Upload folder using huggingface_hub
Browse files- adapter_model.bin +1 -1
- all_results.json +12 -12
- eval_results.json +9 -9
- runs/Mar23_19-24-50_5b81996377db/events.out.tfevents.1711221893.5b81996377db.34.0 +3 -0
- runs/Mar23_19-25-37_5b81996377db/events.out.tfevents.1711222554.5b81996377db.34.1 +3 -0
- runs/Mar23_19-38-54_5b81996377db/events.out.tfevents.1711222735.5b81996377db.34.2 +3 -0
- runs/Mar23_19-38-54_5b81996377db/events.out.tfevents.1711242642.5b81996377db.34.3 +3 -0
- train_results.json +3 -3
- trainer_state.json +108 -108
adapter_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 160086542
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9bfec96e37ef6634530ef3293f303bd6acefe9e3e6d6abfe2f6ce7328f075f8a
|
| 3 |
size 160086542
|
all_results.json
CHANGED
|
@@ -1,17 +1,17 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_f1_score": 0.
|
| 5 |
-
"eval_gmean": 0.
|
| 6 |
-
"eval_loss": 0.
|
| 7 |
-
"eval_precision": 0.
|
| 8 |
-
"eval_recall": 0.
|
| 9 |
-
"eval_runtime":
|
| 10 |
-
"eval_samples_per_second": 0.
|
| 11 |
-
"eval_steps_per_second": 0.
|
| 12 |
"total_flos": 1.0505428534296576e+17,
|
| 13 |
-
"train_loss": 0.
|
| 14 |
-
"train_runtime":
|
| 15 |
-
"train_samples_per_second": 0.
|
| 16 |
"train_steps_per_second": 0.004
|
| 17 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"eval_accuracy": 0.87,
|
| 4 |
+
"eval_f1_score": 0.8705157155935307,
|
| 5 |
+
"eval_gmean": 0.8715093639362524,
|
| 6 |
+
"eval_loss": 0.26869142055511475,
|
| 7 |
+
"eval_precision": 0.8726060606060605,
|
| 8 |
+
"eval_recall": 0.87,
|
| 9 |
+
"eval_runtime": 308.1312,
|
| 10 |
+
"eval_samples_per_second": 0.325,
|
| 11 |
+
"eval_steps_per_second": 0.042,
|
| 12 |
"total_flos": 1.0505428534296576e+17,
|
| 13 |
+
"train_loss": 0.3734391375285823,
|
| 14 |
+
"train_runtime": 19598.5588,
|
| 15 |
+
"train_samples_per_second": 0.268,
|
| 16 |
"train_steps_per_second": 0.004
|
| 17 |
}
|
eval_results.json
CHANGED
|
@@ -1,12 +1,12 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_f1_score": 0.
|
| 5 |
-
"eval_gmean": 0.
|
| 6 |
-
"eval_loss": 0.
|
| 7 |
-
"eval_precision": 0.
|
| 8 |
-
"eval_recall": 0.
|
| 9 |
-
"eval_runtime":
|
| 10 |
-
"eval_samples_per_second": 0.
|
| 11 |
-
"eval_steps_per_second": 0.
|
| 12 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"eval_accuracy": 0.87,
|
| 4 |
+
"eval_f1_score": 0.8705157155935307,
|
| 5 |
+
"eval_gmean": 0.8715093639362524,
|
| 6 |
+
"eval_loss": 0.26869142055511475,
|
| 7 |
+
"eval_precision": 0.8726060606060605,
|
| 8 |
+
"eval_recall": 0.87,
|
| 9 |
+
"eval_runtime": 308.1312,
|
| 10 |
+
"eval_samples_per_second": 0.325,
|
| 11 |
+
"eval_steps_per_second": 0.042
|
| 12 |
}
|
runs/Mar23_19-24-50_5b81996377db/events.out.tfevents.1711221893.5b81996377db.34.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:17b7f6ddf7da1b4c3d5ca2dbbec9ee6edf27a0f95df49e9da2f593ae094de293
|
| 3 |
+
size 4897
|
runs/Mar23_19-25-37_5b81996377db/events.out.tfevents.1711222554.5b81996377db.34.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fb051d3bdeadbd01f9f6b23c3a70adbb9441732a12af5659daf265b4f78b4ef1
|
| 3 |
+
size 538
|
runs/Mar23_19-38-54_5b81996377db/events.out.tfevents.1711222735.5b81996377db.34.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8c66ef2a5b80daad688e46d7f99812a91756665300a7f0211c0eec0273085fb8
|
| 3 |
+
size 12485
|
runs/Mar23_19-38-54_5b81996377db/events.out.tfevents.1711242642.5b81996377db.34.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8cfe8a27bf8a62f29eb53c5dab4c88c26413d2838bb1da25cd5deb0ac551c6fb
|
| 3 |
+
size 605
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
"total_flos": 1.0505428534296576e+17,
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second": 0.
