split_test_model / trainer_state.json
hxl520
Update from split_test_model
dc42f7b
raw
history blame contribute delete
No virus
22.7 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"global_step": 90939,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02,
"learning_rate": 4.9725090445243514e-05,
"loss": 2.5887,
"step": 500
},
{
"epoch": 0.03,
"learning_rate": 4.945018089048703e-05,
"loss": 2.5306,
"step": 1000
},
{
"epoch": 0.05,
"learning_rate": 4.917527133573055e-05,
"loss": 2.5434,
"step": 1500
},
{
"epoch": 0.07,
"learning_rate": 4.8900361780974064e-05,
"loss": 2.4976,
"step": 2000
},
{
"epoch": 0.08,
"learning_rate": 4.8625452226217576e-05,
"loss": 2.4353,
"step": 2500
},
{
"epoch": 0.1,
"learning_rate": 4.835054267146109e-05,
"loss": 2.4423,
"step": 3000
},
{
"epoch": 0.12,
"learning_rate": 4.807563311670461e-05,
"loss": 2.4762,
"step": 3500
},
{
"epoch": 0.13,
"learning_rate": 4.780072356194812e-05,
"loss": 2.3285,
"step": 4000
},
{
"epoch": 0.15,
"learning_rate": 4.752581400719164e-05,
"loss": 2.3804,
"step": 4500
},
{
"epoch": 0.16,
"learning_rate": 4.725090445243515e-05,
"loss": 2.3788,
"step": 5000
},
{
"epoch": 0.18,
"learning_rate": 4.697599489767866e-05,
"loss": 2.4602,
"step": 5500
},
{
"epoch": 0.2,
"learning_rate": 4.670108534292218e-05,
"loss": 2.4195,
"step": 6000
},
{
"epoch": 0.21,
"learning_rate": 4.642617578816569e-05,
"loss": 2.3998,
"step": 6500
},
{
"epoch": 0.23,
"learning_rate": 4.615126623340921e-05,
"loss": 2.3741,
"step": 7000
},
{
"epoch": 0.25,
"learning_rate": 4.5876356678652724e-05,
"loss": 2.3213,
"step": 7500
},
{
"epoch": 0.26,
"learning_rate": 4.560144712389624e-05,
"loss": 2.3207,
"step": 8000
},
{
"epoch": 0.28,
"learning_rate": 4.5326537569139755e-05,
"loss": 2.3541,
"step": 8500
},
{
"epoch": 0.3,
"learning_rate": 4.505162801438327e-05,
"loss": 2.2957,
"step": 9000
},
{
"epoch": 0.31,
"learning_rate": 4.4776718459626786e-05,
"loss": 2.3184,
"step": 9500
},
{
"epoch": 0.33,
"learning_rate": 4.45018089048703e-05,
"loss": 2.2181,
"step": 10000
},
{
"epoch": 0.35,
"learning_rate": 4.422689935011382e-05,
"loss": 2.1771,
"step": 10500
},
{
"epoch": 0.36,
"learning_rate": 4.395198979535733e-05,
"loss": 2.2259,
"step": 11000
},
{
"epoch": 0.38,
"learning_rate": 4.367708024060084e-05,
"loss": 2.2873,
"step": 11500
},
{
"epoch": 0.4,
"learning_rate": 4.340217068584435e-05,
"loss": 2.2289,
"step": 12000
},
{
"epoch": 0.41,
"learning_rate": 4.312726113108788e-05,
"loss": 2.2046,
"step": 12500
},
{
"epoch": 0.43,
"learning_rate": 4.285235157633139e-05,
"loss": 2.2066,
"step": 13000
},
{
"epoch": 0.45,
"learning_rate": 4.25774420215749e-05,
"loss": 2.1925,
"step": 13500
},
{
"epoch": 0.46,
"learning_rate": 4.2302532466818415e-05,
"loss": 2.1787,
"step": 14000
},
{
"epoch": 0.48,
"learning_rate": 4.2027622912061934e-05,
"loss": 2.259,
