ryan04152024_ALLDATA / trainer_state.json
rshrott's picture
Training in progress, step 100
5f39d17 verified
raw
history blame
21.3 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0100219229564673,
"eval_steps": 100,
"global_step": 3225,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"grad_norm": 0.6999576091766357,
"learning_rate": 4.96124031007752e-05,
"loss": 0.533,
"step": 25
},
{
"epoch": 0.02,
"grad_norm": 0.682187020778656,
"learning_rate": 4.922480620155039e-05,
"loss": 0.3801,
"step": 50
},
{
"epoch": 0.02,
"grad_norm": 0.6211907267570496,
"learning_rate": 4.883720930232558e-05,
"loss": 0.3996,
"step": 75
},
{
"epoch": 0.03,
"grad_norm": 2.131941080093384,
"learning_rate": 4.8449612403100775e-05,
"loss": 0.3272,
"step": 100
},
{
"epoch": 0.04,
"grad_norm": 0.6068922877311707,
"learning_rate": 4.8062015503875976e-05,
"loss": 0.3431,
"step": 125
},
{
"epoch": 0.05,
"grad_norm": 2.6034927368164062,
"learning_rate": 4.7674418604651164e-05,
"loss": 0.3298,
"step": 150
},
{
"epoch": 0.05,
"grad_norm": 2.252187490463257,
"learning_rate": 4.728682170542636e-05,
"loss": 0.3048,
"step": 175
},
{
"epoch": 0.06,
"grad_norm": 0.42226940393447876,
"learning_rate": 4.6899224806201553e-05,
"loss": 0.2993,
"step": 200
},
{
"epoch": 0.07,
"grad_norm": 0.5101619958877563,
"learning_rate": 4.651162790697675e-05,
"loss": 0.3291,
"step": 225
},
{
"epoch": 0.08,
"grad_norm": 0.44378426671028137,
"learning_rate": 4.6124031007751936e-05,
"loss": 0.3254,
"step": 250
},
{
"epoch": 0.09,
"grad_norm": 0.39292019605636597,
"learning_rate": 4.573643410852713e-05,
"loss": 0.2908,
"step": 275
},
{
"epoch": 0.09,
"grad_norm": 0.3389347791671753,
"learning_rate": 4.5348837209302326e-05,
"loss": 0.3058,
"step": 300
},
{
"epoch": 0.1,
"grad_norm": 0.3901497423648834,
"learning_rate": 4.496124031007753e-05,
"loss": 0.343,
"step": 325
},
{
"epoch": 0.11,
"grad_norm": 2.7542033195495605,
"learning_rate": 4.4573643410852715e-05,
"loss": 0.2969,
"step": 350
},
{
"epoch": 0.12,
"grad_norm": 0.3395516872406006,
"learning_rate": 4.418604651162791e-05,
"loss": 0.2907,
"step": 375
},
{
"epoch": 0.13,
"grad_norm": 0.3581525385379791,
"learning_rate": 4.3813953488372095e-05,
"loss": 0.2783,
"step": 400
},
{
"epoch": 0.13,
"grad_norm": 0.6102267503738403,
"learning_rate": 4.342635658914729e-05,
"loss": 0.266,
"step": 425
},
{
"epoch": 0.14,
"grad_norm": 1.409443974494934,
"learning_rate": 4.3038759689922484e-05,
"loss": 0.2898,
"step": 450
},
{
"epoch": 0.15,
"grad_norm": 2.3266422748565674,
"learning_rate": 4.265116279069767e-05,
"loss": 0.2967,
"step": 475
},
{
"epoch": 0.16,
"grad_norm": 1.4810105562210083,
"learning_rate": 4.2263565891472874e-05,
"loss": 0.2621,
"step": 500
},
{
"epoch": 0.16,
