madonna / trainer_state.json
AlekseyKorshuk's picture
huggingartists
d439efe
raw history blame
No virus
39.6 kB
{
"best_metric": 1.1562458276748657,
"best_model_checkpoint": "output/madonna/checkpoint-1561",
"epoch": 7.0,
"global_step": 1561,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02,
"learning_rate": 0.0001370283488463827,
"loss": 2.5046,
"step": 5
},
{
"epoch": 0.05,
"learning_rate": 0.00013651425439773298,
"loss": 2.4348,
"step": 10
},
{
"epoch": 0.07,
"learning_rate": 0.00013566028939181128,
"loss": 2.5907,
"step": 15
},
{
"epoch": 0.09,
"learning_rate": 0.00013447072741690948,
"loss": 2.4827,
"step": 20
},
{
"epoch": 0.11,
"learning_rate": 0.00013295152152507697,
"loss": 2.4145,
"step": 25
},
{
"epoch": 0.14,
"learning_rate": 0.0001311102744406261,
"loss": 2.1817,
"step": 30
},
{
"epoch": 0.16,
"learning_rate": 0.00012895620051300705,
"loss": 2.3325,
"step": 35
},
{
"epoch": 0.18,
"learning_rate": 0.0001265000796044546,
"loss": 2.0885,
"step": 40
},
{
"epoch": 0.2,
"learning_rate": 0.00012375420314317213,
"loss": 2.1847,
"step": 45
},
{
"epoch": 0.23,
"learning_rate": 0.00012073231261202464,
"loss": 2.1728,
"step": 50
},
{
"epoch": 0.25,
"learning_rate": 0.00011744953078056836,
"loss": 2.0993,
"step": 55
},
{
"epoch": 0.27,
"learning_rate": 0.00011392228602455959,
"loss": 2.3687,
"step": 60
},
{
"epoch": 0.29,
"learning_rate": 0.0001101682301116794,
"loss": 1.9625,
"step": 65
},
{
"epoch": 0.32,
"learning_rate": 0.0001062061498649071,
"loss": 2.0389,
"step": 70
},
{
"epoch": 0.34,
"learning_rate": 0.00010205587314561534,
"loss": 2.098,
"step": 75
},
{
"epoch": 0.36,
"learning_rate": 9.773816962688569e-05,
"loss": 2.1281,
"step": 80
},
{
"epoch": 0.38,
"learning_rate": 9.327464685361516e-05,
"loss": 1.9415,
"step": 85
},
{
"epoch": 0.41,
"learning_rate": 8.868764210957135e-05,
"loss": 2.3485,
"step": 90
},
{
"epoch": 0.43,
"learning_rate": 8.400011063253747e-05,
"loss": 2.1515,
"step": 95
},
{
"epoch": 0.45,
"learning_rate": 7.923551073696397e-05,
"loss": 1.8548,
"step": 100
},
{
"epoch": 0.47,
"learning_rate": 7.441768641901993e-05,
"loss": 1.969,
"step": 105
},
{
"epoch": 0.5,
"learning_rate": 6.957074803153629e-05,
"loss": 2.0221,
"step": 110
},
{
"epoch": 0.52,
"learning_rate": 6.471895162599185e-05,
"loss": 2.0988,
"step": 115
},
{
"epoch": 0.54,
"learning_rate": 5.9886577565364115e-05,
"loss": 1.7345,
"step": 120
},
{
"epoch": 0.56,
"learning_rate": 5.509780901531541e-05,
"loss": 1.9466,
"step": 125
},
{
"epoch": 0.59,
"learning_rate": 5.037661092179389e-05,
"loss": 2.1125,
"step": 130
},
{
"epoch": 0.61,
"learning_rate": 4.574661008069485e-05,
"loss": 2.0488,
"step": 135
},
{
"epoch": 0.63,
"learning_rate": 4.123097689976244e-05,
"loss": 2.0849,
"step": 140
},
{
"epoch": 0.65,
"learning_rate": 3.685230944444331e-05,
"loss": 1.8891,
"step": 145
},
{
"epoch": 0.68,
"learning_rate": 3.263252034797391e-05,
"loss": 1.8136,
"step": 150
},
{
"epoch": 0.7,
"learning_rate": 2.8592727151649647e-05,
"loss": 1.9213,
"step": 155
},
{
"epoch": 0.72,
"learning_rate": 2.475314662405725e-05,
"loss": 1.9865,
"step": 160
},
{
"epoch": 0.74,
"learning_rate": 2.113299358814034e-05,
"loss": 2.2742,
"step": 165
},
{
"epoch": 0.77,
"learning_rate": 1.7750384762408767e-05,
"loss": 1.8535,
"step": 170
},
{
"epoch": 0.79,
"learning_rate": 1.4622248097508915e-05,
"loss": 2.1606,
"step": 175
},
{
"epoch": 0.81,
"learning_rate": 1.1764238061872434e-05,
"loss": 1.7923,
"step": 180
},
{
"epoch": 0.83,
"learning_rate": 9.190657300387505e-06,
"loss": 1.8476,
"step": 185
},
{
"epoch": 0.86,
"learning_rate": 6.91438505814559e-06,
"loss": 1.9976,
"step": 190
},
{
"epoch": 0.88,
"learning_rate": 4.9468127274594585e-06,
