Hprophetnet-large / trainer_state.json
omidvaramin's picture
Upload 7 files
b2db47d
raw
history blame contribute delete
No virus
24.9 kB
{
"best_metric": 2.340223550796509,
"best_model_checkpoint": "results/models/prophetnet-large-uncased-cnndm-NewsRoom/checkpoint-97402",
"epoch": 2.0,
"global_step": 97402,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1.9948871686413012e-05,
"loss": 3.115,
"step": 500
},
{
"epoch": 0.02,
"learning_rate": 1.9897640705529663e-05,
"loss": 2.9868,
"step": 1000
},
{
"epoch": 0.03,
"learning_rate": 1.9846307057349954e-05,
"loss": 2.9327,
"step": 1500
},
{
"epoch": 0.04,
"learning_rate": 1.9794973409170245e-05,
"loss": 2.9211,
"step": 2000
},
{
"epoch": 0.05,
"learning_rate": 1.9743639760990536e-05,
"loss": 2.9204,
"step": 2500
},
{
"epoch": 0.06,
"learning_rate": 1.9692306112810827e-05,
"loss": 2.8705,
"step": 3000
},
{
"epoch": 0.07,
"learning_rate": 1.9640972464631118e-05,
"loss": 2.8643,
"step": 3500
},
{
"epoch": 0.08,
"learning_rate": 1.958974148374777e-05,
"loss": 2.8546,
"step": 4000
},
{
"epoch": 0.09,
"learning_rate": 1.953840783556806e-05,
"loss": 2.8215,
"step": 4500
},
{
"epoch": 0.1,
"learning_rate": 1.948707418738835e-05,
"loss": 2.8266,
"step": 5000
},
{
"epoch": 0.11,
"learning_rate": 1.9435740539208642e-05,
"loss": 2.7956,
"step": 5500
},
{
"epoch": 0.12,
"learning_rate": 1.9384509558325293e-05,
"loss": 2.807,
"step": 6000
},
{
"epoch": 0.13,
"learning_rate": 1.9333175910145584e-05,
"loss": 2.8149,
"step": 6500
},
{
"epoch": 0.14,
"learning_rate": 1.9281842261965875e-05,
"loss": 2.8123,
"step": 7000
},
{
"epoch": 0.15,
"learning_rate": 1.9230508613786166e-05,
"loss": 2.7988,
"step": 7500
},
{
"epoch": 0.16,
"learning_rate": 1.9179174965606457e-05,
"loss": 2.7968,
"step": 8000
},
{
"epoch": 0.17,
"learning_rate": 1.9128046652019467e-05,
"loss": 2.8055,
"step": 8500
},
{
"epoch": 0.18,
"learning_rate": 1.9076815671136118e-05,
"loss": 2.7762,
"step": 9000
},
{
"epoch": 0.2,
"learning_rate": 1.902548202295641e-05,
"loss": 2.7965,
"step": 9500
},
{
"epoch": 0.21,
"learning_rate": 1.89741483747767e-05,
"loss": 2.7588,
"step": 10000
},
{
"epoch": 0.22,
"learning_rate": 1.892281472659699e-05,
"loss": 2.7493,
"step": 10500
},
{
"epoch": 0.23,
"learning_rate": 1.8871481078417282e-05,
"loss": 2.741,
"step": 11000
},
{
"epoch": 0.24,
"learning_rate": 1.8820147430237573e-05,
"loss": 2.7593,
"step": 11500
},
{
"epoch": 0.25,
"learning_rate": 1.8768813782057864e-05,
"loss": 2.7392,
"step": 12000
},
{
"epoch": 0.26,
"learning_rate": 1.8717685468470875e-05,
"loss": 2.7177,
"step": 12500
},
{
"epoch": 0.27,
"learning_rate": 1.8666351820291166e-05,
"loss": 2.726,
"step": 13000