"grad_norm": 3.7368390560150146,
"learning_rate": 4.187596899224807e-05,
"loss": 0.2766,
"step": 525
},
{
"epoch": 0.17,
"grad_norm": 1.7454067468643188,
"learning_rate": 4.1488372093023256e-05,
"loss": 0.2435,
"step": 550
},
{
"epoch": 0.18,
"grad_norm": 0.34005963802337646,
"learning_rate": 4.110077519379845e-05,
"loss": 0.2407,
"step": 575
},
{
"epoch": 0.19,
"grad_norm": 1.942692518234253,
"learning_rate": 4.0713178294573646e-05,
"loss": 0.2556,
"step": 600
},
{
"epoch": 0.2,
"grad_norm": 1.3751680850982666,
"learning_rate": 4.032558139534884e-05,
"loss": 0.2352,
"step": 625
},
{
"epoch": 0.2,
"grad_norm": 1.911611795425415,
"learning_rate": 3.993798449612403e-05,
"loss": 0.2635,
"step": 650
},
{
"epoch": 0.21,
"grad_norm": 2.1268646717071533,
"learning_rate": 3.955038759689922e-05,
"loss": 0.2473,
"step": 675
},
{
"epoch": 0.22,
"grad_norm": 1.866083025932312,
"learning_rate": 3.9162790697674425e-05,
"loss": 0.238,
"step": 700
},
{
"epoch": 0.23,
"grad_norm": 0.8285239338874817,
"learning_rate": 3.877519379844962e-05,
"loss": 0.2666,
"step": 725
},
{
"epoch": 0.23,
"grad_norm": 0.8000108003616333,
"learning_rate": 3.838759689922481e-05,
"loss": 0.2343,
"step": 750
},
{
"epoch": 0.24,
"grad_norm": 0.6987760066986084,
"learning_rate": 3.8e-05,
"loss": 0.207,
"step": 775
},
{
"epoch": 0.25,
"grad_norm": 1.039829134941101,
"learning_rate": 3.76124031007752e-05,
"loss": 0.2469,
"step": 800
},
{
"epoch": 0.26,
"grad_norm": 1.2499618530273438,
"learning_rate": 3.722480620155039e-05,
"loss": 0.2589,
"step": 825
},
{
"epoch": 0.27,
"grad_norm": 2.926412582397461,
"learning_rate": 3.683720930232558e-05,
"loss": 0.1872,
"step": 850
},
{
"epoch": 0.27,
"grad_norm": 5.37480354309082,
"learning_rate": 3.6449612403100774e-05,
"loss": 0.2797,
"step": 875
},
{
"epoch": 0.28,
"grad_norm": 1.7981369495391846,
"learning_rate": 3.6077519379844966e-05,
"loss": 0.2423,
"step": 900
},
{
"epoch": 0.29,
"grad_norm": 1.1063029766082764,
"learning_rate": 3.568992248062016e-05,
"loss": 0.2528,
"step": 925
},
{
"epoch": 0.3,
"grad_norm": 0.5154077410697937,
"learning_rate": 3.530232558139535e-05,
"loss": 0.1966,
"step": 950
},
{
"epoch": 0.31,
"grad_norm": 1.958801507949829,
"learning_rate": 3.4914728682170543e-05,
"loss": 0.2569,
"step": 975
},
{
"epoch": 0.31,
"grad_norm": 0.43896540999412537,
"learning_rate": 3.452713178294574e-05,
"loss": 0.2294,
"step": 1000
},
{
"epoch": 0.32,
"grad_norm": 4.940056800842285,
"learning_rate": 3.413953488372093e-05,
"loss": 0.2138,
"step": 1025
},
{
"epoch": 0.33,
"grad_norm": 2.0871469974517822,
"learning_rate": 3.375193798449612e-05,
"loss": 0.2486,
"step": 1050
},
{
"epoch": 0.34,
"grad_norm": 0.4600803554058075,
"learning_rate": 3.3364341085271316e-05,
"loss": 0.1784,
