Training in progress, step 100
Browse files- all_results.json +11 -6
- eval_results.json +8 -0
- pytorch_model.bin +1 -1
- runs/May15_17-41-07_601c3d0c9bcd/events.out.tfevents.1715795412.601c3d0c9bcd.2403.3 +3 -0
- runs/May15_17-51-41_601c3d0c9bcd/events.out.tfevents.1715795517.601c3d0c9bcd.2403.4 +3 -0
- test_results.json +8 -0
- train_results.json +6 -6
- trainer_state.json +271 -64
- training_args.bin +1 -1
all_results.json
CHANGED
|
@@ -1,8 +1,13 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"
|
| 6 |
-
"
|
| 7 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.9591836734693877,
|
| 4 |
+
"eval_loss": 0.17216908931732178,
|
| 5 |
+
"eval_runtime": 3.4667,
|
| 6 |
+
"eval_samples_per_second": 28.269,
|
| 7 |
+
"eval_steps_per_second": 3.75,
|
| 8 |
+
"total_flos": 6.129616378514227e+17,
|
| 9 |
+
"train_loss": 0.1937787554860115,
|
| 10 |
+
"train_runtime": 492.9488,
|
| 11 |
+
"train_samples_per_second": 16.046,
|
| 12 |
+
"train_steps_per_second": 1.014
|
| 13 |
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.9591836734693877,
|
| 4 |
+
"eval_loss": 0.17216908931732178,
|
| 5 |
+
"eval_runtime": 2.3887,
|
| 6 |
+
"eval_samples_per_second": 41.026,
|
| 7 |
+
"eval_steps_per_second": 5.442
|
| 8 |
+
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 343269162
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3ede0e8d4c0f611742667a390b1ea2dce854b835570a79c7e5420295720b0a74
|
| 3 |
size 343269162
|
runs/May15_17-41-07_601c3d0c9bcd/events.out.tfevents.1715795412.601c3d0c9bcd.2403.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:87dc7717cf4564b6a5b256aab57e9ec889f1819b796dcc838017ca715c9864a0
|
| 3 |
+
size 734
|
runs/May15_17-51-41_601c3d0c9bcd/events.out.tfevents.1715795517.601c3d0c9bcd.2403.4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8e48d72ad24d03f9797125b8afb10a881216de95746fd2cf5ec91f249266fb4a
|
| 3 |
+
size 5801
|
test_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.9591836734693877,
|
| 4 |
+
"eval_loss": 0.17216908931732178,
|
| 5 |
+
"eval_runtime": 3.4667,
|
| 6 |
+
"eval_samples_per_second": 28.269,
|
| 7 |
+
"eval_steps_per_second": 3.75
|
| 8 |
+
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 1.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"total_flos": 6.129616378514227e+17,
|
| 4 |
+
"train_loss": 0.1937787554860115,
|
| 5 |
+
"train_runtime": 492.9488,
|
| 6 |
+
"train_samples_per_second": 16.046,
|
| 7 |
+
"train_steps_per_second": 1.014
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,163 +1,370 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "./vit-accident-image/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
"epoch": 0.2,
|
| 12 |
-
"learning_rate": 0.
|
| 13 |
-
"loss": 0.
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.4,
|
| 18 |
-
"learning_rate": 0.
|
| 19 |
-
"loss": 0.
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
"epoch": 0.6,
|
| 24 |
-
"learning_rate": 0.
|
| 25 |
-
"loss": 0.
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.8,
|
| 30 |
-
"learning_rate": 0.
|
| 31 |
-
"loss": 0.
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 1.0,
|
| 36 |
-
"learning_rate": 0.
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.2,
|
| 42 |
-
"learning_rate": 0.
|
| 43 |
-
"loss": 0.
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 1.4,
|
| 48 |
-
"learning_rate": 0.
|
| 49 |
-
"loss": 0.
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
"epoch": 1.6,
|
| 54 |
-
"learning_rate": 0.
|
| 55 |
-
"loss": 0.
