anas-awadalla's picture
End of training
fa326f4
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"global_step": 21900,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 2.9863013698630136e-05,
"loss": 1.1148,
"step": 100
},
{
"epoch": 0.02,
"learning_rate": 2.9726027397260275e-05,
"loss": 1.0017,
"step": 200
},
{
"epoch": 0.03,
"learning_rate": 2.958904109589041e-05,
"loss": 0.9283,
"step": 300
},
{
"epoch": 0.04,
"learning_rate": 2.945205479452055e-05,
"loss": 0.8957,
"step": 400
},
{
"epoch": 0.05,
"learning_rate": 2.9315068493150685e-05,
"loss": 0.757,
"step": 500
},
{
"epoch": 0.05,
"learning_rate": 2.9178082191780824e-05,
"loss": 0.7515,
"step": 600
},
{
"epoch": 0.06,
"learning_rate": 2.9041095890410956e-05,
"loss": 0.7115,
"step": 700
},
{
"epoch": 0.07,
"learning_rate": 2.8904109589041095e-05,
"loss": 0.7266,
"step": 800
},
{
"epoch": 0.08,
"learning_rate": 2.876712328767123e-05,
"loss": 0.6896,
"step": 900
},
{
"epoch": 0.09,
"learning_rate": 2.863013698630137e-05,
"loss": 0.6487,
"step": 1000
},
{
"epoch": 0.1,
"learning_rate": 2.8493150684931505e-05,
"loss": 0.6126,
"step": 1100
},
{
"epoch": 0.11,
"learning_rate": 2.8356164383561644e-05,
"loss": 0.6426,
"step": 1200
},
{
"epoch": 0.12,
"learning_rate": 2.821917808219178e-05,
"loss": 0.6459,
"step": 1300
},
{
"epoch": 0.13,
"learning_rate": 2.808219178082192e-05,
"loss": 0.6529,
"step": 1400
},
{
"epoch": 0.14,
"learning_rate": 2.7945205479452054e-05,
"loss": 0.5985,
"step": 1500
},
{
"epoch": 0.15,
"learning_rate": 2.7808219178082193e-05,
"loss": 0.6394,
"step": 1600
},
{
"epoch": 0.16,
"learning_rate": 2.767123287671233e-05,
"loss": 0.6565,
"step": 1700
},
{
"epoch": 0.16,
"learning_rate": 2.7534246575342468e-05,
"loss": 0.5846,
"step": 1800
},
{
"epoch": 0.17,
"learning_rate": 2.7397260273972603e-05,
"loss": 0.5446,
"step": 1900
},
{
"epoch": 0.18,
"learning_rate": 2.726027397260274e-05,
"loss": 0.582,
"step": 2000
},
{
"epoch": 0.19,
"learning_rate": 2.7123287671232875e-05,
"loss": 0.5802,
"step": 2100
},
{
"epoch": 0.2,
"learning_rate": 2.6986301369863014e-05,
"loss": 0.5676,
"step": 2200
},
{
"epoch": 0.21,
"learning_rate": 2.684931506849315e-05,
"loss": 0.5694,
"step": 2300
},
{
"epoch": 0.22,
"learning_rate": 2.6712328767123288e-05,
"loss": 0.5531,
"step": 2400
},
{
"epoch": 0.23,
"learning_rate": 2.6575342465753424e-05,
"loss": 0.5641,
"step": 2500
},
{
"epoch": 0.24,
"learning_rate": 2.6438356164383563e-05,
"loss": 0.5514,
"step": 2600
},
{
"epoch": 0.25,
"learning_rate": 2.6301369863013698e-05,
"loss": 0.5453,
"step": 2700
},
{
"epoch": 0.26,
"learning_rate": 2.6164383561643837e-05,
"loss": 0.5783,
"step": 2800
},
{
"epoch": 0.26,
"learning_rate": 2.6027397260273973e-05,
"loss": 0.548,
"step": 2900
},
{
"epoch": 0.27,
"learning_rate": 2.589041095890411e-05,
"loss": 0.605,
"step": 3000
},
{
"epoch": 0.28,
"learning_rate": 2.5753424657534247e-05,
