xlmr-chatgptdetect-noisy / trainer_state.json
ANTOUN Wissam
added files
79ed695
{
"best_metric": 0.9725928597187162,
"best_model_checkpoint": "/scratch_results/multi/full-1.0_0.5_0.5/xlm-roberta/gradient_accumulation_steps-4-learning_rate-2e-05-epochs-5-lr_scheduler-linear-warmup_steps-0.1/SEED-1/checkpoint-25614",
"epoch": 4.999912165129556,
"global_step": 42690,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.6849379245724996e-07,
"loss": 0.6875,
"step": 100
},
{
"epoch": 0.02,
"learning_rate": 9.369875849144999e-07,
"loss": 0.6518,
"step": 200
},
{
"epoch": 0.04,
"learning_rate": 1.4007964394471774e-06,
"loss": 0.5616,
"step": 300
},
{
"epoch": 0.05,
"learning_rate": 1.8646052939798549e-06,
"loss": 0.4713,
"step": 400
},
{
"epoch": 0.06,
"learning_rate": 2.333099086437105e-06,
"loss": 0.4066,
"step": 500
},
{
"epoch": 0.07,
"learning_rate": 2.7969079409697823e-06,
"loss": 0.325,
"step": 600
},
{
"epoch": 0.08,
"learning_rate": 3.26071679550246e-06,
"loss": 0.2452,
"step": 700
},
{
"epoch": 0.09,
"learning_rate": 3.7292105879597097e-06,
"loss": 0.1638,
"step": 800
},
{
"epoch": 0.11,
"learning_rate": 4.1977043804169594e-06,
"loss": 0.1392,
"step": 900
},
{
"epoch": 0.12,
"learning_rate": 4.66619817287421e-06,
"loss": 0.1285,
"step": 1000
},
{
"epoch": 0.13,
"learning_rate": 5.13469196533146e-06,
"loss": 0.1207,
"step": 1100
},
{
"epoch": 0.14,
"learning_rate": 5.6031857577887095e-06,
"loss": 0.1013,
"step": 1200
},
{
"epoch": 0.15,
"learning_rate": 6.07167955024596e-06,
"loss": 0.12,
"step": 1300
},
{
"epoch": 0.16,
"learning_rate": 6.535488404778637e-06,
"loss": 0.0914,
"step": 1400
},
{
"epoch": 0.18,
"learning_rate": 7.003982197235887e-06,
"loss": 0.1162,
"step": 1500
},
{
"epoch": 0.19,
"learning_rate": 7.472475989693137e-06,
"loss": 0.1136,
"step": 1600
},
{
"epoch": 0.2,
"learning_rate": 7.940969782150387e-06,
"loss": 0.0809,
"step": 1700
},
{
"epoch": 0.21,
"learning_rate": 8.409463574607638e-06,
"loss": 0.0871,
"step": 1800
},
{
"epoch": 0.22,
"learning_rate": 8.877957367064887e-06,
"loss": 0.088,
"step": 1900
},
{
"epoch": 0.23,
"learning_rate": 9.346451159522137e-06,
"loss": 0.0985,
"step": 2000
},
{
"epoch": 0.25,
"learning_rate": 9.814944951979388e-06,
"loss": 0.0847,
"step": 2100
},
{
"epoch": 0.26,
"learning_rate": 1.0283438744436638e-05,
"loss": 0.0807,
"step": 2200
},
{
"epoch": 0.27,
"learning_rate": 1.0751932536893887e-05,
"loss": 0.092,
"step": 2300
},
{
"epoch": 0.28,
"learning_rate": 1.1220426329351138e-05,
"loss": 0.1086,
"step": 2400
},
{
"epoch": 0.29,
"learning_rate": 1.1688920121808387e-05,
"loss": 0.071,
"step": 2500
},
{
"epoch": 0.3,
"learning_rate": 1.2157413914265637e-05,
"loss": 0.105,
"step": 2600
},
{
"epoch": 0.32,
"learning_rate": 1.2625907706722886e-05,
"loss": 0.0757,
"step": 2700
},
{
"epoch": 0.33,
"learning_rate": 1.3094401499180137e-05,
"loss": 0.0639,
"step": 2800
},
{
"epoch": 0.34,
"learning_rate": 1.3562895291637387e-05,
"loss": 0.0778,
"step": 2900
},
{
"epoch": 0.35,
"learning_rate": 1.4031389084094638e-05,
"loss": 0.0774,
"step": 3000
},
{
"epoch": 0.36,
"learning_rate": 1.4499882876551887e-05,
"loss": 0.0616,
"step": 3100
},
{
"epoch": 0.37,
"learning_rate": 1.4968376669009137e-05,
"loss": 0.0723,
"step": 3200
},
{
"epoch": 0.39,
"learning_rate": 1.5436870461466388e-05,
"loss": 0.0645,
"step": 3300
},
{
"epoch": 0.4,
"learning_rate": 1.5905364253923635e-05,
"loss": 0.0496,
"step": 3400
},
{
"epoch": 0.41,
"learning_rate": 1.6373858046380886e-05,
"loss": 0.0579,
"step": 3500
},
{
"epoch": 0.42,
"learning_rate": 1.6842351838838136e-05,
"loss": 0.0513,
"step": 3600
},
{
"epoch": 0.43,
"learning_rate": 1.7310845631295387e-05,
"loss": 0.0463,
"step": 3700
},
{
"epoch": 0.45,
"learning_rate": 1.7779339423752638e-05,
"loss": 0.0603,
"step": 3800
},
{
"epoch": 0.46,
"learning_rate": 1.8247833216209888e-05,
"loss": 0.0506,
"step": 3900
},
{
"epoch": 0.47,
"learning_rate": 1.8716327008667135e-05,
"loss": 0.0607,
"step": 4000
},
{
"epoch": 0.48,
"learning_rate": 1.9184820801124386e-05,
"loss": 0.0588,
"step": 4100
},
{
"epoch": 0.49,
"learning_rate": 1.9653314593581637e-05,
"loss": 0.0486,
"step": 4200
},
{
"epoch": 0.5,
"learning_rate": 1.998646573488457e-05,
"loss": 0.0644,
"step": 4300
},
{
"epoch": 0.52,
"learning_rate": 1.9934410869055988e-05,
"loss": 0.0433,
"step": 4400
},
{
"epoch": 0.53,
"learning_rate": 1.9882356003227403e-05,
"loss": 0.0462,
"step": 4500
},
{
"epoch": 0.54,
"learning_rate": 1.983030113739882e-05,
"loss": 0.049,
"step": 4600
},
{
"epoch": 0.55,
"learning_rate": 1.9778246271570237e-05,
"loss": 0.0394,
"step": 4700
},
{
"epoch": 0.56,
"learning_rate": 1.9726191405741653e-05,
"loss": 0.0394,
"step": 4800
},
{
"epoch": 0.57,
"learning_rate": 1.967413653991307e-05,
"loss": 0.0501,
"step": 4900
},
{
"epoch": 0.59,
"learning_rate": 1.9622081674084487e-05,
"loss": 0.0421,
"step": 5000
},
{
"epoch": 0.6,
"learning_rate": 1.9570026808255903e-05,
"loss": 0.0514,
"step": 5100
},
{
"epoch": 0.61,
"learning_rate": 1.951797194242732e-05,
