Training checkpoint at step 17000
Browse files- trainer_state.json +366 -6
trainer_state.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"best_global_step":
|
| 3 |
-
"best_metric": 2.
|
| 4 |
-
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_8x1024_mem32_bs64_hf_armt_dmem64/run_21/checkpoint-
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 100,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -5768,6 +5768,366 @@
|
|
| 5768 |
"eval_samples_per_second": 3.19,
|
| 5769 |
"eval_steps_per_second": 1.595,
|
| 5770 |
"step": 16000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5771 |
}
|
| 5772 |
],
|
| 5773 |
"logging_steps": 25,
|
|
@@ -5787,7 +6147,7 @@
|
|
| 5787 |
"attributes": {}
|
| 5788 |
}
|
| 5789 |
},
|
| 5790 |
-
"total_flos": 5.
|
| 5791 |
"train_batch_size": 1,
|
| 5792 |
"trial_name": null,
|
| 5793 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 17000,
|
| 3 |
+
"best_metric": 2.394216775894165,
|
| 4 |
+
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_8x1024_mem32_bs64_hf_armt_dmem64/run_21/checkpoint-17000",
|
| 5 |
+
"epoch": 0.34,
|
| 6 |
"eval_steps": 100,
|
| 7 |
+
"global_step": 17000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 5768 |
"eval_samples_per_second": 3.19,
|
| 5769 |
"eval_steps_per_second": 1.595,
|
| 5770 |
"step": 16000
|
| 5771 |
+
},
|
| 5772 |
+
{
|
| 5773 |
+
"epoch": 0.3205,
|
| 5774 |
+
"grad_norm": 0.5490540509150809,
|
| 5775 |
+
"learning_rate": 7.550222222222223e-06,
|
| 5776 |
+
"loss": 2.3908,
|
| 5777 |
+
"step": 16025
|
| 5778 |
+
},
|
| 5779 |
+
{
|
| 5780 |
+
"epoch": 0.321,
|
| 5781 |
+
"grad_norm": 0.5604566538327537,
|
| 5782 |
+
"learning_rate": 7.5446666666666665e-06,
|
| 5783 |
+
"loss": 2.3816,
|
| 5784 |
+
"step": 16050
|
| 5785 |
+
},
|
| 5786 |
+
{
|
| 5787 |
+
"epoch": 0.3215,
|
| 5788 |
+
"grad_norm": 0.5482351645184266,
|
| 5789 |
+
"learning_rate": 7.539111111111112e-06,
|
| 5790 |
+
"loss": 2.3783,
|
| 5791 |
+
"step": 16075
|
| 5792 |
+
},
|
| 5793 |
+
{
|
| 5794 |
+
"epoch": 0.322,
|
| 5795 |
+
"grad_norm": 0.5738611670880387,
|
| 5796 |
+
"learning_rate": 7.533555555555556e-06,
|
| 5797 |
+
"loss": 2.3807,
|
| 5798 |
+
"step": 16100
|
| 5799 |
+
},
|
| 5800 |
+
{
|
| 5801 |
+
"epoch": 0.322,
|
| 5802 |
+
"eval_loss": 2.3955187797546387,
|
| 5803 |
+
"eval_runtime": 31.7782,
|
| 5804 |
+
"eval_samples_per_second": 3.21,
|
| 5805 |
+
"eval_steps_per_second": 1.605,
|
| 5806 |
+
"step": 16100
|
| 5807 |
+
},
|
| 5808 |
+
{
|
| 5809 |
+
"epoch": 0.3225,
|
| 5810 |
+
"grad_norm": 0.6007459037823811,
|
| 5811 |
