End of training
Browse files- README.md +5 -5
- all_results.json +19 -0
- egy_training_log.txt +2 -0
- eval_results.json +13 -0
- train_results.json +9 -0
- train_vs_val_loss.png +0 -0
- trainer_state.json +317 -0
README.md
CHANGED
@@ -17,11 +17,11 @@ should probably proofread and complete it, then remove this comment. -->
|
|
17 |
|
18 |
This model is a fine-tuned version of [aubmindlab/aragpt2-base](https://huggingface.co/aubmindlab/aragpt2-base) on an unknown dataset.
|
19 |
It achieves the following results on the evaluation set:
|
20 |
-
- Loss: 0.
|
21 |
-
- Bleu: 0.
|
22 |
-
- Rouge1: 0.
|
23 |
-
- Rouge2: 0.
|
24 |
-
- Rougel: 0.
|
25 |
|
26 |
## Model description
|
27 |
|
|
|
17 |
|
18 |
This model is a fine-tuned version of [aubmindlab/aragpt2-base](https://huggingface.co/aubmindlab/aragpt2-base) on an unknown dataset.
|
19 |
It achieves the following results on the evaluation set:
|
20 |
+
- Loss: 0.0534
|
21 |
+
- Bleu: 0.0428
|
22 |
+
- Rouge1: 0.3139
|
23 |
+
- Rouge2: 0.1104
|
24 |
+
- Rougel: 0.3097
|
25 |
|
26 |
## Model description
|
27 |
|
all_results.json
ADDED
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 14.0,
|
3 |
+
"eval_bleu": 0.04280893852047394,
|
4 |
+
"eval_loss": 0.05337703973054886,
|
5 |
+
"eval_rouge1": 0.31388693258006284,
|
6 |
+
"eval_rouge2": 0.11036995985896458,
|
7 |
+
"eval_rougeL": 0.3096668301159336,
|
8 |
+
"eval_runtime": 7.0173,
|
9 |
+
"eval_samples": 304,
|
10 |
+
"eval_samples_per_second": 43.322,
|
11 |
+
"eval_steps_per_second": 5.415,
|
12 |
+
"perplexity": 1.0548272819463649,
|
13 |
+
"total_flos": 8918419636224000.0,
|
14 |
+
"train_loss": 0.46412634738241815,
|
15 |
+
"train_runtime": 1276.9352,
|
16 |
+
"train_samples": 1219,
|
17 |
+
"train_samples_per_second": 19.093,
|
18 |
+
"train_steps_per_second": 2.396
|
19 |
+
}
|
egy_training_log.txt
CHANGED
@@ -304,3 +304,5 @@ INFO:root:Epoch 13.0: Train Loss = 0.0331, Eval Loss = 0.05399588495492935
|
|
304 |
INFO:absl:Using default tokenizer.
|
305 |
INFO:root:Epoch 14.0: Train Loss = 0.0308, Eval Loss = 0.05515788868069649
|
306 |
INFO:absl:Using default tokenizer.
|
|
|
|
|
|
304 |
INFO:absl:Using default tokenizer.
|
305 |
INFO:root:Epoch 14.0: Train Loss = 0.0308, Eval Loss = 0.05515788868069649
|
306 |
INFO:absl:Using default tokenizer.
|
307 |
+
INFO:__main__:*** Evaluate ***
|
308 |
+
INFO:absl:Using default tokenizer.