},
{
"epoch": 0.28,
"learning_rate": 1.8615018172111457e-05,
"loss": 2.7165,
"step": 13500
},
{
"epoch": 0.29,
"learning_rate": 1.8563684523931748e-05,
"loss": 2.7333,
"step": 14000
},
{
"epoch": 0.3,
"learning_rate": 1.851235087575204e-05,
"loss": 2.725,
"step": 14500
},
{
"epoch": 0.31,
"learning_rate": 1.846101722757233e-05,
"loss": 2.728,
"step": 15000
},
{
"epoch": 0.32,
"learning_rate": 1.840968357939262e-05,
"loss": 2.7335,
"step": 15500
},
{
"epoch": 0.33,
"learning_rate": 1.8358349931212912e-05,
"loss": 2.7167,
"step": 16000
},
{
"epoch": 0.34,
"learning_rate": 1.8307016283033203e-05,
"loss": 2.7273,
"step": 16500
},
{
"epoch": 0.35,
"learning_rate": 1.8255682634853494e-05,
"loss": 2.6938,
"step": 17000
},
{
"epoch": 0.36,
"learning_rate": 1.8204348986673785e-05,
"loss": 2.7068,
"step": 17500
},
{
"epoch": 0.37,
"learning_rate": 1.8153015338494076e-05,
"loss": 2.7108,
"step": 18000
},
{
"epoch": 0.38,
"learning_rate": 1.8101681690314367e-05,
"loss": 2.6894,
"step": 18500
},
{
"epoch": 0.39,
"learning_rate": 1.805045070943102e-05,
"loss": 2.7194,
"step": 19000
},
{
"epoch": 0.4,
"learning_rate": 1.799911706125131e-05,
"loss": 2.6978,
"step": 19500
},
{
"epoch": 0.41,
"learning_rate": 1.7947783413071603e-05,
"loss": 2.6842,
"step": 20000
},
{
"epoch": 0.42,
"learning_rate": 1.789644976489189e-05,
"loss": 2.7017,
"step": 20500
},
{
"epoch": 0.43,
"learning_rate": 1.7845116116712185e-05,
"loss": 2.6924,
"step": 21000
},
{
"epoch": 0.44,
"learning_rate": 1.7793782468532473e-05,
"loss": 2.7021,
"step": 21500
},
{
"epoch": 0.45,
"learning_rate": 1.7742551487649127e-05,
"loss": 2.6904,
"step": 22000
},
{
"epoch": 0.46,
"learning_rate": 1.7691217839469418e-05,
"loss": 2.6983,
"step": 22500
},
{
"epoch": 0.47,
"learning_rate": 1.7639986858586066e-05,
"loss": 2.7012,
"step": 23000
},
{
"epoch": 0.48,
"learning_rate": 1.758865321040636e-05,
"loss": 2.6839,
"step": 23500
},
{
"epoch": 0.49,
"learning_rate": 1.7537319562226648e-05,
"loss": 2.6706,
"step": 24000
},
{
"epoch": 0.5,
"learning_rate": 1.7486088581343302e-05,
"loss": 2.6806,
"step": 24500
},
{
"epoch": 0.51,
"learning_rate": 1.7434857600459953e-05,
"loss": 2.6741,
"step": 25000
},
{
"epoch": 0.52,
"learning_rate": 1.738352395228024e-05,
"loss": 2.6666,
"step": 25500
},
{
"epoch": 0.53,
"learning_rate": 1.7332190304100535e-05,
"loss": 2.6669,
"step": 26000
},
{
"epoch": 0.54,
"learning_rate": 1.7280856655920822e-05,
"loss": 2.6774,
"step": 26500
},
{
"epoch": 0.55,
"learning_rate": 1.7229523007741117e-05,
"loss": 2.6755,
"step": 27000
},
{
"epoch": 0.56,
"learning_rate": 1.7178189359561404e-05,
"loss": 2.6688,
"step": 27500
},
{
"epoch": 0.57,