"loss": 0.0364,
"step": 5200
},
{
"epoch": 0.62,
"learning_rate": 1.9465917076598737e-05,
"loss": 0.0476,
"step": 5300
},
{
"epoch": 0.63,
"learning_rate": 1.9413862210770153e-05,
"loss": 0.0499,
"step": 5400
},
{
"epoch": 0.64,
"learning_rate": 1.936180734494157e-05,
"loss": 0.0449,
"step": 5500
},
{
"epoch": 0.66,
"learning_rate": 1.9309752479112987e-05,
"loss": 0.0413,
"step": 5600
},
{
"epoch": 0.67,
"learning_rate": 1.9257697613284402e-05,
"loss": 0.0274,
"step": 5700
},
{
"epoch": 0.68,
"learning_rate": 1.9205642747455818e-05,
"loss": 0.0368,
"step": 5800
},
{
"epoch": 0.69,
"learning_rate": 1.9154108430285523e-05,
"loss": 0.0424,
"step": 5900
},
{
"epoch": 0.7,
"learning_rate": 1.9102053564456938e-05,
"loss": 0.0557,
"step": 6000
},
{
"epoch": 0.71,
"learning_rate": 1.9049998698628357e-05,
"loss": 0.0355,
"step": 6100
},
{
"epoch": 0.73,
"learning_rate": 1.8997943832799772e-05,
"loss": 0.0469,
"step": 6200
},
{
"epoch": 0.74,
"learning_rate": 1.8945888966971188e-05,
"loss": 0.0576,
"step": 6300
},
{
"epoch": 0.75,
"learning_rate": 1.8893834101142607e-05,
"loss": 0.033,
"step": 6400
},
{
"epoch": 0.76,
"learning_rate": 1.8841779235314022e-05,
"loss": 0.0379,
"step": 6500
},
{
"epoch": 0.77,
"learning_rate": 1.8789724369485438e-05,
"loss": 0.0386,
"step": 6600
},
{
"epoch": 0.78,
"learning_rate": 1.8737669503656856e-05,
"loss": 0.0474,
"step": 6700
},
{
"epoch": 0.8,
"learning_rate": 1.8685614637828272e-05,
"loss": 0.0352,
"step": 6800
},
{
"epoch": 0.81,
"learning_rate": 1.863355977199969e-05,
"loss": 0.0332,
"step": 6900
},
{
"epoch": 0.82,
"learning_rate": 1.8581504906171106e-05,
"loss": 0.0357,
"step": 7000
},
{
"epoch": 0.83,
"learning_rate": 1.8529450040342522e-05,
"loss": 0.0373,
"step": 7100
},
{
"epoch": 0.84,
"learning_rate": 1.847739517451394e-05,
"loss": 0.0273,
"step": 7200
},
{
"epoch": 0.85,
"learning_rate": 1.8425340308685356e-05,
"loss": 0.0299,
"step": 7300
},
{
"epoch": 0.87,
"learning_rate": 1.837328544285677e-05,
"loss": 0.0279,
"step": 7400
},
{
"epoch": 0.88,
"learning_rate": 1.8321230577028187e-05,
"loss": 0.0321,
"step": 7500
},
{
"epoch": 0.89,
"learning_rate": 1.8269175711199606e-05,
"loss": 0.0257,
"step": 7600
},
{
"epoch": 0.9,
"learning_rate": 1.8217120845371025e-05,
"loss": 0.0223,
"step": 7700
},
{
"epoch": 0.91,
"learning_rate": 1.816506597954244e-05,
"loss": 0.0312,
"step": 7800
},
{
"epoch": 0.93,
"learning_rate": 1.8113011113713856e-05,
"loss": 0.0243,
"step": 7900
},
{
"epoch": 0.94,
"learning_rate": 1.8060956247885275e-05,
"loss": 0.0266,
"step": 8000
},
{
"epoch": 0.95,
"learning_rate": 1.800890138205669e-05,
"loss": 0.0311,
"step": 8100
},
{
"epoch": 0.96,
"learning_rate": 1.7956846516228105e-05,
"loss": 0.0175,
"step": 8200
},
{
"epoch": 0.97,
"learning_rate": 1.790479165039952e-05,
"loss": 0.0357,
"step": 8300
},
{
"epoch": 0.98,
"learning_rate": 1.785273678457094e-05,
"loss": 0.0252,
"step": 8400
},
{
"epoch": 1.0,
"learning_rate": 1.7800681918742355e-05,
"loss": 0.0317,
"step": 8500
},
{
"epoch": 1.0,
"eval_f1": 0.9491644678979771,
"eval_loss": 0.17317870259284973,
"eval_runtime": 47.425,
"eval_samples_per_second": 721.582,
"eval_steps_per_second": 90.206,
"step": 8538
},
{
"epoch": 1.01,
"learning_rate": 1.774862705291377e-05,
"loss": 0.0238,
"step": 8600
},
{
"epoch": 1.02,
"learning_rate": 1.769657218708519e-05,
"loss": 0.0224,
"step": 8700
},
{
"epoch": 1.03,
"learning_rate": 1.7644517321256605e-05,
"loss": 0.015,
"step": 8800
},
{
"epoch": 1.04,
"learning_rate": 1.7592462455428024e-05,
"loss": 0.0261,
"step": 8900
},
{
"epoch": 1.05,
"learning_rate": 1.754040758959944e-05,
"loss": 0.0234,
"step": 9000
},
{
"epoch": 1.07,
"learning_rate": 1.7488352723770855e-05,
"loss": 0.0261,
"step": 9100
},
{
"epoch": 1.08,
"learning_rate": 1.7436297857942274e-05,
"loss": 0.0168,
"step": 9200
},
{
"epoch": 1.09,
"learning_rate": 1.738424299211369e-05,
"loss": 0.0154,
"step": 9300
},
{
"epoch": 1.1,
"learning_rate": 1.7332708674943394e-05,
"loss": 0.0279,
"step": 9400
},
{
"epoch": 1.11,
"learning_rate": 1.728065380911481e-05,
"loss": 0.0219,
"step": 9500
},
{
"epoch": 1.12,
"learning_rate": 1.7228598943286225e-05,
"loss": 0.0265,
"step": 9600
},
{
"epoch": 1.14,
"learning_rate": 1.717654407745764e-05,
"loss": 0.0183,
"step": 9700
},
{
"epoch": 1.15,
"learning_rate": 1.712448921162906e-05,
"loss": 0.0254,
"step": 9800
},
{
"epoch": 1.16,
"learning_rate": 1.7072434345800475e-05,
"loss": 0.0167,
"step": 9900
},
{
"epoch": 1.17,
"learning_rate": 1.702037947997189e-05,
"loss": 0.0143,
"step": 10000
},
{
"epoch": 1.18,
"learning_rate": 1.696832461414331e-05,
"loss": 0.0197,
"step": 10100
},
{
"epoch": 1.19,
"learning_rate": 1.6916269748314725e-05,
"loss": 0.0304,
"step": 10200
},
{
"epoch": 1.21,
"learning_rate": 1.686421488248614e-05,
"loss": 0.0105,
"step": 10300
},
{
"epoch": 1.22,
"learning_rate": 1.681216001665756e-05,
"loss": 0.033,
"step": 10400
},
{
"epoch": 1.23,
"learning_rate": 1.6760105150828974e-05,
"loss": 0.0195,
