hwm21's picture
Training in progress, step 13500, checkpoint
7778af0 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 207.69230769230768,
"eval_steps": 500,
"global_step": 13500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 1.5384615384615383,
"grad_norm": 28.85655403137207,
"learning_rate": 9.9812e-06,
"loss": 1.0794,
"step": 100
},
{
"epoch": 3.076923076923077,
"grad_norm": 40.480037689208984,
"learning_rate": 9.961400000000001e-06,
"loss": 1.1026,
"step": 200
},
{
"epoch": 4.615384615384615,
"grad_norm": 22.348379135131836,
"learning_rate": 9.941600000000002e-06,
"loss": 1.1211,
"step": 300
},
{
"epoch": 6.153846153846154,
"grad_norm": 49.462806701660156,
"learning_rate": 9.921600000000002e-06,
"loss": 1.1217,
"step": 400
},
{
"epoch": 7.6923076923076925,
"grad_norm": 37.26523208618164,
"learning_rate": 9.901600000000002e-06,
"loss": 1.0993,
"step": 500
},
{
"epoch": 9.23076923076923,
"grad_norm": 265.8311767578125,
"learning_rate": 9.8816e-06,
"loss": 1.1576,
"step": 600
},
{
"epoch": 10.76923076923077,
"grad_norm": 48.439048767089844,
"learning_rate": 9.8616e-06,
"loss": 1.1029,
"step": 700
},
{
"epoch": 12.307692307692308,
"grad_norm": 200.80213928222656,
"learning_rate": 9.8416e-06,
"loss": 1.1375,
"step": 800
},
{
"epoch": 13.846153846153847,
"grad_norm": 54.096126556396484,
"learning_rate": 9.821600000000001e-06,
"loss": 1.1134,
"step": 900
},
{
"epoch": 15.384615384615385,
"grad_norm": 72.55537414550781,
"learning_rate": 9.801600000000001e-06,
"loss": 1.1364,
"step": 1000
},
{
"epoch": 16.923076923076923,
"grad_norm": 45.86960983276367,
"learning_rate": 9.781600000000001e-06,
"loss": 1.1259,
"step": 1100
},
{
"epoch": 18.46153846153846,
"grad_norm": 44.21461868286133,
"learning_rate": 9.761600000000002e-06,
"loss": 1.1144,
"step": 1200
},
{
"epoch": 20.0,
"grad_norm": 49.014503479003906,
"learning_rate": 9.741600000000002e-06,
"loss": 1.1151,
"step": 1300
},
{
"epoch": 21.53846153846154,
"grad_norm": 62.7434196472168,
"learning_rate": 9.7216e-06,
"loss": 1.1129,
"step": 1400
},
{
"epoch": 23.076923076923077,
"grad_norm": 48.607032775878906,
"learning_rate": 9.7016e-06,
"loss": 1.1189,
"step": 1500
},
{
"epoch": 24.615384615384617,
"grad_norm": 43.288368225097656,
"learning_rate": 9.6816e-06,
"loss": 1.102,
"step": 1600
},
{
"epoch": 26.153846153846153,
"grad_norm": 33.83266067504883,
"learning_rate": 9.6616e-06,
"loss": 1.1091,
"step": 1700
},
{
"epoch": 27.692307692307693,
"grad_norm": 54.347755432128906,
"learning_rate": 9.641600000000001e-06,
"loss": 1.1478,
"step": 1800
},
{
"epoch": 29.23076923076923,
"grad_norm": 68.06307220458984,
"learning_rate": 9.621600000000001e-06,
"loss": 1.1384,
"step": 1900
},
{
"epoch": 30.76923076923077,
"grad_norm": 24.473976135253906,
"learning_rate": 9.601600000000001e-06,
"loss": 1.1218,
"step": 2000
},
{
"epoch": 32.30769230769231,
"grad_norm": 33.08134460449219,
"learning_rate": 9.5818e-06,
"loss": 1.0807,
"step": 2100
},
{
"epoch": 33.84615384615385,
