psychiq / checkpoint-107000 /trainer_state.json
derenrich's picture
End of training
0de621f
raw history blame
No virus
26.6 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.28621104721144847,
"global_step": 107000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.9973251303998932e-05,
"loss": 3.5144,
"step": 500
},
{
"epoch": 0.0,
"learning_rate": 1.9946502607997863e-05,
"loss": 2.2536,
"step": 1000
},
{
"epoch": 0.0,
"learning_rate": 1.991975391199679e-05,
"loss": 2.0076,
"step": 1500
},
{
"epoch": 0.01,
"learning_rate": 1.9893005215995724e-05,
"loss": 1.77,
"step": 2000
},
{
"epoch": 0.01,
"learning_rate": 1.986625651999465e-05,
"loss": 1.6057,
"step": 2500
},
{
"epoch": 0.01,
"learning_rate": 1.983950782399358e-05,
"loss": 1.4921,
"step": 3000
},
{
"epoch": 0.01,
"learning_rate": 1.981275912799251e-05,
"loss": 1.4723,
"step": 3500
},
{
"epoch": 0.01,
"learning_rate": 1.9786010431991442e-05,
"loss": 1.3909,
"step": 4000
},
{
"epoch": 0.01,
"learning_rate": 1.9759261735990373e-05,
"loss": 1.3073,
"step": 4500
},
{
"epoch": 0.01,
"learning_rate": 1.9732513039989303e-05,
"loss": 1.1996,
"step": 5000
},
{
"epoch": 0.01,
"learning_rate": 1.970576434398823e-05,
"loss": 1.1979,
"step": 5500
},
{
"epoch": 0.02,
"learning_rate": 1.9679015647987164e-05,
"loss": 1.1541,
"step": 6000
},
{
"epoch": 0.02,
"learning_rate": 1.965226695198609e-05,
"loss": 1.1183,
"step": 6500
},
{
"epoch": 0.02,
"learning_rate": 1.962551825598502e-05,
"loss": 1.0842,
"step": 7000
},
{
"epoch": 0.02,
"learning_rate": 1.9598769559983952e-05,
"loss": 1.0402,
"step": 7500
},
{
"epoch": 0.02,
"learning_rate": 1.9572020863982883e-05,
"loss": 1.0074,
"step": 8000
},
{
"epoch": 0.02,
"learning_rate": 1.9545272167981813e-05,
"loss": 1.0338,
"step": 8500
},
{
"epoch": 0.02,
"learning_rate": 1.951852347198074e-05,
"loss": 0.9924,
"step": 9000
},
{
"epoch": 0.03,
"learning_rate": 1.9491774775979674e-05,
"loss": 0.9631,
"step": 9500
},
{
"epoch": 0.03,
"learning_rate": 1.94650260799786e-05,
"loss": 0.9394,
"step": 10000
},
{
"epoch": 0.03,
"learning_rate": 1.943827738397753e-05,
"loss": 0.9232,
"step": 10500
},
{
"epoch": 0.03,
"learning_rate": 1.9411528687976462e-05,
"loss": 0.8787,
"step": 11000
},
{
"epoch": 0.03,
"learning_rate": 1.9384779991975393e-05,
"loss": 0.8995,
"step": 11500
},
{
"epoch": 0.03,
"learning_rate": 1.9358031295974323e-05,
"loss": 0.8935,
"step": 12000
},
{
"epoch": 0.03,
"learning_rate": 1.9331282599973254e-05,
"loss": 0.8847,
"step": 12500
},
{
"epoch": 0.03,
"learning_rate": 1.9304533903972184e-05,
"loss": 0.8572,
"step": 13000
},
{
"epoch": 0.04,
"learning_rate": 1.9277785207971115e-05,
"loss": 0.8075,
"step": 13500
},
{
"epoch": 0.04,
"learning_rate": 1.925103651197004e-05,
"loss": 0.8435,
"step": 14000
},
{
"epoch": 0.04,
"learning_rate": 1.9224287815968975e-05,
"loss": 0.8183,
