lombardata
commited on
Commit
•
e91a92d
1
Parent(s):
74a2d73
🍻 cheers
Browse files- README.md +10 -6
- all_results.json +16 -0
- config.json +1 -1
- eval_results.json +12 -0
- train_results.json +8 -0
- trainer_state.json +349 -0
README.md
CHANGED
@@ -1,7 +1,11 @@
|
|
1 |
---
|
|
|
|
|
2 |
license: apache-2.0
|
3 |
base_model: facebook/dinov2-giant
|
4 |
tags:
|
|
|
|
|
5 |
- generated_from_trainer
|
6 |
metrics:
|
7 |
- accuracy
|
@@ -15,13 +19,13 @@ should probably proofread and complete it, then remove this comment. -->
|
|
15 |
|
16 |
# dinov2-giant-2024_01_02-kornia_img-size518_batch-size32_epochs20_freeze
|
17 |
|
18 |
-
This model is a fine-tuned version of [facebook/dinov2-giant](https://huggingface.co/facebook/dinov2-giant) on the
|
19 |
It achieves the following results on the evaluation set:
|
20 |
-
- Loss: 0.
|
21 |
-
- F1 Micro: 0.
|
22 |
-
- F1 Macro: 0.
|
23 |
-
- Roc Auc: 0.
|
24 |
-
- Accuracy: 0.
|
25 |
- Learning Rate: 0.001
|
26 |
|
27 |
## Model description
|
|
|
1 |
---
|
2 |
+
language:
|
3 |
+
- eng
|
4 |
license: apache-2.0
|
5 |
base_model: facebook/dinov2-giant
|
6 |
tags:
|
7 |
+
- multilabel-image-classification
|
8 |
+
- multilabel
|
9 |
- generated_from_trainer
|
10 |
metrics:
|
11 |
- accuracy
|
|
|
19 |
|
20 |
# dinov2-giant-2024_01_02-kornia_img-size518_batch-size32_epochs20_freeze
|
21 |
|
22 |
+
This model is a fine-tuned version of [facebook/dinov2-giant](https://huggingface.co/facebook/dinov2-giant) on the multilabel_complete_dataset dataset.
|
23 |
It achieves the following results on the evaluation set:
|
24 |
+
- Loss: 0.1266
|
25 |
+
- F1 Micro: 0.8142
|
26 |
+
- F1 Macro: 0.7719
|
27 |
+
- Roc Auc: 0.8801
|
28 |
+
- Accuracy: 0.5121
|
29 |
- Learning Rate: 0.001
|
30 |
|
31 |
## Model description
|
all_results.json
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 20.0,
|
3 |
+
"eval_accuracy": 0.5121436114044351,
|
4 |
+
"eval_f1_macro": 0.7719123466949233,
|
5 |
+
"eval_f1_micro": 0.8141891891891891,
|
6 |
+
"eval_loss": 0.12663225829601288,
|
7 |
+
"eval_roc_auc": 0.8801205728506106,
|
8 |
+
"eval_runtime": 1345.3407,
|
9 |
+
"eval_samples_per_second": 2.112,
|
10 |
+
"eval_steps_per_second": 0.066,
|
11 |
+
"learning_rate": 0.001,
|
12 |
+
"train_loss": 0.07645123254007367,
|
13 |
+
"train_runtime": 44768.9701,
|
14 |
+
"train_samples_per_second": 3.828,
|
15 |
+
"train_steps_per_second": 0.12
|
16 |
+
}
|
config.json
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
{
|
2 |
-
"_name_or_path": "facebook/dinov2-
|
3 |
"apply_layernorm": true,
|
4 |
"architectures": [
|
5 |
"NewheadDinov2ForImageClassification"
|
|
|
1 |
{
|
2 |
+
"_name_or_path": "facebook/dinov2-giant2024_01_03",
|
3 |
"apply_layernorm": true,
|
4 |
"architectures": [
|
5 |
"NewheadDinov2ForImageClassification"
|
eval_results.json
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 20.0,
|
3 |
+
"eval_accuracy": 0.5121436114044351,
|
4 |
+
"eval_f1_macro": 0.7719123466949233,
|
5 |
+
"eval_f1_micro": 0.8141891891891891,
