Training in progress, step 744, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 167832240
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2b3fa76795aac7ce31b919a79ce0da184e31cd62feb3a8b1a2eb838957202798
|
| 3 |
size 167832240
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 85723732
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e6b0f55b2c27d4899beec2319887ce2031f6991a2c282f12dc8403a58f0d4431
|
| 3 |
size 85723732
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:390f0dbac2add99b9de0bdfc18171c87964e9eac7b5a9d79ded1784da24c00e8
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e22f1029354fc2004de93e539443b580b540e5122ae66043f3e66f977d6dd066
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 1.267533302307129,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-700",
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 100,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -4971,6 +4971,314 @@
|
|
| 4971 |
"eval_samples_per_second": 6.272,
|
| 4972 |
"eval_steps_per_second": 1.568,
|
| 4973 |
"step": 700
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4974 |
}
|
| 4975 |
],
|
| 4976 |
"logging_steps": 1,
|
|
@@ -4994,12 +5302,12 @@
|
|
| 4994 |
"should_evaluate": false,
|
| 4995 |
"should_log": false,
|
| 4996 |
"should_save": true,
|
| 4997 |
-
"should_training_stop":
|
| 4998 |
},
|
| 4999 |
"attributes": {}
|
| 5000 |
}
|
| 5001 |
},
|
| 5002 |
-
"total_flos": 2.
|
| 5003 |
"train_batch_size": 4,
|
| 5004 |
"trial_name": null,
|
| 5005 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 1.267533302307129,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-700",
|
| 4 |
+
"epoch": 0.0776081259045806,
|
| 5 |
"eval_steps": 100,
|
| 6 |
+
"global_step": 744,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4971 |
"eval_samples_per_second": 6.272,
|
| 4972 |
"eval_steps_per_second": 1.568,
|
| 4973 |
"step": 700
|
| 4974 |
+
},
|
| 4975 |
+
{
|
| 4976 |
+
"epoch": 0.07312271002568684,
|
| 4977 |
+
"grad_norm": 0.5641259551048279,
|
| 4978 |
+
"learning_rate": 1.688839258059971e-06,
|
| 4979 |
+
"loss": 1.3232,
|
| 4980 |
+
"step": 701
|
| 4981 |
+
},
|
| 4982 |
+
{
|
| 4983 |
+
"epoch": 0.0732270220228704,
|
| 4984 |
+
"grad_norm": 0.5656439661979675,
|
| 4985 |
+
"learning_rate": 1.6114111754051974e-06,
|
| 4986 |
+
"loss": 1.2328,
|
| 4987 |
+
"step": 702
|
| 4988 |
+
},
|
| 4989 |
+
{
|
| 4990 |
+
"epoch": 0.07333133402005398,
|
| 4991 |
+
"grad_norm": 0.5426310896873474,
|
| 4992 |
+
"learning_rate": 1.5357854948895634e-06,
|
| 4993 |
+
"loss": 1.251,
|
| 4994 |
+
"step": 703
|
| 4995 |
+
},
|
| 4996 |
+
{
|
| 4997 |
+
"epoch": 0.07343564601723755,
|
| 4998 |
+
"grad_norm": 0.5175634622573853,
|
| 4999 |
+
"learning_rate": 1.4619636019164606e-06,
|
| 5000 |
+
"loss": 1.1482,
|
| 5001 |
+
"step": 704
|
| 5002 |
+
},
|
| 5003 |
+
{
|
| 5004 |
+
"epoch": 0.07353995801442113,
|
| 5005 |
+
"grad_norm": 0.5635023713111877,
|
| 5006 |
+
"learning_rate": 1.3899468488453583e-06,
|
| 5007 |
+
"loss": 1.2063,
|
| 5008 |
+
"step": 705
|
| 5009 |
+
},
|
| 5010 |
+
{
|
| 5011 |
