codet5-k8s-compacter / checkpoint-7075 /trainer_state.json
gpol13's picture
Upload folder using huggingface_hub
97c01c3 verified
Raw
History Blame Contribute Delete
26.8 kB
{
"best_global_step": 7075,
"best_metric": 0.9260658786475363,
"best_model_checkpoint": "/kaggle/working/codet5-k8s-compacter/checkpoint-7075",
"epoch": 5.0,
"eval_steps": 500,
"global_step": 7075,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0353356890459364,
"grad_norm": 126821.5,
"learning_rate": 4.9653710247349825e-05,
"loss": 4.63,
"step": 50
},
{
"epoch": 0.0706713780918728,
"grad_norm": 98924.359375,
"learning_rate": 4.930035335689046e-05,
"loss": 4.1957,
"step": 100
},
{
"epoch": 0.10600706713780919,
"grad_norm": 70176.9296875,
"learning_rate": 4.89469964664311e-05,
"loss": 3.8596,
"step": 150
},
{
"epoch": 0.1413427561837456,
"grad_norm": 51479.76171875,
"learning_rate": 4.859363957597173e-05,
"loss": 3.6109,
"step": 200
},
{
"epoch": 0.17667844522968199,
"grad_norm": 36380.80859375,
"learning_rate": 4.8240282685512374e-05,
"loss": 3.5303,
"step": 250
},
{
"epoch": 0.21201413427561838,
"grad_norm": 88439.21875,
"learning_rate": 4.788692579505301e-05,
"loss": 3.3827,
"step": 300
},
{
"epoch": 0.24734982332155478,
"grad_norm": 42557.44140625,
"learning_rate": 4.753356890459364e-05,
"loss": 3.354,
"step": 350
},
{
"epoch": 0.2826855123674912,
"grad_norm": 49749.4296875,
"learning_rate": 4.718021201413428e-05,
"loss": 3.368,
"step": 400
},
{
"epoch": 0.31802120141342755,
"grad_norm": 44532.0390625,
"learning_rate": 4.682685512367491e-05,
"loss": 3.2346,
"step": 450
},
{
"epoch": 0.35335689045936397,
"grad_norm": 115820.296875,
"learning_rate": 4.647349823321555e-05,
"loss": 3.0554,
"step": 500
},
{
"epoch": 0.38869257950530034,
"grad_norm": 167465.140625,
"learning_rate": 4.612014134275618e-05,
"loss": 2.7072,
"step": 550
},
{
"epoch": 0.42402826855123676,
"grad_norm": 163194.265625,
"learning_rate": 4.5766784452296823e-05,
"loss": 2.3809,
"step": 600
},
{
"epoch": 0.45936395759717313,
"grad_norm": 31167.71875,
"learning_rate": 4.541342756183746e-05,
"loss": 2.306,
"step": 650
},
{
"epoch": 0.49469964664310956,
"grad_norm": 32075.869140625,
"learning_rate": 4.50600706713781e-05,
"loss": 2.2414,
"step": 700
},
{
"epoch": 0.5300353356890459,
"grad_norm": 39896.47265625,
"learning_rate": 4.470671378091873e-05,
"loss": 2.2537,
"step": 750
},
{
"epoch": 0.5653710247349824,
"grad_norm": 31361.541015625,
"learning_rate": 4.435335689045937e-05,
"loss": 2.1922,
"step": 800
},
{
"epoch": 0.6007067137809188,
"grad_norm": 31486.953125,
"learning_rate": 4.4000000000000006e-05,
"loss": 2.134,
"step": 850
},
{
"epoch": 0.6360424028268551,
"grad_norm": 507986.625,
"learning_rate": 4.364664310954063e-05,
"loss": 2.1755,
"step": 900
},
{
"epoch": 0.6713780918727915,
"grad_norm": 50896.109375,
"learning_rate": 4.329328621908127e-05,
"loss": 2.1437,
"step": 950
},
{
"epoch": 0.7067137809187279,
"grad_norm": 32003.013671875,
"learning_rate": 4.293992932862191e-05,