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 1.8,
|
| 60 |
-
"learning_rate": 0.
|
| 61 |
-
"loss": 0.
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
"epoch": 2.0,
|
| 66 |
-
"learning_rate": 0.
|
| 67 |
-
"loss": 0.
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 2.0,
|
| 72 |
-
"eval_accuracy": 0.
|
| 73 |
-
"eval_loss": 0.
|
| 74 |
-
"eval_runtime": 2.
|
| 75 |
-
"eval_samples_per_second": 43.
|
| 76 |
-
"eval_steps_per_second": 5.
|
| 77 |
"step": 100
|
| 78 |
},
|
| 79 |
{
|
| 80 |
"epoch": 2.2,
|
| 81 |
-
"learning_rate":
|
| 82 |
-
"loss": 0.
|
| 83 |
"step": 110
|
| 84 |
},
|
| 85 |
{
|
| 86 |
"epoch": 2.4,
|
| 87 |
-
"learning_rate":
|
| 88 |
-
"loss": 0.
|
| 89 |
"step": 120
|
| 90 |
},
|
| 91 |
{
|
| 92 |
"epoch": 2.6,
|
| 93 |
-
"learning_rate":
|
| 94 |
-
"loss": 0.
|
| 95 |
"step": 130
|
| 96 |
},
|
| 97 |
{
|
| 98 |
"epoch": 2.8,
|
| 99 |
-
"learning_rate":
|
| 100 |
-
"loss": 0.
|
| 101 |
"step": 140
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 3.0,
|
| 105 |
-
"learning_rate":
|
| 106 |
-
"loss": 0.
|
| 107 |
"step": 150
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 3.2,
|
| 111 |
-
"learning_rate":
|
| 112 |
-
"loss": 0.
|
| 113 |
"step": 160
|
| 114 |
},
|
| 115 |
{
|
| 116 |
"epoch": 3.4,
|
| 117 |
-
"learning_rate":
|
| 118 |
-
"loss": 0.
|
| 119 |
"step": 170
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 3.6,
|
| 123 |
-
"learning_rate":
|
| 124 |
-
"loss": 0.
|
| 125 |
"step": 180
|
| 126 |
},
|
| 127 |
{
|
| 128 |
"epoch": 3.8,
|
| 129 |
-
"learning_rate":
|
| 130 |
-
"loss": 0.
|
| 131 |
"step": 190
|
| 132 |
},
|
| 133 |
{
|
| 134 |
"epoch": 4.0,
|
| 135 |
-
"learning_rate": 0.
|
| 136 |
-
"loss": 0.
|
| 137 |
"step": 200
|
| 138 |
},
|
| 139 |
{
|
| 140 |
"epoch": 4.0,
|
| 141 |
-
"eval_accuracy": 0.
|
| 142 |
-
"eval_loss": 0.
|
| 143 |
-
"eval_runtime":
|
| 144 |
-
"eval_samples_per_second":
|
| 145 |
-
"eval_steps_per_second":
|
| 146 |
"step": 200
|
| 147 |
},
|
| 148 |
{
|
| 149 |
-
"epoch": 4.
