      "epoch": 18.265199161425578,
      "grad_norm": 0.5013428926467896,
      "learning_rate": 1.1378996664976577e-06,
      "loss": 0.0103,
      "num_input_tokens_seen": 22724624,
      "step": 34850
    },
    {
      "epoch": 18.26781970649895,
      "grad_norm": 0.7226726412773132,
      "learning_rate": 1.134491748850741e-06,
      "loss": 0.0264,
      "num_input_tokens_seen": 22727600,
      "step": 34855
    },
    {
      "epoch": 18.270440251572328,
      "grad_norm": 0.1551104038953781,
      "learning_rate": 1.13108882359588e-06,
      "loss": 0.0019,
      "num_input_tokens_seen": 22730608,
      "step": 34860
    },
    {
      "epoch": 18.2730607966457,
      "grad_norm": 0.5126400589942932,
      "learning_rate": 1.127690891444924e-06,
      "loss": 0.0097,
      "num_input_tokens_seen": 22733424,
      "step": 34865
    },
    {
      "epoch": 18.27568134171908,
      "grad_norm": 0.517725944519043,
      "learning_rate": 1.1242979531086829e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 22737008,
      "step": 34870
    },
    {
      "epoch": 18.278301886792452,
      "grad_norm": 0.4696555435657501,
      "learning_rate": 1.1209100092969244e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 22740272,
      "step": 34875
    },
    {
      "epoch": 18.28092243186583,
      "grad_norm": 0.8332765698432922,
      "learning_rate": 1.1175270607183625e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 22744464,
      "step": 34880
    },
    {
      "epoch": 18.283542976939202,
      "grad_norm": 0.38878417015075684,
      "learning_rate": 1.114149108080678e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 22747152,
      "step": 34885
    },
    {
      "epoch": 18.28616352201258,
      "grad_norm": 0.6923092603683472,
      "learning_rate": 1.1107761520904925e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 22751408,
      "step": 34890
    },
    {
      "epoch": 18.288784067085953,
      "grad_norm": 0.4010806381702423,
      "learning_rate": 1.1074081934533908e-06,
      "loss": 0.0132,
      "num_input_tokens_seen": 22755568,
      "step": 34895
    },
    {
      "epoch": 18.29140461215933,
      "grad_norm": 0.9869435429573059,
      "learning_rate": 1.1040452328739153e-06,
      "loss": 0.007,
      "num_input_tokens_seen": 22758576,
      "step": 34900
    },
    {
      "epoch": 18.294025157232703,
      "grad_norm": 0.404784619808197,
      "learning_rate": 1.1006872710555615e-06,
      "loss": 0.0129,
      "num_input_tokens_seen": 22762128,
      "step": 34905
    },
    {
      "epoch": 18.29664570230608,
      "grad_norm": 0.3330172002315521,
      "learning_rate": 1.0973343087007703e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 22764784,
      "step": 34910
    },
    {
      "epoch": 18.299266247379457,
      "grad_norm": 0.0544254295527935,
      "learning_rate": 1.0939863465109473e-06,
      "loss": 0.0133,
      "num_input_tokens_seen": 22768304,
      "step": 34915
    },
    {
      "epoch": 18.30188679245283,
      "grad_norm": 0.42908501625061035,
      "learning_rate": 1.0906433851864433e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 22771632,
      "step": 34920
    },
    {
      "epoch": 18.304507337526207,
      "grad_norm": 0.39624637365341187,
      "learning_rate": 1.0873054254265713e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 22774928,
      "step": 34925
    },
    {
      "epoch": 18.30712788259958,
      "grad_norm": 0.3503292202949524,
      "learning_rate": 1.0839724679295976e-06,
      "loss": 0.0027,
      "num_input_tokens_seen": 22778640,
      "step": 34930
    },
    {
      "epoch": 18.309748427672957,
      "grad_norm": 1.0851678848266602,
      "learning_rate": 1.0806445133927311e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 22781936,
      "step": 34935
    },
    {
      "epoch": 18.31236897274633,
      "grad_norm": 0.13210122287273407,
      "learning_rate": 1.0773215625121457e-06,
      "loss": 0.0025,
      "num_input_tokens_seen": 22784528,
      "step": 34940
    },
    {
      "epoch": 18.314989517819708,
      "grad_norm": 1.1204094886779785,
      "learning_rate": 1.0740036159829715e-06,
      "loss": 0.0144,
      "num_input_tokens_seen": 22790704,
      "step": 34945
    },
    {
      "epoch": 18.31761006289308,
      "grad_norm": 2.811542510986328,
      "learning_rate": 1.0706906744992724e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 22794320,
      "step": 34950
    },
    {
      "epoch": 18.320230607966458,
      "grad_norm": 0.7315271496772766,
      "learning_rate": 1.0673827387540857e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 22798544,
      "step": 34955
    },
    {
      "epoch": 18.32285115303983,
      "grad_norm": 0.1308407038450241,
      "learning_rate": 1.0640798094393916e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 22801392,
      "step": 34960
    },
    {
      "epoch": 18.32547169811321,
      "grad_norm": 2.8668391704559326,
      "learning_rate": 1.060781887246126e-06,
      "loss": 0.0255,
      "num_input_tokens_seen": 22804880,
      "step": 34965
    },
    {
      "epoch": 18.328092243186582,
      "grad_norm": 1.6580861806869507,
      "learning_rate": 1.057488972864179e-06,
      "loss": 0.011,
      "num_input_tokens_seen": 22808112,
      "step": 34970
    },
    {
      "epoch": 18.33071278825996,
      "grad_norm": 0.4589141309261322,
      "learning_rate": 1.0542010669823855e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 22814544,
      "step": 34975
    },
    {
      "epoch": 18.333333333333332,
      "grad_norm": 0.3995929956436157,
      "learning_rate": 1.0509181702885395e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 22817200,
      "step": 34980
    },
    {
      "epoch": 18.33595387840671,
      "grad_norm": 0.36502134799957275,
      "learning_rate": 1.047640283469392e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 22820048,
      "step": 34985
    },
    {
      "epoch": 18.338574423480082,
      "grad_norm": 0.358593612909317,
      "learning_rate": 1.0443674072106308e-06,
      "loss": 0.0023,
      "num_input_tokens_seen": 22823280,
      "step": 34990
    },
    {
      "epoch": 18.34119496855346,
      "grad_norm": 0.4562133848667145,
      "learning_rate": 1.041099542196905e-06,
      "loss": 0.0027,
      "num_input_tokens_seen": 22826224,
      "step": 34995
    },
    {
      "epoch": 18.343815513626833,
      "grad_norm": 0.09055852890014648,
      "learning_rate": 1.0378366891118234e-06,
      "loss": 0.0015,
      "num_input_tokens_seen": 22829360,
      "step": 35000
    },
    {
      "epoch": 18.34643605870021,
      "grad_norm": 0.23166286945343018,
      "learning_rate": 1.0345788486379265e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 22832688,
      "step": 35005
    },
    {
      "epoch": 18.349056603773583,
      "grad_norm": 0.23627056181430817,
      "learning_rate": 1.0313260214567272e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 22836176,
      "step": 35010
    },
    {
      "epoch": 18.35167714884696,
      "grad_norm": 0.1735783964395523,
      "learning_rate": 1.0280782082486729e-06,
      "loss": 0.0026,
      "num_input_tokens_seen": 22839088,
      "step": 35015
    },
    {
      "epoch": 18.354297693920337,
      "grad_norm": 3.0885446071624756,
      "learning_rate": 1.02483540969317e-06,
      "loss": 0.0026,
      "num_input_tokens_seen": 22842384,
      "step": 35020
    },
    {
      "epoch": 18.35691823899371,
      "grad_norm": 0.4780610203742981,
      "learning_rate": 1.0215976264685763e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 22845232,
      "step": 35025
    },
    {
      "epoch": 18.359538784067087,
      "grad_norm": 0.27425000071525574,
      "learning_rate": 1.0183648592522021e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 22848016,
      "step": 35030
    },
    {
      "epoch": 18.36215932914046,
      "grad_norm": 0.4933680593967438,
      "learning_rate": 1.0151371087203015e-06,
      "loss": 0.0179,
      "num_input_tokens_seen": 22850672,
      "step": 35035
    },
    {
      "epoch": 18.364779874213838,
      "grad_norm": 0.22428734600543976,
      "learning_rate": 1.0119143755480815e-06,
      "loss": 0.0024,
      "num_input_tokens_seen": 22853584,
      "step": 35040
    },
    {
      "epoch": 18.36740041928721,
      "grad_norm": 0.6918540000915527,
      "learning_rate": 1.0086966604097026e-06,
      "loss": 0.0089,
      "num_input_tokens_seen": 22857424,
      "step": 35045
    },
    {
      "epoch": 18.370020964360588,
      "grad_norm": 0.4174116551876068,
      "learning_rate": 1.0054839639782766e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 22861456,
      "step": 35050
    },
    {
      "epoch": 18.37264150943396,
      "grad_norm": 0.3310222327709198,
      "learning_rate": 1.00227628692586e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 22864752,
      "step": 35055
    },
    {
      "epoch": 18.37526205450734,
      "grad_norm": 0.32798099517822266,
      "learning_rate": 9.990736299234692e-07,
      "loss": 0.0037,
      "num_input_tokens_seen": 22868624,
      "step": 35060
    },
    {
      "epoch": 18.377882599580712,
      "grad_norm": 0.10563766956329346,
      "learning_rate": 9.958759936410573e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 22871760,
      "step": 35065
    },
    {
      "epoch": 18.38050314465409,
      "grad_norm": 0.20540645718574524,
      "learning_rate": 9.926833787475304e-07,
      "loss": 0.0066,
      "num_input_tokens_seen": 22875248,
      "step": 35070
    },
    {
      "epoch": 18.383123689727462,
      "grad_norm": 0.537834882736206,
      "learning_rate": 9.89495785910749e-07,
      "loss": 0.0037,
      "num_input_tokens_seen": 22878640,
      "step": 35075
    },
    {
      "epoch": 18.38574423480084,
      "grad_norm": 0.7511901259422302,
      "learning_rate": 9.86313215797524e-07,
      "loss": 0.0024,
      "num_input_tokens_seen": 22881584,
      "step": 35080
    },
    {
      "epoch": 18.388364779874212,
      "grad_norm": 0.11047112941741943,
      "learning_rate": 9.831356690736149e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 22884368,
      "step": 35085
    },
    {
      "epoch": 18.39098532494759,
      "grad_norm": 4.3229169845581055,
      "learning_rate": 9.799631464037256e-07,
      "loss": 0.0026,
      "num_input_tokens_seen": 22886896,
      "step": 35090
    },
    {
      "epoch": 18.393605870020963,
      "grad_norm": 3.984623670578003,
      "learning_rate": 9.76795648451509e-07,
      "loss": 0.0075,
      "num_input_tokens_seen": 22890448,
      "step": 35095
    },
    {
      "epoch": 18.39622641509434,
      "grad_norm": 1.362396240234375,
      "learning_rate": 9.73633175879568e-07,
      "loss": 0.0097,
      "num_input_tokens_seen": 22894224,
      "step": 35100
    },
    {
      "epoch": 18.398846960167713,
      "grad_norm": 0.7750764489173889,
      "learning_rate": 9.704757293494627e-07,
      "loss": 0.0055,
      "num_input_tokens_seen": 22897680,
      "step": 35105
    },
    {
      "epoch": 18.40146750524109,
      "grad_norm": 0.1537904292345047,
      "learning_rate": 9.673233095216922e-07,
      "loss": 0.0058,
      "num_input_tokens_seen": 22900880,
      "step": 35110
    },
    {
      "epoch": 18.404088050314467,
      "grad_norm": 0.3104638457298279,
      "learning_rate": 9.641759170557068e-07,
      "loss": 0.0123,
      "num_input_tokens_seen": 22903984,
      "step": 35115
    },
    {
      "epoch": 18.40670859538784,
      "grad_norm": 0.5424119830131531,
      "learning_rate": 9.610335526099024e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 22907216,
      "step": 35120
    },
    {
      "epoch": 18.409329140461217,
      "grad_norm": 0.17728587985038757,
      "learning_rate": 9.578962168416278e-07,
      "loss": 0.0063,
      "num_input_tokens_seen": 22910352,
      "step": 35125
    },
    {
      "epoch": 18.41194968553459,
      "grad_norm": 0.21701078116893768,
      "learning_rate": 9.547639104071775e-07,
      "loss": 0.0193,
      "num_input_tokens_seen": 22913776,
      "step": 35130
    },
    {
      "epoch": 18.414570230607968,
      "grad_norm": 0.5162636637687683,
      "learning_rate": 9.516366339617944e-07,
      "loss": 0.0444,
      "num_input_tokens_seen": 22917584,
      "step": 35135
    },
    {
      "epoch": 18.41719077568134,
      "grad_norm": 0.1986401081085205,
      "learning_rate": 9.485143881596714e-07,
      "loss": 0.0061,
      "num_input_tokens_seen": 22920592,
      "step": 35140
    },
    {
      "epoch": 18.419811320754718,
      "grad_norm": 0.2699951231479645,
      "learning_rate": 9.453971736539447e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 22923248,
      "step": 35145
    },
    {
      "epoch": 18.42243186582809,
      "grad_norm": 0.29060232639312744,
      "learning_rate": 9.42284991096698e-07,
      "loss": 0.0031,
      "num_input_tokens_seen": 22926832,
      "step": 35150
    },
    {
      "epoch": 18.42505241090147,
      "grad_norm": 0.30436012148857117,
      "learning_rate": 9.391778411389634e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 22930256,
      "step": 35155
    },
    {
      "epoch": 18.427672955974842,
      "grad_norm": 0.28200221061706543,
      "learning_rate": 9.360757244307267e-07,
      "loss": 0.0746,
      "num_input_tokens_seen": 22933552,
      "step": 35160
    },
    {
      "epoch": 18.43029350104822,
      "grad_norm": 0.1489429473876953,
      "learning_rate": 9.329786416209157e-07,
      "loss": 0.004,
      "num_input_tokens_seen": 22936112,
      "step": 35165
    },
    {
      "epoch": 18.432914046121592,
      "grad_norm": 1.039015769958496,
      "learning_rate": 9.298865933574014e-07,
      "loss": 0.0057,
      "num_input_tokens_seen": 22939376,
      "step": 35170
    },
    {
      "epoch": 18.43553459119497,
      "grad_norm": 0.23047813773155212,
      "learning_rate": 9.267995802870022e-07,
      "loss": 0.0028,
      "num_input_tokens_seen": 22942416,
      "step": 35175
    },
    {
      "epoch": 18.438155136268342,
      "grad_norm": 1.2202961444854736,
      "learning_rate": 9.237176030554878e-07,
      "loss": 0.0149,
      "num_input_tokens_seen": 22945520,
      "step": 35180
    },
    {
      "epoch": 18.44077568134172,
      "grad_norm": 0.2194661647081375,
      "learning_rate": 9.206406623075758e-07,
      "loss": 0.0582,
      "num_input_tokens_seen": 22948240,
      "step": 35185
    },
    {
      "epoch": 18.443396226415093,
      "grad_norm": 0.169776052236557,
      "learning_rate": 9.175687586869292e-07,
      "loss": 0.0059,
      "num_input_tokens_seen": 22951536,
      "step": 35190
    },
    {
      "epoch": 18.44601677148847,
      "grad_norm": 3.6988000869750977,
      "learning_rate": 9.145018928361476e-07,
      "loss": 0.0072,
      "num_input_tokens_seen": 22955280,
      "step": 35195
    },
    {
      "epoch": 18.448637316561843,
      "grad_norm": 0.443694144487381,
      "learning_rate": 9.114400653967903e-07,
      "loss": 0.0057,
      "num_input_tokens_seen": 22958576,
      "step": 35200
    },
    {
      "epoch": 18.45125786163522,
      "grad_norm": 0.616320788860321,
      "learning_rate": 9.083832770093531e-07,
      "loss": 0.0065,
      "num_input_tokens_seen": 22961328,
      "step": 35205
    },
    {
      "epoch": 18.453878406708597,
      "grad_norm": 0.3521451950073242,
      "learning_rate": 9.053315283132802e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 22964880,
      "step": 35210
    },
    {
      "epoch": 18.45649895178197,
      "grad_norm": 0.11287438869476318,
      "learning_rate": 9.022848199469663e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 22968848,
      "step": 35215
    },
    {
      "epoch": 18.459119496855347,
      "grad_norm": 0.7263496518135071,
      "learning_rate": 8.992431525477518e-07,
      "loss": 0.0184,
      "num_input_tokens_seen": 22972944,
      "step": 35220
    },
    {
      "epoch": 18.46174004192872,
      "grad_norm": 0.31066372990608215,
      "learning_rate": 8.962065267519082e-07,
      "loss": 0.0035,
      "num_input_tokens_seen": 22976208,
      "step": 35225
    },
    {
      "epoch": 18.464360587002098,
      "grad_norm": 0.3262045085430145,
      "learning_rate": 8.931749431946718e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 22978832,
      "step": 35230
    },
    {
      "epoch": 18.46698113207547,
      "grad_norm": 0.2645035684108734,
      "learning_rate": 8.901484025102103e-07,
      "loss": 0.0034,
      "num_input_tokens_seen": 22982096,
      "step": 35235
    },
    {
      "epoch": 18.469601677148848,
      "grad_norm": 0.5415165424346924,
      "learning_rate": 8.871269053316422e-07,
      "loss": 0.0357,
      "num_input_tokens_seen": 22986032,
      "step": 35240
    },
    {
      "epoch": 18.47222222222222,
      "grad_norm": 0.36184388399124146,
      "learning_rate": 8.841104522910343e-07,
      "loss": 0.0041,
      "num_input_tokens_seen": 22989168,
      "step": 35245
    },
    {
      "epoch": 18.4748427672956,
      "grad_norm": 0.1254987269639969,
      "learning_rate": 8.8109904401939e-07,
      "loss": 0.0102,
      "num_input_tokens_seen": 22992784,
      "step": 35250
    },
    {
      "epoch": 18.47746331236897,
      "grad_norm": 0.7821846008300781,
      "learning_rate": 8.780926811466639e-07,
      "loss": 0.0031,
      "num_input_tokens_seen": 22995824,
      "step": 35255
    },
    {
      "epoch": 18.48008385744235,
      "grad_norm": 0.1272430568933487,
      "learning_rate": 8.7509136430175e-07,
      "loss": 0.0068,
      "num_input_tokens_seen": 22998928,
      "step": 35260
    },
    {
      "epoch": 18.482704402515722,
      "grad_norm": 0.81815105676651,
      "learning_rate": 8.720950941124877e-07,
      "loss": 0.0049,
      "num_input_tokens_seen": 23001424,