+
"learning_rate": 7.528000000000001e-06,
|
| 5812 |
+
"loss": 2.3908,
|
| 5813 |
+
"step": 16125
|
| 5814 |
+
},
|
| 5815 |
+
{
|
| 5816 |
+
"epoch": 0.323,
|
| 5817 |
+
"grad_norm": 0.5719140015142068,
|
| 5818 |
+
"learning_rate": 7.522444444444446e-06,
|
| 5819 |
+
"loss": 2.379,
|
| 5820 |
+
"step": 16150
|
| 5821 |
+
},
|
| 5822 |
+
{
|
| 5823 |
+
"epoch": 0.3235,
|
| 5824 |
+
"grad_norm": 0.5722843141001409,
|
| 5825 |
+
"learning_rate": 7.516888888888889e-06,
|
| 5826 |
+
"loss": 2.3831,
|
| 5827 |
+
"step": 16175
|
| 5828 |
+
},
|
| 5829 |
+
{
|
| 5830 |
+
"epoch": 0.324,
|
| 5831 |
+
"grad_norm": 0.5500359198684006,
|
| 5832 |
+
"learning_rate": 7.511333333333334e-06,
|
| 5833 |
+
"loss": 2.3899,
|
| 5834 |
+
"step": 16200
|
| 5835 |
+
},
|
| 5836 |
+
{
|
| 5837 |
+
"epoch": 0.324,
|
| 5838 |
+
"eval_loss": 2.3954145908355713,
|
| 5839 |
+
"eval_runtime": 31.9265,
|
| 5840 |
+
"eval_samples_per_second": 3.195,
|
| 5841 |
+
"eval_steps_per_second": 1.597,
|
| 5842 |
+
"step": 16200
|
| 5843 |
+
},
|
| 5844 |
+
{
|
| 5845 |
+
"epoch": 0.3245,
|
| 5846 |
+
"grad_norm": 0.5988197648020003,
|
| 5847 |
+
"learning_rate": 7.505777777777778e-06,
|
| 5848 |
+
"loss": 2.3768,
|
| 5849 |
+
"step": 16225
|
| 5850 |
+
},
|
| 5851 |
+
{
|
| 5852 |
+
"epoch": 0.325,
|
| 5853 |
+
"grad_norm": 0.566314534087209,
|
| 5854 |
+
"learning_rate": 7.5002222222222235e-06,
|
| 5855 |
+
"loss": 2.3731,
|
| 5856 |
+
"step": 16250
|
| 5857 |
+
},
|
| 5858 |
+
{
|
| 5859 |
+
"epoch": 0.3255,
|
| 5860 |
+
"grad_norm": 0.5462158611596983,
|
| 5861 |
+
"learning_rate": 7.494666666666667e-06,
|
| 5862 |
+
"loss": 2.3821,
|
| 5863 |
+
"step": 16275
|
| 5864 |
+
},
|
| 5865 |
+
{
|
| 5866 |
+
"epoch": 0.326,
|
| 5867 |
+
"grad_norm": 0.5546038414202229,
|
| 5868 |
+
"learning_rate": 7.4891111111111114e-06,
|
| 5869 |
+
"loss": 2.3725,
|
| 5870 |
+
"step": 16300
|
| 5871 |
+
},
|
| 5872 |
+
{
|
| 5873 |
+
"epoch": 0.326,
|
| 5874 |
+
"eval_loss": 2.395524501800537,
|
| 5875 |
+
"eval_runtime": 31.8126,
|
| 5876 |
+
"eval_samples_per_second": 3.206,
|
| 5877 |
+
"eval_steps_per_second": 1.603,
|
| 5878 |
+
"step": 16300
|
| 5879 |
+
},
|
| 5880 |
+
{
|
| 5881 |
+
"epoch": 0.3265,
|
| 5882 |
+
"grad_norm": 0.5596467845027929,
|
| 5883 |
+
"learning_rate": 7.483555555555556e-06,
|
| 5884 |
+
"loss": 2.3843,
|
| 5885 |
+
"step": 16325
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 0.327,
|
| 5889 |
+
"grad_norm": 0.5815120805791782,
|
| 5890 |
+
"learning_rate": 7.478000000000001e-06,
|
| 5891 |
+
"loss": 2.3815,
|
| 5892 |
+
"step": 16350