|
6 |
+
"eval_loss": 0.12663225829601288,
|
7 |
+
"eval_roc_auc": 0.8801205728506106,
|
8 |
+
"eval_runtime": 1345.3407,
|
9 |
+
"eval_samples_per_second": 2.112,
|
10 |
+
"eval_steps_per_second": 0.066,
|
11 |
+
"learning_rate": 0.001
|
12 |
+
}
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 20.0,
|
3 |
+
"learning_rate": 0.001,
|
4 |
+
"train_loss": 0.07645123254007367,
|
5 |
+
"train_runtime": 44768.9701,
|
6 |
+
"train_samples_per_second": 3.828,
|
7 |
+
"train_steps_per_second": 0.12
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,349 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.1276090443134308,
|
3 |
+
"best_model_checkpoint": "/home1/datawork/mcontini/models/multilabel/huggingface/dinov2-giant-2024_01_02-kornia_img-size518_batch-size32_epochs20_freeze/checkpoint-5360",
|
4 |
+
"epoch": 20.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 5360,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 1.0,
|
13 |
+
"eval_accuracy": 0.4055019649874955,
|
14 |
+
"eval_f1_macro": 0.5114163498644834,
|
15 |
+
"eval_f1_micro": 0.6258493771234429,
|
16 |
+
"eval_loss": 0.22310702502727509,
|
17 |
+
"eval_roc_auc": 0.7462668565240048,
|
18 |
+
"eval_runtime": 1321.1574,
|
19 |
+
"eval_samples_per_second": 2.119,
|
20 |
+
"eval_steps_per_second": 0.067,
|
21 |
+
"learning_rate": 0.01,
|
22 |
+
"step": 268
|
23 |
+
},
|
24 |
+
{
|
25 |
+
"epoch": 1.87,
|
26 |
+
"learning_rate": 0.01,
|
27 |
+
"loss": 0.2273,
|
28 |
+
"step": 500
|
29 |
+
},
|
30 |
+
{
|
31 |
+
"epoch": 2.0,
|
32 |
+
"eval_accuracy": 0.38120757413361916,
|
33 |
+
"eval_f1_macro": 0.45109440187622923,
|
34 |
+
"eval_f1_micro": 0.6105779637780503,
|
35 |
+
"eval_loss": 0.25052765011787415,
|
36 |
+
"eval_roc_auc": 0.7359905994056075,
|
37 |
+
"eval_runtime": 1321.0391,
|
38 |
+
"eval_samples_per_second": 2.119,
|
39 |
+
"eval_steps_per_second": 0.067,
|
40 |
+
"learning_rate": 0.01,
|
41 |
+
"step": 536
|
42 |
+
},
|
43 |
+
{
|
44 |
+
"epoch": 3.0,
|
45 |
+
"eval_accuracy": 0.41764916041443373,
|
46 |
+
"eval_f1_macro": 0.6951782991070653,
|
47 |
+
"eval_f1_micro": 0.7531037857011804,
|
48 |
+
"eval_loss": 0.17819169163703918,
|
49 |
+
"eval_roc_auc": 0.8424985182786072,
|
50 |
+
"eval_runtime": 1324.6672,
|
51 |
+
"eval_samples_per_second": 2.113,
|
52 |
+
"eval_steps_per_second": 0.066,
|
53 |
+
"learning_rate": 0.01,
|
54 |
+
"step": 804
|
55 |
+
},
|
56 |
+
{
|
57 |
+
"epoch": 3.73,
|
58 |
+
"learning_rate": 0.01,
|
59 |
+
"loss": 0.196,
|
60 |
+
"step": 1000
|
61 |
+
},
|
62 |
+
{
|
63 |
+
"epoch": 4.0,
|
64 |
+
"eval_accuracy": 0.4240800285816363,
|
65 |
+
"eval_f1_macro": 0.6667105102591024,
|
66 |
+
"eval_f1_micro": 0.7645973453961074,
|
67 |
+
"eval_loss": 0.15775558352470398,
|
68 |
+
"eval_roc_auc": 0.8561540860224954,
|
69 |
+
"eval_runtime": 1324.5164,
|
70 |
+
"eval_samples_per_second": 2.113,
|
71 |
+
"eval_steps_per_second": 0.066,
|
72 |
+
"learning_rate": 0.01,
|
73 |
+
"step": 1072
|
74 |
+
},
|
75 |
+
{
|
76 |
+
"epoch": 5.0,
|
77 |
