MHGanainy/gpt2-xl-lora-multi-5
Browse files- README.md +2 -0
- all_results.json +13 -0
- eval_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +308 -0
README.md
CHANGED
@@ -15,6 +15,8 @@ should probably proofread and complete it, then remove this comment. -->
|
|
15 |
# gpt2-xl-lora-multi-5
|
16 |
|
17 |
This model is a fine-tuned version of [openai-community/gpt2-xl](https://huggingface.co/openai-community/gpt2-xl) on an unknown dataset.
|
|
|
|
|
18 |
|
19 |
## Model description
|
20 |
|
|
|
15 |
# gpt2-xl-lora-multi-5
|
16 |
|
17 |
This model is a fine-tuned version of [openai-community/gpt2-xl](https://huggingface.co/openai-community/gpt2-xl) on an unknown dataset.
|
18 |
+
It achieves the following results on the evaluation set:
|
19 |
+
- Loss: 2.0046
|
20 |
|
21 |
## Model description
|
22 |
|
all_results.json
ADDED
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 1.0,
|
3 |
+
"eval_loss": 2.004640817642212,
|
4 |
+
"eval_runtime": 786.2446,
|
5 |
+
"eval_samples_per_second": 7.809,
|
6 |
+
"eval_steps_per_second": 0.488,
|
7 |
+
"perplexity": 7.423427053812971,
|
8 |
+
"total_flos": 5.5431121920956826e+17,
|
9 |
+
"train_loss": 2.4680882914209215,
|
10 |
+
"train_runtime": 1212.9814,
|
11 |
+
"train_samples_per_second": 50.165,
|
12 |
+
"train_steps_per_second": 3.136
|
13 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 1.0,
|
3 |
+
"eval_loss": 2.004640817642212,
|
4 |
+
"eval_runtime": 786.2446,
|
5 |
+
"eval_samples_per_second": 7.809,
|
6 |
+
"eval_steps_per_second": 0.488,
|
7 |
+
"perplexity": 7.423427053812971
|
8 |
+
}
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 1.0,
|
3 |
+
"total_flos": 5.5431121920956826e+17,
|
4 |
+
"train_loss": 2.4680882914209215,
|
5 |
+
"train_runtime": 1212.9814,
|
6 |
+
"train_samples_per_second": 50.165,
|
7 |
+
"train_steps_per_second": 3.136
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,308 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 3804,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.026288117770767613,
|
13 |
+
"grad_norm": 0.21158406138420105,
|
14 |
+
"learning_rate": 6.377383300460224e-07,
|
15 |
+
"loss": 3.4637,
|
16 |
+
"step": 100
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.052576235541535225,
|
20 |
+
"grad_norm": 0.17568713426589966,
|
21 |
+
"learning_rate": 1.2886259040105195e-06,
|
22 |
+
"loss": 3.5087,
|
23 |
+
"step": 200
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.07886435331230283,
|
27 |
+
"grad_norm": 0.6034949421882629,
|
28 |
+
"learning_rate": 1.946088099934254e-06,
|
29 |
+
"loss": 3.4185,
|
30 |
+
"step": 300
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.10515247108307045,
|
34 |
+
"grad_norm": 0.34301501512527466,
|
35 |
+
"learning_rate": 2.603550295857988e-06,
|
36 |
+
"loss": 3.2854,
|
37 |
+
"step": 400
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.13144058885383805,
|
41 |
+
"grad_norm": 3.483755588531494,
|
42 |
+
"learning_rate": 3.2610124917817228e-06,
|
43 |
+
"loss": 3.1951,
|
44 |
+
"step": 500
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.15772870662460567,
|
48 |
+
"grad_norm": 0.43615999817848206,
|
49 |
+
"learning_rate": 3.918474687705457e-06,
|
50 |
+
"loss": 3.0889,
|
51 |
+
"step": 600
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.18401682439537329,
|
55 |
+
"grad_norm": 0.3760946989059448,
|
56 |
+
"learning_rate": 4.575936883629192e-06,
|
57 |
+
"loss": 2.8217,
|
58 |
+
"step": 700
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.2103049421661409,
|
62 |
+
"grad_norm": 0.2940825819969177,
|
63 |
+
"learning_rate": 5.2333990795529265e-06,
|
64 |
+
"loss": 2.5901,
|
65 |
+
"step": 800
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.23659305993690852,