      "step": 35265
    },
    {
      "epoch": 18.4853249475891,
      "grad_norm": 0.476151704788208,
      "learning_rate": 8.691038712056732e-07,
      "loss": 0.0043,
      "num_input_tokens_seen": 23004784,
      "step": 35270
    },
    {
      "epoch": 18.487945492662472,
      "grad_norm": 0.312792032957077,
      "learning_rate": 8.66117696207025e-07,
      "loss": 0.0069,
      "num_input_tokens_seen": 23008048,
      "step": 35275
    },
    {
      "epoch": 18.49056603773585,
      "grad_norm": 0.3362555503845215,
      "learning_rate": 8.631365697412186e-07,
      "loss": 0.0019,
      "num_input_tokens_seen": 23010416,
      "step": 35280
    },
    {
      "epoch": 18.493186582809223,
      "grad_norm": 0.36623212695121765,
      "learning_rate": 8.601604924318802e-07,
      "loss": 0.0058,
      "num_input_tokens_seen": 23013616,
      "step": 35285
    },
    {
      "epoch": 18.4958071278826,
      "grad_norm": 0.7485036849975586,
      "learning_rate": 8.571894649015588e-07,
      "loss": 0.0119,
      "num_input_tokens_seen": 23017808,
      "step": 35290
    },
    {
      "epoch": 18.498427672955973,
      "grad_norm": 0.36228305101394653,
      "learning_rate": 8.542234877717686e-07,
      "loss": 0.0018,
      "num_input_tokens_seen": 23020720,
      "step": 35295
    },
    {
      "epoch": 18.50104821802935,
      "grad_norm": 0.17903897166252136,
      "learning_rate": 8.512625616629549e-07,
      "loss": 0.0018,
      "num_input_tokens_seen": 23024560,
      "step": 35300
    },
    {
      "epoch": 18.503668763102727,
      "grad_norm": 0.5465940237045288,
      "learning_rate": 8.483066871945083e-07,
      "loss": 0.0038,
      "num_input_tokens_seen": 23027856,
      "step": 35305
    },
    {
      "epoch": 18.5062893081761,
      "grad_norm": 0.20119629800319672,
      "learning_rate": 8.453558649847676e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 23031024,
      "step": 35310
    },
    {
      "epoch": 18.508909853249477,
      "grad_norm": 0.1800525039434433,
      "learning_rate": 8.424100956510112e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 23034192,
      "step": 35315
    },
    {
      "epoch": 18.51153039832285,
      "grad_norm": 0.2775696814060211,
      "learning_rate": 8.394693798094544e-07,
      "loss": 0.0147,
      "num_input_tokens_seen": 23037680,
      "step": 35320
    },
    {
      "epoch": 18.514150943396228,
      "grad_norm": 0.6922610998153687,
      "learning_rate": 8.365337180752691e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 23040784,
      "step": 35325
    },
    {
      "epoch": 18.5167714884696,
      "grad_norm": 0.7760804891586304,
      "learning_rate": 8.336031110625558e-07,
      "loss": 0.0061,
      "num_input_tokens_seen": 23043888,
      "step": 35330
    },
    {
      "epoch": 18.519392033542978,
      "grad_norm": 0.5387576818466187,
      "learning_rate": 8.306775593843657e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 23046928,
      "step": 35335
    },
    {
      "epoch": 18.52201257861635,
      "grad_norm": 0.6347417235374451,
      "learning_rate": 8.277570636526955e-07,
      "loss": 0.0037,
      "num_input_tokens_seen": 23052016,
      "step": 35340
    },
    {
      "epoch": 18.52463312368973,
      "grad_norm": 0.6731374859809875,
      "learning_rate": 8.24841624478473e-07,
      "loss": 0.0057,
      "num_input_tokens_seen": 23055632,
      "step": 35345
    },
    {
      "epoch": 18.5272536687631,
      "grad_norm": 0.4722333252429962,
      "learning_rate": 8.219312424715825e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 23058640,
      "step": 35350
    },
    {
      "epoch": 18.52987421383648,
      "grad_norm": 0.4711783528327942,
      "learning_rate": 8.190259182408345e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 23061488,
      "step": 35355
    },
    {
      "epoch": 18.532494758909852,
      "grad_norm": 0.13556012511253357,
      "learning_rate": 8.161256523939925e-07,
      "loss": 0.0033,
      "num_input_tokens_seen": 23064016,
      "step": 35360
    },
    {
      "epoch": 18.53511530398323,
      "grad_norm": 0.8651562929153442,
      "learning_rate": 8.132304455377576e-07,
      "loss": 0.0051,
      "num_input_tokens_seen": 23066576,
      "step": 35365
    },
    {
      "epoch": 18.537735849056602,
      "grad_norm": 0.4776149094104767,
      "learning_rate": 8.103402982777813e-07,
      "loss": 0.0235,
      "num_input_tokens_seen": 23069936,
      "step": 35370
    },
    {
      "epoch": 18.54035639412998,
      "grad_norm": 1.0905390977859497,
      "learning_rate": 8.07455211218644e-07,
      "loss": 0.0248,
      "num_input_tokens_seen": 23073392,
      "step": 35375
    },
    {
      "epoch": 18.542976939203353,
      "grad_norm": 1.5209704637527466,
      "learning_rate": 8.045751849638738e-07,
      "loss": 0.0202,
      "num_input_tokens_seen": 23079568,
      "step": 35380
    },
    {
      "epoch": 18.54559748427673,
      "grad_norm": 0.536507248878479,
      "learning_rate": 8.017002201159363e-07,
      "loss": 0.0038,
      "num_input_tokens_seen": 23082480,
      "step": 35385
    },
    {
      "epoch": 18.548218029350103,
      "grad_norm": 0.5291569828987122,
      "learning_rate": 7.988303172762418e-07,
      "loss": 0.0159,
      "num_input_tokens_seen": 23085328,
      "step": 35390
    },
    {
      "epoch": 18.55083857442348,
      "grad_norm": 0.46981552243232727,
      "learning_rate": 7.959654770451463e-07,
      "loss": 0.0049,
      "num_input_tokens_seen": 23088208,
      "step": 35395
    },
    {
      "epoch": 18.553459119496857,
      "grad_norm": 0.2779792249202728,
      "learning_rate": 7.9310570002194e-07,
      "loss": 0.0031,
      "num_input_tokens_seen": 23091376,
      "step": 35400
    },
    {
      "epoch": 18.55607966457023,
      "grad_norm": 0.8569138646125793,
      "learning_rate": 7.902509868048552e-07,
      "loss": 0.0063,
      "num_input_tokens_seen": 23094832,
      "step": 35405
    },
    {
      "epoch": 18.558700209643607,
      "grad_norm": 0.6644048690795898,
      "learning_rate": 7.874013379910616e-07,
      "loss": 0.0062,
      "num_input_tokens_seen": 23097808,
      "step": 35410
    },
    {
      "epoch": 18.56132075471698,
      "grad_norm": 0.18133030831813812,
      "learning_rate": 7.845567541766769e-07,
      "loss": 0.0074,
      "num_input_tokens_seen": 23100464,
      "step": 35415
    },
    {
      "epoch": 18.563941299790358,
      "grad_norm": 0.7066510319709778,
      "learning_rate": 7.817172359567526e-07,
      "loss": 0.0026,
      "num_input_tokens_seen": 23103760,
      "step": 35420
    },
    {
      "epoch": 18.56656184486373,
      "grad_norm": 0.22081533074378967,
      "learning_rate": 7.788827839252888e-07,
      "loss": 0.0026,
      "num_input_tokens_seen": 23106832,
      "step": 35425
    },
    {
      "epoch": 18.569182389937108,
      "grad_norm": 1.584452509880066,
      "learning_rate": 7.760533986752222e-07,
      "loss": 0.0163,
      "num_input_tokens_seen": 23110832,
      "step": 35430
    },
    {
      "epoch": 18.57180293501048,
      "grad_norm": 0.2604838013648987,
      "learning_rate": 7.732290807984155e-07,
      "loss": 0.002,
      "num_input_tokens_seen": 23113968,
      "step": 35435
    },
    {
      "epoch": 18.57442348008386,
      "grad_norm": 0.19439977407455444,
      "learning_rate": 7.704098308856933e-07,
      "loss": 0.0026,
      "num_input_tokens_seen": 23116944,
      "step": 35440
    },
    {
      "epoch": 18.57704402515723,
      "grad_norm": 1.377752661705017,
      "learning_rate": 7.675956495268088e-07,
      "loss": 0.0178,
      "num_input_tokens_seen": 23120112,
      "step": 35445
    },
    {
      "epoch": 18.57966457023061,
      "grad_norm": 0.26368558406829834,
      "learning_rate": 7.647865373104551e-07,
      "loss": 0.0114,
      "num_input_tokens_seen": 23124176,
      "step": 35450
    },
    {
      "epoch": 18.582285115303982,
      "grad_norm": 0.5970939993858337,
      "learning_rate": 7.619824948242732e-07,
      "loss": 0.0026,
      "num_input_tokens_seen": 23126832,
      "step": 35455
    },
    {
      "epoch": 18.58490566037736,
      "grad_norm": 0.3316933214664459,
      "learning_rate": 7.5918352265483e-07,
      "loss": 0.0041,
      "num_input_tokens_seen": 23129904,
      "step": 35460
    },
    {
      "epoch": 18.587526205450732,
      "grad_norm": 1.2219500541687012,
      "learning_rate": 7.563896213876348e-07,
      "loss": 0.0102,
      "num_input_tokens_seen": 23133200,
      "step": 35465
    },
    {
      "epoch": 18.59014675052411,
      "grad_norm": 0.15566088259220123,
      "learning_rate": 7.53600791607148e-07,
      "loss": 0.0088,
      "num_input_tokens_seen": 23136496,
      "step": 35470
    },
    {
      "epoch": 18.592767295597483,
      "grad_norm": 0.4314323961734772,
      "learning_rate": 7.508170338967585e-07,
      "loss": 0.005,
      "num_input_tokens_seen": 23139472,
      "step": 35475
    },
    {
      "epoch": 18.59538784067086,
      "grad_norm": 0.3138628304004669,
      "learning_rate": 7.480383488387976e-07,
      "loss": 0.006,
      "num_input_tokens_seen": 23142384,
      "step": 35480
    },
    {
      "epoch": 18.598008385744233,
      "grad_norm": 0.4578585922718048,
      "learning_rate": 7.452647370145338e-07,
      "loss": 0.0222,
      "num_input_tokens_seen": 23145328,
      "step": 35485
    },
    {
      "epoch": 18.60062893081761,
      "grad_norm": 0.300345778465271,
      "learning_rate": 7.424961990041696e-07,
      "loss": 0.0061,
      "num_input_tokens_seen": 23148912,
      "step": 35490
    },
    {
      "epoch": 18.603249475890987,
      "grad_norm": 0.2658504247665405,
      "learning_rate": 7.397327353868583e-07,
      "loss": 0.0032,
      "num_input_tokens_seen": 23151632,
      "step": 35495
    },
    {
      "epoch": 18.60587002096436,
      "grad_norm": 0.545375406742096,
      "learning_rate": 7.369743467406848e-07,
      "loss": 0.0039,
      "num_input_tokens_seen": 23154032,
      "step": 35500
    },
    {
      "epoch": 18.608490566037737,
      "grad_norm": 0.7261425256729126,
      "learning_rate": 7.34221033642668e-07,
      "loss": 0.0133,
      "num_input_tokens_seen": 23156976,
      "step": 35505
    },
    {
      "epoch": 18.61111111111111,
      "grad_norm": 0.36102405190467834,
      "learning_rate": 7.314727966687751e-07,
      "loss": 0.0028,
      "num_input_tokens_seen": 23159472,
      "step": 35510
    },
    {
      "epoch": 18.613731656184488,
      "grad_norm": 0.4068121314048767,
      "learning_rate": 7.287296363939045e-07,
      "loss": 0.0037,
      "num_input_tokens_seen": 23162672,
      "step": 35515
    },
    {
      "epoch": 18.61635220125786,
      "grad_norm": 0.08679217845201492,
      "learning_rate": 7.259915533918888e-07,
      "loss": 0.008,
      "num_input_tokens_seen": 23165456,
      "step": 35520
    },
    {
      "epoch": 18.618972746331238,
      "grad_norm": 0.1676519215106964,
      "learning_rate": 7.232585482355086e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 23168656,
      "step": 35525
    },
    {
      "epoch": 18.62159329140461,
      "grad_norm": 0.3126485347747803,
      "learning_rate": 7.205306214964736e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 23171856,
      "step": 35530
    },
    {
      "epoch": 18.62421383647799,
      "grad_norm": 0.06222597137093544,
      "learning_rate": 7.178077737454436e-07,
      "loss": 0.0067,
      "num_input_tokens_seen": 23175856,
      "step": 35535
    },
    {
      "epoch": 18.62683438155136,
      "grad_norm": 0.6392042636871338,
      "learning_rate": 7.150900055519993e-07,
      "loss": 0.006,
      "num_input_tokens_seen": 23178928,
      "step": 35540
    },
    {
      "epoch": 18.62945492662474,
      "grad_norm": 0.3370489776134491,
      "learning_rate": 7.123773174846693e-07,
      "loss": 0.0174,
      "num_input_tokens_seen": 23182384,
      "step": 35545
    },
    {
      "epoch": 18.632075471698112,
      "grad_norm": 0.23366984724998474,
      "learning_rate": 7.096697101109134e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 23186192,
      "step": 35550
    },
    {
      "epoch": 18.63469601677149,
      "grad_norm": 1.2186285257339478,
      "learning_rate": 7.069671839971371e-07,
      "loss": 0.0098,
      "num_input_tokens_seen": 23191056,
      "step": 35555
    },
    {
      "epoch": 18.637316561844862,
      "grad_norm": 0.3586328625679016,
      "learning_rate": 7.042697397086795e-07,
      "loss": 0.0097,
      "num_input_tokens_seen": 23194160,
      "step": 35560
    },
    {
      "epoch": 18.63993710691824,
      "grad_norm": 0.446320116519928,
      "learning_rate": 7.015773778098089e-07,
      "loss": 0.0039,
      "num_input_tokens_seen": 23197392,
      "step": 35565
    },
    {
      "epoch": 18.642557651991613,
      "grad_norm": 0.2322114259004593,
      "learning_rate": 6.988900988637415e-07,
      "loss": 0.0038,
      "num_input_tokens_seen": 23201008,
      "step": 35570
    },
    {
      "epoch": 18.64517819706499,
      "grad_norm": 0.9453302621841431,
      "learning_rate": 6.962079034326218e-07,
      "loss": 0.0303,
      "num_input_tokens_seen": 23208176,
      "step": 35575
    },
    {
      "epoch": 18.647798742138363,
      "grad_norm": 0.3875398635864258,
      "learning_rate": 6.935307920775346e-07,
      "loss": 0.0026,
      "num_input_tokens_seen": 23210768,
      "step": 35580
    },
    {
      "epoch": 18.65041928721174,
      "grad_norm": 0.20788052678108215,
      "learning_rate": 6.90858765358507e-07,
      "loss": 0.0207,
      "num_input_tokens_seen": 23214480,
      "step": 35585
    },
    {
      "epoch": 18.653039832285117,
      "grad_norm": 0.8969646692276001,
      "learning_rate": 6.881918238344915e-07,
      "loss": 0.0029,
      "num_input_tokens_seen": 23217296,
      "step": 35590
    },
    {
      "epoch": 18.65566037735849,
      "grad_norm": 0.13210611045360565,
      "learning_rate": 6.855299680633808e-07,
      "loss": 0.0024,
      "num_input_tokens_seen": 23220272,
      "step": 35595
    },
    {
      "epoch": 18.658280922431867,
      "grad_norm": 0.3244079649448395,
      "learning_rate": 6.828731986020104e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 23224912,
      "step": 35600
    },
    {
      "epoch": 18.66090146750524,
      "grad_norm": 0.530887246131897,
      "learning_rate": 6.802215160061381e-07,
      "loss": 0.0032,
      "num_input_tokens_seen": 23227600,
      "step": 35605
    },
    {
      "epoch": 18.663522012578618,
      "grad_norm": 4.46806001663208,
      "learning_rate": 6.775749208304704e-07,
      "loss": 0.0206,
      "num_input_tokens_seen": 23230576,
      "step": 35610
    },
    {
      "epoch": 18.66614255765199,
      "grad_norm": 12.119919776916504,
      "learning_rate": 6.749334136286478e-07,
      "loss": 0.0041,
      "num_input_tokens_seen": 23233360,
      "step": 35615
    },
    {
      "epoch": 18.668763102725368,
      "grad_norm": 0.4039601683616638,
      "learning_rate": 6.722969949532365e-07,
      "loss": 0.0664,
      "num_input_tokens_seen": 23237328,
      "step": 35620
    },
    {
      "epoch": 18.67138364779874,
      "grad_norm": 0.8129379153251648,
      "learning_rate": 6.69665665355751e-07,
      "loss": 0.0047,
      "num_input_tokens_seen": 23240592,
      "step": 35625
    },
    {
      "epoch": 18.67400419287212,
      "grad_norm": 0.2083873450756073,
      "learning_rate": 6.670394253866314e-07,
      "loss": 0.0035,
      "num_input_tokens_seen": 23243472,
      "step": 35630
    },
    {
      "epoch": 18.67662473794549,
      "grad_norm": 0.4148973226547241,
      "learning_rate": 6.644182755952605e-07,
      "loss": 0.0018,
      "num_input_tokens_seen": 23246032,
      "step": 35635
    },
    {
      "epoch": 18.67924528301887,
      "grad_norm": 0.2641749382019043,
      "learning_rate": 6.618022165299526e-07,
      "loss": 0.0025,
      "num_input_tokens_seen": 23248304,
      "step": 35640
    },
    {
      "epoch": 18.681865828092242,
      "grad_norm": 0.43102043867111206,
      "learning_rate": 6.591912487379531e-07,
      "loss": 0.0024,
      "num_input_tokens_seen": 23251632,
      "step": 35645
    },
    {
      "epoch": 18.68448637316562,
      "grad_norm": 0.09896684437990189,
      "learning_rate": 6.565853727654503e-07,
      "loss": 0.0097,
      "num_input_tokens_seen": 23255216,
      "step": 35650
    },
    {
      "epoch": 18.687106918238992,
      "grad_norm": 1.3964647054672241,
      "learning_rate": 6.539845891575636e-07,
      "loss": 0.0058,
      "num_input_tokens_seen": 23258736,
      "step": 35655
    },
    {
      "epoch": 18.68972746331237,
      "grad_norm": 0.46674031019210815,
      "learning_rate": 6.513888984583466e-07,
      "loss": 0.0024,
      "num_input_tokens_seen": 23261872,
      "step": 35660
    },
    {
      "epoch": 18.692348008385743,
      "grad_norm": 1.4721075296401978,
      "learning_rate": 6.487983012107873e-07,
      "loss": 0.0179,
      "num_input_tokens_seen": 23265008,
      "step": 35665
    },
    {
      "epoch": 18.69496855345912,
      "grad_norm": 0.9883663058280945,
      "learning_rate": 6.462127979568133e-07,
      "loss": 0.0077,
      "num_input_tokens_seen": 23268912,
      "step": 35670
    },
    {
      "epoch": 18.697589098532493,
      "grad_norm": 0.5022590756416321,
      "learning_rate": 6.436323892372781e-07,
      "loss": 0.001,
      "num_input_tokens_seen": 23271344,
      "step": 35675
    },
    {
      "epoch": 18.70020964360587,
      "grad_norm": 1.8257784843444824,
      "learning_rate": 6.410570755919721e-07,
      "loss": 0.0048,
      "num_input_tokens_seen": 23274960,
      "step": 35680
    },
    {
      "epoch": 18.702830188679247,
      "grad_norm": 0.4772072434425354,
      "learning_rate": 6.384868575596281e-07,
