Upload with huggingface_hub
Browse files- config.json +3 -39
- optimizer.pt +2 -2
- pytorch_model.bin +2 -2
- rng_state.pth +1 -1
- tokenizer.json +1 -6
- trainer_state.json +68 -248
- training_args.bin +1 -1
config.json
CHANGED
@@ -10,51 +10,15 @@
|
|
10 |
"hidden_size": 768,
|
11 |
"id2label": {
|
12 |
"0": "O",
|
13 |
-
"1": "B-
|
14 |
-
"2": "I-
|
15 |
-
"3": "B-mahalle",
|
16 |
-
"4": "I-mahalle",
|
17 |
-
"5": "B-ad-soyad",
|
18 |
-
"6": "I-ad-soyad",
|
19 |
-
"7": "B-kat",
|
20 |
-
"8": "I-kat",
|
21 |
-
"9": "B-ilce",
|
22 |
-
"10": "I-ilce",
|
23 |
-
"11": "B-sokak",
|
24 |
-
"12": "I-sokak",
|
25 |
-
"13": "B-dis kapi no",
|
26 |
-
"14": "I-dis kapi no",
|
27 |
-
"15": "B-Apartman/Site",
|
28 |
-
"16": "I-Apartman/Site",
|
29 |
-
"17": "B-ic kapi no",
|
30 |
-
"18": "I-ic kapi no",
|
31 |
-
"19": "B-il",
|
32 |
-
"20": "I-il"
|
33 |
},
|
34 |
"initializer_range": 0.02,
|
35 |
"intermediate_size": 3072,
|
36 |
"label2id": {
|
37 |
"LABEL_0": 0,
|
38 |
"LABEL_1": 1,
|
39 |
-
"
|
40 |
-
"LABEL_11": 11,
|
41 |
-
"LABEL_12": 12,
|
42 |
-
"LABEL_13": 13,
|
43 |
-
"LABEL_14": 14,
|
44 |
-
"LABEL_15": 15,
|
45 |
-
"LABEL_16": 16,
|
46 |
-
"LABEL_17": 17,
|
47 |
-
"LABEL_18": 18,
|
48 |
-
"LABEL_19": 19,
|
49 |
-
"LABEL_2": 2,
|
50 |
-
"LABEL_20": 20,
|
51 |
-
"LABEL_3": 3,
|
52 |
-
"LABEL_4": 4,
|
53 |
-
"LABEL_5": 5,
|
54 |
-
"LABEL_6": 6,
|
55 |
-
"LABEL_7": 7,
|
56 |
-
"LABEL_8": 8,
|
57 |
-
"LABEL_9": 9
|
58 |
},
|
59 |
"layer_norm_eps": 1e-12,
|
60 |
"max_position_embeddings": 512,
|
|
|
10 |
"hidden_size": 768,
|
11 |
"id2label": {
|
12 |
"0": "O",
|
13 |
+
"1": "B-ad",
|
14 |
+
"2": "I-ad"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15 |
},
|
16 |
"initializer_range": 0.02,
|
17 |
"intermediate_size": 3072,
|
18 |
"label2id": {
|
19 |
"LABEL_0": 0,
|
20 |
"LABEL_1": 1,
|
21 |
+
"LABEL_2": 2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
22 |
},
|
23 |
"layer_norm_eps": 1e-12,
|
24 |
"max_position_embeddings": 512,
|
optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:924bd7591d0c099bd620b921883035bbc7927ab04e165d077e35edc059e2e614
|
3 |
+
size 880347685
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1593ece4347bde1e59c22e58361992517553ef1f6541719eaecc4be565972006
|
3 |
+
size 440186673
|
rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14503
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0a3d7a5d3a3c1c5ea48da76f85e4f8796f66078cbf70ecb50b807da5d9008fba
|
3 |
size 14503
|
tokenizer.json
CHANGED
@@ -1,11 +1,6 @@
|
|
1 |
{
|
2 |
"version": "1.0",
|
3 |
-
"truncation":
|
4 |
-
"direction": "Right",
|
5 |
-
"max_length": 512,
|
6 |
-
"strategy": "LongestFirst",
|
7 |
-
"stride": 0
|
8 |
-
},
|
9 |
"padding": null,
|
10 |
"added_tokens": [
|
11 |
{
|
|
|
1 |
{
|
2 |
"version": "1.0",
|
3 |
+
"truncation": null,
|
|
|
|
|
|
|
|
|
|
|
4 |
"padding": null,
|
5 |
"added_tokens": [
|
6 |
{
|
trainer_state.json
CHANGED
@@ -1,6 +1,6 @@
|
|
1 |
{
|
2 |
-
"best_metric": 0.