"grad_norm": 32.798397064208984,
"learning_rate": 9.5618e-06,
"loss": 1.1548,
"step": 2200
},
{
"epoch": 35.38461538461539,
"grad_norm": 49.251651763916016,
"learning_rate": 9.5418e-06,
"loss": 1.1056,
"step": 2300
},
{
"epoch": 36.92307692307692,
"grad_norm": 47.84526062011719,
"learning_rate": 9.521800000000001e-06,
"loss": 1.169,
"step": 2400
},
{
"epoch": 38.46153846153846,
"grad_norm": 66.61646270751953,
"learning_rate": 9.501800000000001e-06,
"loss": 1.1505,
"step": 2500
},
{
"epoch": 40.0,
"grad_norm": 84.58892059326172,
"learning_rate": 9.481800000000001e-06,
"loss": 1.1117,
"step": 2600
},
{
"epoch": 41.53846153846154,
"grad_norm": 86.26738739013672,
"learning_rate": 9.461800000000001e-06,
"loss": 1.0914,
"step": 2700
},
{
"epoch": 43.07692307692308,
"grad_norm": 106.57470703125,
"learning_rate": 9.441800000000002e-06,
"loss": 1.1232,
"step": 2800
},
{
"epoch": 44.61538461538461,
"grad_norm": 133.2903289794922,
"learning_rate": 9.4218e-06,
"loss": 1.1544,
"step": 2900
},
{
"epoch": 46.15384615384615,
"grad_norm": 75.39508819580078,
"learning_rate": 9.4018e-06,
"loss": 1.1699,
"step": 3000
},
{
"epoch": 47.69230769230769,
"grad_norm": 45.91976547241211,
"learning_rate": 9.3818e-06,
"loss": 1.1652,
"step": 3100
},
{
"epoch": 49.23076923076923,
"grad_norm": 37.38508605957031,
"learning_rate": 9.3618e-06,
"loss": 1.1386,
"step": 3200
},
{
"epoch": 50.76923076923077,
"grad_norm": 106.42298126220703,
"learning_rate": 9.341800000000001e-06,
"loss": 1.0903,
"step": 3300
},
{
"epoch": 52.30769230769231,
"grad_norm": 59.256103515625,
"learning_rate": 9.321800000000001e-06,
"loss": 1.1266,
"step": 3400
},
{
"epoch": 53.84615384615385,
"grad_norm": 61.493003845214844,
"learning_rate": 9.301800000000001e-06,
"loss": 1.1047,
"step": 3500
},
{
"epoch": 55.38461538461539,
"grad_norm": 149.27308654785156,
"learning_rate": 9.281800000000001e-06,
"loss": 1.1132,
"step": 3600
},
{
"epoch": 56.92307692307692,
"grad_norm": 81.58853912353516,
"learning_rate": 9.2618e-06,
"loss": 1.083,
"step": 3700
},
{
"epoch": 58.46153846153846,
"grad_norm": 39.950843811035156,
"learning_rate": 9.2418e-06,
"loss": 1.09,
"step": 3800
},
{
"epoch": 60.0,
"grad_norm": 66.99285888671875,
"learning_rate": 9.2218e-06,
"loss": 1.1161,
"step": 3900
},
{
"epoch": 61.53846153846154,
"grad_norm": 28.769887924194336,
"learning_rate": 9.2018e-06,
"loss": 1.1061,
"step": 4000
},
{
"epoch": 63.07692307692308,
"grad_norm": 37.438724517822266,
"learning_rate": 9.1818e-06,
"loss": 1.0812,
"step": 4100
},
{
"epoch": 64.61538461538461,
"grad_norm": 44.86701965332031,
"learning_rate": 9.161800000000001e-06,
"loss": 1.0923,
"step": 4200
},
{
"epoch": 66.15384615384616,
"grad_norm": 64.42572021484375,
"learning_rate": 9.142000000000002e-06,
"loss": 1.0781,
"step": 4300
},
{
"epoch": 67.6923076923077,
"grad_norm": 51.16594696044922,
"learning_rate": 9.122e-06,
"loss": 1.0446,
"step": 4400
},
{
"epoch": 69.23076923076923,
"grad_norm": 39.816341400146484,