"loss": 1.9706,
"step": 195
},
{
"epoch": 0.9,
"learning_rate": 3.2977868407019207e-06,
"loss": 1.982,
"step": 200
},
{
"epoch": 0.92,
"learning_rate": 1.975559794252193e-06,
"loss": 1.9318,
"step": 205
},
{
"epoch": 0.95,
"learning_rate": 9.867485501471999e-07,
"loss": 2.035,
"step": 210
},
{
"epoch": 0.97,
"learning_rate": 3.363015221117509e-07,
"loss": 1.9653,
"step": 215
},
{
"epoch": 0.99,
"learning_rate": 2.7473811683140093e-08,
"loss": 1.8408,
"step": 220
},
{
"epoch": 1.0,
"eval_loss": 1.924851655960083,
"eval_runtime": 14.2025,
"eval_samples_per_second": 22.672,
"eval_steps_per_second": 2.887,
"step": 222
},
{
"epoch": 1.0,
"eval_loss": 1.7440342903137207,
"eval_runtime": 17.6835,
"eval_samples_per_second": 17.361,
"eval_steps_per_second": 2.205,
"step": 224
},
{
"epoch": 1.0,
"learning_rate": 6.746689293153874e-09,
"loss": 2.1578,
"step": 225
},
{
"epoch": 1.03,
"learning_rate": 2.427415035313232e-07,
"loss": 1.867,
"step": 230
},
{
"epoch": 1.05,
"learning_rate": 8.147449052190903e-07,
"loss": 1.7004,
"step": 235
},
{
"epoch": 1.07,
"learning_rate": 1.7199452243268996e-06,
"loss": 1.8748,
"step": 240
},
{
"epoch": 1.09,
"learning_rate": 2.9538929687704672e-06,
"loss": 1.7175,
"step": 245
},
{
"epoch": 1.12,
"learning_rate": 4.510522695790453e-06,
"loss": 1.9134,
"step": 250
},
{
"epoch": 1.14,
"learning_rate": 6.382182826501689e-06,
"loss": 1.8779,
"step": 255
},
{
"epoch": 1.16,
"learning_rate": 8.559673257059505e-06,
"loss": 1.9084,
"step": 260
},
{
"epoch": 1.18,
"learning_rate": 1.1032290581565944e-05,
"loss": 1.7949,
"step": 265
},
{
"epoch": 1.21,
"learning_rate": 1.3787880704424101e-05,
"loss": 1.7769,
"step": 270
},
{
"epoch": 1.23,
"learning_rate": 1.681289858352526e-05,
"loss": 1.724,
"step": 275
},
{
"epoch": 1.25,
"learning_rate": 2.009247481060283e-05,
"loss": 1.8311,
"step": 280
},
{
"epoch": 1.27,
"learning_rate": 2.3610488701478854e-05,
"loss": 1.7348,
"step": 285
},
{
"epoch": 1.29,
"learning_rate": 2.734964753692934e-05,
"loss": 1.768,
"step": 290
},
{
"epoch": 1.32,
"learning_rate": 3.129157156466198e-05,
"loss": 1.7488,
"step": 295
},
{
"epoch": 1.34,
"learning_rate": 3.541688434458052e-05,
"loss": 1.9316,
"step": 300
},
{
"epoch": 1.36,
"learning_rate": 3.970530799324656e-05,
"loss": 1.7607,
"step": 305
},
{
"epoch": 1.38,
"learning_rate": 4.413576285936533e-05,
"loss": 1.6681,
"step": 310
},
{
"epoch": 1.41,
"learning_rate": 4.8686471140343875e-05,
"loss": 2.065,
"step": 315
},
{
"epoch": 1.43,
"learning_rate": 5.333506393059682e-05,
"loss": 1.848,
"step": 320
},
{
"epoch": 1.45,
"learning_rate": 5.805869117540521e-05,
"loss": 1.8802,
"step": 325
},
{
"epoch": 1.47,
"learning_rate": 6.283413398985365e-05,
"loss": 1.7551,
"step": 330
},
{
"epoch": 1.5,
"learning_rate": 6.763791879074358e-05,
"loss": 1.7544,
"step": 335
},
{
"epoch": 1.52,
"learning_rate": 7.244643268047132e-05,
"loss": 1.793,
"step": 340
},
{
"epoch": 1.54,
"learning_rate": 7.723603951570171e-05,
"loss": 1.7917,
"step": 345
},
{
"epoch": 1.56,
"learning_rate": 8.19831960903064e-05,
"loss": 1.63,
"step": 350
},
{
"epoch": 1.58,
"learning_rate": 8.666456786147141e-05,
"loss": 2.0457,
"step": 355
},
{
"epoch": 1.61,
"learning_rate": 9.125714365012444e-05,
"loss": 2.0322,
"step": 360
},
{
"epoch": 1.63,
"learning_rate": 9.5738348751875e-05,
"loss": 1.687,
"step": 365
},
{
"epoch": 1.65,
"learning_rate": 0.00010008615590247305,
"loss": 1.9025,
"step": 370
},
{
"epoch": 1.67,
"learning_rate": 0.00010427919355233763,
"loss": 1.7503,
"step": 375
},
{
"epoch": 1.7,
"learning_rate": 0.00010829685091793463,
"loss": 1.7749,
"step": 380
},
{
"epoch": 1.72,
"learning_rate": 0.00011211937929362609,