      "loss": 0.0096,
      "num_input_tokens_seen": 23278160,
      "step": 35685
    },
    {
      "epoch": 18.70545073375262,
      "grad_norm": 0.16519802808761597,
      "learning_rate": 6.359217356779024e-07,
      "loss": 0.0073,
      "num_input_tokens_seen": 23282192,
      "step": 35690
    },
    {
      "epoch": 18.708071278825997,
      "grad_norm": 0.8352307677268982,
      "learning_rate": 6.333617104833878e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 23285648,
      "step": 35695
    },
    {
      "epoch": 18.71069182389937,
      "grad_norm": 0.1300288885831833,
      "learning_rate": 6.308067825116142e-07,
      "loss": 0.0123,
      "num_input_tokens_seen": 23288304,
      "step": 35700
    },
    {
      "epoch": 18.713312368972748,
      "grad_norm": 0.3015921413898468,
      "learning_rate": 6.282569522970433e-07,
      "loss": 0.0053,
      "num_input_tokens_seen": 23291408,
      "step": 35705
    },
    {
      "epoch": 18.71593291404612,
      "grad_norm": 0.8895524740219116,
      "learning_rate": 6.257122203730731e-07,
      "loss": 0.0092,
      "num_input_tokens_seen": 23294448,
      "step": 35710
    },
    {
      "epoch": 18.718553459119498,
      "grad_norm": 0.5871415138244629,
      "learning_rate": 6.231725872720251e-07,
      "loss": 0.0057,
      "num_input_tokens_seen": 23297232,
      "step": 35715
    },
    {
      "epoch": 18.72117400419287,
      "grad_norm": 0.7611002326011658,
      "learning_rate": 6.206380535251716e-07,
      "loss": 0.006,
      "num_input_tokens_seen": 23301264,
      "step": 35720
    },
    {
      "epoch": 18.72379454926625,
      "grad_norm": 0.22783689200878143,
      "learning_rate": 6.181086196626995e-07,
      "loss": 0.0176,
      "num_input_tokens_seen": 23304720,
      "step": 35725
    },
    {
      "epoch": 18.72641509433962,
      "grad_norm": 0.28173351287841797,
      "learning_rate": 6.155842862137384e-07,
      "loss": 0.007,
      "num_input_tokens_seen": 23308592,
      "step": 35730
    },
    {
      "epoch": 18.729035639413,
      "grad_norm": 0.5868021249771118,
      "learning_rate": 6.130650537063548e-07,
      "loss": 0.1303,
      "num_input_tokens_seen": 23312080,
      "step": 35735
    },
    {
      "epoch": 18.731656184486372,
      "grad_norm": 2.9929890632629395,
      "learning_rate": 6.105509226675438e-07,
      "loss": 0.0574,
      "num_input_tokens_seen": 23314800,
      "step": 35740
    },
    {
      "epoch": 18.73427672955975,
      "grad_norm": 1.9209318161010742,
      "learning_rate": 6.08041893623229e-07,
      "loss": 0.0161,
      "num_input_tokens_seen": 23317840,
      "step": 35745
    },
    {
      "epoch": 18.736897274633122,
      "grad_norm": 0.1491626501083374,
      "learning_rate": 6.055379670982713e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 23321200,
      "step": 35750
    },
    {
      "epoch": 18.7395178197065,
      "grad_norm": 0.6029341220855713,
      "learning_rate": 6.030391436164628e-07,
      "loss": 0.0051,
      "num_input_tokens_seen": 23323824,
      "step": 35755
    },
    {
      "epoch": 18.742138364779873,
      "grad_norm": 0.47546473145484924,
      "learning_rate": 6.005454237005298e-07,
      "loss": 0.0082,
      "num_input_tokens_seen": 23326928,
      "step": 35760
    },
    {
      "epoch": 18.74475890985325,
      "grad_norm": 2.371925115585327,
      "learning_rate": 5.980568078721327e-07,
      "loss": 0.0419,
      "num_input_tokens_seen": 23330576,
      "step": 35765
    },
    {
      "epoch": 18.747379454926623,
      "grad_norm": 0.8303160071372986,
      "learning_rate": 5.955732966518606e-07,
      "loss": 0.0119,
      "num_input_tokens_seen": 23334832,
      "step": 35770
    },
    {
      "epoch": 18.75,
      "grad_norm": 0.24991516768932343,
      "learning_rate": 5.930948905592343e-07,
      "loss": 0.0023,
      "num_input_tokens_seen": 23337488,
      "step": 35775
    },
    {
      "epoch": 18.752620545073377,
      "grad_norm": 0.29490289092063904,
      "learning_rate": 5.906215901127054e-07,
      "loss": 0.0141,
      "num_input_tokens_seen": 23341296,
      "step": 35780
    },
    {
      "epoch": 18.75524109014675,
      "grad_norm": 0.9848598837852478,
      "learning_rate": 5.881533958296631e-07,
      "loss": 0.0208,
      "num_input_tokens_seen": 23348144,
      "step": 35785
    },
    {
      "epoch": 18.757861635220127,
      "grad_norm": 0.6138462424278259,
      "learning_rate": 5.856903082264248e-07,
      "loss": 0.0057,
      "num_input_tokens_seen": 23352080,
      "step": 35790
    },
    {
      "epoch": 18.7604821802935,
      "grad_norm": 0.10375625640153885,
      "learning_rate": 5.832323278182422e-07,
      "loss": 0.0021,
      "num_input_tokens_seen": 23355536,
      "step": 35795
    },
    {
      "epoch": 18.763102725366878,
      "grad_norm": 0.42405879497528076,
      "learning_rate": 5.807794551192986e-07,
      "loss": 0.0021,
      "num_input_tokens_seen": 23358160,
      "step": 35800
    },
    {
      "epoch": 18.76572327044025,
      "grad_norm": 0.7367733120918274,
      "learning_rate": 5.783316906427e-07,
      "loss": 0.0064,
      "num_input_tokens_seen": 23361296,
      "step": 35805
    },
    {
      "epoch": 18.768343815513628,
      "grad_norm": 0.31823697686195374,
      "learning_rate": 5.758890349004925e-07,
      "loss": 0.0087,
      "num_input_tokens_seen": 23365264,
      "step": 35810
    },
    {
      "epoch": 18.770964360587,
      "grad_norm": 0.11647722870111465,
      "learning_rate": 5.734514884036563e-07,
      "loss": 0.0022,
      "num_input_tokens_seen": 23368208,
      "step": 35815
    },
    {
      "epoch": 18.77358490566038,
      "grad_norm": 0.10885961353778839,
      "learning_rate": 5.710190516620945e-07,
      "loss": 0.004,
      "num_input_tokens_seen": 23371248,
      "step": 35820
    },
    {
      "epoch": 18.77620545073375,
      "grad_norm": 0.22009475529193878,
      "learning_rate": 5.685917251846501e-07,
      "loss": 0.0101,
      "num_input_tokens_seen": 23374960,
      "step": 35825
    },
    {
      "epoch": 18.77882599580713,
      "grad_norm": 0.40183529257774353,
      "learning_rate": 5.661695094790892e-07,
      "loss": 0.0017,
      "num_input_tokens_seen": 23378832,
      "step": 35830
    },
    {
      "epoch": 18.781446540880502,
      "grad_norm": 0.28950250148773193,
      "learning_rate": 5.637524050521064e-07,
      "loss": 0.0186,
      "num_input_tokens_seen": 23382032,
      "step": 35835
    },
    {
      "epoch": 18.78406708595388,
      "grad_norm": 0.13265550136566162,
      "learning_rate": 5.613404124093391e-07,
      "loss": 0.0063,
      "num_input_tokens_seen": 23385456,
      "step": 35840
    },
    {
      "epoch": 18.786687631027252,
      "grad_norm": 0.9830854535102844,
      "learning_rate": 5.589335320553446e-07,
      "loss": 0.008,
      "num_input_tokens_seen": 23388592,
      "step": 35845
    },
    {
      "epoch": 18.78930817610063,
      "grad_norm": 0.28922978043556213,
      "learning_rate": 5.565317644936202e-07,
      "loss": 0.0066,
      "num_input_tokens_seen": 23391824,
      "step": 35850
    },
    {
      "epoch": 18.791928721174003,
      "grad_norm": 0.17627392709255219,
      "learning_rate": 5.541351102265863e-07,
      "loss": 0.0034,
      "num_input_tokens_seen": 23394896,
      "step": 35855
    },
    {
      "epoch": 18.79454926624738,
      "grad_norm": 0.20690934360027313,
      "learning_rate": 5.51743569755589e-07,
      "loss": 0.006,
      "num_input_tokens_seen": 23399888,
      "step": 35860
    },
    {
      "epoch": 18.797169811320753,
      "grad_norm": 0.0901477187871933,
      "learning_rate": 5.493571435809197e-07,
      "loss": 0.0021,
      "num_input_tokens_seen": 23402832,
      "step": 35865
    },
    {
      "epoch": 18.79979035639413,
      "grad_norm": 0.2165970802307129,
      "learning_rate": 5.469758322017849e-07,
      "loss": 0.0221,
      "num_input_tokens_seen": 23405968,
      "step": 35870
    },
    {
      "epoch": 18.802410901467507,
      "grad_norm": 2.4476218223571777,
      "learning_rate": 5.445996361163358e-07,
      "loss": 0.0428,
      "num_input_tokens_seen": 23408656,
      "step": 35875
    },
    {
      "epoch": 18.80503144654088,
      "grad_norm": 0.26713472604751587,
      "learning_rate": 5.422285558216389e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 23411952,
      "step": 35880
    },
    {
      "epoch": 18.807651991614257,
      "grad_norm": 0.691711962223053,
      "learning_rate": 5.398625918137001e-07,
      "loss": 0.0079,
      "num_input_tokens_seen": 23414896,
      "step": 35885
    },
    {
      "epoch": 18.81027253668763,
      "grad_norm": 0.3883988559246063,
      "learning_rate": 5.375017445874486e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 23418224,
      "step": 35890
    },
    {
      "epoch": 18.812893081761008,
      "grad_norm": 0.10232481360435486,
      "learning_rate": 5.351460146367504e-07,
      "loss": 0.0023,
      "num_input_tokens_seen": 23421008,
      "step": 35895
    },
    {
      "epoch": 18.81551362683438,
      "grad_norm": 0.10927422344684601,
      "learning_rate": 5.327954024543946e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 23423792,
      "step": 35900
    },
    {
      "epoch": 18.818134171907758,
      "grad_norm": 0.39327746629714966,
      "learning_rate": 5.304499085321074e-07,
      "loss": 0.0049,
      "num_input_tokens_seen": 23426320,
      "step": 35905
    },
    {
      "epoch": 18.82075471698113,
      "grad_norm": 0.2706437110900879,
      "learning_rate": 5.281095333605352e-07,
      "loss": 0.0112,
      "num_input_tokens_seen": 23429840,
      "step": 35910
    },
    {
      "epoch": 18.82337526205451,
      "grad_norm": 0.36759334802627563,
      "learning_rate": 5.257742774292556e-07,
      "loss": 0.0027,
      "num_input_tokens_seen": 23432720,
      "step": 35915
    },
    {
      "epoch": 18.82599580712788,
      "grad_norm": 0.543059229850769,
      "learning_rate": 5.234441412267837e-07,
      "loss": 0.0061,
      "num_input_tokens_seen": 23436592,
      "step": 35920
    },
    {
      "epoch": 18.82861635220126,
      "grad_norm": 0.1163221225142479,
      "learning_rate": 5.211191252405517e-07,
      "loss": 0.0033,
      "num_input_tokens_seen": 23439856,
      "step": 35925
    },
    {
      "epoch": 18.831236897274632,
      "grad_norm": 0.4671803116798401,
      "learning_rate": 5.187992299569316e-07,
      "loss": 0.0032,
      "num_input_tokens_seen": 23442832,
      "step": 35930
    },
    {
      "epoch": 18.83385744234801,
      "grad_norm": 0.6787527203559875,
      "learning_rate": 5.164844558612131e-07,
      "loss": 0.0137,
      "num_input_tokens_seen": 23446416,
      "step": 35935
    },
    {
      "epoch": 18.836477987421382,
      "grad_norm": 0.4527362585067749,
      "learning_rate": 5.141748034376254e-07,
      "loss": 0.0115,
      "num_input_tokens_seen": 23449104,
      "step": 35940
    },
    {
      "epoch": 18.83909853249476,
      "grad_norm": 1.0329527854919434,
      "learning_rate": 5.118702731693209e-07,
      "loss": 0.0085,
      "num_input_tokens_seen": 23452272,
      "step": 35945
    },
    {
      "epoch": 18.841719077568133,
      "grad_norm": 0.42285409569740295,
      "learning_rate": 5.095708655383751e-07,
      "loss": 0.0131,
      "num_input_tokens_seen": 23455088,
      "step": 35950
    },
    {
      "epoch": 18.84433962264151,
      "grad_norm": 0.26040512323379517,
      "learning_rate": 5.07276581025809e-07,
      "loss": 0.0055,
      "num_input_tokens_seen": 23458704,
      "step": 35955
    },
    {
      "epoch": 18.846960167714883,
      "grad_norm": 0.22455501556396484,
      "learning_rate": 5.049874201115495e-07,
      "loss": 0.0057,
      "num_input_tokens_seen": 23462704,
      "step": 35960
    },
    {
      "epoch": 18.84958071278826,
      "grad_norm": 1.125910997390747,
      "learning_rate": 5.027033832744693e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 23465328,
      "step": 35965
    },
    {
      "epoch": 18.852201257861637,
      "grad_norm": 0.7957437634468079,
      "learning_rate": 5.004244709923639e-07,
      "loss": 0.0208,
      "num_input_tokens_seen": 23468912,
      "step": 35970
    },
    {
      "epoch": 18.85482180293501,
      "grad_norm": 1.6216206550598145,
      "learning_rate": 4.981506837419519e-07,
      "loss": 0.0106,
      "num_input_tokens_seen": 23471984,
      "step": 35975
    },
    {
      "epoch": 18.857442348008387,
      "grad_norm": 0.05380722135305405,
      "learning_rate": 4.958820219988836e-07,
      "loss": 0.005,
      "num_input_tokens_seen": 23475888,
      "step": 35980
    },
    {
      "epoch": 18.86006289308176,
      "grad_norm": 0.28147611021995544,
      "learning_rate": 4.936184862377402e-07,
      "loss": 0.0168,
      "num_input_tokens_seen": 23478704,
      "step": 35985
    },
    {
      "epoch": 18.862683438155138,
      "grad_norm": 1.1655820608139038,
      "learning_rate": 4.913600769320237e-07,
      "loss": 0.0066,
      "num_input_tokens_seen": 23481136,
      "step": 35990
    },
    {
      "epoch": 18.86530398322851,
      "grad_norm": 0.6190304160118103,
      "learning_rate": 4.891067945541728e-07,
      "loss": 0.0026,
      "num_input_tokens_seen": 23483408,
      "step": 35995
    },
    {
      "epoch": 18.867924528301888,
      "grad_norm": 0.704322874546051,
      "learning_rate": 4.868586395755437e-07,
      "loss": 0.0067,
      "num_input_tokens_seen": 23487568,
      "step": 36000
    },
    {
      "epoch": 18.87054507337526,
      "grad_norm": 0.8140106201171875,
      "learning_rate": 4.846156124664242e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 23490512,
      "step": 36005
    },
    {
      "epoch": 18.87316561844864,
      "grad_norm": 0.2659338414669037,
      "learning_rate": 4.823777136960362e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 23493680,
      "step": 36010
    },
    {
      "epoch": 18.87578616352201,
      "grad_norm": 0.7745485305786133,
      "learning_rate": 4.801449437325162e-07,
      "loss": 0.0097,
      "num_input_tokens_seen": 23497680,
      "step": 36015
    },
    {
      "epoch": 18.87840670859539,
      "grad_norm": 0.24460254609584808,
      "learning_rate": 4.779173030429379e-07,
      "loss": 0.002,
      "num_input_tokens_seen": 23500464,
      "step": 36020
    },
    {
      "epoch": 18.881027253668762,
      "grad_norm": 0.22178179025650024,
      "learning_rate": 4.7569479209329804e-07,
      "loss": 0.0032,
      "num_input_tokens_seen": 23503184,
      "step": 36025
    },
    {
      "epoch": 18.88364779874214,
      "grad_norm": 0.5785184502601624,
      "learning_rate": 4.734774113485163e-07,
      "loss": 0.0035,
      "num_input_tokens_seen": 23506064,
      "step": 36030
    },
    {
      "epoch": 18.886268343815512,
      "grad_norm": 0.20389431715011597,
      "learning_rate": 4.7126516127244945e-07,
      "loss": 0.0112,
      "num_input_tokens_seen": 23508848,
      "step": 36035
    },
    {
      "epoch": 18.88888888888889,
      "grad_norm": 0.5251544713973999,
      "learning_rate": 4.690580423278718e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 23512432,
      "step": 36040
    },
    {
      "epoch": 18.891509433962263,
      "grad_norm": 0.3301105201244354,
      "learning_rate": 4.668560549764861e-07,
      "loss": 0.0055,
      "num_input_tokens_seen": 23515312,
      "step": 36045
    },
    {
      "epoch": 18.89412997903564,
      "grad_norm": 2.062116861343384,
      "learning_rate": 4.646591996789268e-07,
      "loss": 0.0038,
      "num_input_tokens_seen": 23519024,
      "step": 36050
    },
    {
      "epoch": 18.896750524109013,
      "grad_norm": 0.2643706202507019,
      "learning_rate": 4.6246747689474847e-07,
      "loss": 0.0038,
      "num_input_tokens_seen": 23522032,
      "step": 36055
    },
    {
      "epoch": 18.89937106918239,
      "grad_norm": 0.2475908100605011,
      "learning_rate": 4.6028088708243434e-07,
      "loss": 0.0037,
      "num_input_tokens_seen": 23524912,
      "step": 36060
    },
    {
      "epoch": 18.901991614255767,
      "grad_norm": 0.12594333291053772,
      "learning_rate": 4.580994306993991e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 23527984,
      "step": 36065
    },
    {
      "epoch": 18.90461215932914,
      "grad_norm": 2.7848217487335205,
      "learning_rate": 4.5592310820196936e-07,
      "loss": 0.0283,
      "num_input_tokens_seen": 23531792,
      "step": 36070
    },
    {
      "epoch": 18.907232704402517,
      "grad_norm": 0.46647825837135315,
      "learning_rate": 4.537519200454116e-07,
      "loss": 0.0135,
      "num_input_tokens_seen": 23535504,
      "step": 36075
    },
    {
      "epoch": 18.90985324947589,
      "grad_norm": 0.19933921098709106,
      "learning_rate": 4.515858666839151e-07,
      "loss": 0.0039,
      "num_input_tokens_seen": 23538864,
      "step": 36080
    },
    {
      "epoch": 18.912473794549268,
      "grad_norm": 0.6094088554382324,
      "learning_rate": 4.494249485705926e-07,
      "loss": 0.0078,
      "num_input_tokens_seen": 23542736,
      "step": 36085
    },
    {
      "epoch": 18.91509433962264,
      "grad_norm": 0.22147423028945923,
      "learning_rate": 4.472691661574824e-07,
      "loss": 0.0192,
      "num_input_tokens_seen": 23547504,
      "step": 36090
    },
    {
      "epoch": 18.917714884696018,
      "grad_norm": 0.6626576781272888,
      "learning_rate": 4.4511851989555156e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 23550224,
      "step": 36095
    },
    {
      "epoch": 18.92033542976939,
      "grad_norm": 0.418096661567688,
      "learning_rate": 4.429730102346874e-07,
      "loss": 0.0043,
      "num_input_tokens_seen": 23553488,
      "step": 36100
    },
    {
      "epoch": 18.92295597484277,