"learning_rate": 9.102e-06,
"loss": 1.1083,
"step": 4500
},
{
"epoch": 70.76923076923077,
"grad_norm": 57.10240173339844,
"learning_rate": 9.082e-06,
"loss": 1.0748,
"step": 4600
},
{
"epoch": 72.3076923076923,
"grad_norm": 98.26517486572266,
"learning_rate": 9.062e-06,
"loss": 1.0624,
"step": 4700
},
{
"epoch": 73.84615384615384,
"grad_norm": 23.59007453918457,
"learning_rate": 9.042e-06,
"loss": 1.0644,
"step": 4800
},
{
"epoch": 75.38461538461539,
"grad_norm": 34.98348617553711,
"learning_rate": 9.022000000000001e-06,
"loss": 1.0859,
"step": 4900
},
{
"epoch": 76.92307692307692,
"grad_norm": 65.94945526123047,
"learning_rate": 9.002000000000001e-06,
"loss": 1.0603,
"step": 5000
},
{
"epoch": 78.46153846153847,
"grad_norm": 49.95417404174805,
"learning_rate": 8.982000000000001e-06,
"loss": 1.0915,
"step": 5100
},
{
"epoch": 80.0,
"grad_norm": 57.85556411743164,
"learning_rate": 8.962e-06,
"loss": 1.0787,
"step": 5200
},
{
"epoch": 81.53846153846153,
"grad_norm": 26.018638610839844,
"learning_rate": 8.942e-06,
"loss": 1.0632,
"step": 5300
},
{
"epoch": 83.07692307692308,
"grad_norm": 58.61573028564453,
"learning_rate": 8.922e-06,
"loss": 1.075,
"step": 5400
},
{
"epoch": 84.61538461538461,
"grad_norm": 122.04191589355469,
"learning_rate": 8.902e-06,
"loss": 1.0724,
"step": 5500
},
{
"epoch": 86.15384615384616,
"grad_norm": 78.37217712402344,
"learning_rate": 8.882e-06,
"loss": 1.0925,
"step": 5600
},
{
"epoch": 87.6923076923077,
"grad_norm": 54.941184997558594,
"learning_rate": 8.862000000000001e-06,
"loss": 1.1195,
"step": 5700
},
{
"epoch": 89.23076923076923,
"grad_norm": 61.23499298095703,
"learning_rate": 8.842000000000001e-06,
"loss": 1.0685,
"step": 5800
},
{
"epoch": 90.76923076923077,
"grad_norm": 89.12802124023438,
"learning_rate": 8.822000000000001e-06,
"loss": 1.0899,
"step": 5900
},
{
"epoch": 92.3076923076923,
"grad_norm": 209.87484741210938,
"learning_rate": 8.802e-06,
"loss": 1.0771,
"step": 6000
},
{
"epoch": 93.84615384615384,
"grad_norm": 55.978092193603516,
"learning_rate": 8.782e-06,
"loss": 1.0657,
"step": 6100
},
{
"epoch": 95.38461538461539,
"grad_norm": 76.39286041259766,
"learning_rate": 8.762e-06,
"loss": 1.0653,
"step": 6200
},
{
"epoch": 96.92307692307692,
"grad_norm": 26.763763427734375,
"learning_rate": 8.742e-06,
"loss": 1.0844,
"step": 6300
},
{
"epoch": 98.46153846153847,
"grad_norm": 37.72929382324219,
"learning_rate": 8.722e-06,
"loss": 1.0753,
"step": 6400
},
{
"epoch": 100.0,
"grad_norm": 33.482521057128906,
"learning_rate": 8.702e-06,
"loss": 1.065,
"step": 6500
},
{
"epoch": 101.53846153846153,
"grad_norm": 38.44742965698242,
"learning_rate": 8.682000000000001e-06,
"loss": 1.0583,
"step": 6600
},
{
"epoch": 103.07692307692308,
"grad_norm": 44.009483337402344,
"learning_rate": 8.662000000000001e-06,
"loss": 1.0613,
"step": 6700
},
{
"epoch": 104.61538461538461,
"grad_norm": 53.22964096069336,
"learning_rate": 8.642e-06,
"loss": 1.079,
"step": 6800
},
{
"epoch": 106.15384615384616,