"step": 14500
},
{
"epoch": 0.49,
"learning_rate": 4.175271335730545e-05,
"loss": 2.1978,
"step": 15000
},
{
"epoch": 0.51,
"learning_rate": 4.1477803802548965e-05,
"loss": 2.1426,
"step": 15500
},
{
"epoch": 0.53,
"learning_rate": 4.120289424779248e-05,
"loss": 2.1855,
"step": 16000
},
{
"epoch": 0.54,
"learning_rate": 4.092798469303599e-05,
"loss": 2.1637,
"step": 16500
},
{
"epoch": 0.56,
"learning_rate": 4.065307513827951e-05,
"loss": 2.1938,
"step": 17000
},
{
"epoch": 0.58,
"learning_rate": 4.037816558352303e-05,
"loss": 2.1819,
"step": 17500
},
{
"epoch": 0.59,
"learning_rate": 4.010325602876654e-05,
"loss": 2.1802,
"step": 18000
},
{
"epoch": 0.61,
"learning_rate": 3.982834647401005e-05,
"loss": 2.1611,
"step": 18500
},
{
"epoch": 0.63,
"learning_rate": 3.955343691925357e-05,
"loss": 2.0925,
"step": 19000
},
{
"epoch": 0.64,
"learning_rate": 3.927852736449708e-05,
"loss": 2.1806,
"step": 19500
},
{
"epoch": 0.66,
"learning_rate": 3.9003617809740594e-05,
"loss": 2.1754,
"step": 20000
},
{
"epoch": 0.68,
"learning_rate": 3.872870825498411e-05,
"loss": 2.1636,
"step": 20500
},
{
"epoch": 0.69,
"learning_rate": 3.8453798700227625e-05,
"loss": 2.0656,
"step": 21000
},
{
"epoch": 0.71,
"learning_rate": 3.8178889145471144e-05,
"loss": 2.198,
"step": 21500
},
{
"epoch": 0.73,
"learning_rate": 3.7903979590714656e-05,
"loss": 2.1448,
"step": 22000
},
{
"epoch": 0.74,
"learning_rate": 3.762907003595817e-05,
"loss": 2.0928,
"step": 22500
},
{
"epoch": 0.76,
"learning_rate": 3.735416048120169e-05,
"loss": 2.0996,
"step": 23000
},
{
"epoch": 0.78,
"learning_rate": 3.7079250926445206e-05,
"loss": 2.0768,
"step": 23500
},
{
"epoch": 0.79,
"learning_rate": 3.680434137168872e-05,
"loss": 2.1074,
"step": 24000
},
{
"epoch": 0.81,
"learning_rate": 3.652943181693223e-05,
"loss": 2.0912,
"step": 24500
},
{
"epoch": 0.82,
"learning_rate": 3.625452226217574e-05,
"loss": 2.1182,
"step": 25000
},
{
"epoch": 0.84,
"learning_rate": 3.597961270741926e-05,
"loss": 2.0224,
"step": 25500
},
{
"epoch": 0.86,
"learning_rate": 3.570470315266278e-05,
"loss": 2.0441,
"step": 26000
},
{
"epoch": 0.87,
"learning_rate": 3.542979359790629e-05,
"loss": 2.0149,
"step": 26500
},
{
"epoch": 0.89,
"learning_rate": 3.5154884043149804e-05,
"loss": 2.1272,
"step": 27000
},
{
"epoch": 0.91,
"learning_rate": 3.4879974488393316e-05,
"loss": 2.0513,
"step": 27500
},
{
"epoch": 0.92,
"learning_rate": 3.4605064933636835e-05,
"loss": 2.0251,
"step": 28000
},
{
"epoch": 0.94,
"learning_rate": 3.4330155378880354e-05,
"loss": 2.0932,
"step": 28500
},
{
"epoch": 0.96,
"learning_rate": 3.4055245824123866e-05,
"loss": 2.0591,
"step": 29000
},
{
"epoch": 0.97,
"learning_rate": 3.378033626936738e-05,
"loss": 2.0504,
"step": 29500
},
{
"epoch": 0.99,
"learning_rate": 3.35054267146109e-05,
"loss": 1.9592,
"step": 30000