"loss": 0.5323,
"step": 3100
},
{
"epoch": 0.29,
"learning_rate": 2.5616438356164386e-05,
"loss": 0.5091,
"step": 3200
},
{
"epoch": 0.3,
"learning_rate": 2.5479452054794518e-05,
"loss": 0.51,
"step": 3300
},
{
"epoch": 0.31,
"learning_rate": 2.5342465753424657e-05,
"loss": 0.5267,
"step": 3400
},
{
"epoch": 0.32,
"learning_rate": 2.5205479452054793e-05,
"loss": 0.5517,
"step": 3500
},
{
"epoch": 0.33,
"learning_rate": 2.5068493150684932e-05,
"loss": 0.5293,
"step": 3600
},
{
"epoch": 0.34,
"learning_rate": 2.4931506849315067e-05,
"loss": 0.5477,
"step": 3700
},
{
"epoch": 0.35,
"learning_rate": 2.4794520547945206e-05,
"loss": 0.5383,
"step": 3800
},
{
"epoch": 0.36,
"learning_rate": 2.4657534246575342e-05,
"loss": 0.5236,
"step": 3900
},
{
"epoch": 0.37,
"learning_rate": 2.452054794520548e-05,
"loss": 0.5478,
"step": 4000
},
{
"epoch": 0.37,
"learning_rate": 2.4383561643835616e-05,
"loss": 0.5282,
"step": 4100
},
{
"epoch": 0.38,
"learning_rate": 2.4246575342465755e-05,
"loss": 0.5012,
"step": 4200
},
{
"epoch": 0.39,
"learning_rate": 2.410958904109589e-05,
"loss": 0.5451,
"step": 4300
},
{
"epoch": 0.4,
"learning_rate": 2.397260273972603e-05,
"loss": 0.5103,
"step": 4400
},
{
"epoch": 0.41,
"learning_rate": 2.3835616438356165e-05,
"loss": 0.5231,
"step": 4500
},
{
"epoch": 0.42,
"learning_rate": 2.36986301369863e-05,
"loss": 0.4836,
"step": 4600
},
{
"epoch": 0.43,
"learning_rate": 2.3561643835616436e-05,
"loss": 0.4855,
"step": 4700
},
{
"epoch": 0.44,
"learning_rate": 2.3424657534246575e-05,
"loss": 0.5002,
"step": 4800
},
{
"epoch": 0.45,
"learning_rate": 2.328767123287671e-05,
"loss": 0.5032,
"step": 4900
},
{
"epoch": 0.46,
"learning_rate": 2.315068493150685e-05,
"loss": 0.4902,
"step": 5000
},
{
"epoch": 0.47,
"learning_rate": 2.3013698630136985e-05,
"loss": 0.4891,
"step": 5100
},
{
"epoch": 0.47,
"learning_rate": 2.2876712328767124e-05,
"loss": 0.4582,
"step": 5200
},
{
"epoch": 0.48,
"learning_rate": 2.273972602739726e-05,
"loss": 0.445,
"step": 5300
},
{
"epoch": 0.49,
"learning_rate": 2.26027397260274e-05,
"loss": 0.4532,
"step": 5400
},
{
"epoch": 0.5,
"learning_rate": 2.2465753424657534e-05,
"loss": 0.4889,
"step": 5500
},
{
"epoch": 0.51,
"learning_rate": 2.2328767123287673e-05,
"loss": 0.4642,
"step": 5600
},
{
"epoch": 0.52,
"learning_rate": 2.219178082191781e-05,
"loss": 0.4235,
"step": 5700
},
{
"epoch": 0.53,
"learning_rate": 2.2054794520547945e-05,
"loss": 0.4771,
"step": 5800
},
{
"epoch": 0.54,
"learning_rate": 2.191780821917808e-05,
"loss": 0.4534,
"step": 5900
},
{
"epoch": 0.55,
"learning_rate": 2.178082191780822e-05,
"loss": 0.4727,
"step": 6000
},
{
"epoch": 0.56,
"learning_rate": 2.1643835616438355e-05,
"loss": 0.4447,
"step": 6100
},
{
"epoch": 0.57,
"learning_rate": 2.1506849315068494e-05,
"loss": 0.48,
"step": 6200
},
{
"epoch": 0.58,
"learning_rate": 2.136986301369863e-05,
"loss": 0.4122,