|
| 7 |
"train_steps_per_second": 0.004
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
"total_flos": 1.0505428534296576e+17,
|
| 4 |
+
"train_loss": 0.3734391375285823,
|
| 5 |
+
"train_runtime": 19598.5588,
|
| 6 |
+
"train_samples_per_second": 0.268,
|
| 7 |
"train_steps_per_second": 0.004
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -10,29 +10,29 @@
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.1,
|
| 13 |
-
"grad_norm":
|
| 14 |
"learning_rate": 1.8500000000000002e-05,
|
| 15 |
-
"loss": 0.
|
| 16 |
"step": 8
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.1,
|
| 20 |
-
"eval_accuracy": 0.
|
| 21 |
-
"eval_f1_score": 0.
|
| 22 |
-
"eval_gmean": 0.
|
| 23 |
-
"eval_loss": 0.
|
| 24 |
-
"eval_precision": 0.
|
| 25 |
-
"eval_recall": 0.
|
| 26 |
-
"eval_runtime":
|
| 27 |
-
"eval_samples_per_second": 0.
|
| 28 |
-
"eval_steps_per_second": 0.
|
| 29 |
"step": 8
|
| 30 |
},
|
| 31 |
{
|
| 32 |
"epoch": 0.19,
|
| 33 |
-
"grad_norm":
|
| 34 |
"learning_rate": 1.65e-05,
|
| 35 |
-
"loss": 0.
|
| 36 |
"step": 16
|
| 37 |
},
|
| 38 |
{
|
|
@@ -40,181 +40,181 @@
|
|
| 40 |
"eval_accuracy": 0.87,
|
| 41 |
"eval_f1_score": 0.8705157155935307,
|
| 42 |
"eval_gmean": 0.8715093639362524,
|
| 43 |
-
"eval_loss": 0.
|
| 44 |
"eval_precision": 0.8726060606060605,
|
| 45 |
"eval_recall": 0.87,
|
| 46 |
-
"eval_runtime":
|
| 47 |
-
"eval_samples_per_second": 0.
|
| 48 |
-
"eval_steps_per_second": 0.
|
| 49 |
"step": 16
|
| 50 |
},
|
| 51 |
{
|
| 52 |
"epoch": 0.29,
|
| 53 |
-
"grad_norm":
|
| 54 |
"learning_rate": 1.45e-05,
|
| 55 |
-
"loss": 0.
|
| 56 |
"step": 24
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 0.29,
|
| 60 |
-
"eval_accuracy": 0.
|
| 61 |
-
"eval_f1_score": 0.
|
| 62 |
-
"eval_gmean": 0.
|
| 63 |
-
"eval_loss": 0.
|
| 64 |
-
"eval_precision": 0.
|
| 65 |
-
"eval_recall": 0.
|
| 66 |
-
"eval_runtime":
|
| 67 |
-
"eval_samples_per_second": 0.
|
| 68 |
-
"eval_steps_per_second": 0.
|
| 69 |
"step": 24
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.39,
|
| 73 |
-
"grad_norm":
|
| 74 |
"learning_rate": 1.25e-05,
|
| 75 |
-
"loss": 0.
|
| 76 |
"step": 32
|
| 77 |
},
|
| 78 |
{
|
| 79 |
"epoch": 0.39,
|
| 80 |
-
"eval_accuracy": 0.