},
{
"epoch": 1.01,
"learning_rate": 3.323051715985441e-05,
"loss": 1.9967,
"step": 30500
},
{
"epoch": 1.02,
"learning_rate": 3.295560760509793e-05,
"loss": 1.8916,
"step": 31000
},
{
"epoch": 1.04,
"learning_rate": 3.268069805034144e-05,
"loss": 1.8464,
"step": 31500
},
{
"epoch": 1.06,
"learning_rate": 3.240578849558495e-05,
"loss": 1.9207,
"step": 32000
},
{
"epoch": 1.07,
"learning_rate": 3.213087894082847e-05,
"loss": 1.8944,
"step": 32500
},
{
"epoch": 1.09,
"learning_rate": 3.185596938607198e-05,
"loss": 1.8422,
"step": 33000
},
{
"epoch": 1.11,
"learning_rate": 3.1581059831315495e-05,
"loss": 1.8327,
"step": 33500
},
{
"epoch": 1.12,
"learning_rate": 3.1306150276559014e-05,
"loss": 1.9038,
"step": 34000
},
{
"epoch": 1.14,
"learning_rate": 3.103124072180253e-05,
"loss": 1.7727,
"step": 34500
},
{
"epoch": 1.15,
"learning_rate": 3.0756331167046045e-05,
"loss": 1.8793,
"step": 35000
},
{
"epoch": 1.17,
"learning_rate": 3.0481421612289557e-05,
"loss": 1.777,
"step": 35500
},
{
"epoch": 1.19,
"learning_rate": 3.0206512057533072e-05,
"loss": 1.8554,
"step": 36000
},
{
"epoch": 1.2,
"learning_rate": 2.993160250277659e-05,
"loss": 1.8257,
"step": 36500
},
{
"epoch": 1.22,
"learning_rate": 2.9656692948020103e-05,
"loss": 1.8659,
"step": 37000
},
{
"epoch": 1.24,
"learning_rate": 2.938178339326362e-05,
"loss": 1.8602,
"step": 37500
},
{
"epoch": 1.25,
"learning_rate": 2.910687383850713e-05,
"loss": 1.845,
"step": 38000
},
{
"epoch": 1.27,
"learning_rate": 2.8831964283750646e-05,
"loss": 1.8808,
"step": 38500
},
{
"epoch": 1.29,
"learning_rate": 2.8557054728994165e-05,
"loss": 1.8281,
"step": 39000
},
{
"epoch": 1.3,
"learning_rate": 2.8282145174237677e-05,
"loss": 1.7822,
"step": 39500
},
{
"epoch": 1.32,
"learning_rate": 2.8007235619481192e-05,
"loss": 1.8515,
"step": 40000
},
{
"epoch": 1.34,
"learning_rate": 2.7732326064724708e-05,
"loss": 1.7762,
"step": 40500
},
{
"epoch": 1.35,
"learning_rate": 2.745741650996822e-05,
"loss": 1.859,
"step": 41000
},
{
"epoch": 1.37,
"learning_rate": 2.7182506955211735e-05,
"loss": 1.7589,
"step": 41500
},
{
"epoch": 1.39,
"learning_rate": 2.6907597400455254e-05,
"loss": 1.8329,
"step": 42000
},
{
"epoch": 1.4,
"learning_rate": 2.6632687845698766e-05,
"loss": 1.8349,
"step": 42500
},
{
"epoch": 1.42,
"learning_rate": 2.6357778290942282e-05,
"loss": 1.7516,
"step": 43000
},
{
"epoch": 1.44,
"learning_rate": 2.6082868736185794e-05,
"loss": 1.7724,
"step": 43500
},
{
"epoch": 1.45,
"learning_rate": 2.580795918142931e-05,
"loss": 1.8026,
"step": 44000
},
{
"epoch": 1.47,
"learning_rate": 2.553304962667283e-05,
"loss": 1.7926,
"step": 44500
},
{
"epoch": 1.48,
"learning_rate": 2.525814007191634e-05,
"loss": 1.7585,
"step": 45000
},
{
"epoch": 1.5,
"learning_rate": 2.4983230517159856e-05,
"loss": 1.8155,
"step": 45500
},
{