"learning_rate": 1.71268557113817e-05,
"loss": 2.6464,
"step": 28000
},
{
"epoch": 0.59,
"learning_rate": 1.7075522063201986e-05,
"loss": 2.6501,
"step": 28500
},
{
"epoch": 0.6,
"learning_rate": 1.702418841502228e-05,
"loss": 2.6459,
"step": 29000
},
{
"epoch": 0.61,
"learning_rate": 1.6972854766842572e-05,
"loss": 2.6579,
"step": 29500
},
{
"epoch": 0.62,
"learning_rate": 1.6921521118662863e-05,
"loss": 2.6277,
"step": 30000
},
{
"epoch": 0.63,
"learning_rate": 1.6870187470483154e-05,
"loss": 2.6606,
"step": 30500
},
{
"epoch": 0.64,
"learning_rate": 1.6818853822303445e-05,
"loss": 2.6642,
"step": 31000
},
{
"epoch": 0.65,
"learning_rate": 1.6767520174123736e-05,
"loss": 2.6157,
"step": 31500
},
{
"epoch": 0.66,
"learning_rate": 1.6716186525944027e-05,
"loss": 2.6221,
"step": 32000
},
{
"epoch": 0.67,
"learning_rate": 1.6664852877764318e-05,
"loss": 2.6417,
"step": 32500
},
{
"epoch": 0.68,
"learning_rate": 1.661351922958461e-05,
"loss": 2.647,
"step": 33000
},
{
"epoch": 0.69,
"learning_rate": 1.656228824870126e-05,
"loss": 2.6216,
"step": 33500
},
{
"epoch": 0.7,
"learning_rate": 1.651095460052155e-05,
"loss": 2.6208,
"step": 34000
},
{
"epoch": 0.71,
"learning_rate": 1.6459620952341842e-05,
"loss": 2.6175,
"step": 34500
},
{
"epoch": 0.72,
"learning_rate": 1.6408287304162133e-05,
"loss": 2.621,
"step": 35000
},
{
"epoch": 0.73,
"learning_rate": 1.6357056323278784e-05,
"loss": 2.6139,
"step": 35500
},
{
"epoch": 0.74,
"learning_rate": 1.6305722675099075e-05,
"loss": 2.6424,
"step": 36000
},
{
"epoch": 0.75,
"learning_rate": 1.6254389026919366e-05,
"loss": 2.6405,
"step": 36500
},
{
"epoch": 0.76,
"learning_rate": 1.6203055378739657e-05,
"loss": 2.5977,
"step": 37000
},
{
"epoch": 0.77,
"learning_rate": 1.6151721730559948e-05,
"loss": 2.6184,
"step": 37500
},
{
"epoch": 0.78,
"learning_rate": 1.6100388082380242e-05,
"loss": 2.6256,
"step": 38000
},
{
"epoch": 0.79,
"learning_rate": 1.604905443420053e-05,
"loss": 2.6289,
"step": 38500
},
{
"epoch": 0.8,
"learning_rate": 1.5997720786020824e-05,
"loss": 2.6286,
"step": 39000
},
{
"epoch": 0.81,
"learning_rate": 1.594659247243383e-05,
"loss": 2.5953,
"step": 39500
},
{
"epoch": 0.82,
"learning_rate": 1.5895258824254122e-05,
"loss": 2.6369,
"step": 40000
},
{
"epoch": 0.83,
"learning_rate": 1.5843925176074413e-05,
"loss": 2.6003,
"step": 40500
},
{
"epoch": 0.84,
"learning_rate": 1.5792591527894704e-05,
"loss": 2.6032,
"step": 41000
},
{
"epoch": 0.85,
"learning_rate": 1.5741257879715e-05,
"loss": 2.6312,
"step": 41500
},
{
"epoch": 0.86,
"learning_rate": 1.5689924231535287e-05,
"loss": 2.6386,
"step": 42000
},
{
"epoch": 0.87,
"learning_rate": 1.563859058335558e-05,