|
69 |
+
"grad_norm": 0.38412269949913025,
|
70 |
+
"learning_rate": 5.89086127547666e-06,
|
71 |
+
"loss": 2.5224,
|
72 |
+
"step": 900
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.2628811777076761,
|
76 |
+
"grad_norm": 0.25396350026130676,
|
77 |
+
"learning_rate": 6.548323471400395e-06,
|
78 |
+
"loss": 2.4714,
|
79 |
+
"step": 1000
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.2891692954784437,
|
83 |
+
"grad_norm": 0.21601170301437378,
|
84 |
+
"learning_rate": 7.20578566732413e-06,
|
85 |
+
"loss": 2.457,
|
86 |
+
"step": 1100
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.31545741324921134,
|
90 |
+
"grad_norm": 0.23218804597854614,
|
91 |
+
"learning_rate": 7.863247863247863e-06,
|
92 |
+
"loss": 2.4262,
|
93 |
+
"step": 1200
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.34174553101997895,
|
97 |
+
"grad_norm": 0.24238590896129608,
|
98 |
+
"learning_rate": 8.5207100591716e-06,
|
99 |
+
"loss": 2.4206,
|
100 |
+
"step": 1300
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.36803364879074657,
|
104 |
+
"grad_norm": 0.4151826500892639,
|
105 |
+
"learning_rate": 9.178172255095332e-06,
|
106 |
+
"loss": 2.3889,
|
107 |
+
"step": 1400
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.3943217665615142,
|
111 |
+
"grad_norm": 0.22028537094593048,
|
112 |
+
"learning_rate": 9.835634451019067e-06,
|
113 |
+
"loss": 2.377,
|
114 |
+
"step": 1500
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.4206098843322818,
|
118 |
+
"grad_norm": 0.23977142572402954,
|
119 |
+
"learning_rate": 1.0493096646942801e-05,
|
120 |
+
"loss": 2.3533,
|
121 |
+
"step": 1600
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.4468980021030494,
|
125 |
+
"grad_norm": 0.2686922252178192,
|
126 |
+
"learning_rate": 1.1150558842866538e-05,
|
127 |
+
"loss": 2.3217,
|
128 |
+
"step": 1700
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.47318611987381703,
|
132 |
+
"grad_norm": 0.2237405776977539,
|
133 |
+
"learning_rate": 1.1808021038790272e-05,
|
134 |
+
"loss": 2.3034,
|
135 |
+
"step": 1800
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.49947423764458465,
|
139 |
+
"grad_norm": 0.35920849442481995,
|
140 |
+
"learning_rate": 1.2465483234714005e-05,
|
141 |
+
"loss": 2.296,
|
142 |
+
"step": 1900
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.5257623554153522,
|
146 |
+
"grad_norm": 0.32759296894073486,
|
147 |
+
"learning_rate": 1.312294543063774e-05,
|
148 |
+
"loss": 2.291,
|
149 |
+
"step": 2000
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.5520504731861199,
|
153 |
+
"grad_norm": 0.3541663587093353,
|
154 |
+
"learning_rate": 1.3780407626561474e-05,
|
155 |
+
"loss": 2.2739,
|
156 |
+
"step": 2100
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.5783385909568874,
|
160 |
+
"grad_norm": 0.3697475492954254,
|
161 |
+
"learning_rate": 1.4437869822485209e-05,
|
162 |
+
"loss": 2.2553,
|
163 |
+
"step": 2200
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 0.6046267087276551,
|
167 |
+
"grad_norm": 0.27857789397239685,
|
168 |
+
"learning_rate": 1.5095332018408943e-05,
|
169 |
+
"loss": 2.2625,
|
170 |
+
"step": 2300
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.6309148264984227,
|
174 |
+
"grad_norm": 0.22623978555202484,
|
175 |
+
"learning_rate": 1.5752794214332678e-05,
|
176 |
+
"loss": 2.2512,
|
177 |
+
"step": 2400
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.6572029442691903,
|
181 |
+
"grad_norm": 0.3197992146015167,
|
182 |
+
"learning_rate": 1.641025641025641e-05,
|
183 |
+
"loss": 2.222,
|
184 |
+
"step": 2500
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.6834910620399579,
|
188 |
+