|
eval_results.json
ADDED
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 14.0,
|
3 |
+
"eval_bleu": 0.04280893852047394,
|
4 |
+
"eval_loss": 0.05337703973054886,
|
5 |
+
"eval_rouge1": 0.31388693258006284,
|
6 |
+
"eval_rouge2": 0.11036995985896458,
|
7 |
+
"eval_rougeL": 0.3096668301159336,
|
8 |
+
"eval_runtime": 7.0173,
|
9 |
+
"eval_samples": 304,
|
10 |
+
"eval_samples_per_second": 43.322,
|
11 |
+
"eval_steps_per_second": 5.415,
|
12 |
+
"perplexity": 1.0548272819463649
|
13 |
+
}
|
train_results.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 14.0,
|
3 |
+
"total_flos": 8918419636224000.0,
|
4 |
+
"train_loss": 0.46412634738241815,
|
5 |
+
"train_runtime": 1276.9352,
|
6 |
+
"train_samples": 1219,
|
7 |
+
"train_samples_per_second": 19.093,
|
8 |
+
"train_steps_per_second": 2.396
|
9 |
+
}
|
train_vs_val_loss.png
ADDED
trainer_state.json
ADDED
@@ -0,0 +1,317 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.05337703973054886,
|
3 |
+
"best_model_checkpoint": "/home/iais_marenpielka/Bouthaina/res_nw_yem_aragpt2-base/checkpoint-1377",
|
4 |
+
"epoch": 14.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 2142,
|
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 |
+
"grad_norm": 0.3274374306201935,
|
14 |
+
"learning_rate": 1.53e-05,
|
15 |
+
"loss": 5.8224,
|
16 |
+
"step": 153
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 1.0,
|
20 |
+
"eval_bleu": 0.00392224521343418,
|
21 |
+
"eval_loss": 0.1128806546330452,
|
22 |
+
"eval_rouge1": 0.07124894337258883,
|
23 |
+
"eval_rouge2": 0.0028587092731829577,
|
24 |
+
"eval_rougeL": 0.06948923152022689,
|
25 |
+
"eval_runtime": 7.1373,
|
26 |
+
"eval_samples_per_second": 42.593,
|
27 |
+
"eval_steps_per_second": 5.324,
|
28 |
+
"step": 153
|
29 |
+
},
|
30 |
+
{
|
31 |
+
"epoch": 2.0,
|
32 |
+
"grad_norm": 0.3334260880947113,
|
33 |
+
"learning_rate": 3.06e-05,
|
34 |
+
"loss": 0.1108,
|
35 |
+
"step": 306
|
36 |
+
},
|
37 |
+
{
|
38 |
+
"epoch": 2.0,
|
39 |
+
"eval_bleu": 0.0,
|
40 |
+
"eval_loss": 0.06912114471197128,
|
41 |
+
"eval_rouge1": 0.09507863230508208,
|
42 |
+
"eval_rouge2": 0.007788772376873063,
|
43 |
+
"eval_rougeL": 0.09317127790126112,
|
44 |
+
"eval_runtime": 6.9813,
|
45 |
+
"eval_samples_per_second": 43.545,
|
46 |
+
"eval_steps_per_second": 5.443,
|
47 |
+
"step": 306
|
48 |
+
},
|
49 |
+
{
|
50 |
+
"epoch": 3.0,
|
51 |
+
"grad_norm": 0.2704195976257324,
|
52 |
+
"learning_rate": 4.5900000000000004e-05,
|
53 |
+
"loss": 0.0775,
|
54 |
+
"step": 459
|
55 |
+
},
|
56 |
+
{
|
57 |
+
"epoch": 3.0,
|
58 |
+
"eval_bleu": 0.006737157818299724,
|
59 |
+
"eval_loss": 0.06276746839284897,
|
60 |
+
"eval_rouge1": 0.12907377035145118,
|
61 |
+