"step": 6300
},
{
"epoch": 0.58,
"learning_rate": 2.1232876712328768e-05,
"loss": 0.4946,
"step": 6400
},
{
"epoch": 0.59,
"learning_rate": 2.1095890410958904e-05,
"loss": 0.4547,
"step": 6500
},
{
"epoch": 0.6,
"learning_rate": 2.0958904109589043e-05,
"loss": 0.4466,
"step": 6600
},
{
"epoch": 0.61,
"learning_rate": 2.0821917808219178e-05,
"loss": 0.482,
"step": 6700
},
{
"epoch": 0.62,
"learning_rate": 2.0684931506849317e-05,
"loss": 0.5153,
"step": 6800
},
{
"epoch": 0.63,
"learning_rate": 2.0547945205479453e-05,
"loss": 0.5148,
"step": 6900
},
{
"epoch": 0.64,
"learning_rate": 2.041095890410959e-05,
"loss": 0.5056,
"step": 7000
},
{
"epoch": 0.65,
"learning_rate": 2.0273972602739724e-05,
"loss": 0.4339,
"step": 7100
},
{
"epoch": 0.66,
"learning_rate": 2.0136986301369863e-05,
"loss": 0.5143,
"step": 7200
},
{
"epoch": 0.67,
"learning_rate": 1.9999999999999998e-05,
"loss": 0.3956,
"step": 7300
},
{
"epoch": 0.68,
"learning_rate": 1.9863013698630137e-05,
"loss": 0.4078,
"step": 7400
},
{
"epoch": 0.68,
"learning_rate": 1.9726027397260273e-05,
"loss": 0.4864,
"step": 7500
},
{
"epoch": 0.69,
"learning_rate": 1.9589041095890412e-05,
"loss": 0.4269,
"step": 7600
},
{
"epoch": 0.7,
"learning_rate": 1.9452054794520547e-05,
"loss": 0.4438,
"step": 7700
},
{
"epoch": 0.71,
"learning_rate": 1.9315068493150686e-05,
"loss": 0.4499,
"step": 7800
},
{
"epoch": 0.72,
"learning_rate": 1.9178082191780822e-05,
"loss": 0.4279,
"step": 7900
},
{
"epoch": 0.73,
"learning_rate": 1.904109589041096e-05,
"loss": 0.4544,
"step": 8000
},
{
"epoch": 0.74,
"learning_rate": 1.8904109589041096e-05,
"loss": 0.4709,
"step": 8100
},
{
"epoch": 0.75,
"learning_rate": 1.8767123287671235e-05,
"loss": 0.4286,
"step": 8200
},
{
"epoch": 0.76,
"learning_rate": 1.863013698630137e-05,
"loss": 0.4371,
"step": 8300
},
{
"epoch": 0.77,
"learning_rate": 1.8493150684931506e-05,
"loss": 0.4226,
"step": 8400
},
{
"epoch": 0.78,
"learning_rate": 1.8356164383561642e-05,
"loss": 0.4145,
"step": 8500
},
{
"epoch": 0.79,
"learning_rate": 1.821917808219178e-05,
"loss": 0.4068,
"step": 8600
},
{
"epoch": 0.79,
"learning_rate": 1.8082191780821916e-05,
"loss": 0.3954,
"step": 8700
},
{
"epoch": 0.8,
"learning_rate": 1.7945205479452055e-05,
"loss": 0.4526,
"step": 8800
},
{
"epoch": 0.81,
"learning_rate": 1.780821917808219e-05,
"loss": 0.4066,
"step": 8900
},
{
"epoch": 0.82,
"learning_rate": 1.767123287671233e-05,
"loss": 0.4264,
"step": 9000
},
{
"epoch": 0.83,
"learning_rate": 1.7534246575342465e-05,
"loss": 0.3907,
"step": 9100
},
{
"epoch": 0.84,
"learning_rate": 1.7397260273972604e-05,
"loss": 0.4977,
"step": 9200
},
{
"epoch": 0.85,
"learning_rate": 1.726027397260274e-05,
"loss": 0.4318,
"step": 9300
},
{
"epoch": 0.86,
"learning_rate": 1.712328767123288e-05,
"loss": 0.4517,
"step": 9400
},
{
"epoch": 0.87,
"learning_rate": 1.6986301369863014e-05,
"loss": 0.4816,
"step": 9500