"step": 14500
},
{
"epoch": 0.04,
"learning_rate": 1.9197539119967903e-05,
"loss": 0.8101,
"step": 15000
},
{
"epoch": 0.04,
"learning_rate": 1.9170790423966833e-05,
"loss": 0.7885,
"step": 15500
},
{
"epoch": 0.04,
"learning_rate": 1.9144041727965764e-05,
"loss": 0.7753,
"step": 16000
},
{
"epoch": 0.04,
"learning_rate": 1.9117293031964694e-05,
"loss": 0.797,
"step": 16500
},
{
"epoch": 0.05,
"learning_rate": 1.9090544335963624e-05,
"loss": 0.784,
"step": 17000
},
{
"epoch": 0.05,
"learning_rate": 1.906379563996255e-05,
"loss": 0.7849,
"step": 17500
},
{
"epoch": 0.05,
"learning_rate": 1.9037046943961482e-05,
"loss": 0.7358,
"step": 18000
},
{
"epoch": 0.05,
"learning_rate": 1.9010298247960413e-05,
"loss": 0.7414,
"step": 18500
},
{
"epoch": 0.05,
"learning_rate": 1.8983549551959343e-05,
"loss": 0.7606,
"step": 19000
},
{
"epoch": 0.05,
"learning_rate": 1.8956800855958274e-05,
"loss": 0.7404,
"step": 19500
},
{
"epoch": 0.05,
"learning_rate": 1.8930052159957204e-05,
"loss": 0.7261,
"step": 20000
},
{
"epoch": 0.05,
"learning_rate": 1.8903303463956134e-05,
"loss": 0.7436,
"step": 20500
},
{
"epoch": 0.06,
"learning_rate": 1.8876554767955065e-05,
"loss": 0.7183,
"step": 21000
},
{
"epoch": 0.06,
"learning_rate": 1.8849806071953992e-05,
"loss": 0.7043,
"step": 21500
},
{
"epoch": 0.06,
"learning_rate": 1.8823057375952926e-05,
"loss": 0.7424,
"step": 22000
},
{
"epoch": 0.06,
"learning_rate": 1.8796308679951853e-05,
"loss": 0.7286,
"step": 22500
},
{
"epoch": 0.06,
"learning_rate": 1.8769559983950784e-05,
"loss": 0.7385,
"step": 23000
},
{
"epoch": 0.06,
"learning_rate": 1.8742811287949714e-05,
"loss": 0.7158,
"step": 23500
},
{
"epoch": 0.06,
"learning_rate": 1.8716062591948644e-05,
"loss": 0.7042,
"step": 24000
},
{
"epoch": 0.07,
"learning_rate": 1.8689313895947575e-05,
"loss": 0.6892,
"step": 24500
},
{
"epoch": 0.07,
"learning_rate": 1.8662565199946505e-05,
"loss": 0.729,
"step": 25000
},
{
"epoch": 0.07,
"learning_rate": 1.8635816503945433e-05,
"loss": 0.6994,
"step": 25500
},
{
"epoch": 0.07,
"learning_rate": 1.8609067807944363e-05,
"loss": 0.6768,
"step": 26000
},
{
"epoch": 0.07,
"learning_rate": 1.8582319111943294e-05,
"loss": 0.6858,
"step": 26500
},
{
"epoch": 0.07,
"learning_rate": 1.8555570415942224e-05,
"loss": 0.693,
"step": 27000
},
{
"epoch": 0.07,
"learning_rate": 1.8528821719941154e-05,
"loss": 0.6863,
"step": 27500
},
{
"epoch": 0.07,
"learning_rate": 1.8502073023940085e-05,
"loss": 0.6703,
"step": 28000
},
{
"epoch": 0.08,
"learning_rate": 1.8475324327939015e-05,
"loss": 0.6785,
"step": 28500
},
{
"epoch": 0.08,
"learning_rate": 1.8448575631937943e-05,
"loss": 0.6847,
"step": 29000
},
{
"epoch": 0.08,
"learning_rate": 1.8421826935936876e-05,
"loss": 0.6583,
"step": 29500
},
{
"epoch": 0.08,
"learning_rate": 1.8395078239935803e-05,
"loss": 0.6659,