|
| 5893 |
+
},
|
| 5894 |
+
{
|
| 5895 |
+
"epoch": 0.3275,
|
| 5896 |
+
"grad_norm": 0.5597449596999192,
|
| 5897 |
+
"learning_rate": 7.4724444444444455e-06,
|
| 5898 |
+
"loss": 2.3732,
|
| 5899 |
+
"step": 16375
|
| 5900 |
+
},
|
| 5901 |
+
{
|
| 5902 |
+
"epoch": 0.328,
|
| 5903 |
+
"grad_norm": 0.5818958282150155,
|
| 5904 |
+
"learning_rate": 7.466888888888889e-06,
|
| 5905 |
+
"loss": 2.3793,
|
| 5906 |
+
"step": 16400
|
| 5907 |
+
},
|
| 5908 |
+
{
|
| 5909 |
+
"epoch": 0.328,
|
| 5910 |
+
"eval_loss": 2.3949294090270996,
|
| 5911 |
+
"eval_runtime": 31.7738,
|
| 5912 |
+
"eval_samples_per_second": 3.21,
|
| 5913 |
+
"eval_steps_per_second": 1.605,
|
| 5914 |
+
"step": 16400
|
| 5915 |
+
},
|
| 5916 |
+
{
|
| 5917 |
+
"epoch": 0.3285,
|
| 5918 |
+
"grad_norm": 0.5662000485734395,
|
| 5919 |
+
"learning_rate": 7.4613333333333334e-06,
|
| 5920 |
+
"loss": 2.3812,
|
| 5921 |
+
"step": 16425
|
| 5922 |
+
},
|
| 5923 |
+
{
|
| 5924 |
+
"epoch": 0.329,
|
| 5925 |
+
"grad_norm": 0.5563577533028059,
|
| 5926 |
+
"learning_rate": 7.455777777777779e-06,
|
| 5927 |
+
"loss": 2.3761,
|
| 5928 |
+
"step": 16450
|
| 5929 |
+
},
|
| 5930 |
+
{
|
| 5931 |
+
"epoch": 0.3295,
|
| 5932 |
+
"grad_norm": 0.5687992956190129,
|
| 5933 |
+
"learning_rate": 7.450222222222223e-06,
|
| 5934 |
+
"loss": 2.381,
|
| 5935 |
+
"step": 16475
|
| 5936 |
+
},
|
| 5937 |
+
{
|
| 5938 |
+
"epoch": 0.33,
|
| 5939 |
+
"grad_norm": 0.5487444076942639,
|
| 5940 |
+
"learning_rate": 7.4446666666666675e-06,
|
| 5941 |
+
"loss": 2.3883,
|
| 5942 |
+
"step": 16500
|
| 5943 |
+
},
|
| 5944 |
+
{
|
| 5945 |
+
"epoch": 0.33,
|
| 5946 |
+
"eval_loss": 2.395174026489258,
|
| 5947 |
+
"eval_runtime": 31.7762,
|
| 5948 |
+
"eval_samples_per_second": 3.21,
|
| 5949 |
+
"eval_steps_per_second": 1.605,
|
| 5950 |
+
"step": 16500
|
| 5951 |
+
},
|
| 5952 |
+
{
|
| 5953 |
+
"epoch": 0.3305,
|
| 5954 |
+
"grad_norm": 0.5469101598299175,
|
| 5955 |
+
"learning_rate": 7.439111111111111e-06,
|
| 5956 |
+
"loss": 2.3766,
|
| 5957 |
+
"step": 16525
|
| 5958 |
+
},
|
| 5959 |
+
{
|
| 5960 |
+
"epoch": 0.331,
|
| 5961 |
+
"grad_norm": 0.5567200858341991,
|
| 5962 |
+
"learning_rate": 7.433555555555556e-06,
|
| 5963 |
+
"loss": 2.3939,
|
| 5964 |
+
"step": 16550
|
| 5965 |
+
},
|
| 5966 |
+
{
|
| 5967 |
+
"epoch": 0.3315,
|
| 5968 |
+
"grad_norm": 0.600536691861987,
|
| 5969 |
+
"learning_rate": 7.428000000000001e-06,
|
| 5970 |
+
"loss": 2.3822,
|
| 5971 |
+
"step": 16575
|
| 5972 |
+
},
|
| 5973 |
+
{
|
| 5974 |
+
"epoch": 0.332,
|
| 5975 |
+
"grad_norm": 0.5505048207350117,