"loss": 2.0918,
"step": 1000
},
{
"epoch": 0.7420494699646644,
"grad_norm": 30772.689453125,
"learning_rate": 4.258657243816255e-05,
"loss": 2.0865,
"step": 1050
},
{
"epoch": 0.7773851590106007,
"grad_norm": 32544.509765625,
"learning_rate": 4.223321554770318e-05,
"loss": 2.1148,
"step": 1100
},
{
"epoch": 0.8127208480565371,
"grad_norm": 30003.00390625,
"learning_rate": 4.187985865724382e-05,
"loss": 2.0311,
"step": 1150
},
{
"epoch": 0.8480565371024735,
"grad_norm": 31737.36328125,
"learning_rate": 4.1526501766784455e-05,
"loss": 1.9837,
"step": 1200
},
{
"epoch": 0.8833922261484098,
"grad_norm": 39721.421875,
"learning_rate": 4.1173144876325096e-05,
"loss": 2.0217,
"step": 1250
},
{
"epoch": 0.9187279151943463,
"grad_norm": 27473.576171875,
"learning_rate": 4.081978798586573e-05,
"loss": 2.0465,
"step": 1300
},
{
"epoch": 0.9540636042402827,
"grad_norm": 28554.697265625,
"learning_rate": 4.0466431095406356e-05,
"loss": 1.9682,
"step": 1350
},
{
"epoch": 0.9893992932862191,
"grad_norm": 246283.546875,
"learning_rate": 4.0113074204947e-05,
"loss": 1.9873,
"step": 1400
},
{
"epoch": 1.0,
"eval_bertscore_f1": 0.9083600320512115,
"eval_bleu": 0.08579427069208953,
"eval_loss": 1.7385830879211426,
"eval_meteor": 0.25803497458339136,
"eval_rouge1": 0.5730890769105835,
"eval_rouge2": 0.2653076932624744,
"eval_runtime": 573.3699,
"eval_samples_per_second": 11.274,
"eval_steps_per_second": 0.705,
"step": 1415
},
{
"epoch": 1.0247349823321554,
"grad_norm": 29727.822265625,
"learning_rate": 3.975971731448763e-05,
"loss": 1.946,
"step": 1450
},
{
"epoch": 1.0600706713780919,
"grad_norm": 103277.5,
"learning_rate": 3.940636042402827e-05,
"loss": 1.9982,
"step": 1500
},
{
"epoch": 1.0954063604240283,
"grad_norm": 25385.19921875,
"learning_rate": 3.9053003533568905e-05,
"loss": 1.9148,
"step": 1550
},
{
"epoch": 1.1307420494699647,
"grad_norm": 31222.15625,
"learning_rate": 3.8699646643109545e-05,
"loss": 1.9448,
"step": 1600
},
{
"epoch": 1.1660777385159011,
"grad_norm": 32575.791015625,
"learning_rate": 3.834628975265018e-05,
"loss": 1.9609,
"step": 1650
},
{
"epoch": 1.2014134275618376,
"grad_norm": 30101.376953125,
"learning_rate": 3.799293286219082e-05,
"loss": 1.943,
"step": 1700
},
{
"epoch": 1.2367491166077738,
"grad_norm": 29648.05859375,
"learning_rate": 3.763957597173145e-05,
"loss": 1.9114,
"step": 1750
},
{
"epoch": 1.2720848056537102,
"grad_norm": 26373.646484375,
"learning_rate": 3.728621908127209e-05,
"loss": 1.9557,
"step": 1800
},
{
"epoch": 1.3074204946996466,
"grad_norm": 26201.498046875,
"learning_rate": 3.693286219081272e-05,
"loss": 1.9525,
"step": 1850
},
{
"epoch": 1.342756183745583,
"grad_norm": 36134.85546875,
"learning_rate": 3.6579505300353355e-05,
"loss": 1.8984,
"step": 1900
},
{
"epoch": 1.3780918727915195,
"grad_norm": 41759.36328125,
"learning_rate": 3.6226148409893995e-05,
"loss": 1.911,
"step": 1950
},
{
"epoch": 1.4134275618374559,
"grad_norm": 28872.177734375,
"learning_rate": 3.587279151943463e-05,