"step": 1075
},
{
"epoch": 0.34,
"grad_norm": 1.7933701276779175,
"learning_rate": 3.297674418604652e-05,
"loss": 0.2486,
"step": 1100
},
{
"epoch": 0.35,
"grad_norm": 0.8704222440719604,
"learning_rate": 3.258914728682171e-05,
"loss": 0.1977,
"step": 1125
},
{
"epoch": 0.36,
"grad_norm": 2.517425775527954,
"learning_rate": 3.22015503875969e-05,
"loss": 0.2128,
"step": 1150
},
{
"epoch": 0.37,
"grad_norm": 1.5435869693756104,
"learning_rate": 3.1813953488372094e-05,
"loss": 0.1889,
"step": 1175
},
{
"epoch": 0.38,
"grad_norm": 0.8101773262023926,
"learning_rate": 3.142635658914729e-05,
"loss": 0.2274,
"step": 1200
},
{
"epoch": 0.38,
"grad_norm": 0.910406231880188,
"learning_rate": 3.1038759689922484e-05,
"loss": 0.1939,
"step": 1225
},
{
"epoch": 0.39,
"grad_norm": 1.1105294227600098,
"learning_rate": 3.065116279069767e-05,
"loss": 0.2712,
"step": 1250
},
{
"epoch": 0.4,
"grad_norm": 1.6309148073196411,
"learning_rate": 3.0263565891472867e-05,
"loss": 0.2098,
"step": 1275
},
{
"epoch": 0.41,
"grad_norm": 0.6428541541099548,
"learning_rate": 2.9875968992248065e-05,
"loss": 0.1944,
"step": 1300
},
{
"epoch": 0.41,
"grad_norm": 1.2899640798568726,
"learning_rate": 2.948837209302326e-05,
"loss": 0.232,
"step": 1325
},
{
"epoch": 0.42,
"grad_norm": 0.35119500756263733,
"learning_rate": 2.910077519379845e-05,
"loss": 0.2279,
"step": 1350
},
{
"epoch": 0.43,
"grad_norm": 0.7236185669898987,
"learning_rate": 2.8713178294573645e-05,
"loss": 0.1759,
"step": 1375
},
{
"epoch": 0.44,
"grad_norm": 0.7780153155326843,
"learning_rate": 2.8325581395348837e-05,
"loss": 0.2006,
"step": 1400
},
{
"epoch": 0.45,
"grad_norm": 2.4104530811309814,
"learning_rate": 2.793798449612403e-05,
"loss": 0.2128,
"step": 1425
},
{
"epoch": 0.45,
"grad_norm": 0.5463195443153381,
"learning_rate": 2.7550387596899223e-05,
"loss": 0.1777,
"step": 1450
},
{
"epoch": 0.46,
"grad_norm": 0.48027583956718445,
"learning_rate": 2.7162790697674418e-05,
"loss": 0.1887,
"step": 1475
},
{
"epoch": 0.47,
"grad_norm": 1.2007440328598022,
"learning_rate": 2.6775193798449616e-05,
"loss": 0.2016,
"step": 1500
},
{
"epoch": 0.48,
"grad_norm": 1.51267409324646,
"learning_rate": 2.638759689922481e-05,
"loss": 0.174,
"step": 1525
},
{
"epoch": 0.49,
"grad_norm": 2.8868677616119385,
"learning_rate": 2.6000000000000002e-05,
"loss": 0.2046,
"step": 1550
},
{
"epoch": 0.49,
"grad_norm": 2.025702476501465,
"learning_rate": 2.5612403100775197e-05,
"loss": 0.1859,
"step": 1575
},
{
"epoch": 0.5,
"grad_norm": 1.341884732246399,
"learning_rate": 2.5224806201550388e-05,
"loss": 0.2007,
"step": 1600
},
{
"epoch": 0.51,
"grad_norm": 0.914476752281189,
"learning_rate": 2.4837209302325583e-05,
"loss": 0.1863,
"step": 1625
},
{
"epoch": 0.52,