      "grad_norm": 1.7678180932998657,
      "learning_rate": 4.4083263762370873e-07,
      "loss": 0.0235,
      "num_input_tokens_seen": 23556816,
      "step": 36105
    },
    {
      "epoch": 18.92557651991614,
      "grad_norm": 14.654979705810547,
      "learning_rate": 4.386974025103574e-07,
      "loss": 0.0024,
      "num_input_tokens_seen": 23559280,
      "step": 36110
    },
    {
      "epoch": 18.92819706498952,
      "grad_norm": 0.5929399728775024,
      "learning_rate": 4.365673053413011e-07,
      "loss": 0.0057,
      "num_input_tokens_seen": 23562416,
      "step": 36115
    },
    {
      "epoch": 18.930817610062892,
      "grad_norm": 0.6084365248680115,
      "learning_rate": 4.34442346562125e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 23566032,
      "step": 36120
    },
    {
      "epoch": 18.93343815513627,
      "grad_norm": 0.06675386428833008,
      "learning_rate": 4.323225266173542e-07,
      "loss": 0.0078,
      "num_input_tokens_seen": 23568880,
      "step": 36125
    },
    {
      "epoch": 18.936058700209642,
      "grad_norm": 0.0752144604921341,
      "learning_rate": 4.302078459504283e-07,
      "loss": 0.0029,
      "num_input_tokens_seen": 23571216,
      "step": 36130
    },
    {
      "epoch": 18.93867924528302,
      "grad_norm": 0.16802819073200226,
      "learning_rate": 4.2809830500371294e-07,
      "loss": 0.0072,
      "num_input_tokens_seen": 23575216,
      "step": 36135
    },
    {
      "epoch": 18.941299790356393,
      "grad_norm": 0.16883741319179535,
      "learning_rate": 4.2599390421850236e-07,
      "loss": 0.0081,
      "num_input_tokens_seen": 23578576,
      "step": 36140
    },
    {
      "epoch": 18.94392033542977,
      "grad_norm": 0.408761590719223,
      "learning_rate": 4.2389464403501113e-07,
      "loss": 0.0023,
      "num_input_tokens_seen": 23580848,
      "step": 36145
    },
    {
      "epoch": 18.946540880503143,
      "grad_norm": 0.17494463920593262,
      "learning_rate": 4.218005248923823e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 23583440,
      "step": 36150
    },
    {
      "epoch": 18.94916142557652,
      "grad_norm": 0.8182193636894226,
      "learning_rate": 4.1971154722867946e-07,
      "loss": 0.0099,
      "num_input_tokens_seen": 23586640,
      "step": 36155
    },
    {
      "epoch": 18.951781970649897,
      "grad_norm": 0.6446055769920349,
      "learning_rate": 4.176277114808946e-07,
      "loss": 0.0033,
      "num_input_tokens_seen": 23589968,
      "step": 36160
    },
    {
      "epoch": 18.95440251572327,
      "grad_norm": 0.34633493423461914,
      "learning_rate": 4.155490180849431e-07,
      "loss": 0.0028,
      "num_input_tokens_seen": 23592368,
      "step": 36165
    },
    {
      "epoch": 18.957023060796647,
      "grad_norm": 0.392355352640152,
      "learning_rate": 4.1347546747566304e-07,
      "loss": 0.0026,
      "num_input_tokens_seen": 23595248,
      "step": 36170
    },
    {
      "epoch": 18.95964360587002,
      "grad_norm": 0.14957790076732635,
      "learning_rate": 4.114070600868131e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 23598384,
      "step": 36175
    },
    {
      "epoch": 18.962264150943398,
      "grad_norm": 0.20324820280075073,
      "learning_rate": 4.093437963510888e-07,
      "loss": 0.0124,
      "num_input_tokens_seen": 23602128,
      "step": 36180
    },
    {
      "epoch": 18.96488469601677,
      "grad_norm": 0.816929042339325,
      "learning_rate": 4.072856767000949e-07,
      "loss": 0.004,
      "num_input_tokens_seen": 23605232,
      "step": 36185
    },
    {
      "epoch": 18.967505241090148,
      "grad_norm": 0.23900477588176727,
      "learning_rate": 4.052327015643703e-07,
      "loss": 0.0041,
      "num_input_tokens_seen": 23608272,
      "step": 36190
    },
    {
      "epoch": 18.97012578616352,
      "grad_norm": 0.5137242674827576,
      "learning_rate": 4.031848713733744e-07,
      "loss": 0.0025,
      "num_input_tokens_seen": 23610576,
      "step": 36195
    },
    {
      "epoch": 18.9727463312369,
      "grad_norm": 0.5291231870651245,
      "learning_rate": 4.011421865554837e-07,
      "loss": 0.0079,
      "num_input_tokens_seen": 23614000,
      "step": 36200
    },
    {
      "epoch": 18.97536687631027,
      "grad_norm": 0.38191890716552734,
      "learning_rate": 3.9910464753801224e-07,
      "loss": 0.0029,
      "num_input_tokens_seen": 23617264,
      "step": 36205
    },
    {
      "epoch": 18.97798742138365,
      "grad_norm": 0.2449396252632141,
      "learning_rate": 3.9707225474718557e-07,
      "loss": 0.0022,
      "num_input_tokens_seen": 23620240,
      "step": 36210
    },
    {
      "epoch": 18.980607966457022,
      "grad_norm": 0.5501219034194946,
      "learning_rate": 3.950450086081553e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 23623216,
      "step": 36215
    },
    {
      "epoch": 18.9832285115304,
      "grad_norm": 0.18926264345645905,
      "learning_rate": 3.930229095450072e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 23626064,
      "step": 36220
    },
    {
      "epoch": 18.985849056603772,
      "grad_norm": 0.17602820694446564,
      "learning_rate": 3.9100595798073345e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 23629360,
      "step": 36225
    },
    {
      "epoch": 18.98846960167715,
      "grad_norm": 0.47274383902549744,
      "learning_rate": 3.8899415433725493e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 23631792,
      "step": 36230
    },
    {
      "epoch": 18.991090146750523,
      "grad_norm": 0.3890947699546814,
      "learning_rate": 3.8698749903542666e-07,
      "loss": 0.0055,
      "num_input_tokens_seen": 23636400,
      "step": 36235
    },
    {
      "epoch": 18.9937106918239,
      "grad_norm": 0.6674619913101196,
      "learning_rate": 3.8498599249501e-07,
      "loss": 0.0126,
      "num_input_tokens_seen": 23639056,
      "step": 36240
    },
    {
      "epoch": 18.996331236897273,
      "grad_norm": 0.16280710697174072,
      "learning_rate": 3.8298963513470345e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 23642672,
      "step": 36245
    },
    {
      "epoch": 18.99895178197065,
      "grad_norm": 0.11316465586423874,
      "learning_rate": 3.8099842737211734e-07,
      "loss": 0.0034,
      "num_input_tokens_seen": 23644848,
      "step": 36250
    },
    {
      "epoch": 19.0,
      "eval_loss": 1.5741678476333618,
      "eval_runtime": 11.8886,
      "eval_samples_per_second": 71.329,
      "eval_steps_per_second": 17.832,
      "num_input_tokens_seen": 23645440,
      "step": 36252
    },
    {
      "epoch": 19.001572327044027,
      "grad_norm": 0.35749906301498413,
      "learning_rate": 3.7901236962379627e-07,
      "loss": 0.002,
      "num_input_tokens_seen": 23647040,
      "step": 36255
    },
    {
      "epoch": 19.0041928721174,
      "grad_norm": 0.28740057349205017,
      "learning_rate": 3.7703146230519126e-07,
      "loss": 0.0024,
      "num_input_tokens_seen": 23649760,
      "step": 36260
    },
    {
      "epoch": 19.006813417190777,
      "grad_norm": 0.08543849736452103,
      "learning_rate": 3.7505570583069306e-07,
      "loss": 0.0019,
      "num_input_tokens_seen": 23652672,
      "step": 36265
    },
    {
      "epoch": 19.00943396226415,
      "grad_norm": 0.13805502653121948,
      "learning_rate": 3.7308510061360435e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 23655584,
      "step": 36270
    },
    {
      "epoch": 19.012054507337528,
      "grad_norm": 0.08910404145717621,
      "learning_rate": 3.7111964706615374e-07,
      "loss": 0.0061,
      "num_input_tokens_seen": 23658496,
      "step": 36275
    },
    {
      "epoch": 19.0146750524109,
      "grad_norm": 0.5493512749671936,
      "learning_rate": 3.691593455994929e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 23660928,
      "step": 36280
    },
    {
      "epoch": 19.017295597484278,
      "grad_norm": 0.14017826318740845,
      "learning_rate": 3.672041966236911e-07,
      "loss": 0.0248,
      "num_input_tokens_seen": 23664448,
      "step": 36285
    },
    {
      "epoch": 19.01991614255765,
      "grad_norm": 0.44873204827308655,
      "learning_rate": 3.6525420054774884e-07,
      "loss": 0.0232,
      "num_input_tokens_seen": 23668128,
      "step": 36290
    },
    {
      "epoch": 19.02253668763103,
      "grad_norm": 0.4777994155883789,
      "learning_rate": 3.6330935777957597e-07,
      "loss": 0.016,
      "num_input_tokens_seen": 23672128,
      "step": 36295
    },
    {
      "epoch": 19.0251572327044,
      "grad_norm": 0.47624439001083374,
      "learning_rate": 3.613696687260193e-07,
      "loss": 0.0039,
      "num_input_tokens_seen": 23675392,
      "step": 36300
    },
    {
      "epoch": 19.02777777777778,
      "grad_norm": 0.461616188287735,
      "learning_rate": 3.59435133792832e-07,
      "loss": 0.0208,
      "num_input_tokens_seen": 23679136,
      "step": 36305
    },
    {
      "epoch": 19.030398322851152,
      "grad_norm": 0.677054762840271,
      "learning_rate": 3.575057533847043e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 23681920,
      "step": 36310
    },
    {
      "epoch": 19.03301886792453,
      "grad_norm": 0.3388068675994873,
      "learning_rate": 3.555815279052327e-07,
      "loss": 0.0039,
      "num_input_tokens_seen": 23685824,
      "step": 36315
    },
    {
      "epoch": 19.035639412997902,
      "grad_norm": 1.1923842430114746,
      "learning_rate": 3.5366245775694805e-07,
      "loss": 0.0101,
      "num_input_tokens_seen": 23689344,
      "step": 36320
    },
    {
      "epoch": 19.03825995807128,
      "grad_norm": 0.39770349860191345,
      "learning_rate": 3.517485433412987e-07,
      "loss": 0.0033,
      "num_input_tokens_seen": 23692384,
      "step": 36325
    },
    {
      "epoch": 19.040880503144653,
      "grad_norm": 0.2558232545852661,
      "learning_rate": 3.4983978505865044e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 23695936,
      "step": 36330
    },
    {
      "epoch": 19.04350104821803,
      "grad_norm": 0.36208564043045044,
      "learning_rate": 3.47936183308295e-07,
      "loss": 0.0071,
      "num_input_tokens_seen": 23698848,
      "step": 36335
    },
    {
      "epoch": 19.046121593291403,
      "grad_norm": 0.3018174171447754,
      "learning_rate": 3.460377384884472e-07,
      "loss": 0.0028,
      "num_input_tokens_seen": 23702464,
      "step": 36340
    },
    {
      "epoch": 19.04874213836478,
      "grad_norm": 0.24646106362342834,
      "learning_rate": 3.4414445099623384e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 23705344,
      "step": 36345
    },
    {
      "epoch": 19.051362683438157,
      "grad_norm": 0.24865682423114777,
      "learning_rate": 3.42256321227713e-07,
      "loss": 0.0447,
      "num_input_tokens_seen": 23708992,
      "step": 36350
    },
    {
      "epoch": 19.05398322851153,
      "grad_norm": 0.17233341932296753,
      "learning_rate": 3.403733495778605e-07,
      "loss": 0.0027,
      "num_input_tokens_seen": 23711808,
      "step": 36355
    },
    {
      "epoch": 19.056603773584907,
      "grad_norm": 1.5303245782852173,
      "learning_rate": 3.3849553644056955e-07,
      "loss": 0.0066,
      "num_input_tokens_seen": 23714720,
      "step": 36360
    },
    {
      "epoch": 19.05922431865828,
      "grad_norm": 0.1851615160703659,
      "learning_rate": 3.36622882208662e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 23717984,
      "step": 36365
    },
    {
      "epoch": 19.061844863731658,
      "grad_norm": 1.2528215646743774,
      "learning_rate": 3.347553872738718e-07,
      "loss": 0.0124,
      "num_input_tokens_seen": 23721184,
      "step": 36370
    },
    {
      "epoch": 19.06446540880503,
      "grad_norm": 0.33199968934059143,
      "learning_rate": 3.3289305202685583e-07,
      "loss": 0.0124,
      "num_input_tokens_seen": 23724384,
      "step": 36375
    },
    {
      "epoch": 19.067085953878408,
      "grad_norm": 0.742325484752655,
      "learning_rate": 3.310358768571997e-07,
      "loss": 0.0071,
      "num_input_tokens_seen": 23727680,
      "step": 36380
    },
    {
      "epoch": 19.06970649895178,
      "grad_norm": 6.941628456115723,
      "learning_rate": 3.2918386215339827e-07,
      "loss": 0.0225,
      "num_input_tokens_seen": 23730464,
      "step": 36385
    },
    {
      "epoch": 19.072327044025158,
      "grad_norm": 0.22437618672847748,
      "learning_rate": 3.2733700830287205e-07,
      "loss": 0.0025,
      "num_input_tokens_seen": 23733888,
      "step": 36390
    },
    {
      "epoch": 19.07494758909853,
      "grad_norm": 0.3872188925743103,
      "learning_rate": 3.254953156919649e-07,
      "loss": 0.0037,
      "num_input_tokens_seen": 23736736,
      "step": 36395
    },
    {
      "epoch": 19.07756813417191,
      "grad_norm": 0.7046974897384644,
      "learning_rate": 3.2365878470593527e-07,
      "loss": 0.0061,
      "num_input_tokens_seen": 23739872,
      "step": 36400
    },
    {
      "epoch": 19.080188679245282,
      "grad_norm": 0.9306637048721313,
      "learning_rate": 3.2182741572896747e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 23742592,
      "step": 36405
    },
    {
      "epoch": 19.08280922431866,
      "grad_norm": 0.24866437911987305,
      "learning_rate": 3.2000120914415786e-07,
      "loss": 0.0115,
      "num_input_tokens_seen": 23747296,
      "step": 36410
    },
    {
      "epoch": 19.085429769392032,
      "grad_norm": 0.621563732624054,
      "learning_rate": 3.181801653335315e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 23751040,
      "step": 36415
    },
    {
      "epoch": 19.08805031446541,
      "grad_norm": 1.7821385860443115,
      "learning_rate": 3.163642846780307e-07,
      "loss": 0.0073,
      "num_input_tokens_seen": 23754976,
      "step": 36420
    },
    {
      "epoch": 19.090670859538783,
      "grad_norm": 0.4566211700439453,
      "learning_rate": 3.1455356755751565e-07,
      "loss": 0.0136,
      "num_input_tokens_seen": 23758464,
      "step": 36425
    },
    {
      "epoch": 19.09329140461216,
      "grad_norm": 0.44157081842422485,
      "learning_rate": 3.127480143507666e-07,
      "loss": 0.0091,
      "num_input_tokens_seen": 23762880,
      "step": 36430
    },
    {
      "epoch": 19.095911949685533,
      "grad_norm": 0.7179908752441406,
      "learning_rate": 3.1094762543548704e-07,
      "loss": 0.005,
      "num_input_tokens_seen": 23765664,
      "step": 36435
    },
    {
      "epoch": 19.09853249475891,
      "grad_norm": 0.4045579135417938,
      "learning_rate": 3.091524011882951e-07,
      "loss": 0.0278,
      "num_input_tokens_seen": 23770016,
      "step": 36440
    },
    {
      "epoch": 19.101153039832287,
      "grad_norm": 0.2964622378349304,
      "learning_rate": 3.073623419847349e-07,
      "loss": 0.0204,
      "num_input_tokens_seen": 23773408,
      "step": 36445
    },
    {
      "epoch": 19.10377358490566,
      "grad_norm": 0.42192283272743225,
      "learning_rate": 3.0557744819926235e-07,
      "loss": 0.0039,
      "num_input_tokens_seen": 23776672,
      "step": 36450
    },
    {
      "epoch": 19.106394129979037,
      "grad_norm": 0.053325653076171875,
      "learning_rate": 3.037977202052594e-07,
      "loss": 0.0016,
      "num_input_tokens_seen": 23779264,
      "step": 36455
    },
    {
      "epoch": 19.10901467505241,
      "grad_norm": 0.32733842730522156,
      "learning_rate": 3.020231583750255e-07,
      "loss": 0.0039,
      "num_input_tokens_seen": 23783104,
      "step": 36460
    },
    {
      "epoch": 19.111635220125788,
      "grad_norm": 0.16550537943840027,
      "learning_rate": 3.002537630797747e-07,
      "loss": 0.0023,
      "num_input_tokens_seen": 23785856,
      "step": 36465
    },
    {
      "epoch": 19.11425576519916,
      "grad_norm": 0.16351011395454407,
      "learning_rate": 2.984895346896499e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 23789344,
      "step": 36470
    },
    {
      "epoch": 19.116876310272538,
      "grad_norm": 0.7985907196998596,
      "learning_rate": 2.967304735737031e-07,
      "loss": 0.008,
      "num_input_tokens_seen": 23793056,
      "step": 36475
    },
    {
      "epoch": 19.11949685534591,
      "grad_norm": 0.08691010624170303,
      "learning_rate": 2.9497658009991214e-07,
      "loss": 0.0016,
      "num_input_tokens_seen": 23795552,
      "step": 36480
    },
    {
      "epoch": 19.122117400419288,
      "grad_norm": 0.29853355884552,
      "learning_rate": 2.932278546351724e-07,
      "loss": 0.0019,
      "num_input_tokens_seen": 23797920,
      "step": 36485
    },
    {
      "epoch": 19.12473794549266,
      "grad_norm": 0.703374981880188,
      "learning_rate": 2.9148429754529137e-07,
      "loss": 0.006,
      "num_input_tokens_seen": 23800864,
      "step": 36490
    },
    {
      "epoch": 19.12735849056604,
      "grad_norm": 0.8070768713951111,
      "learning_rate": 2.897459091950078e-07,
      "loss": 0.0114,
      "num_input_tokens_seen": 23804480,
      "step": 36495
    },
    {
      "epoch": 19.129979035639412,
      "grad_norm": 0.1369275152683258,
      "learning_rate": 2.880126899479724e-07,
      "loss": 0.0021,
      "num_input_tokens_seen": 23807296,
      "step": 36500
    },
    {
      "epoch": 19.13259958071279,
      "grad_norm": 0.20868083834648132,
      "learning_rate": 2.862846401667507e-07,
      "loss": 0.0034,
      "num_input_tokens_seen": 23810432,
      "step": 36505
    },
    {
      "epoch": 19.135220125786162,
      "grad_norm": 0.19978809356689453,
      "learning_rate": 2.8456176021283697e-07,
      "loss": 0.0034,
      "num_input_tokens_seen": 23813088,
      "step": 36510
    },
    {
      "epoch": 19.13784067085954,
      "grad_norm": 0.11419611424207687,
      "learning_rate": 2.8284405044662897e-07,