"grad_norm": 32.453739166259766,
"learning_rate": 8.622e-06,
"loss": 1.0564,
"step": 6900
},
{
"epoch": 107.6923076923077,
"grad_norm": 56.385257720947266,
"learning_rate": 8.602e-06,
"loss": 1.0638,
"step": 7000
},
{
"epoch": 109.23076923076923,
"grad_norm": 75.76576232910156,
"learning_rate": 8.582e-06,
"loss": 1.0451,
"step": 7100
},
{
"epoch": 110.76923076923077,
"grad_norm": 43.19460678100586,
"learning_rate": 8.562e-06,
"loss": 1.0651,
"step": 7200
},
{
"epoch": 112.3076923076923,
"grad_norm": 130.46304321289062,
"learning_rate": 8.542e-06,
"loss": 1.0489,
"step": 7300
},
{
"epoch": 113.84615384615384,
"grad_norm": 81.6989974975586,
"learning_rate": 8.522e-06,
"loss": 1.0275,
"step": 7400
},
{
"epoch": 115.38461538461539,
"grad_norm": 125.24699401855469,
"learning_rate": 8.502000000000001e-06,
"loss": 1.0498,
"step": 7500
},
{
"epoch": 116.92307692307692,
"grad_norm": 54.86334228515625,
"learning_rate": 8.482e-06,
"loss": 1.0327,
"step": 7600
},
{
"epoch": 118.46153846153847,
"grad_norm": 42.080753326416016,
"learning_rate": 8.462e-06,
"loss": 1.0425,
"step": 7700
},
{
"epoch": 120.0,
"grad_norm": 73.9937973022461,
"learning_rate": 8.442e-06,
"loss": 1.0433,
"step": 7800
},
{
"epoch": 121.53846153846153,
"grad_norm": 107.1471939086914,
"learning_rate": 8.422e-06,
"loss": 1.0573,
"step": 7900
},
{
"epoch": 123.07692307692308,
"grad_norm": 239.5729217529297,
"learning_rate": 8.402e-06,
"loss": 1.0388,
"step": 8000
},
{
"epoch": 124.61538461538461,
"grad_norm": 39.29356384277344,
"learning_rate": 8.382e-06,
"loss": 1.0647,
"step": 8100
},
{
"epoch": 126.15384615384616,
"grad_norm": 146.09300231933594,
"learning_rate": 8.362e-06,
"loss": 1.0166,
"step": 8200
},
{
"epoch": 127.6923076923077,
"grad_norm": 90.30326080322266,
"learning_rate": 8.342e-06,
"loss": 1.0677,
"step": 8300
},
{
"epoch": 129.23076923076923,
"grad_norm": 20.28913116455078,
"learning_rate": 8.322000000000001e-06,
"loss": 1.0504,
"step": 8400
},
{
"epoch": 130.76923076923077,
"grad_norm": 44.32182312011719,
"learning_rate": 8.302000000000001e-06,
"loss": 1.0628,
"step": 8500
},
{
"epoch": 132.30769230769232,
"grad_norm": 54.270172119140625,
"learning_rate": 8.282000000000001e-06,
"loss": 1.0361,
"step": 8600
},
{
"epoch": 133.84615384615384,
"grad_norm": 66.50403594970703,
"learning_rate": 8.262000000000002e-06,
"loss": 1.029,
"step": 8700
},
{
"epoch": 135.3846153846154,
"grad_norm": 36.06448745727539,
"learning_rate": 8.242000000000002e-06,
"loss": 1.0412,
"step": 8800
},
{
"epoch": 136.92307692307693,
"grad_norm": 61.25200653076172,
"learning_rate": 8.222000000000002e-06,
"loss": 1.0436,
"step": 8900
},
{
"epoch": 138.46153846153845,
"grad_norm": 75.82115936279297,
"learning_rate": 8.202e-06,
"loss": 1.0203,
"step": 9000
},
{
"epoch": 140.0,
"grad_norm": 82.14342498779297,
"learning_rate": 8.182e-06,
"loss": 1.0414,
"step": 9100
},
{
"epoch": 141.53846153846155,
"grad_norm": 40.55479431152344,
"learning_rate": 8.162e-06,
"loss": 1.0394,
"step": 9200