"eval_rouge2": 0.015683124441348124,
|
62 |
+
"eval_rougeL": 0.1286203798342947,
|
63 |
+
"eval_runtime": 6.9851,
|
64 |
+
"eval_samples_per_second": 43.521,
|
65 |
+
"eval_steps_per_second": 5.44,
|
66 |
+
"step": 459
|
67 |
+
},
|
68 |
+
{
|
69 |
+
"epoch": 4.0,
|
70 |
+
"grad_norm": 0.32800641655921936,
|
71 |
+
"learning_rate": 4.7812500000000003e-05,
|
72 |
+
"loss": 0.0678,
|
73 |
+
"step": 612
|
74 |
+
},
|
75 |
+
{
|
76 |
+
"epoch": 4.0,
|
77 |
+
"eval_bleu": 0.008597461521382226,
|
78 |
+
"eval_loss": 0.059174273163080215,
|
79 |
+
"eval_rouge1": 0.1524066357986381,
|
80 |
+
"eval_rouge2": 0.027263211934264564,
|
81 |
+
"eval_rougeL": 0.14917670820737453,
|
82 |
+
"eval_runtime": 6.9134,
|
83 |
+
"eval_samples_per_second": 43.973,
|
84 |
+
"eval_steps_per_second": 5.497,
|
85 |
+
"step": 612
|
86 |
+
},
|
87 |
+
{
|
88 |
+
"epoch": 5.0,
|
89 |
+
"grad_norm": 0.18473878502845764,
|
90 |
+
"learning_rate": 4.482421875e-05,
|
91 |
+
"loss": 0.0603,
|
92 |
+
"step": 765
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 5.0,
|
96 |
+
"eval_bleu": 0.016158903735540513,
|
97 |
+
"eval_loss": 0.05656920000910759,
|
98 |
+
"eval_rouge1": 0.19186830417118805,
|
99 |
+
"eval_rouge2": 0.04134465720083692,
|
100 |
+
"eval_rougeL": 0.18829939413002822,
|
101 |
+
"eval_runtime": 6.9842,
|
102 |
+
"eval_samples_per_second": 43.527,
|
103 |
+
"eval_steps_per_second": 5.441,
|
104 |
+
"step": 765
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 6.0,
|
108 |
+
"grad_norm": 0.2642894685268402,
|
109 |
+
"learning_rate": 4.18359375e-05,
|
110 |
+
"loss": 0.0547,
|
111 |
+
"step": 918
|
112 |
+
},
|
113 |
+
{
|
114 |
+
"epoch": 6.0,
|
115 |
+
"eval_bleu": 0.018741315914778903,
|
116 |
+
"eval_loss": 0.05464606359601021,
|
117 |
+
"eval_rouge1": 0.2238907288379966,
|
118 |
+
"eval_rouge2": 0.05989126658873175,
|
119 |
+
"eval_rougeL": 0.22184683032173025,
|
120 |
+
"eval_runtime": 7.0034,
|
121 |
+
"eval_samples_per_second": 43.408,
|
122 |
+
"eval_steps_per_second": 5.426,
|
123 |
+
"step": 918
|
124 |
+
},
|
125 |
+
{
|
126 |
+
"epoch": 7.0,
|
127 |
+
"grad_norm": 0.24372832477092743,
|
128 |
+
"learning_rate": 3.884765625e-05,
|
129 |
+
"loss": 0.0498,
|
130 |
+
"step": 1071
|
131 |
+
},
|
132 |
+
{
|
133 |
+
"epoch": 7.0,
|
134 |
+
"eval_bleu": 0.029468363597052377,
|
135 |
+
"eval_loss": 0.054019927978515625,
|
136 |
+
"eval_rouge1": 0.26839447635173574,
|
137 |
+
"eval_rouge2": 0.07325536872796896,
|
138 |
+
"eval_rougeL": 0.2638108570075386,
|
139 |
+
"eval_runtime": 6.9888,
|
140 |
+
"eval_samples_per_second": 43.498,
|
141 |
+
"eval_steps_per_second": 5.437,
|
142 |
+
"step": 1071
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 8.0,
|
146 |
+