"step": 30000
},
{
"epoch": 0.08,
"learning_rate": 1.8368329543934734e-05,
"loss": 0.7055,
"step": 30500
},
{
"epoch": 0.08,
"learning_rate": 1.8341580847933664e-05,
"loss": 0.6751,
"step": 31000
},
{
"epoch": 0.08,
"learning_rate": 1.8314832151932595e-05,
"loss": 0.6628,
"step": 31500
},
{
"epoch": 0.09,
"learning_rate": 1.8288083455931525e-05,
"loss": 0.6568,
"step": 32000
},
{
"epoch": 0.09,
"learning_rate": 1.8261334759930456e-05,
"loss": 0.694,
"step": 32500
},
{
"epoch": 0.09,
"learning_rate": 1.8234586063929386e-05,
"loss": 0.663,
"step": 33000
},
{
"epoch": 0.09,
"learning_rate": 1.8207837367928317e-05,
"loss": 0.6523,
"step": 33500
},
{
"epoch": 0.09,
"learning_rate": 1.8181088671927244e-05,
"loss": 0.6485,
"step": 34000
},
{
"epoch": 0.09,
"learning_rate": 1.8154339975926174e-05,
"loss": 0.6472,
"step": 34500
},
{
"epoch": 0.09,
"learning_rate": 1.8127591279925105e-05,
"loss": 0.6857,
"step": 35000
},
{
"epoch": 0.09,
"learning_rate": 1.8100842583924035e-05,
"loss": 0.6471,
"step": 35500
},
{
"epoch": 0.1,
"learning_rate": 1.8074093887922966e-05,
"loss": 0.6687,
"step": 36000
},
{
"epoch": 0.1,
"learning_rate": 1.8047345191921893e-05,
"loss": 0.6438,
"step": 36500
},
{
"epoch": 0.1,
"learning_rate": 1.8020596495920827e-05,
"loss": 0.6579,
"step": 37000
},
{
"epoch": 0.1,
"learning_rate": 1.7993847799919754e-05,
"loss": 0.657,
"step": 37500
},
{
"epoch": 0.1,
"learning_rate": 1.7967099103918684e-05,
"loss": 0.6459,
"step": 38000
},
{
"epoch": 0.1,
"learning_rate": 1.7940350407917615e-05,
"loss": 0.644,
"step": 38500
},
{
"epoch": 0.1,
"learning_rate": 1.7913601711916545e-05,
"loss": 0.6419,
"step": 39000
},
{
"epoch": 0.11,
"learning_rate": 1.7886853015915476e-05,
"loss": 0.6392,
"step": 39500
},
{
"epoch": 0.11,
"learning_rate": 1.7860104319914406e-05,
"loss": 0.6446,
"step": 40000
},
{
"epoch": 0.11,
"learning_rate": 1.7833355623913337e-05,
"loss": 0.6546,
"step": 40500
},
{
"epoch": 0.11,
"learning_rate": 1.7806606927912267e-05,
"loss": 0.6212,
"step": 41000
},
{
"epoch": 0.11,
"learning_rate": 1.7779858231911194e-05,
"loss": 0.6387,
"step": 41500
},
{
"epoch": 0.11,
"learning_rate": 1.7753109535910128e-05,
"loss": 0.6368,
"step": 42000
},
{
"epoch": 0.11,
"learning_rate": 1.7726360839909055e-05,
"loss": 0.6526,
"step": 42500
},
{
"epoch": 0.12,
"learning_rate": 1.7699612143907986e-05,
"loss": 0.6342,
"step": 43000
},
{
"epoch": 0.12,
"learning_rate": 1.7672863447906916e-05,
"loss": 0.6234,
"step": 43500
},
{
"epoch": 0.12,
"learning_rate": 1.7646114751905843e-05,
"loss": 0.6028,
"step": 44000
},
{
"epoch": 0.12,
"learning_rate": 1.7619366055904777e-05,
"loss": 0.6556,
"step": 44500
},
{
"epoch": 0.12,
"learning_rate": 1.7592617359903704e-05,
"loss": 0.625,
"step": 45000
},
{
"epoch": 0.12,
"learning_rate": 1.7565868663902635e-05,
"loss": 0.6198,