"epoch": 1.52,
"learning_rate": 2.4708320962403368e-05,
"loss": 1.7969,
"step": 46000
},
{
"epoch": 1.53,
"learning_rate": 2.4433411407646887e-05,
"loss": 1.8457,
"step": 46500
},
{
"epoch": 1.55,
"learning_rate": 2.41585018528904e-05,
"loss": 1.7194,
"step": 47000
},
{
"epoch": 1.57,
"learning_rate": 2.3883592298133914e-05,
"loss": 1.7549,
"step": 47500
},
{
"epoch": 1.58,
"learning_rate": 2.360868274337743e-05,
"loss": 1.8323,
"step": 48000
},
{
"epoch": 1.6,
"learning_rate": 2.3333773188620945e-05,
"loss": 1.8174,
"step": 48500
},
{
"epoch": 1.62,
"learning_rate": 2.305886363386446e-05,
"loss": 1.7714,
"step": 49000
},
{
"epoch": 1.63,
"learning_rate": 2.2783954079107976e-05,
"loss": 1.8349,
"step": 49500
},
{
"epoch": 1.65,
"learning_rate": 2.250904452435149e-05,
"loss": 1.7694,
"step": 50000
},
{
"epoch": 1.67,
"learning_rate": 2.2234134969595004e-05,
"loss": 1.7458,
"step": 50500
},
{
"epoch": 1.68,
"learning_rate": 2.195922541483852e-05,
"loss": 1.7347,
"step": 51000
},
{
"epoch": 1.7,
"learning_rate": 2.168431586008203e-05,
"loss": 1.6837,
"step": 51500
},
{
"epoch": 1.72,
"learning_rate": 2.140940630532555e-05,
"loss": 1.7488,
"step": 52000
},
{
"epoch": 1.73,
"learning_rate": 2.1134496750569062e-05,
"loss": 1.7719,
"step": 52500
},
{
"epoch": 1.75,
"learning_rate": 2.085958719581258e-05,
"loss": 1.7886,
"step": 53000
},
{
"epoch": 1.76,
"learning_rate": 2.0584677641056093e-05,
"loss": 1.7716,
"step": 53500
},
{
"epoch": 1.78,
"learning_rate": 2.030976808629961e-05,
"loss": 1.7544,
"step": 54000
},
{
"epoch": 1.8,
"learning_rate": 2.0034858531543124e-05,
"loss": 1.6598,
"step": 54500
},
{
"epoch": 1.81,
"learning_rate": 1.975994897678664e-05,
"loss": 1.7198,
"step": 55000
},
{
"epoch": 1.83,
"learning_rate": 1.948503942203015e-05,
"loss": 1.7645,
"step": 55500
},
{
"epoch": 1.85,
"learning_rate": 1.9210129867273667e-05,
"loss": 1.6864,
"step": 56000
},
{
"epoch": 1.86,
"learning_rate": 1.8935220312517183e-05,
"loss": 1.6056,
"step": 56500
},
{
"epoch": 1.88,
"learning_rate": 1.8660310757760698e-05,
"loss": 1.7577,
"step": 57000
},
{
"epoch": 1.9,
"learning_rate": 1.8385401203004214e-05,
"loss": 1.7205,
"step": 57500
},
{
"epoch": 1.91,
"learning_rate": 1.8110491648247726e-05,
"loss": 1.732,
"step": 58000
},
{
"epoch": 1.93,
"learning_rate": 1.7835582093491245e-05,
"loss": 1.7373,
"step": 58500
},
{
"epoch": 1.95,
"learning_rate": 1.7560672538734757e-05,
"loss": 1.7185,
"step": 59000
},
{
"epoch": 1.96,
"learning_rate": 1.7285762983978272e-05,
"loss": 1.7288,
"step": 59500
},
{
"epoch": 1.98,
"learning_rate": 1.7010853429221788e-05,
"loss": 1.7187,
"step": 60000
},
{
"epoch": 2.0,
"learning_rate": 1.6735943874465303e-05,
"loss": 1.6532,
"step": 60500
},
{
"epoch": 2.01,
"learning_rate": 1.646103431970882e-05,
"loss": 1.6617,
"step": 61000
},
{
"epoch": 2.03,