"loss": 1.8871,
"step": 2000
},
{
"epoch": 1.4487632508833923,
"grad_norm": 28179.990234375,
"learning_rate": 3.551943462897527e-05,
"loss": 1.8729,
"step": 2050
},
{
"epoch": 1.4840989399293285,
"grad_norm": 27540.845703125,
"learning_rate": 3.51660777385159e-05,
"loss": 1.8075,
"step": 2100
},
{
"epoch": 1.5194346289752652,
"grad_norm": 26713.314453125,
"learning_rate": 3.4812720848056544e-05,
"loss": 1.8286,
"step": 2150
},
{
"epoch": 1.5547703180212014,
"grad_norm": 29773.23828125,
"learning_rate": 3.445936395759717e-05,
"loss": 1.7446,
"step": 2200
},
{
"epoch": 1.5901060070671378,
"grad_norm": 28969.7421875,
"learning_rate": 3.410600706713781e-05,
"loss": 1.8114,
"step": 2250
},
{
"epoch": 1.6254416961130742,
"grad_norm": 35276.55859375,
"learning_rate": 3.3752650176678445e-05,
"loss": 1.8169,
"step": 2300
},
{
"epoch": 1.6607773851590106,
"grad_norm": 36119.03515625,
"learning_rate": 3.3399293286219085e-05,
"loss": 1.8459,
"step": 2350
},
{
"epoch": 1.696113074204947,
"grad_norm": 28967.146484375,
"learning_rate": 3.304593639575972e-05,
"loss": 1.8705,
"step": 2400
},
{
"epoch": 1.7314487632508833,
"grad_norm": 30086.880859375,
"learning_rate": 3.269257950530035e-05,
"loss": 1.7781,
"step": 2450
},
{
"epoch": 1.76678445229682,
"grad_norm": 108925.2421875,
"learning_rate": 3.233922261484099e-05,
"loss": 1.7466,
"step": 2500
},
{
"epoch": 1.802120141342756,
"grad_norm": 32525.8515625,
"learning_rate": 3.198586572438163e-05,
"loss": 1.8022,
"step": 2550
},
{
"epoch": 1.8374558303886925,
"grad_norm": 28746.19921875,
"learning_rate": 3.163250883392227e-05,
"loss": 1.7927,
"step": 2600
},
{
"epoch": 1.872791519434629,
"grad_norm": 32453.943359375,
"learning_rate": 3.1279151943462894e-05,
"loss": 1.7421,
"step": 2650
},
{
"epoch": 1.9081272084805654,
"grad_norm": 29018.982421875,
"learning_rate": 3.0925795053003535e-05,
"loss": 1.692,
"step": 2700
},
{
"epoch": 1.9434628975265018,
"grad_norm": 31040.447265625,
"learning_rate": 3.057243816254417e-05,
"loss": 1.7915,
"step": 2750
},
{
"epoch": 1.978798586572438,
"grad_norm": 45870.765625,
"learning_rate": 3.021908127208481e-05,
"loss": 1.7336,
"step": 2800
},
{
"epoch": 2.0,
"eval_bertscore_f1": 0.9169509901323973,
"eval_bleu": 0.12235700547901165,
"eval_loss": 1.5117281675338745,
"eval_meteor": 0.3177099828910354,
"eval_rouge1": 0.6191385072228212,
"eval_rouge2": 0.34179551513048145,
"eval_runtime": 554.3192,
"eval_samples_per_second": 11.661,
"eval_steps_per_second": 0.729,
"step": 2830
},
{
"epoch": 2.0141342756183747,
"grad_norm": 54409.53515625,
"learning_rate": 2.9865724381625443e-05,
"loss": 1.7279,
"step": 2850
},
{
"epoch": 2.049469964664311,
"grad_norm": 28479.771484375,
"learning_rate": 2.9512367491166083e-05,
"loss": 1.7633,
"step": 2900
},
{
"epoch": 2.0848056537102475,
"grad_norm": 31366.7265625,
"learning_rate": 2.9159010600706717e-05,
"loss": 1.6949,
"step": 2950
},
{
"epoch": 2.1201413427561837,
"grad_norm": 29734.166015625,
"learning_rate": 2.8805653710247347e-05,