"step": 45500
},
{
"epoch": 0.12,
"learning_rate": 1.7539119967901565e-05,
"loss": 0.6538,
"step": 46000
},
{
"epoch": 0.12,
"learning_rate": 1.7512371271900496e-05,
"loss": 0.6265,
"step": 46500
},
{
"epoch": 0.13,
"learning_rate": 1.7485622575899426e-05,
"loss": 0.6375,
"step": 47000
},
{
"epoch": 0.13,
"learning_rate": 1.7458873879898357e-05,
"loss": 0.6325,
"step": 47500
},
{
"epoch": 0.13,
"learning_rate": 1.7432125183897287e-05,
"loss": 0.6258,
"step": 48000
},
{
"epoch": 0.13,
"learning_rate": 1.7405376487896218e-05,
"loss": 0.6235,
"step": 48500
},
{
"epoch": 0.13,
"learning_rate": 1.7378627791895145e-05,
"loss": 0.6386,
"step": 49000
},
{
"epoch": 0.13,
"learning_rate": 1.735187909589408e-05,
"loss": 0.6243,
"step": 49500
},
{
"epoch": 0.13,
"learning_rate": 1.7325130399893006e-05,
"loss": 0.6471,
"step": 50000
},
{
"epoch": 0.14,
"learning_rate": 1.7298381703891936e-05,
"loss": 0.6396,
"step": 50500
},
{
"epoch": 0.14,
"learning_rate": 1.7271633007890867e-05,
"loss": 0.6164,
"step": 51000
},
{
"epoch": 0.14,
"learning_rate": 1.7244884311889797e-05,
"loss": 0.5981,
"step": 51500
},
{
"epoch": 0.14,
"learning_rate": 1.7218135615888728e-05,
"loss": 0.6505,
"step": 52000
},
{
"epoch": 0.14,
"learning_rate": 1.7191386919887655e-05,
"loss": 0.651,
"step": 52500
},
{
"epoch": 0.14,
"learning_rate": 1.716463822388659e-05,
"loss": 0.629,
"step": 53000
},
{
"epoch": 0.14,
"learning_rate": 1.7137889527885516e-05,
"loss": 0.6393,
"step": 53500
},
{
"epoch": 0.14,
"learning_rate": 1.7111140831884446e-05,
"loss": 0.6168,
"step": 54000
},
{
"epoch": 0.15,
"learning_rate": 1.7084392135883377e-05,
"loss": 0.6253,
"step": 54500
},
{
"epoch": 0.15,
"learning_rate": 1.7057643439882307e-05,
"loss": 0.625,
"step": 55000
},
{
"epoch": 0.15,
"learning_rate": 1.7030894743881238e-05,
"loss": 0.6131,
"step": 55500
},
{
"epoch": 0.15,
"learning_rate": 1.7004146047880168e-05,
"loss": 0.6018,
"step": 56000
},
{
"epoch": 0.15,
"learning_rate": 1.6977397351879095e-05,
"loss": 0.6389,
"step": 56500
},
{
"epoch": 0.15,
"learning_rate": 1.695064865587803e-05,
"loss": 0.6068,
"step": 57000
},
{
"epoch": 0.15,
"learning_rate": 1.6923899959876956e-05,
"loss": 0.6238,
"step": 57500
},
{
"epoch": 0.16,
"learning_rate": 1.6897151263875887e-05,
"loss": 0.6017,
"step": 58000
},
{
"epoch": 0.16,
"learning_rate": 1.6870402567874817e-05,
"loss": 0.6052,
"step": 58500
},
{
"epoch": 0.16,
"learning_rate": 1.6843653871873748e-05,
"loss": 0.6021,
"step": 59000
},
{
"epoch": 0.16,
"learning_rate": 1.6816905175872678e-05,
"loss": 0.6168,
"step": 59500
},
{
"epoch": 0.16,
"learning_rate": 1.6790156479871605e-05,
"loss": 0.6086,
"step": 60000
},
{
"epoch": 0.16,
"learning_rate": 1.676340778387054e-05,
"loss": 0.5905,
"step": 60500
},
{
"epoch": 0.16,
"learning_rate": 1.6736659087869466e-05,
"loss": 0.6268,
"step": 61000
