Training checkpoint at step 12000
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,
|
|
@@ -3968,6 +3968,366 @@
|
|
| 3968 |
"eval_samples_per_second": 3.217,
|
| 3969 |
"eval_steps_per_second": 1.609,
|
| 3970 |
"step": 11000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3971 |
}
|
| 3972 |
],
|
| 3973 |
"logging_steps": 25,
|
|
@@ -3987,7 +4347,7 @@
|
|
| 3987 |
"attributes": {}
|
| 3988 |
}
|
| 3989 |
},
|
| 3990 |
-
"total_flos": 3.
|
| 3991 |
"train_batch_size": 1,
|
| 3992 |
"trial_name": null,
|
| 3993 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 12000,
|
| 3 |
+
"best_metric": 2.4031572341918945,
|
| 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-12000",
|
| 5 |
+
"epoch": 0.24,
|
| 6 |
"eval_steps": 100,
|
| 7 |
+
"global_step": 12000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 3968 |
"eval_samples_per_second": 3.217,
|
| 3969 |
"eval_steps_per_second": 1.609,
|
| 3970 |
"step": 11000
|
| 3971 |
+
},
|
| 3972 |
+
{
|
| 3973 |
+
"epoch": 0.2205,
|
| 3974 |
+
"grad_norm": 0.619520274382602,
|
| 3975 |
+
"learning_rate": 8.661333333333335e-06,
|
| 3976 |
+
"loss": 2.3868,
|
| 3977 |
+
"step": 11025
|
| 3978 |
+
},
|
| 3979 |
+
{
|
| 3980 |
+
"epoch": 0.221,
|
| 3981 |
+
"grad_norm": 0.5973378822756289,
|
| 3982 |
+
"learning_rate": 8.655777777777778e-06,
|
| 3983 |
+
"loss": 2.398,
|
| 3984 |
+
"step": 11050
|
| 3985 |
+
},
|
| 3986 |
+
{
|
| 3987 |
+
"epoch": 0.2215,
|
| 3988 |
+
"grad_norm": 0.6143187669490118,
|
| 3989 |
+
"learning_rate": 8.650222222222223e-06,
|
| 3990 |
+
"loss": 2.387,
|
| 3991 |
+
"step": 11075
|
| 3992 |
+
},
|
| 3993 |
+
{
|
| 3994 |
+
"epoch": 0.222,
|
| 3995 |
+
"grad_norm": 0.5804040103557917,
|
| 3996 |
+
"learning_rate": 8.644666666666669e-06,
|
| 3997 |
+
"loss": 2.3951,
|
| 3998 |
+
"step": 11100
|
| 3999 |
+
},
|
| 4000 |
+
{
|
| 4001 |
+
"epoch": 0.222,
|
| 4002 |
+
"eval_loss": 2.4050545692443848,
|
| 4003 |
+
"eval_runtime": 31.7713,
|
| 4004 |
+
"eval_samples_per_second": 3.21,
|
| 4005 |
+
"eval_steps_per_second": 1.605,
|
| 4006 |
+
"step": 11100
|
| 4007 |
+
},
|
| 4008 |
+
{
|
| 4009 |
+
"epoch": 0.2225,
|
| 4010 |
+
"grad_norm": 0.5833158956225722,
|
| 4011 |
+
"learning_rate": 8.639111111111112e-06,
|
| 4012 |
+
"loss": 2.3854,
|
| 4013 |
+
"step": 11125
|
| 4014 |
+
},
|
| 4015 |
+
{
|
| 4016 |
+
"epoch": 0.223,
|
| 4017 |
+
"grad_norm": 0.5741811771851818,
|
| 4018 |
+
"learning_rate": 8.633555555555556e-06,
|
| 4019 |
+
"loss": 2.3866,
|
| 4020 |
+