"loss": 1.6852,
"step": 385
},
{
"epoch": 1.74,
"learning_rate": 0.00011572798912599424,
"loss": 1.8013,
"step": 390
},
{
"epoch": 1.76,
"learning_rate": 0.00011910494237347491,
"loss": 1.5069,
"step": 395
},
{
"epoch": 1.79,
"learning_rate": 0.00012223363969730684,
"loss": 1.5859,
"step": 400
},
{
"epoch": 1.81,
"learning_rate": 0.00012509870205521334,
"loss": 1.5894,
"step": 405
},
{
"epoch": 1.83,
"learning_rate": 0.00012768604629674322,
"loss": 1.9085,
"step": 410
},
{
"epoch": 1.85,
"learning_rate": 0.00012998295438868449,
"loss": 1.8982,
"step": 415
},
{
"epoch": 1.88,
"learning_rate": 0.00013197813593027427,
"loss": 1.9507,
"step": 420
},
{
"epoch": 1.9,
"learning_rate": 0.00013366178365091335,
"loss": 1.7504,
"step": 425
},
{
"epoch": 1.92,
"learning_rate": 0.00013502562161758715,
"loss": 1.7843,
"step": 430
},
{
"epoch": 1.94,
"learning_rate": 0.00013606294591503093,
"loss": 1.6083,
"step": 435
},
{
"epoch": 1.96,
"learning_rate": 0.00013676865759867644,
"loss": 1.5864,
"step": 440
},
{
"epoch": 1.99,
"learning_rate": 0.00013713928775840084,
"loss": 1.8849,
"step": 445
},
{
"epoch": 2.0,
"eval_loss": 1.5412778854370117,
"eval_runtime": 17.7637,
"eval_samples_per_second": 17.282,
"eval_steps_per_second": 2.195,
"step": 448
},
{
"epoch": 2.02,
"learning_rate": 0.00013709110969999672,
"loss": 1.6748,
"step": 450
},
{
"epoch": 2.04,
"learning_rate": 0.0001366493352528768,
"loss": 1.7921,
"step": 455
},
{
"epoch": 2.06,
"learning_rate": 0.00013587006036997336,
"loss": 1.6524,
"step": 460
},
{
"epoch": 2.09,
"learning_rate": 0.00013475714997680845,
"loss": 1.5413,
"step": 465
},
{
"epoch": 2.11,
"learning_rate": 0.00013331612371205717,
"loss": 1.5745,
"step": 470
},
{
"epoch": 2.13,
"learning_rate": 0.0001315541285521084,
"loss": 1.4608,
"step": 475
},
{
"epoch": 2.15,
"learning_rate": 0.0001294799033646105,
"loss": 1.637,
"step": 480
},
{
"epoch": 2.17,
"learning_rate": 0.00012710373556680405,
"loss": 1.5212,
"step": 485
},
{
"epoch": 2.2,
"learning_rate": 0.00012443741010360104,
"loss": 1.3742,
"step": 490
},
{
"epoch": 2.22,
"learning_rate": 0.00012149415099846082,
"loss": 1.5598,
"step": 495
},
{
"epoch": 2.24,
"learning_rate": 0.0001182885557669493,
"loss": 1.5528,
"step": 500
},
{
"epoch": 2.26,
"learning_rate": 0.00011483652301826763,
"loss": 1.3306,
"step": 505
},
{
"epoch": 2.29,
"learning_rate": 0.00011115517360381999,
"loss": 1.6203,
"step": 510
},
{
"epoch": 2.31,
"learning_rate": 0.00010726276570389622,
"loss": 1.5496,
"step": 515
},
{
"epoch": 2.33,
"learning_rate": 0.00010317860427360839,
"loss": 1.5886,
"step": 520
},
{
"epoch": 2.35,
"learning_rate": 9.892294529719824e-05,
"loss": 1.5832,
"step": 525
},
{
"epoch": 2.38,
"learning_rate": 9.451689532557822e-05,
"loss": 1.374,
"step": 530
},
{
"epoch": 2.4,
"learning_rate": 8.998230679536645e-05,
"loss": 1.615,
"step": 535
},
{
"epoch": 2.42,
"learning_rate": 8.534166964859326e-05,
"loss": 1.2303,
"step": 540
},
{
"epoch": 2.44,
"learning_rate": 8.061799979060907e-05,
"loss": 1.5842,
"step": 545
},
{
"epoch": 2.47,
"learning_rate": 7.5834724939402e-05,
"loss": 1.6827,
"step": 550
},
{
"epoch": 2.49,
"learning_rate": 7.10155684324705e-05,
"loss": 1.3135,
"step": 555
},
{
"epoch": 2.51,
"learning_rate": 6.618443156752956e-05,
"loss": 1.2597,
"step": 560
},
{
"epoch": 2.53,
"learning_rate": 6.136527506059804e-05,
"loss": 1.3309,
"step": 565
},
{
"epoch": 2.56,
"learning_rate": 5.658200020939097e-05,
"loss": 1.3751,
"step": 570
},
{
"epoch": 2.58,
"learning_rate": 5.185833035140678e-05,
"loss": 1.4378,
"step": 575
},
{
"epoch": 2.6,
"learning_rate": 4.721769320463358e-05,
"loss": 1.3661,
"step": 580
},
{
"epoch": 2.62,