"grad_norm": 0.8603270649909973,
"learning_rate": 2.4465116279069768e-05,
"loss": 0.2002,
"step": 1650
},
{
"epoch": 0.52,
"grad_norm": 3.7743961811065674,
"learning_rate": 2.4077519379844962e-05,
"loss": 0.1991,
"step": 1675
},
{
"epoch": 0.53,
"grad_norm": 2.3465445041656494,
"learning_rate": 2.3689922480620154e-05,
"loss": 0.1901,
"step": 1700
},
{
"epoch": 0.54,
"grad_norm": 1.6707566976547241,
"learning_rate": 2.3302325581395352e-05,
"loss": 0.1805,
"step": 1725
},
{
"epoch": 0.55,
"grad_norm": 0.9617774486541748,
"learning_rate": 2.2914728682170543e-05,
"loss": 0.2044,
"step": 1750
},
{
"epoch": 0.56,
"grad_norm": 1.740809440612793,
"learning_rate": 2.2527131782945738e-05,
"loss": 0.2502,
"step": 1775
},
{
"epoch": 0.56,
"grad_norm": 1.6407257318496704,
"learning_rate": 2.213953488372093e-05,
"loss": 0.2071,
"step": 1800
},
{
"epoch": 0.57,
"grad_norm": 0.5815154314041138,
"learning_rate": 2.1751937984496127e-05,
"loss": 0.1962,
"step": 1825
},
{
"epoch": 0.58,
"grad_norm": 2.898057699203491,
"learning_rate": 2.136434108527132e-05,
"loss": 0.168,
"step": 1850
},
{
"epoch": 0.59,
"grad_norm": 0.7204940319061279,
"learning_rate": 2.0976744186046513e-05,
"loss": 0.1808,
"step": 1875
},
{
"epoch": 0.6,
"grad_norm": 1.5870317220687866,
"learning_rate": 2.0589147286821705e-05,
"loss": 0.1811,
"step": 1900
},
{
"epoch": 0.6,
"grad_norm": 0.8534879088401794,
"learning_rate": 2.0201550387596903e-05,
"loss": 0.2235,
"step": 1925
},
{
"epoch": 0.61,
"grad_norm": 1.5544407367706299,
"learning_rate": 1.9813953488372094e-05,
"loss": 0.192,
"step": 1950
},
{
"epoch": 0.62,
"grad_norm": 2.284710645675659,
"learning_rate": 1.942635658914729e-05,
"loss": 0.2307,
"step": 1975
},
{
"epoch": 0.63,
"grad_norm": 2.3412628173828125,
"learning_rate": 1.903875968992248e-05,
"loss": 0.1858,
"step": 2000
},
{
"epoch": 0.63,
"grad_norm": 0.5765611529350281,
"learning_rate": 1.8651162790697675e-05,
"loss": 0.1833,
"step": 2025
},
{
"epoch": 0.64,
"grad_norm": 1.0757735967636108,
"learning_rate": 1.826356589147287e-05,
"loss": 0.209,
"step": 2050
},
{
"epoch": 0.65,
"grad_norm": 1.0615739822387695,
"learning_rate": 1.7875968992248064e-05,
"loss": 0.1591,
"step": 2075
},
{
"epoch": 0.66,
"grad_norm": 0.5009509921073914,
"learning_rate": 1.7488372093023256e-05,
"loss": 0.1512,
"step": 2100
},
{
"epoch": 0.67,
"grad_norm": 2.7069976329803467,
"learning_rate": 1.710077519379845e-05,
"loss": 0.2202,
"step": 2125
},
{
"epoch": 0.67,
"grad_norm": 1.9919098615646362,
"learning_rate": 1.6713178294573645e-05,
"loss": 0.2132,
"step": 2150
},
{
"epoch": 0.68,
"grad_norm": 2.6684300899505615,
"learning_rate": 1.6325581395348837e-05,
"loss": 0.1887,
"step": 2175
},
{
"epoch": 0.69,
"grad_norm": 2.843564987182617,
