bloom-7b1-heuristics-de / trainer_state.json
atsuki-yamaguchi's picture
Upload folder using huggingface_hub
6f29aa0 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.0631463871386596,
"eval_steps": 500,
"global_step": 8000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 5e-06,
"loss": 8.0409,
"step": 5
},
{
"epoch": 0.0,
"learning_rate": 1e-05,
"loss": 7.9137,
"step": 10
},
{
"epoch": 0.0,
"learning_rate": 1.5e-05,
"loss": 7.72,
"step": 15
},
{
"epoch": 0.0,
"learning_rate": 2e-05,
"loss": 7.3026,
"step": 20
},
{
"epoch": 0.0,
"learning_rate": 2.5e-05,
"loss": 6.8062,
"step": 25
},
{
"epoch": 0.0,
"learning_rate": 3e-05,
"loss": 6.4934,
"step": 30
},
{
"epoch": 0.0,
"learning_rate": 3.5e-05,
"loss": 6.1984,
"step": 35
},
{
"epoch": 0.0,
"learning_rate": 4e-05,
"loss": 6.0188,
"step": 40
},
{
"epoch": 0.0,
"learning_rate": 4.5e-05,
"loss": 5.79,
"step": 45
},
{
"epoch": 0.0,
"learning_rate": 5e-05,
"loss": 5.6608,
"step": 50
},
{
"epoch": 0.0,
"learning_rate": 5.500000000000001e-05,
"loss": 5.5568,
"step": 55
},
{
"epoch": 0.0,
"learning_rate": 6e-05,
"loss": 5.3403,
"step": 60
},
{
"epoch": 0.0,
"learning_rate": 6.500000000000001e-05,
"loss": 5.1958,
"step": 65
},
{
"epoch": 0.0,
"learning_rate": 7e-05,
"loss": 4.9844,
"step": 70
},
{
"epoch": 0.0,
"learning_rate": 7.500000000000001e-05,
"loss": 4.8401,
"step": 75
},
{
"epoch": 0.0,
"learning_rate": 8e-05,
"loss": 4.7795,
"step": 80
},
{
"epoch": 0.0,
"learning_rate": 8.5e-05,
"loss": 4.6467,
"step": 85
},
{
"epoch": 0.0,
"learning_rate": 9e-05,
"loss": 4.5467,
"step": 90
},
{
"epoch": 0.0,
"learning_rate": 9.5e-05,
"loss": 4.4902,
"step": 95
},
{
"epoch": 0.0,
"learning_rate": 0.0001,
"loss": 4.4359,
"step": 100
},
{
"epoch": 0.0,
"learning_rate": 9.999999961506464e-05,
"loss": 4.3332,
"step": 105
},
{
"epoch": 0.0,
"learning_rate": 9.999999846025851e-05,
"loss": 4.3142,
"step": 110
},
{
"epoch": 0.0,
"learning_rate": 9.999999653558166e-05,
"loss": 4.211,
"step": 115
},
{
"epoch": 0.0,
"learning_rate": 9.999999384103412e-05,
"loss": 4.2021,
"step": 120
},
{
"epoch": 0.0,
"learning_rate": 9.999999037661592e-05,
"loss": 4.1034,
"step": 125
},
{
"epoch": 0.0,
"learning_rate": 9.999998614232713e-05,
"loss": 4.0342,
"step": 130
},
{
"epoch": 0.0,
"learning_rate": 9.99999811381678e-05,
"loss": 3.8535,
"step": 135
},
{
"epoch": 0.0,
"learning_rate": 9.9999975364138e-05,
"loss": 3.8476,
"step": 140
},
{
"epoch": 0.0,
"learning_rate": 9.999996882023784e-05,
"loss": 3.8555,
"step": 145
},
{
"epoch": 0.0,
"learning_rate": 9.99999615064674e-05,
"loss": 3.8873,
"step": 150
},
{
"epoch": 0.0,
"learning_rate": 9.99999534228268e-05,
"loss": 3.8591,
"step": 155
},
{
"epoch": 0.0,
"learning_rate": 9.999994456931619e-05,
"loss": 3.8088,
"step": 160
},
{
"epoch": 0.0,
"learning_rate": 9.999993494593566e-05,
"loss": 3.7316,
"step": 165
},
{
"epoch": 0.0,
"learning_rate": 9.999992455268539e-05,
"loss": 3.7493,
"step": 170
},
{
"epoch": 0.0,
"learning_rate": 9.999991338956554e-05,
"loss": 3.637,
"step": 175
},
{
"epoch": 0.0,
"learning_rate": 9.999990145657626e-05,
"loss": 3.68,
"step": 180
},
{
"epoch": 0.0,
"learning_rate": 9.999988875371776e-05,
"loss": 3.702,
"step": 185
},
{
"epoch": 0.0,
"learning_rate": 9.999987528099021e-05,
"loss": 3.6511,
"step": 190
},
{
"epoch": 0.0,
"learning_rate": 9.999986103839384e-05,
"loss": 3.5953,
"step": 195
},
{
"epoch": 0.0,
"learning_rate": 9.999984602592884e-05,
"loss": 3.6343,
"step": 200
},
{
"epoch": 0.0,
"learning_rate": 9.999983024359549e-05,
"loss": 3.4832,
"step": 205
},
{
"epoch": 0.0,
"learning_rate": 9.9999813691394e-05,
"loss": 3.6435,
"step": 210
},
{
"epoch": 0.0,
"learning_rate": 9.999979636932461e-05,
"loss": 3.5427,
"step": 215
},
{
"epoch": 0.0,
"learning_rate": 9.999977827738762e-05,
"loss": 3.5104,
"step": 220
},
{
"epoch": 0.0,
"learning_rate": 9.999975941558328e-05,
"loss": 3.4561,
"step": 225
},
{
"epoch": 0.0,
"learning_rate": 9.999973978391191e-05,
"loss": 3.5724,
"step": 230
},
{
"epoch": 0.0,
"learning_rate": 9.99997193823738e-05,
"loss": 3.5048,
"step": 235
},
{
"epoch": 0.0,
"learning_rate": 9.999969821096924e-05,
"loss": 3.5426,
"step": 240
},
{
"epoch": 0.0,
"learning_rate": 9.99996762696986e-05,
"loss": 3.4377,
"step": 245
},
{
"epoch": 0.0,
"learning_rate": 9.999965355856218e-05,
"loss": 3.3224,
"step": 250
},
{
"epoch": 0.0,
"learning_rate": 9.999963007756035e-05,
"loss": 3.3984,
"step": 255
},
{
"epoch": 0.0,
"learning_rate": 9.999960582669346e-05,
"loss": 3.3484,
"step": 260
},
{
"epoch": 0.0,
"learning_rate": 9.99995808059619e-05,
"loss": 3.3324,
"step": 265
},
{
"epoch": 0.0,
"learning_rate": 9.999955501536603e-05,
"loss": 3.4808,
"step": 270
},
{
"epoch": 0.0,
"learning_rate": 9.999952845490626e-05,
"loss": 3.3514,
"step": 275
},
{
"epoch": 0.0,
"learning_rate": 9.999950112458302e-05,
"loss": 3.4077,
"step": 280
},
{
"epoch": 0.0,
"learning_rate": 9.99994730243967e-05,
"loss": 3.436,
"step": 285
},
{
"epoch": 0.0,
"learning_rate": 9.999944415434774e-05,
"loss": 3.4406,
"step": 290
},
{
"epoch": 0.0,
"learning_rate": 9.99994145144366e-05,
"loss": 3.2683,
"step": 295
},
{
"epoch": 0.0,
"learning_rate": 9.999938410466371e-05,
"loss": 3.349,
"step": 300
},
{
"epoch": 0.0,
"learning_rate": 9.999935292502958e-05,
"loss": 3.3269,
"step": 305
},
{
"epoch": 0.0,
"learning_rate": 9.999932097553463e-05,
"loss": 3.3056,
"step": 310
},
{
"epoch": 0.0,
"learning_rate": 9.999928825617941e-05,
"loss": 3.3348,
"step": 315
},
{
"epoch": 0.0,
"learning_rate": 9.999925476696439e-05,
"loss": 3.2625,
"step": 320
},
{
"epoch": 0.0,
"learning_rate": 9.99992205078901e-05,
"loss": 3.3096,
"step": 325
},
{
"epoch": 0.0,
"learning_rate": 9.999918547895705e-05,
"loss": 3.3747,
"step": 330
},
{
"epoch": 0.0,
"learning_rate": 9.999914968016579e-05,
"loss": 3.2122,
"step": 335
},
{
"epoch": 0.0,
"learning_rate": 9.999911311151689e-05,
"loss": 3.1712,
"step": 340
},
{
"epoch": 0.0,
"learning_rate": 9.999907577301089e-05,
"loss": 3.1477,
"step": 345
},
{
"epoch": 0.0,
"learning_rate": 9.999903766464837e-05,
"loss": 3.1879,
"step": 350
},
{
"epoch": 0.0,
"learning_rate": 9.99989987864299e-05,
"loss": 3.2017,
"step": 355
},
{
"epoch": 0.0,
"learning_rate": 9.999895913835613e-05,
"loss": 3.2648,
"step": 360
},
{
"epoch": 0.0,
"learning_rate": 9.999891872042762e-05,
"loss": 3.2209,
"step": 365
},
{
"epoch": 0.0,
"learning_rate": 9.9998877532645e-05,
"loss": 3.1836,
"step": 370
},
{
"epoch": 0.0,
"learning_rate": 9.999883557500892e-05,
"loss": 3.1445,
"step": 375
},
{
"epoch": 0.0,
"learning_rate": 9.999879284752003e-05,
"loss": 3.0949,
"step": 380
},
{
"epoch": 0.0,
"learning_rate": 9.999874935017897e-05,
"loss": 3.1131,
"step": 385
},
{
"epoch": 0.0,
"learning_rate": 9.999870508298641e-05,
"loss": 3.126,
"step": 390
},
{
"epoch": 0.0,
"learning_rate": 9.999866004594303e-05,
"loss": 3.1439,
"step": 395
},
{
"epoch": 0.0,
"learning_rate": 9.999861423904954e-05,
"loss": 3.1464,
"step": 400
},
{
"epoch": 0.0,
"learning_rate": 9.999856766230665e-05,
"loss": 3.0833,
"step": 405
},
{
"epoch": 0.0,
"learning_rate": 9.999852031571506e-05,
"loss": 3.2007,
"step": 410
},
{
"epoch": 0.0,
"learning_rate": 9.999847219927549e-05,
"loss": 3.1336,
"step": 415
},
{
"epoch": 0.0,
"learning_rate": 9.999842331298872e-05,
"loss": 3.1676,
"step": 420
},
{
"epoch": 0.0,
"learning_rate": 9.999837365685544e-05,
"loss": 3.1989,
"step": 425
},
{
"epoch": 0.0,
"learning_rate": 9.999832323087649e-05,
"loss": 3.1047,
"step": 430
},
{
"epoch": 0.0,
"learning_rate": 9.999827203505259e-05,
"loss": 3.0741,
"step": 435
},
{
"epoch": 0.0,
"learning_rate": 9.999822006938455e-05,
"loss": 3.2121,
"step": 440
},
{
"epoch": 0.0,
"learning_rate": 9.999816733387316e-05,
"loss": 3.147,
"step": 445
},
{
"epoch": 0.0,
"learning_rate": 9.999811382851924e-05,
"loss": 3.061,
"step": 450
},
{
"epoch": 0.0,
"learning_rate": 9.999805955332362e-05,
"loss": 3.0255,
"step": 455
},
{
"epoch": 0.0,
"learning_rate": 9.999800450828711e-05,
"loss": 3.1169,
"step": 460
},
{
"epoch": 0.0,
"learning_rate": 9.99979486934106e-05,
"loss": 3.0572,
"step": 465
},
{
"epoch": 0.0,
"learning_rate": 9.99978921086949e-05,
"loss": 3.0659,
"step": 470
},
{
"epoch": 0.0,
"learning_rate": 9.999783475414093e-05,
"loss": 3.0767,
"step": 475
},
{
"epoch": 0.0,
"learning_rate": 9.999777662974952e-05,
"loss": 2.9625,
"step": 480
},
{
"epoch": 0.0,
"learning_rate": 9.999771773552162e-05,
"loss": 3.0783,
"step": 485
},
{
"epoch": 0.0,
"learning_rate": 9.999765807145811e-05,
"loss": 2.9836,
"step": 490
},
{
"epoch": 0.0,
"learning_rate": 9.99975976375599e-05,
"loss": 3.1429,
"step": 495
},
{
"epoch": 0.0,
"learning_rate": 9.999753643382793e-05,
"loss": 2.9218,
"step": 500
},
{
"epoch": 0.0,
"learning_rate": 9.999747446026315e-05,
"loss": 3.1355,
"step": 505
},
{
"epoch": 0.0,
"learning_rate": 9.999741171686649e-05,
"loss": 3.0135,
"step": 510
},
{
"epoch": 0.0,
"learning_rate": 9.999734820363895e-05,
"loss": 3.0287,
"step": 515
},
{
"epoch": 0.0,
"learning_rate": 9.99972839205815e-05,
"loss": 3.0462,
"step": 520
},
{
"epoch": 0.0,
"learning_rate": 9.999721886769511e-05,
"loss": 3.1352,
"step": 525
},
{
"epoch": 0.0,
"learning_rate": 9.999715304498078e-05,
"loss": 3.0078,
"step": 530
},
{
"epoch": 0.0,
"learning_rate": 9.999708645243955e-05,
"loss": 3.0395,
"step": 535
},
{
"epoch": 0.0,
"learning_rate": 9.999701909007242e-05,
"loss": 2.9824,
"step": 540
},
{
"epoch": 0.0,
"learning_rate": 9.999695095788044e-05,
"loss": 3.0293,
"step": 545
},
{
"epoch": 0.0,
"learning_rate": 9.999688205586468e-05,
"loss": 2.9781,
"step": 550
},
{
"epoch": 0.0,
"learning_rate": 9.999681238402615e-05,
"loss": 2.9944,
"step": 555
},
{
"epoch": 0.0,
"learning_rate": 9.999674194236597e-05,
"loss": 3.0274,
"step": 560
},
{
"epoch": 0.0,
"learning_rate": 9.999667073088519e-05,
"loss": 3.0585,
"step": 565
},
{
"epoch": 0.0,
"learning_rate": 9.999659874958493e-05,
"loss": 2.9889,
"step": 570
},
{
"epoch": 0.0,
"learning_rate": 9.99965259984663e-05,
"loss": 3.0903,
"step": 575
},
{
"epoch": 0.0,
"learning_rate": 9.999645247753039e-05,
"loss": 2.9978,
"step": 580
},
{
"epoch": 0.0,
"learning_rate": 9.999637818677836e-05,
"loss": 3.0721,
"step": 585
},
{
"epoch": 0.0,
"learning_rate": 9.999630312621136e-05,
"loss": 3.0069,
"step": 590
},
{
"epoch": 0.0,
"learning_rate": 9.999622729583051e-05,
"loss": 3.0295,
"step": 595
},
{
"epoch": 0.0,
"learning_rate": 9.999615069563701e-05,
"loss": 2.9868,
"step": 600
},
{
"epoch": 0.0,
"learning_rate": 9.999607332563205e-05,
"loss": 3.0555,
"step": 605
},
{
"epoch": 0.0,
"learning_rate": 9.999599518581676e-05,
"loss": 3.0633,
"step": 610
},
{
"epoch": 0.0,
"learning_rate": 9.999591627619242e-05,
"loss": 2.9342,
"step": 615
},
{
"epoch": 0.0,
"learning_rate": 9.999583659676019e-05,
"loss": 3.0641,
"step": 620
},
{
"epoch": 0.0,
"learning_rate": 9.999575614752133e-05,
"loss": 3.0858,
"step": 625
},
{
"epoch": 0.0,
"learning_rate": 9.999567492847706e-05,
"loss": 2.9561,
"step": 630
},
{
"epoch": 0.01,
"learning_rate": 9.999559293962864e-05,
"loss": 2.9022,
"step": 635
},
{
"epoch": 0.01,
"learning_rate": 9.99955101809773e-05,
"loss": 3.0236,
"step": 640
},
{
"epoch": 0.01,
"learning_rate": 9.999542665252437e-05,
"loss": 3.0396,
"step": 645
},
{
"epoch": 0.01,
"learning_rate": 9.999534235427109e-05,
"loss": 2.9634,
"step": 650
},
{
"epoch": 0.01,
"learning_rate": 9.999525728621879e-05,
"loss": 3.0073,
"step": 655
},
{
"epoch": 0.01,
"learning_rate": 9.999517144836876e-05,
"loss": 2.9314,
"step": 660
},
{
"epoch": 0.01,
"learning_rate": 9.999508484072233e-05,
"loss": 3.0553,
"step": 665
},
{
"epoch": 0.01,
"learning_rate": 9.999499746328083e-05,
"loss": 2.9452,
"step": 670
},
{
"epoch": 0.01,
"learning_rate": 9.99949093160456e-05,
"loss": 2.9199,
"step": 675
},
{
"epoch": 0.01,
"learning_rate": 9.9994820399018e-05,
"loss": 2.9884,
"step": 680
},
{
"epoch": 0.01,
"learning_rate": 9.999473071219941e-05,
"loss": 2.9186,
"step": 685
},
{
"epoch": 0.01,
"learning_rate": 9.999464025559121e-05,
"loss": 2.9333,
"step": 690
},
{
"epoch": 0.01,
"learning_rate": 9.999454902919478e-05,
"loss": 2.9592,
"step": 695
},
{
"epoch": 0.01,
"learning_rate": 9.999445703301154e-05,
"loss": 2.8528,
"step": 700
},
{
"epoch": 0.01,
"learning_rate": 9.999436426704287e-05,
"loss": 2.9291,
"step": 705
},
{
"epoch": 0.01,
"learning_rate": 9.999427073129026e-05,
"loss": 3.0115,
"step": 710
},
{
"epoch": 0.01,
"learning_rate": 9.999417642575509e-05,
"loss": 2.9579,
"step": 715
},
{
"epoch": 0.01,
"learning_rate": 9.999408135043886e-05,
"loss": 2.988,
"step": 720
},
{
"epoch": 0.01,
"learning_rate": 9.999398550534298e-05,
"loss": 2.8694,
"step": 725
},
{
"epoch": 0.01,
"learning_rate": 9.999388889046899e-05,
"loss": 2.8258,
"step": 730
},
{
"epoch": 0.01,
"learning_rate": 9.999379150581833e-05,
"loss": 2.8642,
"step": 735
},
{
"epoch": 0.01,
"learning_rate": 9.999369335139251e-05,
"loss": 2.8994,
"step": 740
},
{
"epoch": 0.01,
"learning_rate": 9.999359442719306e-05,
"loss": 2.9326,
"step": 745
},
{
"epoch": 0.01,
"learning_rate": 9.999349473322149e-05,
"loss": 3.0304,
"step": 750
},
{
"epoch": 0.01,
"learning_rate": 9.999339426947932e-05,
"loss": 2.9733,
"step": 755
},
{
"epoch": 0.01,
"learning_rate": 9.999329303596812e-05,
"loss": 2.8832,
"step": 760
},
{
"epoch": 0.01,
"learning_rate": 9.999319103268945e-05,
"loss": 2.9704,
"step": 765
},
{
"epoch": 0.01,
"learning_rate": 9.999308825964485e-05,
"loss": 2.9157,
"step": 770
},
{
"epoch": 0.01,
"learning_rate": 9.999298471683593e-05,
"loss": 2.9275,
"step": 775
},
{
"epoch": 0.01,
"learning_rate": 9.999288040426429e-05,
"loss": 2.8408,
"step": 780
},
{
"epoch": 0.01,
"learning_rate": 9.999277532193151e-05,
"loss": 2.8029,
"step": 785
},
{
"epoch": 0.01,
"learning_rate": 9.999266946983923e-05,
"loss": 2.7996,
"step": 790
},
{
"epoch": 0.01,
"learning_rate": 9.999256284798906e-05,
"loss": 2.85,
"step": 795
},
{
"epoch": 0.01,
"learning_rate": 9.999245545638267e-05,
"loss": 2.8564,
"step": 800
},
{
"epoch": 0.01,
"learning_rate": 9.999234729502167e-05,
"loss": 2.9438,
"step": 805
},
{
"epoch": 0.01,
"learning_rate": 9.999223836390777e-05,
"loss": 2.9317,
"step": 810
},
{
"epoch": 0.01,
"learning_rate": 9.999212866304262e-05,
"loss": 2.8375,
"step": 815
},
{
"epoch": 0.01,
"learning_rate": 9.999201819242794e-05,
"loss": 2.8157,
"step": 820
},
{
"epoch": 0.01,
"learning_rate": 9.99919069520654e-05,
"loss": 2.8463,
"step": 825
},
{
"epoch": 0.01,
"learning_rate": 9.99917949419567e-05,
"loss": 2.8026,
"step": 830
},
{
"epoch": 0.01,
"learning_rate": 9.999168216210362e-05,
"loss": 2.7495,
"step": 835
},
{
"epoch": 0.01,
"learning_rate": 9.999156861250784e-05,
"loss": 2.8326,
"step": 840
},
{
"epoch": 0.01,
"learning_rate": 9.999145429317114e-05,
"loss": 2.8583,
"step": 845
},
{
"epoch": 0.01,
"learning_rate": 9.999133920409527e-05,
"loss": 2.8714,
"step": 850
},
{
"epoch": 0.01,
"learning_rate": 9.999122334528199e-05,
"loss": 2.7641,
"step": 855
},
{
"epoch": 0.01,
"learning_rate": 9.999110671673313e-05,
"loss": 2.8282,
"step": 860
},
{
"epoch": 0.01,
"learning_rate": 9.999098931845044e-05,
"loss": 2.9202,
"step": 865
},
{
"epoch": 0.01,
"learning_rate": 9.999087115043573e-05,
"loss": 2.8696,
"step": 870
},
{
"epoch": 0.01,
"learning_rate": 9.999075221269083e-05,
"loss": 2.8167,
"step": 875
},
{
"epoch": 0.01,
"learning_rate": 9.999063250521759e-05,
"loss": 2.8631,
"step": 880
},
{
"epoch": 0.01,
"learning_rate": 9.999051202801781e-05,
"loss": 2.8423,
"step": 885
},
{
"epoch": 0.01,
"learning_rate": 9.999039078109341e-05,
"loss": 2.828,
"step": 890
},
{
"epoch": 0.01,
"learning_rate": 9.999026876444617e-05,
"loss": 2.7923,
"step": 895
},
{
"epoch": 0.01,
"learning_rate": 9.999014597807805e-05,
"loss": 2.808,
"step": 900
},
{
"epoch": 0.01,
"learning_rate": 9.99900224219909e-05,
"loss": 2.8048,
"step": 905
},
{
"epoch": 0.01,
"learning_rate": 9.998989809618662e-05,
"loss": 2.7426,
"step": 910
},
{
"epoch": 0.01,
"learning_rate": 9.998977300066715e-05,
"loss": 2.7983,
"step": 915
},
{
"epoch": 0.01,
"learning_rate": 9.998964713543437e-05,
"loss": 2.8439,
"step": 920
},
{
"epoch": 0.01,
"learning_rate": 9.998952050049026e-05,
"loss": 2.8567,
"step": 925
},
{
"epoch": 0.01,
"learning_rate": 9.998939309583677e-05,
"loss": 2.9089,
"step": 930
},
{
"epoch": 0.01,
"learning_rate": 9.998926492147585e-05,
"loss": 2.8719,
"step": 935
},
{
"epoch": 0.01,
"learning_rate": 9.998913597740945e-05,
"loss": 2.8931,
"step": 940
},
{
"epoch": 0.01,
"learning_rate": 9.99890062636396e-05,
"loss": 2.7969,
"step": 945
},
{
"epoch": 0.01,
"learning_rate": 9.998887578016827e-05,
"loss": 2.7051,
"step": 950
},
{
"epoch": 0.01,
"learning_rate": 9.998874452699747e-05,
"loss": 2.9098,
"step": 955
},
{
"epoch": 0.01,
"learning_rate": 9.998861250412922e-05,
"loss": 2.7896,
"step": 960
},
{
"epoch": 0.01,
"learning_rate": 9.998847971156556e-05,
"loss": 2.8215,
"step": 965
},
{
"epoch": 0.01,
"learning_rate": 9.998834614930855e-05,
"loss": 2.7873,
"step": 970
},
{
"epoch": 0.01,
"learning_rate": 9.99882118173602e-05,
"loss": 2.8684,
"step": 975
},
{
"epoch": 0.01,
"learning_rate": 9.998807671572263e-05,
"loss": 2.7201,
"step": 980
},
{
"epoch": 0.01,
"learning_rate": 9.998794084439789e-05,
"loss": 2.7839,
"step": 985
},
{
"epoch": 0.01,
"learning_rate": 9.998780420338809e-05,
"loss": 2.7237,
"step": 990
},
{
"epoch": 0.01,
"learning_rate": 9.998766679269531e-05,
"loss": 2.8067,
"step": 995
},
{
"epoch": 0.01,
"learning_rate": 9.998752861232168e-05,
"loss": 2.8399,
"step": 1000
},
{
"epoch": 0.01,
"learning_rate": 9.998738966226933e-05,
"loss": 2.8257,
"step": 1005
},
{
"epoch": 0.01,
"learning_rate": 9.998724994254041e-05,
"loss": 2.7325,
"step": 1010
},
{
"epoch": 0.01,
"learning_rate": 9.998710945313705e-05,
"loss": 2.7778,
"step": 1015
},
{
"epoch": 0.01,
"learning_rate": 9.998696819406141e-05,
"loss": 2.837,