|
3 |
-
"best_model_checkpoint": "/mounts/work/akoksal/
|
4 |
"epoch": 5.0,
|
5 |
"global_step": 240,
|
6 |
"is_hyper_param_search": false,
|
@@ -10,297 +10,117 @@
|
|
10 |
{
|
11 |
"epoch": 1.0,
|
12 |
"learning_rate": 4.4444444444444447e-05,
|
13 |
-
"loss":
|
14 |
"step": 48
|
15 |
},
|
16 |
{
|
17 |
"epoch": 1.0,
|
18 |
-
"
|
19 |
-
"
|
20 |
-
"
|
21 |
-
"
|
22 |
-
"
|
23 |
-
"
|
24 |
-
"
|
25 |
-
"
|
26 |
-
"
|
27 |
-
"
|
28 |
-
"
|
29 |
-
"
|
30 |
-
"eval_ic kapi no_f1": 0.0,
|
31 |
-
"eval_ic kapi no_precision": 0.0,
|
32 |
-
"eval_ic kapi no_recall": 0.0,
|
33 |
-
"eval_ic kapi no_support": 10,
|
34 |
-
"eval_il_f1": 0.7532467532467533,
|
35 |
-
"eval_il_precision": 0.696,
|
36 |
-
"eval_il_recall": 0.8207547169811321,
|
37 |
-
"eval_il_support": 106,
|
38 |
-
"eval_ilce_f1": 0.5732484076433121,
|
39 |
-
"eval_ilce_precision": 0.6164383561643836,
|
40 |
-
"eval_ilce_recall": 0.5357142857142857,
|
41 |
-
"eval_ilce_support": 84,
|
42 |
-
"eval_kat_f1": 0.0,
|
43 |
-
"eval_kat_precision": 0.0,
|
44 |
-
"eval_kat_recall": 0.0,
|
45 |
-
"eval_kat_support": 13,
|
46 |
-
"eval_loss": 0.34245046973228455,
|
47 |
-
"eval_mahalle_f1": 0.6117647058823529,
|
48 |
-
"eval_mahalle_precision": 0.5454545454545454,
|
49 |
-
"eval_mahalle_recall": 0.6964285714285714,
|
50 |
-
"eval_mahalle_support": 112,
|
51 |
-
"eval_no_f1": 0.33557046979865773,
|
52 |
-
"eval_no_precision": 0.2336448598130841,
|
53 |
-
"eval_no_recall": 0.5952380952380952,
|
54 |
-
"eval_no_support": 42,
|
55 |
-
"eval_overall_accuracy": 0.9031231350706187,
|
56 |
-
"eval_overall_f1": 0.5896103896103896,
|
57 |
-
"eval_overall_precision": 0.530373831775701,
|
58 |
-
"eval_overall_recall": 0.6637426900584795,
|
59 |
-
"eval_runtime": 0.5655,
|
60 |
-
"eval_samples_per_second": 339.518,
|
61 |
-
"eval_sokak_f1": 0.5461538461538461,
|
62 |
-
"eval_sokak_precision": 0.47651006711409394,
|
63 |
-
"eval_sokak_recall": 0.6396396396396397,
|
64 |
-
"eval_sokak_support": 111,
|
65 |
-
"eval_steps_per_second": 10.61,
|
66 |
"step": 48
|
67 |
},
|
68 |
{
|
69 |
"epoch": 2.0,
|
70 |
"learning_rate": 3.3333333333333335e-05,
|
71 |
-
"loss": 0.