      "loss": 0.0315,
      "num_input_tokens_seen": 23815968,
      "step": 36515
    },
    {
      "epoch": 19.140461215932913,
      "grad_norm": 0.29920223355293274,
      "learning_rate": 2.8113151122745874e-07,
      "loss": 0.0029,
      "num_input_tokens_seen": 23819392,
      "step": 36520
    },
    {
      "epoch": 19.14308176100629,
      "grad_norm": 0.17375461757183075,
      "learning_rate": 2.7942414291357023e-07,
      "loss": 0.0105,
      "num_input_tokens_seen": 23822144,
      "step": 36525
    },
    {
      "epoch": 19.145702306079663,
      "grad_norm": 0.20915226638317108,
      "learning_rate": 2.777219458621222e-07,
      "loss": 0.0031,
      "num_input_tokens_seen": 23826144,
      "step": 36530
    },
    {
      "epoch": 19.14832285115304,
      "grad_norm": 0.4692058265209198,
      "learning_rate": 2.7602492042919923e-07,
      "loss": 0.002,
      "num_input_tokens_seen": 23829536,
      "step": 36535
    },
    {
      "epoch": 19.150943396226417,
      "grad_norm": 0.3055820167064667,
      "learning_rate": 2.7433306696979523e-07,
      "loss": 0.0022,
      "num_input_tokens_seen": 23833408,
      "step": 36540
    },
    {
      "epoch": 19.15356394129979,
      "grad_norm": 0.10176043957471848,
      "learning_rate": 2.7264638583782973e-07,
      "loss": 0.0016,
      "num_input_tokens_seen": 23835808,
      "step": 36545
    },
    {
      "epoch": 19.156184486373167,
      "grad_norm": 0.3083440363407135,
      "learning_rate": 2.7096487738613177e-07,
      "loss": 0.03,
      "num_input_tokens_seen": 23838624,
      "step": 36550
    },
    {
      "epoch": 19.15880503144654,
      "grad_norm": 0.16325730085372925,
      "learning_rate": 2.6928854196646147e-07,
      "loss": 0.0025,
      "num_input_tokens_seen": 23841312,
      "step": 36555
    },
    {
      "epoch": 19.161425576519918,
      "grad_norm": 0.4569627046585083,
      "learning_rate": 2.6761737992948565e-07,
      "loss": 0.0018,
      "num_input_tokens_seen": 23844416,
      "step": 36560
    },
    {
      "epoch": 19.16404612159329,
      "grad_norm": 0.6563942432403564,
      "learning_rate": 2.6595139162479134e-07,
      "loss": 0.004,
      "num_input_tokens_seen": 23847744,
      "step": 36565
    },
    {
      "epoch": 19.166666666666668,
      "grad_norm": 2.557725191116333,
      "learning_rate": 2.6429057740088303e-07,
      "loss": 0.0257,
      "num_input_tokens_seen": 23850144,
      "step": 36570
    },
    {
      "epoch": 19.16928721174004,
      "grad_norm": 3.2201156616210938,
      "learning_rate": 2.6263493760518564e-07,
      "loss": 0.0372,
      "num_input_tokens_seen": 23853952,
      "step": 36575
    },
    {
      "epoch": 19.171907756813418,
      "grad_norm": 0.44866612553596497,
      "learning_rate": 2.6098447258404157e-07,
      "loss": 0.0093,
      "num_input_tokens_seen": 23857376,
      "step": 36580
    },
    {
      "epoch": 19.17452830188679,
      "grad_norm": 0.949286699295044,
      "learning_rate": 2.593391826827107e-07,
      "loss": 0.0093,
      "num_input_tokens_seen": 23861056,
      "step": 36585
    },
    {
      "epoch": 19.17714884696017,
      "grad_norm": 0.16220076382160187,
      "learning_rate": 2.5769906824536225e-07,
      "loss": 0.0019,
      "num_input_tokens_seen": 23863968,
      "step": 36590
    },
    {
      "epoch": 19.179769392033542,
      "grad_norm": 0.46297189593315125,
      "learning_rate": 2.560641296150967e-07,
      "loss": 0.0033,
      "num_input_tokens_seen": 23866912,
      "step": 36595
    },
    {
      "epoch": 19.18238993710692,
      "grad_norm": 0.5222681760787964,
      "learning_rate": 2.544343671339211e-07,
      "loss": 0.0032,
      "num_input_tokens_seen": 23869856,
      "step": 36600
    },
    {
      "epoch": 19.185010482180292,
      "grad_norm": 0.14116930961608887,
      "learning_rate": 2.5280978114276264e-07,
      "loss": 0.0024,
      "num_input_tokens_seen": 23873088,
      "step": 36605
    },
    {
      "epoch": 19.18763102725367,
      "grad_norm": 0.22554320096969604,
      "learning_rate": 2.511903719814662e-07,
      "loss": 0.0105,
      "num_input_tokens_seen": 23876160,
      "step": 36610
    },
    {
      "epoch": 19.190251572327043,
      "grad_norm": 3.8502845764160156,
      "learning_rate": 2.4957613998879695e-07,
      "loss": 0.0145,
      "num_input_tokens_seen": 23878880,
      "step": 36615
    },
    {
      "epoch": 19.19287211740042,
      "grad_norm": 0.12931139767169952,
      "learning_rate": 2.479670855024291e-07,
      "loss": 0.0111,
      "num_input_tokens_seen": 23882912,
      "step": 36620
    },
    {
      "epoch": 19.195492662473793,
      "grad_norm": 0.45267733931541443,
      "learning_rate": 2.463632088589629e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 23886560,
      "step": 36625
    },
    {
      "epoch": 19.19811320754717,
      "grad_norm": 3.4318528175354004,
      "learning_rate": 2.447645103939078e-07,
      "loss": 0.0163,
      "num_input_tokens_seen": 23889728,
      "step": 36630
    },
    {
      "epoch": 19.200733752620547,
      "grad_norm": 0.6213622689247131,
      "learning_rate": 2.431709904416962e-07,
      "loss": 0.0057,
      "num_input_tokens_seen": 23893408,
      "step": 36635
    },
    {
      "epoch": 19.20335429769392,
      "grad_norm": 0.1373426765203476,
      "learning_rate": 2.4158264933567264e-07,
      "loss": 0.0025,
      "num_input_tokens_seen": 23896960,
      "step": 36640
    },
    {
      "epoch": 19.205974842767297,
      "grad_norm": 0.7353290319442749,
      "learning_rate": 2.399994874080963e-07,
      "loss": 0.002,
      "num_input_tokens_seen": 23899936,
      "step": 36645
    },
    {
      "epoch": 19.20859538784067,
      "grad_norm": 0.4963071942329407,
      "learning_rate": 2.3842150499015504e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 23903360,
      "step": 36650
    },
    {
      "epoch": 19.211215932914047,
      "grad_norm": 0.1352425515651703,
      "learning_rate": 2.368487024119348e-07,
      "loss": 0.0025,
      "num_input_tokens_seen": 23906784,
      "step": 36655
    },
    {
      "epoch": 19.21383647798742,
      "grad_norm": 0.12355730682611465,
      "learning_rate": 2.3528108000245573e-07,
      "loss": 0.0048,
      "num_input_tokens_seen": 23910240,
      "step": 36660
    },
    {
      "epoch": 19.216457023060798,
      "grad_norm": 0.23355819284915924,
      "learning_rate": 2.3371863808964445e-07,
      "loss": 0.0051,
      "num_input_tokens_seen": 23913888,
      "step": 36665
    },
    {
      "epoch": 19.21907756813417,
      "grad_norm": 0.16011129319667816,
      "learning_rate": 2.32161377000345e-07,
      "loss": 0.0068,
      "num_input_tokens_seen": 23917312,
      "step": 36670
    },
    {
      "epoch": 19.221698113207548,
      "grad_norm": 0.5063791871070862,
      "learning_rate": 2.306092970603163e-07,
      "loss": 0.0022,
      "num_input_tokens_seen": 23920032,
      "step": 36675
    },
    {
      "epoch": 19.22431865828092,
      "grad_norm": 0.3123050928115845,
      "learning_rate": 2.2906239859424306e-07,
      "loss": 0.0055,
      "num_input_tokens_seen": 23924000,
      "step": 36680
    },
    {
      "epoch": 19.2269392033543,
      "grad_norm": 0.5614033937454224,
      "learning_rate": 2.2752068192571084e-07,
      "loss": 0.0054,
      "num_input_tokens_seen": 23928128,
      "step": 36685
    },
    {
      "epoch": 19.229559748427672,
      "grad_norm": 0.24076588451862335,
      "learning_rate": 2.2598414737723396e-07,
      "loss": 0.0075,
      "num_input_tokens_seen": 23931712,
      "step": 36690
    },
    {
      "epoch": 19.23218029350105,
      "grad_norm": 2.522092580795288,
      "learning_rate": 2.2445279527023578e-07,
      "loss": 0.0521,
      "num_input_tokens_seen": 23934912,
      "step": 36695
    },
    {
      "epoch": 19.234800838574422,
      "grad_norm": 3.1774630546569824,
      "learning_rate": 2.2292662592506007e-07,
      "loss": 0.0165,
      "num_input_tokens_seen": 23938560,
      "step": 36700
    },
    {
      "epoch": 19.2374213836478,
      "grad_norm": 0.8696791529655457,
      "learning_rate": 2.2140563966095985e-07,
      "loss": 0.0096,
      "num_input_tokens_seen": 23941440,
      "step": 36705
    },
    {
      "epoch": 19.240041928721173,
      "grad_norm": 0.2415711134672165,
      "learning_rate": 2.1988983679611386e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 23944736,
      "step": 36710
    },
    {
      "epoch": 19.24266247379455,
      "grad_norm": 0.8919480443000793,
      "learning_rate": 2.1837921764760462e-07,
      "loss": 0.048,
      "num_input_tokens_seen": 23949376,
      "step": 36715
    },
    {
      "epoch": 19.245283018867923,
      "grad_norm": 0.7609349489212036,
      "learning_rate": 2.1687378253144043e-07,
      "loss": 0.0059,
      "num_input_tokens_seen": 23952640,
      "step": 36720
    },
    {
      "epoch": 19.2479035639413,
      "grad_norm": 0.48438000679016113,
      "learning_rate": 2.153735317625416e-07,
      "loss": 0.0107,
      "num_input_tokens_seen": 23956992,
      "step": 36725
    },
    {
      "epoch": 19.250524109014677,
      "grad_norm": 0.30232131481170654,
      "learning_rate": 2.1387846565474045e-07,
      "loss": 0.0028,
      "num_input_tokens_seen": 23960192,
      "step": 36730
    },
    {
      "epoch": 19.25314465408805,
      "grad_norm": 0.1469998061656952,
      "learning_rate": 2.123885845207896e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 23963040,
      "step": 36735
    },
    {
      "epoch": 19.255765199161427,
      "grad_norm": 0.1429663598537445,
      "learning_rate": 2.1090388867235356e-07,
      "loss": 0.0041,
      "num_input_tokens_seen": 23965664,
      "step": 36740
    },
    {
      "epoch": 19.2583857442348,
      "grad_norm": 0.5115342140197754,
      "learning_rate": 2.0942437842001728e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 23968992,
      "step": 36745
    },
    {
      "epoch": 19.261006289308177,
      "grad_norm": 0.5821546912193298,
      "learning_rate": 2.0795005407327483e-07,
      "loss": 0.0894,
      "num_input_tokens_seen": 23972032,
      "step": 36750
    },
    {
      "epoch": 19.26362683438155,
      "grad_norm": 0.38531219959259033,
      "learning_rate": 2.0648091594053508e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 23975040,
      "step": 36755
    },
    {
      "epoch": 19.266247379454928,
      "grad_norm": 0.33671605587005615,
      "learning_rate": 2.0501696432913275e-07,
      "loss": 0.0082,
      "num_input_tokens_seen": 23977408,
      "step": 36760
    },
    {
      "epoch": 19.2688679245283,
      "grad_norm": 0.39665380120277405,
      "learning_rate": 2.035581995453034e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 23980032,
      "step": 36765
    },
    {
      "epoch": 19.271488469601678,
      "grad_norm": 0.6251835823059082,
      "learning_rate": 2.0210462189420577e-07,
      "loss": 0.0035,
      "num_input_tokens_seen": 23982720,
      "step": 36770
    },
    {
      "epoch": 19.27410901467505,
      "grad_norm": 0.8591946959495544,
      "learning_rate": 2.006562316799132e-07,
      "loss": 0.0101,
      "num_input_tokens_seen": 23985984,
      "step": 36775
    },
    {
      "epoch": 19.27672955974843,
      "grad_norm": 1.4055733680725098,
      "learning_rate": 1.992130292054084e-07,
      "loss": 0.0163,
      "num_input_tokens_seen": 23989504,
      "step": 36780
    },
    {
      "epoch": 19.279350104821802,
      "grad_norm": 0.12322840094566345,
      "learning_rate": 1.9777501477259708e-07,
      "loss": 0.0094,
      "num_input_tokens_seen": 23992160,
      "step": 36785
    },
    {
      "epoch": 19.28197064989518,
      "grad_norm": 0.9225424528121948,
      "learning_rate": 1.9634218868229414e-07,
      "loss": 0.0269,
      "num_input_tokens_seen": 23998816,
      "step": 36790
    },
    {
      "epoch": 19.284591194968552,
      "grad_norm": 0.8124634027481079,
      "learning_rate": 1.9491455123423207e-07,
      "loss": 0.0072,
      "num_input_tokens_seen": 24003040,
      "step": 36795
    },
    {
      "epoch": 19.28721174004193,
      "grad_norm": 0.5083751678466797,
      "learning_rate": 1.9349210272705253e-07,
      "loss": 0.0068,
      "num_input_tokens_seen": 24006272,
      "step": 36800
    },
    {
      "epoch": 19.289832285115303,
      "grad_norm": 0.2492956817150116,
      "learning_rate": 1.920748434583175e-07,
      "loss": 0.0192,
      "num_input_tokens_seen": 24009792,
      "step": 36805
    },
    {
      "epoch": 19.29245283018868,
      "grad_norm": 0.515361487865448,
      "learning_rate": 1.906627737245009e-07,
      "loss": 0.0033,
      "num_input_tokens_seen": 24012960,
      "step": 36810
    },
    {
      "epoch": 19.295073375262053,
      "grad_norm": 0.35303327441215515,
      "learning_rate": 1.8925589382099152e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 24016416,
      "step": 36815
    },
    {
      "epoch": 19.29769392033543,
      "grad_norm": 0.3206177353858948,
      "learning_rate": 1.8785420404209553e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 24019520,
      "step": 36820
    },
    {
      "epoch": 19.300314465408803,
      "grad_norm": 0.594407856464386,
      "learning_rate": 1.864577046810284e-07,
      "loss": 0.0068,
      "num_input_tokens_seen": 24023168,
      "step": 36825
    },
    {
      "epoch": 19.30293501048218,
      "grad_norm": 0.4324367940425873,
      "learning_rate": 1.8506639602992038e-07,
      "loss": 0.0032,
      "num_input_tokens_seen": 24026368,
      "step": 36830
    },
    {
      "epoch": 19.305555555555557,
      "grad_norm": 0.06977701187133789,
      "learning_rate": 1.836802783798164e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 24028960,
      "step": 36835
    },
    {
      "epoch": 19.30817610062893,
      "grad_norm": 0.322182297706604,
      "learning_rate": 1.822993520206817e-07,
      "loss": 0.0037,
      "num_input_tokens_seen": 24031392,
      "step": 36840
    },
    {
      "epoch": 19.310796645702307,
      "grad_norm": 0.29945504665374756,
      "learning_rate": 1.8092361724138528e-07,
      "loss": 0.0032,
      "num_input_tokens_seen": 24034368,
      "step": 36845
    },
    {
      "epoch": 19.31341719077568,
      "grad_norm": 0.18179979920387268,
      "learning_rate": 1.7955307432971914e-07,
      "loss": 0.0074,
      "num_input_tokens_seen": 24037472,
      "step": 36850
    },
    {
      "epoch": 19.316037735849058,
      "grad_norm": 1.6297842264175415,
      "learning_rate": 1.7818772357238445e-07,
      "loss": 0.0069,
      "num_input_tokens_seen": 24040672,
      "step": 36855
    },
    {
      "epoch": 19.31865828092243,
      "grad_norm": 0.1883315145969391,
      "learning_rate": 1.7682756525499166e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 24044192,
      "step": 36860
    },
    {
      "epoch": 19.321278825995808,
      "grad_norm": 1.3420958518981934,
      "learning_rate": 1.7547259966207708e-07,
      "loss": 0.0054,
      "num_input_tokens_seen": 24047520,
      "step": 36865
    },
    {
      "epoch": 19.32389937106918,
      "grad_norm": 0.8124817609786987,
      "learning_rate": 1.7412282707708062e-07,
      "loss": 0.0155,
      "num_input_tokens_seen": 24050944,
      "step": 36870
    },
    {
      "epoch": 19.32651991614256,
      "grad_norm": 0.27334001660346985,
      "learning_rate": 1.7277824778235974e-07,
      "loss": 0.0017,
      "num_input_tokens_seen": 24053440,
      "step": 36875
    },
    {
      "epoch": 19.329140461215932,
      "grad_norm": 0.21241305768489838,
      "learning_rate": 1.714388620591867e-07,
      "loss": 0.0043,
      "num_input_tokens_seen": 24056480,
      "step": 36880
    },
    {
      "epoch": 19.33176100628931,
      "grad_norm": 0.30458924174308777,
      "learning_rate": 1.701046701877429e-07,
      "loss": 0.005,
      "num_input_tokens_seen": 24060768,
      "step": 36885
    },
    {
      "epoch": 19.334381551362682,
      "grad_norm": 0.5188015103340149,
      "learning_rate": 1.6877567244713011e-07,
      "loss": 0.0035,
      "num_input_tokens_seen": 24063200,
      "step": 36890
    },
    {
      "epoch": 19.33700209643606,
      "grad_norm": 0.9672465920448303,
      "learning_rate": 1.6745186911535372e-07,
      "loss": 0.0286,
      "num_input_tokens_seen": 24066272,
      "step": 36895
    },
    {
      "epoch": 19.339622641509433,
      "grad_norm": 0.5466281175613403,
      "learning_rate": 1.6613326046934497e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 24068960,
      "step": 36900
    },
    {
      "epoch": 19.34224318658281,
      "grad_norm": 0.64110267162323,
      "learning_rate": 1.64819846784936e-07,
      "loss": 0.0112,
      "num_input_tokens_seen": 24071936,
      "step": 36905
    },
    {
      "epoch": 19.344863731656183,
      "grad_norm": 1.3487931489944458,
      "learning_rate": 1.6351162833688204e-07,
      "loss": 0.0034,
      "num_input_tokens_seen": 24074752,
      "step": 36910
    },
    {
      "epoch": 19.34748427672956,
      "grad_norm": 0.5908254384994507,
      "learning_rate": 1.6220860539884475e-07,
      "loss": 0.0053,
      "num_input_tokens_seen": 24078240,
      "step": 36915
    },
    {
      "epoch": 19.350104821802937,
      "grad_norm": 0.4247349500656128,
      "learning_rate": 1.6091077824340327e-07,
      "loss": 0.0031,
      "num_input_tokens_seen": 24081920,
      "step": 36920
    },
    {
      "epoch": 19.35272536687631,
      "grad_norm": 0.22709383070468903,
      "learning_rate": 1.5961814714204605e-07,
      "loss": 0.0088,
      "num_input_tokens_seen": 24085952,
      "step": 36925
    },
    {
      "epoch": 19.355345911949687,
      "grad_norm": 0.9792500138282776,
      "learning_rate": 1.5833071236518172e-07,
      "loss": 0.0073,
      "num_input_tokens_seen": 24088672,