"step": 1020
},
{
"epoch": 0.01,
"learning_rate": 9.99868261653157e-05,
"loss": 2.7561,
"step": 1025
},
{
"epoch": 0.01,
"learning_rate": 9.998668336690206e-05,
"loss": 2.7672,
"step": 1030
},
{
"epoch": 0.01,
"learning_rate": 9.998653979882271e-05,
"loss": 2.6946,
"step": 1035
},
{
"epoch": 0.01,
"learning_rate": 9.998639546107989e-05,
"loss": 2.826,
"step": 1040
},
{
"epoch": 0.01,
"learning_rate": 9.998625035367577e-05,
"loss": 2.7059,
"step": 1045
},
{
"epoch": 0.01,
"learning_rate": 9.998610447661261e-05,
"loss": 2.824,
"step": 1050
},
{
"epoch": 0.01,
"learning_rate": 9.998595782989266e-05,
"loss": 2.8651,
"step": 1055
},
{
"epoch": 0.01,
"learning_rate": 9.998581041351818e-05,
"loss": 2.7566,
"step": 1060
},
{
"epoch": 0.01,
"learning_rate": 9.998566222749142e-05,
"loss": 2.7231,
"step": 1065
},
{
"epoch": 0.01,
"learning_rate": 9.998551327181468e-05,
"loss": 2.8682,
"step": 1070
},
{
"epoch": 0.01,
"learning_rate": 9.998536354649026e-05,
"loss": 2.8013,
"step": 1075
},
{
"epoch": 0.01,
"learning_rate": 9.998521305152045e-05,
"loss": 2.7334,
"step": 1080
},
{
"epoch": 0.01,
"learning_rate": 9.998506178690755e-05,
"loss": 2.808,
"step": 1085
},
{
"epoch": 0.01,
"learning_rate": 9.998490975265393e-05,
"loss": 2.7687,
"step": 1090
},
{
"epoch": 0.01,
"learning_rate": 9.99847569487619e-05,
"loss": 2.7807,
"step": 1095
},
{
"epoch": 0.01,
"learning_rate": 9.998460337523383e-05,
"loss": 2.7571,
"step": 1100
},
{
"epoch": 0.01,
"learning_rate": 9.998444903207207e-05,
"loss": 2.8142,
"step": 1105
},
{
"epoch": 0.01,
"learning_rate": 9.998429391927902e-05,
"loss": 2.7817,
"step": 1110
},
{
"epoch": 0.01,
"learning_rate": 9.998413803685705e-05,
"loss": 2.7352,
"step": 1115
},
{
"epoch": 0.01,
"learning_rate": 9.998398138480855e-05,
"loss": 2.7677,
"step": 1120
},
{
"epoch": 0.01,
"learning_rate": 9.998382396313597e-05,
"loss": 2.7399,
"step": 1125
},
{
"epoch": 0.01,
"learning_rate": 9.998366577184168e-05,
"loss": 2.7109,
"step": 1130
},
{
"epoch": 0.01,
"learning_rate": 9.998350681092817e-05,
"loss": 2.7372,
"step": 1135
},
{
"epoch": 0.01,
"learning_rate": 9.998334708039785e-05,
"loss": 2.689,
"step": 1140
},
{
"epoch": 0.01,
"learning_rate": 9.998318658025318e-05,
"loss": 2.7408,
"step": 1145
},
{
"epoch": 0.01,
"learning_rate": 9.998302531049667e-05,
"loss": 2.7858,
"step": 1150
},
{
"epoch": 0.01,
"learning_rate": 9.998286327113076e-05,
"loss": 2.7493,
"step": 1155
},
{
"epoch": 0.01,
"learning_rate": 9.998270046215797e-05,
"loss": 2.7294,
"step": 1160
},
{
"epoch": 0.01,
"learning_rate": 9.998253688358078e-05,
"loss": 2.7713,
"step": 1165
},
{
"epoch": 0.01,
"learning_rate": 9.998237253540174e-05,
"loss": 2.774,
"step": 1170
},
{
"epoch": 0.01,
"learning_rate": 9.998220741762338e-05,
"loss": 2.7874,
"step": 1175
},
{
"epoch": 0.01,
"learning_rate": 9.998204153024821e-05,
"loss": 2.7994,
"step": 1180
},
{
"epoch": 0.01,
"learning_rate": 9.998187487327882e-05,
"loss": 2.7134,
"step": 1185
},
{
"epoch": 0.01,
"learning_rate": 9.998170744671775e-05,
"loss": 2.8666,
"step": 1190
},
{
"epoch": 0.01,
"learning_rate": 9.998153925056758e-05,
"loss": 2.8025,
"step": 1195
},
{
"epoch": 0.01,
"learning_rate": 9.998137028483092e-05,
"loss": 2.7092,
"step": 1200
},
{
"epoch": 0.01,
"learning_rate": 9.998120054951035e-05,
"loss": 2.7364,
"step": 1205
},
{
"epoch": 0.01,
"learning_rate": 9.99810300446085e-05,
"loss": 2.7442,
"step": 1210
},
{
"epoch": 0.01,
"learning_rate": 9.9980858770128e-05,
"loss": 2.7776,
"step": 1215
},
{
"epoch": 0.01,
"learning_rate": 9.998068672607146e-05,
"loss": 2.6824,
"step": 1220
},
{
"epoch": 0.01,
"learning_rate": 9.998051391244155e-05,
"loss": 2.6489,
"step": 1225
},
{
"epoch": 0.01,
"learning_rate": 9.998034032924092e-05,
"loss": 2.727,
"step": 1230
},
{
"epoch": 0.01,
"learning_rate": 9.998016597647225e-05,
"loss": 2.6416,
"step": 1235
},
{
"epoch": 0.01,
"learning_rate": 9.997999085413823e-05,
"loss": 2.7362,
"step": 1240
},
{
"epoch": 0.01,
"learning_rate": 9.997981496224153e-05,
"loss": 2.796,
"step": 1245
},
{
"epoch": 0.01,
"learning_rate": 9.99796383007849e-05,
"loss": 2.8211,
"step": 1250
},
{
"epoch": 0.01,
"learning_rate": 9.997946086977103e-05,
"loss": 2.735,
"step": 1255
},
{
"epoch": 0.01,
"learning_rate": 9.997928266920266e-05,
"loss": 2.8364,
"step": 1260
},
{
"epoch": 0.01,
"learning_rate": 9.997910369908254e-05,
"loss": 2.7425,
"step": 1265
},
{
"epoch": 0.01,
"learning_rate": 9.99789239594134e-05,
"loss": 2.8218,
"step": 1270
},
{
"epoch": 0.01,
"learning_rate": 9.997874345019805e-05,
"loss": 2.7977,
"step": 1275
},
{
"epoch": 0.01,
"learning_rate": 9.997856217143921e-05,
"loss": 2.7087,
"step": 1280
},
{
"epoch": 0.01,
"learning_rate": 9.997838012313974e-05,
"loss": 2.7913,
"step": 1285
},
{
"epoch": 0.01,
"learning_rate": 9.99781973053024e-05,
"loss": 2.8349,
"step": 1290
},
{
"epoch": 0.01,
"learning_rate": 9.997801371793001e-05,
"loss": 2.6015,
"step": 1295
},
{
"epoch": 0.01,
"learning_rate": 9.997782936102542e-05,
"loss": 2.7237,
"step": 1300
},
{
"epoch": 0.01,
"learning_rate": 9.997764423459144e-05,
"loss": 2.7821,
"step": 1305
},
{
"epoch": 0.01,
"learning_rate": 9.997745833863092e-05,
"loss": 2.7198,
"step": 1310
},
{
"epoch": 0.01,
"learning_rate": 9.997727167314676e-05,
"loss": 2.7406,
"step": 1315
},
{
"epoch": 0.01,
"learning_rate": 9.997708423814178e-05,
"loss": 2.6309,
"step": 1320
},
{
"epoch": 0.01,
"learning_rate": 9.99768960336189e-05,
"loss": 2.7655,
"step": 1325
},
{
"epoch": 0.01,
"learning_rate": 9.997670705958102e-05,
"loss": 2.7317,
"step": 1330
},
{
"epoch": 0.01,
"learning_rate": 9.997651731603104e-05,
"loss": 2.6619,
"step": 1335
},
{
"epoch": 0.01,
"learning_rate": 9.997632680297188e-05,
"loss": 2.6723,
"step": 1340
},
{
"epoch": 0.01,
"learning_rate": 9.997613552040648e-05,
"loss": 2.7652,
"step": 1345
},
{
"epoch": 0.01,
"learning_rate": 9.997594346833779e-05,
"loss": 2.7044,
"step": 1350
},
{
"epoch": 0.01,
"learning_rate": 9.997575064676875e-05,
"loss": 2.7218,
"step": 1355
},
{
"epoch": 0.01,
"learning_rate": 9.997555705570232e-05,
"loss": 2.7637,
"step": 1360
},
{
"epoch": 0.01,
"learning_rate": 9.997536269514152e-05,
"loss": 2.7444,
"step": 1365
},
{
"epoch": 0.01,
"learning_rate": 9.99751675650893e-05,
"loss": 2.7691,
"step": 1370
},
{
"epoch": 0.01,
"learning_rate": 9.99749716655487e-05,
"loss": 2.7685,
"step": 1375
},
{
"epoch": 0.01,
"learning_rate": 9.997477499652273e-05,
"loss": 2.7795,
"step": 1380
},
{
"epoch": 0.01,
"learning_rate": 9.99745775580144e-05,
"loss": 2.7228,
"step": 1385
},
{
"epoch": 0.01,
"learning_rate": 9.997437935002676e-05,
"loss": 2.8142,
"step": 1390
},
{
"epoch": 0.01,
"learning_rate": 9.997418037256285e-05,
"loss": 2.6732,
"step": 1395
},
{
"epoch": 0.01,
"learning_rate": 9.997398062562575e-05,
"loss": 2.7128,
"step": 1400
},
{
"epoch": 0.01,
"learning_rate": 9.997378010921856e-05,
"loss": 2.7015,
"step": 1405
},
{
"epoch": 0.01,
"learning_rate": 9.99735788233443e-05,
"loss": 2.7631,
"step": 1410
},
{
"epoch": 0.01,
"learning_rate": 9.997337676800613e-05,
"loss": 2.8036,
"step": 1415
},
{
"epoch": 0.01,
"learning_rate": 9.997317394320713e-05,
"loss": 2.7363,
"step": 1420
},
{
"epoch": 0.01,
"learning_rate": 9.997297034895044e-05,
"loss": 2.6103,
"step": 1425
},
{
"epoch": 0.01,
"learning_rate": 9.99727659852392e-05,
"loss": 2.6435,
"step": 1430
},
{
"epoch": 0.01,
"learning_rate": 9.997256085207652e-05,
"loss": 2.7487,
"step": 1435
},
{
"epoch": 0.01,
"learning_rate": 9.997235494946557e-05,
"loss": 2.6269,
"step": 1440
},
{
"epoch": 0.01,
"learning_rate": 9.997214827740955e-05,
"loss": 2.7691,
"step": 1445
},
{
"epoch": 0.01,
"learning_rate": 9.997194083591164e-05,
"loss": 2.7558,
"step": 1450
},
{
"epoch": 0.01,
"learning_rate": 9.9971732624975e-05,
"loss": 2.708,
"step": 1455
},
{
"epoch": 0.01,
"learning_rate": 9.997152364460287e-05,
"loss": 2.6844,
"step": 1460
},
{
"epoch": 0.01,
"learning_rate": 9.997131389479844e-05,
"loss": 2.7028,
"step": 1465
},
{
"epoch": 0.01,
"learning_rate": 9.997110337556494e-05,
"loss": 2.7616,
"step": 1470
},
{
"epoch": 0.01,
"learning_rate": 9.997089208690564e-05,
"loss": 2.623,
"step": 1475
},
{
"epoch": 0.01,
"learning_rate": 9.997068002882377e-05,
"loss": 2.6733,
"step": 1480
},
{
"epoch": 0.01,
"learning_rate": 9.997046720132261e-05,
"loss": 2.7081,
"step": 1485
},
{
"epoch": 0.01,
"learning_rate": 9.997025360440541e-05,
"loss": 2.6995,
"step": 1490
},
{
"epoch": 0.01,
"learning_rate": 9.997003923807551e-05,
"loss": 2.7587,
"step": 1495
},
{
"epoch": 0.01,
"learning_rate": 9.996982410233615e-05,
"loss": 2.7226,
"step": 1500
},
{
"epoch": 0.01,
"learning_rate": 9.996960819719068e-05,
"loss": 2.7019,
"step": 1505
},
{
"epoch": 0.01,
"learning_rate": 9.996939152264242e-05,
"loss": 2.6982,
"step": 1510
},
{
"epoch": 0.01,
"learning_rate": 9.99691740786947e-05,
"loss": 2.7043,
"step": 1515
},
{
"epoch": 0.01,
"learning_rate": 9.996895586535086e-05,
"loss": 2.6791,
"step": 1520
},
{
"epoch": 0.01,
"learning_rate": 9.996873688261428e-05,
"loss": 2.565,
"step": 1525
},
{
"epoch": 0.01,
"learning_rate": 9.996851713048831e-05,
"loss": 2.6534,
"step": 1530
},
{
"epoch": 0.01,
"learning_rate": 9.996829660897637e-05,
"loss": 2.7249,
"step": 1535
},
{
"epoch": 0.01,
"learning_rate": 9.996807531808181e-05,
"loss": 2.7101,
"step": 1540
},
{
"epoch": 0.01,
"learning_rate": 9.996785325780806e-05,
"loss": 2.6748,
"step": 1545
},
{
"epoch": 0.01,
"learning_rate": 9.996763042815853e-05,
"loss": 2.7194,
"step": 1550
},
{
"epoch": 0.01,
"learning_rate": 9.996740682913668e-05,
"loss": 2.7557,
"step": 1555
},
{
"epoch": 0.01,
"learning_rate": 9.996718246074591e-05,
"loss": 2.7593,
"step": 1560
},
{
"epoch": 0.01,
"learning_rate": 9.99669573229897e-05,
"loss": 2.662,
"step": 1565
},
{
"epoch": 0.01,
"learning_rate": 9.996673141587151e-05,
"loss": 2.6794,
"step": 1570
},
{
"epoch": 0.01,
"learning_rate": 9.996650473939482e-05,
"loss": 2.5802,
"step": 1575
},
{
"epoch": 0.01,
"learning_rate": 9.996627729356314e-05,
"loss": 2.6492,
"step": 1580
},
{
"epoch": 0.01,
"learning_rate": 9.996604907837993e-05,
"loss": 2.7238,
"step": 1585
},
{
"epoch": 0.01,
"learning_rate": 9.996582009384874e-05,
"loss": 2.6468,
"step": 1590
},
{
"epoch": 0.01,
"learning_rate": 9.996559033997307e-05,
"loss": 2.6055,
"step": 1595
},
{
"epoch": 0.01,
"learning_rate": 9.996535981675648e-05,
"loss": 2.6955,
"step": 1600
},
{
"epoch": 0.01,
"learning_rate": 9.996512852420251e-05,
"loss": 2.6257,
"step": 1605
},
{
"epoch": 0.01,
"learning_rate": 9.996489646231471e-05,
"loss": 2.6435,
"step": 1610
},
{
"epoch": 0.01,
"learning_rate": 9.996466363109667e-05,
"loss": 2.6242,
"step": 1615
},
{
"epoch": 0.01,
"learning_rate": 9.996443003055196e-05,
"loss": 2.6643,
"step": 1620
},
{
"epoch": 0.01,
"learning_rate": 9.99641956606842e-05,
"loss": 2.6947,
"step": 1625
},
{
"epoch": 0.01,
"learning_rate": 9.996396052149698e-05,
"loss": 2.6988,
"step": 1630
},
{
"epoch": 0.01,
"learning_rate": 9.996372461299392e-05,
"loss": 2.6383,
"step": 1635
},
{
"epoch": 0.01,
"learning_rate": 9.996348793517866e-05,
"loss": 2.7534,
"step": 1640
},
{
"epoch": 0.01,
"learning_rate": 9.996325048805484e-05,
"loss": 2.6885,
"step": 1645
},
{
"epoch": 0.01,
"learning_rate": 9.996301227162612e-05,
"loss": 2.6516,
"step": 1650
},
{
"epoch": 0.01,
"learning_rate": 9.996277328589618e-05,
"loss": 2.7059,
"step": 1655
},
{
"epoch": 0.01,
"learning_rate": 9.996253353086866e-05,
"loss": 2.6935,
"step": 1660
},
{
"epoch": 0.01,
"learning_rate": 9.996229300654729e-05,
"loss": 2.5786,
"step": 1665
},
{
"epoch": 0.01,
"learning_rate": 9.996205171293576e-05,
"loss": 2.6647,
"step": 1670
},
{
"epoch": 0.01,
"learning_rate": 9.996180965003778e-05,
"loss": 2.7188,
"step": 1675
},
{
"epoch": 0.01,
"learning_rate": 9.996156681785708e-05,
"loss": 2.6856,
"step": 1680
},
{
"epoch": 0.01,
"learning_rate": 9.996132321639741e-05,
"loss": 2.6024,
"step": 1685
},
{
"epoch": 0.01,
"learning_rate": 9.996107884566252e-05,
"loss": 2.5822,
"step": 1690
},
{
"epoch": 0.01,
"learning_rate": 9.996083370565616e-05,
"loss": 2.652,
"step": 1695
},
{
"epoch": 0.01,
"learning_rate": 9.99605877963821e-05,
"loss": 2.5987,
"step": 1700
},
{
"epoch": 0.01,
"learning_rate": 9.996034111784413e-05,
"loss": 2.6432,
"step": 1705
},
{
"epoch": 0.01,
"learning_rate": 9.996009367004607e-05,
"loss": 2.7128,
"step": 1710
},
{
"epoch": 0.01,
"learning_rate": 9.995984545299171e-05,
"loss": 2.6575,
"step": 1715
},
{
"epoch": 0.01,
"learning_rate": 9.995959646668488e-05,
"loss": 2.6654,
"step": 1720
},
{
"epoch": 0.01,
"learning_rate": 9.995934671112941e-05,
"loss": 2.6713,
"step": 1725
},
{
"epoch": 0.01,
"learning_rate": 9.995909618632915e-05,
"loss": 2.7157,
"step": 1730
},
{
"epoch": 0.01,
"learning_rate": 9.995884489228794e-05,
"loss": 2.6613,
"step": 1735
},
{
"epoch": 0.01,
"learning_rate": 9.995859282900967e-05,
"loss": 2.6212,
"step": 1740
},
{
"epoch": 0.01,
"learning_rate": 9.995833999649821e-05,
"loss": 2.6473,
"step": 1745
},
{
"epoch": 0.01,
"learning_rate": 9.995808639475746e-05,
"loss": 2.6193,
"step": 1750
},
{
"epoch": 0.01,
"learning_rate": 9.995783202379131e-05,
"loss": 2.5961,
"step": 1755
},
{
"epoch": 0.01,
"learning_rate": 9.99575768836037e-05,
"loss": 2.6321,
"step": 1760
},
{
"epoch": 0.01,
"learning_rate": 9.995732097419855e-05,
"loss": 2.7201,
"step": 1765
},
{
"epoch": 0.01,
"learning_rate": 9.995706429557978e-05,
"loss": 2.6548,
"step": 1770
},
{
"epoch": 0.01,
"learning_rate": 9.995680684775137e-05,
"loss": 2.5922,
"step": 1775
},
{
"epoch": 0.01,
"learning_rate": 9.995654863071729e-05,
"loss": 2.6131,
"step": 1780
},
{
"epoch": 0.01,
"learning_rate": 9.995628964448147e-05,
"loss": 2.6951,
"step": 1785
},
{
"epoch": 0.01,
"learning_rate": 9.995602988904793e-05,
"loss": 2.6755,
"step": 1790
},
{
"epoch": 0.01,
"learning_rate": 9.995576936442068e-05,
"loss": 2.7042,
"step": 1795
},
{
"epoch": 0.01,
"learning_rate": 9.995550807060371e-05,
"loss": 2.6368,
"step": 1800
},
{
"epoch": 0.01,
"learning_rate": 9.995524600760105e-05,
"loss": 2.5698,
"step": 1805
},
{
"epoch": 0.01,
"learning_rate": 9.995498317541674e-05,
"loss": 2.6368,
"step": 1810
},
{
"epoch": 0.01,
"learning_rate": 9.995471957405483e-05,
"loss": 2.5622,
"step": 1815
},
{
"epoch": 0.01,
"learning_rate": 9.995445520351935e-05,
"loss": 2.6346,
"step": 1820
},
{
"epoch": 0.01,
"learning_rate": 9.995419006381442e-05,
"loss": 2.6094,
"step": 1825
},
{
"epoch": 0.01,
"learning_rate": 9.995392415494407e-05,
"loss": 2.665,
"step": 1830
},
{
"epoch": 0.01,
"learning_rate": 9.995365747691243e-05,
"loss": 2.7206,
"step": 1835
},
{
"epoch": 0.01,
"learning_rate": 9.995339002972359e-05,
"loss": 2.7288,
"step": 1840
},
{
"epoch": 0.01,
"learning_rate": 9.995312181338168e-05,
"loss": 2.6274,
"step": 1845
},
{
"epoch": 0.01,
"learning_rate": 9.995285282789082e-05,
"loss": 2.6638,
"step": 1850
},
{
"epoch": 0.01,
"learning_rate": 9.995258307325515e-05,
"loss": 2.6872,
"step": 1855
},
{
"epoch": 0.01,
"learning_rate": 9.995231254947883e-05,
"loss": 2.6438,
"step": 1860
},
{
"epoch": 0.01,
"learning_rate": 9.995204125656603e-05,
"loss": 2.6049,
"step": 1865
},
{
"epoch": 0.01,
"learning_rate": 9.995176919452092e-05,
"loss": 2.6126,
"step": 1870
},
{
"epoch": 0.01,
"learning_rate": 9.995149636334769e-05,
"loss": 2.6059,
"step": 1875
},
{
"epoch": 0.01,
"learning_rate": 9.995122276305054e-05,
"loss": 2.5121,
"step": 1880
},
{
"epoch": 0.01,
"learning_rate": 9.995094839363368e-05,
"loss": 2.6934,
"step": 1885
},
{
"epoch": 0.01,
"learning_rate": 9.995067325510132e-05,
"loss": 2.5923,
"step": 1890
},
{
"epoch": 0.01,
"learning_rate": 9.995039734745774e-05,
"loss": 2.6695,
"step": 1895
},
{
"epoch": 0.01,
"learning_rate": 9.995012067070715e-05,
"loss": 2.6797,
"step": 1900
},
{
"epoch": 0.02,
"learning_rate": 9.994984322485383e-05,
"loss": 2.5872,
"step": 1905
},
{
"epoch": 0.02,
"learning_rate": 9.994956500990205e-05,
"loss": 2.5689,
"step": 1910
},
{
"epoch": 0.02,
"learning_rate": 9.994928602585608e-05,
"loss": 2.6608,
"step": 1915
},
{
"epoch": 0.02,
"learning_rate": 9.994900627272023e-05,
"loss": 2.5504,
"step": 1920
},
{
"epoch": 0.02,
"learning_rate": 9.99487257504988e-05,
"loss": 2.6144,
"step": 1925
},
{
"epoch": 0.02,
"learning_rate": 9.994844445919612e-05,
"loss": 2.6,
"step": 1930
},
{
"epoch": 0.02,
"learning_rate": 9.994816239881648e-05,
"loss": 2.6435,
"step": 1935
},
{
"epoch": 0.02,
"learning_rate": 9.99478795693643e-05,
"loss": 2.6704,
"step": 1940
},
{
"epoch": 0.02,
"learning_rate": 9.994759597084384e-05,
"loss": 2.623,
"step": 1945
},
{
"epoch": 0.02,
"learning_rate": 9.994731160325955e-05,
"loss": 2.6469,
"step": 1950
},
{
"epoch": 0.02,
"learning_rate": 9.994702646661577e-05,
"loss": 2.685,
"step": 1955
},
{
"epoch": 0.02,
"learning_rate": 9.994674056091689e-05,
"loss": 2.6562,
"step": 1960
},
{
"epoch": 0.02,
"learning_rate": 9.994645388616732e-05,
"loss": 2.6218,
"step": 1965
},
{
"epoch": 0.02,
"learning_rate": 9.994616644237148e-05,
"loss": 2.4988,
"step": 1970
},
{
"epoch": 0.02,
"learning_rate": 9.994587822953378e-05,
"loss": 2.5552,
"step": 1975
},
{
"epoch": 0.02,
"learning_rate": 9.994558924765866e-05,
"loss": 2.6182,
"step": 1980
},
{
"epoch": 0.02,
"learning_rate": 9.99452994967506e-05,
"loss": 2.6767,
"step": 1985
},
{
"epoch": 0.02,
"learning_rate": 9.994500897681401e-05,
"loss": 2.7063,
"step": 1990
},
{
"epoch": 0.02,
"learning_rate": 9.994471768785339e-05,
"loss": 2.6532,
"step": 1995
},
{
"epoch": 0.02,
"learning_rate": 9.994442562987324e-05,
"loss": 2.6418,
"step": 2000
},
{
"epoch": 0.02,
"learning_rate": 9.994413280287803e-05,
"loss": 2.5203,
"step": 2005
},
{
"epoch": 0.02,
"learning_rate": 9.994383920687229e-05,
"loss": 2.5248,
"step": 2010
},
{
"epoch": 0.02,
"learning_rate": 9.994354484186053e-05,
"loss": 2.6817,
"step": 2015
},
{
"epoch": 0.02,