|
72 |
"step": 96
|
73 |
},
|
74 |
{
|
75 |
"epoch": 2.0,
|
76 |
-
"
|
77 |
-
"
|
78 |
-
"
|
79 |
-
"
|
80 |
-
"
|
81 |
-
"
|
82 |
-
"
|
83 |
-
"
|
84 |
-
"
|
85 |
-
"
|
86 |
-
"
|
87 |
-
"
|
88 |
-
"eval_ic kapi no_f1": 0.0,
|
89 |
-
"eval_ic kapi no_precision": 0.0,
|
90 |
-
"eval_ic kapi no_recall": 0.0,
|
91 |
-
"eval_ic kapi no_support": 10,
|
92 |
-
"eval_il_f1": 0.8622222222222222,
|
93 |
-
"eval_il_precision": 0.8151260504201681,
|
94 |
-
"eval_il_recall": 0.9150943396226415,
|
95 |
-
"eval_il_support": 106,
|
96 |
-
"eval_ilce_f1": 0.819672131147541,
|
97 |
-
"eval_ilce_precision": 0.7575757575757576,
|
98 |
-
"eval_ilce_recall": 0.8928571428571429,
|
99 |
-
"eval_ilce_support": 84,
|
100 |
-
"eval_kat_f1": 0.3157894736842105,
|
101 |
-
"eval_kat_precision": 0.24,
|
102 |
-
"eval_kat_recall": 0.46153846153846156,
|
103 |
-
"eval_kat_support": 13,
|
104 |
-
"eval_loss": 0.28523704409599304,
|
105 |
-
"eval_mahalle_f1": 0.6302521008403361,
|
106 |
-
"eval_mahalle_precision": 0.5952380952380952,
|
107 |
-
"eval_mahalle_recall": 0.6696428571428571,
|
108 |
-
"eval_mahalle_support": 112,
|
109 |
-
"eval_no_f1": 0.4509803921568628,
|
110 |
-
"eval_no_precision": 0.38333333333333336,
|
111 |
-
"eval_no_recall": 0.5476190476190477,
|
112 |
-
"eval_no_support": 42,
|
113 |
-
"eval_overall_accuracy": 0.9154565347125522,
|
114 |
-
"eval_overall_f1": 0.68733153638814,
|
115 |
-
"eval_overall_precision": 0.6375,
|
116 |
-
"eval_overall_recall": 0.7456140350877193,
|
117 |
-
"eval_runtime": 0.5594,
|
118 |
-
"eval_samples_per_second": 343.207,
|
119 |
-
"eval_sokak_f1": 0.6382978723404256,
|
120 |
-
"eval_sokak_precision": 0.6048387096774194,
|
121 |
-
"eval_sokak_recall": 0.6756756756756757,
|
122 |
-
"eval_sokak_support": 111,
|
123 |
-
"eval_steps_per_second": 10.725,
|
124 |
"step": 96
|
125 |
},
|
126 |
{
|
127 |
"epoch": 3.0,
|
128 |
"learning_rate": 2.2222222222222223e-05,
|
129 |
-
"loss": 0.