"loss": 2.6112,
"step": 42500
},
{
"epoch": 0.88,
"learning_rate": 1.558725693517587e-05,
"loss": 2.6156,
"step": 43000
},
{
"epoch": 0.89,
"learning_rate": 1.5535923286996163e-05,
"loss": 2.5918,
"step": 43500
},
{
"epoch": 0.9,
"learning_rate": 1.548458963881645e-05,
"loss": 2.6063,
"step": 44000
},
{
"epoch": 0.91,
"learning_rate": 1.5433358657933105e-05,
"loss": 2.6263,
"step": 44500
},
{
"epoch": 0.92,
"learning_rate": 1.5382025009753392e-05,
"loss": 2.6083,
"step": 45000
},
{
"epoch": 0.93,
"learning_rate": 1.5330691361573687e-05,
"loss": 2.602,
"step": 45500
},
{
"epoch": 0.94,
"learning_rate": 1.5279357713393978e-05,
"loss": 2.5989,
"step": 46000
},
{
"epoch": 0.95,
"learning_rate": 1.5228024065214269e-05,
"loss": 2.5898,
"step": 46500
},
{
"epoch": 0.97,
"learning_rate": 1.5176690417034558e-05,
"loss": 2.6033,
"step": 47000
},
{
"epoch": 0.98,
"learning_rate": 1.5125356768854851e-05,
"loss": 2.5861,
"step": 47500
},
{
"epoch": 0.99,
"learning_rate": 1.50741257879715e-05,
"loss": 2.5959,
"step": 48000
},
{
"epoch": 1.0,
"learning_rate": 1.5022792139791791e-05,
"loss": 2.6006,
"step": 48500
},
{
"epoch": 1.0,
"eval_gen_len": 13.17,
"eval_loss": 2.431262493133545,
"eval_rouge1": 36.6311,
"eval_rouge2": 19.2397,
"eval_rougeL": 34.5043,
"eval_rougeLsum": 34.9589,
"eval_runtime": 12.9254,
"eval_samples_per_second": 7.737,
"eval_steps_per_second": 0.542,
"step": 48701
},
{
"epoch": 1.01,
"learning_rate": 1.4971458491612084e-05,
"loss": 2.4662,
"step": 49000
},
{
"epoch": 1.02,
"learning_rate": 1.4920124843432375e-05,
"loss": 2.3465,
"step": 49500
},
{
"epoch": 1.03,
"learning_rate": 1.4868791195252666e-05,
"loss": 2.3791,
"step": 50000
},
{
"epoch": 1.04,
"learning_rate": 1.4817457547072957e-05,
"loss": 2.3492,
"step": 50500
},
{
"epoch": 1.05,
"learning_rate": 1.4766123898893248e-05,
"loss": 2.3735,
"step": 51000
},
{
"epoch": 1.06,
"learning_rate": 1.471479025071354e-05,
"loss": 2.3669,
"step": 51500
},
{
"epoch": 1.07,
"learning_rate": 1.466345660253383e-05,
"loss": 2.3797,
"step": 52000
},
{
"epoch": 1.08,
"learning_rate": 1.4612122954354123e-05,
"loss": 2.3744,
"step": 52500
},
{
"epoch": 1.09,
"learning_rate": 1.4560891973470772e-05,
"loss": 2.3798,
"step": 53000
},
{
"epoch": 1.1,
"learning_rate": 1.4509558325291063e-05,
"loss": 2.3719,
"step": 53500
},
{
"epoch": 1.11,
"learning_rate": 1.4458224677111354e-05,
"loss": 2.3724,
"step": 54000
},
{
"epoch": 1.12,
"learning_rate": 1.4406993696228005e-05,
"loss": 2.3723,
"step": 54500
},
{
"epoch": 1.13,
"learning_rate": 1.4355660048048297e-05,
"loss": 2.3588,
"step": 55000
},
{
"epoch": 1.14,
"learning_rate": 1.4304326399868587e-05,