"learning_rate": 4.268310467442181e-05,
"loss": 1.4564,
"step": 585
},
{
"epoch": 2.65,
"learning_rate": 3.827705470280179e-05,
"loss": 1.2126,
"step": 590
},
{
"epoch": 2.67,
"learning_rate": 3.4021395726391656e-05,
"loss": 1.3859,
"step": 595
},
{
"epoch": 2.69,
"learning_rate": 2.9937234296103807e-05,
"loss": 1.4162,
"step": 600
},
{
"epoch": 2.71,
"learning_rate": 2.6044826396180043e-05,
"loss": 1.3521,
"step": 605
},
{
"epoch": 2.74,
"learning_rate": 2.236347698173244e-05,
"loss": 1.4423,
"step": 610
},
{
"epoch": 2.76,
"learning_rate": 1.8911444233050724e-05,
"loss": 1.2658,
"step": 615
},
{
"epoch": 2.78,
"learning_rate": 1.5705849001539244e-05,
"loss": 1.2766,
"step": 620
},
{
"epoch": 2.8,
"learning_rate": 1.2762589896399027e-05,
"loss": 1.2592,
"step": 625
},
{
"epoch": 2.83,
"learning_rate": 1.0096264433195999e-05,
"loss": 1.4557,
"step": 630
},
{
"epoch": 2.85,
"learning_rate": 7.72009663538956e-06,
"loss": 1.4331,
"step": 635
},
{
"epoch": 2.87,
"learning_rate": 5.6458714478916046e-06,
"loss": 1.5097,
"step": 640
},
{
"epoch": 2.89,
"learning_rate": 3.883876287942861e-06,
"loss": 1.474,
"step": 645
},
{
"epoch": 2.91,
"learning_rate": 2.4428500231915692e-06,
"loss": 1.4667,
"step": 650
},
{
"epoch": 2.94,
"learning_rate": 1.32993963002664e-06,
"loss": 1.3699,
"step": 655
},
{
"epoch": 2.96,
"learning_rate": 5.506647471231935e-07,
"loss": 1.4343,
"step": 660
},
{
"epoch": 2.98,
"learning_rate": 1.0889030000328246e-07,
"loss": 1.3816,
"step": 665
},
{
"epoch": 3.0,
"eval_loss": 1.3440579175949097,
"eval_runtime": 2.2108,
"eval_samples_per_second": 142.485,
"eval_steps_per_second": 18.093,
"step": 669
},
{
"epoch": 3.0,
"learning_rate": 6.807332383425125e-09,
"loss": 1.2189,
"step": 670
},
{
"epoch": 3.03,
"learning_rate": 2.4492213937922245e-07,
"loss": 1.2422,
"step": 675
},
{
"epoch": 3.05,
"learning_rate": 8.22053756462089e-07,
"loss": 1.3055,
"step": 680
},
{
"epoch": 3.07,
"learning_rate": 1.735339816622185e-06,
"loss": 1.4475,
"step": 685
},
{
"epoch": 3.09,
"learning_rate": 2.980250746687993e-06,
"loss": 1.2442,
"step": 690
},
{
"epoch": 3.12,
"learning_rate": 4.550612232394036e-06,
"loss": 1.4277,
"step": 695
},
{
"epoch": 3.14,
"learning_rate": 6.438635840777816e-06,
"loss": 1.3777,
"step": 700
},
{
"epoch": 3.16,
"learning_rate": 8.634957648029665e-06,
"loss": 1.3328,
"step": 705
},
{
"epoch": 3.18,
"learning_rate": 1.1128684681215911e-05,
"loss": 1.3339,
"step": 710
},
{
"epoch": 3.21,
"learning_rate": 1.3907448943539908e-05,
"loss": 1.2034,
"step": 715
},
{
"epoch": 3.23,
"learning_rate": 1.6957468755195406e-05,
"loss": 1.1218,
"step": 720
},
{
"epoch": 3.25,
"learning_rate": 2.0263617105584107e-05,
"loss": 1.2544,
"step": 725
},
{
"epoch": 3.27,
"learning_rate": 2.3809496677893602e-05,
"loss": 1.1479,
"step": 730
},
{
"epoch": 3.3,
"learning_rate": 2.757752117394104e-05,
"loss": 1.1837,
"step": 735
},
{
"epoch": 3.32,
"learning_rate": 3.1549002535941176e-05,
"loss": 1.3242,
"step": 740
},
{
"epoch": 3.34,
"learning_rate": 3.570424363260841e-05,
"loss": 1.1755,
"step": 745
},
{
"epoch": 3.36,
"learning_rate": 4.002263594990341e-05,
"loss": 1.2206,
"step": 750
},
{
"epoch": 3.39,
"learning_rate": 4.448276180191424e-05,
"loss": 1.219,
"step": 755
},
{
"epoch": 3.41,
"learning_rate": 4.9062500554940324e-05,
"loss": 1.3741,
"step": 760
},
{
"epoch": 3.43,
"learning_rate": 5.373913833794688e-05,
"loss": 1.3206,
"step": 765
},
{
"epoch": 3.45,
"learning_rate": 5.848948069526606e-05,
"loss": 1.2842,
"step": 770
},
{
"epoch": 3.48,
"learning_rate": 6.32899676228247e-05,
"loss": 1.4653,
"step": 775
},
{
"epoch": 3.5,
"learning_rate": 6.811679041736369e-05,
"loss": 1.0829,