|
| 81 |
-
"eval_f1_score": 0.
|
| 82 |
-
"eval_gmean": 0.
|
| 83 |
-
"eval_loss": 0.
|
| 84 |
-
"eval_precision": 0.
|
| 85 |
-
"eval_recall": 0.
|
| 86 |
-
"eval_runtime":
|
| 87 |
-
"eval_samples_per_second": 0.
|
| 88 |
-
"eval_steps_per_second": 0.
|
| 89 |
"step": 32
|
| 90 |
},
|
| 91 |
{
|
| 92 |
"epoch": 0.49,
|
| 93 |
-
"grad_norm":
|
| 94 |
"learning_rate": 1.0500000000000001e-05,
|
| 95 |
-
"loss": 0.
|
| 96 |
"step": 40
|
| 97 |
},
|
| 98 |
{
|
| 99 |
"epoch": 0.49,
|
| 100 |
-
"eval_accuracy": 0.
|
| 101 |
-
"eval_f1_score": 0.
|
| 102 |
-
"eval_gmean": 0.
|
| 103 |
-
"eval_loss": 0.
|
| 104 |
-
"eval_precision": 0.
|
| 105 |
-
"eval_recall": 0.
|
| 106 |
-
"eval_runtime":
|
| 107 |
-
"eval_samples_per_second": 0.
|
| 108 |
-
"eval_steps_per_second": 0.
|
| 109 |
"step": 40
|
| 110 |
},
|
| 111 |
{
|
| 112 |
"epoch": 0.58,
|
| 113 |
-
"grad_norm":
|
| 114 |
"learning_rate": 8.5e-06,
|
| 115 |
-
"loss": 0.
|
| 116 |
"step": 48
|
| 117 |
},
|
| 118 |
{
|
| 119 |
"epoch": 0.58,
|
| 120 |
-
"eval_accuracy": 0.
|
| 121 |
-
"eval_f1_score": 0.
|
| 122 |
-
"eval_gmean": 0.
|
| 123 |
-
"eval_loss": 0.
|
| 124 |
-
"eval_precision": 0.
|
| 125 |
-
"eval_recall": 0.
|
| 126 |
-
"eval_runtime":
|
| 127 |
-
"eval_samples_per_second": 0.
|
| 128 |
-
"eval_steps_per_second": 0.
|
| 129 |
"step": 48
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.68,
|
| 133 |
-
"grad_norm":
|
| 134 |
"learning_rate": 6.5000000000000004e-06,
|
| 135 |
-
"loss": 0.
|
| 136 |
"step": 56
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.68,
|
| 140 |
-
"eval_accuracy": 0.
|
| 141 |
-
"eval_f1_score": 0.
|
| 142 |
-
"eval_gmean": 0.
|
| 143 |
-
"eval_loss": 0.
|
| 144 |
-
"eval_precision": 0.
|
| 145 |
-
"eval_recall": 0.
|
| 146 |
-
"eval_runtime":
|
| 147 |
-
"eval_samples_per_second": 0.
|
| 148 |
-
"eval_steps_per_second": 0.
|
| 149 |
"step": 56
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 0.78,
|
| 153 |
-
"grad_norm":
|
| 154 |
"learning_rate": 4.5e-06,
|
| 155 |
-
"loss": 0.
|
| 156 |
"step": 64
|
| 157 |
},
|
| 158 |
{
|
| 159 |
"epoch": 0.78,
|
| 160 |
-
"eval_accuracy": 0.
|
| 161 |
-
"eval_f1_score": 0.
|
| 162 |
-
"eval_gmean": 0.
|
| 163 |
-
"eval_loss": 0.
|
| 164 |
-
"eval_precision": 0.
|
| 165 |
-
"eval_recall": 0.
|
| 166 |
-
"eval_runtime":
|
| 167 |
-
"eval_samples_per_second": 0.
|
| 168 |
-
"eval_steps_per_second": 0.
|
| 169 |
"step": 64
|
| 170 |
},
|
| 171 |
{
|
| 172 |
"epoch": 0.88,
|
| 173 |
-
"grad_norm":
|
| 174 |
"learning_rate": 2.5e-06,
|
| 175 |
-
"loss": 0.