      "step": 36930
    },
    {
      "epoch": 19.35796645702306,
      "grad_norm": 0.8113270401954651,
      "learning_rate": 1.5704847418212266e-07,
      "loss": 0.0047,
      "num_input_tokens_seen": 24092064,
      "step": 36935
    },
    {
      "epoch": 19.360587002096437,
      "grad_norm": 1.018370509147644,
      "learning_rate": 1.5577143286109875e-07,
      "loss": 0.0091,
      "num_input_tokens_seen": 24095328,
      "step": 36940
    },
    {
      "epoch": 19.36320754716981,
      "grad_norm": 0.6854138970375061,
      "learning_rate": 1.5449958866925185e-07,
      "loss": 0.0038,
      "num_input_tokens_seen": 24098560,
      "step": 36945
    },
    {
      "epoch": 19.365828092243188,
      "grad_norm": 1.6245956420898438,
      "learning_rate": 1.532329418726386e-07,
      "loss": 0.0135,
      "num_input_tokens_seen": 24101696,
      "step": 36950
    },
    {
      "epoch": 19.36844863731656,
      "grad_norm": 0.12527212500572205,
      "learning_rate": 1.5197149273623036e-07,
      "loss": 0.0261,
      "num_input_tokens_seen": 24104448,
      "step": 36955
    },
    {
      "epoch": 19.371069182389938,
      "grad_norm": 0.15730999410152435,
      "learning_rate": 1.5071524152389936e-07,
      "loss": 0.0074,
      "num_input_tokens_seen": 24107520,
      "step": 36960
    },
    {
      "epoch": 19.37368972746331,
      "grad_norm": 0.884133517742157,
      "learning_rate": 1.4946418849844934e-07,
      "loss": 0.0041,
      "num_input_tokens_seen": 24110304,
      "step": 36965
    },
    {
      "epoch": 19.37631027253669,
      "grad_norm": 0.31717735528945923,
      "learning_rate": 1.4821833392157923e-07,
      "loss": 0.0033,
      "num_input_tokens_seen": 24113792,
      "step": 36970
    },
    {
      "epoch": 19.378930817610062,
      "grad_norm": 0.227358877658844,
      "learning_rate": 1.4697767805390838e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 24116224,
      "step": 36975
    },
    {
      "epoch": 19.38155136268344,
      "grad_norm": 0.44603002071380615,
      "learning_rate": 1.4574222115497084e-07,
      "loss": 0.0017,
      "num_input_tokens_seen": 24120480,
      "step": 36980
    },
    {
      "epoch": 19.384171907756812,
      "grad_norm": 0.33086585998535156,
      "learning_rate": 1.4451196348320706e-07,
      "loss": 0.0114,
      "num_input_tokens_seen": 24123424,
      "step": 36985
    },
    {
      "epoch": 19.38679245283019,
      "grad_norm": 1.415076494216919,
      "learning_rate": 1.4328690529597232e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 24126528,
      "step": 36990
    },
    {
      "epoch": 19.389412997903563,
      "grad_norm": 0.88449627161026,
      "learning_rate": 1.4206704684953943e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 24131360,
      "step": 36995
    },
    {
      "epoch": 19.39203354297694,
      "grad_norm": 1.0967544317245483,
      "learning_rate": 1.4085238839908476e-07,
      "loss": 0.0068,
      "num_input_tokens_seen": 24133920,
      "step": 37000
    },
    {
      "epoch": 19.394654088050313,
      "grad_norm": 0.07836625725030899,
      "learning_rate": 1.3964293019870235e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 24137536,
      "step": 37005
    },
    {
      "epoch": 19.39727463312369,
      "grad_norm": 1.5929652452468872,
      "learning_rate": 1.3843867250139807e-07,
      "loss": 0.0057,
      "num_input_tokens_seen": 24140448,
      "step": 37010
    },
    {
      "epoch": 19.399895178197063,
      "grad_norm": 0.4503779709339142,
      "learning_rate": 1.3723961555908715e-07,
      "loss": 0.0127,
      "num_input_tokens_seen": 24143488,
      "step": 37015
    },
    {
      "epoch": 19.40251572327044,
      "grad_norm": 0.2569425702095032,
      "learning_rate": 1.3604575962259947e-07,
      "loss": 0.0041,
      "num_input_tokens_seen": 24146944,
      "step": 37020
    },
    {
      "epoch": 19.405136268343817,
      "grad_norm": 1.3043417930603027,
      "learning_rate": 1.3485710494167692e-07,
      "loss": 0.0083,
      "num_input_tokens_seen": 24150944,
      "step": 37025
    },
    {
      "epoch": 19.40775681341719,
      "grad_norm": 0.28878921270370483,
      "learning_rate": 1.3367365176496783e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 24153504,
      "step": 37030
    },
    {
      "epoch": 19.410377358490567,
      "grad_norm": 0.7366212606430054,
      "learning_rate": 1.324954003400436e-07,
      "loss": 0.0063,
      "num_input_tokens_seen": 24156480,
      "step": 37035
    },
    {
      "epoch": 19.41299790356394,
      "grad_norm": 0.29576587677001953,
      "learning_rate": 1.313223509133793e-07,
      "loss": 0.0058,
      "num_input_tokens_seen": 24160192,
      "step": 37040
    },
    {
      "epoch": 19.415618448637318,
      "grad_norm": 0.11474431306123734,
      "learning_rate": 1.301545037303592e-07,
      "loss": 0.0024,
      "num_input_tokens_seen": 24162560,
      "step": 37045
    },
    {
      "epoch": 19.41823899371069,
      "grad_norm": 0.897369384765625,
      "learning_rate": 1.2899185903529066e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 24165568,
      "step": 37050
    },
    {
      "epoch": 19.420859538784068,
      "grad_norm": 1.010769248008728,
      "learning_rate": 1.278344170713791e-07,
      "loss": 0.0071,
      "num_input_tokens_seen": 24168608,
      "step": 37055
    },
    {
      "epoch": 19.42348008385744,
      "grad_norm": 0.29598748683929443,
      "learning_rate": 1.2668217808075033e-07,
      "loss": 0.0055,
      "num_input_tokens_seen": 24171552,
      "step": 37060
    },
    {
      "epoch": 19.42610062893082,
      "grad_norm": 0.7951231002807617,
      "learning_rate": 1.255351423044393e-07,
      "loss": 0.0049,
      "num_input_tokens_seen": 24174336,
      "step": 37065
    },
    {
      "epoch": 19.428721174004192,
      "grad_norm": 0.19954517483711243,
      "learning_rate": 1.2439330998239572e-07,
      "loss": 0.0031,
      "num_input_tokens_seen": 24178400,
      "step": 37070
    },
    {
      "epoch": 19.43134171907757,
      "grad_norm": 0.1729251891374588,
      "learning_rate": 1.2325668135347578e-07,
      "loss": 0.0106,
      "num_input_tokens_seen": 24181312,
      "step": 37075
    },
    {
      "epoch": 19.433962264150942,
      "grad_norm": 0.11988227069377899,
      "learning_rate": 1.221252566554476e-07,
      "loss": 0.076,
      "num_input_tokens_seen": 24184544,
      "step": 37080
    },
    {
      "epoch": 19.43658280922432,
      "grad_norm": 0.3352515995502472,
      "learning_rate": 1.2099903612499685e-07,
      "loss": 0.0019,
      "num_input_tokens_seen": 24186720,
      "step": 37085
    },
    {
      "epoch": 19.439203354297693,
      "grad_norm": 0.250588595867157,
      "learning_rate": 1.198780199977101e-07,
      "loss": 0.0135,
      "num_input_tokens_seen": 24189984,
      "step": 37090
    },
    {
      "epoch": 19.44182389937107,
      "grad_norm": 0.14208613336086273,
      "learning_rate": 1.1876220850809416e-07,
      "loss": 0.0048,
      "num_input_tokens_seen": 24192448,
      "step": 37095
    },
    {
      "epoch": 19.444444444444443,
      "grad_norm": 0.2995854616165161,
      "learning_rate": 1.1765160188956514e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 24196128,
      "step": 37100
    },
    {
      "epoch": 19.44706498951782,
      "grad_norm": 0.05755804479122162,
      "learning_rate": 1.1654620037445107e-07,
      "loss": 0.0019,
      "num_input_tokens_seen": 24199104,
      "step": 37105
    },
    {
      "epoch": 19.449685534591197,
      "grad_norm": 0.11222023516893387,
      "learning_rate": 1.1544600419398088e-07,
      "loss": 0.0055,
      "num_input_tokens_seen": 24202176,
      "step": 37110
    },
    {
      "epoch": 19.45230607966457,
      "grad_norm": 0.5000672340393066,
      "learning_rate": 1.1435101357831213e-07,
      "loss": 0.0053,
      "num_input_tokens_seen": 24205824,
      "step": 37115
    },
    {
      "epoch": 19.454926624737947,
      "grad_norm": 0.37852370738983154,
      "learning_rate": 1.1326122875650047e-07,
      "loss": 0.0155,
      "num_input_tokens_seen": 24208640,
      "step": 37120
    },
    {
      "epoch": 19.45754716981132,
      "grad_norm": 0.11466235667467117,
      "learning_rate": 1.1217664995651633e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 24215456,
      "step": 37125
    },
    {
      "epoch": 19.460167714884697,
      "grad_norm": 0.27922186255455017,
      "learning_rate": 1.1109727740523935e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 24218752,
      "step": 37130
    },
    {
      "epoch": 19.46278825995807,
      "grad_norm": 2.807563066482544,
      "learning_rate": 1.1002311132846944e-07,
      "loss": 0.0043,
      "num_input_tokens_seen": 24221632,
      "step": 37135
    },
    {
      "epoch": 19.465408805031448,
      "grad_norm": 0.3659830689430237,
      "learning_rate": 1.0895415195090186e-07,
      "loss": 0.0134,
      "num_input_tokens_seen": 24225824,
      "step": 37140
    },
    {
      "epoch": 19.46802935010482,
      "grad_norm": 0.32416778802871704,
      "learning_rate": 1.0789039949615221e-07,
      "loss": 0.0076,
      "num_input_tokens_seen": 24228960,
      "step": 37145
    },
    {
      "epoch": 19.470649895178198,
      "grad_norm": 0.7050501704216003,
      "learning_rate": 1.0683185418674801e-07,
      "loss": 0.0037,
      "num_input_tokens_seen": 24232352,
      "step": 37150
    },
    {
      "epoch": 19.47327044025157,
      "grad_norm": 0.24104726314544678,
      "learning_rate": 1.0577851624412604e-07,
      "loss": 0.0027,
      "num_input_tokens_seen": 24235456,
      "step": 37155
    },
    {
      "epoch": 19.47589098532495,
      "grad_norm": 0.4337958097457886,
      "learning_rate": 1.0473038588862672e-07,
      "loss": 0.0043,
      "num_input_tokens_seen": 24238016,
      "step": 37160
    },
    {
      "epoch": 19.478511530398322,
      "grad_norm": 0.029929080978035927,
      "learning_rate": 1.0368746333951351e-07,
      "loss": 0.015,
      "num_input_tokens_seen": 24240672,
      "step": 37165
    },
    {
      "epoch": 19.4811320754717,
      "grad_norm": 0.16667981445789337,
      "learning_rate": 1.0264974881494805e-07,
      "loss": 0.0095,
      "num_input_tokens_seen": 24243520,
      "step": 37170
    },
    {
      "epoch": 19.483752620545072,
      "grad_norm": 0.2899142801761627,
      "learning_rate": 1.0161724253201222e-07,
      "loss": 0.0131,
      "num_input_tokens_seen": 24246784,
      "step": 37175
    },
    {
      "epoch": 19.48637316561845,
      "grad_norm": 0.6173955798149109,
      "learning_rate": 1.0058994470669436e-07,
      "loss": 0.0237,
      "num_input_tokens_seen": 24250848,
      "step": 37180
    },
    {
      "epoch": 19.488993710691823,
      "grad_norm": 0.14492860436439514,
      "learning_rate": 9.956785555388926e-08,
      "loss": 0.0024,
      "num_input_tokens_seen": 24254208,
      "step": 37185
    },
    {
      "epoch": 19.4916142557652,
      "grad_norm": 1.057827115058899,
      "learning_rate": 9.855097528741475e-08,
      "loss": 0.0241,
      "num_input_tokens_seen": 24260896,
      "step": 37190
    },
    {
      "epoch": 19.494234800838573,
      "grad_norm": 3.1472384929656982,
      "learning_rate": 9.753930411998402e-08,
      "loss": 0.0051,
      "num_input_tokens_seen": 24263232,
      "step": 37195
    },
    {
      "epoch": 19.49685534591195,
      "grad_norm": 0.9424697160720825,
      "learning_rate": 9.653284226322501e-08,
      "loss": 0.0105,
      "num_input_tokens_seen": 24266784,
      "step": 37200
    },
    {
      "epoch": 19.499475890985323,
      "grad_norm": 0.3748774230480194,
      "learning_rate": 9.553158992768318e-08,
      "loss": 0.003,
      "num_input_tokens_seen": 24270112,
      "step": 37205
    },
    {
      "epoch": 19.5020964360587,
      "grad_norm": 1.2231062650680542,
      "learning_rate": 9.453554732281044e-08,
      "loss": 0.0075,
      "num_input_tokens_seen": 24273184,
      "step": 37210
    },
    {
      "epoch": 19.504716981132077,
      "grad_norm": 0.46253716945648193,
      "learning_rate": 9.354471465696236e-08,
      "loss": 0.0078,
      "num_input_tokens_seen": 24276224,
      "step": 37215
    },
    {
      "epoch": 19.50733752620545,
      "grad_norm": 0.030400998890399933,
      "learning_rate": 9.255909213741199e-08,
      "loss": 0.0082,
      "num_input_tokens_seen": 24281024,
      "step": 37220
    },
    {
      "epoch": 19.509958071278827,
      "grad_norm": 0.46878141164779663,
      "learning_rate": 9.157867997034164e-08,
      "loss": 0.0129,
      "num_input_tokens_seen": 24283840,
      "step": 37225
    },
    {
      "epoch": 19.5125786163522,
      "grad_norm": 0.1752505749464035,
      "learning_rate": 9.060347836084004e-08,
      "loss": 0.0028,
      "num_input_tokens_seen": 24286240,
      "step": 37230
    },
    {
      "epoch": 19.515199161425578,
      "grad_norm": 0.140499547123909,
      "learning_rate": 8.96334875129079e-08,
      "loss": 0.0018,
      "num_input_tokens_seen": 24288992,
      "step": 37235
    },
    {
      "epoch": 19.51781970649895,
      "grad_norm": 0.4632833003997803,
      "learning_rate": 8.866870762946067e-08,
      "loss": 0.005,
      "num_input_tokens_seen": 24292288,
      "step": 37240
    },
    {
      "epoch": 19.520440251572328,
      "grad_norm": 0.34793248772621155,
      "learning_rate": 8.77091389123147e-08,
      "loss": 0.0105,
      "num_input_tokens_seen": 24295584,
      "step": 37245
    },
    {
      "epoch": 19.5230607966457,
      "grad_norm": 0.17903564870357513,
      "learning_rate": 8.675478156220385e-08,
      "loss": 0.0051,
      "num_input_tokens_seen": 24298400,
      "step": 37250
    },
    {
      "epoch": 19.52568134171908,
      "grad_norm": 0.3360633850097656,
      "learning_rate": 8.580563577876566e-08,
      "loss": 0.0028,
      "num_input_tokens_seen": 24301280,
      "step": 37255
    },
    {
      "epoch": 19.528301886792452,
      "grad_norm": 0.3629761338233948,
      "learning_rate": 8.486170176055242e-08,
      "loss": 0.0037,
      "num_input_tokens_seen": 24303968,
      "step": 37260
    },
    {
      "epoch": 19.53092243186583,
      "grad_norm": 0.30614233016967773,
      "learning_rate": 8.392297970502282e-08,
      "loss": 0.0105,
      "num_input_tokens_seen": 24307456,
      "step": 37265
    },
    {
      "epoch": 19.533542976939202,
      "grad_norm": 0.23689621686935425,
      "learning_rate": 8.298946980855315e-08,
      "loss": 0.01,
      "num_input_tokens_seen": 24310912,
      "step": 37270
    },
    {
      "epoch": 19.53616352201258,
      "grad_norm": 0.24503366649150848,
      "learning_rate": 8.206117226641497e-08,
      "loss": 0.0029,
      "num_input_tokens_seen": 24313728,
      "step": 37275
    },
    {
      "epoch": 19.538784067085953,
      "grad_norm": 0.4058052599430084,
      "learning_rate": 8.113808727280015e-08,
      "loss": 0.0053,
      "num_input_tokens_seen": 24316928,
      "step": 37280
    },
    {
      "epoch": 19.54140461215933,
      "grad_norm": 0.07809674739837646,
      "learning_rate": 8.022021502080979e-08,
      "loss": 0.0015,
      "num_input_tokens_seen": 24319552,
      "step": 37285
    },
    {
      "epoch": 19.544025157232703,
      "grad_norm": 0.13120369613170624,
      "learning_rate": 7.930755570244863e-08,
      "loss": 0.0162,
      "num_input_tokens_seen": 24321984,
      "step": 37290
    },
    {
      "epoch": 19.54664570230608,
      "grad_norm": 0.30455318093299866,
      "learning_rate": 7.840010950863619e-08,
      "loss": 0.0044,
      "num_input_tokens_seen": 24324928,
      "step": 37295
    },
    {
      "epoch": 19.549266247379457,
      "grad_norm": 0.12654249370098114,
      "learning_rate": 7.749787662920116e-08,
      "loss": 0.0018,
      "num_input_tokens_seen": 24328064,
      "step": 37300
    },
    {
      "epoch": 19.55188679245283,
      "grad_norm": 0.13420164585113525,
      "learning_rate": 7.660085725288146e-08,
      "loss": 0.0054,
      "num_input_tokens_seen": 24331136,
      "step": 37305
    },
    {
      "epoch": 19.554507337526207,
      "grad_norm": 0.0715843066573143,
      "learning_rate": 7.570905156731867e-08,
      "loss": 0.0037,
      "num_input_tokens_seen": 24334112,
      "step": 37310
    },
    {
      "epoch": 19.55712788259958,
      "grad_norm": 0.15468615293502808,
      "learning_rate": 7.482245975907188e-08,
      "loss": 0.0082,
      "num_input_tokens_seen": 24338048,
      "step": 37315
    },
    {
      "epoch": 19.559748427672957,
      "grad_norm": 0.902385413646698,
      "learning_rate": 7.394108201360383e-08,
      "loss": 0.0105,
      "num_input_tokens_seen": 24341632,
      "step": 37320
    },
    {
      "epoch": 19.56236897274633,
      "grad_norm": 0.291456401348114,
      "learning_rate": 7.306491851529206e-08,
      "loss": 0.0053,
      "num_input_tokens_seen": 24345088,
      "step": 37325
    },
    {
      "epoch": 19.564989517819708,
      "grad_norm": 0.5867154002189636,
      "learning_rate": 7.219396944741774e-08,
      "loss": 0.0047,
      "num_input_tokens_seen": 24348736,
      "step": 37330
    },
    {
      "epoch": 19.56761006289308,
      "grad_norm": 0.4128708839416504,
      "learning_rate": 7.132823499217123e-08,
      "loss": 0.0025,
      "num_input_tokens_seen": 24351776,
      "step": 37335
    },
    {
      "epoch": 19.570230607966458,
      "grad_norm": 3.321275234222412,
      "learning_rate": 7.046771533065766e-08,
      "loss": 0.0102,
      "num_input_tokens_seen": 24355296,
      "step": 37340
    },
    {
      "epoch": 19.57285115303983,
      "grad_norm": 1.9757970571517944,
      "learning_rate": 6.961241064288859e-08,
      "loss": 0.0049,
      "num_input_tokens_seen": 24359168,
      "step": 37345
    },
    {
      "epoch": 19.57547169811321,