"learning_rate": 9.994324970784727e-05,
"loss": 2.636,
"step": 2020
},
{
"epoch": 0.02,
"learning_rate": 9.994295380483709e-05,
"loss": 2.6579,
"step": 2025
},
{
"epoch": 0.02,
"learning_rate": 9.99426571328345e-05,
"loss": 2.6527,
"step": 2030
},
{
"epoch": 0.02,
"learning_rate": 9.994235969184411e-05,
"loss": 2.5002,
"step": 2035
},
{
"epoch": 0.02,
"learning_rate": 9.994206148187046e-05,
"loss": 2.5534,
"step": 2040
},
{
"epoch": 0.02,
"learning_rate": 9.994176250291818e-05,
"loss": 2.5677,
"step": 2045
},
{
"epoch": 0.02,
"learning_rate": 9.994146275499186e-05,
"loss": 2.654,
"step": 2050
},
{
"epoch": 0.02,
"learning_rate": 9.994116223809611e-05,
"loss": 2.5609,
"step": 2055
},
{
"epoch": 0.02,
"learning_rate": 9.994086095223557e-05,
"loss": 2.615,
"step": 2060
},
{
"epoch": 0.02,
"learning_rate": 9.994055889741485e-05,
"loss": 2.5987,
"step": 2065
},
{
"epoch": 0.02,
"learning_rate": 9.994025607363862e-05,
"loss": 2.4983,
"step": 2070
},
{
"epoch": 0.02,
"learning_rate": 9.993995248091155e-05,
"loss": 2.6197,
"step": 2075
},
{
"epoch": 0.02,
"learning_rate": 9.993964811923831e-05,
"loss": 2.6273,
"step": 2080
},
{
"epoch": 0.02,
"learning_rate": 9.993934298862358e-05,
"loss": 2.5427,
"step": 2085
},
{
"epoch": 0.02,
"learning_rate": 9.993903708907205e-05,
"loss": 2.5767,
"step": 2090
},
{
"epoch": 0.02,
"learning_rate": 9.993873042058845e-05,
"loss": 2.6893,
"step": 2095
},
{
"epoch": 0.02,
"learning_rate": 9.993842298317749e-05,
"loss": 2.566,
"step": 2100
},
{
"epoch": 0.02,
"learning_rate": 9.993811477684392e-05,
"loss": 2.5451,
"step": 2105
},
{
"epoch": 0.02,
"learning_rate": 9.993780580159246e-05,
"loss": 2.5609,
"step": 2110
},
{
"epoch": 0.02,
"learning_rate": 9.993749605742788e-05,
"loss": 2.6141,
"step": 2115
},
{
"epoch": 0.02,
"learning_rate": 9.993718554435495e-05,
"loss": 2.5711,
"step": 2120
},
{
"epoch": 0.02,
"learning_rate": 9.993687426237846e-05,
"loss": 2.5801,
"step": 2125
},
{
"epoch": 0.02,
"learning_rate": 9.993656221150318e-05,
"loss": 2.659,
"step": 2130
},
{
"epoch": 0.02,
"learning_rate": 9.993624939173394e-05,
"loss": 2.5476,
"step": 2135
},
{
"epoch": 0.02,
"learning_rate": 9.993593580307553e-05,
"loss": 2.5786,
"step": 2140
},
{
"epoch": 0.02,
"learning_rate": 9.99356214455328e-05,
"loss": 2.5907,
"step": 2145
},
{
"epoch": 0.02,
"learning_rate": 9.993530631911059e-05,
"loss": 2.6088,
"step": 2150
},
{
"epoch": 0.02,
"learning_rate": 9.993499042381373e-05,
"loss": 2.6549,
"step": 2155
},
{
"epoch": 0.02,
"learning_rate": 9.993467375964711e-05,
"loss": 2.5518,
"step": 2160
},
{
"epoch": 0.02,
"learning_rate": 9.99343563266156e-05,
"loss": 2.6655,
"step": 2165
},
{
"epoch": 0.02,
"learning_rate": 9.993403812472408e-05,
"loss": 2.5188,
"step": 2170
},
{
"epoch": 0.02,
"learning_rate": 9.993371915397745e-05,
"loss": 2.5168,
"step": 2175
},
{
"epoch": 0.02,
"learning_rate": 9.993339941438062e-05,
"loss": 2.5417,
"step": 2180
},
{
"epoch": 0.02,
"learning_rate": 9.993307890593853e-05,
"loss": 2.5925,
"step": 2185
},
{
"epoch": 0.02,
"learning_rate": 9.993275762865609e-05,
"loss": 2.5889,
"step": 2190
},
{
"epoch": 0.02,
"learning_rate": 9.993243558253826e-05,
"loss": 2.6004,
"step": 2195
},
{
"epoch": 0.02,
"learning_rate": 9.993211276759e-05,
"loss": 2.4859,
"step": 2200
},
{
"epoch": 0.02,
"learning_rate": 9.993178918381628e-05,
"loss": 2.5298,
"step": 2205
},
{
"epoch": 0.02,
"learning_rate": 9.993146483122207e-05,
"loss": 2.6023,
"step": 2210
},
{
"epoch": 0.02,
"learning_rate": 9.993113970981238e-05,
"loss": 2.5757,
"step": 2215
},
{
"epoch": 0.02,
"learning_rate": 9.993081381959221e-05,
"loss": 2.4852,
"step": 2220
},
{
"epoch": 0.02,
"learning_rate": 9.99304871605666e-05,
"loss": 2.6316,
"step": 2225
},
{
"epoch": 0.02,
"learning_rate": 9.993015973274053e-05,
"loss": 2.6782,
"step": 2230
},
{
"epoch": 0.02,
"learning_rate": 9.992983153611908e-05,
"loss": 2.5189,
"step": 2235
},
{
"epoch": 0.02,
"learning_rate": 9.99295025707073e-05,
"loss": 2.5596,
"step": 2240
},
{
"epoch": 0.02,
"learning_rate": 9.992917283651024e-05,
"loss": 2.5624,
"step": 2245
},
{
"epoch": 0.02,
"learning_rate": 9.992884233353298e-05,
"loss": 2.6578,
"step": 2250
},
{
"epoch": 0.02,
"learning_rate": 9.992851106178063e-05,
"loss": 2.4975,
"step": 2255
},
{
"epoch": 0.02,
"learning_rate": 9.992817902125826e-05,
"loss": 2.6095,
"step": 2260
},
{
"epoch": 0.02,
"learning_rate": 9.9927846211971e-05,
"loss": 2.577,
"step": 2265
},
{
"epoch": 0.02,
"learning_rate": 9.992751263392397e-05,
"loss": 2.5584,
"step": 2270
},
{
"epoch": 0.02,
"learning_rate": 9.992717828712232e-05,
"loss": 2.599,
"step": 2275
},
{
"epoch": 0.02,
"learning_rate": 9.992684317157117e-05,
"loss": 2.611,
"step": 2280
},
{
"epoch": 0.02,
"learning_rate": 9.992650728727571e-05,
"loss": 2.5793,
"step": 2285
},
{
"epoch": 0.02,
"learning_rate": 9.99261706342411e-05,
"loss": 2.5897,
"step": 2290
},
{
"epoch": 0.02,
"learning_rate": 9.99258332124725e-05,
"loss": 2.5731,
"step": 2295
},
{
"epoch": 0.02,
"learning_rate": 9.992549502197514e-05,
"loss": 2.557,
"step": 2300
},
{
"epoch": 0.02,
"learning_rate": 9.992515606275423e-05,
"loss": 2.6484,
"step": 2305
},
{
"epoch": 0.02,
"learning_rate": 9.992481633481495e-05,
"loss": 2.6383,
"step": 2310
},
{
"epoch": 0.02,
"learning_rate": 9.992447583816257e-05,
"loss": 2.5671,
"step": 2315
},
{
"epoch": 0.02,
"learning_rate": 9.992413457280231e-05,
"loss": 2.5776,
"step": 2320
},
{
"epoch": 0.02,
"learning_rate": 9.992379253873943e-05,
"loss": 2.6134,
"step": 2325
},
{
"epoch": 0.02,
"learning_rate": 9.99234497359792e-05,
"loss": 2.6245,
"step": 2330
},
{
"epoch": 0.02,
"learning_rate": 9.992310616452689e-05,
"loss": 2.5677,
"step": 2335
},
{
"epoch": 0.02,
"learning_rate": 9.99227618243878e-05,
"loss": 2.6514,
"step": 2340
},
{
"epoch": 0.02,
"learning_rate": 9.992241671556724e-05,
"loss": 2.5463,
"step": 2345
},
{
"epoch": 0.02,
"learning_rate": 9.99220708380705e-05,
"loss": 2.6168,
"step": 2350
},
{
"epoch": 0.02,
"learning_rate": 9.99217241919029e-05,
"loss": 2.5815,
"step": 2355
},
{
"epoch": 0.02,
"learning_rate": 9.992137677706982e-05,
"loss": 2.4965,
"step": 2360
},
{
"epoch": 0.02,
"learning_rate": 9.99210285935766e-05,
"loss": 2.621,
"step": 2365
},
{
"epoch": 0.02,
"learning_rate": 9.992067964142856e-05,
"loss": 2.6669,
"step": 2370
},
{
"epoch": 0.02,
"learning_rate": 9.992032992063109e-05,
"loss": 2.5371,
"step": 2375
},
{
"epoch": 0.02,
"learning_rate": 9.99199794311896e-05,
"loss": 2.5767,
"step": 2380
},
{
"epoch": 0.02,
"learning_rate": 9.991962817310947e-05,
"loss": 2.5748,
"step": 2385
},
{
"epoch": 0.02,
"learning_rate": 9.991927614639612e-05,
"loss": 2.5674,
"step": 2390
},
{
"epoch": 0.02,
"learning_rate": 9.991892335105493e-05,
"loss": 2.5685,
"step": 2395
},
{
"epoch": 0.02,
"learning_rate": 9.991856978709137e-05,
"loss": 2.5905,
"step": 2400
},
{
"epoch": 0.02,
"learning_rate": 9.99182154545109e-05,
"loss": 2.5204,
"step": 2405
},
{
"epoch": 0.02,
"learning_rate": 9.991786035331893e-05,
"loss": 2.5912,
"step": 2410
},
{
"epoch": 0.02,
"learning_rate": 9.991750448352096e-05,
"loss": 2.5464,
"step": 2415
},
{
"epoch": 0.02,
"learning_rate": 9.991714784512243e-05,
"loss": 2.5941,
"step": 2420
},
{
"epoch": 0.02,
"learning_rate": 9.991679043812887e-05,
"loss": 2.5966,
"step": 2425
},
{
"epoch": 0.02,
"learning_rate": 9.99164322625458e-05,
"loss": 2.6034,
"step": 2430
},
{
"epoch": 0.02,
"learning_rate": 9.991607331837867e-05,
"loss": 2.5438,
"step": 2435
},
{
"epoch": 0.02,
"learning_rate": 9.991571360563306e-05,
"loss": 2.4839,
"step": 2440
},
{
"epoch": 0.02,
"learning_rate": 9.99153531243145e-05,
"loss": 2.6219,
"step": 2445
},
{
"epoch": 0.02,
"learning_rate": 9.991499187442852e-05,
"loss": 2.6151,
"step": 2450
},
{
"epoch": 0.02,
"learning_rate": 9.99146298559807e-05,
"loss": 2.5115,
"step": 2455
},
{
"epoch": 0.02,
"learning_rate": 9.991426706897661e-05,
"loss": 2.5843,
"step": 2460
},
{
"epoch": 0.02,
"learning_rate": 9.991390351342185e-05,
"loss": 2.4691,
"step": 2465
},
{
"epoch": 0.02,
"learning_rate": 9.991353918932198e-05,
"loss": 2.6279,
"step": 2470
},
{
"epoch": 0.02,
"learning_rate": 9.991317409668264e-05,
"loss": 2.6623,
"step": 2475
},
{
"epoch": 0.02,
"learning_rate": 9.991280823550945e-05,
"loss": 2.5449,
"step": 2480
},
{
"epoch": 0.02,
"learning_rate": 9.991244160580803e-05,
"loss": 2.5897,
"step": 2485
},
{
"epoch": 0.02,
"learning_rate": 9.991207420758404e-05,
"loss": 2.624,
"step": 2490
},
{
"epoch": 0.02,
"learning_rate": 9.991170604084314e-05,
"loss": 2.6293,
"step": 2495
},
{
"epoch": 0.02,
"learning_rate": 9.991133710559097e-05,
"loss": 2.504,
"step": 2500
},
{
"epoch": 0.02,
"learning_rate": 9.991096740183324e-05,
"loss": 2.5858,
"step": 2505
},
{
"epoch": 0.02,
"learning_rate": 9.991059692957563e-05,
"loss": 2.5528,
"step": 2510
},
{
"epoch": 0.02,
"learning_rate": 9.991022568882384e-05,
"loss": 2.5473,
"step": 2515
},
{
"epoch": 0.02,
"learning_rate": 9.99098536795836e-05,
"loss": 2.5127,
"step": 2520
},
{
"epoch": 0.02,
"learning_rate": 9.990948090186063e-05,
"loss": 2.5326,
"step": 2525
},
{
"epoch": 0.02,
"learning_rate": 9.990910735566065e-05,
"loss": 2.6132,
"step": 2530
},
{
"epoch": 0.02,
"learning_rate": 9.990873304098946e-05,
"loss": 2.4558,
"step": 2535
},
{
"epoch": 0.02,
"learning_rate": 9.990835795785278e-05,
"loss": 2.5565,
"step": 2540
},
{
"epoch": 0.02,
"learning_rate": 9.990798210625642e-05,
"loss": 2.5634,
"step": 2545
},
{
"epoch": 0.02,
"learning_rate": 9.990760548620613e-05,
"loss": 2.4665,
"step": 2550
},
{
"epoch": 0.02,
"learning_rate": 9.990722809770773e-05,
"loss": 2.5741,
"step": 2555
},
{
"epoch": 0.02,
"learning_rate": 9.990684994076702e-05,
"loss": 2.5126,
"step": 2560
},
{
"epoch": 0.02,
"learning_rate": 9.990647101538983e-05,
"loss": 2.6132,
"step": 2565
},
{
"epoch": 0.02,
"learning_rate": 9.990609132158201e-05,
"loss": 2.6722,
"step": 2570
},
{
"epoch": 0.02,
"learning_rate": 9.990571085934938e-05,
"loss": 2.5621,
"step": 2575
},
{
"epoch": 0.02,
"learning_rate": 9.990532962869783e-05,
"loss": 2.6266,
"step": 2580
},
{
"epoch": 0.02,
"learning_rate": 9.990494762963318e-05,
"loss": 2.5936,
"step": 2585
},
{
"epoch": 0.02,
"learning_rate": 9.990456486216136e-05,
"loss": 2.6362,
"step": 2590
},
{
"epoch": 0.02,
"learning_rate": 9.990418132628824e-05,
"loss": 2.5906,
"step": 2595
},
{
"epoch": 0.02,
"learning_rate": 9.990379702201974e-05,
"loss": 2.5915,
"step": 2600
},
{
"epoch": 0.02,
"learning_rate": 9.990341194936173e-05,
"loss": 2.6048,
"step": 2605
},
{
"epoch": 0.02,
"learning_rate": 9.990302610832021e-05,
"loss": 2.6206,
"step": 2610
},
{
"epoch": 0.02,
"learning_rate": 9.990263949890108e-05,
"loss": 2.5932,
"step": 2615
},
{
"epoch": 0.02,
"learning_rate": 9.990225212111029e-05,
"loss": 2.6049,
"step": 2620
},
{
"epoch": 0.02,
"learning_rate": 9.990186397495383e-05,
"loss": 2.5169,
"step": 2625
},
{
"epoch": 0.02,
"learning_rate": 9.990147506043766e-05,
"loss": 2.6169,
"step": 2630
},
{
"epoch": 0.02,
"learning_rate": 9.990108537756775e-05,
"loss": 2.579,
"step": 2635
},
{
"epoch": 0.02,
"learning_rate": 9.990069492635015e-05,
"loss": 2.4828,
"step": 2640
},
{
"epoch": 0.02,
"learning_rate": 9.990030370679082e-05,
"loss": 2.5224,
"step": 2645
},
{
"epoch": 0.02,
"learning_rate": 9.989991171889581e-05,
"loss": 2.5835,
"step": 2650
},
{
"epoch": 0.02,
"learning_rate": 9.989951896267115e-05,
"loss": 2.5358,
"step": 2655
},
{
"epoch": 0.02,
"learning_rate": 9.989912543812289e-05,
"loss": 2.5426,
"step": 2660
},
{
"epoch": 0.02,
"learning_rate": 9.989873114525709e-05,
"loss": 2.5323,
"step": 2665
},
{
"epoch": 0.02,
"learning_rate": 9.989833608407981e-05,
"loss": 2.5252,
"step": 2670
},
{
"epoch": 0.02,
"learning_rate": 9.989794025459715e-05,
"loss": 2.5383,
"step": 2675
},
{
"epoch": 0.02,
"learning_rate": 9.989754365681519e-05,
"loss": 2.5719,
"step": 2680
},
{
"epoch": 0.02,
"learning_rate": 9.989714629074006e-05,
"loss": 2.5591,
"step": 2685
},
{
"epoch": 0.02,
"learning_rate": 9.989674815637784e-05,
"loss": 2.5106,
"step": 2690
},
{
"epoch": 0.02,
"learning_rate": 9.989634925373468e-05,
"loss": 2.6532,
"step": 2695
},
{
"epoch": 0.02,
"learning_rate": 9.989594958281673e-05,
"loss": 2.4906,
"step": 2700
},
{
"epoch": 0.02,
"learning_rate": 9.989554914363015e-05,
"loss": 2.4999,
"step": 2705
},
{
"epoch": 0.02,
"learning_rate": 9.989514793618109e-05,
"loss": 2.4766,
"step": 2710
},
{
"epoch": 0.02,
"learning_rate": 9.989474596047572e-05,
"loss": 2.5695,
"step": 2715
},
{
"epoch": 0.02,
"learning_rate": 9.989434321652024e-05,
"loss": 2.4705,
"step": 2720
},
{
"epoch": 0.02,
"learning_rate": 9.989393970432085e-05,
"loss": 2.5359,
"step": 2725
},
{
"epoch": 0.02,
"learning_rate": 9.989353542388378e-05,
"loss": 2.4943,
"step": 2730
},
{
"epoch": 0.02,
"learning_rate": 9.989313037521524e-05,
"loss": 2.5836,
"step": 2735
},
{
"epoch": 0.02,
"learning_rate": 9.989272455832145e-05,
"loss": 2.5616,
"step": 2740
},
{
"epoch": 0.02,
"learning_rate": 9.989231797320869e-05,
"loss": 2.4282,
"step": 2745
},
{
"epoch": 0.02,
"learning_rate": 9.989191061988319e-05,
"loss": 2.6391,
"step": 2750
},
{
"epoch": 0.02,
"learning_rate": 9.989150249835125e-05,
"loss": 2.5295,
"step": 2755
},
{
"epoch": 0.02,
"learning_rate": 9.989109360861913e-05,
"loss": 2.6409,
"step": 2760
},
{
"epoch": 0.02,
"learning_rate": 9.989068395069313e-05,
"loss": 2.481,
"step": 2765
},
{
"epoch": 0.02,
"learning_rate": 9.989027352457958e-05,
"loss": 2.5188,
"step": 2770
},
{
"epoch": 0.02,
"learning_rate": 9.988986233028478e-05,
"loss": 2.46,
"step": 2775
},
{
"epoch": 0.02,
"learning_rate": 9.988945036781507e-05,
"loss": 2.4218,
"step": 2780
},
{
"epoch": 0.02,
"learning_rate": 9.988903763717679e-05,
"loss": 2.4891,
"step": 2785
},
{
"epoch": 0.02,
"learning_rate": 9.988862413837628e-05,
"loss": 2.5554,
"step": 2790
},
{
"epoch": 0.02,
"learning_rate": 9.988820987141993e-05,
"loss": 2.634,
"step": 2795
},
{
"epoch": 0.02,
"learning_rate": 9.988779483631409e-05,
"loss": 2.5343,
"step": 2800
},
{
"epoch": 0.02,
"learning_rate": 9.98873790330652e-05,
"loss": 2.5503,
"step": 2805
},
{
"epoch": 0.02,
"learning_rate": 9.988696246167961e-05,
"loss": 2.5922,
"step": 2810
},
{
"epoch": 0.02,
"learning_rate": 9.988654512216378e-05,
"loss": 2.4938,
"step": 2815
},
{
"epoch": 0.02,
"learning_rate": 9.988612701452409e-05,
"loss": 2.4578,
"step": 2820
},
{
"epoch": 0.02,
"learning_rate": 9.988570813876701e-05,
"loss": 2.4034,
"step": 2825
},
{
"epoch": 0.02,
"learning_rate": 9.988528849489897e-05,
"loss": 2.5934,
"step": 2830
},
{
"epoch": 0.02,
"learning_rate": 9.988486808292646e-05,
"loss": 2.5703,
"step": 2835
},
{
"epoch": 0.02,
"learning_rate": 9.988444690285592e-05,
"loss": 2.5102,
"step": 2840
},
{
"epoch": 0.02,
"learning_rate": 9.988402495469385e-05,
"loss": 2.6149,
"step": 2845
},
{
"epoch": 0.02,
"learning_rate": 9.988360223844673e-05,
"loss": 2.5666,
"step": 2850
},
{
"epoch": 0.02,
"learning_rate": 9.98831787541211e-05,
"loss": 2.5636,
"step": 2855
},
{
"epoch": 0.02,
"learning_rate": 9.988275450172346e-05,
"loss": 2.5993,
"step": 2860
},
{
"epoch": 0.02,
"learning_rate": 9.988232948126037e-05,
"loss": 2.6198,
"step": 2865
},
{
"epoch": 0.02,
"learning_rate": 9.988190369273833e-05,
"loss": 2.5102,
"step": 2870
},
{
"epoch": 0.02,
"learning_rate": 9.988147713616391e-05,
"loss": 2.4577,
"step": 2875
},
{
"epoch": 0.02,
"learning_rate": 9.988104981154371e-05,
"loss": 2.5193,
"step": 2880
},
{
"epoch": 0.02,
"learning_rate": 9.988062171888429e-05,
"loss": 2.5052,
"step": 2885
},
{
"epoch": 0.02,
"learning_rate": 9.988019285819221e-05,
"loss": 2.5434,
"step": 2890
},
{
"epoch": 0.02,
"learning_rate": 9.987976322947411e-05,
"loss": 2.5902,
"step": 2895
},
{
"epoch": 0.02,
"learning_rate": 9.98793328327366e-05,
"loss": 2.4246,
"step": 2900
},
{
"epoch": 0.02,
"learning_rate": 9.98789016679863e-05,
"loss": 2.4512,
"step": 2905
},
{
"epoch": 0.02,
"learning_rate": 9.987846973522985e-05,
"loss": 2.4555,
"step": 2910
},
{
"epoch": 0.02,
"learning_rate": 9.98780370344739e-05,
"loss": 2.5373,
"step": 2915
},
{
"epoch": 0.02,
"learning_rate": 9.987760356572512e-05,
"loss": 2.5184,
"step": 2920
},
{
"epoch": 0.02,
"learning_rate": 9.987716932899018e-05,
"loss": 2.4406,
"step": 2925
},
{
"epoch": 0.02,
"learning_rate": 9.987673432427576e-05,
"loss": 2.5333,
"step": 2930
},
{
"epoch": 0.02,
"learning_rate": 9.987629855158855e-05,
"loss": 2.6413,
"step": 2935
},
{
"epoch": 0.02,
"learning_rate": 9.98758620109353e-05,
"loss": 2.6341,
"step": 2940
},
{
"epoch": 0.02,
"learning_rate": 9.98754247023227e-05,
"loss": 2.5173,
"step": 2945
},
{
"epoch": 0.02,
"learning_rate": 9.987498662575747e-05,
"loss": 2.555,
"step": 2950
},
{
"epoch": 0.02,
"learning_rate": 9.987454778124639e-05,
"loss": 2.5186,
"step": 2955
},
{
"epoch": 0.02,
"learning_rate": 9.987410816879619e-05,
"loss": 2.525,
"step": 2960
},
{
"epoch": 0.02,
"learning_rate": 9.987366778841365e-05,
"loss": 2.5631,
"step": 2965
},
{
"epoch": 0.02,
"learning_rate": 9.987322664010554e-05,
"loss": 2.5736,
"step": 2970
},
{
"epoch": 0.02,
"learning_rate": 9.987278472387867e-05,
"loss": 2.5567,
"step": 2975
},
{
"epoch": 0.02,
"learning_rate": 9.987234203973986e-05,
"loss": 2.6388,
"step": 2980
},
{
"epoch": 0.02,
"learning_rate": 9.987189858769586e-05,
"loss": 2.5009,
"step": 2985
},
{
"epoch": 0.02,
"learning_rate": 9.987145436775357e-05,
"loss": 2.4188,
"step": 2990
},
{
"epoch": 0.02,
"learning_rate": 9.98710093799198e-05,
"loss": 2.6081,
"step": 2995
},
{
"epoch": 0.02,
"learning_rate": 9.987056362420138e-05,
"loss": 2.497,
"step": 3000
},
{
"epoch": 0.02,
"learning_rate": 9.98701171006052e-05,
"loss": 2.5281,
"step": 3005
},
{
"epoch": 0.02,
"learning_rate": 9.986966980913817e-05,
"loss": 2.5494,
"step": 3010
},
{
"epoch": 0.02,
"learning_rate": 9.986922174980709e-05,
"loss": 2.5669,
"step": 3015
},
{
"epoch": 0.02,