|
| 176 |
"step": 72
|
| 177 |
},
|
| 178 |
{
|
| 179 |
"epoch": 0.88,
|
| 180 |
-
"eval_accuracy": 0.
|
| 181 |
-
"eval_f1_score": 0.
|
| 182 |
-
"eval_gmean": 0.
|
| 183 |
-
"eval_loss": 0.
|
| 184 |
-
"eval_precision": 0.
|
| 185 |
-
"eval_recall": 0.
|
| 186 |
-
"eval_runtime":
|
| 187 |
-
"eval_samples_per_second": 0.
|
| 188 |
-
"eval_steps_per_second": 0.
|
| 189 |
"step": 72
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.97,
|
| 193 |
-
"grad_norm":
|
| 194 |
"learning_rate": 5.000000000000001e-07,
|
| 195 |
-
"loss": 0.
|
| 196 |
"step": 80
|
| 197 |
},
|
| 198 |
{
|
| 199 |
"epoch": 0.97,
|
| 200 |
-
"eval_accuracy": 0.
|
| 201 |
-
"eval_f1_score": 0.
|
| 202 |
-
"eval_gmean": 0.
|
| 203 |
-
"eval_loss": 0.
|
| 204 |
-
"eval_precision": 0.
|
| 205 |
-
"eval_recall": 0.
|
| 206 |
-
"eval_runtime":
|
| 207 |
-
"eval_samples_per_second": 0.
|
| 208 |
-
"eval_steps_per_second": 0.
|
| 209 |
"step": 80
|
| 210 |
},
|
| 211 |
{
|
| 212 |
"epoch": 1.0,
|
| 213 |
"step": 82,
|
| 214 |
"total_flos": 1.0505428534296576e+17,
|
| 215 |
-
"train_loss": 0.
|
| 216 |
-
"train_runtime":
|
| 217 |
-
"train_samples_per_second": 0.
|
| 218 |
"train_steps_per_second": 0.004
|
| 219 |
}
|
| 220 |
],
|
|
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.1,
|
| 13 |
+
"grad_norm": 6.96875,
|
| 14 |
"learning_rate": 1.8500000000000002e-05,
|
| 15 |
+
"loss": 0.3825,
|
| 16 |
"step": 8
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.1,
|
| 20 |
+
"eval_accuracy": 0.87,
|
| 21 |
+
"eval_f1_score": 0.8705157155935307,
|
| 22 |
+
"eval_gmean": 0.8715093639362524,
|
| 23 |
+
"eval_loss": 0.27127930521965027,
|
| 24 |
+
"eval_precision": 0.8726060606060605,
|
| 25 |
+
"eval_recall": 0.87,
|
| 26 |
+
"eval_runtime": 308.2568,
|
| 27 |
+
"eval_samples_per_second": 0.324,
|
| 28 |
+
"eval_steps_per_second": 0.042,
|
| 29 |
"step": 8
|
| 30 |
},
|
| 31 |
{
|
| 32 |
"epoch": 0.19,
|
| 33 |
+
"grad_norm": 8.0625,
|
| 34 |
"learning_rate": 1.65e-05,
|
| 35 |
+
"loss": 0.3684,
|
| 36 |
"step": 16
|
| 37 |
},
|
| 38 |
{
|
|
|
|
| 40 |
"eval_accuracy": 0.87,
|
| 41 |
"eval_f1_score": 0.8705157155935307,
|
| 42 |
"eval_gmean": 0.8715093639362524,
|
| 43 |
+
"eval_loss": 0.27037107944488525,
|
| 44 |
"eval_precision": 0.8726060606060605,
|
| 45 |
"eval_recall": 0.87,
|
| 46 |
+
"eval_runtime": 308.6792,
|
| 47 |
+
"eval_samples_per_second": 0.324,
|
| 48 |
+
"eval_steps_per_second": 0.042,
|
| 49 |
"step": 16
|
| 50 |
},
|
| 51 |
{
|
| 52 |
"epoch": 0.29,
|
| 53 |
+
"grad_norm": 7.1875,
|
| 54 |
"learning_rate": 1.45e-05,
|
| 55 |
+
"loss": 0.3846,
|
| 56 |
"step": 24