|
130 |
"step": 144
|
131 |
},
|
132 |
{
|
133 |
"epoch": 3.0,
|
134 |
-
"
|
135 |
-
"
|
136 |
-
"
|
137 |
-
"
|
138 |
-
"
|
139 |
-
"
|
140 |
-
"
|
141 |
-
"
|
142 |
-
"
|
143 |
-
"
|
144 |
-
"
|
145 |
-
"
|
146 |
-
"eval_ic kapi no_f1": 0.26666666666666666,
|
147 |
-
"eval_ic kapi no_precision": 0.4,
|
148 |
-
"eval_ic kapi no_recall": 0.2,
|
149 |
-
"eval_ic kapi no_support": 10,
|
150 |
-
"eval_il_f1": 0.9245283018867925,
|
151 |
-
"eval_il_precision": 0.9245283018867925,
|
152 |
-
"eval_il_recall": 0.9245283018867925,
|
153 |
-
"eval_il_support": 106,
|
154 |
-
"eval_ilce_f1": 0.8222222222222222,
|
155 |
-
"eval_ilce_precision": 0.7708333333333334,
|
156 |
-
"eval_ilce_recall": 0.8809523809523809,
|
157 |
-
"eval_ilce_support": 84,
|
158 |
-
"eval_kat_f1": 0.47058823529411764,
|
159 |
-
"eval_kat_precision": 0.38095238095238093,
|
160 |
-
"eval_kat_recall": 0.6153846153846154,
|
161 |
-
"eval_kat_support": 13,
|
162 |
-
"eval_loss": 0.25738969445228577,
|
163 |
-
"eval_mahalle_f1": 0.6297872340425532,
|
164 |
-
"eval_mahalle_precision": 0.6016260162601627,
|
165 |
-
"eval_mahalle_recall": 0.6607142857142857,
|
166 |
-
"eval_mahalle_support": 112,
|
167 |
-
"eval_no_f1": 0.37606837606837606,
|
168 |
-
"eval_no_precision": 0.29333333333333333,
|
169 |
-
"eval_no_recall": 0.5238095238095238,
|
170 |
-
"eval_no_support": 42,
|
171 |
-
"eval_overall_accuracy": 0.9242092699423116,
|
172 |
-
"eval_overall_f1": 0.7053206002728514,
|
173 |
-
"eval_overall_precision": 0.6611253196930946,
|
174 |
-
"eval_overall_recall": 0.7558479532163743,
|
175 |
-
"eval_runtime": 0.5539,
|
176 |
-
"eval_samples_per_second": 346.644,
|
177 |
-
"eval_sokak_f1": 0.6978723404255319,
|
178 |
-
"eval_sokak_precision": 0.6612903225806451,
|
179 |
-
"eval_sokak_recall": 0.7387387387387387,
|
180 |
-
"eval_sokak_support": 111,
|
181 |
-
"eval_steps_per_second": 10.833,
|
182 |
"step": 144
|
183 |
},
|
184 |
{
|
185 |
"epoch": 4.0,
|
186 |
"learning_rate": 1.1111111111111112e-05,
|
187 |
-
"loss": 0.