"step": 780
},
{
"epoch": 3.52,
"learning_rate": 7.294600975911534e-05,
"loss": 1.2317,
"step": 785
},
{
"epoch": 3.54,
"learning_rate": 7.775367444229201e-05,
"loss": 1.0955,
"step": 790
},
{
"epoch": 3.57,
"learning_rate": 8.251594016452641e-05,
"loss": 1.3158,
"step": 795
},
{
"epoch": 3.59,
"learning_rate": 8.720918778610773e-05,
"loss": 1.2593,
"step": 800
},
{
"epoch": 3.61,
"learning_rate": 9.181014047249159e-05,
"loss": 1.1672,
"step": 805
},
{
"epoch": 3.63,
"learning_rate": 9.629597913909913e-05,
"loss": 1.2407,
"step": 810
},
{
"epoch": 3.65,
"learning_rate": 0.0001006444556258389,
"loss": 1.2637,
"step": 815
},
{
"epoch": 3.68,
"learning_rate": 0.00010483400304004644,
"loss": 1.4097,
"step": 820
},
{
"epoch": 3.7,
"learning_rate": 0.00010884384272058186,
"loss": 1.3197,
"step": 825
},
{
"epoch": 3.72,
"learning_rate": 0.00011265408729257823,
"loss": 1.0697,
"step": 830
},
{
"epoch": 3.74,
"learning_rate": 0.00011624583930172985,
"loss": 1.079,
"step": 835
},
{
"epoch": 3.77,
"learning_rate": 0.00011960128493892567,
"loss": 1.2747,
"step": 840
},
{
"epoch": 3.79,
"learning_rate": 0.0001227037823903907,
"loss": 1.3586,
"step": 845
},
{
"epoch": 3.81,
"learning_rate": 0.000125537944375147,
"loss": 1.3766,
"step": 850
},
{
"epoch": 3.83,
"learning_rate": 0.00012808971446044075,
"loss": 1.2691,
"step": 855
},
{
"epoch": 3.86,
"learning_rate": 0.00013034643677663536,
"loss": 1.1551,
"step": 860
},
{
"epoch": 3.88,
"learning_rate": 0.00013229691878581222,
"loss": 1.3372,
"step": 865
},
{
"epoch": 3.9,
"learning_rate": 0.00013393148679276723,
"loss": 1.2608,
"step": 870
},
{
"epoch": 3.92,
"learning_rate": 0.00013524203392308896,
"loss": 1.1078,
"step": 875
},
{
"epoch": 3.95,
"learning_rate": 0.00013622206033036527,
"loss": 1.0532,
"step": 880
},
{
"epoch": 3.97,
"learning_rate": 0.00013686670543310324,
"loss": 1.1992,
"step": 885
},
{
"epoch": 3.99,
"learning_rate": 0.00013717277202148013,
"loss": 1.1418,
"step": 890
},
{
"epoch": 4.0,
"eval_loss": 1.3141199350357056,
"eval_runtime": 2.2159,
"eval_samples_per_second": 142.153,
"eval_steps_per_second": 18.051,
"step": 892
},
{
"epoch": 4.01,
"learning_rate": 0.000137138742114364,
"loss": 1.0312,
"step": 895
},
{
"epoch": 4.04,
"learning_rate": 0.00013676478448795875,
"loss": 1.0096,
"step": 900
},
{
"epoch": 4.06,
"learning_rate": 0.00013605275383873428,
"loss": 1.1906,
"step": 905
},
{
"epoch": 4.08,
"learning_rate": 0.00013500618158479363,
"loss": 1.1664,
"step": 910
},
{
"epoch": 4.1,
"learning_rate": 0.00013363025835129812,
"loss": 1.0422,
"step": 915
},
{
"epoch": 4.13,
"learning_rate": 0.00013193180822681808,
"loss": 1.0758,
"step": 920
},
{
"epoch": 4.15,
"learning_rate": 0.0001299192549182867,
"loss": 1.3852,
"step": 925
},
{
"epoch": 4.17,
"learning_rate": 0.0001276025799724176,
"loss": 1.2089,
"step": 930
},
{
"epoch": 4.19,
"learning_rate": 0.00012499327327079299,
"loss": 1.1252,
"step": 935
},
{
"epoch": 4.22,
"learning_rate": 0.00012210427604414836,
"loss": 1.2521,
"step": 940
},
{
"epoch": 4.24,
"learning_rate": 0.00011894991668848237,
"loss": 1.2428,
"step": 945
},
{
"epoch": 4.26,
"learning_rate": 0.00011554583970132328,
"loss": 1.1082,
"step": 950
},
{
"epoch": 4.28,
"learning_rate": 0.00011190892809059987,
"loss": 1.0971,
"step": 955
},
{
"epoch": 4.3,
"learning_rate": 0.00010805721964094202,
"loss": 1.0657,
"step": 960
},
{
"epoch": 4.33,
"learning_rate": 0.00010400981745270263,
"loss": 1.0439,
"step": 965
},
{
"epoch": 4.35,
"learning_rate": 9.978679519739164e-05,
"loss": 1.1785,
"step": 970
},
{
"epoch": 4.37,
"learning_rate": 9.540909755942318e-05,
"loss": 1.1708,
"step": 975
},
{
"epoch": 4.39,