"step": 11150
|
| 4021 |
+
},
|
| 4022 |
+
{
|
| 4023 |
+
"epoch": 0.2235,
|
| 4024 |
+
"grad_norm": 0.5856955103294486,
|
| 4025 |
+
"learning_rate": 8.628000000000001e-06,
|
| 4026 |
+
"loss": 2.4058,
|
| 4027 |
+
"step": 11175
|
| 4028 |
+
},
|
| 4029 |
+
{
|
| 4030 |
+
"epoch": 0.224,
|
| 4031 |
+
"grad_norm": 0.5685596699989746,
|
| 4032 |
+
"learning_rate": 8.622444444444446e-06,
|
| 4033 |
+
"loss": 2.3953,
|
| 4034 |
+
"step": 11200
|
| 4035 |
+
},
|
| 4036 |
+
{
|
| 4037 |
+
"epoch": 0.224,
|
| 4038 |
+
"eval_loss": 2.4051928520202637,
|
| 4039 |
+
"eval_runtime": 35.481,
|
| 4040 |
+
"eval_samples_per_second": 2.875,
|
| 4041 |
+
"eval_steps_per_second": 1.437,
|
| 4042 |
+
"step": 11200
|
| 4043 |
+
},
|
| 4044 |
+
{
|
| 4045 |
+
"epoch": 0.2245,
|
| 4046 |
+
"grad_norm": 0.5854297741723825,
|
| 4047 |
+
"learning_rate": 8.61688888888889e-06,
|
| 4048 |
+
"loss": 2.3977,
|
| 4049 |
+
"step": 11225
|
| 4050 |
+
},
|
| 4051 |
+
{
|
| 4052 |
+
"epoch": 0.225,
|
| 4053 |
+
"grad_norm": 0.582929503102295,
|
| 4054 |
+
"learning_rate": 8.611333333333333e-06,
|
| 4055 |
+
"loss": 2.3948,
|
| 4056 |
+
"step": 11250
|
| 4057 |
+
},
|
| 4058 |
+
{
|
| 4059 |
+
"epoch": 0.2255,
|
| 4060 |
+
"grad_norm": 0.5839207937169353,
|
| 4061 |
+
"learning_rate": 8.605777777777779e-06,
|
| 4062 |
+
"loss": 2.4104,
|
| 4063 |
+
"step": 11275
|
| 4064 |
+
},
|
| 4065 |
+
{
|
| 4066 |
+
"epoch": 0.226,
|
| 4067 |
+
"grad_norm": 0.5568849917729087,
|
| 4068 |
+
"learning_rate": 8.600222222222224e-06,
|
| 4069 |
+
"loss": 2.4011,
|
| 4070 |
+
"step": 11300
|
| 4071 |
+
},
|
| 4072 |
+
{
|
| 4073 |
+
"epoch": 0.226,
|
| 4074 |
+
"eval_loss": 2.404717445373535,
|
| 4075 |
+
"eval_runtime": 31.9835,
|
| 4076 |
+
"eval_samples_per_second": 3.189,
|
| 4077 |
+
"eval_steps_per_second": 1.595,
|
| 4078 |
+
"step": 11300
|
| 4079 |
+
},
|
| 4080 |
+
{
|
| 4081 |
+
"epoch": 0.2265,
|
| 4082 |
+
"grad_norm": 0.5549969270675909,
|
| 4083 |
+
"learning_rate": 8.594666666666668e-06,
|
| 4084 |
+
"loss": 2.3965,
|
| 4085 |
+
"step": 11325
|
| 4086 |
+
},
|
| 4087 |
+
{
|
| 4088 |
+
"epoch": 0.227,
|
| 4089 |
+
"grad_norm": 0.5606539732290856,
|
| 4090 |
+
"learning_rate": 8.589111111111111e-06,
|
| 4091 |
+
"loss": 2.3921,
|
| 4092 |
+
"step": 11350
|
| 4093 |
+
},
|
| 4094 |
+
{
|
| 4095 |
+
"epoch": 0.2275,
|
| 4096 |
+
"grad_norm": 0.5626929771754517,
|
| 4097 |
+
"learning_rate": 8.583555555555556e-06,
|
| 4098 |
+
"loss": 2.3912,
|
| 4099 |
+
"step": 11375
|
| 4100 |