|
188 |
"step": 192
|
189 |
},
|
190 |
{
|
191 |
"epoch": 4.0,
|
192 |
-
"
|
193 |
-
"
|
194 |
-
"
|
195 |
-
"
|
196 |
-
"
|
197 |
-
"
|
198 |
-
"
|
199 |
-
"
|
200 |
-
"
|
201 |
-
"
|
202 |
-
"
|
203 |
-
"
|
204 |
-
"eval_ic kapi no_f1": 0.1818181818181818,
|
205 |
-
"eval_ic kapi no_precision": 0.16666666666666666,
|
206 |
-
"eval_ic kapi no_recall": 0.2,
|
207 |
-
"eval_ic kapi no_support": 10,
|
208 |
-
"eval_il_f1": 0.9194312796208531,
|
209 |
-
"eval_il_precision": 0.9238095238095239,
|
210 |
-
"eval_il_recall": 0.9150943396226415,
|
211 |
-
"eval_il_support": 106,
|
212 |
-
"eval_ilce_f1": 0.8287292817679559,
|
213 |
-
"eval_ilce_precision": 0.7731958762886598,
|
214 |
-
"eval_ilce_recall": 0.8928571428571429,
|
215 |
-
"eval_ilce_support": 84,
|
216 |
-
"eval_kat_f1": 0.4242424242424242,
|
217 |
-
"eval_kat_precision": 0.35,
|
218 |
-
"eval_kat_recall": 0.5384615384615384,
|
219 |
-
"eval_kat_support": 13,
|
220 |
-
"eval_loss": 0.252394437789917,
|
221 |
-
"eval_mahalle_f1": 0.6666666666666667,
|
222 |
-
"eval_mahalle_precision": 0.6470588235294118,
|
223 |
-
"eval_mahalle_recall": 0.6875,
|
224 |
-
"eval_mahalle_support": 112,
|
225 |
-
"eval_no_f1": 0.4571428571428571,
|
226 |
-
"eval_no_precision": 0.38095238095238093,
|
227 |
-
"eval_no_recall": 0.5714285714285714,
|
228 |
-
"eval_no_support": 42,
|
229 |
-
"eval_overall_accuracy": 0.926795305351104,
|
230 |
-
"eval_overall_f1": 0.708825573314802,
|
231 |
-
"eval_overall_precision": 0.6754966887417219,
|
232 |
-
"eval_overall_recall": 0.7456140350877193,
|
233 |
-
"eval_runtime": 0.5605,
|
234 |
-
"eval_samples_per_second": 342.58,
|
235 |
-
"eval_sokak_f1": 0.6233766233766234,
|
236 |
-
"eval_sokak_precision": 0.6,
|
237 |
-
"eval_sokak_recall": 0.6486486486486487,
|
238 |
-
"eval_sokak_support": 111,
|
239 |
-
"eval_steps_per_second": 10.706,
|
240 |
"step": 192
|
241 |
},
|
242 |
{
|
243 |
"epoch": 5.0,
|
244 |
"learning_rate": 0.0,
|
245 |
-
"loss": 0.
|
246 |
"step": 240
|
247 |
},
|
248 |
{
|
249 |
"epoch": 5.0,
|
250 |
-
"
|
251 |
-
"
|
252 |
-
"
|
253 |
-
"
|
254 |
-
"
|
255 |
-
"
|
256 |
-
"
|
257 |
-
"
|
258 |
-
"
|
259 |
-
"
|
260 |
-
"
|
261 |
-
"
|
262 |
-
"eval_ic kapi no_f1": 0.2727272727272727,
|
263 |
-
"eval_ic kapi no_precision": 0.25,
|
264 |
-
"eval_ic kapi no_recall": 0.3,
|
265 |
-
"eval_ic kapi no_support": 10,
|
266 |
-
"eval_il_f1": 0.9142857142857143,
|
267 |
-
"eval_il_precision": 0.9230769230769231,
|
268 |
-
"eval_il_recall": 0.9056603773584906,
|
269 |
-
"eval_il_support": 106,
|
270 |
-
"eval_ilce_f1": 0.8379888268156424,
|
271 |
-
"eval_ilce_precision": 0.7894736842105263,
|
272 |
-
"eval_ilce_recall": 0.8928571428571429,
|
273 |
-
"eval_ilce_support": 84,
|
274 |
-
"eval_kat_f1": 0.4571428571428572,
|
275 |
-
"eval_kat_precision": 0.36363636363636365,
|
276 |
-
"eval_kat_recall": 0.6153846153846154,