"loss": 1.7121,
"step": 3000
},
{
"epoch": 2.1554770318021204,
"grad_norm": 30663.30859375,
"learning_rate": 2.8452296819787988e-05,
"loss": 1.6896,
"step": 3050
},
{
"epoch": 2.1908127208480566,
"grad_norm": 28285.623046875,
"learning_rate": 2.809893992932862e-05,
"loss": 1.6855,
"step": 3100
},
{
"epoch": 2.2261484098939928,
"grad_norm": 29568.185546875,
"learning_rate": 2.7745583038869262e-05,
"loss": 1.7287,
"step": 3150
},
{
"epoch": 2.2614840989399294,
"grad_norm": 73635.6015625,
"learning_rate": 2.7392226148409892e-05,
"loss": 1.7696,
"step": 3200
},
{
"epoch": 2.2968197879858656,
"grad_norm": 29761.19140625,
"learning_rate": 2.7038869257950533e-05,
"loss": 1.6635,
"step": 3250
},
{
"epoch": 2.3321554770318023,
"grad_norm": 30002.23046875,
"learning_rate": 2.6685512367491167e-05,
"loss": 1.7323,
"step": 3300
},
{
"epoch": 2.3674911660777385,
"grad_norm": 26734.421875,
"learning_rate": 2.6332155477031807e-05,
"loss": 1.7232,
"step": 3350
},
{
"epoch": 2.402826855123675,
"grad_norm": 27280.53515625,
"learning_rate": 2.597879858657244e-05,
"loss": 1.63,
"step": 3400
},
{
"epoch": 2.4381625441696113,
"grad_norm": 28405.10546875,
"learning_rate": 2.562544169611307e-05,
"loss": 1.6773,
"step": 3450
},
{
"epoch": 2.4734982332155475,
"grad_norm": 26917.61328125,
"learning_rate": 2.5272084805653712e-05,
"loss": 1.6261,
"step": 3500
},
{
"epoch": 2.508833922261484,
"grad_norm": 28608.21484375,
"learning_rate": 2.491872791519435e-05,
"loss": 1.6923,
"step": 3550
},
{
"epoch": 2.5441696113074204,
"grad_norm": 32485.646484375,
"learning_rate": 2.4565371024734986e-05,
"loss": 1.699,
"step": 3600
},
{
"epoch": 2.579505300353357,
"grad_norm": 31671.697265625,
"learning_rate": 2.421201413427562e-05,
"loss": 1.6546,
"step": 3650
},
{
"epoch": 2.614840989399293,
"grad_norm": 34171.91796875,
"learning_rate": 2.3858657243816253e-05,
"loss": 1.725,
"step": 3700
},
{
"epoch": 2.65017667844523,
"grad_norm": 37400.90625,
"learning_rate": 2.350530035335689e-05,
"loss": 1.6396,
"step": 3750
},
{
"epoch": 2.685512367491166,
"grad_norm": 30312.548828125,
"learning_rate": 2.3151943462897528e-05,
"loss": 1.6359,
"step": 3800
},
{
"epoch": 2.7208480565371023,
"grad_norm": 78248.390625,
"learning_rate": 2.2798586572438165e-05,
"loss": 1.6089,
"step": 3850
},
{
"epoch": 2.756183745583039,
"grad_norm": 37866.140625,
"learning_rate": 2.24452296819788e-05,
"loss": 1.5952,
"step": 3900
},
{
"epoch": 2.791519434628975,
"grad_norm": 29562.447265625,
"learning_rate": 2.2091872791519436e-05,
"loss": 1.6323,
"step": 3950
},
{
"epoch": 2.8268551236749118,
"grad_norm": 32963.109375,
"learning_rate": 2.1738515901060073e-05,
"loss": 1.6153,
"step": 4000
},
{
"epoch": 2.862190812720848,
"grad_norm": 34929.30859375,
"learning_rate": 2.138515901060071e-05,
"loss": 1.6296,
"step": 4050
},
{
"epoch": 2.8975265017667846,
"grad_norm": 41500.45703125,
"learning_rate": 2.1031802120141344e-05,
"loss": 1.6259,
"step": 4100
},
{
"epoch": 2.932862190812721,