"step": 10500
},
{
"epoch": 1.24,
"learning_rate": 1.6708050285000393e-05,
"loss": 0.0314,
"step": 10600
},
{
"epoch": 1.25,
"learning_rate": 1.665599541917181e-05,
"loss": 0.0163,
"step": 10700
},
{
"epoch": 1.26,
"learning_rate": 1.6603940553343224e-05,
"loss": 0.0156,
"step": 10800
},
{
"epoch": 1.28,
"learning_rate": 1.6551885687514643e-05,
"loss": 0.014,
"step": 10900
},
{
"epoch": 1.29,
"learning_rate": 1.649983082168606e-05,
"loss": 0.0128,
"step": 11000
},
{
"epoch": 1.3,
"learning_rate": 1.6447775955857474e-05,
"loss": 0.0167,
"step": 11100
},
{
"epoch": 1.31,
"learning_rate": 1.639572109002889e-05,
"loss": 0.0174,
"step": 11200
},
{
"epoch": 1.32,
"learning_rate": 1.6343666224200308e-05,
"loss": 0.0218,
"step": 11300
},
{
"epoch": 1.34,
"learning_rate": 1.6291611358371727e-05,
"loss": 0.0072,
"step": 11400
},
{
"epoch": 1.35,
"learning_rate": 1.6239556492543143e-05,
"loss": 0.0199,
"step": 11500
},
{
"epoch": 1.36,
"learning_rate": 1.6187501626714558e-05,
"loss": 0.0197,
"step": 11600
},
{
"epoch": 1.37,
"learning_rate": 1.6135446760885977e-05,
"loss": 0.0166,
"step": 11700
},
{
"epoch": 1.38,
"learning_rate": 1.6083391895057392e-05,
"loss": 0.0173,
"step": 11800
},
{
"epoch": 1.39,
"learning_rate": 1.6031337029228808e-05,
"loss": 0.0157,
"step": 11900
},
{
"epoch": 1.41,
"learning_rate": 1.5979282163400223e-05,
"loss": 0.0207,
"step": 12000
},
{
"epoch": 1.42,
"learning_rate": 1.5927227297571642e-05,
"loss": 0.0155,
"step": 12100
},
{
"epoch": 1.43,
"learning_rate": 1.5875172431743058e-05,
"loss": 0.0095,
"step": 12200
},
{
"epoch": 1.44,
"learning_rate": 1.5823117565914473e-05,
"loss": 0.0172,
"step": 12300
},
{
"epoch": 1.45,
"learning_rate": 1.5771062700085892e-05,
"loss": 0.0194,
"step": 12400
},
{
"epoch": 1.46,
"learning_rate": 1.571900783425731e-05,
"loss": 0.0145,
"step": 12500
},
{
"epoch": 1.48,
"learning_rate": 1.5666952968428726e-05,
"loss": 0.0188,
"step": 12600
},
{
"epoch": 1.49,
"learning_rate": 1.5614898102600142e-05,
"loss": 0.0149,
"step": 12700
},
{
"epoch": 1.5,
"learning_rate": 1.5562843236771557e-05,
"loss": 0.0169,
"step": 12800
},
{
"epoch": 1.51,
"learning_rate": 1.5510788370942976e-05,
"loss": 0.0167,
"step": 12900
},
{
"epoch": 1.52,
"learning_rate": 1.545873350511439e-05,
"loss": 0.0122,
"step": 13000
},
{
"epoch": 1.53,
"learning_rate": 1.5406678639285807e-05,
"loss": 0.0163,
"step": 13100
},
{
"epoch": 1.55,
"learning_rate": 1.5355144322115512e-05,
"loss": 0.0205,
"step": 13200
},
{
"epoch": 1.56,
"learning_rate": 1.5303089456286927e-05,
"loss": 0.0157,
"step": 13300
},
{
"epoch": 1.57,
"learning_rate": 1.5251034590458344e-05,
"loss": 0.0134,
"step": 13400
},
{
"epoch": 1.58,
"learning_rate": 1.5198979724629762e-05,
"loss": 0.0227,
"step": 13500
},
{
"epoch": 1.59,
"learning_rate": 1.5146924858801177e-05,
"loss": 0.0219,
"step": 13600
},
{
"epoch": 1.6,
"learning_rate": 1.5094869992972594e-05,
"loss": 0.013,
"step": 13700
},
{
"epoch": 1.62,
"learning_rate": 1.504281512714401e-05,
"loss": 0.0135,
"step": 13800
},
{
"epoch": 1.63,
"learning_rate": 1.4990760261315427e-05,
"loss": 0.0207,
"step": 13900
},
{
"epoch": 1.64,
"learning_rate": 1.4938705395486844e-05,
"loss": 0.0174,
"step": 14000
},
{
"epoch": 1.65,
"learning_rate": 1.4886650529658261e-05,
"loss": 0.006,
"step": 14100
},
{
"epoch": 1.66,
"learning_rate": 1.4834595663829678e-05,
"loss": 0.0108,
"step": 14200
},
{
"epoch": 1.67,
"learning_rate": 1.4782540798001096e-05,
"loss": 0.0123,
"step": 14300
},
{
"epoch": 1.69,
"learning_rate": 1.4730485932172511e-05,
"loss": 0.0156,
"step": 14400
},
{
"epoch": 1.7,
"learning_rate": 1.4678431066343928e-05,
"loss": 0.0262,
"step": 14500
},
{
"epoch": 1.71,
"learning_rate": 1.4626376200515344e-05,
"loss": 0.0136,
"step": 14600
},
{
"epoch": 1.72,
"learning_rate": 1.457432133468676e-05,
"loss": 0.0098,
"step": 14700
},
{
"epoch": 1.73,
"learning_rate": 1.4522266468858178e-05,
"loss": 0.0138,
"step": 14800
},
{
"epoch": 1.75,
"learning_rate": 1.4470211603029593e-05,
"loss": 0.0132,
"step": 14900
},
{
"epoch": 1.76,
"learning_rate": 1.441815673720101e-05,
"loss": 0.0106,
"step": 15000
},
{
"epoch": 1.77,
"learning_rate": 1.436610187137243e-05,
"loss": 0.015,
"step": 15100
},
{
"epoch": 1.78,
"learning_rate": 1.4314047005543845e-05,
"loss": 0.0167,
"step": 15200
},
{
"epoch": 1.79,
"learning_rate": 1.4261992139715262e-05,
"loss": 0.0084,
"step": 15300
},
{
"epoch": 1.8,
"learning_rate": 1.4209937273886678e-05,
"loss": 0.0188,
"step": 15400
},
{
"epoch": 1.82,
"learning_rate": 1.4157882408058095e-05,
"loss": 0.0088,
"step": 15500
},
{
"epoch": 1.83,
"learning_rate": 1.4106348090887796e-05,
"loss": 0.0171,
"step": 15600
},
{
"epoch": 1.84,
"learning_rate": 1.4054293225059215e-05,
"loss": 0.0168,
"step": 15700
},
{
"epoch": 1.85,
"learning_rate": 1.400223835923063e-05,
"loss": 0.009,
"step": 15800
},
{
"epoch": 1.86,
"learning_rate": 1.3950183493402048e-05,
"loss": 0.0078,
"step": 15900
},
{
"epoch": 1.87,
"learning_rate": 1.3898128627573465e-05,