"learning_rate": 1.618612476495233e-05,
"loss": 1.5789,
"step": 61500
},
{
"epoch": 2.05,
"learning_rate": 1.5911215210195846e-05,
"loss": 1.5465,
"step": 62000
},
{
"epoch": 2.06,
"learning_rate": 1.563630565543936e-05,
"loss": 1.5776,
"step": 62500
},
{
"epoch": 2.08,
"learning_rate": 1.5361396100682877e-05,
"loss": 1.5464,
"step": 63000
},
{
"epoch": 2.09,
"learning_rate": 1.508648654592639e-05,
"loss": 1.6179,
"step": 63500
},
{
"epoch": 2.11,
"learning_rate": 1.4811576991169906e-05,
"loss": 1.5108,
"step": 64000
},
{
"epoch": 2.13,
"learning_rate": 1.453666743641342e-05,
"loss": 1.5434,
"step": 64500
},
{
"epoch": 2.14,
"learning_rate": 1.4261757881656937e-05,
"loss": 1.5069,
"step": 65000
},
{
"epoch": 2.16,
"learning_rate": 1.3986848326900451e-05,
"loss": 1.5761,
"step": 65500
},
{
"epoch": 2.18,
"learning_rate": 1.3711938772143965e-05,
"loss": 1.551,
"step": 66000
},
{
"epoch": 2.19,
"learning_rate": 1.3437029217387482e-05,
"loss": 1.5682,
"step": 66500
},
{
"epoch": 2.21,
"learning_rate": 1.3162119662630996e-05,
"loss": 1.5375,
"step": 67000
},
{
"epoch": 2.23,
"learning_rate": 1.288721010787451e-05,
"loss": 1.5341,
"step": 67500
},
{
"epoch": 2.24,
"learning_rate": 1.2612300553118025e-05,
"loss": 1.4952,
"step": 68000
},
{
"epoch": 2.26,
"learning_rate": 1.2337390998361539e-05,
"loss": 1.5409,
"step": 68500
},
{
"epoch": 2.28,
"learning_rate": 1.2062481443605054e-05,
"loss": 1.5029,
"step": 69000
},
{
"epoch": 2.29,
"learning_rate": 1.178757188884857e-05,
"loss": 1.5773,
"step": 69500
},
{
"epoch": 2.31,
"learning_rate": 1.1512662334092083e-05,
"loss": 1.5365,
"step": 70000
},
{
"epoch": 2.33,
"learning_rate": 1.1237752779335599e-05,
"loss": 1.541,
"step": 70500
},
{
"epoch": 2.34,
"learning_rate": 1.0962843224579114e-05,
"loss": 1.5085,
"step": 71000
},
{
"epoch": 2.36,
"learning_rate": 1.068793366982263e-05,
"loss": 1.5077,
"step": 71500
},
{
"epoch": 2.38,
"learning_rate": 1.0413024115066144e-05,
"loss": 1.4844,
"step": 72000
},
{
"epoch": 2.39,
"learning_rate": 1.0138114560309659e-05,
"loss": 1.5759,
"step": 72500
},
{
"epoch": 2.41,
"learning_rate": 9.863205005553173e-06,
"loss": 1.5878,
"step": 73000
},
{
"epoch": 2.42,
"learning_rate": 9.588295450796688e-06,
"loss": 1.5286,
"step": 73500
},
{
"epoch": 2.44,
"learning_rate": 9.313385896040202e-06,
"loss": 1.4713,
"step": 74000
},
{
"epoch": 2.46,
"learning_rate": 9.038476341283717e-06,
"loss": 1.4904,
"step": 74500
},
{
"epoch": 2.47,
"learning_rate": 8.763566786527233e-06,
"loss": 1.5195,
"step": 75000
},
{
"epoch": 2.49,
"learning_rate": 8.488657231770748e-06,
"loss": 1.5317,
"step": 75500
},
{
"epoch": 2.51,
"learning_rate": 8.213747677014262e-06,
"loss": 1.4875,
"step": 76000
},
{
"epoch": 2.52,
"learning_rate": 7.938838122257778e-06,
"loss": 1.5204,
"step": 76500
},
{
"epoch": 2.54,