+
"eval_accuracy": 0.3551268310110754,
|
78 |
+
"eval_f1_macro": 0.6462786847563172,
|
79 |
+
"eval_f1_micro": 0.7289891034631567,
|
80 |
+
"eval_loss": 0.1978054791688919,
|
81 |
+
"eval_roc_auc": 0.8616257579280331,
|
82 |
+
"eval_runtime": 1318.8019,
|
83 |
+
"eval_samples_per_second": 2.122,
|
84 |
+
"eval_steps_per_second": 0.067,
|
85 |
+
"learning_rate": 0.01,
|
86 |
+
"step": 1340
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 5.6,
|
90 |
+
"learning_rate": 0.01,
|
91 |
+
"loss": 0.1916,
|
92 |
+
"step": 1500
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 6.0,
|
96 |
+
"eval_accuracy": 0.45480528760271527,
|
97 |
+
"eval_f1_macro": 0.6154722345590861,
|
98 |
+
"eval_f1_micro": 0.7533968253968253,
|
99 |
+
"eval_loss": 0.1570044308900833,
|
100 |
+
"eval_roc_auc": 0.8332111249225472,
|
101 |
+
"eval_runtime": 1322.0736,
|
102 |
+
"eval_samples_per_second": 2.117,
|
103 |
+
"eval_steps_per_second": 0.067,
|
104 |
+
"learning_rate": 0.01,
|
105 |
+
"step": 1608
|
106 |
+
},
|
107 |
+
{
|
108 |
+
"epoch": 7.0,
|
109 |
+
"eval_accuracy": 0.4076455877098964,
|
110 |
+
"eval_f1_macro": 0.7033703129441511,
|
111 |
+
"eval_f1_micro": 0.7711244893571275,
|
112 |
+
"eval_loss": 0.17040091753005981,
|
113 |
+
"eval_roc_auc": 0.8893350315039463,
|
114 |
+
"eval_runtime": 1321.072,
|
115 |
+
"eval_samples_per_second": 2.119,
|
116 |
+
"eval_steps_per_second": 0.067,
|
117 |
+
"learning_rate": 0.01,
|
118 |
+
"step": 1876
|
119 |
+
},
|
120 |
+
{
|
121 |
+
"epoch": 7.46,
|
122 |
+
"learning_rate": 0.01,
|
123 |
+
"loss": 0.1935,
|
124 |
+
"step": 2000
|
125 |
+
},
|
126 |
+
{
|
127 |
+
"epoch": 8.0,
|
128 |
+
"eval_accuracy": 0.44873168988924617,
|
129 |
+
"eval_f1_macro": 0.7240467456541528,
|
130 |
+
"eval_f1_micro": 0.7782660807031561,
|
131 |
+
"eval_loss": 0.1583939492702484,
|
132 |
+
"eval_roc_auc": 0.87591640182751,
|
133 |
+
"eval_runtime": 1324.3647,
|
134 |
+
"eval_samples_per_second": 2.113,
|
135 |
+
"eval_steps_per_second": 0.066,
|
136 |
+
"learning_rate": 0.01,
|
137 |
+
"step": 2144
|
138 |
+
},
|
139 |
+
{
|
140 |
+
"epoch": 9.0,
|
141 |
+
"eval_accuracy": 0.44337263308324404,
|
142 |
+
"eval_f1_macro": 0.7026377863872897,
|
143 |
+
"eval_f1_micro": 0.7724530831099197,
|
144 |
+
"eval_loss": 0.16136306524276733,
|
145 |
+
"eval_roc_auc": 0.878715414884403,
|
146 |
+
"eval_runtime": 1334.0015,
|
147 |
+
"eval_samples_per_second": 2.098,
|
148 |
+
"eval_steps_per_second": 0.066,
|
149 |
+
"learning_rate": 0.01,
|
150 |
+
"step": 2412
|
151 |
+
},
|
152 |
+
{
|
153 |
+
"epoch": 9.33,
|
154 |
+
"learning_rate": 0.01,
|
155 |
+
"loss": 0.1945,
|
156 |
+
"step": 2500
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 10.0,
|
160 |
+
"eval_accuracy": 0.43658449446230796,
|
161 |
+
"eval_f1_macro": 0.6244582432023046,
|
162 |
+
"eval_f1_micro": 0.743831546311846,
|
163 |
+
"eval_loss": 0.1569277048110962,
|
164 |
+
"eval_roc_auc": 0.8239477455598103,
|
165 |
+
"eval_runtime": 1341.7823,
|
166 |
+
"eval_samples_per_second": 2.086,