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 0.29,
|
| 60 |
+
"eval_accuracy": 0.87,
|
| 61 |
+
"eval_f1_score": 0.8705157155935307,
|
| 62 |
+
"eval_gmean": 0.8715093639362524,
|
| 63 |
+
"eval_loss": 0.26976072788238525,
|
| 64 |
+
"eval_precision": 0.8726060606060605,
|
| 65 |
+
"eval_recall": 0.87,
|
| 66 |
+
"eval_runtime": 308.4781,
|
| 67 |
+
"eval_samples_per_second": 0.324,
|
| 68 |
+
"eval_steps_per_second": 0.042,
|
| 69 |
"step": 24
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.39,
|
| 73 |
+
"grad_norm": 6.96875,
|
| 74 |
"learning_rate": 1.25e-05,
|
| 75 |
+
"loss": 0.368,
|
| 76 |
"step": 32
|
| 77 |
},
|
| 78 |
{
|
| 79 |
"epoch": 0.39,
|
| 80 |
+
"eval_accuracy": 0.87,
|
| 81 |
+
"eval_f1_score": 0.8705157155935307,
|
| 82 |
+
"eval_gmean": 0.8715093639362524,
|
| 83 |
+
"eval_loss": 0.26942384243011475,
|
| 84 |
+
"eval_precision": 0.8726060606060605,
|
| 85 |
+
"eval_recall": 0.87,
|
| 86 |
+
"eval_runtime": 308.2201,
|
| 87 |
+
"eval_samples_per_second": 0.324,
|
| 88 |
+
"eval_steps_per_second": 0.042,
|
| 89 |
"step": 32
|
| 90 |
},
|
| 91 |
{
|
| 92 |
"epoch": 0.49,
|
| 93 |
+
"grad_norm": 6.34375,
|
| 94 |
"learning_rate": 1.0500000000000001e-05,
|
| 95 |
+
"loss": 0.3954,
|
| 96 |
"step": 40
|
| 97 |
},
|
| 98 |
{
|
| 99 |
"epoch": 0.49,
|
| 100 |
+
"eval_accuracy": 0.87,
|
| 101 |
+
"eval_f1_score": 0.8705157155935307,
|
| 102 |
+
"eval_gmean": 0.8715093639362524,
|
| 103 |
+
"eval_loss": 0.26915037631988525,
|
| 104 |
+
"eval_precision": 0.8726060606060605,
|
| 105 |
+
"eval_recall": 0.87,
|
| 106 |
+
"eval_runtime": 308.3621,
|
| 107 |
+
"eval_samples_per_second": 0.324,
|
| 108 |
+
"eval_steps_per_second": 0.042,
|
| 109 |
"step": 40
|
| 110 |
},
|
| 111 |
{
|
| 112 |
"epoch": 0.58,
|
| 113 |
+
"grad_norm": 5.28125,
|
| 114 |
"learning_rate": 8.5e-06,
|
| 115 |
+
"loss": 0.3952,
|
| 116 |
"step": 48
|
| 117 |
},
|
| 118 |
{
|
| 119 |
"epoch": 0.58,
|
| 120 |
+
"eval_accuracy": 0.87,
|
| 121 |
+
"eval_f1_score": 0.8705157155935307,
|
| 122 |
+
"eval_gmean": 0.8715093639362524,
|
| 123 |
+
"eval_loss": 0.2691015601158142,
|
| 124 |
+
"eval_precision": 0.8726060606060605,
|
| 125 |
+
"eval_recall": 0.87,
|
| 126 |
+
"eval_runtime": 307.9578,
|
| 127 |
+
"eval_samples_per_second": 0.325,
|
| 128 |
+
"eval_steps_per_second": 0.042,
|
| 129 |
"step": 48
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.68,
|
| 133 |
+
"grad_norm": 5.65625,
|
| 134 |
"learning_rate": 6.5000000000000004e-06,
|
| 135 |
+
"loss": 0.368,
|
| 136 |
"step": 56
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.68,
|
| 140 |
+
"eval_accuracy": 0.87,