|
| 150 |
-
"
|
| 151 |
-
"
|
| 152 |
-
"
|
| 153 |
-
|
| 154 |
-
|
| 155 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 156 |
}
|
| 157 |
],
|
| 158 |
-
"max_steps":
|
| 159 |
-
"num_train_epochs":
|
| 160 |
-
"total_flos":
|
| 161 |
"trial_name": null,
|
| 162 |
"trial_params": null
|
| 163 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.17216908931732178,
|
| 3 |
+
"best_model_checkpoint": "./vit-accident-image/checkpoint-400",
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"global_step": 500,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
"epoch": 0.2,
|
| 12 |
+
"learning_rate": 0.000196,
|
| 13 |
+
"loss": 0.6688,
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.4,
|
| 18 |
+
"learning_rate": 0.000192,
|
| 19 |
+
"loss": 0.6088,
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
"epoch": 0.6,
|
| 24 |
+
"learning_rate": 0.000188,
|
| 25 |
+
"loss": 0.6081,
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.8,
|
| 30 |
+
"learning_rate": 0.00018400000000000003,
|
| 31 |
+
"loss": 0.5727,
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 1.0,
|
| 36 |
+
"learning_rate": 0.00018,
|
| 37 |
+
"loss": 0.5506,
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.2,
|
| 42 |
+
"learning_rate": 0.00017600000000000002,
|
| 43 |
+
"loss": 0.3946,
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 1.4,
|
| 48 |
+
"learning_rate": 0.000172,
|
| 49 |
+
"loss": 0.3847,
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
"epoch": 1.6,
|
| 54 |
+
"learning_rate": 0.000168,
|
| 55 |
+
"loss": 0.5908,
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 1.8,
|
| 60 |
+
"learning_rate": 0.000164,
|
| 61 |
+
"loss": 0.3266,
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
"epoch": 2.0,
|
| 66 |
+
"learning_rate": 0.00016,
|
| 67 |
+
"loss": 0.5025,
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 2.0,
|
| 72 |
+
"eval_accuracy": 0.8775510204081632,
|
| 73 |
+
"eval_loss": 0.3267006278038025,
|
| 74 |
+
"eval_runtime": 2.247,
|
| 75 |
+
"eval_samples_per_second": 43.613,
|
| 76 |
+
"eval_steps_per_second": 5.785,
|
| 77 |
"step": 100
|
| 78 |
},
|
| 79 |
{
|
| 80 |
"epoch": 2.2,
|
| 81 |
+
"learning_rate": 0.00015600000000000002,
|
| 82 |
+
"loss": 0.287,
|
| 83 |
"step": 110
|
| 84 |
},
|
| 85 |
{
|
| 86 |
"epoch": 2.4,
|
| 87 |
+
"learning_rate": 0.000152,
|
| 88 |
+
"loss": 0.2996,
|
| 89 |
"step": 120
|
| 90 |
},
|
| 91 |
{
|
| 92 |
"epoch": 2.6,
|
| 93 |
+
"learning_rate": 0.000148,
|
| 94 |
+
"loss": 0.1329,
|
| 95 |
"step": 130
|
| 96 |
},
|
| 97 |
{
|
| 98 |
"epoch": 2.8,
|
| 99 |
+
"learning_rate": 0.000144,
|
| 100 |
+
"loss": 0.2857,
|
| 101 |
"step": 140
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 3.0,
|
| 105 |
+
"learning_rate": 0.00014,
|
| 106 |
+
"loss": 0.2885,
|
| 107 |
"step": 150
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 3.2,
|
| 111 |
+
"learning_rate": 0.00013600000000000003,
|
| 112 |
+
"loss": 0.244,
|
| 113 |
"step": 160
|
| 114 |
},
|
| 115 |
{
|
| 116 |
"epoch": 3.4,
|
| 117 |
+
"learning_rate": 0.000132,
|
| 118 |
+
"loss": 0.1873,
|
| 119 |
"step": 170
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 3.6,
|
| 123 |
+
"learning_rate": 0.00012800000000000002,
|
| 124 |
+
"loss": 0.1446,
|
| 125 |
"step": 180
|
| 126 |
},
|
| 127 |
{
|
| 128 |
"epoch": 3.8,
|
| 129 |
+
"learning_rate": 0.000124,
|
| 130 |
+
"loss": 0.1887,