"grad_norm": 0.2509661316871643,
|
147 |
+
"learning_rate": 3.5859375e-05,
|
148 |
+
"loss": 0.0456,
|
149 |
+
"step": 1224
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 8.0,
|
153 |
+
"eval_bleu": 0.029195772701040013,
|
154 |
+
"eval_loss": 0.053611643612384796,
|
155 |
+
"eval_rouge1": 0.2883971192764231,
|
156 |
+
"eval_rouge2": 0.08184845936724905,
|
157 |
+
"eval_rougeL": 0.28406572737215363,
|
158 |
+
"eval_runtime": 6.9871,
|
159 |
+
"eval_samples_per_second": 43.509,
|
160 |
+
"eval_steps_per_second": 5.439,
|
161 |
+
"step": 1224
|
162 |
+
},
|
163 |
+
{
|
164 |
+
"epoch": 9.0,
|
165 |
+
"grad_norm": 0.23149649798870087,
|
166 |
+
"learning_rate": 3.287109375e-05,
|
167 |
+
"loss": 0.0419,
|
168 |
+
"step": 1377
|
169 |
+
},
|
170 |
+
{
|
171 |
+
"epoch": 9.0,
|
172 |
+
"eval_bleu": 0.04280893852047394,
|
173 |
+
"eval_loss": 0.05337703973054886,
|
174 |
+
"eval_rouge1": 0.31388693258006284,
|
175 |
+
"eval_rouge2": 0.11036995985896458,
|
176 |
+
"eval_rougeL": 0.3096668301159336,
|
177 |
+
"eval_runtime": 6.9883,
|
178 |
+
"eval_samples_per_second": 43.501,
|
179 |
+
"eval_steps_per_second": 5.438,
|
180 |
+
"step": 1377
|
181 |
+
},
|
182 |
+
{
|
183 |
+
"epoch": 10.0,
|
184 |
+
"grad_norm": 0.2521679401397705,
|
185 |
+
"learning_rate": 2.9882812500000002e-05,
|
186 |
+
"loss": 0.0385,
|
187 |
+
"step": 1530
|
188 |
+
},
|
189 |
+
{
|
190 |
+
"epoch": 10.0,
|
191 |
+
"eval_bleu": 0.046113511822263795,
|
192 |
+
"eval_loss": 0.05344715714454651,
|
193 |
+
"eval_rouge1": 0.32550355167793193,
|
194 |
+
"eval_rouge2": 0.11175887195986847,
|
195 |
+
"eval_rougeL": 0.3185482802099305,
|
196 |
+
"eval_runtime": 6.9943,
|
197 |
+
"eval_samples_per_second": 43.464,
|
198 |
+
"eval_steps_per_second": 5.433,
|
199 |
+
"step": 1530
|
200 |
+
},
|
201 |
+
{
|
202 |
+
"epoch": 11.0,
|
203 |
+
"grad_norm": 0.23900777101516724,
|
204 |
+
"learning_rate": 2.689453125e-05,
|
205 |
+
"loss": 0.0354,
|
206 |
+
"step": 1683
|
207 |
+
},
|
208 |
+
{
|
209 |
+
"epoch": 11.0,
|
210 |
+
"eval_bleu": 0.047259792068284935,
|
211 |
+
"eval_loss": 0.05403715744614601,
|
212 |
+
"eval_rouge1": 0.3357586244098568,
|
213 |
+
"eval_rouge2": 0.12191279657273882,
|
214 |
+
"eval_rougeL": 0.3287623290777512,
|
215 |
+
"eval_runtime": 7.0002,
|
216 |
+
"eval_samples_per_second": 43.427,
|
217 |
+
"eval_steps_per_second": 5.428,
|
218 |
+
"step": 1683
|
219 |
+
},
|
220 |
+
{
|
221 |
+
"epoch": 12.0,
|
222 |
+
"grad_norm": 0.24781078100204468,
|
223 |
+
"learning_rate": 2.3906250000000002e-05,
|
224 |
+
"loss": 0.0331,
|
225 |
+
"step": 1836
|
226 |
+
},
|
227 |
+
{
|
228 |
+
"epoch": 12.0,
|
229 |
+
"eval_bleu": 0.0475663160301358,
|
230 |
+
"eval_loss": 0.05399588495492935,
|