"learning_rate": 9.089843635795119e-05,
"loss": 1.0215,
"step": 980
},
{
"epoch": 4.42,
"learning_rate": 8.627718286398852e-05,
"loss": 1.1393,
"step": 985
},
{
"epoch": 4.44,
"learning_rate": 8.156825684687972e-05,
"loss": 1.0526,
"step": 990
},
{
"epoch": 4.46,
"learning_rate": 7.679501290041993e-05,
"loss": 0.9887,
"step": 995
},
{
"epoch": 4.48,
"learning_rate": 7.198112461239741e-05,
"loss": 0.8743,
"step": 1000
},
{
"epoch": 4.51,
"learning_rate": 6.715046715204013e-05,
"loss": 0.9096,
"step": 1005
},
{
"epoch": 4.53,
"learning_rate": 6.232699885769075e-05,
"loss": 1.2684,
"step": 1010
},
{
"epoch": 4.55,
"learning_rate": 5.753464241199265e-05,
"loss": 1.0412,
"step": 1015
},
{
"epoch": 4.57,
"learning_rate": 5.279716619391422e-05,
"loss": 0.9954,
"step": 1020
},
{
"epoch": 4.6,
"learning_rate": 4.813806639606615e-05,
"loss": 0.8753,
"step": 1025
},
{
"epoch": 4.62,
"learning_rate": 4.358045049196426e-05,
"loss": 0.8638,
"step": 1030
},
{
"epoch": 4.64,
"learning_rate": 3.9146922631201556e-05,
"loss": 1.0574,
"step": 1035
},
{
"epoch": 4.66,
"learning_rate": 3.485947153092735e-05,
"loss": 0.8752,
"step": 1040
},
{
"epoch": 4.69,
"learning_rate": 3.073936141965131e-05,
"loss": 1.0359,
"step": 1045
},
{
"epoch": 4.71,
"learning_rate": 2.680702657425298e-05,
"loss": 1.0701,
"step": 1050
},
{
"epoch": 4.73,
"learning_rate": 2.3081969973255665e-05,
"loss": 0.9777,
"step": 1055
},
{
"epoch": 4.75,
"learning_rate": 1.9582666569008116e-05,
"loss": 0.9861,
"step": 1060
},
{
"epoch": 4.78,
"learning_rate": 1.632647165850912e-05,
"loss": 1.0365,
"step": 1065
},
{
"epoch": 4.8,
"learning_rate": 1.3329534807322873e-05,
"loss": 0.9034,
"step": 1070
},
{
"epoch": 4.82,
"learning_rate": 1.0606719753493918e-05,
"loss": 1.0476,
"step": 1075
},
{
"epoch": 4.84,
"learning_rate": 8.171530688706377e-06,
"loss": 1.0288,
"step": 1080
},
{
"epoch": 4.87,
"learning_rate": 6.036045282308682e-06,
"loss": 1.1233,
"step": 1085
},
{
"epoch": 4.89,
"learning_rate": 4.210854780381334e-06,
"loss": 0.9158,
"step": 1090
},
{
"epoch": 4.91,
"learning_rate": 2.705011476932832e-06,
"loss": 1.0037,
"step": 1095
},
{
"epoch": 4.93,
"learning_rate": 1.5259838177494246e-06,
"loss": 1.279,
"step": 1100
},
{
"epoch": 4.96,
"learning_rate": 6.796193595677371e-07,
"loss": 1.0122,
"step": 1105
},
{
"epoch": 4.98,
"learning_rate": 1.7011576827836676e-07,
"loss": 1.058,
"step": 1110
},
{
"epoch": 5.0,
"learning_rate": 0.0,
"loss": 1.0917,
"step": 1115
},
{
"epoch": 5.0,
"eval_loss": 1.2071579694747925,
"eval_runtime": 2.221,
"eval_samples_per_second": 141.829,
"eval_steps_per_second": 18.01,
"step": 1115
},
{
"epoch": 5.02,
"learning_rate": 1.7011576827835916e-07,
"loss": 0.9624,
"step": 1120
},
{
"epoch": 5.04,
"learning_rate": 6.796193595677219e-07,
"loss": 1.0903,
"step": 1125
},
{
"epoch": 5.07,
"learning_rate": 1.5259838177494094e-06,
"loss": 0.8195,
"step": 1130
},
{
"epoch": 5.09,
"learning_rate": 2.705011476932809e-06,
"loss": 0.9007,
"step": 1135
},
{
"epoch": 5.11,
"learning_rate": 4.210854780381303e-06,
"loss": 0.6803,
"step": 1140
},
{
"epoch": 5.13,
"learning_rate": 6.036045282308651e-06,
"loss": 0.9312,
"step": 1145
},
{
"epoch": 5.16,
"learning_rate": 8.171530688706338e-06,
"loss": 1.0329,
"step": 1150
},
{
"epoch": 5.18,
"learning_rate": 1.0606719753493872e-05,
"loss": 0.9101,
"step": 1155
},
{
"epoch": 5.2,
"learning_rate": 1.3329534807322828e-05,
"loss": 0.8229,
"step": 1160
},
{
"epoch": 5.22,
"learning_rate": 1.6326471658509066e-05,
"loss": 1.0693,
"step": 1165
},
{
"epoch": 5.25,
"learning_rate": 1.9582666569008055e-05,
"loss": 0.7779,
"step": 1170
},
{
"epoch": 5.27,
"learning_rate": 2.3081969973255604e-05,
"loss": 0.7962,