},
{
"epoch": 0.88,
"learning_rate": 1.6849315068493153e-05,
"loss": 0.4393,
"step": 9600
},
{
"epoch": 0.89,
"learning_rate": 1.6712328767123286e-05,
"loss": 0.4194,
"step": 9700
},
{
"epoch": 0.89,
"learning_rate": 1.6575342465753425e-05,
"loss": 0.4023,
"step": 9800
},
{
"epoch": 0.9,
"learning_rate": 1.643835616438356e-05,
"loss": 0.3988,
"step": 9900
},
{
"epoch": 0.91,
"learning_rate": 1.63013698630137e-05,
"loss": 0.4305,
"step": 10000
},
{
"epoch": 0.92,
"learning_rate": 1.6164383561643835e-05,
"loss": 0.385,
"step": 10100
},
{
"epoch": 0.93,
"learning_rate": 1.6027397260273974e-05,
"loss": 0.4318,
"step": 10200
},
{
"epoch": 0.94,
"learning_rate": 1.589041095890411e-05,
"loss": 0.3975,
"step": 10300
},
{
"epoch": 0.95,
"learning_rate": 1.5753424657534248e-05,
"loss": 0.41,
"step": 10400
},
{
"epoch": 0.96,
"learning_rate": 1.5616438356164384e-05,
"loss": 0.4079,
"step": 10500
},
{
"epoch": 0.97,
"learning_rate": 1.5479452054794523e-05,
"loss": 0.4153,
"step": 10600
},
{
"epoch": 0.98,
"learning_rate": 1.5342465753424658e-05,
"loss": 0.4091,
"step": 10700
},
{
"epoch": 0.99,
"learning_rate": 1.5205479452054795e-05,
"loss": 0.4031,
"step": 10800
},
{
"epoch": 1.0,
"learning_rate": 1.5068493150684931e-05,
"loss": 0.3917,
"step": 10900
},
{
"epoch": 1.0,
"learning_rate": 1.4931506849315068e-05,
"loss": 0.3244,
"step": 11000
},
{
"epoch": 1.01,
"learning_rate": 1.4794520547945205e-05,
"loss": 0.3067,
"step": 11100
},
{
"epoch": 1.02,
"learning_rate": 1.4657534246575343e-05,
"loss": 0.2667,
"step": 11200
},
{
"epoch": 1.03,
"learning_rate": 1.4520547945205478e-05,
"loss": 0.3103,
"step": 11300
},
{
"epoch": 1.04,
"learning_rate": 1.4383561643835615e-05,
"loss": 0.3499,
"step": 11400
},
{
"epoch": 1.05,
"learning_rate": 1.4246575342465753e-05,
"loss": 0.2896,
"step": 11500
},
{
"epoch": 1.06,
"learning_rate": 1.410958904109589e-05,
"loss": 0.3046,
"step": 11600
},
{
"epoch": 1.07,
"learning_rate": 1.3972602739726027e-05,
"loss": 0.3087,
"step": 11700
},
{
"epoch": 1.08,
"learning_rate": 1.3835616438356164e-05,
"loss": 0.3079,
"step": 11800
},
{
"epoch": 1.09,
"learning_rate": 1.3698630136986302e-05,
"loss": 0.2548,
"step": 11900
},
{
"epoch": 1.1,
"learning_rate": 1.3561643835616437e-05,
"loss": 0.2825,
"step": 12000
},
{
"epoch": 1.11,
"learning_rate": 1.3424657534246575e-05,
"loss": 0.2897,
"step": 12100
},
{
"epoch": 1.11,
"learning_rate": 1.3287671232876712e-05,
"loss": 0.3052,
"step": 12200
},
{
"epoch": 1.12,
"learning_rate": 1.3150684931506849e-05,
"loss": 0.2706,
"step": 12300
},
{
"epoch": 1.13,
"learning_rate": 1.3013698630136986e-05,
"loss": 0.3199,
"step": 12400
},
{
"epoch": 1.14,
"learning_rate": 1.2876712328767124e-05,
"loss": 0.3175,
"step": 12500
},
{
"epoch": 1.15,
"learning_rate": 1.2739726027397259e-05,
"loss": 0.3024,
"step": 12600
},
{
"epoch": 1.16,
"learning_rate": 1.2602739726027396e-05,