"learning_rate": 1.593798449612403e-05,
"loss": 0.2035,
"step": 2200
},
{
"epoch": 0.7,
"grad_norm": 2.938154697418213,
"learning_rate": 1.5550387596899226e-05,
"loss": 0.1946,
"step": 2225
},
{
"epoch": 0.7,
"grad_norm": 2.365872859954834,
"learning_rate": 1.516279069767442e-05,
"loss": 0.159,
"step": 2250
},
{
"epoch": 0.71,
"grad_norm": 2.8501169681549072,
"learning_rate": 1.4775193798449614e-05,
"loss": 0.1569,
"step": 2275
},
{
"epoch": 0.72,
"grad_norm": 0.6475769877433777,
"learning_rate": 1.4387596899224807e-05,
"loss": 0.1626,
"step": 2300
},
{
"epoch": 0.73,
"grad_norm": 3.125209093093872,
"learning_rate": 1.4000000000000001e-05,
"loss": 0.2249,
"step": 2325
},
{
"epoch": 0.74,
"grad_norm": 0.6424142718315125,
"learning_rate": 1.3612403100775195e-05,
"loss": 0.1583,
"step": 2350
},
{
"epoch": 0.74,
"grad_norm": 1.333817481994629,
"learning_rate": 1.3224806201550388e-05,
"loss": 0.1505,
"step": 2375
},
{
"epoch": 0.75,
"grad_norm": 0.30924907326698303,
"learning_rate": 1.283720930232558e-05,
"loss": 0.211,
"step": 2400
},
{
"epoch": 0.76,
"grad_norm": 0.9411177039146423,
"learning_rate": 1.2449612403100775e-05,
"loss": 0.1582,
"step": 2425
},
{
"epoch": 0.77,
"grad_norm": 0.6235800981521606,
"learning_rate": 1.206201550387597e-05,
"loss": 0.1762,
"step": 2450
},
{
"epoch": 0.78,
"grad_norm": 1.076593279838562,
"learning_rate": 1.1674418604651163e-05,
"loss": 0.2124,
"step": 2475
},
{
"epoch": 0.78,
"grad_norm": 1.0768756866455078,
"learning_rate": 1.1286821705426358e-05,
"loss": 0.2067,
"step": 2500
},
{
"epoch": 0.79,
"grad_norm": 1.3747373819351196,
"learning_rate": 1.089922480620155e-05,
"loss": 0.1959,
"step": 2525
},
{
"epoch": 0.8,
"grad_norm": 3.307833194732666,
"learning_rate": 1.0511627906976746e-05,
"loss": 0.2249,
"step": 2550
},
{
"epoch": 0.81,
"grad_norm": 1.8046529293060303,
"learning_rate": 1.0124031007751939e-05,
"loss": 0.1566,
"step": 2575
},
{
"epoch": 0.81,
"grad_norm": 1.890415072441101,
"learning_rate": 9.736434108527132e-06,
"loss": 0.1621,
"step": 2600
},
{
"epoch": 0.82,
"grad_norm": 1.755682110786438,
"learning_rate": 9.348837209302326e-06,
"loss": 0.202,
"step": 2625
},
{
"epoch": 0.83,
"grad_norm": 0.6612691879272461,
"learning_rate": 8.96124031007752e-06,
"loss": 0.1863,
"step": 2650
},
{
"epoch": 0.84,
"grad_norm": 2.275036334991455,
"learning_rate": 8.573643410852714e-06,
"loss": 0.173,
"step": 2675
},
{
"epoch": 0.85,
"grad_norm": 0.7513935565948486,
"learning_rate": 8.186046511627907e-06,
"loss": 0.1884,
"step": 2700
},
{
"epoch": 0.85,
"grad_norm": 1.9435423612594604,
"learning_rate": 7.798449612403102e-06,
"loss": 0.1795,
"step": 2725
},
{
"epoch": 0.86,
"grad_norm": 0.5520192384719849,
"learning_rate": 7.410852713178294e-06,
"loss": 0.1908,