"learning_rate": 9.986877292261893e-05,
"loss": 2.4999,
"step": 3020
},
{
"epoch": 0.02,
"learning_rate": 9.986832332758058e-05,
"loss": 2.5715,
"step": 3025
},
{
"epoch": 0.02,
"learning_rate": 9.986787296469895e-05,
"loss": 2.5708,
"step": 3030
},
{
"epoch": 0.02,
"learning_rate": 9.986742183398099e-05,
"loss": 2.6132,
"step": 3035
},
{
"epoch": 0.02,
"learning_rate": 9.986696993543366e-05,
"loss": 2.591,
"step": 3040
},
{
"epoch": 0.02,
"learning_rate": 9.986651726906387e-05,
"loss": 2.5843,
"step": 3045
},
{
"epoch": 0.02,
"learning_rate": 9.986606383487862e-05,
"loss": 2.5986,
"step": 3050
},
{
"epoch": 0.02,
"learning_rate": 9.98656096328849e-05,
"loss": 2.4665,
"step": 3055
},
{
"epoch": 0.02,
"learning_rate": 9.98651546630897e-05,
"loss": 2.5431,
"step": 3060
},
{
"epoch": 0.02,
"learning_rate": 9.986469892550001e-05,
"loss": 2.4558,
"step": 3065
},
{
"epoch": 0.02,
"learning_rate": 9.986424242012287e-05,
"loss": 2.5506,
"step": 3070
},
{
"epoch": 0.02,
"learning_rate": 9.986378514696528e-05,
"loss": 2.5463,
"step": 3075
},
{
"epoch": 0.02,
"learning_rate": 9.98633271060343e-05,
"loss": 2.4526,
"step": 3080
},
{
"epoch": 0.02,
"learning_rate": 9.9862868297337e-05,
"loss": 2.4713,
"step": 3085
},
{
"epoch": 0.02,
"learning_rate": 9.98624087208804e-05,
"loss": 2.4111,
"step": 3090
},
{
"epoch": 0.02,
"learning_rate": 9.98619483766716e-05,
"loss": 2.4782,
"step": 3095
},
{
"epoch": 0.02,
"learning_rate": 9.98614872647177e-05,
"loss": 2.5791,
"step": 3100
},
{
"epoch": 0.02,
"learning_rate": 9.986102538502579e-05,
"loss": 2.4983,
"step": 3105
},
{
"epoch": 0.02,
"learning_rate": 9.986056273760297e-05,
"loss": 2.3883,
"step": 3110
},
{
"epoch": 0.02,
"learning_rate": 9.986009932245637e-05,
"loss": 2.4637,
"step": 3115
},
{
"epoch": 0.02,
"learning_rate": 9.985963513959313e-05,
"loss": 2.4957,
"step": 3120
},
{
"epoch": 0.02,
"learning_rate": 9.985917018902041e-05,
"loss": 2.4633,
"step": 3125
},
{
"epoch": 0.02,
"learning_rate": 9.985870447074535e-05,
"loss": 2.5292,
"step": 3130
},
{
"epoch": 0.02,
"learning_rate": 9.985823798477512e-05,
"loss": 2.5051,
"step": 3135
},
{
"epoch": 0.02,
"learning_rate": 9.985777073111691e-05,
"loss": 2.5481,
"step": 3140
},
{
"epoch": 0.02,
"learning_rate": 9.985730270977791e-05,
"loss": 2.5249,
"step": 3145
},
{
"epoch": 0.02,
"learning_rate": 9.985683392076534e-05,
"loss": 2.5668,
"step": 3150
},
{
"epoch": 0.02,
"learning_rate": 9.98563643640864e-05,
"loss": 2.5712,
"step": 3155
},
{
"epoch": 0.02,
"learning_rate": 9.985589403974833e-05,
"loss": 2.5443,
"step": 3160
},
{
"epoch": 0.02,
"learning_rate": 9.985542294775838e-05,
"loss": 2.4432,
"step": 3165
},
{
"epoch": 0.03,
"learning_rate": 9.985495108812377e-05,
"loss": 2.3771,
"step": 3170
},
{
"epoch": 0.03,
"learning_rate": 9.985447846085181e-05,
"loss": 2.4931,
"step": 3175
},
{
"epoch": 0.03,
"learning_rate": 9.985400506594974e-05,
"loss": 2.5313,
"step": 3180
},
{
"epoch": 0.03,
"learning_rate": 9.985353090342487e-05,
"loss": 2.5195,
"step": 3185
},
{
"epoch": 0.03,
"learning_rate": 9.985305597328449e-05,
"loss": 2.577,
"step": 3190
},
{
"epoch": 0.03,
"learning_rate": 9.985258027553593e-05,
"loss": 2.5203,
"step": 3195
},
{
"epoch": 0.03,
"learning_rate": 9.98521038101865e-05,
"loss": 2.5843,
"step": 3200
},
{
"epoch": 0.03,
"learning_rate": 9.985162657724354e-05,
"loss": 2.5346,
"step": 3205
},
{
"epoch": 0.03,
"learning_rate": 9.98511485767144e-05,
"loss": 2.4874,
"step": 3210
},
{
"epoch": 0.03,
"learning_rate": 9.985066980860642e-05,
"loss": 2.5721,
"step": 3215
},
{
"epoch": 0.03,
"learning_rate": 9.9850190272927e-05,
"loss": 2.5926,
"step": 3220
},
{
"epoch": 0.03,
"learning_rate": 9.984970996968352e-05,
"loss": 2.4917,
"step": 3225
},
{
"epoch": 0.03,
"learning_rate": 9.984922889888335e-05,
"loss": 2.4411,
"step": 3230
},
{
"epoch": 0.03,
"learning_rate": 9.984874706053393e-05,
"loss": 2.4835,
"step": 3235
},
{
"epoch": 0.03,
"learning_rate": 9.984826445464266e-05,
"loss": 2.4013,
"step": 3240
},
{
"epoch": 0.03,
"learning_rate": 9.984778108121698e-05,
"loss": 2.4119,
"step": 3245
},
{
"epoch": 0.03,
"learning_rate": 9.984729694026431e-05,
"loss": 2.5511,
"step": 3250
},
{
"epoch": 0.03,
"learning_rate": 9.984681203179214e-05,
"loss": 2.4924,
"step": 3255
},
{
"epoch": 0.03,
"learning_rate": 9.984632635580792e-05,
"loss": 2.489,
"step": 3260
},
{
"epoch": 0.03,
"learning_rate": 9.984583991231913e-05,
"loss": 2.5243,
"step": 3265
},
{
"epoch": 0.03,
"learning_rate": 9.984535270133324e-05,
"loss": 2.5499,
"step": 3270
},
{
"epoch": 0.03,
"learning_rate": 9.984486472285777e-05,
"loss": 2.4994,
"step": 3275
},
{
"epoch": 0.03,
"learning_rate": 9.984437597690024e-05,
"loss": 2.4315,
"step": 3280
},
{
"epoch": 0.03,
"learning_rate": 9.984388646346817e-05,
"loss": 2.4635,
"step": 3285
},
{
"epoch": 0.03,
"learning_rate": 9.984339618256909e-05,
"loss": 2.5705,
"step": 3290
},
{
"epoch": 0.03,
"learning_rate": 9.984290513421057e-05,
"loss": 2.5505,
"step": 3295
},
{
"epoch": 0.03,
"learning_rate": 9.984241331840015e-05,
"loss": 2.5131,
"step": 3300
},
{
"epoch": 0.03,
"learning_rate": 9.984192073514538e-05,
"loss": 2.6361,
"step": 3305
},
{
"epoch": 0.03,
"learning_rate": 9.98414273844539e-05,
"loss": 2.5756,
"step": 3310
},
{
"epoch": 0.03,
"learning_rate": 9.984093326633325e-05,
"loss": 2.5964,
"step": 3315
},
{
"epoch": 0.03,
"learning_rate": 9.98404383807911e-05,
"loss": 2.5278,
"step": 3320
},
{
"epoch": 0.03,
"learning_rate": 9.983994272783502e-05,
"loss": 2.5841,
"step": 3325
},
{
"epoch": 0.03,
"learning_rate": 9.983944630747266e-05,
"loss": 2.5163,
"step": 3330
},
{
"epoch": 0.03,
"learning_rate": 9.983894911971167e-05,
"loss": 2.5362,
"step": 3335
},
{
"epoch": 0.03,
"learning_rate": 9.983845116455969e-05,
"loss": 2.4797,
"step": 3340
},
{
"epoch": 0.03,
"learning_rate": 9.983795244202439e-05,
"loss": 2.6148,
"step": 3345
},
{
"epoch": 0.03,
"learning_rate": 9.983745295211347e-05,
"loss": 2.5263,
"step": 3350
},
{
"epoch": 0.03,
"learning_rate": 9.983695269483459e-05,
"loss": 2.5832,
"step": 3355
},
{
"epoch": 0.03,
"learning_rate": 9.983645167019546e-05,
"loss": 2.4736,
"step": 3360
},
{
"epoch": 0.03,
"learning_rate": 9.983594987820381e-05,
"loss": 2.5781,
"step": 3365
},
{
"epoch": 0.03,
"learning_rate": 9.983544731886735e-05,
"loss": 2.5449,
"step": 3370
},
{
"epoch": 0.03,
"learning_rate": 9.983494399219386e-05,
"loss": 2.4939,
"step": 3375
},
{
"epoch": 0.03,
"learning_rate": 9.983443989819103e-05,
"loss": 2.4699,
"step": 3380
},
{
"epoch": 0.03,
"learning_rate": 9.983393503686665e-05,
"loss": 2.4089,
"step": 3385
},
{
"epoch": 0.03,
"learning_rate": 9.983342940822849e-05,
"loss": 2.5499,
"step": 3390
},
{
"epoch": 0.03,
"learning_rate": 9.983292301228432e-05,
"loss": 2.5165,
"step": 3395
},
{
"epoch": 0.03,
"learning_rate": 9.983241584904197e-05,
"loss": 2.5357,
"step": 3400
},
{
"epoch": 0.03,
"learning_rate": 9.983190791850925e-05,
"loss": 2.4769,
"step": 3405
},
{
"epoch": 0.03,
"learning_rate": 9.983139922069394e-05,
"loss": 2.5239,
"step": 3410
},
{
"epoch": 0.03,
"learning_rate": 9.983088975560391e-05,
"loss": 2.4193,
"step": 3415
},
{
"epoch": 0.03,
"learning_rate": 9.983037952324699e-05,
"loss": 2.4663,
"step": 3420
},
{
"epoch": 0.03,
"learning_rate": 9.982986852363102e-05,
"loss": 2.5382,
"step": 3425
},
{
"epoch": 0.03,
"learning_rate": 9.98293567567639e-05,
"loss": 2.5092,
"step": 3430
},
{
"epoch": 0.03,
"learning_rate": 9.982884422265349e-05,
"loss": 2.5545,
"step": 3435
},
{
"epoch": 0.03,
"learning_rate": 9.982833092130768e-05,
"loss": 2.4647,
"step": 3440
},
{
"epoch": 0.03,
"learning_rate": 9.982781685273438e-05,
"loss": 2.4582,
"step": 3445
},
{
"epoch": 0.03,
"learning_rate": 9.982730201694153e-05,
"loss": 2.5022,
"step": 3450
},
{
"epoch": 0.03,
"learning_rate": 9.982678641393701e-05,
"loss": 2.4989,
"step": 3455
},
{
"epoch": 0.03,
"learning_rate": 9.98262700437288e-05,
"loss": 2.4925,
"step": 3460
},
{
"epoch": 0.03,
"learning_rate": 9.982575290632481e-05,
"loss": 2.6004,
"step": 3465
},
{
"epoch": 0.03,
"learning_rate": 9.982523500173304e-05,
"loss": 2.4672,
"step": 3470
},
{
"epoch": 0.03,
"learning_rate": 9.982471632996144e-05,
"loss": 2.59,
"step": 3475
},
{
"epoch": 0.03,
"learning_rate": 9.9824196891018e-05,
"loss": 2.4516,
"step": 3480
},
{
"epoch": 0.03,
"learning_rate": 9.982367668491075e-05,
"loss": 2.5149,
"step": 3485
},
{
"epoch": 0.03,
"learning_rate": 9.982315571164765e-05,
"loss": 2.5058,
"step": 3490
},
{
"epoch": 0.03,
"learning_rate": 9.982263397123675e-05,
"loss": 2.4954,
"step": 3495
},
{
"epoch": 0.03,
"learning_rate": 9.982211146368608e-05,
"loss": 2.5242,
"step": 3500
},
{
"epoch": 0.03,
"learning_rate": 9.982158818900369e-05,
"loss": 2.4384,
"step": 3505
},
{
"epoch": 0.03,
"learning_rate": 9.982106414719762e-05,
"loss": 2.5181,
"step": 3510
},
{
"epoch": 0.03,
"learning_rate": 9.982053933827595e-05,
"loss": 2.4915,
"step": 3515
},
{
"epoch": 0.03,
"learning_rate": 9.982001376224677e-05,
"loss": 2.4907,
"step": 3520
},
{
"epoch": 0.03,
"learning_rate": 9.981948741911816e-05,
"loss": 2.5326,
"step": 3525
},
{
"epoch": 0.03,
"learning_rate": 9.981896030889821e-05,
"loss": 2.549,
"step": 3530
},
{
"epoch": 0.03,
"learning_rate": 9.981843243159507e-05,
"loss": 2.3802,
"step": 3535
},
{
"epoch": 0.03,
"learning_rate": 9.981790378721684e-05,
"loss": 2.499,
"step": 3540
},
{
"epoch": 0.03,
"learning_rate": 9.98173743757717e-05,
"loss": 2.491,
"step": 3545
},
{
"epoch": 0.03,
"learning_rate": 9.981684419726775e-05,
"loss": 2.4371,
"step": 3550
},
{
"epoch": 0.03,
"learning_rate": 9.981631325171318e-05,
"loss": 2.3461,
"step": 3555
},
{
"epoch": 0.03,
"learning_rate": 9.981578153911616e-05,
"loss": 2.4708,
"step": 3560
},
{
"epoch": 0.03,
"learning_rate": 9.981524905948487e-05,
"loss": 2.3976,
"step": 3565
},
{
"epoch": 0.03,
"learning_rate": 9.981471581282753e-05,
"loss": 2.4421,
"step": 3570
},
{
"epoch": 0.03,
"learning_rate": 9.981418179915234e-05,
"loss": 2.5103,
"step": 3575
},
{
"epoch": 0.03,
"learning_rate": 9.981364701846752e-05,
"loss": 2.537,
"step": 3580
},
{
"epoch": 0.03,
"learning_rate": 9.981311147078128e-05,
"loss": 2.5119,
"step": 3585
},
{
"epoch": 0.03,
"learning_rate": 9.981257515610192e-05,
"loss": 2.5055,
"step": 3590
},
{
"epoch": 0.03,
"learning_rate": 9.981203807443764e-05,
"loss": 2.5406,
"step": 3595
},
{
"epoch": 0.03,
"learning_rate": 9.981150022579676e-05,
"loss": 2.4755,
"step": 3600
},
{
"epoch": 0.03,
"learning_rate": 9.981096161018754e-05,
"loss": 2.5475,
"step": 3605
},
{
"epoch": 0.03,
"learning_rate": 9.981042222761826e-05,
"loss": 2.5503,
"step": 3610
},
{
"epoch": 0.03,
"learning_rate": 9.980988207809726e-05,
"loss": 2.4972,
"step": 3615
},
{
"epoch": 0.03,
"learning_rate": 9.98093411616328e-05,
"loss": 2.4677,
"step": 3620
},
{
"epoch": 0.03,
"learning_rate": 9.980879947823327e-05,
"loss": 2.4781,
"step": 3625
},
{
"epoch": 0.03,
"learning_rate": 9.980825702790696e-05,
"loss": 2.5027,
"step": 3630
},
{
"epoch": 0.03,
"learning_rate": 9.980771381066227e-05,
"loss": 2.4611,
"step": 3635
},
{
"epoch": 0.03,
"learning_rate": 9.980716982650753e-05,
"loss": 2.3804,
"step": 3640
},
{
"epoch": 0.03,
"learning_rate": 9.980662507545113e-05,
"loss": 2.3927,
"step": 3645
},
{
"epoch": 0.03,
"learning_rate": 9.980607955750146e-05,
"loss": 2.5181,
"step": 3650
},
{
"epoch": 0.03,
"learning_rate": 9.98055332726669e-05,
"loss": 2.4685,
"step": 3655
},
{
"epoch": 0.03,
"learning_rate": 9.980498622095588e-05,
"loss": 2.5104,
"step": 3660
},
{
"epoch": 0.03,
"learning_rate": 9.980443840237681e-05,
"loss": 2.5798,
"step": 3665
},
{
"epoch": 0.03,
"learning_rate": 9.980388981693816e-05,
"loss": 2.3567,
"step": 3670
},
{
"epoch": 0.03,
"learning_rate": 9.980334046464832e-05,
"loss": 2.4839,
"step": 3675
},
{
"epoch": 0.03,
"learning_rate": 9.980279034551579e-05,
"loss": 2.6321,
"step": 3680
},
{
"epoch": 0.03,
"learning_rate": 9.980223945954902e-05,
"loss": 2.5261,
"step": 3685
},
{
"epoch": 0.03,
"learning_rate": 9.980168780675651e-05,
"loss": 2.4431,
"step": 3690
},
{
"epoch": 0.03,
"learning_rate": 9.980113538714675e-05,
"loss": 2.4281,
"step": 3695
},
{
"epoch": 0.03,
"learning_rate": 9.980058220072825e-05,
"loss": 2.4993,
"step": 3700
},
{
"epoch": 0.03,
"learning_rate": 9.98000282475095e-05,
"loss": 2.5749,
"step": 3705
},
{
"epoch": 0.03,
"learning_rate": 9.979947352749906e-05,
"loss": 2.4565,
"step": 3710
},
{
"epoch": 0.03,
"learning_rate": 9.979891804070545e-05,
"loss": 2.4835,
"step": 3715
},
{
"epoch": 0.03,
"learning_rate": 9.979836178713724e-05,
"loss": 2.4432,
"step": 3720
},
{
"epoch": 0.03,
"learning_rate": 9.979780476680298e-05,
"loss": 2.4162,
"step": 3725
},
{
"epoch": 0.03,
"learning_rate": 9.979724697971125e-05,
"loss": 2.5042,
"step": 3730
},
{
"epoch": 0.03,
"learning_rate": 9.979668842587065e-05,
"loss": 2.4409,
"step": 3735
},
{
"epoch": 0.03,
"learning_rate": 9.979612910528976e-05,
"loss": 2.4441,
"step": 3740
},
{
"epoch": 0.03,
"learning_rate": 9.979556901797723e-05,
"loss": 2.6701,
"step": 3745
},
{
"epoch": 0.03,
"learning_rate": 9.979500816394165e-05,
"loss": 2.481,
"step": 3750
},
{
"epoch": 0.03,
"learning_rate": 9.979444654319165e-05,
"loss": 2.4403,
"step": 3755
},
{
"epoch": 0.03,
"learning_rate": 9.979388415573591e-05,
"loss": 2.4052,
"step": 3760
},
{
"epoch": 0.03,
"learning_rate": 9.979332100158307e-05,
"loss": 2.4247,
"step": 3765
},
{
"epoch": 0.03,
"learning_rate": 9.979275708074179e-05,
"loss": 2.4894,
"step": 3770
},
{
"epoch": 0.03,
"learning_rate": 9.979219239322078e-05,
"loss": 2.5155,
"step": 3775
},
{
"epoch": 0.03,
"learning_rate": 9.979162693902871e-05,
"loss": 2.4897,
"step": 3780
},
{
"epoch": 0.03,
"learning_rate": 9.97910607181743e-05,
"loss": 2.4329,
"step": 3785
},
{
"epoch": 0.03,
"learning_rate": 9.979049373066626e-05,
"loss": 2.5458,
"step": 3790
},
{
"epoch": 0.03,
"learning_rate": 9.978992597651334e-05,
"loss": 2.4431,
"step": 3795
},
{
"epoch": 0.03,
"learning_rate": 9.978935745572424e-05,
"loss": 2.4022,
"step": 3800
},
{
"epoch": 0.03,
"learning_rate": 9.978878816830776e-05,
"loss": 2.4928,
"step": 3805
},
{
"epoch": 0.03,
"learning_rate": 9.978821811427263e-05,
"loss": 2.4174,
"step": 3810
},
{
"epoch": 0.03,
"learning_rate": 9.978764729362766e-05,
"loss": 2.4849,
"step": 3815
},
{
"epoch": 0.03,
"learning_rate": 9.978707570638162e-05,
"loss": 2.3698,
"step": 3820
},
{
"epoch": 0.03,
"learning_rate": 9.978650335254332e-05,
"loss": 2.435,
"step": 3825
},
{
"epoch": 0.03,
"learning_rate": 9.978593023212155e-05,
"loss": 2.4361,
"step": 3830
},
{
"epoch": 0.03,
"learning_rate": 9.978535634512515e-05,
"loss": 2.4449,
"step": 3835
},
{
"epoch": 0.03,
"learning_rate": 9.978478169156298e-05,
"loss": 2.428,
"step": 3840
},
{
"epoch": 0.03,
"learning_rate": 9.978420627144385e-05,
"loss": 2.4926,
"step": 3845
},
{
"epoch": 0.03,
"learning_rate": 9.978363008477665e-05,
"loss": 2.359,
"step": 3850
},
{
"epoch": 0.03,
"learning_rate": 9.978305313157022e-05,
"loss": 2.4484,
"step": 3855
},
{
"epoch": 0.03,
"learning_rate": 9.978247541183348e-05,
"loss": 2.5426,
"step": 3860
},
{
"epoch": 0.03,
"learning_rate": 9.978189692557529e-05,
"loss": 2.3577,
"step": 3865
},
{
"epoch": 0.03,
"learning_rate": 9.978131767280456e-05,
"loss": 2.5071,
"step": 3870
},
{
"epoch": 0.03,
"learning_rate": 9.978073765353026e-05,
"loss": 2.4406,
"step": 3875
},
{
"epoch": 0.03,
"learning_rate": 9.978015686776125e-05,
"loss": 2.4656,
"step": 3880
},
{
"epoch": 0.03,
"learning_rate": 9.977957531550652e-05,
"loss": 2.537,
"step": 3885
},
{
"epoch": 0.03,
"learning_rate": 9.977899299677502e-05,
"loss": 2.4374,
"step": 3890
},
{
"epoch": 0.03,
"learning_rate": 9.97784099115757e-05,
"loss": 2.521,
"step": 3895
},
{
"epoch": 0.03,
"learning_rate": 9.977782605991754e-05,
"loss": 2.4189,
"step": 3900
},
{
"epoch": 0.03,
"learning_rate": 9.977724144180953e-05,
"loss": 2.4391,
"step": 3905
},
{
"epoch": 0.03,
"learning_rate": 9.977665605726067e-05,
"loss": 2.3809,
"step": 3910
},
{
"epoch": 0.03,
"learning_rate": 9.977606990628001e-05,
"loss": 2.4366,
"step": 3915
},
{
"epoch": 0.03,
"learning_rate": 9.977548298887651e-05,
"loss": 2.4703,
"step": 3920
},
{
"epoch": 0.03,
"learning_rate": 9.977489530505927e-05,
"loss": 2.423,
"step": 3925
},
{
"epoch": 0.03,
"learning_rate": 9.97743068548373e-05,
"loss": 2.5142,
"step": 3930
},
{
"epoch": 0.03,
"learning_rate": 9.977371763821966e-05,
"loss": 2.4254,
"step": 3935
},
{
"epoch": 0.03,
"learning_rate": 9.977312765521545e-05,
"loss": 2.4951,
"step": 3940
},
{
"epoch": 0.03,
"learning_rate": 9.977253690583372e-05,
"loss": 2.5265,
"step": 3945
},
{
"epoch": 0.03,
"learning_rate": 9.97719453900836e-05,
"loss": 2.5094,
"step": 3950
},
{
"epoch": 0.03,
"learning_rate": 9.977135310797418e-05,
"loss": 2.423,
"step": 3955
},
{
"epoch": 0.03,
"learning_rate": 9.977076005951458e-05,
"loss": 2.4241,
"step": 3960
},
{
"epoch": 0.03,
"learning_rate": 9.977016624471392e-05,
"loss": 2.4672,
"step": 3965
},
{
"epoch": 0.03,
"learning_rate": 9.976957166358137e-05,
"loss": 2.4234,
"step": 3970
},
{
"epoch": 0.03,
"learning_rate": 9.976897631612607e-05,
"loss": 2.4141,
"step": 3975
},
{
"epoch": 0.03,
"learning_rate": 9.97683802023572e-05,
"loss": 2.3811,
"step": 3980
},
{
"epoch": 0.03,
"learning_rate": 9.976778332228389e-05,
"loss": 2.3569,
"step": 3985
},
{
"epoch": 0.03,
"learning_rate": 9.97671856759154e-05,
"loss": 2.4439,
"step": 3990
},
{
"epoch": 0.03,
"learning_rate": 9.976658726326088e-05,
"loss": 2.4663,
"step": 3995
},
{
"epoch": 0.03,
"learning_rate": 9.976598808432957e-05,
"loss": 2.4188,
"step": 4000
},
{
"epoch": 0.03,
"learning_rate": 9.976538813913068e-05,
"loss": 2.3859,
"step": 4005
},
{
"epoch": 0.03,
"learning_rate": 9.976478742767347e-05,
"loss": 2.3342,
"step": 4010
},
{
"epoch": 0.03,
"learning_rate": 9.976418594996716e-05,
"loss": 2.4667,