},
{
"epoch": 143.07692307692307,
"grad_norm": 62.4966926574707,
"learning_rate": 8.142000000000001e-06,
"loss": 1.0621,
"step": 9300
},
{
"epoch": 144.6153846153846,
"grad_norm": 40.539161682128906,
"learning_rate": 8.122000000000001e-06,
"loss": 1.0346,
"step": 9400
},
{
"epoch": 146.15384615384616,
"grad_norm": 65.81285858154297,
"learning_rate": 8.102000000000001e-06,
"loss": 1.0561,
"step": 9500
},
{
"epoch": 147.69230769230768,
"grad_norm": 64.11259460449219,
"learning_rate": 8.082000000000002e-06,
"loss": 1.0203,
"step": 9600
},
{
"epoch": 149.23076923076923,
"grad_norm": 68.70179748535156,
"learning_rate": 8.062000000000002e-06,
"loss": 1.0413,
"step": 9700
},
{
"epoch": 150.76923076923077,
"grad_norm": 49.53428268432617,
"learning_rate": 8.0422e-06,
"loss": 1.0199,
"step": 9800
},
{
"epoch": 152.30769230769232,
"grad_norm": 33.95280075073242,
"learning_rate": 8.022200000000001e-06,
"loss": 1.0278,
"step": 9900
},
{
"epoch": 153.84615384615384,
"grad_norm": 75.1406021118164,
"learning_rate": 8.002200000000001e-06,
"loss": 1.0093,
"step": 10000
},
{
"epoch": 155.3846153846154,
"grad_norm": 33.94615173339844,
"learning_rate": 7.982200000000001e-06,
"loss": 1.0431,
"step": 10100
},
{
"epoch": 156.92307692307693,
"grad_norm": 40.253597259521484,
"learning_rate": 7.962200000000001e-06,
"loss": 0.9965,
"step": 10200
},
{
"epoch": 158.46153846153845,
"grad_norm": 45.14350128173828,
"learning_rate": 7.942200000000002e-06,
"loss": 1.0309,
"step": 10300
},
{
"epoch": 160.0,
"grad_norm": 21.002464294433594,
"learning_rate": 7.922200000000002e-06,
"loss": 1.0016,
"step": 10400
},
{
"epoch": 161.53846153846155,
"grad_norm": 35.12867736816406,
"learning_rate": 7.9022e-06,
"loss": 1.0121,
"step": 10500
},
{
"epoch": 163.07692307692307,
"grad_norm": 79.60009002685547,
"learning_rate": 7.8822e-06,
"loss": 1.0231,
"step": 10600
},
{
"epoch": 164.6153846153846,
"grad_norm": 226.0706787109375,
"learning_rate": 7.8622e-06,
"loss": 1.0305,
"step": 10700
},
{
"epoch": 166.15384615384616,
"grad_norm": 23.051137924194336,
"learning_rate": 7.842200000000001e-06,
"loss": 0.9855,
"step": 10800
},
{
"epoch": 167.69230769230768,
"grad_norm": 45.090057373046875,
"learning_rate": 7.822200000000001e-06,
"loss": 1.0162,
"step": 10900
},
{
"epoch": 169.23076923076923,
"grad_norm": 30.125905990600586,
"learning_rate": 7.802200000000001e-06,
"loss": 1.0,
"step": 11000
},
{
"epoch": 170.76923076923077,
"grad_norm": 69.0078125,
"learning_rate": 7.782200000000001e-06,
"loss": 1.0363,
"step": 11100
},
{
"epoch": 172.30769230769232,
"grad_norm": 31.446842193603516,
"learning_rate": 7.762200000000002e-06,
"loss": 1.0092,
"step": 11200
},
{
"epoch": 173.84615384615384,
"grad_norm": 72.3807144165039,
"learning_rate": 7.7422e-06,
"loss": 1.0052,
"step": 11300
},
{
"epoch": 175.3846153846154,
"grad_norm": 42.25762176513672,
"learning_rate": 7.7222e-06,
"loss": 1.0183,
"step": 11400
},
{
"epoch": 176.92307692307693,
"grad_norm": 37.23869323730469,