|
| 131 |
"step": 190
|
| 132 |
},
|
| 133 |
{
|
| 134 |
"epoch": 4.0,
|
| 135 |
+
"learning_rate": 0.00012,
|
| 136 |
+
"loss": 0.204,
|
| 137 |
"step": 200
|
| 138 |
},
|
| 139 |
{
|
| 140 |
"epoch": 4.0,
|
| 141 |
+
"eval_accuracy": 0.9081632653061225,
|
| 142 |
+
"eval_loss": 0.29408177733421326,
|
| 143 |
+
"eval_runtime": 3.0909,
|
| 144 |
+
"eval_samples_per_second": 31.706,
|
| 145 |
+
"eval_steps_per_second": 4.206,
|
| 146 |
"step": 200
|
| 147 |
},
|
| 148 |
{
|
| 149 |
+
"epoch": 4.2,
|
| 150 |
+
"learning_rate": 0.000116,
|
| 151 |
+
"loss": 0.0822,
|
| 152 |
+
"step": 210
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 4.4,
|
| 156 |
+
"learning_rate": 0.00011200000000000001,
|
| 157 |
+
"loss": 0.0912,
|
| 158 |
+
"step": 220
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 4.6,
|
| 162 |
+
"learning_rate": 0.00010800000000000001,
|
| 163 |
+
"loss": 0.1816,
|
| 164 |
+
"step": 230
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 4.8,
|
| 168 |
+
"learning_rate": 0.00010400000000000001,
|
| 169 |
+
"loss": 0.1005,
|
| 170 |
+
"step": 240
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 5.0,
|
| 174 |
+
"learning_rate": 0.0001,
|
| 175 |
+
"loss": 0.1167,
|
| 176 |
+
"step": 250
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 5.2,
|
| 180 |
+
"learning_rate": 9.6e-05,
|
| 181 |
+
"loss": 0.1081,
|
| 182 |
+
"step": 260
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 5.4,
|
| 186 |
+
"learning_rate": 9.200000000000001e-05,
|
| 187 |
+
"loss": 0.0973,
|
| 188 |
+
"step": 270
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 5.6,
|
| 192 |
+
"learning_rate": 8.800000000000001e-05,
|
| 193 |
+
"loss": 0.1825,
|
| 194 |
+
"step": 280
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 5.8,
|
| 198 |
+
"learning_rate": 8.4e-05,
|
| 199 |
+
"loss": 0.0941,
|
| 200 |
+
"step": 290
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 6.0,
|
| 204 |
+
"learning_rate": 8e-05,
|
| 205 |
+
"loss": 0.1425,
|
| 206 |
+
"step": 300
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 6.0,
|
| 210 |
+
"eval_accuracy": 0.9081632653061225,
|
| 211 |
+
"eval_loss": 0.34069785475730896,
|
| 212 |
+
"eval_runtime": 2.9263,
|
| 213 |
+
"eval_samples_per_second": 33.489,
|
| 214 |
+
"eval_steps_per_second": 4.442,
|
| 215 |
+
"step": 300
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 6.2,
|
| 219 |
+
"learning_rate": 7.6e-05,
|
| 220 |
+
"loss": 0.093,
|
| 221 |
+
"step": 310
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 6.4,
|
| 225 |
+
"learning_rate": 7.2e-05,
|
| 226 |
+
"loss": 0.0847,
|
| 227 |
+
"step": 320
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 6.6,
|
| 231 |
+
"learning_rate": 6.800000000000001e-05,
|
| 232 |
+
"loss": 0.0561,
|
| 233 |
+
"step": 330
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 6.8,
|
| 237 |
+
"learning_rate": 6.400000000000001e-05,
|
| 238 |
+
"loss": 0.0832,
|
| 239 |
+
"step": 340
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 7.0,
|
| 243 |
+
"learning_rate": 6e-05,
|
| 244 |
+
"loss": 0.0663,
|
| 245 |
+
"step": 350
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 7.2,
|
| 249 |
+
"learning_rate": 5.6000000000000006e-05,
|
| 250 |
+
"loss": 0.0626,
|
| 251 |
+
"step": 360
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 7.4,
|
| 255 |
+
"learning_rate": 5.2000000000000004e-05,
|
| 256 |
+
"loss": 0.0282,