"loss": 0.2762,
"step": 12700
},
{
"epoch": 1.17,
"learning_rate": 1.2465753424657534e-05,
"loss": 0.288,
"step": 12800
},
{
"epoch": 1.18,
"learning_rate": 1.2328767123287671e-05,
"loss": 0.2675,
"step": 12900
},
{
"epoch": 1.19,
"learning_rate": 1.2191780821917808e-05,
"loss": 0.2542,
"step": 13000
},
{
"epoch": 1.2,
"learning_rate": 1.2054794520547945e-05,
"loss": 0.313,
"step": 13100
},
{
"epoch": 1.21,
"learning_rate": 1.1917808219178083e-05,
"loss": 0.2568,
"step": 13200
},
{
"epoch": 1.21,
"learning_rate": 1.1780821917808218e-05,
"loss": 0.2729,
"step": 13300
},
{
"epoch": 1.22,
"learning_rate": 1.1643835616438355e-05,
"loss": 0.3091,
"step": 13400
},
{
"epoch": 1.23,
"learning_rate": 1.1506849315068493e-05,
"loss": 0.3367,
"step": 13500
},
{
"epoch": 1.24,
"learning_rate": 1.136986301369863e-05,
"loss": 0.2847,
"step": 13600
},
{
"epoch": 1.25,
"learning_rate": 1.1232876712328767e-05,
"loss": 0.2602,
"step": 13700
},
{
"epoch": 1.26,
"learning_rate": 1.1095890410958904e-05,
"loss": 0.2789,
"step": 13800
},
{
"epoch": 1.27,
"learning_rate": 1.095890410958904e-05,
"loss": 0.3117,
"step": 13900
},
{
"epoch": 1.28,
"learning_rate": 1.0821917808219177e-05,
"loss": 0.2818,
"step": 14000
},
{
"epoch": 1.29,
"learning_rate": 1.0684931506849315e-05,
"loss": 0.2976,
"step": 14100
},
{
"epoch": 1.3,
"learning_rate": 1.0547945205479452e-05,
"loss": 0.2871,
"step": 14200
},
{
"epoch": 1.31,
"learning_rate": 1.0410958904109589e-05,
"loss": 0.2976,
"step": 14300
},
{
"epoch": 1.32,
"learning_rate": 1.0273972602739726e-05,
"loss": 0.2619,
"step": 14400
},
{
"epoch": 1.32,
"learning_rate": 1.0136986301369862e-05,
"loss": 0.2654,
"step": 14500
},
{
"epoch": 1.33,
"learning_rate": 9.999999999999999e-06,
"loss": 0.28,
"step": 14600
},
{
"epoch": 1.34,
"learning_rate": 9.863013698630136e-06,
"loss": 0.3073,
"step": 14700
},
{
"epoch": 1.35,
"learning_rate": 9.726027397260274e-06,
"loss": 0.2485,
"step": 14800
},
{
"epoch": 1.36,
"learning_rate": 9.589041095890411e-06,
"loss": 0.2682,
"step": 14900
},
{
"epoch": 1.37,
"learning_rate": 9.452054794520548e-06,
"loss": 0.2881,
"step": 15000
},
{
"epoch": 1.38,
"learning_rate": 9.315068493150685e-06,
"loss": 0.2718,
"step": 15100
},
{
"epoch": 1.39,
"learning_rate": 9.178082191780821e-06,
"loss": 0.2753,
"step": 15200
},
{
"epoch": 1.4,
"learning_rate": 9.041095890410958e-06,
"loss": 0.2856,
"step": 15300
},
{
"epoch": 1.41,
"learning_rate": 8.904109589041095e-06,
"loss": 0.2777,
"step": 15400
},
{
"epoch": 1.42,
"learning_rate": 8.767123287671233e-06,
"loss": 0.266,
"step": 15500
},
{
"epoch": 1.42,
"learning_rate": 8.63013698630137e-06,
"loss": 0.2611,
"step": 15600
},
{
"epoch": 1.43,
"learning_rate": 8.493150684931507e-06,
"loss": 0.2434,
"step": 15700
},
{
"epoch": 1.44,
"learning_rate": 8.356164383561643e-06,
"loss": 0.2699,
"step": 15800
},
{
"epoch": 1.45,
"learning_rate": 8.21917808219178e-06,