"grad_norm": 0.2565959692001343,
|
189 |
+
"learning_rate": 1.7067718606180147e-05,
|
190 |
+
"loss": 2.2295,
|
191 |
+
"step": 2600
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.7097791798107256,
|
195 |
+
"grad_norm": 0.24300645291805267,
|
196 |
+
"learning_rate": 1.772518080210388e-05,
|
197 |
+
"loss": 2.217,
|
198 |
+
"step": 2700
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 0.7360672975814931,
|
202 |
+
"grad_norm": 0.21543020009994507,
|
203 |
+
"learning_rate": 1.8382642998027616e-05,
|
204 |
+
"loss": 2.234,
|
205 |
+
"step": 2800
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 0.7623554153522608,
|
209 |
+
"grad_norm": 0.27316907048225403,
|
210 |
+
"learning_rate": 1.904010519395135e-05,
|
211 |
+
"loss": 2.2144,
|
212 |
+
"step": 2900
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.7886435331230284,
|
216 |
+
"grad_norm": 0.3314705789089203,
|
217 |
+
"learning_rate": 1.9697567389875082e-05,
|
218 |
+
"loss": 2.2068,
|
219 |
+
"step": 3000
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.814931650893796,
|
223 |
+
"grad_norm": 0.2859819531440735,
|
224 |
+
"learning_rate": 1.975319567905163e-05,
|
225 |
+
"loss": 2.1974,
|
226 |
+
"step": 3100
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 0.8412197686645636,
|
230 |
+
"grad_norm": 0.44442227482795715,
|
231 |
+
"learning_rate": 1.8051220835680475e-05,
|
232 |
+
"loss": 2.1802,
|
233 |
+
"step": 3200
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 0.8675078864353313,
|
237 |
+
"grad_norm": 0.35564538836479187,
|
238 |
+
"learning_rate": 1.5000000000000002e-05,
|
239 |
+
"loss": 2.1839,
|
240 |
+
"step": 3300
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 0.8937960042060988,
|
244 |
+
"grad_norm": 0.28175088763237,
|
245 |
+
"learning_rate": 1.1110865250488047e-05,
|
246 |
+
"loss": 2.1736,
|
247 |
+
"step": 3400
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 0.9200841219768665,
|
251 |
+
"grad_norm": 0.28838351368904114,
|
252 |
+
"learning_rate": 7.0355686110209266e-06,
|
253 |
+
"loss": 2.1717,
|
254 |
+
"step": 3500
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.9463722397476341,
|
258 |
+
"grad_norm": 0.3840157687664032,
|
259 |
+
"learning_rate": 3.4570596330386775e-06,
|
260 |
+
"loss": 2.1674,
|
261 |
+
"step": 3600
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 0.9726603575184016,
|
265 |
+
"grad_norm": 0.2805449366569519,
|
266 |
+
"learning_rate": 9.750348092257368e-07,
|
267 |
+
"loss": 2.1741,
|
268 |
+
"step": 3700
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 0.9989484752891693,
|
272 |
+
"grad_norm": 0.3055085837841034,
|
273 |
+
"learning_rate": 5.438766451707489e-09,
|
274 |
+
"loss": 2.1764,
|
275 |
+
"step": 3800
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 1.0,
|
279 |
+
"step": 3804,
|
280 |
+
"total_flos": 5.5431121920956826e+17,
|
281 |
+
"train_loss": 2.4680882914209215,
|
282 |
+
"train_runtime": 1212.9814,
|
283 |
+
"train_samples_per_second": 50.165,
|
284 |
+
"train_steps_per_second": 3.136
|
285 |
+
}
|
286 |
+
],
|
287 |
+
"logging_steps": 100,
|
288 |
+
"max_steps": 3804,
|
289 |
+
"num_input_tokens_seen": 0,
|
290 |
+
"num_train_epochs": 1,
|
291 |
+
"save_steps": 500,
|
292 |
+
"stateful_callbacks": {
|
293 |
+
"TrainerControl": {
|
294 |
+
"args": {
|
295 |
+
"should_epoch_stop": false,
|
296 |
+
"should_evaluate": false,
|
297 |
+
"should_log": false,
|
298 |
+
"should_save": true,
|
299 |
+
"should_training_stop": true
|
300 |
+
},
|
301 |
+
"attributes": {}
|
302 |
+
}
|
303 |
+
},
|
304 |
+
"total_flos": 5.5431121920956826e+17,
|
305 |
+
"train_batch_size": 2,
|
306 |
+
"trial_name": null,
|
307 |
+
"trial_params": null
|
308 |
+
}
|