"loss": 2.3719,
"step": 55500
},
{
"epoch": 1.15,
"learning_rate": 1.4253095418985237e-05,
"loss": 2.3747,
"step": 56000
},
{
"epoch": 1.16,
"learning_rate": 1.4201761770805528e-05,
"loss": 2.3708,
"step": 56500
},
{
"epoch": 1.17,
"learning_rate": 1.415042812262582e-05,
"loss": 2.3665,
"step": 57000
},
{
"epoch": 1.18,
"learning_rate": 1.409909447444611e-05,
"loss": 2.3773,
"step": 57500
},
{
"epoch": 1.19,
"learning_rate": 1.4047760826266402e-05,
"loss": 2.3692,
"step": 58000
},
{
"epoch": 1.2,
"learning_rate": 1.3996427178086693e-05,
"loss": 2.3853,
"step": 58500
},
{
"epoch": 1.21,
"learning_rate": 1.3945093529906985e-05,
"loss": 2.3764,
"step": 59000
},
{
"epoch": 1.22,
"learning_rate": 1.3893759881727276e-05,
"loss": 2.3565,
"step": 59500
},
{
"epoch": 1.23,
"learning_rate": 1.3842426233547567e-05,
"loss": 2.3514,
"step": 60000
},
{
"epoch": 1.24,
"learning_rate": 1.3791092585367858e-05,
"loss": 2.3657,
"step": 60500
},
{
"epoch": 1.25,
"learning_rate": 1.373975893718815e-05,
"loss": 2.3637,
"step": 61000
},
{
"epoch": 1.26,
"learning_rate": 1.368842528900844e-05,
"loss": 2.3653,
"step": 61500
},
{
"epoch": 1.27,
"learning_rate": 1.363719430812509e-05,
"loss": 2.3765,
"step": 62000
},
{
"epoch": 1.28,
"learning_rate": 1.3585860659945382e-05,
"loss": 2.3799,
"step": 62500
},
{
"epoch": 1.29,
"learning_rate": 1.3534527011765672e-05,
"loss": 2.3681,
"step": 63000
},
{
"epoch": 1.3,
"learning_rate": 1.3483193363585964e-05,
"loss": 2.3751,
"step": 63500
},
{
"epoch": 1.31,
"learning_rate": 1.3431859715406255e-05,
"loss": 2.3787,
"step": 64000
},
{
"epoch": 1.32,
"learning_rate": 1.3380526067226546e-05,
"loss": 2.3769,
"step": 64500
},
{
"epoch": 1.33,
"learning_rate": 1.3329192419046837e-05,
"loss": 2.3727,
"step": 65000
},
{
"epoch": 1.34,
"learning_rate": 1.327796143816349e-05,
"loss": 2.3642,
"step": 65500
},
{
"epoch": 1.36,
"learning_rate": 1.322662778998378e-05,
"loss": 2.3593,
"step": 66000
},
{
"epoch": 1.37,
"learning_rate": 1.317539680910043e-05,
"loss": 2.3817,
"step": 66500
},
{
"epoch": 1.38,
"learning_rate": 1.3124063160920721e-05,
"loss": 2.3743,
"step": 67000
},
{
"epoch": 1.39,
"learning_rate": 1.3072729512741014e-05,
"loss": 2.402,
"step": 67500
},
{
"epoch": 1.4,
"learning_rate": 1.3021395864561303e-05,
"loss": 2.3731,
"step": 68000
},
{
"epoch": 1.41,
"learning_rate": 1.2970062216381596e-05,
"loss": 2.3722,
"step": 68500
},
{
"epoch": 1.42,
"learning_rate": 1.2918728568201885e-05,
"loss": 2.3703,
"step": 69000
},
{
"epoch": 1.43,
"learning_rate": 1.2867394920022178e-05,
"loss": 2.3565,
"step": 69500
},
{
"epoch": 1.44,
"learning_rate": 1.2816061271842469e-05,
"loss": 2.3842,
"step": 70000