"loss": 0.0108,
"step": 16000
},
{
"epoch": 1.89,
"learning_rate": 1.3846594310403166e-05,
"loss": 0.0072,
"step": 16100
},
{
"epoch": 1.9,
"learning_rate": 1.3794539444574582e-05,
"loss": 0.0089,
"step": 16200
},
{
"epoch": 1.91,
"learning_rate": 1.3742484578746e-05,
"loss": 0.0055,
"step": 16300
},
{
"epoch": 1.92,
"learning_rate": 1.3690429712917418e-05,
"loss": 0.0177,
"step": 16400
},
{
"epoch": 1.93,
"learning_rate": 1.3638374847088833e-05,
"loss": 0.008,
"step": 16500
},
{
"epoch": 1.94,
"learning_rate": 1.358631998126025e-05,
"loss": 0.0127,
"step": 16600
},
{
"epoch": 1.96,
"learning_rate": 1.3534265115431666e-05,
"loss": 0.0069,
"step": 16700
},
{
"epoch": 1.97,
"learning_rate": 1.3482210249603083e-05,
"loss": 0.0089,
"step": 16800
},
{
"epoch": 1.98,
"learning_rate": 1.3430155383774498e-05,
"loss": 0.0094,
"step": 16900
},
{
"epoch": 1.99,
"learning_rate": 1.3378100517945916e-05,
"loss": 0.008,
"step": 17000
},
{
"epoch": 2.0,
"eval_f1": 0.9269947608004809,
"eval_loss": 0.3541407883167267,
"eval_runtime": 56.6762,
"eval_samples_per_second": 603.799,
"eval_steps_per_second": 75.481,
"step": 17076
},
{
"epoch": 2.0,
"learning_rate": 1.3326045652117333e-05,
"loss": 0.0091,
"step": 17100
},
{
"epoch": 2.01,
"learning_rate": 1.3273990786288748e-05,
"loss": 0.0075,
"step": 17200
},
{
"epoch": 2.03,
"learning_rate": 1.3221935920460167e-05,
"loss": 0.0084,
"step": 17300
},
{
"epoch": 2.04,
"learning_rate": 1.3169881054631584e-05,
"loss": 0.0032,
"step": 17400
},
{
"epoch": 2.05,
"learning_rate": 1.3117826188803e-05,
"loss": 0.0113,
"step": 17500
},
{
"epoch": 2.06,
"learning_rate": 1.3065771322974417e-05,
"loss": 0.0053,
"step": 17600
},
{
"epoch": 2.07,
"learning_rate": 1.3013716457145832e-05,
"loss": 0.0077,
"step": 17700
},
{
"epoch": 2.08,
"learning_rate": 1.296166159131725e-05,
"loss": 0.0058,
"step": 17800
},
{
"epoch": 2.1,
"learning_rate": 1.2909606725488667e-05,
"loss": 0.0085,
"step": 17900
},
{
"epoch": 2.11,
"learning_rate": 1.2857551859660082e-05,
"loss": 0.0029,
"step": 18000
},
{
"epoch": 2.12,
"learning_rate": 1.28054969938315e-05,
"loss": 0.0059,
"step": 18100
},
{
"epoch": 2.13,
"learning_rate": 1.2753442128002915e-05,
"loss": 0.0079,
"step": 18200
},
{
"epoch": 2.14,
"learning_rate": 1.2701387262174332e-05,
"loss": 0.004,
"step": 18300
},
{
"epoch": 2.16,
"learning_rate": 1.264933239634575e-05,
"loss": 0.0096,
"step": 18400
},
{
"epoch": 2.17,
"learning_rate": 1.2597277530517166e-05,
"loss": 0.0063,
"step": 18500
},
{
"epoch": 2.18,
"learning_rate": 1.2545222664688583e-05,
"loss": 0.0029,
"step": 18600
},
{
"epoch": 2.19,
"learning_rate": 1.249316779886e-05,
"loss": 0.0119,
"step": 18700
},
{
"epoch": 2.2,
"learning_rate": 1.2441112933031416e-05,
"loss": 0.0088,
"step": 18800
},
{
"epoch": 2.21,
"learning_rate": 1.2389058067202833e-05,
"loss": 0.004,
"step": 18900
},
{
"epoch": 2.23,
"learning_rate": 1.2337003201374249e-05,
"loss": 0.0066,
"step": 19000
},
{
"epoch": 2.24,
"learning_rate": 1.2284948335545666e-05,
"loss": 0.0058,
"step": 19100
},
{
"epoch": 2.25,
"learning_rate": 1.2232893469717081e-05,
"loss": 0.005,
"step": 19200
},
{
"epoch": 2.26,
"learning_rate": 1.2180838603888498e-05,
"loss": 0.0121,
"step": 19300
},
{
"epoch": 2.27,
"learning_rate": 1.2128783738059917e-05,
"loss": 0.0039,
"step": 19400
},
{
"epoch": 2.28,
"learning_rate": 1.2076728872231333e-05,
"loss": 0.0053,
"step": 19500
},
{
"epoch": 2.3,
"learning_rate": 1.202467400640275e-05,
"loss": 0.0041,
"step": 19600
},
{
"epoch": 2.31,
"learning_rate": 1.1972619140574167e-05,
"loss": 0.019,
"step": 19700
},
{
"epoch": 2.32,
"learning_rate": 1.1920564274745583e-05,
"loss": 0.0065,
"step": 19800
},
{
"epoch": 2.33,
"learning_rate": 1.1868509408917e-05,
"loss": 0.0099,
"step": 19900
},
{
"epoch": 2.34,
"learning_rate": 1.1816454543088415e-05,
"loss": 0.0029,
"step": 20000
},
{
"epoch": 2.35,
"learning_rate": 1.1764399677259832e-05,
"loss": 0.0045,
"step": 20100
},
{
"epoch": 2.37,
"learning_rate": 1.1712865360089535e-05,
"loss": 0.013,
"step": 20200
},
{
"epoch": 2.38,
"learning_rate": 1.1660810494260953e-05,
"loss": 0.003,
"step": 20300
},
{
"epoch": 2.39,
"learning_rate": 1.1608755628432368e-05,
"loss": 0.0062,
"step": 20400
},
{
"epoch": 2.4,
"learning_rate": 1.1556700762603785e-05,
"loss": 0.0054,
"step": 20500
},
{
"epoch": 2.41,
"learning_rate": 1.1504645896775202e-05,
"loss": 0.0123,
"step": 20600
},
{
"epoch": 2.42,
"learning_rate": 1.1452591030946618e-05,
"loss": 0.0095,
"step": 20700
},
{
"epoch": 2.44,
"learning_rate": 1.1400536165118035e-05,
"loss": 0.009,
"step": 20800
},
{
"epoch": 2.45,
"learning_rate": 1.134848129928945e-05,
"loss": 0.0061,
"step": 20900
},
{
"epoch": 2.46,
"learning_rate": 1.129642643346087e-05,
"loss": 0.0005,
"step": 21000
},
{
"epoch": 2.47,
"learning_rate": 1.1244371567632287e-05,
"loss": 0.0015,
"step": 21100
},
{
"epoch": 2.48,
"learning_rate": 1.1192316701803702e-05,
"loss": 0.0019,
"step": 21200
},
{
"epoch": 2.49,
"learning_rate": 1.114026183597512e-05,