"step": 4015
},
{
"epoch": 0.03,
"learning_rate": 9.976358370602103e-05,
"loss": 2.4712,
"step": 4020
},
{
"epoch": 0.03,
"learning_rate": 9.976298069584438e-05,
"loss": 2.3967,
"step": 4025
},
{
"epoch": 0.03,
"learning_rate": 9.976237691944643e-05,
"loss": 2.4175,
"step": 4030
},
{
"epoch": 0.03,
"learning_rate": 9.976177237683653e-05,
"loss": 2.5705,
"step": 4035
},
{
"epoch": 0.03,
"learning_rate": 9.976116706802395e-05,
"loss": 2.4835,
"step": 4040
},
{
"epoch": 0.03,
"learning_rate": 9.976056099301805e-05,
"loss": 2.4658,
"step": 4045
},
{
"epoch": 0.03,
"learning_rate": 9.975995415182812e-05,
"loss": 2.4888,
"step": 4050
},
{
"epoch": 0.03,
"learning_rate": 9.975934654446355e-05,
"loss": 2.4535,
"step": 4055
},
{
"epoch": 0.03,
"learning_rate": 9.975873817093367e-05,
"loss": 2.4171,
"step": 4060
},
{
"epoch": 0.03,
"learning_rate": 9.975812903124782e-05,
"loss": 2.4423,
"step": 4065
},
{
"epoch": 0.03,
"learning_rate": 9.975751912541544e-05,
"loss": 2.4689,
"step": 4070
},
{
"epoch": 0.03,
"learning_rate": 9.975690845344588e-05,
"loss": 2.53,
"step": 4075
},
{
"epoch": 0.03,
"learning_rate": 9.975629701534854e-05,
"loss": 2.4122,
"step": 4080
},
{
"epoch": 0.03,
"learning_rate": 9.975568481113285e-05,
"loss": 2.4387,
"step": 4085
},
{
"epoch": 0.03,
"learning_rate": 9.975507184080823e-05,
"loss": 2.4948,
"step": 4090
},
{
"epoch": 0.03,
"learning_rate": 9.975445810438413e-05,
"loss": 2.4701,
"step": 4095
},
{
"epoch": 0.03,
"learning_rate": 9.975384360186999e-05,
"loss": 2.4457,
"step": 4100
},
{
"epoch": 0.03,
"learning_rate": 9.975322833327527e-05,
"loss": 2.5128,
"step": 4105
},
{
"epoch": 0.03,
"learning_rate": 9.975261229860944e-05,
"loss": 2.3839,
"step": 4110
},
{
"epoch": 0.03,
"learning_rate": 9.975199549788199e-05,
"loss": 2.3738,
"step": 4115
},
{
"epoch": 0.03,
"learning_rate": 9.975137793110243e-05,
"loss": 2.5266,
"step": 4120
},
{
"epoch": 0.03,
"learning_rate": 9.975075959828026e-05,
"loss": 2.3991,
"step": 4125
},
{
"epoch": 0.03,
"learning_rate": 9.975014049942499e-05,
"loss": 2.4292,
"step": 4130
},
{
"epoch": 0.03,
"learning_rate": 9.974952063454615e-05,
"loss": 2.4484,
"step": 4135
},
{
"epoch": 0.03,
"learning_rate": 9.974890000365332e-05,
"loss": 2.4619,
"step": 4140
},
{
"epoch": 0.03,
"learning_rate": 9.9748278606756e-05,
"loss": 2.4383,
"step": 4145
},
{
"epoch": 0.03,
"learning_rate": 9.974765644386382e-05,
"loss": 2.4086,
"step": 4150
},
{
"epoch": 0.03,
"learning_rate": 9.974703351498631e-05,
"loss": 2.4246,
"step": 4155
},
{
"epoch": 0.03,
"learning_rate": 9.974640982013307e-05,
"loss": 2.4526,
"step": 4160
},
{
"epoch": 0.03,
"learning_rate": 9.974578535931374e-05,
"loss": 2.4776,
"step": 4165
},
{
"epoch": 0.03,
"learning_rate": 9.974516013253789e-05,
"loss": 2.4841,
"step": 4170
},
{
"epoch": 0.03,
"learning_rate": 9.974453413981517e-05,
"loss": 2.4132,
"step": 4175
},
{
"epoch": 0.03,
"learning_rate": 9.974390738115521e-05,
"loss": 2.4743,
"step": 4180
},
{
"epoch": 0.03,
"learning_rate": 9.974327985656767e-05,
"loss": 2.4804,
"step": 4185
},
{
"epoch": 0.03,
"learning_rate": 9.974265156606221e-05,
"loss": 2.4502,
"step": 4190
},
{
"epoch": 0.03,
"learning_rate": 9.974202250964849e-05,
"loss": 2.4775,
"step": 4195
},
{
"epoch": 0.03,
"learning_rate": 9.974139268733623e-05,
"loss": 2.5803,
"step": 4200
},
{
"epoch": 0.03,
"learning_rate": 9.974076209913508e-05,
"loss": 2.5239,
"step": 4205
},
{
"epoch": 0.03,
"learning_rate": 9.974013074505477e-05,
"loss": 2.4058,
"step": 4210
},
{
"epoch": 0.03,
"learning_rate": 9.973949862510503e-05,
"loss": 2.4783,
"step": 4215
},
{
"epoch": 0.03,
"learning_rate": 9.97388657392956e-05,
"loss": 2.5459,
"step": 4220
},
{
"epoch": 0.03,
"learning_rate": 9.97382320876362e-05,
"loss": 2.5023,
"step": 4225
},
{
"epoch": 0.03,
"learning_rate": 9.973759767013662e-05,
"loss": 2.4503,
"step": 4230
},
{
"epoch": 0.03,
"learning_rate": 9.973696248680659e-05,
"loss": 2.4341,
"step": 4235
},
{
"epoch": 0.03,
"learning_rate": 9.97363265376559e-05,
"loss": 2.4958,
"step": 4240
},
{
"epoch": 0.03,
"learning_rate": 9.973568982269436e-05,
"loss": 2.4286,
"step": 4245
},
{
"epoch": 0.03,
"learning_rate": 9.973505234193177e-05,
"loss": 2.4703,
"step": 4250
},
{
"epoch": 0.03,
"learning_rate": 9.973441409537793e-05,
"loss": 2.4102,
"step": 4255
},
{
"epoch": 0.03,
"learning_rate": 9.973377508304269e-05,
"loss": 2.4168,
"step": 4260
},
{
"epoch": 0.03,
"learning_rate": 9.973313530493586e-05,
"loss": 2.4547,
"step": 4265
},
{
"epoch": 0.03,
"learning_rate": 9.973249476106732e-05,
"loss": 2.5417,
"step": 4270
},
{
"epoch": 0.03,
"learning_rate": 9.97318534514469e-05,
"loss": 2.5325,
"step": 4275
},
{
"epoch": 0.03,
"learning_rate": 9.97312113760845e-05,
"loss": 2.45,
"step": 4280
},
{
"epoch": 0.03,
"learning_rate": 9.973056853499001e-05,
"loss": 2.5066,
"step": 4285
},
{
"epoch": 0.03,
"learning_rate": 9.972992492817331e-05,
"loss": 2.4899,
"step": 4290
},
{
"epoch": 0.03,
"learning_rate": 9.972928055564432e-05,
"loss": 2.5845,
"step": 4295
},
{
"epoch": 0.03,
"learning_rate": 9.972863541741297e-05,
"loss": 2.4008,
"step": 4300
},
{
"epoch": 0.03,
"learning_rate": 9.972798951348918e-05,
"loss": 2.4234,
"step": 4305
},
{
"epoch": 0.03,
"learning_rate": 9.972734284388288e-05,
"loss": 2.4782,
"step": 4310
},
{
"epoch": 0.03,
"learning_rate": 9.972669540860406e-05,
"loss": 2.4742,
"step": 4315
},
{
"epoch": 0.03,
"learning_rate": 9.972604720766268e-05,
"loss": 2.5593,
"step": 4320
},
{
"epoch": 0.03,
"learning_rate": 9.97253982410687e-05,
"loss": 2.3879,
"step": 4325
},
{
"epoch": 0.03,
"learning_rate": 9.972474850883213e-05,
"loss": 2.5097,
"step": 4330
},
{
"epoch": 0.03,
"learning_rate": 9.972409801096299e-05,
"loss": 2.3041,
"step": 4335
},
{
"epoch": 0.03,
"learning_rate": 9.972344674747125e-05,
"loss": 2.4937,
"step": 4340
},
{
"epoch": 0.03,
"learning_rate": 9.972279471836698e-05,
"loss": 2.4039,
"step": 4345
},
{
"epoch": 0.03,
"learning_rate": 9.972214192366021e-05,
"loss": 2.41,
"step": 4350
},
{
"epoch": 0.03,
"learning_rate": 9.972148836336098e-05,
"loss": 2.4245,
"step": 4355
},
{
"epoch": 0.03,
"learning_rate": 9.972083403747937e-05,
"loss": 2.5008,
"step": 4360
},
{
"epoch": 0.03,
"learning_rate": 9.972017894602542e-05,
"loss": 2.4767,
"step": 4365
},
{
"epoch": 0.03,
"learning_rate": 9.971952308900925e-05,
"loss": 2.4756,
"step": 4370
},
{
"epoch": 0.03,
"learning_rate": 9.971886646644096e-05,
"loss": 2.4911,
"step": 4375
},
{
"epoch": 0.03,
"learning_rate": 9.971820907833065e-05,
"loss": 2.434,
"step": 4380
},
{
"epoch": 0.03,
"learning_rate": 9.971755092468843e-05,
"loss": 2.3326,
"step": 4385
},
{
"epoch": 0.03,
"learning_rate": 9.971689200552445e-05,
"loss": 2.5042,
"step": 4390
},
{
"epoch": 0.03,
"learning_rate": 9.971623232084886e-05,
"loss": 2.4468,
"step": 4395
},
{
"epoch": 0.03,
"learning_rate": 9.971557187067179e-05,
"loss": 2.498,
"step": 4400
},
{
"epoch": 0.03,
"learning_rate": 9.971491065500345e-05,
"loss": 2.3639,
"step": 4405
},
{
"epoch": 0.03,
"learning_rate": 9.9714248673854e-05,
"loss": 2.4773,
"step": 4410
},
{
"epoch": 0.03,
"learning_rate": 9.971358592723361e-05,
"loss": 2.3826,
"step": 4415
},
{
"epoch": 0.03,
"learning_rate": 9.97129224151525e-05,
"loss": 2.393,
"step": 4420
},
{
"epoch": 0.03,
"learning_rate": 9.971225813762094e-05,
"loss": 2.4121,
"step": 4425
},
{
"epoch": 0.03,
"learning_rate": 9.971159309464906e-05,
"loss": 2.5083,
"step": 4430
},
{
"epoch": 0.04,
"learning_rate": 9.971092728624718e-05,
"loss": 2.5473,
"step": 4435
},
{
"epoch": 0.04,
"learning_rate": 9.97102607124255e-05,
"loss": 2.4578,
"step": 4440
},
{
"epoch": 0.04,
"learning_rate": 9.970959337319433e-05,
"loss": 2.4254,
"step": 4445
},
{
"epoch": 0.04,
"learning_rate": 9.970892526856392e-05,
"loss": 2.4291,
"step": 4450
},
{
"epoch": 0.04,
"learning_rate": 9.970825639854454e-05,
"loss": 2.4437,
"step": 4455
},
{
"epoch": 0.04,
"learning_rate": 9.970758676314652e-05,
"loss": 2.409,
"step": 4460
},
{
"epoch": 0.04,
"learning_rate": 9.970691636238017e-05,
"loss": 2.5235,
"step": 4465
},
{
"epoch": 0.04,
"learning_rate": 9.970624519625578e-05,
"loss": 2.4631,
"step": 4470
},
{
"epoch": 0.04,
"learning_rate": 9.970557326478371e-05,
"loss": 2.4312,
"step": 4475
},
{
"epoch": 0.04,
"learning_rate": 9.97049005679743e-05,
"loss": 2.4631,
"step": 4480
},
{
"epoch": 0.04,
"learning_rate": 9.970422710583791e-05,
"loss": 2.428,
"step": 4485
},
{
"epoch": 0.04,
"learning_rate": 9.970355287838491e-05,
"loss": 2.4882,
"step": 4490
},
{
"epoch": 0.04,
"learning_rate": 9.970287788562567e-05,
"loss": 2.439,
"step": 4495
},
{
"epoch": 0.04,
"learning_rate": 9.97022021275706e-05,
"loss": 2.468,
"step": 4500
},
{
"epoch": 0.04,
"learning_rate": 9.97015256042301e-05,
"loss": 2.4097,
"step": 4505
},
{
"epoch": 0.04,
"learning_rate": 9.97008483156146e-05,
"loss": 2.4334,
"step": 4510
},
{
"epoch": 0.04,
"learning_rate": 9.970017026173448e-05,
"loss": 2.4445,
"step": 4515
},
{
"epoch": 0.04,
"learning_rate": 9.969949144260022e-05,
"loss": 2.4658,
"step": 4520
},
{
"epoch": 0.04,
"learning_rate": 9.969881185822227e-05,
"loss": 2.4308,
"step": 4525
},
{
"epoch": 0.04,
"learning_rate": 9.969813150861111e-05,
"loss": 2.3844,
"step": 4530
},
{
"epoch": 0.04,
"learning_rate": 9.969745039377717e-05,
"loss": 2.5329,
"step": 4535
},
{
"epoch": 0.04,
"learning_rate": 9.969676851373098e-05,
"loss": 2.5195,
"step": 4540
},
{
"epoch": 0.04,
"learning_rate": 9.969608586848301e-05,
"loss": 2.426,
"step": 4545
},
{
"epoch": 0.04,
"learning_rate": 9.969540245804377e-05,
"loss": 2.505,
"step": 4550
},
{
"epoch": 0.04,
"learning_rate": 9.96947182824238e-05,
"loss": 2.4318,
"step": 4555
},
{
"epoch": 0.04,
"learning_rate": 9.969403334163364e-05,
"loss": 2.4645,
"step": 4560
},
{
"epoch": 0.04,
"learning_rate": 9.969334763568383e-05,
"loss": 2.3778,
"step": 4565
},
{
"epoch": 0.04,
"learning_rate": 9.969266116458491e-05,
"loss": 2.3943,
"step": 4570
},
{
"epoch": 0.04,
"learning_rate": 9.969197392834747e-05,
"loss": 2.4365,
"step": 4575
},
{
"epoch": 0.04,
"learning_rate": 9.969128592698207e-05,
"loss": 2.4507,
"step": 4580
},
{
"epoch": 0.04,
"learning_rate": 9.969059716049934e-05,
"loss": 2.4142,
"step": 4585
},
{
"epoch": 0.04,
"learning_rate": 9.968990762890986e-05,
"loss": 2.4332,
"step": 4590
},
{
"epoch": 0.04,
"learning_rate": 9.968921733222425e-05,
"loss": 2.4159,
"step": 4595
},
{
"epoch": 0.04,
"learning_rate": 9.968852627045313e-05,
"loss": 2.5003,
"step": 4600
},
{
"epoch": 0.04,
"learning_rate": 9.968783444360716e-05,
"loss": 2.4103,
"step": 4605
},
{
"epoch": 0.04,
"learning_rate": 9.968714185169698e-05,
"loss": 2.4322,
"step": 4610
},
{
"epoch": 0.04,
"learning_rate": 9.968644849473326e-05,
"loss": 2.52,
"step": 4615
},
{
"epoch": 0.04,
"learning_rate": 9.968575437272667e-05,
"loss": 2.4315,
"step": 4620
},
{
"epoch": 0.04,
"learning_rate": 9.968505948568792e-05,
"loss": 2.4343,
"step": 4625
},
{
"epoch": 0.04,
"learning_rate": 9.968436383362766e-05,
"loss": 2.4706,
"step": 4630
},
{
"epoch": 0.04,
"learning_rate": 9.968366741655663e-05,
"loss": 2.4297,
"step": 4635
},
{
"epoch": 0.04,
"learning_rate": 9.968297023448558e-05,
"loss": 2.5185,
"step": 4640
},
{
"epoch": 0.04,
"learning_rate": 9.96822722874252e-05,
"loss": 2.4759,
"step": 4645
},
{
"epoch": 0.04,
"learning_rate": 9.968157357538625e-05,
"loss": 2.3797,
"step": 4650
},
{
"epoch": 0.04,
"learning_rate": 9.968087409837951e-05,
"loss": 2.4191,
"step": 4655
},
{
"epoch": 0.04,
"learning_rate": 9.968017385641574e-05,
"loss": 2.4686,
"step": 4660
},
{
"epoch": 0.04,
"learning_rate": 9.967947284950569e-05,
"loss": 2.4531,
"step": 4665
},
{
"epoch": 0.04,
"learning_rate": 9.967877107766021e-05,
"loss": 2.4087,
"step": 4670
},
{
"epoch": 0.04,
"learning_rate": 9.967806854089006e-05,
"loss": 2.4725,
"step": 4675
},
{
"epoch": 0.04,
"learning_rate": 9.967736523920607e-05,
"loss": 2.4022,
"step": 4680
},
{
"epoch": 0.04,
"learning_rate": 9.967666117261907e-05,
"loss": 2.3114,
"step": 4685
},
{
"epoch": 0.04,
"learning_rate": 9.96759563411399e-05,
"loss": 2.3932,
"step": 4690
},
{
"epoch": 0.04,
"learning_rate": 9.967525074477943e-05,
"loss": 2.4975,
"step": 4695
},
{
"epoch": 0.04,
"learning_rate": 9.96745443835485e-05,
"loss": 2.4636,
"step": 4700
},
{
"epoch": 0.04,
"learning_rate": 9.967383725745799e-05,
"loss": 2.4265,
"step": 4705
},
{
"epoch": 0.04,
"learning_rate": 9.96731293665188e-05,
"loss": 2.4709,
"step": 4710
},
{
"epoch": 0.04,
"learning_rate": 9.967242071074181e-05,
"loss": 2.4743,
"step": 4715
},
{
"epoch": 0.04,
"learning_rate": 9.967171129013797e-05,
"loss": 2.3916,
"step": 4720
},
{
"epoch": 0.04,
"learning_rate": 9.967100110471817e-05,
"loss": 2.369,
"step": 4725
},
{
"epoch": 0.04,
"learning_rate": 9.967029015449334e-05,
"loss": 2.3976,
"step": 4730
},
{
"epoch": 0.04,
"learning_rate": 9.966957843947444e-05,
"loss": 2.4274,
"step": 4735
},
{
"epoch": 0.04,
"learning_rate": 9.966886595967243e-05,
"loss": 2.3975,
"step": 4740
},
{
"epoch": 0.04,
"learning_rate": 9.966815271509828e-05,
"loss": 2.4005,
"step": 4745
},
{
"epoch": 0.04,
"learning_rate": 9.966743870576297e-05,
"loss": 2.4278,
"step": 4750
},
{
"epoch": 0.04,
"learning_rate": 9.96667239316775e-05,
"loss": 2.3963,
"step": 4755
},
{
"epoch": 0.04,
"learning_rate": 9.966600839285286e-05,
"loss": 2.3441,
"step": 4760
},
{
"epoch": 0.04,
"learning_rate": 9.966529208930008e-05,
"loss": 2.4095,
"step": 4765
},
{
"epoch": 0.04,
"learning_rate": 9.96645750210302e-05,
"loss": 2.464,
"step": 4770
},
{
"epoch": 0.04,
"learning_rate": 9.966385718805423e-05,
"loss": 2.3837,
"step": 4775
},
{
"epoch": 0.04,
"learning_rate": 9.966313859038325e-05,
"loss": 2.3771,
"step": 4780
},
{
"epoch": 0.04,
"learning_rate": 9.966241922802832e-05,
"loss": 2.4306,
"step": 4785
},
{
"epoch": 0.04,
"learning_rate": 9.96616991010005e-05,
"loss": 2.3703,
"step": 4790
},
{
"epoch": 0.04,
"learning_rate": 9.96609782093109e-05,
"loss": 2.3597,
"step": 4795
},
{
"epoch": 0.04,
"learning_rate": 9.96602565529706e-05,
"loss": 2.3954,
"step": 4800
},
{
"epoch": 0.04,
"learning_rate": 9.965953413199071e-05,
"loss": 2.4118,
"step": 4805
},
{
"epoch": 0.04,
"learning_rate": 9.965881094638238e-05,
"loss": 2.3658,
"step": 4810
},
{
"epoch": 0.04,
"learning_rate": 9.965808699615674e-05,
"loss": 2.3102,
"step": 4815
},
{
"epoch": 0.04,
"learning_rate": 9.965736228132492e-05,
"loss": 2.3527,
"step": 4820
},
{
"epoch": 0.04,
"learning_rate": 9.965663680189807e-05,
"loss": 2.4761,
"step": 4825
},
{
"epoch": 0.04,
"learning_rate": 9.965591055788738e-05,
"loss": 2.3677,
"step": 4830
},
{
"epoch": 0.04,
"learning_rate": 9.965518354930405e-05,
"loss": 2.5218,
"step": 4835
},
{
"epoch": 0.04,
"learning_rate": 9.965445577615922e-05,
"loss": 2.3519,
"step": 4840
},
{
"epoch": 0.04,
"learning_rate": 9.965372723846415e-05,
"loss": 2.4599,
"step": 4845
},
{
"epoch": 0.04,
"learning_rate": 9.965299793623003e-05,
"loss": 2.4154,
"step": 4850
},
{
"epoch": 0.04,
"learning_rate": 9.96522678694681e-05,
"loss": 2.399,
"step": 4855
},
{
"epoch": 0.04,
"learning_rate": 9.965153703818959e-05,
"loss": 2.5019,
"step": 4860
},
{
"epoch": 0.04,
"learning_rate": 9.965080544240575e-05,
"loss": 2.4092,
"step": 4865
},
{
"epoch": 0.04,
"learning_rate": 9.965007308212786e-05,
"loss": 2.3909,
"step": 4870
},
{
"epoch": 0.04,
"learning_rate": 9.964933995736721e-05,
"loss": 2.4012,
"step": 4875
},
{
"epoch": 0.04,
"learning_rate": 9.964860606813504e-05,
"loss": 2.3833,
"step": 4880
},
{
"epoch": 0.04,
"learning_rate": 9.96478714144427e-05,
"loss": 2.4573,
"step": 4885
},
{
"epoch": 0.04,
"learning_rate": 9.964713599630146e-05,
"loss": 2.4358,
"step": 4890
},
{
"epoch": 0.04,
"learning_rate": 9.964639981372267e-05,
"loss": 2.4069,
"step": 4895
},
{
"epoch": 0.04,
"learning_rate": 9.964566286671766e-05,
"loss": 2.4341,
"step": 4900
},
{
"epoch": 0.04,
"learning_rate": 9.964492515529777e-05,
"loss": 2.5132,
"step": 4905
},
{
"epoch": 0.04,
"learning_rate": 9.964418667947438e-05,
"loss": 2.4031,
"step": 4910
},
{
"epoch": 0.04,
"learning_rate": 9.964344743925884e-05,
"loss": 2.3754,
"step": 4915
},
{
"epoch": 0.04,
"learning_rate": 9.964270743466253e-05,
"loss": 2.3648,
"step": 4920
},
{
"epoch": 0.04,
"learning_rate": 9.964196666569686e-05,
"loss": 2.4817,
"step": 4925
},
{
"epoch": 0.04,
"learning_rate": 9.964122513237322e-05,
"loss": 2.436,
"step": 4930
},
{
"epoch": 0.04,
"learning_rate": 9.964048283470305e-05,
"loss": 2.434,
"step": 4935
},
{
"epoch": 0.04,
"learning_rate": 9.963973977269776e-05,
"loss": 2.4269,
"step": 4940
},
{
"epoch": 0.04,
"learning_rate": 9.963899594636879e-05,
"loss": 2.3994,
"step": 4945
},
{
"epoch": 0.04,
"learning_rate": 9.96382513557276e-05,
"loss": 2.3923,
"step": 4950
},
{
"epoch": 0.04,
"learning_rate": 9.963750600078565e-05,
"loss": 2.4796,
"step": 4955
},
{
"epoch": 0.04,
"learning_rate": 9.963675988155443e-05,
"loss": 2.3955,
"step": 4960
},
{
"epoch": 0.04,
"learning_rate": 9.963601299804541e-05,
"loss": 2.414,
"step": 4965
},
{
"epoch": 0.04,
"learning_rate": 9.963526535027012e-05,
"loss": 2.3382,
"step": 4970
},
{
"epoch": 0.04,
"learning_rate": 9.963451693824003e-05,
"loss": 2.2808,
"step": 4975
},
{
"epoch": 0.04,
"learning_rate": 9.963376776196669e-05,
"loss": 2.4828,
"step": 4980
},
{
"epoch": 0.04,
"learning_rate": 9.963301782146163e-05,
"loss": 2.4452,
"step": 4985
},
{
"epoch": 0.04,
"learning_rate": 9.96322671167364e-05,
"loss": 2.4239,
"step": 4990
},
{
"epoch": 0.04,
"learning_rate": 9.963151564780256e-05,
"loss": 2.4351,
"step": 4995
},
{
"epoch": 0.04,
"learning_rate": 9.963076341467165e-05,
"loss": 2.3344,
"step": 5000
},
{
"epoch": 0.04,
"learning_rate": 9.963001041735531e-05,
"loss": 2.4623,
"step": 5005
},
{
"epoch": 0.04,
"learning_rate": 9.962925665586509e-05,
"loss": 2.4029,
"step": 5010
},
{
"epoch": 0.04,
"learning_rate": 9.962850213021262e-05,