+
"epoch": 0.07364427001160471,
|
| 5012 |
+
"grad_norm": 0.518434464931488,
|
| 5013 |
+
"learning_rate": 1.319736554966955e-06,
|
| 5014 |
+
"loss": 1.4171,
|
| 5015 |
+
"step": 706
|
| 5016 |
+
},
|
| 5017 |
+
{
|
| 5018 |
+
"epoch": 0.07374858200878828,
|
| 5019 |
+
"grad_norm": 0.5022286176681519,
|
| 5020 |
+
"learning_rate": 1.2513340064790102e-06,
|
| 5021 |
+
"loss": 1.1245,
|
| 5022 |
+
"step": 707
|
| 5023 |
+
},
|
| 5024 |
+
{
|
| 5025 |
+
"epoch": 0.07385289400597186,
|
| 5026 |
+
"grad_norm": 0.5017629265785217,
|
| 5027 |
+
"learning_rate": 1.1847404564628185e-06,
|
| 5028 |
+
"loss": 1.144,
|
| 5029 |
+
"step": 708
|
| 5030 |
+
},
|
| 5031 |
+
{
|
| 5032 |
+
"epoch": 0.07395720600315543,
|
| 5033 |
+
"grad_norm": 0.5079241991043091,
|
| 5034 |
+
"learning_rate": 1.1199571248602382e-06,
|
| 5035 |
+
"loss": 1.3892,
|
| 5036 |
+
"step": 709
|
| 5037 |
+
},
|
| 5038 |
+
{
|
| 5039 |
+
"epoch": 0.07406151800033901,
|
| 5040 |
+
"grad_norm": 0.558506190776825,
|
| 5041 |
+
"learning_rate": 1.0569851984513103e-06,
|
| 5042 |
+
"loss": 1.2448,
|
| 5043 |
+
"step": 710
|
| 5044 |
+
},
|
| 5045 |
+
{
|
| 5046 |
+
"epoch": 0.0741658299975226,
|
| 5047 |
+
"grad_norm": 0.5008535981178284,
|
| 5048 |
+
"learning_rate": 9.958258308325973e-07,
|
| 5049 |
+
"loss": 1.2267,
|
| 5050 |
+
"step": 711
|
| 5051 |
+
},
|
| 5052 |
+
{
|
| 5053 |
+
"epoch": 0.07427014199470616,
|
| 5054 |
+
"grad_norm": 0.53139728307724,
|
| 5055 |
+
"learning_rate": 9.364801423959235e-07,
|
| 5056 |
+
"loss": 1.2265,
|
| 5057 |
+
"step": 712
|
| 5058 |
+
},
|
| 5059 |
+
{
|
| 5060 |
+
"epoch": 0.07437445399188974,
|
| 5061 |
+
"grad_norm": 0.5449005961418152,
|
| 5062 |
+
"learning_rate": 8.78949220308023e-07,
|
| 5063 |
+
"loss": 1.3256,
|
| 5064 |
+
"step": 713
|
| 5065 |
+
},
|
| 5066 |
+
{
|
| 5067 |
+
"epoch": 0.07447876598907331,
|
| 5068 |
+
"grad_norm": 0.5834050178527832,
|
| 5069 |
+
"learning_rate": 8.232341184904457e-07,
|
| 5070 |
+
"loss": 1.2626,
|
| 5071 |
+
"step": 714
|
| 5072 |
+
},
|
| 5073 |
+
{
|
| 5074 |
+
"epoch": 0.0745830779862569,
|
| 5075 |
+
"grad_norm": 0.5395832657814026,
|
| 5076 |
+
"learning_rate": 7.693358576003617e-07,
|
| 5077 |
+
"loss": 1.2074,
|
| 5078 |
+
"step": 715
|
| 5079 |
+
},
|
| 5080 |
+
{
|
| 5081 |
+
"epoch": 0.07468738998344047,
|
| 5082 |
+
"grad_norm": 0.5254425406455994,
|
| 5083 |
+
"learning_rate": 7.172554250118535e-07,
|
| 5084 |
+
"loss": 1.2065,
|
| 5085 |
+
"step": 716
|
| 5086 |
+
},
|
| 5087 |
+
{
|
| 5088 |
+
"epoch": 0.07479170198062404,
|
| 5089 |
+
"grad_norm": 0.5222790837287903,
|
| 5090 |
+
"learning_rate": 6.66993774797775e-07,
|
| 5091 |
+
"loss": 1.1553,
|
| 5092 |
+
"step": 717
|
| 5093 |
+
},
|
| 5094 |
+
{
|
| 5095 |
+
"epoch": 0.07489601397780762,
|
| 5096 |
+
"grad_norm": 0.5531771183013916,
|
| 5097 |
+
"learning_rate": 6.185518277123214e-07,
|
| 5098 |
+
"loss": 1.1405,
|
| 5099 |
+
"step": 718
|
| 5100 |
+
},
|
| 5101 |
+
{
|
| 5102 |
+
"epoch": 0.0750003259749912,
|
| 5103 |
+
"grad_norm": 0.5848060250282288,
|
| 5104 |
+
"learning_rate": 5.719304711741535e-07,