      "grad_norm": 1.0830347537994385,
      "learning_rate": 6.876232110778202e-08,
      "loss": 0.0091,
      "num_input_tokens_seen": 24362720,
      "step": 37350
    },
    {
      "epoch": 19.578092243186582,
      "grad_norm": 0.2902049124240875,
      "learning_rate": 6.791744690316515e-08,
      "loss": 0.0025,
      "num_input_tokens_seen": 24365824,
      "step": 37355
    },
    {
      "epoch": 19.58071278825996,
      "grad_norm": 0.9071994423866272,
      "learning_rate": 6.707778820577715e-08,
      "loss": 0.0048,
      "num_input_tokens_seen": 24368928,
      "step": 37360
    },
    {
      "epoch": 19.583333333333332,
      "grad_norm": 0.22329844534397125,
      "learning_rate": 6.624334519126919e-08,
      "loss": 0.0026,
      "num_input_tokens_seen": 24372192,
      "step": 37365
    },
    {
      "epoch": 19.58595387840671,
      "grad_norm": 0.3286275863647461,
      "learning_rate": 6.541411803419329e-08,
      "loss": 0.003,
      "num_input_tokens_seen": 24374912,
      "step": 37370
    },
    {
      "epoch": 19.588574423480082,
      "grad_norm": 0.414485901594162,
      "learning_rate": 6.459010690801626e-08,
      "loss": 0.0045,
      "num_input_tokens_seen": 24378176,
      "step": 37375
    },
    {
      "epoch": 19.59119496855346,
      "grad_norm": 0.2850298583507538,
      "learning_rate": 6.377131198510855e-08,
      "loss": 0.0037,
      "num_input_tokens_seen": 24381760,
      "step": 37380
    },
    {
      "epoch": 19.593815513626833,
      "grad_norm": 0.4448111951351166,
      "learning_rate": 6.295773343675537e-08,
      "loss": 0.0055,
      "num_input_tokens_seen": 24386144,
      "step": 37385
    },
    {
      "epoch": 19.59643605870021,
      "grad_norm": 0.20997630059719086,
      "learning_rate": 6.214937143314837e-08,
      "loss": 0.0321,
      "num_input_tokens_seen": 24389600,
      "step": 37390
    },
    {
      "epoch": 19.599056603773583,
      "grad_norm": 0.131125345826149,
      "learning_rate": 6.134622614338559e-08,
      "loss": 0.0016,
      "num_input_tokens_seen": 24392160,
      "step": 37395
    },
    {
      "epoch": 19.60167714884696,
      "grad_norm": 0.1431571990251541,
      "learning_rate": 6.05482977354771e-08,
      "loss": 0.004,
      "num_input_tokens_seen": 24395584,
      "step": 37400
    },
    {
      "epoch": 19.604297693920337,
      "grad_norm": 0.20762786269187927,
      "learning_rate": 5.975558637634215e-08,
      "loss": 0.0019,
      "num_input_tokens_seen": 24397696,
      "step": 37405
    },
    {
      "epoch": 19.60691823899371,
      "grad_norm": 0.6610403656959534,
      "learning_rate": 5.896809223180366e-08,
      "loss": 0.0048,
      "num_input_tokens_seen": 24401696,
      "step": 37410
    },
    {
      "epoch": 19.609538784067087,
      "grad_norm": 0.28878021240234375,
      "learning_rate": 5.8185815466596516e-08,
      "loss": 0.0149,
      "num_input_tokens_seen": 24404416,
      "step": 37415
    },
    {
      "epoch": 19.61215932914046,
      "grad_norm": 0.13827279210090637,
      "learning_rate": 5.7408756244370365e-08,
      "loss": 0.0022,
      "num_input_tokens_seen": 24407008,
      "step": 37420
    },
    {
      "epoch": 19.614779874213838,
      "grad_norm": 0.48628586530685425,
      "learning_rate": 5.663691472766741e-08,
      "loss": 0.0025,
      "num_input_tokens_seen": 24410208,
      "step": 37425
    },
    {
      "epoch": 19.61740041928721,
      "grad_norm": 0.324675977230072,
      "learning_rate": 5.587029107795572e-08,
      "loss": 0.004,
      "num_input_tokens_seen": 24415168,
      "step": 37430
    },
    {
      "epoch": 19.620020964360588,
      "grad_norm": 0.38713085651397705,
      "learning_rate": 5.510888545559867e-08,
      "loss": 0.0043,
      "num_input_tokens_seen": 24417984,
      "step": 37435
    },
    {
      "epoch": 19.62264150943396,
      "grad_norm": 0.3566564619541168,
      "learning_rate": 5.435269801987997e-08,
      "loss": 0.0041,
      "num_input_tokens_seen": 24421376,
      "step": 37440
    },
    {
      "epoch": 19.62526205450734,
      "grad_norm": 0.1779756098985672,
      "learning_rate": 5.360172892897863e-08,
      "loss": 0.0016,
      "num_input_tokens_seen": 24424544,
      "step": 37445
    },
    {
      "epoch": 19.627882599580712,
      "grad_norm": 0.32051151990890503,
      "learning_rate": 5.285597833999401e-08,
      "loss": 0.0041,
      "num_input_tokens_seen": 24427488,
      "step": 37450
    },
    {
      "epoch": 19.63050314465409,
      "grad_norm": 0.047865282744169235,
      "learning_rate": 5.2115446408926295e-08,
      "loss": 0.0049,
      "num_input_tokens_seen": 24430336,
      "step": 37455
    },
    {
      "epoch": 19.633123689727462,
      "grad_norm": 0.27663570642471313,
      "learning_rate": 5.138013329068769e-08,
      "loss": 0.0127,
      "num_input_tokens_seen": 24435136,
      "step": 37460
    },
    {
      "epoch": 19.63574423480084,
      "grad_norm": 0.3440513610839844,
      "learning_rate": 5.065003913909405e-08,
      "loss": 0.0048,
      "num_input_tokens_seen": 24438848,
      "step": 37465
    },
    {
      "epoch": 19.638364779874212,
      "grad_norm": 0.46727392077445984,
      "learning_rate": 4.9925164106875975e-08,
      "loss": 0.0023,
      "num_input_tokens_seen": 24441248,
      "step": 37470
    },
    {
      "epoch": 19.64098532494759,
      "grad_norm": 0.238470196723938,
      "learning_rate": 4.9205508345670506e-08,
      "loss": 0.0025,
      "num_input_tokens_seen": 24443840,
      "step": 37475
    },
    {
      "epoch": 19.643605870020963,
      "grad_norm": 0.3142942786216736,
      "learning_rate": 4.8491072006015546e-08,
      "loss": 0.0039,
      "num_input_tokens_seen": 24447200,
      "step": 37480
    },
    {
      "epoch": 19.64622641509434,
      "grad_norm": 0.25102218985557556,
      "learning_rate": 4.778185523736933e-08,
      "loss": 0.0049,
      "num_input_tokens_seen": 24450368,
      "step": 37485
    },
    {
      "epoch": 19.648846960167717,
      "grad_norm": 0.091954305768013,
      "learning_rate": 4.7077858188085414e-08,
      "loss": 0.0048,
      "num_input_tokens_seen": 24452992,
      "step": 37490
    },
    {
      "epoch": 19.65146750524109,
      "grad_norm": 0.14042891561985016,
      "learning_rate": 4.6379081005437644e-08,
      "loss": 0.002,
      "num_input_tokens_seen": 24455264,
      "step": 37495
    },
    {
      "epoch": 19.654088050314467,
      "grad_norm": 0.16608814895153046,
      "learning_rate": 4.568552383560076e-08,
      "loss": 0.0262,
      "num_input_tokens_seen": 24458304,
      "step": 37500
    },
    {
      "epoch": 19.65670859538784,
      "grad_norm": 0.1828642040491104,
      "learning_rate": 4.4997186823655945e-08,
      "loss": 0.0031,
      "num_input_tokens_seen": 24461824,
      "step": 37505
    },
    {
      "epoch": 19.659329140461217,
      "grad_norm": 0.5026172995567322,
      "learning_rate": 4.431407011360189e-08,
      "loss": 0.0034,
      "num_input_tokens_seen": 24464736,
      "step": 37510
    },
    {
      "epoch": 19.66194968553459,
      "grad_norm": 0.4783326983451843,
      "learning_rate": 4.3636173848332627e-08,
      "loss": 0.0028,
      "num_input_tokens_seen": 24467264,
      "step": 37515
    },
    {
      "epoch": 19.664570230607968,
      "grad_norm": 2.0441882610321045,
      "learning_rate": 4.2963498169656946e-08,
      "loss": 0.0109,
      "num_input_tokens_seen": 24470912,
      "step": 37520
    },
    {
      "epoch": 19.66719077568134,
      "grad_norm": 0.3109152615070343,
      "learning_rate": 4.229604321829561e-08,
      "loss": 0.0051,
      "num_input_tokens_seen": 24474304,
      "step": 37525
    },
    {
      "epoch": 19.669811320754718,
      "grad_norm": 0.05265980586409569,
      "learning_rate": 4.16338091338675e-08,
      "loss": 0.0042,
      "num_input_tokens_seen": 24477280,
      "step": 37530
    },
    {
      "epoch": 19.67243186582809,
      "grad_norm": 0.25612497329711914,
      "learning_rate": 4.097679605490623e-08,
      "loss": 0.0019,
      "num_input_tokens_seen": 24479456,
      "step": 37535
    },
    {
      "epoch": 19.67505241090147,
      "grad_norm": 0.11336565762758255,
      "learning_rate": 4.032500411885465e-08,
      "loss": 0.0041,
      "num_input_tokens_seen": 24482624,
      "step": 37540
    },
    {
      "epoch": 19.677672955974842,
      "grad_norm": 0.18862946331501007,
      "learning_rate": 3.967843346205369e-08,
      "loss": 0.0017,
      "num_input_tokens_seen": 24485792,
      "step": 37545
    },
    {
      "epoch": 19.68029350104822,
      "grad_norm": 0.08702176064252853,
      "learning_rate": 3.903708421976182e-08,
      "loss": 0.0021,
      "num_input_tokens_seen": 24488576,
      "step": 37550
    },
    {
      "epoch": 19.682914046121592,
      "grad_norm": 0.5808188915252686,
      "learning_rate": 3.840095652614395e-08,
      "loss": 0.0069,
      "num_input_tokens_seen": 24491136,
      "step": 37555
    },
    {
      "epoch": 19.68553459119497,
      "grad_norm": 0.4277207851409912,
      "learning_rate": 3.777005051426863e-08,
      "loss": 0.0826,
      "num_input_tokens_seen": 24494528,
      "step": 37560
    },
    {
      "epoch": 19.688155136268342,
      "grad_norm": 0.24966847896575928,
      "learning_rate": 3.71443663161164e-08,
      "loss": 0.0242,
      "num_input_tokens_seen": 24498816,
      "step": 37565
    },
    {
      "epoch": 19.69077568134172,
      "grad_norm": 0.9092435240745544,
      "learning_rate": 3.652390406256867e-08,
      "loss": 0.0068,
      "num_input_tokens_seen": 24502368,
      "step": 37570
    },
    {
      "epoch": 19.693396226415093,
      "grad_norm": 0.19745217263698578,
      "learning_rate": 3.5908663883421624e-08,
      "loss": 0.0031,
      "num_input_tokens_seen": 24505440,
      "step": 37575
    },
    {
      "epoch": 19.69601677148847,
      "grad_norm": 0.8043612837791443,
      "learning_rate": 3.529864590737786e-08,
      "loss": 0.0223,
      "num_input_tokens_seen": 24508512,
      "step": 37580
    },
    {
      "epoch": 19.698637316561843,
      "grad_norm": 0.2986477017402649,
      "learning_rate": 3.4693850262046415e-08,
      "loss": 0.0035,
      "num_input_tokens_seen": 24511360,
      "step": 37585
    },
    {
      "epoch": 19.70125786163522,
      "grad_norm": 0.3084336817264557,
      "learning_rate": 3.409427707394275e-08,
      "loss": 0.0031,
      "num_input_tokens_seen": 24514048,
      "step": 37590
    },
    {
      "epoch": 19.703878406708597,
      "grad_norm": 1.523566722869873,
      "learning_rate": 3.349992646848599e-08,
      "loss": 0.0059,
      "num_input_tokens_seen": 24517088,
      "step": 37595
    },
    {
      "epoch": 19.70649895178197,
      "grad_norm": 0.2530877888202667,
      "learning_rate": 3.291079857001556e-08,
      "loss": 0.0092,
      "num_input_tokens_seen": 24519808,
      "step": 37600
    },
    {
      "epoch": 19.709119496855347,
      "grad_norm": 0.40311819314956665,
      "learning_rate": 3.2326893501763436e-08,
      "loss": 0.0109,
      "num_input_tokens_seen": 24524320,
      "step": 37605
    },
    {
      "epoch": 19.71174004192872,
      "grad_norm": 0.30464547872543335,
      "learning_rate": 3.174821138588191e-08,
      "loss": 0.0073,
      "num_input_tokens_seen": 24528192,
      "step": 37610
    },
    {
      "epoch": 19.714360587002098,
      "grad_norm": 0.42952701449394226,
      "learning_rate": 3.11747523434186e-08,
      "loss": 0.0298,
      "num_input_tokens_seen": 24530880,
      "step": 37615
    },
    {
      "epoch": 19.71698113207547,
      "grad_norm": 0.39840108156204224,
      "learning_rate": 3.060651649434143e-08,
      "loss": 0.0038,
      "num_input_tokens_seen": 24533952,
      "step": 37620
    },
    {
      "epoch": 19.719601677148848,
      "grad_norm": 0.3750562369823456,
      "learning_rate": 3.004350395751088e-08,
      "loss": 0.0096,
      "num_input_tokens_seen": 24537152,
      "step": 37625
    },
    {
      "epoch": 19.72222222222222,
      "grad_norm": 0.3062137961387634,
      "learning_rate": 2.94857148507105e-08,
      "loss": 0.0009,
      "num_input_tokens_seen": 24539776,
      "step": 37630
    },
    {
      "epoch": 19.7248427672956,
      "grad_norm": 0.8709806203842163,
      "learning_rate": 2.893314929061641e-08,
      "loss": 0.0068,
      "num_input_tokens_seen": 24542944,
      "step": 37635
    },
    {
      "epoch": 19.72746331236897,
      "grad_norm": 0.3207685649394989,
      "learning_rate": 2.8385807392825037e-08,
      "loss": 0.0031,
      "num_input_tokens_seen": 24545504,
      "step": 37640
    },
    {
      "epoch": 19.73008385744235,
      "grad_norm": 0.16242559254169464,
      "learning_rate": 2.7843689271830897e-08,
      "loss": 0.0034,
      "num_input_tokens_seen": 24548736,
      "step": 37645
    },
    {
      "epoch": 19.732704402515722,
      "grad_norm": 0.4300368130207062,
      "learning_rate": 2.7306795041040502e-08,
      "loss": 0.0106,
      "num_input_tokens_seen": 24552000,
      "step": 37650
    },
    {
      "epoch": 19.7353249475891,
      "grad_norm": 0.241311177611351,
      "learning_rate": 2.677512481276401e-08,
      "loss": 0.0027,
      "num_input_tokens_seen": 24555968,
      "step": 37655
    },
    {
      "epoch": 19.737945492662472,
      "grad_norm": 1.5870418548583984,
      "learning_rate": 2.6248678698220785e-08,
      "loss": 0.0203,
      "num_input_tokens_seen": 24559104,
      "step": 37660
    },
    {
      "epoch": 19.74056603773585,
      "grad_norm": 0.5619472861289978,
      "learning_rate": 2.5727456807542183e-08,
      "loss": 0.0084,
      "num_input_tokens_seen": 24561824,
      "step": 37665
    },
    {
      "epoch": 19.743186582809223,
      "grad_norm": 0.23968254029750824,
      "learning_rate": 2.5211459249754875e-08,
      "loss": 0.0027,
      "num_input_tokens_seen": 24564864,
      "step": 37670
    },
    {
      "epoch": 19.7458071278826,
      "grad_norm": 0.09214484691619873,
      "learning_rate": 2.4700686132803076e-08,
      "loss": 0.004,
      "num_input_tokens_seen": 24568224,
      "step": 37675
    },
    {
      "epoch": 19.748427672955973,
      "grad_norm": 0.5113641023635864,
      "learning_rate": 2.4195137563534643e-08,
      "loss": 0.0071,
      "num_input_tokens_seen": 24573312,
      "step": 37680
    },
    {
      "epoch": 19.75104821802935,
      "grad_norm": 0.3569447994232178,
      "learning_rate": 2.369481364770387e-08,
      "loss": 0.0029,
      "num_input_tokens_seen": 24577248,
      "step": 37685
    },
    {
      "epoch": 19.753668763102727,
      "grad_norm": 0.13466165959835052,
      "learning_rate": 2.3199714489974266e-08,
      "loss": 0.0042,
      "num_input_tokens_seen": 24580544,
      "step": 37690
    },
    {
      "epoch": 19.7562893081761,
      "grad_norm": 1.1281354427337646,
      "learning_rate": 2.2709840193912978e-08,
      "loss": 0.0021,
      "num_input_tokens_seen": 24582944,
      "step": 37695
    },
    {
      "epoch": 19.758909853249477,
      "grad_norm": 0.397726833820343,
      "learning_rate": 2.2225190861999146e-08,
      "loss": 0.0083,
      "num_input_tokens_seen": 24585728,
      "step": 37700
    },
    {
      "epoch": 19.76153039832285,
      "grad_norm": 0.16463439166545868,
      "learning_rate": 2.1745766595612783e-08,
      "loss": 0.0079,
      "num_input_tokens_seen": 24588640,
      "step": 37705
    },
    {
      "epoch": 19.764150943396228,
      "grad_norm": 0.19684284925460815,
      "learning_rate": 2.1271567495045886e-08,
      "loss": 0.0085,
      "num_input_tokens_seen": 24592064,
      "step": 37710
    },
    {
      "epoch": 19.7667714884696,
      "grad_norm": 0.10359736531972885,
      "learning_rate": 2.0802593659491333e-08,
      "loss": 0.0263,
      "num_input_tokens_seen": 24598944,
      "step": 37715
    },
    {
      "epoch": 19.769392033542978,
      "grad_norm": 0.3740951716899872,
      "learning_rate": 2.0338845187059528e-08,
      "loss": 0.0046,
      "num_input_tokens_seen": 24601152,
      "step": 37720
    },
    {
      "epoch": 19.77201257861635,
      "grad_norm": 0.3180868327617645,
      "learning_rate": 1.988032217475899e-08,
      "loss": 0.0042,
      "num_input_tokens_seen": 24603904,
      "step": 37725
    },
    {
      "epoch": 19.77463312368973,
      "grad_norm": 0.3409057557582855,
      "learning_rate": 1.942702471850466e-08,
      "loss": 0.0024,
      "num_input_tokens_seen": 24606240,
      "step": 37730
    },
    {
      "epoch": 19.7772536687631,
      "grad_norm": 0.4017205536365509,
      "learning_rate": 1.897895291312346e-08,
      "loss": 0.0047,
      "num_input_tokens_seen": 24609088,
      "step": 37735
    },
    {
      "epoch": 19.77987421383648,
      "grad_norm": 0.18128421902656555,
      "learning_rate": 1.8536106852348746e-08,
      "loss": 0.004,
      "num_input_tokens_seen": 24612320,
      "step": 37740
    },
    {
      "epoch": 19.782494758909852,
      "grad_norm": 0.16928978264331818,
      "learning_rate": 1.8098486628814748e-08,
      "loss": 0.0047,
      "num_input_tokens_seen": 24617664,
      "step": 37745
    },
    {
      "epoch": 19.78511530398323,
      "grad_norm": 0.31263434886932373,
      "learning_rate": 1.7666092334067684e-08,
      "loss": 0.0044,
      "num_input_tokens_seen": 24622304,
      "step": 37750
    },
    {
      "epoch": 19.787735849056602,
      "grad_norm": 0.9133978486061096,
      "learning_rate": 1.7238924058562977e-08,
      "loss": 0.0035,
      "num_input_tokens_seen": 24625088,
      "step": 37755
    },
    {
      "epoch": 19.79035639412998,
      "grad_norm": 1.2627753019332886,
      "learning_rate": 1.6816981891656924e-08,
      "loss": 0.0097,
      "num_input_tokens_seen": 24628544,
      "step": 37760
    },
    {
      "epoch": 19.792976939203353,
      "grad_norm": 0.2115083634853363,