|
277 |
-
"eval_kat_support": 13,
|
278 |
-
"eval_loss": 0.26428258419036865,
|
279 |
-
"eval_mahalle_f1": 0.6324786324786325,
|
280 |
-
"eval_mahalle_precision": 0.6065573770491803,
|
281 |
-
"eval_mahalle_recall": 0.6607142857142857,
|
282 |
-
"eval_mahalle_support": 112,
|
283 |
-
"eval_no_f1": 0.4752475247524753,
|
284 |
-
"eval_no_precision": 0.4067796610169492,
|
285 |
-
"eval_no_recall": 0.5714285714285714,
|
286 |
-
"eval_no_support": 42,
|
287 |
-
"eval_overall_accuracy": 0.9273920827531331,
|
288 |
-
"eval_overall_f1": 0.7178770949720671,
|
289 |
-
"eval_overall_precision": 0.6871657754010695,
|
290 |
-
"eval_overall_recall": 0.7514619883040936,
|
291 |
-
"eval_runtime": 0.5624,
|
292 |
-
"eval_samples_per_second": 341.388,
|
293 |
-
"eval_sokak_f1": 0.689655172413793,
|
294 |
-
"eval_sokak_precision": 0.6611570247933884,
|
295 |
-
"eval_sokak_recall": 0.7207207207207207,
|
296 |
-
"eval_sokak_support": 111,
|
297 |
-
"eval_steps_per_second": 10.668,
|
298 |
"step": 240
|
299 |
}
|
300 |
],
|
301 |
"max_steps": 240,
|
302 |
"num_train_epochs": 5,
|
303 |
-
"total_flos":
|
304 |
"trial_name": null,
|
305 |
"trial_params": null
|
306 |
}
|
|
|
1 |
{
|
2 |
+
"best_metric": 0.03221152722835541,
|
3 |
+
"best_model_checkpoint": "/mounts/work/akoksal/earthquake_anon_models/checkpoint-96",
|
4 |
"epoch": 5.0,
|
5 |
"global_step": 240,
|
6 |
"is_hyper_param_search": false,
|
|
|
10 |
{
|
11 |
"epoch": 1.0,
|
12 |
"learning_rate": 4.4444444444444447e-05,
|
13 |
+
"loss": 0.301,
|
14 |
"step": 48
|
15 |
},
|
16 |
{
|
17 |
"epoch": 1.0,
|
18 |
+
"eval_ad_f1": 0.8044280442804429,
|
19 |
+
"eval_ad_precision": 0.7315436241610739,
|
20 |
+
"eval_ad_recall": 0.8934426229508197,
|
21 |
+
"eval_ad_support": 122,
|
22 |
+
"eval_loss": 0.03607584163546562,
|
23 |
+
"eval_overall_accuracy": 0.9892580067634772,
|
24 |
+
"eval_overall_f1": 0.8044280442804429,
|
25 |
+
"eval_overall_precision": 0.7315436241610739,
|
26 |
+
"eval_overall_recall": 0.8934426229508197,
|
27 |
+
"eval_runtime": 0.7431,
|
28 |
+
"eval_samples_per_second": 258.387,
|
29 |
+
"eval_steps_per_second": 8.075,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
30 |
"step": 48
|
31 |
},
|
32 |
{
|
33 |
"epoch": 2.0,
|
34 |
"learning_rate": 3.3333333333333335e-05,
|
35 |
+
"loss": 0.0274,
|
36 |
"step": 96
|
37 |
},
|
38 |
{
|
39 |
"epoch": 2.0,
|
40 |
+
"eval_ad_f1": 0.8286852589641434,
|
41 |
+
"eval_ad_precision": 0.8062015503875969,
|
42 |
+
"eval_ad_recall": 0.8524590163934426,
|
43 |
+
"eval_ad_support": 122,
|
44 |
+
"eval_loss": 0.03221152722835541,
|
45 |
+
"eval_overall_accuracy": 0.9910483389695643,
|
46 |
+
"eval_overall_f1": 0.8286852589641434,
|
47 |
+
"eval_overall_precision": 0.8062015503875969,
|
48 |
+
"eval_overall_recall": 0.8524590163934426,
|
49 |
+
"eval_runtime": 0.7362,
|
50 |
+
"eval_samples_per_second": 260.794,