|
| 5976 |
+
"learning_rate": 7.422444444444445e-06,
|
| 5977 |
+
"loss": 2.378,
|
| 5978 |
+
"step": 16600
|
| 5979 |
+
},
|
| 5980 |
+
{
|
| 5981 |
+
"epoch": 0.332,
|
| 5982 |
+
"eval_loss": 2.39481520652771,
|
| 5983 |
+
"eval_runtime": 31.8394,
|
| 5984 |
+
"eval_samples_per_second": 3.204,
|
| 5985 |
+
"eval_steps_per_second": 1.602,
|
| 5986 |
+
"step": 16600
|
| 5987 |
+
},
|
| 5988 |
+
{
|
| 5989 |
+
"epoch": 0.3325,
|
| 5990 |
+
"grad_norm": 0.5492676702406505,
|
| 5991 |
+
"learning_rate": 7.416888888888889e-06,
|
| 5992 |
+
"loss": 2.3769,
|
| 5993 |
+
"step": 16625
|
| 5994 |
+
},
|
| 5995 |
+
{
|
| 5996 |
+
"epoch": 0.333,
|
| 5997 |
+
"grad_norm": 0.5492443037384863,
|
| 5998 |
+
"learning_rate": 7.411333333333334e-06,
|
| 5999 |
+
"loss": 2.3701,
|
| 6000 |
+
"step": 16650
|
| 6001 |
+
},
|
| 6002 |
+
{
|
| 6003 |
+
"epoch": 0.3335,
|
| 6004 |
+
"grad_norm": 0.5857568383624908,
|
| 6005 |
+
"learning_rate": 7.405777777777778e-06,
|
| 6006 |
+
"loss": 2.381,
|
| 6007 |
+
"step": 16675
|
| 6008 |
+
},
|
| 6009 |
+
{
|
| 6010 |
+
"epoch": 0.334,
|
| 6011 |
+
"grad_norm": 0.5647204860919086,
|
| 6012 |
+
"learning_rate": 7.400222222222223e-06,
|
| 6013 |
+
"loss": 2.3819,
|
| 6014 |
+
"step": 16700
|
| 6015 |
+
},
|
| 6016 |
+
{
|
| 6017 |
+
"epoch": 0.334,
|
| 6018 |
+
"eval_loss": 2.394426107406616,
|
| 6019 |
+
"eval_runtime": 31.892,
|
| 6020 |
+
"eval_samples_per_second": 3.198,
|
| 6021 |
+
"eval_steps_per_second": 1.599,
|
| 6022 |
+
"step": 16700
|
| 6023 |
+
},
|
| 6024 |
+
{
|
| 6025 |
+
"epoch": 0.3345,
|
| 6026 |
+
"grad_norm": 0.5730702201176824,
|
| 6027 |
+
"learning_rate": 7.394666666666668e-06,
|
| 6028 |
+
"loss": 2.3857,
|
| 6029 |
+
"step": 16725
|
| 6030 |
+
},
|
| 6031 |
+
{
|
| 6032 |
+
"epoch": 0.335,
|
| 6033 |
+
"grad_norm": 0.5521969424083262,
|
| 6034 |
+
"learning_rate": 7.3891111111111115e-06,
|
| 6035 |
+
"loss": 2.363,
|
| 6036 |
+
"step": 16750
|
| 6037 |
+
},
|
| 6038 |
+
{
|
| 6039 |
+
"epoch": 0.3355,
|
| 6040 |
+
"grad_norm": 0.6057695700506919,
|
| 6041 |
+
"learning_rate": 7.383555555555556e-06,
|
| 6042 |
+
"loss": 2.3848,
|
| 6043 |
+
"step": 16775
|
| 6044 |
+
},
|
| 6045 |
+
{
|
| 6046 |
+
"epoch": 0.336,
|
| 6047 |
+
"grad_norm": 0.5749986280132275,
|
| 6048 |
+
"learning_rate": 7.378e-06,
|
| 6049 |
+
"loss": 2.389,
|
| 6050 |
+
"step": 16800
|
| 6051 |
+
},
|
| 6052 |
+
{
|
| 6053 |
+
"epoch": 0.336,
|
| 6054 |
+
"eval_loss": 2.3945508003234863,
|
| 6055 |
+
"eval_runtime": 31.7463,
|
| 6056 |
+
"eval_samples_per_second": 3.213,
|
| 6057 |
+