},
{
"epoch": 1.45,
"learning_rate": 1.276472762366276e-05,
"loss": 2.3783,
"step": 70500
},
{
"epoch": 1.46,
"learning_rate": 1.2713393975483051e-05,
"loss": 2.3662,
"step": 71000
},
{
"epoch": 1.47,
"learning_rate": 1.2662060327303342e-05,
"loss": 2.3613,
"step": 71500
},
{
"epoch": 1.48,
"learning_rate": 1.2610726679123633e-05,
"loss": 2.3799,
"step": 72000
},
{
"epoch": 1.49,
"learning_rate": 1.2559495698240282e-05,
"loss": 2.3457,
"step": 72500
},
{
"epoch": 1.5,
"learning_rate": 1.2508162050060575e-05,
"loss": 2.3773,
"step": 73000
},
{
"epoch": 1.51,
"learning_rate": 1.2456931069177226e-05,
"loss": 2.3736,
"step": 73500
},
{
"epoch": 1.52,
"learning_rate": 1.2405597420997517e-05,
"loss": 2.3625,
"step": 74000
},
{
"epoch": 1.53,
"learning_rate": 1.2354263772817808e-05,
"loss": 2.3818,
"step": 74500
},
{
"epoch": 1.54,
"learning_rate": 1.2302930124638099e-05,
"loss": 2.3538,
"step": 75000
},
{
"epoch": 1.55,
"learning_rate": 1.225159647645839e-05,
"loss": 2.381,
"step": 75500
},
{
"epoch": 1.56,
"learning_rate": 1.2200262828278682e-05,
"loss": 2.3648,
"step": 76000
},
{
"epoch": 1.57,
"learning_rate": 1.2148929180098972e-05,
"loss": 2.3679,
"step": 76500
},
{
"epoch": 1.58,
"learning_rate": 1.2097595531919265e-05,
"loss": 2.3437,
"step": 77000
},
{
"epoch": 1.59,
"learning_rate": 1.2046364551035914e-05,
"loss": 2.36,
"step": 77500
},
{
"epoch": 1.6,
"learning_rate": 1.1995030902856206e-05,
"loss": 2.3563,
"step": 78000
},
{
"epoch": 1.61,
"learning_rate": 1.1943697254676496e-05,
"loss": 2.3474,
"step": 78500
},
{
"epoch": 1.62,
"learning_rate": 1.1892363606496788e-05,
"loss": 2.3577,
"step": 79000
},
{
"epoch": 1.63,
"learning_rate": 1.1841029958317078e-05,
"loss": 2.3795,
"step": 79500
},
{
"epoch": 1.64,
"learning_rate": 1.178969631013737e-05,
"loss": 2.3727,
"step": 80000
},
{
"epoch": 1.65,
"learning_rate": 1.1738362661957661e-05,
"loss": 2.3259,
"step": 80500
},
{
"epoch": 1.66,
"learning_rate": 1.1687029013777953e-05,
"loss": 2.3806,
"step": 81000
},
{
"epoch": 1.67,
"learning_rate": 1.1635798032894603e-05,
"loss": 2.3705,
"step": 81500
},
{
"epoch": 1.68,
"learning_rate": 1.1584464384714893e-05,
"loss": 2.3839,
"step": 82000
},
{
"epoch": 1.69,
"learning_rate": 1.1533130736535185e-05,
"loss": 2.3448,
"step": 82500
},
{
"epoch": 1.7,
"learning_rate": 1.1481797088355475e-05,
"loss": 2.3528,
"step": 83000
},
{
"epoch": 1.71,
"learning_rate": 1.1430463440175767e-05,
"loss": 2.3618,
"step": 83500
},
{
"epoch": 1.72,
"learning_rate": 1.1379129791996057e-05,
"loss": 2.3763,
"step": 84000
},
{
"epoch": 1.74,
"learning_rate": 1.132789881111271e-05,
"loss": 2.3391,
"step": 84500
},
{
"epoch": 1.75,