      "learning_rate": 1.6400265921615032e-08,
      "loss": 0.0223,
      "num_input_tokens_seen": 24631776,
      "step": 37765
    },
    {
      "epoch": 19.79559748427673,
      "grad_norm": 0.2117316871881485,
      "learning_rate": 1.598877623560924e-08,
      "loss": 0.0035,
      "num_input_tokens_seen": 24634976,
      "step": 37770
    },
    {
      "epoch": 19.798218029350103,
      "grad_norm": 0.27432507276535034,
      "learning_rate": 1.5582512919717908e-08,
      "loss": 0.0025,
      "num_input_tokens_seen": 24637696,
      "step": 37775
    },
    {
      "epoch": 19.80083857442348,
      "grad_norm": 0.23300984501838684,
      "learning_rate": 1.5181476058928613e-08,
      "loss": 0.0035,
      "num_input_tokens_seen": 24640864,
      "step": 37780
    },
    {
      "epoch": 19.803459119496857,
      "grad_norm": 0.3224647045135498,
      "learning_rate": 1.4785665737129805e-08,
      "loss": 0.0048,
      "num_input_tokens_seen": 24646784,
      "step": 37785
    },
    {
      "epoch": 19.80607966457023,
      "grad_norm": 0.3706931173801422,
      "learning_rate": 1.4395082037127471e-08,
      "loss": 0.0022,
      "num_input_tokens_seen": 24650048,
      "step": 37790
    },
    {
      "epoch": 19.808700209643607,
      "grad_norm": 0.20720653235912323,
      "learning_rate": 1.4009725040620148e-08,
      "loss": 0.0066,
      "num_input_tokens_seen": 24653728,
      "step": 37795
    },
    {
      "epoch": 19.81132075471698,
      "grad_norm": 0.04898753762245178,
      "learning_rate": 1.3629594828223902e-08,
      "loss": 0.0041,
      "num_input_tokens_seen": 24656288,
      "step": 37800
    },
    {
      "epoch": 19.813941299790358,
      "grad_norm": 0.22914710640907288,
      "learning_rate": 1.3254691479458458e-08,
      "loss": 0.0028,
      "num_input_tokens_seen": 24659744,
      "step": 37805
    },
    {
      "epoch": 19.81656184486373,
      "grad_norm": 1.4910703897476196,
      "learning_rate": 1.2885015072747197e-08,
      "loss": 0.0151,
      "num_input_tokens_seen": 24663488,
      "step": 37810
    },
    {
      "epoch": 19.819182389937108,
      "grad_norm": 0.5856581330299377,
      "learning_rate": 1.2520565685419927e-08,
      "loss": 0.0019,
      "num_input_tokens_seen": 24666272,
      "step": 37815
    },
    {
      "epoch": 19.82180293501048,
      "grad_norm": 0.2383277714252472,
      "learning_rate": 1.2161343393721214e-08,
      "loss": 0.0018,
      "num_input_tokens_seen": 24670400,
      "step": 37820
    },
    {
      "epoch": 19.82442348008386,
      "grad_norm": 0.669457197189331,
      "learning_rate": 1.1807348272790952e-08,
      "loss": 0.0043,
      "num_input_tokens_seen": 24673280,
      "step": 37825
    },
    {
      "epoch": 19.82704402515723,
      "grad_norm": 0.5843515396118164,
      "learning_rate": 1.1458580396683794e-08,
      "loss": 0.0084,
      "num_input_tokens_seen": 24676320,
      "step": 37830
    },
    {
      "epoch": 19.82966457023061,
      "grad_norm": 0.24057815968990326,
      "learning_rate": 1.1115039838358044e-08,
      "loss": 0.0019,
      "num_input_tokens_seen": 24680960,
      "step": 37835
    },
    {
      "epoch": 19.832285115303982,
      "grad_norm": 1.0873693227767944,
      "learning_rate": 1.0776726669678439e-08,
      "loss": 0.0049,
      "num_input_tokens_seen": 24685792,
      "step": 37840
    },
    {
      "epoch": 19.83490566037736,
      "grad_norm": 1.6922736167907715,
      "learning_rate": 1.044364096141337e-08,
      "loss": 0.0034,
      "num_input_tokens_seen": 24688448,
      "step": 37845
    },
    {
      "epoch": 19.837526205450732,
      "grad_norm": 0.36329185962677,
      "learning_rate": 1.0115782783243211e-08,
      "loss": 0.0039,
      "num_input_tokens_seen": 24691552,
      "step": 37850
    },
    {
      "epoch": 19.84014675052411,
      "grad_norm": 0.8344599604606628,
      "learning_rate": 9.793152203751988e-09,
      "loss": 0.003,
      "num_input_tokens_seen": 24694368,
      "step": 37855
    },
    {
      "epoch": 19.842767295597483,
      "grad_norm": 0.06556183099746704,
      "learning_rate": 9.475749290430158e-09,
      "loss": 0.0212,
      "num_input_tokens_seen": 24696608,
      "step": 37860
    },
    {
      "epoch": 19.84538784067086,
      "grad_norm": 0.058775894343853,
      "learning_rate": 9.16357410967461e-09,
      "loss": 0.0118,
      "num_input_tokens_seen": 24699872,
      "step": 37865
    },
    {
      "epoch": 19.848008385744233,
      "grad_norm": 0.29275423288345337,
      "learning_rate": 8.856626726788664e-09,
      "loss": 0.0198,
      "num_input_tokens_seen": 24704608,
      "step": 37870
    },
    {
      "epoch": 19.85062893081761,
      "grad_norm": 0.5817102789878845,
      "learning_rate": 8.554907205982066e-09,
      "loss": 0.0034,
      "num_input_tokens_seen": 24708544,
      "step": 37875
    },
    {
      "epoch": 19.853249475890987,
      "grad_norm": 0.12963718175888062,
      "learning_rate": 8.258415610373772e-09,
      "loss": 0.0024,
      "num_input_tokens_seen": 24710880,
      "step": 37880
    },
    {
      "epoch": 19.85587002096436,
      "grad_norm": 1.1454113721847534,
      "learning_rate": 7.967152001980838e-09,
      "loss": 0.0117,
      "num_input_tokens_seen": 24715008,
      "step": 37885
    },
    {
      "epoch": 19.858490566037737,
      "grad_norm": 0.3906896412372589,
      "learning_rate": 7.681116441737856e-09,
      "loss": 0.0031,
      "num_input_tokens_seen": 24718240,
      "step": 37890
    },
    {
      "epoch": 19.86111111111111,
      "grad_norm": 0.2620212137699127,
      "learning_rate": 7.4003089894747464e-09,
      "loss": 0.0027,
      "num_input_tokens_seen": 24720800,
      "step": 37895
    },
    {
      "epoch": 19.863731656184488,
      "grad_norm": 0.5028545260429382,
      "learning_rate": 7.12472970393896e-09,
      "loss": 0.0045,
      "num_input_tokens_seen": 24723392,
      "step": 37900
    },
    {
      "epoch": 19.86635220125786,
      "grad_norm": 0.2589717507362366,
      "learning_rate": 6.854378642776049e-09,
      "loss": 0.0022,
      "num_input_tokens_seen": 24725792,
      "step": 37905
    },
    {
      "epoch": 19.868972746331238,
      "grad_norm": 0.2027866095304489,
      "learning_rate": 6.589255862538002e-09,
      "loss": 0.0019,
      "num_input_tokens_seen": 24728672,
      "step": 37910
    },
    {
      "epoch": 19.87159329140461,
      "grad_norm": 0.8676725625991821,
      "learning_rate": 6.32936141869156e-09,
      "loss": 0.0168,
      "num_input_tokens_seen": 24732384,
      "step": 37915
    },
    {
      "epoch": 19.87421383647799,
      "grad_norm": 0.2854346036911011,
      "learning_rate": 6.074695365596017e-09,
      "loss": 0.0104,
      "num_input_tokens_seen": 24735744,
      "step": 37920
    },
    {
      "epoch": 19.87683438155136,
      "grad_norm": 0.5634627342224121,
      "learning_rate": 5.8252577565309775e-09,
      "loss": 0.0029,
      "num_input_tokens_seen": 24738496,
      "step": 37925
    },
    {
      "epoch": 19.87945492662474,
      "grad_norm": 0.3961026966571808,
      "learning_rate": 5.581048643671371e-09,
      "loss": 0.0031,
      "num_input_tokens_seen": 24742080,
      "step": 37930
    },
    {
      "epoch": 19.882075471698112,
      "grad_norm": 1.853985071182251,
      "learning_rate": 5.342068078106888e-09,
      "loss": 0.0071,
      "num_input_tokens_seen": 24744512,
      "step": 37935
    },
    {
      "epoch": 19.88469601677149,
      "grad_norm": 0.24342124164104462,
      "learning_rate": 5.1083161098253175e-09,
      "loss": 0.003,
      "num_input_tokens_seen": 24748224,
      "step": 37940
    },
    {
      "epoch": 19.887316561844862,
      "grad_norm": 0.6143083572387695,
      "learning_rate": 4.87979278772921e-09,
      "loss": 0.0077,
      "num_input_tokens_seen": 24751232,
      "step": 37945
    },
    {
      "epoch": 19.88993710691824,
      "grad_norm": 0.1349313110113144,
      "learning_rate": 4.656498159619216e-09,
      "loss": 0.0021,
      "num_input_tokens_seen": 24754208,
      "step": 37950
    },
    {
      "epoch": 19.892557651991613,
      "grad_norm": 1.9057343006134033,
      "learning_rate": 4.438432272210746e-09,
      "loss": 0.0047,
      "num_input_tokens_seen": 24757440,
      "step": 37955
    },
    {
      "epoch": 19.89517819706499,
      "grad_norm": 0.28491660952568054,
      "learning_rate": 4.225595171114538e-09,
      "loss": 0.0025,
      "num_input_tokens_seen": 24759808,
      "step": 37960
    },
    {
      "epoch": 19.897798742138363,
      "grad_norm": 0.3329765796661377,
      "learning_rate": 4.017986900858861e-09,
      "loss": 0.0017,
      "num_input_tokens_seen": 24761952,
      "step": 37965
    },
    {
      "epoch": 19.90041928721174,
      "grad_norm": 0.4638674259185791,
      "learning_rate": 3.815607504870089e-09,
      "loss": 0.0082,
      "num_input_tokens_seen": 24765280,
      "step": 37970
    },
    {
      "epoch": 19.903039832285117,
      "grad_norm": 0.18716749548912048,
      "learning_rate": 3.618457025486577e-09,
      "loss": 0.0018,
      "num_input_tokens_seen": 24768320,
      "step": 37975
    },
    {
      "epoch": 19.90566037735849,
      "grad_norm": 0.24616563320159912,
      "learning_rate": 3.426535503947559e-09,
      "loss": 0.0081,
      "num_input_tokens_seen": 24770912,
      "step": 37980
    },
    {
      "epoch": 19.908280922431867,
      "grad_norm": 2.4469876289367676,
      "learning_rate": 3.2398429803986997e-09,
      "loss": 0.0059,
      "num_input_tokens_seen": 24774176,
      "step": 37985
    },
    {
      "epoch": 19.91090146750524,
      "grad_norm": 0.5124332308769226,
      "learning_rate": 3.058379493900421e-09,
      "loss": 0.0059,
      "num_input_tokens_seen": 24777120,
      "step": 37990
    },
    {
      "epoch": 19.913522012578618,
      "grad_norm": 0.3254033923149109,
      "learning_rate": 2.882145082405696e-09,
      "loss": 0.0043,
      "num_input_tokens_seen": 24779712,
      "step": 37995
    },
    {
      "epoch": 19.91614255765199,
      "grad_norm": 1.4778589010238647,
      "learning_rate": 2.711139782785033e-09,
      "loss": 0.0243,
      "num_input_tokens_seen": 24783200,
      "step": 38000
    },
    {
      "epoch": 19.918763102725368,
      "grad_norm": 1.7760319709777832,
      "learning_rate": 2.5453636308098163e-09,
      "loss": 0.0172,
      "num_input_tokens_seen": 24786592,
      "step": 38005
    },
    {
      "epoch": 19.92138364779874,
      "grad_norm": 0.22094838321208954,
      "learning_rate": 2.3848166611606383e-09,
      "loss": 0.0048,
      "num_input_tokens_seen": 24789664,
      "step": 38010
    },
    {
      "epoch": 19.92400419287212,
      "grad_norm": 0.5393492579460144,
      "learning_rate": 2.2294989074161943e-09,
      "loss": 0.0063,
      "num_input_tokens_seen": 24792288,
      "step": 38015
    },
    {
      "epoch": 19.92662473794549,
      "grad_norm": 0.37469321489334106,
      "learning_rate": 2.0794104020727125e-09,
      "loss": 0.0049,
      "num_input_tokens_seen": 24796000,
      "step": 38020
    },
    {
      "epoch": 19.92924528301887,
      "grad_norm": 0.15399907529354095,
      "learning_rate": 1.9345511765272993e-09,
      "loss": 0.0036,
      "num_input_tokens_seen": 24798976,
      "step": 38025
    },
    {
      "epoch": 19.931865828092242,
      "grad_norm": 0.16936391592025757,
      "learning_rate": 1.794921261077942e-09,
      "loss": 0.0021,
      "num_input_tokens_seen": 24801760,
      "step": 38030
    },
    {
      "epoch": 19.93448637316562,
      "grad_norm": 0.12439881265163422,
      "learning_rate": 1.6605206849373833e-09,
      "loss": 0.0023,
      "num_input_tokens_seen": 24805376,
      "step": 38035
    },
    {
      "epoch": 19.937106918238992,
      "grad_norm": 1.9014005661010742,
      "learning_rate": 1.531349476219246e-09,
      "loss": 0.0084,
      "num_input_tokens_seen": 24809440,
      "step": 38040
    },
    {
      "epoch": 19.93972746331237,
      "grad_norm": 0.1447361260652542,
      "learning_rate": 1.4074076619463584e-09,
      "loss": 0.0037,
      "num_input_tokens_seen": 24813472,
      "step": 38045
    },
    {
      "epoch": 19.942348008385743,
      "grad_norm": 0.1800856590270996,
      "learning_rate": 1.2886952680452036e-09,
      "loss": 0.0028,
      "num_input_tokens_seen": 24816608,
      "step": 38050
    },
    {
      "epoch": 19.94496855345912,
      "grad_norm": 0.212967649102211,
      "learning_rate": 1.1752123193459197e-09,
      "loss": 0.0253,
      "num_input_tokens_seen": 24819776,
      "step": 38055
    },
    {
      "epoch": 19.947589098532493,
      "grad_norm": 0.34799113869667053,
      "learning_rate": 1.0669588395934017e-09,
      "loss": 0.0067,
      "num_input_tokens_seen": 24822560,
      "step": 38060
    },
    {
      "epoch": 19.95020964360587,
      "grad_norm": 0.4007241427898407,
      "learning_rate": 9.639348514278722e-10,
      "loss": 0.0034,
      "num_input_tokens_seen": 24825600,
      "step": 38065
    },
    {
      "epoch": 19.952830188679247,
      "grad_norm": 0.6776245832443237,
      "learning_rate": 8.661403764043119e-10,
      "loss": 0.0076,
      "num_input_tokens_seen": 24828224,
      "step": 38070
    },
    {
      "epoch": 19.95545073375262,
      "grad_norm": 0.23342901468276978,
      "learning_rate": 7.735754349785795e-10,
      "loss": 0.0064,
      "num_input_tokens_seen": 24830848,
      "step": 38075
    },
    {
      "epoch": 19.958071278825997,
      "grad_norm": 0.2884101867675781,
      "learning_rate": 6.862400465157403e-10,
      "loss": 0.0111,
      "num_input_tokens_seen": 24834304,
      "step": 38080
    },
    {
      "epoch": 19.96069182389937,
      "grad_norm": 0.09356354176998138,
      "learning_rate": 6.041342292817387e-10,
      "loss": 0.0013,
      "num_input_tokens_seen": 24838112,
      "step": 38085
    },
    {
      "epoch": 19.963312368972748,
      "grad_norm": 0.6371324062347412,
      "learning_rate": 5.272580004572758e-10,
      "loss": 0.0046,
      "num_input_tokens_seen": 24842240,
      "step": 38090
    },
    {
      "epoch": 19.96593291404612,
      "grad_norm": 0.2653566002845764,
      "learning_rate": 4.556113761211567e-10,
      "loss": 0.0052,
      "num_input_tokens_seen": 24845824,
      "step": 38095
    },
    {
      "epoch": 19.968553459119498,
      "grad_norm": 1.7268937826156616,
      "learning_rate": 3.8919437125861657e-10,
      "loss": 0.0058,
      "num_input_tokens_seen": 24848704,
      "step": 38100
    },
    {
      "epoch": 19.97117400419287,
      "grad_norm": 0.4848364591598511,
      "learning_rate": 3.280069997696478e-10,
      "loss": 0.0037,
      "num_input_tokens_seen": 24853120,
      "step": 38105
    },
    {
      "epoch": 19.97379454926625,
      "grad_norm": 0.4345487058162689,
      "learning_rate": 2.720492744467951e-10,
      "loss": 0.009,
      "num_input_tokens_seen": 24856800,
      "step": 38110
    },
    {
      "epoch": 19.97641509433962,
      "grad_norm": 0.4478537142276764,
      "learning_rate": 2.2132120700291136e-10,
      "loss": 0.0138,
      "num_input_tokens_seen": 24859392,
      "step": 38115
    },
    {
      "epoch": 19.979035639413,
      "grad_norm": 0.25862205028533936,
      "learning_rate": 1.7582280804340213e-10,
      "loss": 0.0921,
      "num_input_tokens_seen": 24863328,
      "step": 38120
    },
    {
      "epoch": 19.981656184486372,
      "grad_norm": 0.5385345220565796,
      "learning_rate": 1.3555408708842977e-10,
      "loss": 0.0107,
      "num_input_tokens_seen": 24866496,
      "step": 38125
    },
    {
      "epoch": 19.98427672955975,
      "grad_norm": 0.4953327178955078,
      "learning_rate": 1.0051505256458704e-10,
      "loss": 0.0063,
      "num_input_tokens_seen": 24871328,
      "step": 38130
    },
    {
      "epoch": 19.986897274633122,
      "grad_norm": 0.3853899836540222,
      "learning_rate": 7.070571179657037e-11,
      "loss": 0.0037,
      "num_input_tokens_seen": 24874336,
      "step": 38135
    },
    {
      "epoch": 19.9895178197065,
      "grad_norm": 0.38260406255722046,
      "learning_rate": 4.612607102383315e-11,
      "loss": 0.0056,
      "num_input_tokens_seen": 24877024,
      "step": 38140
    },
    {
      "epoch": 19.992138364779873,
      "grad_norm": 0.026595933362841606,
      "learning_rate": 2.677613538670798e-11,
      "loss": 0.0095,
      "num_input_tokens_seen": 24880096,
      "step": 38145
    },
    {
      "epoch": 19.99475890985325,
      "grad_norm": 0.25116854906082153,
      "learning_rate": 1.2655908931957782e-11,
      "loss": 0.003,
      "num_input_tokens_seen": 24883200,
      "step": 38150
    },
    {
      "epoch": 19.997379454926623,
      "grad_norm": 0.24277228116989136,
      "learning_rate": 3.765394615551366e-12,
      "loss": 0.0024,
      "num_input_tokens_seen": 24885664,
      "step": 38155
    },
    {
      "epoch": 20.0,
      "grad_norm": 0.15647614002227783,
      "learning_rate": 1.0459429711229262e-13,
      "loss": 0.0014,
      "num_input_tokens_seen": 24887720,
      "step": 38160
    },
    {
      "epoch": 20.0,
      "eval_loss": 1.5826364755630493,
      "eval_runtime": 11.9404,
      "eval_samples_per_second": 71.019,
      "eval_steps_per_second": 17.755,
      "num_input_tokens_seen": 24887720,
      "step": 38160
    },
    {
      "epoch": 20.0,
      "num_input_tokens_seen": 24887720,
      "step": 38160,
      "total_flos": 1.1228749030814515e+18,
      "train_loss": 0.19066539223493073,
      "train_runtime": 8807.1105,
      "train_samples_per_second": 17.325,
      "train_steps_per_second": 4.333
    }
  ],
  "logging_steps": 5,
  "max_steps": 38160,
  "num_input_tokens_seen": 24887720,
  "num_train_epochs": 20,
  "save_steps": 1908,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1228749030814515e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}