"learning_rate": 7.663928567501293e-06,
"loss": 1.4699,
"step": 77000
},
{
"epoch": 2.56,
"learning_rate": 7.389019012744808e-06,
"loss": 1.4663,
"step": 77500
},
{
"epoch": 2.57,
"learning_rate": 7.1141094579883215e-06,
"loss": 1.52,
"step": 78000
},
{
"epoch": 2.59,
"learning_rate": 6.839199903231837e-06,
"loss": 1.4954,
"step": 78500
},
{
"epoch": 2.61,
"learning_rate": 6.5642903484753525e-06,
"loss": 1.4794,
"step": 79000
},
{
"epoch": 2.62,
"learning_rate": 6.289380793718867e-06,
"loss": 1.5041,
"step": 79500
},
{
"epoch": 2.64,
"learning_rate": 6.014471238962382e-06,
"loss": 1.4915,
"step": 80000
},
{
"epoch": 2.66,
"learning_rate": 5.739561684205896e-06,
"loss": 1.5566,
"step": 80500
},
{
"epoch": 2.67,
"learning_rate": 5.464652129449412e-06,
"loss": 1.4002,
"step": 81000
},
{
"epoch": 2.69,
"learning_rate": 5.1897425746929264e-06,
"loss": 1.4538,
"step": 81500
},
{
"epoch": 2.71,
"learning_rate": 4.914833019936441e-06,
"loss": 1.4771,
"step": 82000
},
{
"epoch": 2.72,
"learning_rate": 4.639923465179956e-06,
"loss": 1.457,
"step": 82500
},
{
"epoch": 2.74,
"learning_rate": 4.365013910423471e-06,
"loss": 1.4573,
"step": 83000
},
{
"epoch": 2.75,
"learning_rate": 4.090104355666986e-06,
"loss": 1.4708,
"step": 83500
},
{
"epoch": 2.77,
"learning_rate": 3.815194800910501e-06,
"loss": 1.4716,
"step": 84000
},
{
"epoch": 2.79,
"learning_rate": 3.5402852461540154e-06,
"loss": 1.4135,
"step": 84500
},
{
"epoch": 2.8,
"learning_rate": 3.2653756913975305e-06,
"loss": 1.4447,
"step": 85000
},
{
"epoch": 2.82,
"learning_rate": 2.990466136641045e-06,
"loss": 1.53,
"step": 85500
},
{
"epoch": 2.84,
"learning_rate": 2.71555658188456e-06,
"loss": 1.5049,
"step": 86000
},
{
"epoch": 2.85,
"learning_rate": 2.4406470271280748e-06,
"loss": 1.4019,
"step": 86500
},
{
"epoch": 2.87,
"learning_rate": 2.16573747237159e-06,
"loss": 1.4247,
"step": 87000
},
{
"epoch": 2.89,
"learning_rate": 1.8908279176151046e-06,
"loss": 1.4441,
"step": 87500
},
{
"epoch": 2.9,
"learning_rate": 1.6159183628586195e-06,
"loss": 1.4318,
"step": 88000
},
{
"epoch": 2.92,
"learning_rate": 1.3410088081021345e-06,
"loss": 1.4432,
"step": 88500
},
{
"epoch": 2.94,
"learning_rate": 1.0660992533456494e-06,
"loss": 1.4239,
"step": 89000
},
{
"epoch": 2.95,
"learning_rate": 7.911896985891643e-07,
"loss": 1.5041,
"step": 89500
},
{
"epoch": 2.97,
"learning_rate": 5.16280143832679e-07,
"loss": 1.4287,
"step": 90000
},
{
"epoch": 2.99,
"learning_rate": 2.4137058907619396e-07,
"loss": 1.3748,
"step": 90500
},
{
"epoch": 3.0,
"step": 90939,
"total_flos": 1.4497996464171216e+16,
"train_loss": 1.8377529225624027,
"train_runtime": 9670.2298,
"train_samples_per_second": 75.23,
"train_steps_per_second": 9.404
}
],
"max_steps": 90939,
"num_train_epochs": 3,
"total_flos": 1.4497996464171216e+16,
"trial_name": null,
"trial_params": null
}