"loss": 0.2832,
"step": 15900
},
{
"epoch": 1.46,
"learning_rate": 8.082191780821917e-06,
"loss": 0.2502,
"step": 16000
},
{
"epoch": 1.47,
"learning_rate": 7.945205479452055e-06,
"loss": 0.2921,
"step": 16100
},
{
"epoch": 1.48,
"learning_rate": 7.808219178082192e-06,
"loss": 0.3346,
"step": 16200
},
{
"epoch": 1.49,
"learning_rate": 7.671232876712329e-06,
"loss": 0.3443,
"step": 16300
},
{
"epoch": 1.5,
"learning_rate": 7.5342465753424655e-06,
"loss": 0.2922,
"step": 16400
},
{
"epoch": 1.51,
"learning_rate": 7.397260273972603e-06,
"loss": 0.2725,
"step": 16500
},
{
"epoch": 1.52,
"learning_rate": 7.260273972602739e-06,
"loss": 0.2671,
"step": 16600
},
{
"epoch": 1.53,
"learning_rate": 7.123287671232876e-06,
"loss": 0.2702,
"step": 16700
},
{
"epoch": 1.53,
"learning_rate": 6.986301369863014e-06,
"loss": 0.2703,
"step": 16800
},
{
"epoch": 1.54,
"learning_rate": 6.849315068493151e-06,
"loss": 0.2809,
"step": 16900
},
{
"epoch": 1.55,
"learning_rate": 6.712328767123287e-06,
"loss": 0.2579,
"step": 17000
},
{
"epoch": 1.56,
"learning_rate": 6.5753424657534245e-06,
"loss": 0.2386,
"step": 17100
},
{
"epoch": 1.57,
"learning_rate": 6.438356164383562e-06,
"loss": 0.2569,
"step": 17200
},
{
"epoch": 1.58,
"learning_rate": 6.301369863013698e-06,
"loss": 0.2703,
"step": 17300
},
{
"epoch": 1.59,
"learning_rate": 6.1643835616438354e-06,
"loss": 0.2332,
"step": 17400
},
{
"epoch": 1.6,
"learning_rate": 6.027397260273973e-06,
"loss": 0.2649,
"step": 17500
},
{
"epoch": 1.61,
"learning_rate": 5.890410958904109e-06,
"loss": 0.2677,
"step": 17600
},
{
"epoch": 1.62,
"learning_rate": 5.753424657534246e-06,
"loss": 0.2745,
"step": 17700
},
{
"epoch": 1.63,
"learning_rate": 5.616438356164384e-06,
"loss": 0.2344,
"step": 17800
},
{
"epoch": 1.63,
"learning_rate": 5.47945205479452e-06,
"loss": 0.2369,
"step": 17900
},
{
"epoch": 1.64,
"learning_rate": 5.342465753424657e-06,
"loss": 0.2512,
"step": 18000
},
{
"epoch": 1.65,
"learning_rate": 5.2054794520547945e-06,
"loss": 0.2674,
"step": 18100
},
{
"epoch": 1.66,
"learning_rate": 5.068493150684931e-06,
"loss": 0.2839,
"step": 18200
},
{
"epoch": 1.67,
"learning_rate": 4.931506849315068e-06,
"loss": 0.236,
"step": 18300
},
{
"epoch": 1.68,
"learning_rate": 4.7945205479452054e-06,
"loss": 0.2421,
"step": 18400
},
{
"epoch": 1.69,
"learning_rate": 4.657534246575343e-06,
"loss": 0.2448,
"step": 18500
},
{
"epoch": 1.7,
"learning_rate": 4.520547945205479e-06,
"loss": 0.2428,
"step": 18600
},
{
"epoch": 1.71,
"learning_rate": 4.383561643835616e-06,
"loss": 0.2438,
"step": 18700
},
{
"epoch": 1.72,
"learning_rate": 4.246575342465754e-06,
"loss": 0.2305,
"step": 18800
},
{
"epoch": 1.73,
"learning_rate": 4.10958904109589e-06,
"loss": 0.2358,
"step": 18900
},
{
"epoch": 1.74,
"learning_rate": 3.972602739726027e-06,
"loss": 0.2731,
"step": 19000
},
{
"epoch": 1.74,
"learning_rate": 3.8356164383561645e-06,