},
{
"epoch": 0.16,
"learning_rate": 1.6709910391868397e-05,
"loss": 0.626,
"step": 61500
},
{
"epoch": 0.17,
"learning_rate": 1.6683161695867327e-05,
"loss": 0.631,
"step": 62000
},
{
"epoch": 0.17,
"learning_rate": 1.6656412999866258e-05,
"loss": 0.5936,
"step": 62500
},
{
"epoch": 0.17,
"learning_rate": 1.6629664303865188e-05,
"loss": 0.6137,
"step": 63000
},
{
"epoch": 0.17,
"learning_rate": 1.660291560786412e-05,
"loss": 0.6185,
"step": 63500
},
{
"epoch": 0.17,
"learning_rate": 1.6576166911863046e-05,
"loss": 0.6009,
"step": 64000
},
{
"epoch": 0.17,
"learning_rate": 1.654941821586198e-05,
"loss": 0.6038,
"step": 64500
},
{
"epoch": 0.17,
"learning_rate": 1.6522669519860907e-05,
"loss": 0.6081,
"step": 65000
},
{
"epoch": 0.18,
"learning_rate": 1.6495920823859837e-05,
"loss": 0.6032,
"step": 65500
},
{
"epoch": 0.18,
"learning_rate": 1.6469172127858768e-05,
"loss": 0.6166,
"step": 66000
},
{
"epoch": 0.18,
"learning_rate": 1.6442423431857698e-05,
"loss": 0.6125,
"step": 66500
},
{
"epoch": 0.18,
"learning_rate": 1.641567473585663e-05,
"loss": 0.5948,
"step": 67000
},
{
"epoch": 0.18,
"learning_rate": 1.638892603985556e-05,
"loss": 0.5957,
"step": 67500
},
{
"epoch": 0.18,
"learning_rate": 1.636217734385449e-05,
"loss": 0.5965,
"step": 68000
},
{
"epoch": 0.18,
"learning_rate": 1.6335428647853417e-05,
"loss": 0.5997,
"step": 68500
},
{
"epoch": 0.18,
"learning_rate": 1.6308679951852347e-05,
"loss": 0.5736,
"step": 69000
},
{
"epoch": 0.19,
"learning_rate": 1.6281931255851278e-05,
"loss": 0.6116,
"step": 69500
},
{
"epoch": 0.19,
"learning_rate": 1.6255182559850208e-05,
"loss": 0.6081,
"step": 70000
},
{
"epoch": 0.19,
"learning_rate": 1.622843386384914e-05,
"loss": 0.5891,
"step": 70500
},
{
"epoch": 0.19,
"learning_rate": 1.620168516784807e-05,
"loss": 0.5942,
"step": 71000
},
{
"epoch": 0.19,
"learning_rate": 1.6174936471847e-05,
"loss": 0.6041,
"step": 71500
},
{
"epoch": 0.19,
"learning_rate": 1.614818777584593e-05,
"loss": 0.5901,
"step": 72000
},
{
"epoch": 0.19,
"learning_rate": 1.6121439079844857e-05,
"loss": 0.5926,
"step": 72500
},
{
"epoch": 0.2,
"learning_rate": 1.609469038384379e-05,
"loss": 0.6103,
"step": 73000
},
{
"epoch": 0.2,
"learning_rate": 1.6067941687842718e-05,
"loss": 0.6269,
"step": 73500
},
{
"epoch": 0.2,
"learning_rate": 1.604119299184165e-05,
"loss": 0.5939,
"step": 74000
},
{
"epoch": 0.2,
"learning_rate": 1.601444429584058e-05,
"loss": 0.6014,
"step": 74500
},
{
"epoch": 0.2,
"learning_rate": 1.598769559983951e-05,
"loss": 0.5945,
"step": 75000
},
{
"epoch": 0.2,
"learning_rate": 1.596094690383844e-05,
"loss": 0.6079,
"step": 75500
},
{
"epoch": 0.2,
"learning_rate": 1.593419820783737e-05,
"loss": 0.5995,
"step": 76000
},
{
"epoch": 0.2,
"learning_rate": 1.5907449511836298e-05,
"loss": 0.5761,
"step": 76500
},
{
"epoch": 0.21,