|
167 |
+
"eval_steps_per_second": 0.066,
|
168 |
+
"learning_rate": 0.01,
|
169 |
+
"step": 2680
|
170 |
+
},
|
171 |
+
{
|
172 |
+
"epoch": 11.0,
|
173 |
+
"eval_accuracy": 0.42979635584137194,
|
174 |
+
"eval_f1_macro": 0.6986025660751581,
|
175 |
+
"eval_f1_micro": 0.763875932835821,
|
176 |
+
"eval_loss": 0.1665951907634735,
|
177 |
+
"eval_roc_auc": 0.8613857432852049,
|
178 |
+
"eval_runtime": 1333.2352,
|
179 |
+
"eval_samples_per_second": 2.099,
|
180 |
+
"eval_steps_per_second": 0.066,
|
181 |
+
"learning_rate": 0.01,
|
182 |
+
"step": 2948
|
183 |
+
},
|
184 |
+
{
|
185 |
+
"epoch": 11.19,
|
186 |
+
"learning_rate": 0.01,
|
187 |
+
"loss": 0.1951,
|
188 |
+
"step": 3000
|
189 |
+
},
|
190 |
+
{
|
191 |
+
"epoch": 12.0,
|
192 |
+
"eval_accuracy": 0.44765987852804573,
|
193 |
+
"eval_f1_macro": 0.6291100964373895,
|
194 |
+
"eval_f1_micro": 0.744811534234176,
|
195 |
+
"eval_loss": 0.15847544372081757,
|
196 |
+
"eval_roc_auc": 0.8242335503635296,
|
197 |
+
"eval_runtime": 1336.9094,
|
198 |
+
"eval_samples_per_second": 2.094,
|
199 |
+
"eval_steps_per_second": 0.066,
|
200 |
+
"learning_rate": 0.01,
|
201 |
+
"step": 3216
|
202 |
+
},
|
203 |
+
{
|
204 |
+
"epoch": 13.0,
|
205 |
+
"eval_accuracy": 0.43801357627724186,
|
206 |
+
"eval_f1_macro": 0.6650314316976844,
|
207 |
+
"eval_f1_micro": 0.7623590633130962,
|
208 |
+
"eval_loss": 0.15646952390670776,
|
209 |
+
"eval_roc_auc": 0.844307766178341,
|
210 |
+
"eval_runtime": 1326.5168,
|
211 |
+
"eval_samples_per_second": 2.11,
|
212 |
+
"eval_steps_per_second": 0.066,
|
213 |
+
"learning_rate": 0.01,
|
214 |
+
"step": 3484
|
215 |
+
},
|
216 |
+
{
|
217 |
+
"epoch": 13.06,
|
218 |
+
"learning_rate": 0.01,
|
219 |
+
"loss": 0.1953,
|
220 |
+
"step": 3500
|
221 |
+
},
|
222 |
+
{
|
223 |
+
"epoch": 14.0,
|
224 |
+
"eval_accuracy": 0.4465880671668453,
|
225 |
+
"eval_f1_macro": 0.5021716379213181,
|
226 |
+
"eval_f1_micro": 0.6728278270994033,
|
227 |
+
"eval_loss": 0.17276929318904877,
|
228 |
+
"eval_roc_auc": 0.7638863294586612,
|
229 |
+
"eval_runtime": 1334.5311,
|
230 |
+
"eval_samples_per_second": 2.097,
|
231 |
+
"eval_steps_per_second": 0.066,
|
232 |
+
"learning_rate": 0.01,
|
233 |
+
"step": 3752
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 14.93,
|
237 |
+
"learning_rate": 0.01,
|
238 |
+
"loss": 0.1945,
|
239 |
+
"step": 4000
|
240 |
+
},
|
241 |
+
{
|
242 |
+
"epoch": 15.0,
|
243 |
+
"eval_accuracy": 0.4555198285101822,
|
244 |
+
"eval_f1_macro": 0.6523865470596933,
|
245 |
+
"eval_f1_micro": 0.7441088615997346,
|
246 |
+
"eval_loss": 0.15650007128715515,
|
247 |
+
"eval_roc_auc": 0.8176688310877148,
|
248 |
+
"eval_runtime": 1329.7371,
|
249 |
+
"eval_samples_per_second": 2.105,
|
250 |
+
"eval_steps_per_second": 0.066,
|
251 |
+
"learning_rate": 0.01,
|
252 |
+
"step": 4020
|
253 |
+
},
|
254 |
+
{
|
255 |
+
"epoch": 16.0,
|
256 |
+
"eval_accuracy": 0.4580207216863165,
|
257 |
+
"eval_f1_macro": 0.643940672540041,
|
258 |
+
"eval_f1_micro": 0.7514664626370825,