"grad_norm": 55486.66796875,
"learning_rate": 2.067844522968198e-05,
"loss": 1.6189,
"step": 4150
},
{
"epoch": 2.968197879858657,
"grad_norm": 29259.208984375,
"learning_rate": 2.0325088339222618e-05,
"loss": 1.5875,
"step": 4200
},
{
"epoch": 3.0,
"eval_bertscore_f1": 0.9231123085880634,
"eval_bleu": 0.1599837453099706,
"eval_loss": 1.3867945671081543,
"eval_meteor": 0.37567315454997113,
"eval_rouge1": 0.6499931426853252,
"eval_rouge2": 0.3996259880420389,
"eval_runtime": 552.2348,
"eval_samples_per_second": 11.705,
"eval_steps_per_second": 0.732,
"step": 4245
},
{
"epoch": 3.0035335689045937,
"grad_norm": 30802.6484375,
"learning_rate": 1.997173144876325e-05,
"loss": 1.5837,
"step": 4250
},
{
"epoch": 3.03886925795053,
"grad_norm": 25378.61328125,
"learning_rate": 1.9618374558303885e-05,
"loss": 1.5962,
"step": 4300
},
{
"epoch": 3.0742049469964665,
"grad_norm": 26053.240234375,
"learning_rate": 1.9265017667844522e-05,
"loss": 1.6072,
"step": 4350
},
{
"epoch": 3.1095406360424027,
"grad_norm": 26053.06640625,
"learning_rate": 1.891166077738516e-05,
"loss": 1.5856,
"step": 4400
},
{
"epoch": 3.1448763250883394,
"grad_norm": 50342.1796875,
"learning_rate": 1.8558303886925797e-05,
"loss": 1.5687,
"step": 4450
},
{
"epoch": 3.1802120141342756,
"grad_norm": 44797.23046875,
"learning_rate": 1.8204946996466434e-05,
"loss": 1.6209,
"step": 4500
},
{
"epoch": 3.215547703180212,
"grad_norm": 29664.0,
"learning_rate": 1.7851590106007067e-05,
"loss": 1.621,
"step": 4550
},
{
"epoch": 3.2508833922261484,
"grad_norm": 28708.9609375,
"learning_rate": 1.7498233215547705e-05,
"loss": 1.636,
"step": 4600
},
{
"epoch": 3.2862190812720846,
"grad_norm": 29774.0390625,
"learning_rate": 1.714487632508834e-05,
"loss": 1.561,
"step": 4650
},
{
"epoch": 3.3215547703180213,
"grad_norm": 29480.68359375,
"learning_rate": 1.679151943462898e-05,
"loss": 1.5591,
"step": 4700
},
{
"epoch": 3.3568904593639575,
"grad_norm": 38181.2109375,
"learning_rate": 1.643816254416961e-05,
"loss": 1.6705,
"step": 4750
},
{
"epoch": 3.392226148409894,
"grad_norm": 33917.00390625,
"learning_rate": 1.6084805653710246e-05,
"loss": 1.5799,
"step": 4800
},
{
"epoch": 3.4275618374558303,
"grad_norm": 31117.16796875,
"learning_rate": 1.5731448763250883e-05,
"loss": 1.5259,
"step": 4850
},
{
"epoch": 3.462897526501767,
"grad_norm": 33511.91796875,
"learning_rate": 1.537809187279152e-05,
"loss": 1.5974,
"step": 4900
},
{
"epoch": 3.498233215547703,
"grad_norm": 31202.59765625,
"learning_rate": 1.5024734982332156e-05,
"loss": 1.6172,
"step": 4950
},
{
"epoch": 3.53356890459364,
"grad_norm": 33110.30859375,
"learning_rate": 1.4671378091872793e-05,
"loss": 1.5364,
"step": 5000
},
{
"epoch": 3.568904593639576,
"grad_norm": 35678.1484375,
"learning_rate": 1.4318021201413428e-05,
"loss": 1.5857,
"step": 5050
},
{
"epoch": 3.604240282685512,
"grad_norm": 49993.73046875,
"learning_rate": 1.3964664310954066e-05,
"loss": 1.6595,
"step": 5100
},
{
"epoch": 3.639575971731449,