"loss": 0.0027,
"step": 21300
},
{
"epoch": 2.51,
"learning_rate": 1.1088206970146536e-05,
"loss": 0.004,
"step": 21400
},
{
"epoch": 2.52,
"learning_rate": 1.1036152104317952e-05,
"loss": 0.0075,
"step": 21500
},
{
"epoch": 2.53,
"learning_rate": 1.0984097238489369e-05,
"loss": 0.007,
"step": 21600
},
{
"epoch": 2.54,
"learning_rate": 1.0932042372660784e-05,
"loss": 0.0039,
"step": 21700
},
{
"epoch": 2.55,
"learning_rate": 1.0879987506832202e-05,
"loss": 0.0145,
"step": 21800
},
{
"epoch": 2.56,
"learning_rate": 1.0827932641003617e-05,
"loss": 0.0071,
"step": 21900
},
{
"epoch": 2.58,
"learning_rate": 1.0775877775175034e-05,
"loss": 0.0168,
"step": 22000
},
{
"epoch": 2.59,
"learning_rate": 1.0723822909346453e-05,
"loss": 0.0095,
"step": 22100
},
{
"epoch": 2.6,
"learning_rate": 1.0671768043517869e-05,
"loss": 0.0033,
"step": 22200
},
{
"epoch": 2.61,
"learning_rate": 1.0619713177689286e-05,
"loss": 0.0022,
"step": 22300
},
{
"epoch": 2.62,
"learning_rate": 1.0567658311860703e-05,
"loss": 0.0022,
"step": 22400
},
{
"epoch": 2.64,
"learning_rate": 1.0515603446032118e-05,
"loss": 0.0082,
"step": 22500
},
{
"epoch": 2.65,
"learning_rate": 1.0463548580203536e-05,
"loss": 0.007,
"step": 22600
},
{
"epoch": 2.66,
"learning_rate": 1.0411493714374951e-05,
"loss": 0.0043,
"step": 22700
},
{
"epoch": 2.67,
"learning_rate": 1.0359438848546368e-05,
"loss": 0.0109,
"step": 22800
},
{
"epoch": 2.68,
"learning_rate": 1.0307383982717785e-05,
"loss": 0.0023,
"step": 22900
},
{
"epoch": 2.69,
"learning_rate": 1.02553291168892e-05,
"loss": 0.0037,
"step": 23000
},
{
"epoch": 2.71,
"learning_rate": 1.0203794799718904e-05,
"loss": 0.0072,
"step": 23100
},
{
"epoch": 2.72,
"learning_rate": 1.0151739933890321e-05,
"loss": 0.0096,
"step": 23200
},
{
"epoch": 2.73,
"learning_rate": 1.0099685068061738e-05,
"loss": 0.0072,
"step": 23300
},
{
"epoch": 2.74,
"learning_rate": 1.0047630202233154e-05,
"loss": 0.0021,
"step": 23400
},
{
"epoch": 2.75,
"learning_rate": 9.99557533640457e-06,
"loss": 0.0075,
"step": 23500
},
{
"epoch": 2.76,
"learning_rate": 9.943520470575988e-06,
"loss": 0.0071,
"step": 23600
},
{
"epoch": 2.78,
"learning_rate": 9.891465604747405e-06,
"loss": 0.0028,
"step": 23700
},
{
"epoch": 2.79,
"learning_rate": 9.83941073891882e-06,
"loss": 0.0083,
"step": 23800
},
{
"epoch": 2.8,
"learning_rate": 9.787355873090238e-06,
"loss": 0.0028,
"step": 23900
},
{
"epoch": 2.81,
"learning_rate": 9.735301007261655e-06,
"loss": 0.0016,
"step": 24000
},
{
"epoch": 2.82,
"learning_rate": 9.683246141433072e-06,
"loss": 0.0117,
"step": 24100
},
{
"epoch": 2.83,
"learning_rate": 9.631191275604488e-06,
"loss": 0.0124,
"step": 24200
},
{
"epoch": 2.85,
"learning_rate": 9.579136409775905e-06,
"loss": 0.0033,
"step": 24300
},
{
"epoch": 2.86,
"learning_rate": 9.52708154394732e-06,
"loss": 0.005,
"step": 24400
},
{
"epoch": 2.87,
"learning_rate": 9.475026678118739e-06,
"loss": 0.0001,
"step": 24500
},
{
"epoch": 2.88,
"learning_rate": 9.422971812290155e-06,
"loss": 0.0025,
"step": 24600
},
{
"epoch": 2.89,
"learning_rate": 9.370916946461572e-06,
"loss": 0.0158,
"step": 24700
},
{
"epoch": 2.9,
"learning_rate": 9.318862080632987e-06,
"loss": 0.0042,
"step": 24800
},
{
"epoch": 2.92,
"learning_rate": 9.266807214804404e-06,
"loss": 0.0035,
"step": 24900
},
{
"epoch": 2.93,
"learning_rate": 9.214752348975822e-06,
"loss": 0.0024,
"step": 25000
},
{
"epoch": 2.94,
"learning_rate": 9.162697483147239e-06,
"loss": 0.0053,
"step": 25100
},
{
"epoch": 2.95,
"learning_rate": 9.11116316597694e-06,
"loss": 0.0064,
"step": 25200
},
{
"epoch": 2.96,
"learning_rate": 9.059108300148357e-06,
"loss": 0.0013,
"step": 25300
},
{
"epoch": 2.97,
"learning_rate": 9.007053434319773e-06,
"loss": 0.0058,
"step": 25400
},
{
"epoch": 2.99,
"learning_rate": 8.95499856849119e-06,
"loss": 0.0029,
"step": 25500
},
{
"epoch": 3.0,
"learning_rate": 8.902943702662607e-06,
"loss": 0.0085,
"step": 25600
},
{
"epoch": 3.0,
"eval_f1": 0.9725928597187162,
"eval_loss": 0.11609108746051788,
"eval_runtime": 47.4182,
"eval_samples_per_second": 721.684,
"eval_steps_per_second": 90.218,
"step": 25614
},
{
"epoch": 3.01,
"learning_rate": 8.850888836834024e-06,
"loss": 0.0008,
"step": 25700
},
{
"epoch": 3.02,
"learning_rate": 8.79883397100544e-06,
"loss": 0.0086,
"step": 25800
},
{
"epoch": 3.03,
"learning_rate": 8.746779105176857e-06,
"loss": 0.0001,
"step": 25900
},
{
"epoch": 3.05,
"learning_rate": 8.694724239348274e-06,
"loss": 0.0,
"step": 26000
},
{
"epoch": 3.06,
"learning_rate": 8.642669373519691e-06,
"loss": 0.0,
"step": 26100
},
{
"epoch": 3.07,
"learning_rate": 8.590614507691107e-06,
"loss": 0.0,
"step": 26200
},
{
"epoch": 3.08,
"learning_rate": 8.538559641862524e-06,
"loss": 0.0027,
"step": 26300
},
{
"epoch": 3.09,
"learning_rate": 8.486504776033941e-06,
"loss": 0.0,
"step": 26400
},
{
"epoch": 3.1,
"learning_rate": 8.434449910205356e-06,
"loss": 0.0043,
"step": 26500
},
{
"epoch": 3.12,
"learning_rate": 8.382395044376774e-06,
"loss": 0.004,