"step": 1175
},
{
"epoch": 5.29,
"learning_rate": 2.680702657425292e-05,
"loss": 0.9127,
"step": 1180
},
{
"epoch": 5.31,
"learning_rate": 3.073936141965114e-05,
"loss": 0.8227,
"step": 1185
},
{
"epoch": 5.34,
"learning_rate": 3.4859471530927266e-05,
"loss": 0.9535,
"step": 1190
},
{
"epoch": 5.36,
"learning_rate": 3.914692263120148e-05,
"loss": 0.7736,
"step": 1195
},
{
"epoch": 5.38,
"learning_rate": 4.358045049196419e-05,
"loss": 0.9111,
"step": 1200
},
{
"epoch": 5.4,
"learning_rate": 4.813806639606595e-05,
"loss": 0.9058,
"step": 1205
},
{
"epoch": 5.43,
"learning_rate": 5.279716619391414e-05,
"loss": 0.8259,
"step": 1210
},
{
"epoch": 5.45,
"learning_rate": 5.753464241199256e-05,
"loss": 0.9812,
"step": 1215
},
{
"epoch": 5.47,
"learning_rate": 6.232699885769054e-05,
"loss": 1.0264,
"step": 1220
},
{
"epoch": 5.49,
"learning_rate": 6.715046715203992e-05,
"loss": 0.8185,
"step": 1225
},
{
"epoch": 5.52,
"learning_rate": 7.198112461239733e-05,
"loss": 0.8403,
"step": 1230
},
{
"epoch": 5.54,
"learning_rate": 7.679501290041973e-05,
"loss": 0.914,
"step": 1235
},
{
"epoch": 5.56,
"learning_rate": 8.15682568468795e-05,
"loss": 0.9836,
"step": 1240
},
{
"epoch": 5.58,
"learning_rate": 8.627718286398833e-05,
"loss": 0.9734,
"step": 1245
},
{
"epoch": 5.61,
"learning_rate": 9.089843635795102e-05,
"loss": 0.8468,
"step": 1250
},
{
"epoch": 5.63,
"learning_rate": 9.540909755942299e-05,
"loss": 0.835,
"step": 1255
},
{
"epoch": 5.65,
"learning_rate": 9.978679519739145e-05,
"loss": 1.0896,
"step": 1260
},
{
"epoch": 5.67,
"learning_rate": 0.00010400981745270244,
"loss": 0.7718,
"step": 1265
},
{
"epoch": 5.7,
"learning_rate": 0.00010805721964094184,
"loss": 0.8329,
"step": 1270
},
{
"epoch": 5.72,
"learning_rate": 0.0001119089280905997,
"loss": 0.9622,
"step": 1275
},
{
"epoch": 5.74,
"learning_rate": 0.0001155458397013233,
"loss": 0.8208,
"step": 1280
},
{
"epoch": 5.76,
"learning_rate": 0.00011894991668848222,
"loss": 0.9137,
"step": 1285
},
{
"epoch": 5.78,
"learning_rate": 0.00012210427604414823,
"loss": 1.0149,
"step": 1290
},
{
"epoch": 5.81,
"learning_rate": 0.000124993273270793,
"loss": 0.9856,
"step": 1295
},
{
"epoch": 5.83,
"learning_rate": 0.0001276025799724176,
"loss": 0.9686,
"step": 1300
},
{
"epoch": 5.85,
"learning_rate": 0.0001299192549182866,
"loss": 0.9409,
"step": 1305
},
{
"epoch": 5.87,
"learning_rate": 0.00013193180822681808,
"loss": 1.214,
"step": 1310
},
{
"epoch": 5.9,
"learning_rate": 0.00013363025835129815,
"loss": 0.8486,
"step": 1315
},
{
"epoch": 5.92,
"learning_rate": 0.00013500618158479366,
"loss": 1.0083,
"step": 1320
},
{
"epoch": 5.94,
"learning_rate": 0.00013605275383873428,
"loss": 0.9284,
"step": 1325
},
{
"epoch": 5.96,
"learning_rate": 0.00013676478448795875,
"loss": 1.0467,
"step": 1330
},
{
"epoch": 5.99,
"learning_rate": 0.00013713874211436402,
"loss": 0.9702,
"step": 1335
},
{
"epoch": 6.0,
"eval_loss": 1.2323447465896606,
"eval_runtime": 2.2212,
"eval_samples_per_second": 141.817,
"eval_steps_per_second": 18.008,
"step": 1338
},
{
"epoch": 6.01,
"learning_rate": 0.00013717277202148013,
"loss": 0.9361,
"step": 1340
},
{
"epoch": 6.03,
"learning_rate": 0.00013686670543310324,
"loss": 0.951,
"step": 1345
},
{
"epoch": 6.05,
"learning_rate": 0.00013622206033036527,
"loss": 0.7387,
"step": 1350
},
{
"epoch": 6.08,
"learning_rate": 0.0001352420339230889,
"loss": 0.8237,
"step": 1355
},
{
"epoch": 6.1,
"learning_rate": 0.0001339314867927672,
"loss": 0.8428,
"step": 1360
},
{
"epoch": 6.12,
"learning_rate": 0.00013229691878581222,
"loss": 0.8352,
"step": 1365
},
{
"epoch": 6.14,
"learning_rate": 0.00013034643677663527,
"loss": 0.8816,
"step": 1370
},
{