"grad_norm": 28844.365234375,
"learning_rate": 1.3611307420494701e-05,
"loss": 1.6041,
"step": 5150
},
{
"epoch": 3.674911660777385,
"grad_norm": 498931.9375,
"learning_rate": 1.3257950530035338e-05,
"loss": 1.5895,
"step": 5200
},
{
"epoch": 3.7102473498233217,
"grad_norm": 33296.671875,
"learning_rate": 1.2904593639575973e-05,
"loss": 1.4737,
"step": 5250
},
{
"epoch": 3.745583038869258,
"grad_norm": 31166.392578125,
"learning_rate": 1.2551236749116607e-05,
"loss": 1.5497,
"step": 5300
},
{
"epoch": 3.7809187279151946,
"grad_norm": 34931.58984375,
"learning_rate": 1.2197879858657244e-05,
"loss": 1.5326,
"step": 5350
},
{
"epoch": 3.8162544169611308,
"grad_norm": 46567.40625,
"learning_rate": 1.184452296819788e-05,
"loss": 1.5965,
"step": 5400
},
{
"epoch": 3.851590106007067,
"grad_norm": 31954.55859375,
"learning_rate": 1.1491166077738517e-05,
"loss": 1.5341,
"step": 5450
},
{
"epoch": 3.8869257950530036,
"grad_norm": 46481.6640625,
"learning_rate": 1.1137809187279152e-05,
"loss": 1.4878,
"step": 5500
},
{
"epoch": 3.92226148409894,
"grad_norm": 36655.73046875,
"learning_rate": 1.078445229681979e-05,
"loss": 1.5397,
"step": 5550
},
{
"epoch": 3.9575971731448765,
"grad_norm": 34471.97265625,
"learning_rate": 1.0431095406360425e-05,
"loss": 1.5268,
"step": 5600
},
{
"epoch": 3.9929328621908127,
"grad_norm": 40272.60546875,
"learning_rate": 1.0077738515901062e-05,
"loss": 1.5684,
"step": 5650
},
{
"epoch": 4.0,
"eval_bertscore_f1": 0.925423993738276,
"eval_bleu": 0.17623458339946207,
"eval_loss": 1.3250840902328491,
"eval_meteor": 0.4052628837777222,
"eval_rouge1": 0.6619125904046801,
"eval_rouge2": 0.41874017691116694,
"eval_runtime": 553.906,
"eval_samples_per_second": 11.67,
"eval_steps_per_second": 0.729,
"step": 5660
},
{
"epoch": 4.028268551236749,
"grad_norm": 31037.947265625,
"learning_rate": 9.724381625441696e-06,
"loss": 1.5536,
"step": 5700
},
{
"epoch": 4.063604240282685,
"grad_norm": 35932.71484375,
"learning_rate": 9.371024734982333e-06,
"loss": 1.5144,
"step": 5750
},
{
"epoch": 4.098939929328622,
"grad_norm": 32796.33984375,
"learning_rate": 9.017667844522968e-06,
"loss": 1.543,
"step": 5800
},
{
"epoch": 4.134275618374558,
"grad_norm": 30133.2109375,
"learning_rate": 8.664310954063605e-06,
"loss": 1.5336,
"step": 5850
},
{
"epoch": 4.169611307420495,
"grad_norm": 78945.078125,
"learning_rate": 8.31095406360424e-06,
"loss": 1.5428,
"step": 5900
},
{
"epoch": 4.204946996466431,
"grad_norm": 29020.2890625,
"learning_rate": 7.957597173144876e-06,
"loss": 1.5247,
"step": 5950
},
{
"epoch": 4.240282685512367,
"grad_norm": 48070.6875,
"learning_rate": 7.6042402826855124e-06,
"loss": 1.5446,
"step": 6000
},
{
"epoch": 4.275618374558304,
"grad_norm": 31588.23828125,
"learning_rate": 7.250883392226149e-06,
"loss": 1.5016,
"step": 6050
},
{
"epoch": 4.310954063604241,
"grad_norm": 33308.55859375,
"learning_rate": 6.897526501766785e-06,
"loss": 1.5387,
"step": 6100
},
{
"epoch": 4.3462897526501765,
"grad_norm": 28458.3125,