|
| 141 |
+
"eval_f1_score": 0.8705157155935307,
|
| 142 |
+
"eval_gmean": 0.8715093639362524,
|
| 143 |
+
"eval_loss": 0.2687695324420929,
|
| 144 |
+
"eval_precision": 0.8726060606060605,
|
| 145 |
+
"eval_recall": 0.87,
|
| 146 |
+
"eval_runtime": 308.2935,
|
| 147 |
+
"eval_samples_per_second": 0.324,
|
| 148 |
+
"eval_steps_per_second": 0.042,
|
| 149 |
"step": 56
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 0.78,
|
| 153 |
+
"grad_norm": 6.03125,
|
| 154 |
"learning_rate": 4.5e-06,
|
| 155 |
+
"loss": 0.3826,
|
| 156 |
"step": 64
|
| 157 |
},
|
| 158 |
{
|
| 159 |
"epoch": 0.78,
|
| 160 |
+
"eval_accuracy": 0.87,
|
| 161 |
+
"eval_f1_score": 0.8705157155935307,
|
| 162 |
+
"eval_gmean": 0.8715093639362524,
|
| 163 |
+
"eval_loss": 0.26869142055511475,
|
| 164 |
+
"eval_precision": 0.8726060606060605,
|
| 165 |
+
"eval_recall": 0.87,
|
| 166 |
+
"eval_runtime": 307.9498,
|
| 167 |
+
"eval_samples_per_second": 0.325,
|
| 168 |
+
"eval_steps_per_second": 0.042,
|
| 169 |
"step": 64
|
| 170 |
},
|
| 171 |
{
|
| 172 |
"epoch": 0.88,
|
| 173 |
+
"grad_norm": 10.0,
|
| 174 |
"learning_rate": 2.5e-06,
|
| 175 |
+
"loss": 0.3254,
|
| 176 |
"step": 72
|
| 177 |
},
|
| 178 |
{
|
| 179 |
"epoch": 0.88,
|
| 180 |
+
"eval_accuracy": 0.87,
|
| 181 |
+
"eval_f1_score": 0.8705157155935307,
|
| 182 |
+
"eval_gmean": 0.8715093639362524,
|
| 183 |
+
"eval_loss": 0.26869142055511475,
|
| 184 |
+
"eval_precision": 0.8726060606060605,
|
| 185 |
+
"eval_recall": 0.87,
|
| 186 |
+
"eval_runtime": 308.1218,
|
| 187 |
+
"eval_samples_per_second": 0.325,
|
| 188 |
+
"eval_steps_per_second": 0.042,
|
| 189 |
"step": 72
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.97,
|
| 193 |
+
"grad_norm": 8.75,
|
| 194 |
"learning_rate": 5.000000000000001e-07,
|
| 195 |
+
"loss": 0.3547,
|
| 196 |
"step": 80
|
| 197 |
},
|
| 198 |
{
|
| 199 |
"epoch": 0.97,
|
| 200 |
+
"eval_accuracy": 0.87,
|
| 201 |
+
"eval_f1_score": 0.8705157155935307,
|
| 202 |
+
"eval_gmean": 0.8715093639362524,
|
| 203 |
+
"eval_loss": 0.26869142055511475,
|
| 204 |
+
"eval_precision": 0.8726060606060605,
|
| 205 |
+
"eval_recall": 0.87,
|
| 206 |
+
"eval_runtime": 308.2047,
|
| 207 |
+
"eval_samples_per_second": 0.324,
|
| 208 |
+
"eval_steps_per_second": 0.042,
|
| 209 |
"step": 80
|
| 210 |
},
|
| 211 |
{
|
| 212 |
"epoch": 1.0,
|
| 213 |
"step": 82,
|
| 214 |
"total_flos": 1.0505428534296576e+17,
|
| 215 |
+
"train_loss": 0.3734391375285823,
|
| 216 |
+
"train_runtime": 19598.5588,
|
| 217 |
+
"train_samples_per_second": 0.268,
|
| 218 |
"train_steps_per_second": 0.004
|
| 219 |
}
|
| 220 |
],
|