"step": 26600
},
{
"epoch": 3.13,
"learning_rate": 8.33034017854819e-06,
"loss": 0.0,
"step": 26700
},
{
"epoch": 3.14,
"learning_rate": 8.278285312719608e-06,
"loss": 0.0016,
"step": 26800
},
{
"epoch": 3.15,
"learning_rate": 8.226230446891023e-06,
"loss": 0.0,
"step": 26900
},
{
"epoch": 3.16,
"learning_rate": 8.17417558106244e-06,
"loss": 0.0117,
"step": 27000
},
{
"epoch": 3.17,
"learning_rate": 8.122120715233858e-06,
"loss": 0.0046,
"step": 27100
},
{
"epoch": 3.19,
"learning_rate": 8.07058639806356e-06,
"loss": 0.0025,
"step": 27200
},
{
"epoch": 3.2,
"learning_rate": 8.018531532234976e-06,
"loss": 0.0022,
"step": 27300
},
{
"epoch": 3.21,
"learning_rate": 7.966476666406393e-06,
"loss": 0.011,
"step": 27400
},
{
"epoch": 3.22,
"learning_rate": 7.914421800577809e-06,
"loss": 0.007,
"step": 27500
},
{
"epoch": 3.23,
"learning_rate": 7.862366934749226e-06,
"loss": 0.0081,
"step": 27600
},
{
"epoch": 3.24,
"learning_rate": 7.810312068920643e-06,
"loss": 0.0032,
"step": 27700
},
{
"epoch": 3.26,
"learning_rate": 7.75825720309206e-06,
"loss": 0.0038,
"step": 27800
},
{
"epoch": 3.27,
"learning_rate": 7.706202337263476e-06,
"loss": 0.003,
"step": 27900
},
{
"epoch": 3.28,
"learning_rate": 7.654147471434893e-06,
"loss": 0.007,
"step": 28000
},
{
"epoch": 3.29,
"learning_rate": 7.602092605606309e-06,
"loss": 0.0004,
"step": 28100
},
{
"epoch": 3.3,
"learning_rate": 7.550037739777726e-06,
"loss": 0.0039,
"step": 28200
},
{
"epoch": 3.31,
"learning_rate": 7.497982873949144e-06,
"loss": 0.0,
"step": 28300
},
{
"epoch": 3.33,
"learning_rate": 7.44592800812056e-06,
"loss": 0.003,
"step": 28400
},
{
"epoch": 3.34,
"learning_rate": 7.393873142291976e-06,
"loss": 0.0018,
"step": 28500
},
{
"epoch": 3.35,
"learning_rate": 7.341818276463393e-06,
"loss": 0.0011,
"step": 28600
},
{
"epoch": 3.36,
"learning_rate": 7.289763410634809e-06,
"loss": 0.0013,
"step": 28700
},
{
"epoch": 3.37,
"learning_rate": 7.237708544806227e-06,
"loss": 0.0028,
"step": 28800
},
{
"epoch": 3.38,
"learning_rate": 7.185653678977643e-06,
"loss": 0.0029,
"step": 28900
},
{
"epoch": 3.4,
"learning_rate": 7.13359881314906e-06,
"loss": 0.0009,
"step": 29000
},
{
"epoch": 3.41,
"learning_rate": 7.081543947320476e-06,
"loss": 0.0,
"step": 29100
},
{
"epoch": 3.42,
"learning_rate": 7.029489081491892e-06,
"loss": 0.0,
"step": 29200
},
{
"epoch": 3.43,
"learning_rate": 6.97743421566331e-06,
"loss": 0.0029,
"step": 29300
},
{
"epoch": 3.44,
"learning_rate": 6.9253793498347266e-06,
"loss": 0.0,
"step": 29400
},
{
"epoch": 3.46,
"learning_rate": 6.873324484006143e-06,
"loss": 0.0043,
"step": 29500
},
{
"epoch": 3.47,
"learning_rate": 6.821269618177559e-06,
"loss": 0.0041,
"step": 29600
},
{
"epoch": 3.48,
"learning_rate": 6.769214752348976e-06,
"loss": 0.0,
"step": 29700
},
{
"epoch": 3.49,
"learning_rate": 6.7171598865203935e-06,
"loss": 0.0035,
"step": 29800
},
{
"epoch": 3.5,
"learning_rate": 6.66510502069181e-06,
"loss": 0.0028,
"step": 29900
},
{
"epoch": 3.51,
"learning_rate": 6.613050154863226e-06,
"loss": 0.0031,
"step": 30000
},
{
"epoch": 3.53,
"learning_rate": 6.5609952890346425e-06,
"loss": 0.003,
"step": 30100
},
{
"epoch": 3.54,
"learning_rate": 6.50894042320606e-06,
"loss": 0.0013,
"step": 30200
},
{
"epoch": 3.55,
"learning_rate": 6.456885557377477e-06,
"loss": 0.0087,
"step": 30300
},
{
"epoch": 3.56,
"learning_rate": 6.404830691548893e-06,
"loss": 0.0034,
"step": 30400
},
{
"epoch": 3.57,
"learning_rate": 6.3527758257203095e-06,
"loss": 0.0015,
"step": 30500
},
{
"epoch": 3.58,
"learning_rate": 6.300720959891727e-06,
"loss": 0.0015,
"step": 30600
},
{
"epoch": 3.6,
"learning_rate": 6.248666094063143e-06,
"loss": 0.0014,
"step": 30700
},
{
"epoch": 3.61,
"learning_rate": 6.196611228234559e-06,
"loss": 0.0017,
"step": 30800
},
{
"epoch": 3.62,
"learning_rate": 6.1445563624059764e-06,
"loss": 0.0024,
"step": 30900
},
{
"epoch": 3.63,
"learning_rate": 6.092501496577394e-06,
"loss": 0.0046,
"step": 31000
},
{
"epoch": 3.64,
"learning_rate": 6.04044663074881e-06,
"loss": 0.0048,
"step": 31100
},
{
"epoch": 3.65,
"learning_rate": 5.988391764920226e-06,
"loss": 0.0,
"step": 31200
},
{
"epoch": 3.67,
"learning_rate": 5.9368574477499284e-06,
"loss": 0.0005,
"step": 31300
},
{
"epoch": 3.68,
"learning_rate": 5.8848025819213464e-06,
"loss": 0.0,
"step": 31400
},
{
"epoch": 3.69,
"learning_rate": 5.832747716092763e-06,
"loss": 0.0,
"step": 31500
},
{
"epoch": 3.7,
"learning_rate": 5.781213398922465e-06,
"loss": 0.0054,
"step": 31600
},
{
"epoch": 3.71,
"learning_rate": 5.729158533093881e-06,
"loss": 0.0066,
"step": 31700
},
{
"epoch": 3.72,
"learning_rate": 5.677103667265298e-06,
"loss": 0.0017,
"step": 31800
},
{
"epoch": 3.74,
"learning_rate": 5.625048801436715e-06,
"loss": 0.0017,
"step": 31900
},
{
"epoch": 3.75,
"learning_rate": 5.572993935608132e-06,
"loss": 0.003,
"step": 32000
},
{
"epoch": 3.76,
"learning_rate": 5.520939069779548e-06,
"loss": 0.0,
"step": 32100
},
{
"epoch": 3.77,
"learning_rate": 5.4688842039509646e-06,