|
51 |
+
"eval_steps_per_second": 8.15,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
52 |
"step": 96
|
53 |
},
|
54 |
{
|
55 |
"epoch": 3.0,
|
56 |
"learning_rate": 2.2222222222222223e-05,
|
57 |
+
"loss": 0.0145,
|
58 |
"step": 144
|
59 |
},
|
60 |
{
|
61 |
"epoch": 3.0,
|
62 |
+
"eval_ad_f1": 0.8537549407114624,
|
63 |
+
"eval_ad_precision": 0.8244274809160306,
|
64 |
+
"eval_ad_recall": 0.8852459016393442,
|
65 |
+
"eval_ad_support": 122,
|
66 |
+
"eval_loss": 0.03574167191982269,
|
67 |
+
"eval_overall_accuracy": 0.9916451163715934,
|
68 |
+
"eval_overall_f1": 0.8537549407114624,
|
69 |
+
"eval_overall_precision": 0.8244274809160306,
|
70 |
+
"eval_overall_recall": 0.8852459016393442,
|
71 |
+
"eval_runtime": 0.7344,
|
72 |
+
"eval_samples_per_second": 261.428,
|
73 |
+
"eval_steps_per_second": 8.17,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
74 |
"step": 144
|
75 |
},
|
76 |
{
|
77 |
"epoch": 4.0,
|
78 |
"learning_rate": 1.1111111111111112e-05,
|
79 |
+
"loss": 0.0101,
|
80 |
"step": 192
|
81 |
},
|
82 |
{
|
83 |
"epoch": 4.0,
|
84 |
+
"eval_ad_f1": 0.8582677165354332,
|
85 |
+
"eval_ad_precision": 0.8257575757575758,
|
86 |
+
"eval_ad_recall": 0.8934426229508197,
|
87 |
+
"eval_ad_support": 122,
|
88 |
+
"eval_loss": 0.03515293449163437,
|
89 |
+
"eval_overall_accuracy": 0.9920429679729461,
|
90 |
+
"eval_overall_f1": 0.8582677165354332,
|
91 |
+
"eval_overall_precision": 0.8257575757575758,
|
92 |
+
"eval_overall_recall": 0.8934426229508197,
|
93 |
+
"eval_runtime": 0.7354,
|
94 |
+
"eval_samples_per_second": 261.069,
|
95 |
+
"eval_steps_per_second": 8.158,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
96 |
"step": 192
|
97 |
},
|
98 |
{
|
99 |
"epoch": 5.0,
|
100 |
"learning_rate": 0.0,
|
101 |
+
"loss": 0.0065,
|
102 |
"step": 240
|
103 |
},
|
104 |
{
|
105 |
"epoch": 5.0,
|
106 |
+
"eval_ad_f1": 0.8582677165354332,
|
107 |
+
"eval_ad_precision": 0.8257575757575758,
|
108 |
+
"eval_ad_recall": 0.8934426229508197,
|
109 |
+
"eval_ad_support": 122,
|
110 |
+
"eval_loss": 0.03596705570816994,
|
111 |
+
"eval_overall_accuracy": 0.9922418937736225,
|
112 |
+
"eval_overall_f1": 0.8582677165354332,
|
113 |
+
"eval_overall_precision": 0.8257575757575758,
|
114 |
+
"eval_overall_recall": 0.8934426229508197,
|
115 |
+
"eval_runtime": 0.7358,
|
116 |
+
"eval_samples_per_second": 260.929,
|
117 |
+
"eval_steps_per_second": 8.154,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
118 |
"step": 240
|
119 |
}
|
120 |
],
|
121 |
"max_steps": 240,
|
122 |
"num_train_epochs": 5,
|
123 |
+
"total_flos": 178090660164636.0,
|
124 |
"trial_name": null,
|
125 |
"trial_params": null
|
126 |
}
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3503
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4d095bb03a50a602dd1a1d2499d385fd9440fbcad5a9689576891e2f47365632
|
3 |
size 3503
|