|
| 5105 |
+
"loss": 1.283,
|
| 5106 |
+
"step": 719
|
| 5107 |
+
},
|
| 5108 |
+
{
|
| 5109 |
+
"epoch": 0.07510463797217477,
|
| 5110 |
+
"grad_norm": 0.6206434965133667,
|
| 5111 |
+
"learning_rate": 5.271305592501108e-07,
|
| 5112 |
+
"loss": 1.2983,
|
| 5113 |
+
"step": 720
|
| 5114 |
+
},
|
| 5115 |
+
{
|
| 5116 |
+
"epoch": 0.07520894996935835,
|
| 5117 |
+
"grad_norm": 0.5380674600601196,
|
| 5118 |
+
"learning_rate": 4.841529126396238e-07,
|
| 5119 |
+
"loss": 1.4001,
|
| 5120 |
+
"step": 721
|
| 5121 |
+
},
|
| 5122 |
+
{
|
| 5123 |
+
"epoch": 0.07531326196654192,
|
| 5124 |
+
"grad_norm": 0.5470457673072815,
|
| 5125 |
+
"learning_rate": 4.429983186596265e-07,
|
| 5126 |
+
"loss": 1.2379,
|
| 5127 |
+
"step": 722
|
| 5128 |
+
},
|
| 5129 |
+
{
|
| 5130 |
+
"epoch": 0.0754175739637255,
|
| 5131 |
+
"grad_norm": 0.598996102809906,
|
| 5132 |
+
"learning_rate": 4.036675312301452e-07,
|
| 5133 |
+
"loss": 1.2304,
|
| 5134 |
+
"step": 723
|
| 5135 |
+
},
|
| 5136 |
+
{
|
| 5137 |
+
"epoch": 0.07552188596090909,
|
| 5138 |
+
"grad_norm": 0.521056056022644,
|
| 5139 |
+
"learning_rate": 3.6616127086051e-07,
|
| 5140 |
+
"loss": 1.3609,
|
| 5141 |
+
"step": 724
|
| 5142 |
+
},
|
| 5143 |
+
{
|
| 5144 |
+
"epoch": 0.07562619795809265,
|
| 5145 |
+
"grad_norm": 0.5366652011871338,
|
| 5146 |
+
"learning_rate": 3.3048022463612047e-07,
|
| 5147 |
+
"loss": 1.2715,
|
| 5148 |
+
"step": 725
|
| 5149 |
+
},
|
| 5150 |
+
{
|
| 5151 |
+
"epoch": 0.07573050995527623,
|
| 5152 |
+
"grad_norm": 0.5772047638893127,
|
| 5153 |
+
"learning_rate": 2.9662504620588947e-07,
|
| 5154 |
+
"loss": 1.1976,
|
| 5155 |
+
"step": 726
|
| 5156 |
+
},
|
| 5157 |
+
{
|
| 5158 |
+
"epoch": 0.0758348219524598,
|
| 5159 |
+
"grad_norm": 0.547029435634613,
|
| 5160 |
+
"learning_rate": 2.6459635577026353e-07,
|
| 5161 |
+
"loss": 1.3886,
|
| 5162 |
+
"step": 727
|
| 5163 |
+
},
|
| 5164 |
+
{
|
| 5165 |
+
"epoch": 0.07593913394964338,
|
| 5166 |
+
"grad_norm": 0.5879443287849426,
|
| 5167 |
+
"learning_rate": 2.343947400698432e-07,
|
| 5168 |
+
"loss": 1.3782,
|
| 5169 |
+
"step": 728
|
| 5170 |
+
},
|
| 5171 |
+
{
|
| 5172 |
+
"epoch": 0.07604344594682697,
|
| 5173 |
+
"grad_norm": 0.5763773322105408,
|
| 5174 |
+
"learning_rate": 2.0602075237465823e-07,
|
| 5175 |
+
"loss": 1.2558,
|
| 5176 |
+
"step": 729
|
| 5177 |
+
},
|
| 5178 |
+
{
|
| 5179 |
+
"epoch": 0.07614775794401053,
|
| 5180 |
+
"grad_norm": 0.7472733855247498,
|
| 5181 |
+
"learning_rate": 1.7947491247399806e-07,
|
| 5182 |
+
"loss": 1.4806,
|
| 5183 |
+
"step": 730
|
| 5184 |
+
},
|
| 5185 |
+
{
|
| 5186 |
+
"epoch": 0.07625206994119411,
|
| 5187 |
+
"grad_norm": 0.54900723695755,
|
| 5188 |
+
"learning_rate": 1.5475770666694144e-07,
|
| 5189 |
+
"loss": 1.1513,
|
| 5190 |
+
"step": 731
|
| 5191 |
+
},
|
| 5192 |
+
{
|
| 5193 |
+
"epoch": 0.07635638193837768,
|
| 5194 |
+
"grad_norm": 0.5599141120910645,
|
| 5195 |
+
"learning_rate": 1.318695877533971e-07,
|
| 5196 |
+
"loss": 1.1123,
|
| 5197 |
+
"step": 732
|
| 5198 |
+
},
|
| 5199 |
+
{
|
| 5200 |
+