"epoch": 6.17,
"learning_rate": 0.00012808971446044072,
"loss": 0.781,
"step": 1375
},
{
"epoch": 6.19,
"learning_rate": 0.00012553794437514699,
"loss": 0.7146,
"step": 1380
},
{
"epoch": 6.21,
"learning_rate": 0.0001227037823903906,
"loss": 0.9361,
"step": 1385
},
{
"epoch": 6.23,
"learning_rate": 0.00011960128493892572,
"loss": 0.8889,
"step": 1390
},
{
"epoch": 6.26,
"learning_rate": 0.00011624583930172982,
"loss": 0.9329,
"step": 1395
},
{
"epoch": 6.28,
"learning_rate": 0.0001126540872925782,
"loss": 0.7853,
"step": 1400
},
{
"epoch": 6.3,
"learning_rate": 0.00010884384272058193,
"loss": 0.7188,
"step": 1405
},
{
"epoch": 6.32,
"learning_rate": 0.00010483400304004653,
"loss": 0.7662,
"step": 1410
},
{
"epoch": 6.35,
"learning_rate": 0.00010064445562583886,
"loss": 0.7705,
"step": 1415
},
{
"epoch": 6.37,
"learning_rate": 9.629597913909932e-05,
"loss": 0.9695,
"step": 1420
},
{
"epoch": 6.39,
"learning_rate": 9.181014047249165e-05,
"loss": 0.6245,
"step": 1425
},
{
"epoch": 6.41,
"learning_rate": 8.72091877861077e-05,
"loss": 0.9465,
"step": 1430
},
{
"epoch": 6.43,
"learning_rate": 8.25159401645266e-05,
"loss": 0.8094,
"step": 1435
},
{
"epoch": 6.46,
"learning_rate": 7.775367444229211e-05,
"loss": 0.8367,
"step": 1440
},
{
"epoch": 6.48,
"learning_rate": 7.29460097591153e-05,
"loss": 0.5786,
"step": 1445
},
{
"epoch": 6.5,
"learning_rate": 6.81167904173639e-05,
"loss": 0.8389,
"step": 1450
},
{
"epoch": 6.52,
"learning_rate": 6.328996762282478e-05,
"loss": 0.9198,
"step": 1455
},
{
"epoch": 6.55,
"learning_rate": 5.848948069526602e-05,
"loss": 0.8697,
"step": 1460
},
{
"epoch": 6.57,
"learning_rate": 5.3739138337947083e-05,
"loss": 0.805,
"step": 1465
},
{
"epoch": 6.59,
"learning_rate": 4.90625005549404e-05,
"loss": 0.7912,
"step": 1470
},
{
"epoch": 6.61,
"learning_rate": 4.448276180191432e-05,
"loss": 0.7832,
"step": 1475
},
{
"epoch": 6.64,
"learning_rate": 4.0022635949903595e-05,
"loss": 0.7555,
"step": 1480
},
{
"epoch": 6.66,
"learning_rate": 3.570424363260848e-05,
"loss": 0.88,
"step": 1485
},
{
"epoch": 6.68,
"learning_rate": 3.1549002535941244e-05,
"loss": 0.7924,
"step": 1490
},
{
"epoch": 6.7,
"learning_rate": 2.75775211739412e-05,
"loss": 0.7299,
"step": 1495
},
{
"epoch": 6.73,
"learning_rate": 2.3809496677893663e-05,
"loss": 0.9469,
"step": 1500
},
{
"epoch": 6.75,
"learning_rate": 2.0263617105584168e-05,
"loss": 0.6995,
"step": 1505
},
{
"epoch": 6.77,
"learning_rate": 1.6957468755195542e-05,
"loss": 0.8889,
"step": 1510
},
{
"epoch": 6.79,
"learning_rate": 1.3907448943539954e-05,
"loss": 0.8217,
"step": 1515
},
{
"epoch": 6.82,
"learning_rate": 1.1128684681215955e-05,
"loss": 0.7166,
"step": 1520
},
{
"epoch": 6.84,
"learning_rate": 8.63495764802965e-06,
"loss": 0.873,
"step": 1525
},
{
"epoch": 6.86,
"learning_rate": 6.4386358407778465e-06,
"loss": 0.906,
"step": 1530
},
{
"epoch": 6.88,
"learning_rate": 4.550612232394066e-06,
"loss": 0.9969,
"step": 1535
},
{
"epoch": 6.91,
"learning_rate": 2.9802507466879855e-06,
"loss": 0.6348,
"step": 1540
},
{
"epoch": 6.93,
"learning_rate": 1.735339816622208e-06,
"loss": 0.7365,
"step": 1545
},
{
"epoch": 6.95,
"learning_rate": 8.220537564620966e-07,
"loss": 0.9509,
"step": 1550
},
{
"epoch": 6.97,
"learning_rate": 2.4492213937922245e-07,
"loss": 0.8438,
"step": 1555
},
{
"epoch": 7.0,
"learning_rate": 6.807332383425125e-09,
"loss": 0.9895,
"step": 1560
},
{
"epoch": 7.0,
"eval_loss": 1.1562458276748657,
"eval_runtime": 2.2177,
"eval_samples_per_second": 142.037,
"eval_steps_per_second": 18.036,
"step": 1561
}
],
"max_steps": 1561,
"num_train_epochs": 7,
"total_flos": 1630331633664000.0,
"trial_name": null,
"trial_params": null
}