"learning_rate": 6.544169611307421e-06,
"loss": 1.5398,
"step": 6150
},
{
"epoch": 4.381625441696113,
"grad_norm": 27657.84375,
"learning_rate": 6.190812720848057e-06,
"loss": 1.5227,
"step": 6200
},
{
"epoch": 4.41696113074205,
"grad_norm": 72290.9296875,
"learning_rate": 5.837455830388693e-06,
"loss": 1.5334,
"step": 6250
},
{
"epoch": 4.4522968197879855,
"grad_norm": 90226.515625,
"learning_rate": 5.484098939929329e-06,
"loss": 1.5476,
"step": 6300
},
{
"epoch": 4.487632508833922,
"grad_norm": 70717.1875,
"learning_rate": 5.1307420494699655e-06,
"loss": 1.5561,
"step": 6350
},
{
"epoch": 4.522968197879859,
"grad_norm": 30283.47265625,
"learning_rate": 4.777385159010601e-06,
"loss": 1.4628,
"step": 6400
},
{
"epoch": 4.5583038869257955,
"grad_norm": 29865.373046875,
"learning_rate": 4.424028268551237e-06,
"loss": 1.4948,
"step": 6450
},
{
"epoch": 4.593639575971731,
"grad_norm": 25492.849609375,
"learning_rate": 4.070671378091873e-06,
"loss": 1.5816,
"step": 6500
},
{
"epoch": 4.628975265017668,
"grad_norm": 29844.244140625,
"learning_rate": 3.717314487632509e-06,
"loss": 1.5828,
"step": 6550
},
{
"epoch": 4.6643109540636045,
"grad_norm": 28020.556640625,
"learning_rate": 3.363957597173145e-06,
"loss": 1.5165,
"step": 6600
},
{
"epoch": 4.69964664310954,
"grad_norm": 31423.6796875,
"learning_rate": 3.010600706713781e-06,
"loss": 1.5769,
"step": 6650
},
{
"epoch": 4.734982332155477,
"grad_norm": 50652.4453125,
"learning_rate": 2.6572438162544172e-06,
"loss": 1.5217,
"step": 6700
},
{
"epoch": 4.770318021201414,
"grad_norm": 81618.984375,
"learning_rate": 2.303886925795053e-06,
"loss": 1.5231,
"step": 6750
},
{
"epoch": 4.80565371024735,
"grad_norm": 31952.064453125,
"learning_rate": 1.9505300353356893e-06,
"loss": 1.5409,
"step": 6800
},
{
"epoch": 4.840989399293286,
"grad_norm": 31362.482421875,
"learning_rate": 1.5971731448763252e-06,
"loss": 1.5721,
"step": 6850
},
{
"epoch": 4.876325088339223,
"grad_norm": 62709.55078125,
"learning_rate": 1.2438162544169612e-06,
"loss": 1.5167,
"step": 6900
},
{
"epoch": 4.911660777385159,
"grad_norm": 35775.5625,
"learning_rate": 8.904593639575972e-07,
"loss": 1.5063,
"step": 6950
},
{
"epoch": 4.946996466431095,
"grad_norm": 29569.689453125,
"learning_rate": 5.371024734982332e-07,
"loss": 1.5619,
"step": 7000
},
{
"epoch": 4.982332155477032,
"grad_norm": 44565.48828125,
"learning_rate": 1.8374558303886926e-07,
"loss": 1.5322,
"step": 7050
},
{
"epoch": 5.0,
"eval_bertscore_f1": 0.9260658786475363,
"eval_bleu": 0.1809952487112448,
"eval_loss": 1.3066847324371338,
"eval_meteor": 0.4108362479887506,
"eval_rouge1": 0.6655984244559798,
"eval_rouge2": 0.42503071573389145,
"eval_runtime": 552.7345,
"eval_samples_per_second": 11.695,
"eval_steps_per_second": 0.731,
"step": 7075
}
],
"logging_steps": 50,
"max_steps": 7075,
"num_input_tokens_seen": 0,
"num_train_epochs": 5,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 1.723382794788864e+16,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}