"loss": 0.2617,
"step": 19100
},
{
"epoch": 1.75,
"learning_rate": 3.6986301369863014e-06,
"loss": 0.2378,
"step": 19200
},
{
"epoch": 1.76,
"learning_rate": 3.561643835616438e-06,
"loss": 0.2678,
"step": 19300
},
{
"epoch": 1.77,
"learning_rate": 3.4246575342465754e-06,
"loss": 0.2592,
"step": 19400
},
{
"epoch": 1.78,
"learning_rate": 3.2876712328767123e-06,
"loss": 0.2354,
"step": 19500
},
{
"epoch": 1.79,
"learning_rate": 3.150684931506849e-06,
"loss": 0.2393,
"step": 19600
},
{
"epoch": 1.8,
"learning_rate": 3.0136986301369864e-06,
"loss": 0.2402,
"step": 19700
},
{
"epoch": 1.81,
"learning_rate": 2.876712328767123e-06,
"loss": 0.2333,
"step": 19800
},
{
"epoch": 1.82,
"learning_rate": 2.73972602739726e-06,
"loss": 0.239,
"step": 19900
},
{
"epoch": 1.83,
"learning_rate": 2.6027397260273973e-06,
"loss": 0.2619,
"step": 20000
},
{
"epoch": 1.84,
"learning_rate": 2.465753424657534e-06,
"loss": 0.2502,
"step": 20100
},
{
"epoch": 1.84,
"learning_rate": 2.3287671232876713e-06,
"loss": 0.2267,
"step": 20200
},
{
"epoch": 1.85,
"learning_rate": 2.191780821917808e-06,
"loss": 0.2566,
"step": 20300
},
{
"epoch": 1.86,
"learning_rate": 2.054794520547945e-06,
"loss": 0.2299,
"step": 20400
},
{
"epoch": 1.87,
"learning_rate": 1.9178082191780823e-06,
"loss": 0.2352,
"step": 20500
},
{
"epoch": 1.88,
"learning_rate": 1.780821917808219e-06,
"loss": 0.2394,
"step": 20600
},
{
"epoch": 1.89,
"learning_rate": 1.6438356164383561e-06,
"loss": 0.2366,
"step": 20700
},
{
"epoch": 1.9,
"learning_rate": 1.5068493150684932e-06,
"loss": 0.2282,
"step": 20800
},
{
"epoch": 1.91,
"learning_rate": 1.36986301369863e-06,
"loss": 0.2312,
"step": 20900
},
{
"epoch": 1.92,
"learning_rate": 1.232876712328767e-06,
"loss": 0.2674,
"step": 21000
},
{
"epoch": 1.93,
"learning_rate": 1.095890410958904e-06,
"loss": 0.2318,
"step": 21100
},
{
"epoch": 1.94,
"learning_rate": 9.589041095890411e-07,
"loss": 0.2591,
"step": 21200
},
{
"epoch": 1.95,
"learning_rate": 8.219178082191781e-07,
"loss": 0.228,
"step": 21300
},
{
"epoch": 1.95,
"learning_rate": 6.84931506849315e-07,
"loss": 0.2428,
"step": 21400
},
{
"epoch": 1.96,
"learning_rate": 5.47945205479452e-07,
"loss": 0.2499,
"step": 21500
},
{
"epoch": 1.97,
"learning_rate": 4.1095890410958903e-07,
"loss": 0.2427,
"step": 21600
},
{
"epoch": 1.98,
"learning_rate": 2.73972602739726e-07,
"loss": 0.2236,
"step": 21700
},
{
"epoch": 1.99,
"learning_rate": 1.36986301369863e-07,
"loss": 0.2067,
"step": 21800
},
{
"epoch": 2.0,
"learning_rate": 0.0,
"loss": 0.2336,
"step": 21900
},
{
"epoch": 2.0,
"step": 21900,
"total_flos": 1.4237714703030682e+17,
"train_loss": 0.39270679595807917,
"train_runtime": 10191.0343,
"train_samples_per_second": 17.191,
"train_steps_per_second": 2.149
}
],
"max_steps": 21900,
"num_train_epochs": 2,
"total_flos": 1.4237714703030682e+17,
"trial_name": null,
"trial_params": null
}