|
| 257 |
+
"step": 370
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 7.6,
|
| 261 |
+
"learning_rate": 4.8e-05,
|
| 262 |
+
"loss": 0.0528,
|
| 263 |
+
"step": 380
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 7.8,
|
| 267 |
+
"learning_rate": 4.4000000000000006e-05,
|
| 268 |
+
"loss": 0.0416,
|
| 269 |
+
"step": 390
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 8.0,
|
| 273 |
+
"learning_rate": 4e-05,
|
| 274 |
+
"loss": 0.091,
|
| 275 |
+
"step": 400
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 8.0,
|
| 279 |
+
"eval_accuracy": 0.9591836734693877,
|
| 280 |
+
"eval_loss": 0.17216908931732178,
|
| 281 |
+
"eval_runtime": 2.1538,
|
| 282 |
+
"eval_samples_per_second": 45.502,
|
| 283 |
+
"eval_steps_per_second": 6.036,
|
| 284 |
+
"step": 400
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 8.2,
|
| 288 |
+
"learning_rate": 3.6e-05,
|
| 289 |
+
"loss": 0.0602,
|
| 290 |
+
"step": 410
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 8.4,
|
| 294 |
+
"learning_rate": 3.2000000000000005e-05,
|
| 295 |
+
"loss": 0.0267,
|
| 296 |
+
"step": 420
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 8.6,
|
| 300 |
+
"learning_rate": 2.8000000000000003e-05,
|
| 301 |
+
"loss": 0.0447,
|
| 302 |
+
"step": 430
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 8.8,
|
| 306 |
+
"learning_rate": 2.4e-05,
|
| 307 |
+
"loss": 0.038,
|
| 308 |
+
"step": 440
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 9.0,
|
| 312 |
+
"learning_rate": 2e-05,
|
| 313 |
+
"loss": 0.0393,
|
| 314 |
+
"step": 450
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 9.2,
|
| 318 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 319 |
+
"loss": 0.0271,
|
| 320 |
+
"step": 460
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 9.4,
|
| 324 |
+
"learning_rate": 1.2e-05,
|
| 325 |
+
"loss": 0.0318,
|
| 326 |
+
"step": 470
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 9.6,
|
| 330 |
+
"learning_rate": 8.000000000000001e-06,
|
| 331 |
+
"loss": 0.0465,
|
| 332 |
+
"step": 480
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 9.8,
|
| 336 |
+
"learning_rate": 4.000000000000001e-06,
|
| 337 |
+
"loss": 0.0154,
|
| 338 |
+
"step": 490
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 10.0,
|
| 342 |
+
"learning_rate": 0.0,
|
| 343 |
+
"loss": 0.0328,
|
| 344 |
+
"step": 500
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 10.0,
|
| 348 |
+
"eval_accuracy": 0.9387755102040817,
|
| 349 |
+
"eval_loss": 0.20507687330245972,
|
| 350 |
+
"eval_runtime": 3.1293,
|
| 351 |
+
"eval_samples_per_second": 31.316,
|
| 352 |
+
"eval_steps_per_second": 4.154,
|
| 353 |
+
"step": 500
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 10.0,
|
| 357 |
+
"step": 500,
|
| 358 |
+
"total_flos": 6.129616378514227e+17,
|
| 359 |
+
"train_loss": 0.1937787554860115,
|
| 360 |
+
"train_runtime": 492.9488,
|
| 361 |
+
"train_samples_per_second": 16.046,
|
| 362 |
+
"train_steps_per_second": 1.014
|
| 363 |
}
|
| 364 |
],
|
| 365 |
+
"max_steps": 500,
|
| 366 |
+
"num_train_epochs": 10,
|
| 367 |
+
"total_flos": 6.129616378514227e+17,
|
| 368 |
"trial_name": null,
|
| 369 |
"trial_params": null
|
| 370 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4408
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:daed8b6ce15cc6b3f87734aa2e8979670d0cfb855c83d202deb6e40526c5fca7
|
| 3 |
size 4408
|