+
},
|
| 4101 |
+
{
|
| 4102 |
+
"epoch": 0.228,
|
| 4103 |
+
"grad_norm": 0.5731631708828652,
|
| 4104 |
+
"learning_rate": 8.578000000000002e-06,
|
| 4105 |
+
"loss": 2.3926,
|
| 4106 |
+
"step": 11400
|
| 4107 |
+
},
|
| 4108 |
+
{
|
| 4109 |
+
"epoch": 0.228,
|
| 4110 |
+
"eval_loss": 2.4047322273254395,
|
| 4111 |
+
"eval_runtime": 31.8245,
|
| 4112 |
+
"eval_samples_per_second": 3.205,
|
| 4113 |
+
"eval_steps_per_second": 1.603,
|
| 4114 |
+
"step": 11400
|
| 4115 |
+
},
|
| 4116 |
+
{
|
| 4117 |
+
"epoch": 0.2285,
|
| 4118 |
+
"grad_norm": 0.5661654100374769,
|
| 4119 |
+
"learning_rate": 8.572444444444445e-06,
|
| 4120 |
+
"loss": 2.3951,
|
| 4121 |
+
"step": 11425
|
| 4122 |
+
},
|
| 4123 |
+
{
|
| 4124 |
+
"epoch": 0.229,
|
| 4125 |
+
"grad_norm": 0.5602181256620924,
|
| 4126 |
+
"learning_rate": 8.56688888888889e-06,
|
| 4127 |
+
"loss": 2.3812,
|
| 4128 |
+
"step": 11450
|
| 4129 |
+
},
|
| 4130 |
+
{
|
| 4131 |
+
"epoch": 0.2295,
|
| 4132 |
+
"grad_norm": 0.5950733473289397,
|
| 4133 |
+
"learning_rate": 8.561333333333334e-06,
|
| 4134 |
+
"loss": 2.3963,
|
| 4135 |
+
"step": 11475
|
| 4136 |
+
},
|
| 4137 |
+
{
|
| 4138 |
+
"epoch": 0.23,
|
| 4139 |
+
"grad_norm": 0.5733938863696743,
|
| 4140 |
+
"learning_rate": 8.55577777777778e-06,
|
| 4141 |
+
"loss": 2.3932,
|
| 4142 |
+
"step": 11500
|
| 4143 |
+
},
|
| 4144 |
+
{
|
| 4145 |
+
"epoch": 0.23,
|
| 4146 |
+
"eval_loss": 2.403830051422119,
|
| 4147 |
+
"eval_runtime": 31.7862,
|
| 4148 |
+
"eval_samples_per_second": 3.209,
|
| 4149 |
+
"eval_steps_per_second": 1.604,
|
| 4150 |
+
"step": 11500
|
| 4151 |
+
},
|
| 4152 |
+
{
|
| 4153 |
+
"epoch": 0.2305,
|
| 4154 |
+
"grad_norm": 0.5702512759518216,
|
| 4155 |
+
"learning_rate": 8.550222222222223e-06,
|
| 4156 |
+
"loss": 2.3824,
|
| 4157 |
+
"step": 11525
|
| 4158 |
+
},
|
| 4159 |
+
{
|
| 4160 |
+
"epoch": 0.231,
|
| 4161 |
+
"grad_norm": 0.5749933738625221,
|
| 4162 |
+
"learning_rate": 8.544666666666668e-06,
|
| 4163 |
+
"loss": 2.3674,
|
| 4164 |
+
"step": 11550
|
| 4165 |
+
},
|
| 4166 |
+
{
|
| 4167 |
+
"epoch": 0.2315,
|
| 4168 |
+
"grad_norm": 0.563814842108926,
|
| 4169 |
+
"learning_rate": 8.539111111111112e-06,
|
| 4170 |
+
"loss": 2.3866,
|
| 4171 |
+
"step": 11575
|
| 4172 |
+
},
|
| 4173 |
+
{
|
| 4174 |
+
"epoch": 0.232,
|
| 4175 |
+
"grad_norm": 0.601764608458657,
|
| 4176 |
+
"learning_rate": 8.533555555555557e-06,
|
| 4177 |
+
"loss": 2.3949,
|
| 4178 |
+
"step": 11600
|
| 4179 |
+
},
|
| 4180 |
+
{
|