|
259 |
+
"eval_loss": 0.15759296715259552,
|
260 |
+
"eval_roc_auc": 0.8311410282106092,
|
261 |
+
"eval_runtime": 1337.597,
|
262 |
+
"eval_samples_per_second": 2.093,
|
263 |
+
"eval_steps_per_second": 0.066,
|
264 |
+
"learning_rate": 0.01,
|
265 |
+
"step": 4288
|
266 |
+
},
|
267 |
+
{
|
268 |
+
"epoch": 16.79,
|
269 |
+
"learning_rate": 0.01,
|
270 |
+
"loss": 0.1929,
|
271 |
+
"step": 4500
|
272 |
+
},
|
273 |
+
{
|
274 |
+
"epoch": 17.0,
|
275 |
+
"eval_accuracy": 0.43122543765630583,
|
276 |
+
"eval_f1_macro": 0.5707187678300992,
|
277 |
+
"eval_f1_micro": 0.7358732093873818,
|
278 |
+
"eval_loss": 0.17005115747451782,
|
279 |
+
"eval_roc_auc": 0.833709678625855,
|
280 |
+
"eval_runtime": 1326.7318,
|
281 |
+
"eval_samples_per_second": 2.11,
|
282 |
+
"eval_steps_per_second": 0.066,
|
283 |
+
"learning_rate": 0.01,
|
284 |
+
"step": 4556
|
285 |
+
},
|
286 |
+
{
|
287 |
+
"epoch": 18.0,
|
288 |
+
"eval_accuracy": 0.42300821722043586,
|
289 |
+
"eval_f1_macro": 0.6533638646137062,
|
290 |
+
"eval_f1_micro": 0.7531492248062015,
|
291 |
+
"eval_loss": 0.15990640223026276,
|
292 |
+
"eval_roc_auc": 0.8450542243393845,
|
293 |
+
"eval_runtime": 1329.7371,
|
294 |
+
"eval_samples_per_second": 2.105,
|
295 |
+
"eval_steps_per_second": 0.066,
|
296 |
+
"learning_rate": 0.01,
|
297 |
+
"step": 4824
|
298 |
+
},
|
299 |
+
{
|
300 |
+
"epoch": 18.66,
|
301 |
+
"learning_rate": 0.01,
|
302 |
+
"loss": 0.1952,
|
303 |
+
"step": 5000
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 19.0,
|
307 |
+
"eval_accuracy": 0.45480528760271527,
|
308 |
+
"eval_f1_macro": 0.6657705432619248,
|
309 |
+
"eval_f1_micro": 0.7347428800106717,
|
310 |
+
"eval_loss": 0.16034173965454102,
|
311 |
+
"eval_roc_auc": 0.8117717147474299,
|
312 |
+
"eval_runtime": 1328.0617,
|
313 |
+
"eval_samples_per_second": 2.108,
|
314 |
+
"eval_steps_per_second": 0.066,
|
315 |
+
"learning_rate": 0.01,
|
316 |
+
"step": 5092
|
317 |
+
},
|
318 |
+
{
|
319 |
+
"epoch": 20.0,
|
320 |
+
"eval_accuracy": 0.5262593783494105,
|
321 |
+
"eval_f1_macro": 0.7676828961065644,
|
322 |
+
"eval_f1_micro": 0.8133643372904173,
|
323 |
+
"eval_loss": 0.1276090443134308,
|
324 |
+
"eval_roc_auc": 0.8759327417838362,
|
325 |
+
"eval_runtime": 1330.6724,
|
326 |
+
"eval_samples_per_second": 2.103,
|
327 |
+
"eval_steps_per_second": 0.066,
|
328 |
+
"learning_rate": 0.001,
|
329 |
+
"step": 5360
|
330 |
+
},
|
331 |
+
{
|
332 |
+
"epoch": 20.0,
|
333 |
+
"learning_rate": 0.001,
|
334 |
+
"step": 5360,
|
335 |
+
"total_flos": 9.454976615476292e+20,
|
336 |
+
"train_loss": 0.07645123254007367,
|
337 |
+
"train_runtime": 44768.9701,
|
338 |
+
"train_samples_per_second": 3.828,
|
339 |
+
"train_steps_per_second": 0.12
|
340 |
+
}
|
341 |
+
],
|
342 |
+
"logging_steps": 500,
|
343 |
+
"max_steps": 5360,
|
344 |
+
"num_train_epochs": 20,
|
345 |
+
"save_steps": 500,
|
346 |
+
"total_flos": 9.454976615476292e+20,
|
347 |
+
"trial_name": null,
|
348 |
+
"trial_params": null
|
349 |
+
}
|