"learning_rate": 7.7022e-06,
"loss": 1.0029,
"step": 11500
},
{
"epoch": 178.46153846153845,
"grad_norm": 123.58518981933594,
"learning_rate": 7.6822e-06,
"loss": 1.0043,
"step": 11600
},
{
"epoch": 180.0,
"grad_norm": 36.04173278808594,
"learning_rate": 7.662200000000001e-06,
"loss": 1.0249,
"step": 11700
},
{
"epoch": 181.53846153846155,
"grad_norm": 54.06993103027344,
"learning_rate": 7.642200000000001e-06,
"loss": 1.0203,
"step": 11800
},
{
"epoch": 183.07692307692307,
"grad_norm": 23.067537307739258,
"learning_rate": 7.6222000000000005e-06,
"loss": 1.011,
"step": 11900
},
{
"epoch": 184.6153846153846,
"grad_norm": 38.63265609741211,
"learning_rate": 7.602400000000001e-06,
"loss": 1.0129,
"step": 12000
},
{
"epoch": 186.15384615384616,
"grad_norm": 67.62531280517578,
"learning_rate": 7.5824000000000005e-06,
"loss": 0.9953,
"step": 12100
},
{
"epoch": 187.69230769230768,
"grad_norm": 60.271663665771484,
"learning_rate": 7.562400000000001e-06,
"loss": 0.9949,
"step": 12200
},
{
"epoch": 189.23076923076923,
"grad_norm": 85.98783111572266,
"learning_rate": 7.542400000000001e-06,
"loss": 1.008,
"step": 12300
},
{
"epoch": 190.76923076923077,
"grad_norm": 60.040924072265625,
"learning_rate": 7.522400000000001e-06,
"loss": 0.9893,
"step": 12400
},
{
"epoch": 192.30769230769232,
"grad_norm": 34.971290588378906,
"learning_rate": 7.5024e-06,
"loss": 0.999,
"step": 12500
},
{
"epoch": 193.84615384615384,
"grad_norm": 47.212989807128906,
"learning_rate": 7.4824000000000006e-06,
"loss": 0.993,
"step": 12600
},
{
"epoch": 195.3846153846154,
"grad_norm": 38.17967987060547,
"learning_rate": 7.462400000000001e-06,
"loss": 1.0284,
"step": 12700
},
{
"epoch": 196.92307692307693,
"grad_norm": 47.6786003112793,
"learning_rate": 7.442400000000001e-06,
"loss": 0.9636,
"step": 12800
},
{
"epoch": 198.46153846153845,
"grad_norm": 31.550769805908203,
"learning_rate": 7.422400000000001e-06,
"loss": 1.0082,
"step": 12900
},
{
"epoch": 200.0,
"grad_norm": 176.42813110351562,
"learning_rate": 7.4024000000000005e-06,
"loss": 0.9848,
"step": 13000
},
{
"epoch": 201.53846153846155,
"grad_norm": 26.10734748840332,
"learning_rate": 7.382600000000001e-06,
"loss": 1.0191,
"step": 13100
},
{
"epoch": 203.07692307692307,
"grad_norm": 28.24509048461914,
"learning_rate": 7.3626000000000005e-06,
"loss": 0.9799,
"step": 13200
},
{
"epoch": 204.6153846153846,
"grad_norm": 27.511566162109375,
"learning_rate": 7.342600000000001e-06,
"loss": 0.9935,
"step": 13300
},
{
"epoch": 206.15384615384616,
"grad_norm": 43.895633697509766,
"learning_rate": 7.322600000000001e-06,
"loss": 1.0101,
"step": 13400
},
{
"epoch": 207.69230769230768,
"grad_norm": 73.81085968017578,
"learning_rate": 7.302600000000001e-06,
"loss": 0.9559,
"step": 13500
}
],
"logging_steps": 100,
"max_steps": 50000,
"num_input_tokens_seen": 0,
"num_train_epochs": 770,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 5.160282743808e+19,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}