"loss": 2.4518,
"step": 5015
},
{
"epoch": 0.04,
"learning_rate": 9.96277468404095e-05,
"loss": 2.399,
"step": 5020
},
{
"epoch": 0.04,
"learning_rate": 9.962699078646735e-05,
"loss": 2.3857,
"step": 5025
},
{
"epoch": 0.04,
"learning_rate": 9.962623396839785e-05,
"loss": 2.39,
"step": 5030
},
{
"epoch": 0.04,
"learning_rate": 9.962547638621262e-05,
"loss": 2.4257,
"step": 5035
},
{
"epoch": 0.04,
"learning_rate": 9.962471803992334e-05,
"loss": 2.462,
"step": 5040
},
{
"epoch": 0.04,
"learning_rate": 9.962395892954168e-05,
"loss": 2.445,
"step": 5045
},
{
"epoch": 0.04,
"learning_rate": 9.962319905507934e-05,
"loss": 2.4719,
"step": 5050
},
{
"epoch": 0.04,
"learning_rate": 9.9622438416548e-05,
"loss": 2.4196,
"step": 5055
},
{
"epoch": 0.04,
"learning_rate": 9.962167701395937e-05,
"loss": 2.3855,
"step": 5060
},
{
"epoch": 0.04,
"learning_rate": 9.962091484732521e-05,
"loss": 2.4471,
"step": 5065
},
{
"epoch": 0.04,
"learning_rate": 9.962015191665722e-05,
"loss": 2.3732,
"step": 5070
},
{
"epoch": 0.04,
"learning_rate": 9.961938822196716e-05,
"loss": 2.4251,
"step": 5075
},
{
"epoch": 0.04,
"learning_rate": 9.961862376326678e-05,
"loss": 2.4173,
"step": 5080
},
{
"epoch": 0.04,
"learning_rate": 9.961785854056788e-05,
"loss": 2.4244,
"step": 5085
},
{
"epoch": 0.04,
"learning_rate": 9.961709255388221e-05,
"loss": 2.4758,
"step": 5090
},
{
"epoch": 0.04,
"learning_rate": 9.961632580322156e-05,
"loss": 2.4353,
"step": 5095
},
{
"epoch": 0.04,
"learning_rate": 9.961555828859777e-05,
"loss": 2.4143,
"step": 5100
},
{
"epoch": 0.04,
"learning_rate": 9.961479001002264e-05,
"loss": 2.4233,
"step": 5105
},
{
"epoch": 0.04,
"learning_rate": 9.961402096750799e-05,
"loss": 2.4743,
"step": 5110
},
{
"epoch": 0.04,
"learning_rate": 9.961325116106566e-05,
"loss": 2.4457,
"step": 5115
},
{
"epoch": 0.04,
"learning_rate": 9.961248059070751e-05,
"loss": 2.3348,
"step": 5120
},
{
"epoch": 0.04,
"learning_rate": 9.961170925644542e-05,
"loss": 2.3566,
"step": 5125
},
{
"epoch": 0.04,
"learning_rate": 9.961093715829127e-05,
"loss": 2.4219,
"step": 5130
},
{
"epoch": 0.04,
"learning_rate": 9.961016429625691e-05,
"loss": 2.4318,
"step": 5135
},
{
"epoch": 0.04,
"learning_rate": 9.960939067035426e-05,
"loss": 2.4203,
"step": 5140
},
{
"epoch": 0.04,
"learning_rate": 9.960861628059524e-05,
"loss": 2.5147,
"step": 5145
},
{
"epoch": 0.04,
"learning_rate": 9.960784112699177e-05,
"loss": 2.3704,
"step": 5150
},
{
"epoch": 0.04,
"learning_rate": 9.960706520955578e-05,
"loss": 2.3697,
"step": 5155
},
{
"epoch": 0.04,
"learning_rate": 9.960628852829923e-05,
"loss": 2.4674,
"step": 5160
},
{
"epoch": 0.04,
"learning_rate": 9.960551108323406e-05,
"loss": 2.4262,
"step": 5165
},
{
"epoch": 0.04,
"learning_rate": 9.960473287437224e-05,
"loss": 2.4403,
"step": 5170
},
{
"epoch": 0.04,
"learning_rate": 9.960395390172579e-05,
"loss": 2.4003,
"step": 5175
},
{
"epoch": 0.04,
"learning_rate": 9.960317416530665e-05,
"loss": 2.3976,
"step": 5180
},
{
"epoch": 0.04,
"learning_rate": 9.960239366512686e-05,
"loss": 2.4275,
"step": 5185
},
{
"epoch": 0.04,
"learning_rate": 9.960161240119843e-05,
"loss": 2.3274,
"step": 5190
},
{
"epoch": 0.04,
"learning_rate": 9.96008303735334e-05,
"loss": 2.4106,
"step": 5195
},
{
"epoch": 0.04,
"learning_rate": 9.960004758214379e-05,
"loss": 2.3045,
"step": 5200
},
{
"epoch": 0.04,
"learning_rate": 9.959926402704165e-05,
"loss": 2.4603,
"step": 5205
},
{
"epoch": 0.04,
"learning_rate": 9.959847970823908e-05,
"loss": 2.4201,
"step": 5210
},
{
"epoch": 0.04,
"learning_rate": 9.959769462574813e-05,
"loss": 2.4302,
"step": 5215
},
{
"epoch": 0.04,
"learning_rate": 9.959690877958088e-05,
"loss": 2.3552,
"step": 5220
},
{
"epoch": 0.04,
"learning_rate": 9.959612216974946e-05,
"loss": 2.4144,
"step": 5225
},
{
"epoch": 0.04,
"learning_rate": 9.959533479626596e-05,
"loss": 2.3972,
"step": 5230
},
{
"epoch": 0.04,
"learning_rate": 9.95945466591425e-05,
"loss": 2.4456,
"step": 5235
},
{
"epoch": 0.04,
"learning_rate": 9.959375775839124e-05,
"loss": 2.4972,
"step": 5240
},
{
"epoch": 0.04,
"learning_rate": 9.95929680940243e-05,
"loss": 2.3721,
"step": 5245
},
{
"epoch": 0.04,
"learning_rate": 9.959217766605384e-05,
"loss": 2.4292,
"step": 5250
},
{
"epoch": 0.04,
"learning_rate": 9.959138647449205e-05,
"loss": 2.4042,
"step": 5255
},
{
"epoch": 0.04,
"learning_rate": 9.95905945193511e-05,
"loss": 2.3876,
"step": 5260
},
{
"epoch": 0.04,
"learning_rate": 9.958980180064318e-05,
"loss": 2.4962,
"step": 5265
},
{
"epoch": 0.04,
"learning_rate": 9.958900831838051e-05,
"loss": 2.399,
"step": 5270
},
{
"epoch": 0.04,
"learning_rate": 9.958821407257531e-05,
"loss": 2.4187,
"step": 5275
},
{
"epoch": 0.04,
"learning_rate": 9.958741906323978e-05,
"loss": 2.4635,
"step": 5280
},
{
"epoch": 0.04,
"learning_rate": 9.95866232903862e-05,
"loss": 2.404,
"step": 5285
},
{
"epoch": 0.04,
"learning_rate": 9.958582675402677e-05,
"loss": 2.405,
"step": 5290
},
{
"epoch": 0.04,
"learning_rate": 9.958502945417382e-05,
"loss": 2.4051,
"step": 5295
},
{
"epoch": 0.04,
"learning_rate": 9.958423139083958e-05,
"loss": 2.4178,
"step": 5300
},
{
"epoch": 0.04,
"learning_rate": 9.958343256403635e-05,
"loss": 2.4313,
"step": 5305
},
{
"epoch": 0.04,
"learning_rate": 9.958263297377643e-05,
"loss": 2.4099,
"step": 5310
},
{
"epoch": 0.04,
"learning_rate": 9.958183262007215e-05,
"loss": 2.3535,
"step": 5315
},
{
"epoch": 0.04,
"learning_rate": 9.95810315029358e-05,
"loss": 2.4389,
"step": 5320
},
{
"epoch": 0.04,
"learning_rate": 9.958022962237974e-05,
"loss": 2.3584,
"step": 5325
},
{
"epoch": 0.04,
"learning_rate": 9.957942697841632e-05,
"loss": 2.4267,
"step": 5330
},
{
"epoch": 0.04,
"learning_rate": 9.957862357105785e-05,
"loss": 2.3984,
"step": 5335
},
{
"epoch": 0.04,
"learning_rate": 9.957781940031676e-05,
"loss": 2.4944,
"step": 5340
},
{
"epoch": 0.04,
"learning_rate": 9.957701446620543e-05,
"loss": 2.4801,
"step": 5345
},
{
"epoch": 0.04,
"learning_rate": 9.957620876873621e-05,
"loss": 2.3349,
"step": 5350
},
{
"epoch": 0.04,
"learning_rate": 9.957540230792152e-05,
"loss": 2.4124,
"step": 5355
},
{
"epoch": 0.04,
"learning_rate": 9.957459508377381e-05,
"loss": 2.4176,
"step": 5360
},
{
"epoch": 0.04,
"learning_rate": 9.957378709630548e-05,
"loss": 2.4228,
"step": 5365
},
{
"epoch": 0.04,
"learning_rate": 9.957297834552898e-05,
"loss": 2.3781,
"step": 5370
},
{
"epoch": 0.04,
"learning_rate": 9.957216883145675e-05,
"loss": 2.3298,
"step": 5375
},
{
"epoch": 0.04,
"learning_rate": 9.957135855410127e-05,
"loss": 2.3408,
"step": 5380
},
{
"epoch": 0.04,
"learning_rate": 9.957054751347501e-05,
"loss": 2.4944,
"step": 5385
},
{
"epoch": 0.04,
"learning_rate": 9.956973570959045e-05,
"loss": 2.3879,
"step": 5390
},
{
"epoch": 0.04,
"learning_rate": 9.956892314246012e-05,
"loss": 2.4223,
"step": 5395
},
{
"epoch": 0.04,
"learning_rate": 9.956810981209647e-05,
"loss": 2.4272,
"step": 5400
},
{
"epoch": 0.04,
"learning_rate": 9.956729571851211e-05,
"loss": 2.488,
"step": 5405
},
{
"epoch": 0.04,
"learning_rate": 9.956648086171949e-05,
"loss": 2.4591,
"step": 5410
},
{
"epoch": 0.04,
"learning_rate": 9.956566524173121e-05,
"loss": 2.3961,
"step": 5415
},
{
"epoch": 0.04,
"learning_rate": 9.956484885855982e-05,
"loss": 2.3531,
"step": 5420
},
{
"epoch": 0.04,
"learning_rate": 9.956403171221786e-05,
"loss": 2.3696,
"step": 5425
},
{
"epoch": 0.04,
"learning_rate": 9.956321380271796e-05,
"loss": 2.4035,
"step": 5430
},
{
"epoch": 0.04,
"learning_rate": 9.956239513007268e-05,
"loss": 2.3872,
"step": 5435
},
{
"epoch": 0.04,
"learning_rate": 9.956157569429464e-05,
"loss": 2.3922,
"step": 5440
},
{
"epoch": 0.04,
"learning_rate": 9.956075549539643e-05,
"loss": 2.3719,
"step": 5445
},
{
"epoch": 0.04,
"learning_rate": 9.955993453339072e-05,
"loss": 2.41,
"step": 5450
},
{
"epoch": 0.04,
"learning_rate": 9.955911280829011e-05,
"loss": 2.4428,
"step": 5455
},
{
"epoch": 0.04,
"learning_rate": 9.95582903201073e-05,
"loss": 2.4003,
"step": 5460
},
{
"epoch": 0.04,
"learning_rate": 9.955746706885492e-05,
"loss": 2.3758,
"step": 5465
},
{
"epoch": 0.04,
"learning_rate": 9.955664305454564e-05,
"loss": 2.4299,
"step": 5470
},
{
"epoch": 0.04,
"learning_rate": 9.955581827719216e-05,
"loss": 2.4862,
"step": 5475
},
{
"epoch": 0.04,
"learning_rate": 9.955499273680719e-05,
"loss": 2.3573,
"step": 5480
},
{
"epoch": 0.04,
"learning_rate": 9.955416643340343e-05,
"loss": 2.3851,
"step": 5485
},
{
"epoch": 0.04,
"learning_rate": 9.95533393669936e-05,
"loss": 2.3725,
"step": 5490
},
{
"epoch": 0.04,
"learning_rate": 9.955251153759044e-05,
"loss": 2.4214,
"step": 5495
},
{
"epoch": 0.04,
"learning_rate": 9.955168294520672e-05,
"loss": 2.505,
"step": 5500
},
{
"epoch": 0.04,
"learning_rate": 9.955085358985514e-05,
"loss": 2.3628,
"step": 5505
},
{
"epoch": 0.04,
"learning_rate": 9.955002347154851e-05,
"loss": 2.4788,
"step": 5510
},
{
"epoch": 0.04,
"learning_rate": 9.954919259029962e-05,
"loss": 2.4296,
"step": 5515
},
{
"epoch": 0.04,
"learning_rate": 9.954836094612123e-05,
"loss": 2.4137,
"step": 5520
},
{
"epoch": 0.04,
"learning_rate": 9.954752853902619e-05,
"loss": 2.3694,
"step": 5525
},
{
"epoch": 0.04,
"learning_rate": 9.954669536902728e-05,
"loss": 2.3556,
"step": 5530
},
{
"epoch": 0.04,
"learning_rate": 9.954586143613733e-05,
"loss": 2.4449,
"step": 5535
},
{
"epoch": 0.04,
"learning_rate": 9.954502674036919e-05,
"loss": 2.3977,
"step": 5540
},
{
"epoch": 0.04,
"learning_rate": 9.954419128173572e-05,
"loss": 2.3546,
"step": 5545
},
{
"epoch": 0.04,
"learning_rate": 9.954335506024978e-05,
"loss": 2.3796,
"step": 5550
},
{
"epoch": 0.04,
"learning_rate": 9.954251807592423e-05,
"loss": 2.326,
"step": 5555
},
{
"epoch": 0.04,
"learning_rate": 9.954168032877198e-05,
"loss": 2.3561,
"step": 5560
},
{
"epoch": 0.04,
"learning_rate": 9.95408418188059e-05,
"loss": 2.3969,
"step": 5565
},
{
"epoch": 0.04,
"learning_rate": 9.954000254603894e-05,
"loss": 2.3707,
"step": 5570
},
{
"epoch": 0.04,
"learning_rate": 9.9539162510484e-05,
"loss": 2.3618,
"step": 5575
},
{
"epoch": 0.04,
"learning_rate": 9.953832171215401e-05,
"loss": 2.4498,
"step": 5580
},
{
"epoch": 0.04,
"learning_rate": 9.953748015106193e-05,
"loss": 2.4191,
"step": 5585
},
{
"epoch": 0.04,
"learning_rate": 9.953663782722069e-05,
"loss": 2.4129,
"step": 5590
},
{
"epoch": 0.04,
"learning_rate": 9.953579474064329e-05,
"loss": 2.3424,
"step": 5595
},
{
"epoch": 0.04,
"learning_rate": 9.95349508913427e-05,
"loss": 2.3091,
"step": 5600
},
{
"epoch": 0.04,
"learning_rate": 9.953410627933193e-05,
"loss": 2.5092,
"step": 5605
},
{
"epoch": 0.04,
"learning_rate": 9.953326090462397e-05,
"loss": 2.4146,
"step": 5610
},
{
"epoch": 0.04,
"learning_rate": 9.953241476723181e-05,
"loss": 2.3904,
"step": 5615
},
{
"epoch": 0.04,
"learning_rate": 9.953156786716851e-05,
"loss": 2.3501,
"step": 5620
},
{
"epoch": 0.04,
"learning_rate": 9.953072020444711e-05,
"loss": 2.3537,
"step": 5625
},
{
"epoch": 0.04,
"learning_rate": 9.952987177908066e-05,
"loss": 2.3597,
"step": 5630
},
{
"epoch": 0.04,
"learning_rate": 9.95290225910822e-05,
"loss": 2.4353,
"step": 5635
},
{
"epoch": 0.04,
"learning_rate": 9.952817264046486e-05,
"loss": 2.3653,
"step": 5640
},
{
"epoch": 0.04,
"learning_rate": 9.952732192724167e-05,
"loss": 2.4377,
"step": 5645
},
{
"epoch": 0.04,
"learning_rate": 9.952647045142576e-05,
"loss": 2.4304,
"step": 5650
},
{
"epoch": 0.04,
"learning_rate": 9.952561821303021e-05,
"loss": 2.3757,
"step": 5655
},
{
"epoch": 0.04,
"learning_rate": 9.952476521206818e-05,
"loss": 2.3984,
"step": 5660
},
{
"epoch": 0.04,
"learning_rate": 9.952391144855278e-05,
"loss": 2.4252,
"step": 5665
},
{
"epoch": 0.04,
"learning_rate": 9.952305692249717e-05,
"loss": 2.4178,
"step": 5670
},
{
"epoch": 0.04,
"learning_rate": 9.952220163391451e-05,
"loss": 2.3294,
"step": 5675
},
{
"epoch": 0.04,
"learning_rate": 9.952134558281793e-05,
"loss": 2.4604,
"step": 5680
},
{
"epoch": 0.04,
"learning_rate": 9.952048876922066e-05,
"loss": 2.3853,
"step": 5685
},
{
"epoch": 0.04,
"learning_rate": 9.951963119313588e-05,
"loss": 2.3068,
"step": 5690
},
{
"epoch": 0.04,
"learning_rate": 9.951877285457678e-05,
"loss": 2.4923,
"step": 5695
},
{
"epoch": 0.04,
"learning_rate": 9.951791375355659e-05,
"loss": 2.4028,
"step": 5700
},
{
"epoch": 0.05,
"learning_rate": 9.951705389008852e-05,
"loss": 2.3305,
"step": 5705
},
{
"epoch": 0.05,
"learning_rate": 9.951619326418582e-05,
"loss": 2.4462,
"step": 5710
},
{
"epoch": 0.05,
"learning_rate": 9.951533187586178e-05,
"loss": 2.429,
"step": 5715
},
{
"epoch": 0.05,
"learning_rate": 9.951446972512958e-05,
"loss": 2.4138,
"step": 5720
},
{
"epoch": 0.05,
"learning_rate": 9.951360681200256e-05,
"loss": 2.4755,
"step": 5725
},
{
"epoch": 0.05,
"learning_rate": 9.9512743136494e-05,
"loss": 2.4388,
"step": 5730
},
{
"epoch": 0.05,
"learning_rate": 9.951187869861715e-05,
"loss": 2.4259,
"step": 5735
},
{
"epoch": 0.05,
"learning_rate": 9.951101349838538e-05,
"loss": 2.368,
"step": 5740
},
{
"epoch": 0.05,
"learning_rate": 9.9510147535812e-05,
"loss": 2.4763,
"step": 5745
},
{
"epoch": 0.05,
"learning_rate": 9.950928081091032e-05,
"loss": 2.3205,
"step": 5750
},
{
"epoch": 0.05,
"learning_rate": 9.95084133236937e-05,
"loss": 2.4175,
"step": 5755
},
{
"epoch": 0.05,
"learning_rate": 9.950754507417547e-05,
"loss": 2.35,
"step": 5760
},
{
"epoch": 0.05,
"learning_rate": 9.950667606236906e-05,
"loss": 2.4276,
"step": 5765
},
{
"epoch": 0.05,
"learning_rate": 9.95058062882878e-05,
"loss": 2.3957,
"step": 5770
},
{
"epoch": 0.05,
"learning_rate": 9.950493575194509e-05,
"loss": 2.3747,
"step": 5775
},
{
"epoch": 0.05,
"learning_rate": 9.950406445335434e-05,
"loss": 2.2962,
"step": 5780
},
{
"epoch": 0.05,
"learning_rate": 9.950319239252896e-05,
"loss": 2.3804,
"step": 5785
},
{
"epoch": 0.05,
"learning_rate": 9.95023195694824e-05,
"loss": 2.3853,
"step": 5790
},
{
"epoch": 0.05,
"learning_rate": 9.950144598422809e-05,
"loss": 2.4188,
"step": 5795
},
{
"epoch": 0.05,
"learning_rate": 9.950057163677944e-05,
"loss": 2.3756,
"step": 5800
},
{
"epoch": 0.05,
"learning_rate": 9.949969652714997e-05,
"loss": 2.4255,
"step": 5805
},
{
"epoch": 0.05,
"learning_rate": 9.949882065535312e-05,
"loss": 2.361,
"step": 5810
},
{
"epoch": 0.05,
"learning_rate": 9.949794402140239e-05,
"loss": 2.3949,
"step": 5815
},
{
"epoch": 0.05,
"learning_rate": 9.949706662531127e-05,
"loss": 2.32,
"step": 5820
},
{
"epoch": 0.05,
"learning_rate": 9.949618846709327e-05,
"loss": 2.3942,
"step": 5825
},
{
"epoch": 0.05,
"learning_rate": 9.949530954676194e-05,
"loss": 2.3306,
"step": 5830
},
{
"epoch": 0.05,
"learning_rate": 9.949442986433076e-05,
"loss": 2.3382,
"step": 5835
},
{
"epoch": 0.05,
"learning_rate": 9.949354941981332e-05,
"loss": 2.3445,
"step": 5840
},
{
"epoch": 0.05,
"learning_rate": 9.949266821322315e-05,
"loss": 2.432,
"step": 5845
},
{
"epoch": 0.05,
"learning_rate": 9.949178624457384e-05,
"loss": 2.378,
"step": 5850
},
{
"epoch": 0.05,
"learning_rate": 9.949090351387895e-05,
"loss": 2.4171,
"step": 5855
},
{
"epoch": 0.05,
"learning_rate": 9.949002002115208e-05,
"loss": 2.3872,
"step": 5860
},
{
"epoch": 0.05,
"learning_rate": 9.948913576640683e-05,
"loss": 2.4019,
"step": 5865
},
{
"epoch": 0.05,
"learning_rate": 9.948825074965682e-05,
"loss": 2.4126,
"step": 5870
},
{
"epoch": 0.05,
"learning_rate": 9.948736497091569e-05,
"loss": 2.4297,
"step": 5875
},
{
"epoch": 0.05,
"learning_rate": 9.948647843019705e-05,
"loss": 2.3444,
"step": 5880
},
{
"epoch": 0.05,
"learning_rate": 9.948559112751456e-05,
"loss": 2.3486,
"step": 5885
},
{
"epoch": 0.05,
"learning_rate": 9.948470306288189e-05,
"loss": 2.3978,
"step": 5890
},
{
"epoch": 0.05,
"learning_rate": 9.948381423631273e-05,
"loss": 2.3817,
"step": 5895
},
{
"epoch": 0.05,
"learning_rate": 9.948292464782072e-05,
"loss": 2.3949,
"step": 5900
},
{
"epoch": 0.05,
"learning_rate": 9.948203429741961e-05,
"loss": 2.4012,
"step": 5905
},
{
"epoch": 0.05,
"learning_rate": 9.948114318512307e-05,
"loss": 2.3917,
"step": 5910
},
{
"epoch": 0.05,
"learning_rate": 9.948025131094482e-05,
"loss": 2.4394,
"step": 5915
},
{
"epoch": 0.05,
"learning_rate": 9.947935867489862e-05,
"loss": 2.4485,
"step": 5920
},
{
"epoch": 0.05,
"learning_rate": 9.94784652769982e-05,
"loss": 2.388,
"step": 5925
},
{
"epoch": 0.05,
"learning_rate": 9.947757111725732e-05,
"loss": 2.3887,
"step": 5930
},
{
"epoch": 0.05,
"learning_rate": 9.947667619568974e-05,
"loss": 2.3615,
"step": 5935
},
{
"epoch": 0.05,
"learning_rate": 9.947578051230925e-05,
"loss": 2.3804,
"step": 5940
},
{
"epoch": 0.05,
"learning_rate": 9.947488406712962e-05,
"loss": 2.4414,
"step": 5945
},
{
"epoch": 0.05,
"learning_rate": 9.947398686016469e-05,
"loss": 2.5399,
"step": 5950
},
{
"epoch": 0.05,
"learning_rate": 9.947308889142824e-05,
"loss": 2.3809,
"step": 5955
},
{
"epoch": 0.05,
"learning_rate": 9.94721901609341e-05,
"loss": 2.4101,
"step": 5960
},
{
"epoch": 0.05,
"learning_rate": 9.947129066869614e-05,
"loss": 2.402,
"step": 5965
},
{
"epoch": 0.05,
"learning_rate": 9.947039041472817e-05,
"loss": 2.4402,
"step": 5970
},
{
"epoch": 0.05,
"learning_rate": 9.946948939904407e-05,
"loss": 2.4013,
"step": 5975
},
{
"epoch": 0.05,
"learning_rate": 9.946858762165772e-05,
"loss": 2.4476,
"step": 5980
},
{
"epoch": 0.05,
"learning_rate": 9.946768508258299e-05,
"loss": 2.3988,
"step": 5985
},
{
"epoch": 0.05,
"learning_rate": 9.946678178183377e-05,
"loss": 2.3808,
"step": 5990
},
{
"epoch": 0.05,
"learning_rate": 9.946587771942401e-05,
"loss": 2.4968,
"step": 5995
},
{
"epoch": 0.05,
"learning_rate": 9.946497289536758e-05,
"loss": 2.32,
"step": 6000
},
{
"epoch": 0.05,
"learning_rate": 9.946406730967846e-05,
"loss": 2.4001,
"step": 6005
},
{
"epoch": 0.05,
"learning_rate": 9.946316096237054e-05,
"loss": 2.4336,
"step": 6010
},
{
"epoch": 0.05,