"eval_steps_per_second": 1.606,
|
| 6058 |
+
"step": 16800
|
| 6059 |
+
},
|
| 6060 |
+
{
|
| 6061 |
+
"epoch": 0.3365,
|
| 6062 |
+
"grad_norm": 0.5947076066210849,
|
| 6063 |
+
"learning_rate": 7.372444444444446e-06,
|
| 6064 |
+
"loss": 2.3865,
|
| 6065 |
+
"step": 16825
|
| 6066 |
+
},
|
| 6067 |
+
{
|
| 6068 |
+
"epoch": 0.337,
|
| 6069 |
+
"grad_norm": 0.564221658006085,
|
| 6070 |
+
"learning_rate": 7.366888888888889e-06,
|
| 6071 |
+
"loss": 2.3696,
|
| 6072 |
+
"step": 16850
|
| 6073 |
+
},
|
| 6074 |
+
{
|
| 6075 |
+
"epoch": 0.3375,
|
| 6076 |
+
"grad_norm": 0.5702041520098122,
|
| 6077 |
+
"learning_rate": 7.3613333333333336e-06,
|
| 6078 |
+
"loss": 2.3872,
|
| 6079 |
+
"step": 16875
|
| 6080 |
+
},
|
| 6081 |
+
{
|
| 6082 |
+
"epoch": 0.338,
|
| 6083 |
+
"grad_norm": 0.5538661614565709,
|
| 6084 |
+
"learning_rate": 7.355777777777778e-06,
|
| 6085 |
+
"loss": 2.3828,
|
| 6086 |
+
"step": 16900
|
| 6087 |
+
},
|
| 6088 |
+
{
|
| 6089 |
+
"epoch": 0.338,
|
| 6090 |
+
"eval_loss": 2.3942644596099854,
|
| 6091 |
+
"eval_runtime": 31.8144,
|
| 6092 |
+
"eval_samples_per_second": 3.206,
|
| 6093 |
+
"eval_steps_per_second": 1.603,
|
| 6094 |
+
"step": 16900
|
| 6095 |
+
},
|
| 6096 |
+
{
|
| 6097 |
+
"epoch": 0.3385,
|
| 6098 |
+
"grad_norm": 0.5614412730199092,
|
| 6099 |
+
"learning_rate": 7.350222222222223e-06,
|
| 6100 |
+
"loss": 2.3898,
|
| 6101 |
+
"step": 16925
|
| 6102 |
+
},
|
| 6103 |
+
{
|
| 6104 |
+
"epoch": 0.339,
|
| 6105 |
+
"grad_norm": 0.5656638849693418,
|
| 6106 |
+
"learning_rate": 7.344666666666668e-06,
|
| 6107 |
+
"loss": 2.3639,
|
| 6108 |
+
"step": 16950
|
| 6109 |
+
},
|
| 6110 |
+
{
|
| 6111 |
+
"epoch": 0.3395,
|
| 6112 |
+
"grad_norm": 0.5587793192894792,
|
| 6113 |
+
"learning_rate": 7.339111111111111e-06,
|
| 6114 |
+
"loss": 2.3761,
|
| 6115 |
+
"step": 16975
|
| 6116 |
+
},
|
| 6117 |
+
{
|
| 6118 |
+
"epoch": 0.34,
|
| 6119 |
+
"grad_norm": 0.5537041511919,
|
| 6120 |
+
"learning_rate": 7.3335555555555556e-06,
|
| 6121 |
+
"loss": 2.3785,
|
| 6122 |
+
"step": 17000
|
| 6123 |
+
},
|
| 6124 |
+
{
|
| 6125 |
+
"epoch": 0.34,
|
| 6126 |
+
"eval_loss": 2.394216775894165,
|
| 6127 |
+
"eval_runtime": 31.7287,
|
| 6128 |
+
"eval_samples_per_second": 3.215,
|
| 6129 |
+
"eval_steps_per_second": 1.607,
|
| 6130 |
+
"step": 17000
|
| 6131 |
}
|
| 6132 |
],
|
| 6133 |
"logging_steps": 25,
|
|
|
|
| 6147 |
"attributes": {}
|
| 6148 |
}
|
| 6149 |
},
|
| 6150 |
+
"total_flos": 5.4114439071562465e+19,
|
| 6151 |
"train_batch_size": 1,
|
| 6152 |
"trial_name": null,
|
| 6153 |
"trial_params": null
|