231 |
+
"eval_rouge1": 0.3483203778520053,
|
232 |
+
"eval_rouge2": 0.13122980561384404,
|
233 |
+
"eval_rougeL": 0.3442445629111672,
|
234 |
+
"eval_runtime": 6.9958,
|
235 |
+
"eval_samples_per_second": 43.455,
|
236 |
+
"eval_steps_per_second": 5.432,
|
237 |
+
"step": 1836
|
238 |
+
},
|
239 |
+
{
|
240 |
+
"epoch": 13.0,
|
241 |
+
"grad_norm": 0.3433144986629486,
|
242 |
+
"learning_rate": 2.091796875e-05,
|
243 |
+
"loss": 0.0308,
|
244 |
+
"step": 1989
|
245 |
+
},
|
246 |
+
{
|
247 |
+
"epoch": 13.0,
|
248 |
+
"eval_bleu": 0.05895730441080768,
|
249 |
+
"eval_loss": 0.05515788868069649,
|
250 |
+
"eval_rouge1": 0.3599156447280303,
|
251 |
+
"eval_rouge2": 0.14385576959590118,
|
252 |
+
"eval_rougeL": 0.3539109033921022,
|
253 |
+
"eval_runtime": 6.9936,
|
254 |
+
"eval_samples_per_second": 43.469,
|
255 |
+
"eval_steps_per_second": 5.434,
|
256 |
+
"step": 1989
|
257 |
+
},
|
258 |
+
{
|
259 |
+
"epoch": 14.0,
|
260 |
+
"grad_norm": 0.3219660520553589,
|
261 |
+
"learning_rate": 1.79296875e-05,
|
262 |
+
"loss": 0.0291,
|
263 |
+
"step": 2142
|
264 |
+
},
|
265 |
+
{
|
266 |
+
"epoch": 14.0,
|
267 |
+
"eval_bleu": 0.06246557033315383,
|
268 |
+
"eval_loss": 0.05557234585285187,
|
269 |
+
"eval_rouge1": 0.37373672906337496,
|
270 |
+
"eval_rouge2": 0.14894175508943397,
|
271 |
+
"eval_rougeL": 0.3669928154707829,
|
272 |
+
"eval_runtime": 6.9886,
|
273 |
+
"eval_samples_per_second": 43.499,
|
274 |
+
"eval_steps_per_second": 5.437,
|
275 |
+
"step": 2142
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 14.0,
|
279 |
+
"step": 2142,
|
280 |
+
"total_flos": 8918419636224000.0,
|
281 |
+
"train_loss": 0.46412634738241815,
|
282 |
+
"train_runtime": 1276.9352,
|
283 |
+
"train_samples_per_second": 19.093,
|
284 |
+
"train_steps_per_second": 2.396
|
285 |
+
}
|
286 |
+
],
|
287 |
+
"logging_steps": 500,
|
288 |
+
"max_steps": 3060,
|
289 |
+
"num_input_tokens_seen": 0,
|
290 |
+
"num_train_epochs": 20,
|
291 |
+
"save_steps": 500,
|
292 |
+
"stateful_callbacks": {
|
293 |
+
"EarlyStoppingCallback": {
|
294 |
+
"args": {
|
295 |
+
"early_stopping_patience": 5,
|
296 |
+
"early_stopping_threshold": 0.0
|
297 |
+
},
|
298 |
+
"attributes": {
|
299 |
+
"early_stopping_patience_counter": 0
|
300 |
+
}
|
301 |
+
},
|
302 |
+
"TrainerControl": {
|
303 |
+
"args": {
|
304 |
+
"should_epoch_stop": false,
|
305 |
+
"should_evaluate": false,
|
306 |
+
"should_log": false,
|
307 |
+
"should_save": true,
|
308 |
+
"should_training_stop": true
|
309 |
+
},
|
310 |
+
"attributes": {}
|
311 |
+
}
|
312 |
+
},
|
313 |
+
"total_flos": 8918419636224000.0,
|
314 |
+
"train_batch_size": 8,
|
315 |
+
"trial_name": null,
|
316 |
+
"trial_params": null
|
317 |
+
}
|