"learning_rate": 9.946225385345782e-05,
"loss": 2.4105,
"step": 6015
},
{
"epoch": 0.05,
"learning_rate": 9.946134598295424e-05,
"loss": 2.4,
"step": 6020
},
{
"epoch": 0.05,
"learning_rate": 9.94604373508738e-05,
"loss": 2.4254,
"step": 6025
},
{
"epoch": 0.05,
"learning_rate": 9.945952795723048e-05,
"loss": 2.4307,
"step": 6030
},
{
"epoch": 0.05,
"learning_rate": 9.945861780203828e-05,
"loss": 2.449,
"step": 6035
},
{
"epoch": 0.05,
"learning_rate": 9.945770688531122e-05,
"loss": 2.4111,
"step": 6040
},
{
"epoch": 0.05,
"learning_rate": 9.945679520706332e-05,
"loss": 2.4319,
"step": 6045
},
{
"epoch": 0.05,
"learning_rate": 9.945588276730862e-05,
"loss": 2.4282,
"step": 6050
},
{
"epoch": 0.05,
"learning_rate": 9.945496956606115e-05,
"loss": 2.3619,
"step": 6055
},
{
"epoch": 0.05,
"learning_rate": 9.945405560333502e-05,
"loss": 2.3617,
"step": 6060
},
{
"epoch": 0.05,
"learning_rate": 9.945314087914426e-05,
"loss": 2.3018,
"step": 6065
},
{
"epoch": 0.05,
"learning_rate": 9.945222539350296e-05,
"loss": 2.4243,
"step": 6070
},
{
"epoch": 0.05,
"learning_rate": 9.945130914642524e-05,
"loss": 2.3079,
"step": 6075
},
{
"epoch": 0.05,
"learning_rate": 9.945039213792518e-05,
"loss": 2.4408,
"step": 6080
},
{
"epoch": 0.05,
"learning_rate": 9.94494743680169e-05,
"loss": 2.4326,
"step": 6085
},
{
"epoch": 0.05,
"learning_rate": 9.944855583671455e-05,
"loss": 2.3574,
"step": 6090
},
{
"epoch": 0.05,
"learning_rate": 9.944763654403226e-05,
"loss": 2.3601,
"step": 6095
},
{
"epoch": 0.05,
"learning_rate": 9.94467164899842e-05,
"loss": 2.3834,
"step": 6100
},
{
"epoch": 0.05,
"learning_rate": 9.944579567458452e-05,
"loss": 2.3788,
"step": 6105
},
{
"epoch": 0.05,
"learning_rate": 9.944487409784741e-05,
"loss": 2.4031,
"step": 6110
},
{
"epoch": 0.05,
"learning_rate": 9.944395175978705e-05,
"loss": 2.3998,
"step": 6115
},
{
"epoch": 0.05,
"learning_rate": 9.944302866041764e-05,
"loss": 2.4016,
"step": 6120
},
{
"epoch": 0.05,
"learning_rate": 9.944210479975342e-05,
"loss": 2.3591,
"step": 6125
},
{
"epoch": 0.05,
"learning_rate": 9.944118017780857e-05,
"loss": 2.3345,
"step": 6130
},
{
"epoch": 0.05,
"learning_rate": 9.944025479459735e-05,
"loss": 2.3499,
"step": 6135
},
{
"epoch": 0.05,
"learning_rate": 9.943932865013404e-05,
"loss": 2.426,
"step": 6140
},
{
"epoch": 0.05,
"learning_rate": 9.943840174443284e-05,
"loss": 2.3574,
"step": 6145
},
{
"epoch": 0.05,
"learning_rate": 9.943747407750805e-05,
"loss": 2.3967,
"step": 6150
},
{
"epoch": 0.05,
"learning_rate": 9.943654564937397e-05,
"loss": 2.3874,
"step": 6155
},
{
"epoch": 0.05,
"learning_rate": 9.943561646004488e-05,
"loss": 2.3351,
"step": 6160
},
{
"epoch": 0.05,
"learning_rate": 9.943468650953508e-05,
"loss": 2.3865,
"step": 6165
},
{
"epoch": 0.05,
"learning_rate": 9.943375579785889e-05,
"loss": 2.33,
"step": 6170
},
{
"epoch": 0.05,
"learning_rate": 9.943282432503067e-05,
"loss": 2.3743,
"step": 6175
},
{
"epoch": 0.05,
"learning_rate": 9.943189209106472e-05,
"loss": 2.3752,
"step": 6180
},
{
"epoch": 0.05,
"learning_rate": 9.943095909597541e-05,
"loss": 2.3517,
"step": 6185
},
{
"epoch": 0.05,
"learning_rate": 9.943002533977713e-05,
"loss": 2.4089,
"step": 6190
},
{
"epoch": 0.05,
"learning_rate": 9.942909082248423e-05,
"loss": 2.3981,
"step": 6195
},
{
"epoch": 0.05,
"learning_rate": 9.942815554411109e-05,
"loss": 2.3771,
"step": 6200
},
{
"epoch": 0.05,
"learning_rate": 9.942721950467214e-05,
"loss": 2.2684,
"step": 6205
},
{
"epoch": 0.05,
"learning_rate": 9.942628270418177e-05,
"loss": 2.3326,
"step": 6210
},
{
"epoch": 0.05,
"learning_rate": 9.942534514265442e-05,
"loss": 2.4671,
"step": 6215
},
{
"epoch": 0.05,
"learning_rate": 9.942440682010452e-05,
"loss": 2.4549,
"step": 6220
},
{
"epoch": 0.05,
"learning_rate": 9.942346773654652e-05,
"loss": 2.4365,
"step": 6225
},
{
"epoch": 0.05,
"learning_rate": 9.942252789199487e-05,
"loss": 2.347,
"step": 6230
},
{
"epoch": 0.05,
"learning_rate": 9.942158728646406e-05,
"loss": 2.3408,
"step": 6235
},
{
"epoch": 0.05,
"learning_rate": 9.942064591996855e-05,
"loss": 2.3624,
"step": 6240
},
{
"epoch": 0.05,
"learning_rate": 9.941970379252284e-05,
"loss": 2.3385,
"step": 6245
},
{
"epoch": 0.05,
"learning_rate": 9.941876090414146e-05,
"loss": 2.3938,
"step": 6250
},
{
"epoch": 0.05,
"learning_rate": 9.941781725483889e-05,
"loss": 2.2893,
"step": 6255
},
{
"epoch": 0.05,
"learning_rate": 9.941687284462969e-05,
"loss": 2.4918,
"step": 6260
},
{
"epoch": 0.05,
"learning_rate": 9.941592767352839e-05,
"loss": 2.3324,
"step": 6265
},
{
"epoch": 0.05,
"learning_rate": 9.941498174154954e-05,
"loss": 2.4773,
"step": 6270
},
{
"epoch": 0.05,
"learning_rate": 9.94140350487077e-05,
"loss": 2.3904,
"step": 6275
},
{
"epoch": 0.05,
"learning_rate": 9.941308759501747e-05,
"loss": 2.3859,
"step": 6280
},
{
"epoch": 0.05,
"learning_rate": 9.941213938049342e-05,
"loss": 2.3444,
"step": 6285
},
{
"epoch": 0.05,
"learning_rate": 9.941119040515017e-05,
"loss": 2.5005,
"step": 6290
},
{
"epoch": 0.05,
"learning_rate": 9.941024066900229e-05,
"loss": 2.3875,
"step": 6295
},
{
"epoch": 0.05,
"learning_rate": 9.940929017206445e-05,
"loss": 2.4087,
"step": 6300
},
{
"epoch": 0.05,
"learning_rate": 9.940833891435124e-05,
"loss": 2.3135,
"step": 6305
},
{
"epoch": 0.05,
"learning_rate": 9.940738689587735e-05,
"loss": 2.3755,
"step": 6310
},
{
"epoch": 0.05,
"learning_rate": 9.940643411665742e-05,
"loss": 2.2683,
"step": 6315
},
{
"epoch": 0.05,
"learning_rate": 9.940548057670612e-05,
"loss": 2.4092,
"step": 6320
},
{
"epoch": 0.05,
"learning_rate": 9.940452627603813e-05,
"loss": 2.3856,
"step": 6325
},
{
"epoch": 0.05,
"learning_rate": 9.940357121466813e-05,
"loss": 2.2741,
"step": 6330
},
{
"epoch": 0.05,
"learning_rate": 9.940261539261086e-05,
"loss": 2.3542,
"step": 6335
},
{
"epoch": 0.05,
"learning_rate": 9.940165880988102e-05,
"loss": 2.3049,
"step": 6340
},
{
"epoch": 0.05,
"learning_rate": 9.940070146649333e-05,
"loss": 2.2339,
"step": 6345
},
{
"epoch": 0.05,
"learning_rate": 9.939974336246254e-05,
"loss": 2.3621,
"step": 6350
},
{
"epoch": 0.05,
"learning_rate": 9.93987844978034e-05,
"loss": 2.436,
"step": 6355
},
{
"epoch": 0.05,
"learning_rate": 9.939782487253067e-05,
"loss": 2.3608,
"step": 6360
},
{
"epoch": 0.05,
"learning_rate": 9.939686448665915e-05,
"loss": 2.396,
"step": 6365
},
{
"epoch": 0.05,
"learning_rate": 9.939590334020359e-05,
"loss": 2.3469,
"step": 6370
},
{
"epoch": 0.05,
"learning_rate": 9.939494143317881e-05,
"loss": 2.4006,
"step": 6375
},
{
"epoch": 0.05,
"learning_rate": 9.939397876559962e-05,
"loss": 2.4229,
"step": 6380
},
{
"epoch": 0.05,
"learning_rate": 9.939301533748084e-05,
"loss": 2.3629,
"step": 6385
},
{
"epoch": 0.05,
"learning_rate": 9.939205114883731e-05,
"loss": 2.3331,
"step": 6390
},
{
"epoch": 0.05,
"learning_rate": 9.939108619968386e-05,
"loss": 2.3244,
"step": 6395
},
{
"epoch": 0.05,
"learning_rate": 9.939012049003537e-05,
"loss": 2.405,
"step": 6400
},
{
"epoch": 0.05,
"learning_rate": 9.93891540199067e-05,
"loss": 2.3989,
"step": 6405
},
{
"epoch": 0.05,
"learning_rate": 9.938818678931271e-05,
"loss": 2.3753,
"step": 6410
},
{
"epoch": 0.05,
"learning_rate": 9.938721879826832e-05,
"loss": 2.451,
"step": 6415
},
{
"epoch": 0.05,
"learning_rate": 9.938625004678844e-05,
"loss": 2.3953,
"step": 6420
},
{
"epoch": 0.05,
"learning_rate": 9.938528053488796e-05,
"loss": 2.3559,
"step": 6425
},
{
"epoch": 0.05,
"learning_rate": 9.938431026258182e-05,
"loss": 2.3978,
"step": 6430
},
{
"epoch": 0.05,
"learning_rate": 9.938333922988497e-05,
"loss": 2.3751,
"step": 6435
},
{
"epoch": 0.05,
"learning_rate": 9.938236743681235e-05,
"loss": 2.451,
"step": 6440
},
{
"epoch": 0.05,
"learning_rate": 9.938139488337891e-05,
"loss": 2.4495,
"step": 6445
},
{
"epoch": 0.05,
"learning_rate": 9.938042156959966e-05,
"loss": 2.4667,
"step": 6450
},
{
"epoch": 0.05,
"learning_rate": 9.937944749548956e-05,
"loss": 2.452,
"step": 6455
},
{
"epoch": 0.05,
"learning_rate": 9.937847266106359e-05,
"loss": 2.4043,
"step": 6460
},
{
"epoch": 0.05,
"learning_rate": 9.937749706633682e-05,
"loss": 2.3558,
"step": 6465
},
{
"epoch": 0.05,
"learning_rate": 9.937652071132422e-05,
"loss": 2.3404,
"step": 6470
},
{
"epoch": 0.05,
"learning_rate": 9.937554359604084e-05,
"loss": 2.3435,
"step": 6475
},
{
"epoch": 0.05,
"learning_rate": 9.937456572050172e-05,
"loss": 2.3657,
"step": 6480
},
{
"epoch": 0.05,
"learning_rate": 9.937358708472193e-05,
"loss": 2.3731,
"step": 6485
},
{
"epoch": 0.05,
"learning_rate": 9.937260768871654e-05,
"loss": 2.3385,
"step": 6490
},
{
"epoch": 0.05,
"learning_rate": 9.93716275325006e-05,
"loss": 2.3386,
"step": 6495
},
{
"epoch": 0.05,
"learning_rate": 9.937064661608924e-05,
"loss": 2.3097,
"step": 6500
},
{
"epoch": 0.05,
"learning_rate": 9.936966493949752e-05,
"loss": 2.3703,
"step": 6505
},
{
"epoch": 0.05,
"learning_rate": 9.936868250274059e-05,
"loss": 2.4015,
"step": 6510
},
{
"epoch": 0.05,
"learning_rate": 9.936769930583359e-05,
"loss": 2.3205,
"step": 6515
},
{
"epoch": 0.05,
"learning_rate": 9.93667153487916e-05,
"loss": 2.4496,
"step": 6520
},
{
"epoch": 0.05,
"learning_rate": 9.936573063162984e-05,
"loss": 2.2827,
"step": 6525
},
{
"epoch": 0.05,
"learning_rate": 9.936474515436342e-05,
"loss": 2.3835,
"step": 6530
},
{
"epoch": 0.05,
"learning_rate": 9.936375891700753e-05,
"loss": 2.4261,
"step": 6535
},
{
"epoch": 0.05,
"learning_rate": 9.936277191957737e-05,
"loss": 2.4129,
"step": 6540
},
{
"epoch": 0.05,
"learning_rate": 9.936178416208813e-05,
"loss": 2.3517,
"step": 6545
},
{
"epoch": 0.05,
"learning_rate": 9.9360795644555e-05,
"loss": 2.2883,
"step": 6550
},
{
"epoch": 0.05,
"learning_rate": 9.935980636699322e-05,
"loss": 2.4398,
"step": 6555
},
{
"epoch": 0.05,
"learning_rate": 9.935881632941801e-05,
"loss": 2.4896,
"step": 6560
},
{
"epoch": 0.05,
"learning_rate": 9.935782553184463e-05,
"loss": 2.3903,
"step": 6565
},
{
"epoch": 0.05,
"learning_rate": 9.935683397428833e-05,
"loss": 2.3165,
"step": 6570
},
{
"epoch": 0.05,
"learning_rate": 9.935584165676436e-05,
"loss": 2.3147,
"step": 6575
},
{
"epoch": 0.05,
"learning_rate": 9.935484857928801e-05,
"loss": 2.345,
"step": 6580
},
{
"epoch": 0.05,
"learning_rate": 9.935385474187459e-05,
"loss": 2.343,
"step": 6585
},
{
"epoch": 0.05,
"learning_rate": 9.93528601445394e-05,
"loss": 2.4187,
"step": 6590
},
{
"epoch": 0.05,
"learning_rate": 9.935186478729771e-05,
"loss": 2.3759,
"step": 6595
},
{
"epoch": 0.05,
"learning_rate": 9.93508686701649e-05,
"loss": 2.3217,
"step": 6600
},
{
"epoch": 0.05,
"learning_rate": 9.934987179315627e-05,
"loss": 2.4285,
"step": 6605
},
{
"epoch": 0.05,
"learning_rate": 9.934887415628719e-05,
"loss": 2.3872,
"step": 6610
},
{
"epoch": 0.05,
"learning_rate": 9.934787575957301e-05,
"loss": 2.4104,
"step": 6615
},
{
"epoch": 0.05,
"learning_rate": 9.934687660302912e-05,
"loss": 2.4372,
"step": 6620
},
{
"epoch": 0.05,
"learning_rate": 9.934587668667088e-05,
"loss": 2.4247,
"step": 6625
},
{
"epoch": 0.05,
"learning_rate": 9.93448760105137e-05,
"loss": 2.2106,
"step": 6630
},
{
"epoch": 0.05,
"learning_rate": 9.934387457457299e-05,
"loss": 2.3513,
"step": 6635
},
{
"epoch": 0.05,
"learning_rate": 9.934287237886417e-05,
"loss": 2.3342,
"step": 6640
},
{
"epoch": 0.05,
"learning_rate": 9.934186942340265e-05,
"loss": 2.4268,
"step": 6645
},
{
"epoch": 0.05,
"learning_rate": 9.934086570820392e-05,
"loss": 2.3048,
"step": 6650
},
{
"epoch": 0.05,
"learning_rate": 9.933986123328337e-05,
"loss": 2.3706,
"step": 6655
},
{
"epoch": 0.05,
"learning_rate": 9.933885599865651e-05,
"loss": 2.3671,
"step": 6660
},
{
"epoch": 0.05,
"learning_rate": 9.933785000433882e-05,
"loss": 2.3992,
"step": 6665
},
{
"epoch": 0.05,
"learning_rate": 9.933684325034577e-05,
"loss": 2.3294,
"step": 6670
},
{
"epoch": 0.05,
"learning_rate": 9.933583573669287e-05,
"loss": 2.4166,
"step": 6675
},
{
"epoch": 0.05,
"learning_rate": 9.933482746339564e-05,
"loss": 2.3237,
"step": 6680
},
{
"epoch": 0.05,
"learning_rate": 9.933381843046959e-05,
"loss": 2.2813,
"step": 6685
},
{
"epoch": 0.05,
"learning_rate": 9.933280863793026e-05,
"loss": 2.1867,
"step": 6690
},
{
"epoch": 0.05,
"learning_rate": 9.933179808579321e-05,
"loss": 2.3893,
"step": 6695
},
{
"epoch": 0.05,
"learning_rate": 9.933078677407399e-05,
"loss": 2.4493,
"step": 6700
},
{
"epoch": 0.05,
"learning_rate": 9.932977470278818e-05,
"loss": 2.2811,
"step": 6705
},
{
"epoch": 0.05,
"learning_rate": 9.932876187195135e-05,
"loss": 2.4711,
"step": 6710
},
{
"epoch": 0.05,
"learning_rate": 9.932774828157911e-05,
"loss": 2.3858,
"step": 6715
},
{
"epoch": 0.05,
"learning_rate": 9.932673393168705e-05,
"loss": 2.4135,
"step": 6720
},
{
"epoch": 0.05,
"learning_rate": 9.932571882229081e-05,
"loss": 2.3879,
"step": 6725
},
{
"epoch": 0.05,
"learning_rate": 9.932470295340599e-05,
"loss": 2.4096,
"step": 6730
},
{
"epoch": 0.05,
"learning_rate": 9.932368632504826e-05,
"loss": 2.2986,
"step": 6735
},
{
"epoch": 0.05,
"learning_rate": 9.932266893723325e-05,
"loss": 2.406,
"step": 6740
},
{
"epoch": 0.05,
"learning_rate": 9.932165078997666e-05,
"loss": 2.4007,
"step": 6745
},
{
"epoch": 0.05,
"learning_rate": 9.932063188329412e-05,
"loss": 2.3916,
"step": 6750
},
{
"epoch": 0.05,
"learning_rate": 9.931961221720136e-05,
"loss": 2.4662,
"step": 6755
},
{
"epoch": 0.05,
"learning_rate": 9.931859179171405e-05,
"loss": 2.3746,
"step": 6760
},
{
"epoch": 0.05,
"learning_rate": 9.931757060684792e-05,
"loss": 2.3223,
"step": 6765
},
{
"epoch": 0.05,
"learning_rate": 9.93165486626187e-05,
"loss": 2.3499,
"step": 6770
},
{
"epoch": 0.05,
"learning_rate": 9.93155259590421e-05,
"loss": 2.3228,
"step": 6775
},
{
"epoch": 0.05,
"learning_rate": 9.931450249613389e-05,
"loss": 2.375,
"step": 6780
},
{
"epoch": 0.05,
"learning_rate": 9.931347827390982e-05,
"loss": 2.3417,
"step": 6785
},
{
"epoch": 0.05,
"learning_rate": 9.931245329238566e-05,
"loss": 2.3654,
"step": 6790
},
{
"epoch": 0.05,
"learning_rate": 9.931142755157719e-05,
"loss": 2.4091,
"step": 6795
},
{
"epoch": 0.05,
"learning_rate": 9.931040105150019e-05,
"loss": 2.3769,
"step": 6800
},
{
"epoch": 0.05,
"learning_rate": 9.93093737921705e-05,
"loss": 2.3605,
"step": 6805
},
{
"epoch": 0.05,
"learning_rate": 9.93083457736039e-05,
"loss": 2.3349,
"step": 6810
},
{
"epoch": 0.05,
"learning_rate": 9.930731699581626e-05,
"loss": 2.3785,
"step": 6815
},
{
"epoch": 0.05,
"learning_rate": 9.930628745882339e-05,
"loss": 2.3442,
"step": 6820
},
{
"epoch": 0.05,
"learning_rate": 9.930525716264114e-05,
"loss": 2.3885,
"step": 6825
},
{
"epoch": 0.05,
"learning_rate": 9.93042261072854e-05,
"loss": 2.3758,
"step": 6830
},
{
"epoch": 0.05,
"learning_rate": 9.930319429277202e-05,
"loss": 2.376,
"step": 6835
},
{
"epoch": 0.05,
"learning_rate": 9.93021617191169e-05,
"loss": 2.3369,
"step": 6840
},
{
"epoch": 0.05,
"learning_rate": 9.930112838633592e-05,
"loss": 2.3396,
"step": 6845
},
{
"epoch": 0.05,
"learning_rate": 9.930009429444503e-05,
"loss": 2.3468,
"step": 6850
},
{
"epoch": 0.05,
"learning_rate": 9.929905944346012e-05,
"loss": 2.4011,
"step": 6855
},
{
"epoch": 0.05,
"learning_rate": 9.929802383339713e-05,
"loss": 2.3926,
"step": 6860
},
{
"epoch": 0.05,
"learning_rate": 9.9296987464272e-05,
"loss": 2.4429,
"step": 6865
},
{
"epoch": 0.05,
"learning_rate": 9.929595033610072e-05,
"loss": 2.4369,
"step": 6870
},
{
"epoch": 0.05,
"learning_rate": 9.929491244889921e-05,
"loss": 2.3852,
"step": 6875
},
{
"epoch": 0.05,
"learning_rate": 9.929387380268349e-05,
"loss": 2.3331,
"step": 6880
},
{
"epoch": 0.05,
"learning_rate": 9.929283439746955e-05,
"loss": 2.3485,
"step": 6885
},
{
"epoch": 0.05,
"learning_rate": 9.929179423327336e-05,
"loss": 2.2489,
"step": 6890
},
{
"epoch": 0.05,
"learning_rate": 9.929075331011097e-05,
"loss": 2.4056,
"step": 6895
},
{
"epoch": 0.05,
"learning_rate": 9.928971162799838e-05,
"loss": 2.2964,
"step": 6900
},
{
"epoch": 0.05,
"learning_rate": 9.928866918695165e-05,
"loss": 2.397,
"step": 6905
},
{
"epoch": 0.05,
"learning_rate": 9.928762598698684e-05,
"loss": 2.3699,
"step": 6910
},
{
"epoch": 0.05,
"learning_rate": 9.928658202811999e-05,
"loss": 2.3825,
"step": 6915
},
{
"epoch": 0.05,
"learning_rate": 9.928553731036719e-05,
"loss": 2.2952,
"step": 6920
},
{
"epoch": 0.05,
"learning_rate": 9.928449183374451e-05,
"loss": 2.3189,
"step": 6925
},
{
"epoch": 0.05,
"learning_rate": 9.928344559826806e-05,
"loss": 2.3045,
"step": 6930
},
{
"epoch": 0.05,
"learning_rate": 9.928239860395394e-05,
"loss": 2.4084,
"step": 6935
},
{
"epoch": 0.05,
"learning_rate": 9.928135085081827e-05,
"loss": 2.4062,
"step": 6940
},
{
"epoch": 0.05,
"learning_rate": 9.928030233887721e-05,
"loss": 2.3289,
"step": 6945
},
{
"epoch": 0.05,
"learning_rate": 9.927925306814686e-05,
"loss": 2.2755,
"step": 6950
},
{
"epoch": 0.05,
"learning_rate": 9.927820303864341e-05,
"loss": 2.43,
"step": 6955
},
{
"epoch": 0.05,
"learning_rate": 9.927715225038303e-05,
"loss": 2.4077,
"step": 6960
},
{
"epoch": 0.05,
"learning_rate": 9.927610070338187e-05,
"loss": 2.3135,
"step": 6965
},
{
"epoch": 0.06,
"learning_rate": 9.927504839765615e-05,
"loss": 2.3458,
"step": 6970
},
{
"epoch": 0.06,
"learning_rate": 9.927399533322206e-05,
"loss": 2.405,
"step": 6975
},
{
"epoch": 0.06,
"learning_rate": 9.92729415100958e-05,
"loss": 2.301,
"step": 6980
},
{
"epoch": 0.06,
"learning_rate": 9.927188692829363e-05,
"loss": 2.3007,
"step": 6985
},
{
"epoch": 0.06,
"learning_rate": 9.927083158783177e-05,
"loss": 2.3093,
"step": 6990
},
{
"epoch": 0.06,
"learning_rate": 9.926977548872644e-05,
"loss": 2.3382,
"step": 6995
},
{
"epoch": 0.06,
"learning_rate": 9.926871863099396e-05,
"loss": 2.4066,
"step": 7000
},
{
"epoch": 0.06,
"learning_rate": 9.926766101465056e-05,
"loss": 2.3682,
"step": 7005
},
{
"epoch": 0.06,
"learning_rate": 9.926660263971254e-05,
"loss": 2.4402,
"step": 7010
},
{