"loss": 0.0011,
"step": 32200
},
{
"epoch": 3.78,
"learning_rate": 5.417349886780668e-06,
"loss": 0.0018,
"step": 32300
},
{
"epoch": 3.79,
"learning_rate": 5.365295020952084e-06,
"loss": 0.0023,
"step": 32400
},
{
"epoch": 3.81,
"learning_rate": 5.3132401551235e-06,
"loss": 0.0,
"step": 32500
},
{
"epoch": 3.82,
"learning_rate": 5.261185289294917e-06,
"loss": 0.0017,
"step": 32600
},
{
"epoch": 3.83,
"learning_rate": 5.209130423466335e-06,
"loss": 0.0,
"step": 32700
},
{
"epoch": 3.84,
"learning_rate": 5.157075557637751e-06,
"loss": 0.0001,
"step": 32800
},
{
"epoch": 3.85,
"learning_rate": 5.105020691809167e-06,
"loss": 0.0019,
"step": 32900
},
{
"epoch": 3.86,
"learning_rate": 5.0529658259805835e-06,
"loss": 0.0042,
"step": 33000
},
{
"epoch": 3.88,
"learning_rate": 5.0009109601520015e-06,
"loss": 0.0013,
"step": 33100
},
{
"epoch": 3.89,
"learning_rate": 4.948856094323417e-06,
"loss": 0.0009,
"step": 33200
},
{
"epoch": 3.9,
"learning_rate": 4.896801228494834e-06,
"loss": 0.0021,
"step": 33300
},
{
"epoch": 3.91,
"learning_rate": 4.8447463626662505e-06,
"loss": 0.0031,
"step": 33400
},
{
"epoch": 3.92,
"learning_rate": 4.792691496837668e-06,
"loss": 0.0019,
"step": 33500
},
{
"epoch": 3.94,
"learning_rate": 4.740636631009084e-06,
"loss": 0.0,
"step": 33600
},
{
"epoch": 3.95,
"learning_rate": 4.6885817651805e-06,
"loss": 0.0,
"step": 33700
},
{
"epoch": 3.96,
"learning_rate": 4.6365268993519175e-06,
"loss": 0.0,
"step": 33800
},
{
"epoch": 3.97,
"learning_rate": 4.584472033523334e-06,
"loss": 0.0088,
"step": 33900
},
{
"epoch": 3.98,
"learning_rate": 4.53241716769475e-06,
"loss": 0.0021,
"step": 34000
},
{
"epoch": 3.99,
"learning_rate": 4.480362301866167e-06,
"loss": 0.0015,
"step": 34100
},
{
"epoch": 4.0,
"eval_f1": 0.9515538902358385,
"eval_loss": 0.25566887855529785,
"eval_runtime": 47.5026,
"eval_samples_per_second": 720.403,
"eval_steps_per_second": 90.058,
"step": 34152
},
{
"epoch": 4.01,
"learning_rate": 4.428307436037584e-06,
"loss": 0.0,
"step": 34200
},
{
"epoch": 4.02,
"learning_rate": 4.376773118867287e-06,
"loss": 0.0004,
"step": 34300
},
{
"epoch": 4.03,
"learning_rate": 4.324718253038703e-06,
"loss": 0.0,
"step": 34400
},
{
"epoch": 4.04,
"learning_rate": 4.27266338721012e-06,
"loss": 0.0,
"step": 34500
},
{
"epoch": 4.05,
"learning_rate": 4.2206085213815364e-06,
"loss": 0.0,
"step": 34600
},
{
"epoch": 4.06,
"learning_rate": 4.168553655552954e-06,
"loss": 0.0,
"step": 34700
},
{
"epoch": 4.08,
"learning_rate": 4.11649878972437e-06,
"loss": 0.0,
"step": 34800
},
{
"epoch": 4.09,
"learning_rate": 4.064443923895787e-06,
"loss": 0.0,
"step": 34900
},
{
"epoch": 4.1,
"learning_rate": 4.012389058067203e-06,
"loss": 0.0006,
"step": 35000
},
{
"epoch": 4.11,
"learning_rate": 3.96033419223862e-06,
"loss": 0.0024,
"step": 35100
},
{
"epoch": 4.12,
"learning_rate": 3.908279326410037e-06,
"loss": 0.0024,
"step": 35200
},
{
"epoch": 4.13,
"learning_rate": 3.856224460581453e-06,
"loss": 0.0,
"step": 35300
},
{
"epoch": 4.15,
"learning_rate": 3.80416959475287e-06,
"loss": 0.0,
"step": 35400
},
{
"epoch": 4.16,
"learning_rate": 3.7521147289242867e-06,
"loss": 0.0,
"step": 35500
},
{
"epoch": 4.17,
"learning_rate": 3.700059863095703e-06,
"loss": 0.0,
"step": 35600
},
{
"epoch": 4.18,
"learning_rate": 3.64800499726712e-06,
"loss": 0.0,
"step": 35700
},
{
"epoch": 4.19,
"learning_rate": 3.5959501314385365e-06,
"loss": 0.0,
"step": 35800
},
{
"epoch": 4.2,
"learning_rate": 3.5438952656099537e-06,
"loss": 0.0013,
"step": 35900
},
{
"epoch": 4.22,
"learning_rate": 3.49184039978137e-06,
"loss": 0.002,
"step": 36000
},
{
"epoch": 4.23,
"learning_rate": 3.4397855339527863e-06,
"loss": 0.0,
"step": 36100
},
{
"epoch": 4.24,
"learning_rate": 3.3877306681242035e-06,
"loss": 0.0015,
"step": 36200
},
{
"epoch": 4.25,
"learning_rate": 3.3356758022956198e-06,
"loss": 0.0,
"step": 36300
},
{
"epoch": 4.26,
"learning_rate": 3.283620936467036e-06,
"loss": 0.0,
"step": 36400
},
{
"epoch": 4.27,
"learning_rate": 3.2315660706384533e-06,
"loss": 0.0021,
"step": 36500
},
{
"epoch": 4.29,
"learning_rate": 3.1795112048098696e-06,
"loss": 0.0,
"step": 36600
},
{
"epoch": 4.3,
"learning_rate": 3.1274563389812867e-06,
"loss": 0.0,
"step": 36700
},
{
"epoch": 4.31,
"learning_rate": 3.075401473152703e-06,
"loss": 0.0014,
"step": 36800
},
{
"epoch": 4.32,
"learning_rate": 3.02334660732412e-06,
"loss": 0.0037,
"step": 36900
},
{
"epoch": 4.33,
"learning_rate": 2.9712917414955366e-06,
"loss": 0.0,
"step": 37000
},
{
"epoch": 4.35,
"learning_rate": 2.9192368756669533e-06,
"loss": 0.0013,
"step": 37100
},
{
"epoch": 4.36,
"learning_rate": 2.8677025584966555e-06,
"loss": 0.0011,
"step": 37200
},
{
"epoch": 4.37,
"learning_rate": 2.8156476926680727e-06,
"loss": 0.0031,
"step": 37300
},
{
"epoch": 4.38,
"learning_rate": 2.763592826839489e-06,
"loss": 0.0012,
"step": 37400
},
{
"epoch": 4.39,
"learning_rate": 2.7115379610109053e-06,
"loss": 0.0,
"step": 37500
},
{
"epoch": 4.4,
"learning_rate": 2.6594830951823225e-06,