"learning_rate": 1.5880700815835228e-05,
"loss": 0.6073,
"step": 77000
},
{
"epoch": 0.21,
"learning_rate": 1.585395211983416e-05,
"loss": 0.6005,
"step": 77500
},
{
"epoch": 0.21,
"learning_rate": 1.582720342383309e-05,
"loss": 0.6059,
"step": 78000
},
{
"epoch": 0.21,
"learning_rate": 1.580045472783202e-05,
"loss": 0.5678,
"step": 78500
},
{
"epoch": 0.21,
"learning_rate": 1.577370603183095e-05,
"loss": 0.6119,
"step": 79000
},
{
"epoch": 0.21,
"learning_rate": 1.574695733582988e-05,
"loss": 0.5892,
"step": 79500
},
{
"epoch": 0.21,
"learning_rate": 1.5720208639828808e-05,
"loss": 0.5889,
"step": 80000
},
{
"epoch": 0.22,
"learning_rate": 1.569345994382774e-05,
"loss": 0.5922,
"step": 80500
},
{
"epoch": 0.22,
"learning_rate": 1.566671124782667e-05,
"loss": 0.582,
"step": 81000
},
{
"epoch": 0.22,
"learning_rate": 1.56399625518256e-05,
"loss": 0.5695,
"step": 81500
},
{
"epoch": 0.22,
"learning_rate": 1.561321385582453e-05,
"loss": 0.5804,
"step": 82000
},
{
"epoch": 0.22,
"learning_rate": 1.558646515982346e-05,
"loss": 0.5809,
"step": 82500
},
{
"epoch": 0.22,
"learning_rate": 1.555971646382239e-05,
"loss": 0.5828,
"step": 83000
},
{
"epoch": 0.22,
"learning_rate": 1.553296776782132e-05,
"loss": 0.5754,
"step": 83500
},
{
"epoch": 0.22,
"learning_rate": 1.5506219071820248e-05,
"loss": 0.5772,
"step": 84000
},
{
"epoch": 0.23,
"learning_rate": 1.5479470375819182e-05,
"loss": 0.5668,
"step": 84500
},
{
"epoch": 0.23,
"learning_rate": 1.545272167981811e-05,
"loss": 0.5886,
"step": 85000
},
{
"epoch": 0.23,
"learning_rate": 1.542597298381704e-05,
"loss": 0.5698,
"step": 85500
},
{
"epoch": 0.23,
"learning_rate": 1.539922428781597e-05,
"loss": 0.5891,
"step": 86000
},
{
"epoch": 0.23,
"learning_rate": 1.53724755918149e-05,
"loss": 0.5818,
"step": 86500
},
{
"epoch": 0.23,
"learning_rate": 1.534572689581383e-05,
"loss": 0.58,
"step": 87000
},
{
"epoch": 0.23,
"learning_rate": 1.5318978199812758e-05,
"loss": 0.5969,
"step": 87500
},
{
"epoch": 0.24,
"learning_rate": 1.5292229503811692e-05,
"loss": 0.579,
"step": 88000
},
{
"epoch": 0.24,
"learning_rate": 1.526548080781062e-05,
"loss": 0.5803,
"step": 88500
},
{
"epoch": 0.24,
"learning_rate": 1.5238732111809551e-05,
"loss": 0.5756,
"step": 89000
},
{
"epoch": 0.24,
"learning_rate": 1.521198341580848e-05,
"loss": 0.5794,
"step": 89500
},
{
"epoch": 0.24,
"learning_rate": 1.518523471980741e-05,
"loss": 0.5718,
"step": 90000
},
{
"epoch": 0.24,
"learning_rate": 1.5158486023806341e-05,
"loss": 0.5849,
"step": 90500
},
{
"epoch": 0.24,
"learning_rate": 1.5131737327805272e-05,
"loss": 0.5704,
"step": 91000
},
{
"epoch": 0.24,
"learning_rate": 1.51049886318042e-05,
"loss": 0.5817,
"step": 91500
},
{
"epoch": 0.25,
"learning_rate": 1.5078239935803133e-05,
"loss": 0.5716,
"step": 92000
},
{
"epoch": 0.25,
"learning_rate": 1.5051491239802061e-05,