"epoch": 0.06,
"learning_rate": 9.92655435061962e-05,
"loss": 2.2739,
"step": 7015
},
{
"epoch": 0.06,
"learning_rate": 9.926448361411784e-05,
"loss": 2.3413,
"step": 7020
},
{
"epoch": 0.06,
"learning_rate": 9.926342296349378e-05,
"loss": 2.405,
"step": 7025
},
{
"epoch": 0.06,
"learning_rate": 9.926236155434034e-05,
"loss": 2.4328,
"step": 7030
},
{
"epoch": 0.06,
"learning_rate": 9.926129938667388e-05,
"loss": 2.3451,
"step": 7035
},
{
"epoch": 0.06,
"learning_rate": 9.926023646051075e-05,
"loss": 2.3214,
"step": 7040
},
{
"epoch": 0.06,
"learning_rate": 9.925917277586734e-05,
"loss": 2.3572,
"step": 7045
},
{
"epoch": 0.06,
"learning_rate": 9.925810833275998e-05,
"loss": 2.3211,
"step": 7050
},
{
"epoch": 0.06,
"learning_rate": 9.925704313120508e-05,
"loss": 2.3606,
"step": 7055
},
{
"epoch": 0.06,
"learning_rate": 9.925597717121906e-05,
"loss": 2.3801,
"step": 7060
},
{
"epoch": 0.06,
"learning_rate": 9.925491045281832e-05,
"loss": 2.3279,
"step": 7065
},
{
"epoch": 0.06,
"learning_rate": 9.925384297601928e-05,
"loss": 2.3772,
"step": 7070
},
{
"epoch": 0.06,
"learning_rate": 9.925277474083835e-05,
"loss": 2.4553,
"step": 7075
},
{
"epoch": 0.06,
"learning_rate": 9.925170574729204e-05,
"loss": 2.2811,
"step": 7080
},
{
"epoch": 0.06,
"learning_rate": 9.925063599539677e-05,
"loss": 2.3583,
"step": 7085
},
{
"epoch": 0.06,
"learning_rate": 9.924956548516901e-05,
"loss": 2.3221,
"step": 7090
},
{
"epoch": 0.06,
"learning_rate": 9.924849421662528e-05,
"loss": 2.3138,
"step": 7095
},
{
"epoch": 0.06,
"learning_rate": 9.924742218978203e-05,
"loss": 2.4319,
"step": 7100
},
{
"epoch": 0.06,
"learning_rate": 9.924634940465577e-05,
"loss": 2.3649,
"step": 7105
},
{
"epoch": 0.06,
"learning_rate": 9.924527586126305e-05,
"loss": 2.3692,
"step": 7110
},
{
"epoch": 0.06,
"learning_rate": 9.924420155962037e-05,
"loss": 2.2898,
"step": 7115
},
{
"epoch": 0.06,
"learning_rate": 9.924312649974427e-05,
"loss": 2.3377,
"step": 7120
},
{
"epoch": 0.06,
"learning_rate": 9.924205068165134e-05,
"loss": 2.298,
"step": 7125
},
{
"epoch": 0.06,
"learning_rate": 9.924097410535811e-05,
"loss": 2.357,
"step": 7130
},
{
"epoch": 0.06,
"learning_rate": 9.923989677088116e-05,
"loss": 2.3682,
"step": 7135
},
{
"epoch": 0.06,
"learning_rate": 9.92388186782371e-05,
"loss": 2.3223,
"step": 7140
},
{
"epoch": 0.06,
"learning_rate": 9.92377398274425e-05,
"loss": 2.4141,
"step": 7145
},
{
"epoch": 0.06,
"learning_rate": 9.923666021851398e-05,
"loss": 2.3555,
"step": 7150
},
{
"epoch": 0.06,
"learning_rate": 9.92355798514682e-05,
"loss": 2.3275,
"step": 7155
},
{
"epoch": 0.06,
"learning_rate": 9.923449872632173e-05,
"loss": 2.2707,
"step": 7160
},
{
"epoch": 0.06,
"learning_rate": 9.923341684309128e-05,
"loss": 2.3185,
"step": 7165
},
{
"epoch": 0.06,
"learning_rate": 9.923233420179347e-05,
"loss": 2.2994,
"step": 7170
},
{
"epoch": 0.06,
"learning_rate": 9.923125080244497e-05,
"loss": 2.3234,
"step": 7175
},
{
"epoch": 0.06,
"learning_rate": 9.923016664506248e-05,
"loss": 2.2583,
"step": 7180
},
{
"epoch": 0.06,
"learning_rate": 9.922908172966268e-05,
"loss": 2.3228,
"step": 7185
},
{
"epoch": 0.06,
"learning_rate": 9.922799605626227e-05,
"loss": 2.3452,
"step": 7190
},
{
"epoch": 0.06,
"learning_rate": 9.9226909624878e-05,
"loss": 2.3377,
"step": 7195
},
{
"epoch": 0.06,
"learning_rate": 9.922582243552654e-05,
"loss": 2.3689,
"step": 7200
},
{
"epoch": 0.06,
"learning_rate": 9.922473448822469e-05,
"loss": 2.417,
"step": 7205
},
{
"epoch": 0.06,
"learning_rate": 9.922364578298916e-05,
"loss": 2.3599,
"step": 7210
},
{
"epoch": 0.06,
"learning_rate": 9.922255631983673e-05,
"loss": 2.422,
"step": 7215
},
{
"epoch": 0.06,
"learning_rate": 9.922146609878418e-05,
"loss": 2.2757,
"step": 7220
},
{
"epoch": 0.06,
"learning_rate": 9.922037511984829e-05,
"loss": 2.4058,
"step": 7225
},
{
"epoch": 0.06,
"learning_rate": 9.921928338304584e-05,
"loss": 2.3777,
"step": 7230
},
{
"epoch": 0.06,
"learning_rate": 9.921819088839369e-05,
"loss": 2.3488,
"step": 7235
},
{
"epoch": 0.06,
"learning_rate": 9.92170976359086e-05,
"loss": 2.4269,
"step": 7240
},
{
"epoch": 0.06,
"learning_rate": 9.921600362560744e-05,
"loss": 2.3461,
"step": 7245
},
{
"epoch": 0.06,
"learning_rate": 9.921490885750706e-05,
"loss": 2.3722,
"step": 7250
},
{
"epoch": 0.06,
"learning_rate": 9.92138133316243e-05,
"loss": 2.2829,
"step": 7255
},
{
"epoch": 0.06,
"learning_rate": 9.921271704797602e-05,
"loss": 2.4695,
"step": 7260
},
{
"epoch": 0.06,
"learning_rate": 9.921162000657913e-05,
"loss": 2.262,
"step": 7265
},
{
"epoch": 0.06,
"learning_rate": 9.921052220745048e-05,
"loss": 2.2571,
"step": 7270
},
{
"epoch": 0.06,
"learning_rate": 9.920942365060702e-05,
"loss": 2.3274,
"step": 7275
},
{
"epoch": 0.06,
"learning_rate": 9.920832433606564e-05,
"loss": 2.3853,
"step": 7280
},
{
"epoch": 0.06,
"learning_rate": 9.920722426384326e-05,
"loss": 2.3681,
"step": 7285
},
{
"epoch": 0.06,
"learning_rate": 9.920612343395683e-05,
"loss": 2.3298,
"step": 7290
},
{
"epoch": 0.06,
"learning_rate": 9.92050218464233e-05,
"loss": 2.4227,
"step": 7295
},
{
"epoch": 0.06,
"learning_rate": 9.920391950125963e-05,
"loss": 2.3897,
"step": 7300
},
{
"epoch": 0.06,
"learning_rate": 9.920281639848278e-05,
"loss": 2.4071,
"step": 7305
},
{
"epoch": 0.06,
"learning_rate": 9.920171253810975e-05,
"loss": 2.3699,
"step": 7310
},
{
"epoch": 0.06,
"learning_rate": 9.920060792015754e-05,
"loss": 2.4576,
"step": 7315
},
{
"epoch": 0.06,
"learning_rate": 9.919950254464316e-05,
"loss": 2.3486,
"step": 7320
},
{
"epoch": 0.06,
"learning_rate": 9.919839641158361e-05,
"loss": 2.328,
"step": 7325
},
{
"epoch": 0.06,
"learning_rate": 9.919728952099593e-05,
"loss": 2.2872,
"step": 7330
},
{
"epoch": 0.06,
"learning_rate": 9.919618187289717e-05,
"loss": 2.3114,
"step": 7335
},
{
"epoch": 0.06,
"learning_rate": 9.919507346730437e-05,
"loss": 2.299,
"step": 7340
},
{
"epoch": 0.06,
"learning_rate": 9.919396430423463e-05,
"loss": 2.3634,
"step": 7345
},
{
"epoch": 0.06,
"learning_rate": 9.9192854383705e-05,
"loss": 2.3354,
"step": 7350
},
{
"epoch": 0.06,
"learning_rate": 9.919174370573257e-05,
"loss": 2.3476,
"step": 7355
},
{
"epoch": 0.06,
"learning_rate": 9.919063227033445e-05,
"loss": 2.351,
"step": 7360
},
{
"epoch": 0.06,
"learning_rate": 9.918952007752776e-05,
"loss": 2.3201,
"step": 7365
},
{
"epoch": 0.06,
"learning_rate": 9.91884071273296e-05,
"loss": 2.3329,
"step": 7370
},
{
"epoch": 0.06,
"learning_rate": 9.918729341975712e-05,
"loss": 2.4268,
"step": 7375
},
{
"epoch": 0.06,
"learning_rate": 9.918617895482749e-05,
"loss": 2.398,
"step": 7380
},
{
"epoch": 0.06,
"learning_rate": 9.918506373255784e-05,
"loss": 2.3007,
"step": 7385
},
{
"epoch": 0.06,
"learning_rate": 9.918394775296536e-05,
"loss": 2.425,
"step": 7390
},
{
"epoch": 0.06,
"learning_rate": 9.91828310160672e-05,
"loss": 2.3903,
"step": 7395
},
{
"epoch": 0.06,
"learning_rate": 9.918171352188062e-05,
"loss": 2.4062,
"step": 7400
},
{
"epoch": 0.06,
"learning_rate": 9.918059527042275e-05,
"loss": 2.2876,
"step": 7405
},
{
"epoch": 0.06,
"learning_rate": 9.917947626171086e-05,
"loss": 2.5239,
"step": 7410
},
{
"epoch": 0.06,
"learning_rate": 9.917835649576217e-05,
"loss": 2.4021,
"step": 7415
},
{
"epoch": 0.06,
"learning_rate": 9.91772359725939e-05,
"loss": 2.3304,
"step": 7420
},
{
"epoch": 0.06,
"learning_rate": 9.917611469222333e-05,
"loss": 2.3385,
"step": 7425
},
{
"epoch": 0.06,
"learning_rate": 9.917499265466772e-05,
"loss": 2.3162,
"step": 7430
},
{
"epoch": 0.06,
"learning_rate": 9.917386985994432e-05,
"loss": 2.337,
"step": 7435
},
{
"epoch": 0.06,
"learning_rate": 9.917274630807047e-05,
"loss": 2.3652,
"step": 7440
},
{
"epoch": 0.06,
"learning_rate": 9.917162199906341e-05,
"loss": 2.4653,
"step": 7445
},
{
"epoch": 0.06,
"learning_rate": 9.917049693294049e-05,
"loss": 2.2902,
"step": 7450
},
{
"epoch": 0.06,
"learning_rate": 9.916937110971902e-05,
"loss": 2.3456,
"step": 7455
},
{
"epoch": 0.06,
"learning_rate": 9.916824452941635e-05,
"loss": 2.3954,
"step": 7460
},
{
"epoch": 0.06,
"learning_rate": 9.916711719204981e-05,
"loss": 2.3778,
"step": 7465
},
{
"epoch": 0.06,
"learning_rate": 9.916598909763675e-05,
"loss": 2.2936,
"step": 7470
},
{
"epoch": 0.06,
"learning_rate": 9.916486024619455e-05,
"loss": 2.3188,
"step": 7475
},
{
"epoch": 0.06,
"learning_rate": 9.916373063774062e-05,
"loss": 2.3263,
"step": 7480
},
{
"epoch": 0.06,
"learning_rate": 9.91626002722923e-05,
"loss": 2.3595,
"step": 7485
},
{
"epoch": 0.06,
"learning_rate": 9.916146914986703e-05,
"loss": 2.481,
"step": 7490
},
{
"epoch": 0.06,
"learning_rate": 9.916033727048223e-05,
"loss": 2.2505,
"step": 7495
},
{
"epoch": 0.06,
"learning_rate": 9.915920463415531e-05,
"loss": 2.4084,
"step": 7500
},
{
"epoch": 0.06,
"learning_rate": 9.915807124090371e-05,
"loss": 2.2481,
"step": 7505
},
{
"epoch": 0.06,
"learning_rate": 9.915693709074487e-05,
"loss": 2.3676,
"step": 7510
},
{
"epoch": 0.06,
"learning_rate": 9.91558021836963e-05,
"loss": 2.286,
"step": 7515
},
{
"epoch": 0.06,
"learning_rate": 9.915466651977543e-05,
"loss": 2.3683,
"step": 7520
},
{
"epoch": 0.06,
"learning_rate": 9.915353009899976e-05,
"loss": 2.3834,
"step": 7525
},
{
"epoch": 0.06,
"learning_rate": 9.915239292138678e-05,
"loss": 2.3726,
"step": 7530
},
{
"epoch": 0.06,
"learning_rate": 9.915125498695402e-05,
"loss": 2.4103,
"step": 7535
},
{
"epoch": 0.06,
"learning_rate": 9.9150116295719e-05,
"loss": 2.3567,
"step": 7540
},
{
"epoch": 0.06,
"learning_rate": 9.914897684769921e-05,
"loss": 2.4444,
"step": 7545
},
{
"epoch": 0.06,
"learning_rate": 9.914783664291225e-05,
"loss": 2.3492,
"step": 7550
},
{
"epoch": 0.06,
"learning_rate": 9.914669568137565e-05,
"loss": 2.302,
"step": 7555
},
{
"epoch": 0.06,
"learning_rate": 9.914555396310698e-05,
"loss": 2.3261,
"step": 7560
},
{
"epoch": 0.06,
"learning_rate": 9.914441148812382e-05,
"loss": 2.4048,
"step": 7565
},
{
"epoch": 0.06,
"learning_rate": 9.914326825644376e-05,
"loss": 2.3129,
"step": 7570
},
{
"epoch": 0.06,
"learning_rate": 9.91421242680844e-05,
"loss": 2.3386,
"step": 7575
},
{
"epoch": 0.06,
"learning_rate": 9.914097952306338e-05,
"loss": 2.324,
"step": 7580
},
{
"epoch": 0.06,
"learning_rate": 9.913983402139828e-05,
"loss": 2.3791,
"step": 7585
},
{
"epoch": 0.06,
"learning_rate": 9.913868776310677e-05,
"loss": 2.3458,
"step": 7590
},
{
"epoch": 0.06,
"learning_rate": 9.913754074820649e-05,
"loss": 2.3966,
"step": 7595
},
{
"epoch": 0.06,
"learning_rate": 9.913639297671511e-05,
"loss": 2.2862,
"step": 7600
},
{
"epoch": 0.06,
"learning_rate": 9.91352444486503e-05,
"loss": 2.323,
"step": 7605
},
{
"epoch": 0.06,
"learning_rate": 9.913409516402974e-05,
"loss": 2.2953,
"step": 7610
},
{
"epoch": 0.06,
"learning_rate": 9.913294512287112e-05,
"loss": 2.3351,
"step": 7615
},
{
"epoch": 0.06,
"learning_rate": 9.913179432519215e-05,
"loss": 2.3243,
"step": 7620
},
{
"epoch": 0.06,
"learning_rate": 9.913064277101055e-05,
"loss": 2.308,
"step": 7625
},
{
"epoch": 0.06,
"learning_rate": 9.912949046034406e-05,
"loss": 2.3389,
"step": 7630
},
{
"epoch": 0.06,
"learning_rate": 9.912833739321042e-05,
"loss": 2.4108,
"step": 7635
},
{
"epoch": 0.06,
"learning_rate": 9.912718356962738e-05,
"loss": 2.2862,
"step": 7640
},
{
"epoch": 0.06,
"learning_rate": 9.91260289896127e-05,
"loss": 2.3558,
"step": 7645
},
{
"epoch": 0.06,
"learning_rate": 9.912487365318417e-05,
"loss": 2.3464,
"step": 7650
},
{
"epoch": 0.06,
"learning_rate": 9.912371756035957e-05,
"loss": 2.2672,
"step": 7655
},
{
"epoch": 0.06,
"learning_rate": 9.912256071115671e-05,
"loss": 2.3999,
"step": 7660
},
{
"epoch": 0.06,
"learning_rate": 9.91214031055934e-05,
"loss": 2.3575,
"step": 7665
},
{
"epoch": 0.06,
"learning_rate": 9.912024474368745e-05,
"loss": 2.3922,
"step": 7670
},
{
"epoch": 0.06,
"learning_rate": 9.911908562545671e-05,
"loss": 2.3691,
"step": 7675
},
{
"epoch": 0.06,
"learning_rate": 9.911792575091903e-05,
"loss": 2.3647,
"step": 7680
},
{
"epoch": 0.06,
"learning_rate": 9.911676512009226e-05,
"loss": 2.4268,
"step": 7685
},
{
"epoch": 0.06,
"learning_rate": 9.911560373299426e-05,
"loss": 2.2945,
"step": 7690
},
{
"epoch": 0.06,
"learning_rate": 9.911444158964296e-05,
"loss": 2.2823,
"step": 7695
},
{
"epoch": 0.06,
"learning_rate": 9.91132786900562e-05,
"loss": 2.3925,
"step": 7700
},
{
"epoch": 0.06,
"learning_rate": 9.911211503425191e-05,
"loss": 2.4117,
"step": 7705
},
{
"epoch": 0.06,
"learning_rate": 9.911095062224798e-05,
"loss": 2.296,
"step": 7710
},
{
"epoch": 0.06,
"learning_rate": 9.91097854540624e-05,
"loss": 2.3004,
"step": 7715
},
{
"epoch": 0.06,
"learning_rate": 9.910861952971304e-05,
"loss": 2.3728,
"step": 7720
},
{
"epoch": 0.06,
"learning_rate": 9.910745284921791e-05,
"loss": 2.3438,
"step": 7725
},
{
"epoch": 0.06,
"learning_rate": 9.910628541259495e-05,
"loss": 2.4127,
"step": 7730
},
{
"epoch": 0.06,
"learning_rate": 9.91051172198621e-05,
"loss": 2.4626,
"step": 7735
},
{
"epoch": 0.06,
"learning_rate": 9.910394827103742e-05,
"loss": 2.33,
"step": 7740
},
{
"epoch": 0.06,
"learning_rate": 9.910277856613886e-05,
"loss": 2.3187,
"step": 7745
},
{
"epoch": 0.06,
"learning_rate": 9.910160810518442e-05,
"loss": 2.3519,
"step": 7750
},
{
"epoch": 0.06,
"learning_rate": 9.910043688819218e-05,
"loss": 2.3718,
"step": 7755
},
{
"epoch": 0.06,
"learning_rate": 9.909926491518012e-05,
"loss": 2.3485,
"step": 7760
},
{
"epoch": 0.06,
"learning_rate": 9.909809218616629e-05,
"loss": 2.4477,
"step": 7765
},
{
"epoch": 0.06,
"learning_rate": 9.909691870116877e-05,
"loss": 2.3936,
"step": 7770
},
{
"epoch": 0.06,
"learning_rate": 9.909574446020561e-05,
"loss": 2.3875,
"step": 7775
},
{
"epoch": 0.06,
"learning_rate": 9.90945694632949e-05,
"loss": 2.2542,
"step": 7780
},
{
"epoch": 0.06,
"learning_rate": 9.909339371045474e-05,
"loss": 2.3827,
"step": 7785
},
{
"epoch": 0.06,
"learning_rate": 9.90922172017032e-05,
"loss": 2.3266,
"step": 7790
},
{
"epoch": 0.06,
"learning_rate": 9.909103993705845e-05,
"loss": 2.3195,
"step": 7795
},
{
"epoch": 0.06,
"learning_rate": 9.908986191653857e-05,
"loss": 2.3872,
"step": 7800
},
{
"epoch": 0.06,
"learning_rate": 9.90886831401617e-05,
"loss": 2.3606,
"step": 7805
},
{
"epoch": 0.06,
"learning_rate": 9.908750360794603e-05,
"loss": 2.3853,
"step": 7810
},
{
"epoch": 0.06,
"learning_rate": 9.908632331990968e-05,
"loss": 2.2987,
"step": 7815
},
{
"epoch": 0.06,
"learning_rate": 9.908514227607084e-05,
"loss": 2.2592,
"step": 7820
},
{
"epoch": 0.06,
"learning_rate": 9.90839604764477e-05,
"loss": 2.2975,
"step": 7825
},
{
"epoch": 0.06,
"learning_rate": 9.908277792105845e-05,
"loss": 2.3444,
"step": 7830
},
{
"epoch": 0.06,
"learning_rate": 9.90815946099213e-05,
"loss": 2.3504,
"step": 7835
},
{
"epoch": 0.06,
"learning_rate": 9.908041054305446e-05,
"loss": 2.3073,
"step": 7840
},
{
"epoch": 0.06,
"learning_rate": 9.907922572047618e-05,
"loss": 2.3719,
"step": 7845
},
{
"epoch": 0.06,
"learning_rate": 9.90780401422047e-05,
"loss": 2.3488,
"step": 7850
},
{
"epoch": 0.06,
"learning_rate": 9.907685380825825e-05,
"loss": 2.4012,
"step": 7855
},
{
"epoch": 0.06,
"learning_rate": 9.907566671865512e-05,
"loss": 2.1989,
"step": 7860
},
{
"epoch": 0.06,
"learning_rate": 9.907447887341359e-05,
"loss": 2.3225,
"step": 7865
},
{
"epoch": 0.06,
"learning_rate": 9.907329027255194e-05,
"loss": 2.3684,
"step": 7870
},
{
"epoch": 0.06,
"learning_rate": 9.907210091608848e-05,
"loss": 2.4374,
"step": 7875
},
{
"epoch": 0.06,
"learning_rate": 9.907091080404152e-05,
"loss": 2.3309,
"step": 7880
},
{
"epoch": 0.06,
"learning_rate": 9.906971993642937e-05,
"loss": 2.185,
"step": 7885
},
{
"epoch": 0.06,
"learning_rate": 9.906852831327038e-05,
"loss": 2.3023,
"step": 7890
},
{
"epoch": 0.06,
"learning_rate": 9.906733593458289e-05,
"loss": 2.3231,
"step": 7895
},
{
"epoch": 0.06,
"learning_rate": 9.906614280038529e-05,
"loss": 2.3666,
"step": 7900
},
{
"epoch": 0.06,
"learning_rate": 9.90649489106959e-05,
"loss": 2.2454,
"step": 7905
},
{
"epoch": 0.06,
"learning_rate": 9.906375426553314e-05,
"loss": 2.2789,
"step": 7910
},
{
"epoch": 0.06,
"learning_rate": 9.90625588649154e-05,
"loss": 2.3328,
"step": 7915
},
{
"epoch": 0.06,
"learning_rate": 9.906136270886107e-05,
"loss": 2.3041,
"step": 7920
},
{
"epoch": 0.06,
"learning_rate": 9.90601657973886e-05,
"loss": 2.4075,
"step": 7925
},
{
"epoch": 0.06,
"learning_rate": 9.905896813051636e-05,
"loss": 2.3781,
"step": 7930
},
{
"epoch": 0.06,
"learning_rate": 9.905776970826286e-05,
"loss": 2.2789,
"step": 7935
},
{
"epoch": 0.06,
"learning_rate": 9.905657053064651e-05,
"loss": 2.3442,
"step": 7940
},
{
"epoch": 0.06,
"learning_rate": 9.905537059768578e-05,
"loss": 2.2675,
"step": 7945
},
{
"epoch": 0.06,
"learning_rate": 9.905416990939915e-05,
"loss": 2.2976,
"step": 7950
},
{
"epoch": 0.06,
"learning_rate": 9.905296846580512e-05,
"loss": 2.2297,
"step": 7955
},
{
"epoch": 0.06,
"learning_rate": 9.905176626692217e-05,
"loss": 2.3725,
"step": 7960
},
{
"epoch": 0.06,
"learning_rate": 9.905056331276884e-05,
"loss": 2.3705,
"step": 7965
},
{
"epoch": 0.06,
"learning_rate": 9.904935960336359e-05,
"loss": 2.348,
"step": 7970
},
{
"epoch": 0.06,
"learning_rate": 9.904815513872503e-05,
"loss": 2.318,
"step": 7975
},
{
"epoch": 0.06,
"learning_rate": 9.904694991887164e-05,
"loss": 2.2599,
"step": 7980
},
{
"epoch": 0.06,
"learning_rate": 9.904574394382203e-05,
"loss": 2.2969,
"step": 7985
},
{
"epoch": 0.06,
"learning_rate": 9.904453721359474e-05,
"loss": 2.3593,
"step": 7990
},
{
"epoch": 0.06,
"learning_rate": 9.904332972820837e-05,
"loss": 2.3171,
"step": 7995
},
{
"epoch": 0.06,
"learning_rate": 9.904212148768148e-05,
"loss": 2.4089,
"step": 8000
}
],
"logging_steps": 5,
"max_steps": 126689,
"num_train_epochs": 1,
"save_steps": 1000,
"total_flos": 1.970167211674829e+19,
"trial_name": null,
"trial_params": null
}