"loss": 0.0,
"step": 37600
},
{
"epoch": 4.42,
"learning_rate": 2.6074282293537388e-06,
"loss": 0.0,
"step": 37700
},
{
"epoch": 4.43,
"learning_rate": 2.555373363525156e-06,
"loss": 0.0012,
"step": 37800
},
{
"epoch": 4.44,
"learning_rate": 2.5033184976965723e-06,
"loss": 0.0,
"step": 37900
},
{
"epoch": 4.45,
"learning_rate": 2.451263631867989e-06,
"loss": 0.0033,
"step": 38000
},
{
"epoch": 4.46,
"learning_rate": 2.3992087660394057e-06,
"loss": 0.0,
"step": 38100
},
{
"epoch": 4.47,
"learning_rate": 2.3471539002108225e-06,
"loss": 0.0,
"step": 38200
},
{
"epoch": 4.49,
"learning_rate": 2.2950990343822392e-06,
"loss": 0.0019,
"step": 38300
},
{
"epoch": 4.5,
"learning_rate": 2.243044168553656e-06,
"loss": 0.0,
"step": 38400
},
{
"epoch": 4.51,
"learning_rate": 2.1909893027250727e-06,
"loss": 0.0005,
"step": 38500
},
{
"epoch": 4.52,
"learning_rate": 2.138934436896489e-06,
"loss": 0.0018,
"step": 38600
},
{
"epoch": 4.53,
"learning_rate": 2.0868795710679058e-06,
"loss": 0.0,
"step": 38700
},
{
"epoch": 4.54,
"learning_rate": 2.0348247052393225e-06,
"loss": 0.0018,
"step": 38800
},
{
"epoch": 4.56,
"learning_rate": 1.9827698394107393e-06,
"loss": 0.0,
"step": 38900
},
{
"epoch": 4.57,
"learning_rate": 1.9307149735821556e-06,
"loss": 0.0,
"step": 39000
},
{
"epoch": 4.58,
"learning_rate": 1.8786601077535723e-06,
"loss": 0.0,
"step": 39100
},
{
"epoch": 4.59,
"learning_rate": 1.826605241924989e-06,
"loss": 0.0,
"step": 39200
},
{
"epoch": 4.6,
"learning_rate": 1.7745503760964058e-06,
"loss": 0.0,
"step": 39300
},
{
"epoch": 4.61,
"learning_rate": 1.7224955102678223e-06,
"loss": 0.0,
"step": 39400
},
{
"epoch": 4.63,
"learning_rate": 1.670440644439239e-06,
"loss": 0.0,
"step": 39500
},
{
"epoch": 4.64,
"learning_rate": 1.6189063272689415e-06,
"loss": 0.0018,
"step": 39600
},
{
"epoch": 4.65,
"learning_rate": 1.5668514614403582e-06,
"loss": 0.0,
"step": 39700
},
{
"epoch": 4.66,
"learning_rate": 1.514796595611775e-06,
"loss": 0.0011,
"step": 39800
},
{
"epoch": 4.67,
"learning_rate": 1.4627417297831917e-06,
"loss": 0.0,
"step": 39900
},
{
"epoch": 4.68,
"learning_rate": 1.4106868639546085e-06,
"loss": 0.0,
"step": 40000
},
{
"epoch": 4.7,
"learning_rate": 1.3586319981260248e-06,
"loss": 0.0,
"step": 40100
},
{
"epoch": 4.71,
"learning_rate": 1.3065771322974415e-06,
"loss": 0.0013,
"step": 40200
},
{
"epoch": 4.72,
"learning_rate": 1.2545222664688583e-06,
"loss": 0.0,
"step": 40300
},
{
"epoch": 4.73,
"learning_rate": 1.202467400640275e-06,
"loss": 0.0,
"step": 40400
},
{
"epoch": 4.74,
"learning_rate": 1.1504125348116917e-06,
"loss": 0.0,
"step": 40500
},
{
"epoch": 4.76,
"learning_rate": 1.0983576689831083e-06,
"loss": 0.0,
"step": 40600
},
{
"epoch": 4.77,
"learning_rate": 1.046302803154525e-06,
"loss": 0.0,
"step": 40700
},
{
"epoch": 4.78,
"learning_rate": 9.942479373259415e-07,
"loss": 0.0012,
"step": 40800
},
{
"epoch": 4.79,
"learning_rate": 9.421930714973583e-07,
"loss": 0.0012,
"step": 40900
},
{
"epoch": 4.8,
"learning_rate": 8.901382056687749e-07,
"loss": 0.0,
"step": 41000
},
{
"epoch": 4.81,
"learning_rate": 8.380833398401917e-07,
"loss": 0.0013,
"step": 41100
},
{
"epoch": 4.83,
"learning_rate": 7.860284740116084e-07,
"loss": 0.0,
"step": 41200
},
{
"epoch": 4.84,
"learning_rate": 7.339736081830249e-07,
"loss": 0.0,
"step": 41300
},
{
"epoch": 4.85,
"learning_rate": 6.819187423544417e-07,
"loss": 0.0,
"step": 41400
},
{
"epoch": 4.86,
"learning_rate": 6.298638765258583e-07,
"loss": 0.0001,
"step": 41500
},
{
"epoch": 4.87,
"learning_rate": 5.77809010697275e-07,
"loss": 0.0,
"step": 41600
},
{
"epoch": 4.88,
"learning_rate": 5.257541448686916e-07,
"loss": 0.0008,
"step": 41700
},
{
"epoch": 4.9,
"learning_rate": 4.736992790401083e-07,
"loss": 0.0,
"step": 41800
},
{
"epoch": 4.91,
"learning_rate": 4.21644413211525e-07,
"loss": 0.0,
"step": 41900
},
{
"epoch": 4.92,
"learning_rate": 3.6958954738294165e-07,
"loss": 0.0,
"step": 42000
},
{
"epoch": 4.93,
"learning_rate": 3.1753468155435834e-07,
"loss": 0.0,
"step": 42100
},
{
"epoch": 4.94,
"learning_rate": 2.65479815725775e-07,
"loss": 0.0015,
"step": 42200
},
{
"epoch": 4.95,
"learning_rate": 2.1342494989719166e-07,
"loss": 0.0,
"step": 42300
},
{
"epoch": 4.97,
"learning_rate": 1.6137008406860833e-07,
"loss": 0.0,
"step": 42400
},
{
"epoch": 4.98,
"learning_rate": 1.0931521824002499e-07,
"loss": 0.0,
"step": 42500
},
{
"epoch": 4.99,
"learning_rate": 5.7260352411441666e-08,
"loss": 0.0,
"step": 42600
},
{
"epoch": 5.0,
"eval_f1": 0.9649930701479859,
"eval_loss": 0.22864048182964325,
"eval_runtime": 47.3601,
"eval_samples_per_second": 722.57,
"eval_steps_per_second": 90.329,
"step": 42690
},
{
"epoch": 5.0,
"step": 42690,
"total_flos": 8.986242362554368e+16,
"train_loss": 0.023754316836012614,
"train_runtime": 9371.3538,
"train_samples_per_second": 145.781,
"train_steps_per_second": 4.555
}
],
"max_steps": 42690,
"num_train_epochs": 5,
"total_flos": 8.986242362554368e+16,
"trial_name": null,
"trial_params": null
}