"loss": 0.5957,
"step": 92500
},
{
"epoch": 0.25,
"learning_rate": 1.5024742543800992e-05,
"loss": 0.5686,
"step": 93000
},
{
"epoch": 0.25,
"learning_rate": 1.499799384779992e-05,
"loss": 0.5764,
"step": 93500
},
{
"epoch": 0.25,
"learning_rate": 1.497124515179885e-05,
"loss": 0.5837,
"step": 94000
},
{
"epoch": 0.25,
"learning_rate": 1.4944496455797782e-05,
"loss": 0.5668,
"step": 94500
},
{
"epoch": 0.25,
"learning_rate": 1.491774775979671e-05,
"loss": 0.579,
"step": 95000
},
{
"epoch": 0.26,
"learning_rate": 1.489099906379564e-05,
"loss": 0.5738,
"step": 95500
},
{
"epoch": 0.26,
"learning_rate": 1.486425036779457e-05,
"loss": 0.5675,
"step": 96000
},
{
"epoch": 0.26,
"learning_rate": 1.4837501671793502e-05,
"loss": 0.5844,
"step": 96500
},
{
"epoch": 0.26,
"learning_rate": 1.481075297579243e-05,
"loss": 0.5907,
"step": 97000
},
{
"epoch": 0.26,
"learning_rate": 1.4784004279791361e-05,
"loss": 0.5837,
"step": 97500
},
{
"epoch": 0.26,
"learning_rate": 1.4757255583790292e-05,
"loss": 0.5634,
"step": 98000
},
{
"epoch": 0.26,
"learning_rate": 1.4730506887789222e-05,
"loss": 0.5743,
"step": 98500
},
{
"epoch": 0.26,
"learning_rate": 1.470375819178815e-05,
"loss": 0.5821,
"step": 99000
},
{
"epoch": 0.27,
"learning_rate": 1.4677009495787083e-05,
"loss": 0.5779,
"step": 99500
},
{
"epoch": 0.27,
"learning_rate": 1.4650260799786012e-05,
"loss": 0.569,
"step": 100000
},
{
"epoch": 0.27,
"learning_rate": 1.4623512103784942e-05,
"loss": 0.5907,
"step": 100500
},
{
"epoch": 0.27,
"learning_rate": 1.4596763407783871e-05,
"loss": 0.5851,
"step": 101000
},
{
"epoch": 0.27,
"learning_rate": 1.4570014711782803e-05,
"loss": 0.5933,
"step": 101500
},
{
"epoch": 0.27,
"learning_rate": 1.4543266015781732e-05,
"loss": 0.5823,
"step": 102000
},
{
"epoch": 0.27,
"learning_rate": 1.451651731978066e-05,
"loss": 0.5701,
"step": 102500
},
{
"epoch": 0.28,
"learning_rate": 1.4489768623779591e-05,
"loss": 0.5872,
"step": 103000
},
{
"epoch": 0.28,
"learning_rate": 1.4463019927778522e-05,
"loss": 0.5675,
"step": 103500
},
{
"epoch": 0.28,
"learning_rate": 1.4436271231777452e-05,
"loss": 0.5736,
"step": 104000
},
{
"epoch": 0.28,
"learning_rate": 1.4409522535776381e-05,
"loss": 0.5574,
"step": 104500
},
{
"epoch": 0.28,
"learning_rate": 1.4382773839775313e-05,
"loss": 0.5631,
"step": 105000
},
{
"epoch": 0.28,
"learning_rate": 1.4356025143774242e-05,
"loss": 0.5662,
"step": 105500
},
{
"epoch": 0.28,
"learning_rate": 1.4329276447773172e-05,
"loss": 0.5702,
"step": 106000
},
{
"epoch": 0.28,
"learning_rate": 1.4302527751772101e-05,
"loss": 0.5507,
"step": 106500
},
{
"epoch": 0.29,
"learning_rate": 1.4275779055771033e-05,
"loss": 0.5718,
"step": 107000
}
],
"max_steps": 373850,
"num_train_epochs": 1,
"total_flos": 4.269171558690125e+16,
"trial_name": null,
"trial_params": null
}