"epoch": 0.07646069393556126,
|
| 5201 |
+
"grad_norm": 0.5251076817512512,
|
| 5202 |
+
"learning_rate": 1.1081097502584348e-07,
|
| 5203 |
+
"loss": 1.1586,
|
| 5204 |
+
"step": 733
|
| 5205 |
+
},
|
| 5206 |
+
{
|
| 5207 |
+
"epoch": 0.07656500593274485,
|
| 5208 |
+
"grad_norm": 0.5731672644615173,
|
| 5209 |
+
"learning_rate": 9.158225426160183e-08,
|
| 5210 |
+
"loss": 1.1249,
|
| 5211 |
+
"step": 734
|
| 5212 |
+
},
|
| 5213 |
+
{
|
| 5214 |
+
"epoch": 0.07666931792992841,
|
| 5215 |
+
"grad_norm": 0.5583503842353821,
|
| 5216 |
+
"learning_rate": 7.418377771585273e-08,
|
| 5217 |
+
"loss": 1.184,
|
| 5218 |
+
"step": 735
|
| 5219 |
+
},
|
| 5220 |
+
{
|
| 5221 |
+
"epoch": 0.076773629927112,
|
| 5222 |
+
"grad_norm": 0.7562082409858704,
|
| 5223 |
+
"learning_rate": 5.86158641150969e-08,
|
| 5224 |
+
"loss": 1.3864,
|
| 5225 |
+
"step": 736
|
| 5226 |
+
},
|
| 5227 |
+
{
|
| 5228 |
+
"epoch": 0.07687794192429556,
|
| 5229 |
+
"grad_norm": 0.492931604385376,
|
| 5230 |
+
"learning_rate": 4.487879865133771e-08,
|
| 5231 |
+
"loss": 1.2668,
|
| 5232 |
+
"step": 737
|
| 5233 |
+
},
|
| 5234 |
+
{
|
| 5235 |
+
"epoch": 0.07698225392147914,
|
| 5236 |
+
"grad_norm": 0.5797216296195984,
|
| 5237 |
+
"learning_rate": 3.2972832976918554e-08,
|
| 5238 |
+
"loss": 1.2089,
|
| 5239 |
+
"step": 738
|
| 5240 |
+
},
|
| 5241 |
+
{
|
| 5242 |
+
"epoch": 0.07708656591866273,
|
| 5243 |
+
"grad_norm": 0.5604730844497681,
|
| 5244 |
+
"learning_rate": 2.2898185199826673e-08,
|
| 5245 |
+
"loss": 1.404,
|
| 5246 |
+
"step": 739
|
| 5247 |
+
},
|
| 5248 |
+
{
|
| 5249 |
+
"epoch": 0.07719087791584629,
|
| 5250 |
+
"grad_norm": 0.5185406804084778,
|
| 5251 |
+
"learning_rate": 1.4655039879740706e-08,
|
| 5252 |
+
"loss": 1.2627,
|
| 5253 |
+
"step": 740
|
| 5254 |
+
},
|
| 5255 |
+
{
|
| 5256 |
+
"epoch": 0.07729518991302987,
|
| 5257 |
+
"grad_norm": 0.5420083999633789,
|
| 5258 |
+
"learning_rate": 8.243548024655656e-09,
|
| 5259 |
+
"loss": 1.1388,
|
| 5260 |
+
"step": 741
|
| 5261 |
+
},
|
| 5262 |
+
{
|
| 5263 |
+
"epoch": 0.07739950191021344,
|
| 5264 |
+
"grad_norm": 0.5024222731590271,
|
| 5265 |
+
"learning_rate": 3.663827088085103e-09,
|
| 5266 |
+
"loss": 1.2322,
|
| 5267 |
+
"step": 742
|
| 5268 |
+
},
|
| 5269 |
+
{
|
| 5270 |
+
"epoch": 0.07750381390739702,
|
| 5271 |
+
"grad_norm": 0.6090103387832642,
|
| 5272 |
+
"learning_rate": 9.159609669406877e-10,
|
| 5273 |
+
"loss": 1.3192,
|
| 5274 |
+
"step": 743
|
| 5275 |
+
},
|
| 5276 |
+
{
|
| 5277 |
+
"epoch": 0.0776081259045806,
|
| 5278 |
+
"grad_norm": 0.53780597448349,
|
| 5279 |
+
"learning_rate": 0.0,
|
| 5280 |
+
"loss": 1.13,
|
| 5281 |
+
"step": 744
|
| 5282 |
}
|
| 5283 |
],
|
| 5284 |
"logging_steps": 1,
|
|
|
|
| 5302 |
"should_evaluate": false,
|
| 5303 |
"should_log": false,
|
| 5304 |
"should_save": true,
|
| 5305 |
+
"should_training_stop": true
|
| 5306 |
},
|
| 5307 |
"attributes": {}
|
| 5308 |
}
|
| 5309 |
},
|
| 5310 |
+
"total_flos": 2.381458180229038e+18,
|
| 5311 |
"train_batch_size": 4,
|
| 5312 |
"trial_name": null,
|
| 5313 |
"trial_params": null
|