| 4181 |
+
"epoch": 0.232,
|
| 4182 |
+
"eval_loss": 2.4035561084747314,
|
| 4183 |
+
"eval_runtime": 31.7077,
|
| 4184 |
+
"eval_samples_per_second": 3.217,
|
| 4185 |
+
"eval_steps_per_second": 1.608,
|
| 4186 |
+
"step": 11600
|
| 4187 |
+
},
|
| 4188 |
+
{
|
| 4189 |
+
"epoch": 0.2325,
|
| 4190 |
+
"grad_norm": 0.5674229084100237,
|
| 4191 |
+
"learning_rate": 8.528e-06,
|
| 4192 |
+
"loss": 2.3782,
|
| 4193 |
+
"step": 11625
|
| 4194 |
+
},
|
| 4195 |
+
{
|
| 4196 |
+
"epoch": 0.233,
|
| 4197 |
+
"grad_norm": 0.5660025767055805,
|
| 4198 |
+
"learning_rate": 8.522444444444446e-06,
|
| 4199 |
+
"loss": 2.3811,
|
| 4200 |
+
"step": 11650
|
| 4201 |
+
},
|
| 4202 |
+
{
|
| 4203 |
+
"epoch": 0.2335,
|
| 4204 |
+
"grad_norm": 0.5776196117388842,
|
| 4205 |
+
"learning_rate": 8.51688888888889e-06,
|
| 4206 |
+
"loss": 2.3964,
|
| 4207 |
+
"step": 11675
|
| 4208 |
+
},
|
| 4209 |
+
{
|
| 4210 |
+
"epoch": 0.234,
|
| 4211 |
+
"grad_norm": 0.5815076886720436,
|
| 4212 |
+
"learning_rate": 8.511333333333334e-06,
|
| 4213 |
+
"loss": 2.3907,
|
| 4214 |
+
"step": 11700
|
| 4215 |
+
},
|
| 4216 |
+
{
|
| 4217 |
+
"epoch": 0.234,
|
| 4218 |
+
"eval_loss": 2.4035725593566895,
|
| 4219 |
+
"eval_runtime": 31.7541,
|
| 4220 |
+
"eval_samples_per_second": 3.212,
|
| 4221 |
+
"eval_steps_per_second": 1.606,
|
| 4222 |
+
"step": 11700
|
| 4223 |
+
},
|
| 4224 |
+
{
|
| 4225 |
+
"epoch": 0.2345,
|
| 4226 |
+
"grad_norm": 0.5810635532925048,
|
| 4227 |
+
"learning_rate": 8.505777777777778e-06,
|
| 4228 |
+
"loss": 2.3921,
|
| 4229 |
+
"step": 11725
|
| 4230 |
+
},
|
| 4231 |
+
{
|
| 4232 |
+
"epoch": 0.235,
|
| 4233 |
+
"grad_norm": 0.5635380257098753,
|
| 4234 |
+
"learning_rate": 8.500222222222223e-06,
|
| 4235 |
+
"loss": 2.4062,
|
| 4236 |
+
"step": 11750
|
| 4237 |
+
},
|
| 4238 |
+
{
|
| 4239 |
+
"epoch": 0.2355,
|
| 4240 |
+
"grad_norm": 0.5985004911332629,
|
| 4241 |
+
"learning_rate": 8.494666666666668e-06,
|
| 4242 |
+
"loss": 2.3853,
|
| 4243 |
+
"step": 11775
|
| 4244 |
+
},
|
| 4245 |
+
{
|
| 4246 |
+
"epoch": 0.236,
|
| 4247 |
+
"grad_norm": 0.580078413647693,
|
| 4248 |
+
"learning_rate": 8.489111111111112e-06,
|
| 4249 |
+
"loss": 2.3826,
|
| 4250 |
+
"step": 11800
|
| 4251 |
+
},
|
| 4252 |
+
{
|
| 4253 |
+
"epoch": 0.236,
|
| 4254 |
+
"eval_loss": 2.403505325317383,
|
| 4255 |
+
"eval_runtime": 31.7265,
|
| 4256 |
+
"eval_samples_per_second": 3.215,
|
| 4257 |
+
"eval_steps_per_second": 1.607,
|
| 4258 |
+
"step": 11800
|
| 4259 |
+
},