"learning_rate": 1.1276565162933e-05,
"loss": 2.3545,
"step": 85000
},
{
"epoch": 1.76,
"learning_rate": 1.1225231514753291e-05,
"loss": 2.3352,
"step": 85500
},
{
"epoch": 1.77,
"learning_rate": 1.1173897866573582e-05,
"loss": 2.3718,
"step": 86000
},
{
"epoch": 1.78,
"learning_rate": 1.1122666885690235e-05,
"loss": 2.3774,
"step": 86500
},
{
"epoch": 1.79,
"learning_rate": 1.1071333237510524e-05,
"loss": 2.3454,
"step": 87000
},
{
"epoch": 1.8,
"learning_rate": 1.1020102256627175e-05,
"loss": 2.3579,
"step": 87500
},
{
"epoch": 1.81,
"learning_rate": 1.0968871275743827e-05,
"loss": 2.3349,
"step": 88000
},
{
"epoch": 1.82,
"learning_rate": 1.0917640294860476e-05,
"loss": 2.3655,
"step": 88500
},
{
"epoch": 1.83,
"learning_rate": 1.0866306646680766e-05,
"loss": 2.3605,
"step": 89000
},
{
"epoch": 1.84,
"learning_rate": 1.0814972998501058e-05,
"loss": 2.3601,
"step": 89500
},
{
"epoch": 1.85,
"learning_rate": 1.076363935032135e-05,
"loss": 2.3622,
"step": 90000
},
{
"epoch": 1.86,
"learning_rate": 1.071230570214164e-05,
"loss": 2.3586,
"step": 90500
},
{
"epoch": 1.87,
"learning_rate": 1.0660972053961932e-05,
"loss": 2.3637,
"step": 91000
},
{
"epoch": 1.88,
"learning_rate": 1.0609638405782223e-05,
"loss": 2.3384,
"step": 91500
},
{
"epoch": 1.89,
"learning_rate": 1.0558304757602514e-05,
"loss": 2.3503,
"step": 92000
},
{
"epoch": 1.9,
"learning_rate": 1.0506971109422806e-05,
"loss": 2.3703,
"step": 92500
},
{
"epoch": 1.91,
"learning_rate": 1.0455637461243096e-05,
"loss": 2.3415,
"step": 93000
},
{
"epoch": 1.92,
"learning_rate": 1.0404303813063388e-05,
"loss": 2.3576,
"step": 93500
},
{
"epoch": 1.93,
"learning_rate": 1.0352970164883678e-05,
"loss": 2.3351,
"step": 94000
},
{
"epoch": 1.94,
"learning_rate": 1.030173918400033e-05,
"loss": 2.3393,
"step": 94500
},
{
"epoch": 1.95,
"learning_rate": 1.025040553582062e-05,
"loss": 2.3387,
"step": 95000
},
{
"epoch": 1.96,
"learning_rate": 1.019917455493727e-05,
"loss": 2.3464,
"step": 95500
},
{
"epoch": 1.97,
"learning_rate": 1.0147840906757563e-05,
"loss": 2.3427,
"step": 96000
},
{
"epoch": 1.98,
"learning_rate": 1.0096507258577854e-05,
"loss": 2.3678,
"step": 96500
},
{
"epoch": 1.99,
"learning_rate": 1.0045173610398145e-05,
"loss": 2.3497,
"step": 97000
},
{
"epoch": 2.0,
"eval_gen_len": 13.07,
"eval_loss": 2.340223550796509,
"eval_rouge1": 38.6817,
"eval_rouge2": 21.0493,
"eval_rougeL": 37.0384,
"eval_rougeLsum": 37.1377,
"eval_runtime": 8.7057,
"eval_samples_per_second": 11.487,
"eval_steps_per_second": 0.804,
"step": 97402
}
],
"max_steps": 194804,
"num_train_epochs": 4,
"total_flos": 3.4246205705319014e+18,
"trial_name": null,
"trial_params": null
}