"step": 2750
},
{
"epoch": 0.87,
"grad_norm": 2.2986812591552734,
"learning_rate": 7.023255813953489e-06,
"loss": 0.1767,
"step": 2775
},
{
"epoch": 0.88,
"grad_norm": 2.4153175354003906,
"learning_rate": 6.635658914728682e-06,
"loss": 0.2202,
"step": 2800
},
{
"epoch": 0.88,
"grad_norm": 2.7443273067474365,
"learning_rate": 6.2480620155038765e-06,
"loss": 0.1961,
"step": 2825
},
{
"epoch": 0.89,
"grad_norm": 5.525525093078613,
"learning_rate": 5.86046511627907e-06,
"loss": 0.1827,
"step": 2850
},
{
"epoch": 0.9,
"grad_norm": 2.7762510776519775,
"learning_rate": 5.472868217054264e-06,
"loss": 0.1816,
"step": 2875
},
{
"epoch": 0.91,
"grad_norm": 2.3923604488372803,
"learning_rate": 5.085271317829457e-06,
"loss": 0.1482,
"step": 2900
},
{
"epoch": 0.92,
"grad_norm": 0.8370789289474487,
"learning_rate": 4.697674418604651e-06,
"loss": 0.148,
"step": 2925
},
{
"epoch": 0.92,
"grad_norm": 2.383211374282837,
"learning_rate": 4.310077519379845e-06,
"loss": 0.1772,
"step": 2950
},
{
"epoch": 0.93,
"grad_norm": 0.5339698195457458,
"learning_rate": 3.922480620155039e-06,
"loss": 0.1409,
"step": 2975
},
{
"epoch": 0.94,
"grad_norm": 0.5133029222488403,
"learning_rate": 3.534883720930233e-06,
"loss": 0.1532,
"step": 3000
},
{
"epoch": 0.95,
"grad_norm": 3.2108466625213623,
"learning_rate": 3.1472868217054267e-06,
"loss": 0.1999,
"step": 3025
},
{
"epoch": 0.96,
"grad_norm": 3.6793041229248047,
"learning_rate": 2.7596899224806206e-06,
"loss": 0.1732,
"step": 3050
},
{
"epoch": 0.96,
"grad_norm": 2.0069997310638428,
"learning_rate": 2.372093023255814e-06,
"loss": 0.1839,
"step": 3075
},
{
"epoch": 0.97,
"grad_norm": 0.663833737373352,
"learning_rate": 1.984496124031008e-06,
"loss": 0.1928,
"step": 3100
},
{
"epoch": 0.98,
"grad_norm": 3.452453851699829,
"learning_rate": 1.5968992248062018e-06,
"loss": 0.1304,
"step": 3125
},
{
"epoch": 0.99,
"grad_norm": 0.8049046397209167,
"learning_rate": 1.2093023255813954e-06,
"loss": 0.1972,
"step": 3150
},
{
"epoch": 0.99,
"grad_norm": 0.5933910608291626,
"learning_rate": 8.217054263565891e-07,
"loss": 0.159,
"step": 3175
},
{
"epoch": 1.0,
"grad_norm": 0.493872731924057,
"learning_rate": 4.3410852713178294e-07,
"loss": 0.1212,
"step": 3200
},
{
"epoch": 1.01,
"grad_norm": 1.8325016498565674,
"learning_rate": 4.651162790697675e-08,
"loss": 0.1036,
"step": 3225
},
{
"epoch": 1.01,
"step": 3225,
"total_flos": 3.9986525216152166e+18,
"train_loss": 0.21891093194946762,
"train_runtime": 6427.7582,
"train_samples_per_second": 8.027,
"train_steps_per_second": 0.502
}
],
"logging_steps": 25,
"max_steps": 3225,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 100,
"total_flos": 3.9986525216152166e+18,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}