|
| 4260 |
+
{
|
| 4261 |
+
"epoch": 0.2365,
|
| 4262 |
+
"grad_norm": 0.5560334145179444,
|
| 4263 |
+
"learning_rate": 8.483555555555556e-06,
|
| 4264 |
+
"loss": 2.3829,
|
| 4265 |
+
"step": 11825
|
| 4266 |
+
},
|
| 4267 |
+
{
|
| 4268 |
+
"epoch": 0.237,
|
| 4269 |
+
"grad_norm": 0.5870934042209253,
|
| 4270 |
+
"learning_rate": 8.478e-06,
|
| 4271 |
+
"loss": 2.374,
|
| 4272 |
+
"step": 11850
|
| 4273 |
+
},
|
| 4274 |
+
{
|
| 4275 |
+
"epoch": 0.2375,
|
| 4276 |
+
"grad_norm": 0.5745342448568999,
|
| 4277 |
+
"learning_rate": 8.472444444444446e-06,
|
| 4278 |
+
"loss": 2.3797,
|
| 4279 |
+
"step": 11875
|
| 4280 |
+
},
|
| 4281 |
+
{
|
| 4282 |
+
"epoch": 0.238,
|
| 4283 |
+
"grad_norm": 0.5676573173578097,
|
| 4284 |
+
"learning_rate": 8.46688888888889e-06,
|
| 4285 |
+
"loss": 2.3867,
|
| 4286 |
+
"step": 11900
|
| 4287 |
+
},
|
| 4288 |
+
{
|
| 4289 |
+
"epoch": 0.238,
|
| 4290 |
+
"eval_loss": 2.403400421142578,
|
| 4291 |
+
"eval_runtime": 31.8105,
|
| 4292 |
+
"eval_samples_per_second": 3.206,
|
| 4293 |
+
"eval_steps_per_second": 1.603,
|
| 4294 |
+
"step": 11900
|
| 4295 |
+
},
|
| 4296 |
+
{
|
| 4297 |
+
"epoch": 0.2385,
|
| 4298 |
+
"grad_norm": 0.5701256243606029,
|
| 4299 |
+
"learning_rate": 8.461333333333333e-06,
|
| 4300 |
+
"loss": 2.3832,
|
| 4301 |
+
"step": 11925
|
| 4302 |
+
},
|
| 4303 |
+
{
|
| 4304 |
+
"epoch": 0.239,
|
| 4305 |
+
"grad_norm": 0.5839965205220576,
|
| 4306 |
+
"learning_rate": 8.455777777777778e-06,
|
| 4307 |
+
"loss": 2.3939,
|
| 4308 |
+
"step": 11950
|
| 4309 |
+
},
|
| 4310 |
+
{
|
| 4311 |
+
"epoch": 0.2395,
|
| 4312 |
+
"grad_norm": 0.581600775004578,
|
| 4313 |
+
"learning_rate": 8.450222222222224e-06,
|
| 4314 |
+
"loss": 2.382,
|
| 4315 |
+
"step": 11975
|
| 4316 |
+
},
|
| 4317 |
+
{
|
| 4318 |
+
"epoch": 0.24,
|
| 4319 |
+
"grad_norm": 0.5945113931788275,
|
| 4320 |
+
"learning_rate": 8.444666666666667e-06,
|
| 4321 |
+
"loss": 2.3947,
|
| 4322 |
+
"step": 12000
|
| 4323 |
+
},
|
| 4324 |
+
{
|
| 4325 |
+
"epoch": 0.24,
|
| 4326 |
+
"eval_loss": 2.4031572341918945,
|
| 4327 |
+
"eval_runtime": 31.7154,
|
| 4328 |
+
"eval_samples_per_second": 3.216,
|
| 4329 |
+
"eval_steps_per_second": 1.608,
|
| 4330 |
+
"step": 12000
|
| 4331 |
}
|
| 4332 |
],
|
| 4333 |
"logging_steps": 25,
|
|
|
|
| 4347 |
"attributes": {}
|
| 4348 |
}
|
| 4349 |
},
|
| 4350 |
+
"total_flos": 3.819842757992645e+19,
|
| 4351 |
"train_batch_size": 1,
|
| 4352 |
"trial_name": null,
|
| 4353 |
"trial_params": null
|