Paulo Vitor commited on
Commit
406e7e9
1 Parent(s): 1a07ae8

fine tune swag

Browse files
README.md ADDED
@@ -0,0 +1,74 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ license: mit
5
+ base_model: microsoft/mdeberta-v3-base
6
+ tags:
7
+ - generated_from_trainer
8
+ datasets:
9
+ - swag
10
+ metrics:
11
+ - accuracy
12
+ model-index:
13
+ - name: llm-mdeberta-v3-swag
14
+ results:
15
+ - task:
16
+ name: Multiple Choice
17
+ type: multiple-choice
18
+ dataset:
19
+ name: SWAG
20
+ type: swag
21
+ config: regular
22
+ split: validation
23
+ args: regular
24
+ metrics:
25
+ - name: Accuracy
26
+ type: accuracy
27
+ value: 0.777816653251648
28
+ ---
29
+
30
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
31
+ should probably proofread and complete it, then remove this comment. -->
32
+
33
+ # llm-mdeberta-v3-swag
34
+
35
+ This model is a fine-tuned version of [microsoft/mdeberta-v3-base](https://huggingface.co/microsoft/mdeberta-v3-base) on the SWAG dataset.
36
+ It achieves the following results on the evaluation set:
37
+ - Loss: 0.8202
38
+ - Accuracy: 0.7778
39
+
40
+ ## Model description
41
+
42
+ More information needed
43
+
44
+ ## Intended uses & limitations
45
+
46
+ More information needed
47
+
48
+ ## Training and evaluation data
49
+
50
+ More information needed
51
+
52
+ ## Training procedure
53
+
54
+ ### Training hyperparameters
55
+
56
+ The following hyperparameters were used during training:
57
+ - learning_rate: 5e-05
58
+ - train_batch_size: 8
59
+ - eval_batch_size: 8
60
+ - seed: 42
61
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
62
+ - lr_scheduler_type: linear
63
+ - num_epochs: 3.0
64
+
65
+ ### Training results
66
+
67
+
68
+
69
+ ### Framework versions
70
+
71
+ - Transformers 4.37.0.dev0
72
+ - Pytorch 2.1.0+cu121
73
+ - Datasets 2.15.0
74
+ - Tokenizers 0.15.0
added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "[MASK]": 250101
3
+ }
all_results.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "eval_accuracy": 0.777816653251648,
4
+ "eval_loss": 0.8202171921730042,
5
+ "eval_runtime": 366.8314,
6
+ "eval_samples": 20006,
7
+ "eval_samples_per_second": 54.537,
8
+ "eval_steps_per_second": 6.818,
9
+ "train_loss": 0.6582008492790169,
10
+ "train_runtime": 15084.1881,
11
+ "train_samples": 73546,
12
+ "train_samples_per_second": 14.627,
13
+ "train_steps_per_second": 1.829
14
+ }
config.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "microsoft/mdeberta-v3-base",
3
+ "architectures": [
4
+ "DebertaV2ForMultipleChoice"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "hidden_act": "gelu",
8
+ "hidden_dropout_prob": 0.1,
9
+ "hidden_size": 768,
10
+ "initializer_range": 0.02,
11
+ "intermediate_size": 3072,
12
+ "layer_norm_eps": 1e-07,
13
+ "max_position_embeddings": 512,
14
+ "max_relative_positions": -1,
15
+ "model_type": "deberta-v2",
16
+ "norm_rel_ebd": "layer_norm",
17
+ "num_attention_heads": 12,
18
+ "num_hidden_layers": 12,
19
+ "pad_token_id": 0,
20
+ "pooler_dropout": 0,
21
+ "pooler_hidden_act": "gelu",
22
+ "pooler_hidden_size": 768,
23
+ "pos_att_type": [
24
+ "p2c",
25
+ "c2p"
26
+ ],
27
+ "position_biased_input": false,
28
+ "position_buckets": 256,
29
+ "relative_attention": true,
30
+ "share_att_key": true,
31
+ "torch_dtype": "float32",
32
+ "transformers_version": "4.37.0.dev0",
33
+ "type_vocab_size": 0,
34
+ "vocab_size": 251000
35
+ }
eval_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "eval_accuracy": 0.777816653251648,
4
+ "eval_loss": 0.8202171921730042,
5
+ "eval_runtime": 366.8314,
6
+ "eval_samples": 20006,
7
+ "eval_samples_per_second": 54.537,
8
+ "eval_steps_per_second": 6.818
9
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:49a3c54108f314f93a59b05591eb83b27b3624be5ed7914ffb4b87c9e5d732db
3
+ size 1115265124
runs/Dec17_22-15-06_a41ddc482777/events.out.tfevents.1702851476.a41ddc482777.2398.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:18b5f5614e5eb5a568387966334bb4c4ddf967f790d777e7dcff32fce94ab062
3
+ size 13582
runs/Dec17_22-15-06_a41ddc482777/events.out.tfevents.1702866932.a41ddc482777.2398.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:86176a3404a512a677dc548fbe6b079ba99046255a9c8919f081dac975b54bb4
3
+ size 417
special_tokens_map.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "[CLS]",
3
+ "cls_token": "[CLS]",
4
+ "eos_token": "[SEP]",
5
+ "mask_token": "[MASK]",
6
+ "pad_token": "[PAD]",
7
+ "sep_token": "[SEP]",
8
+ "unk_token": {
9
+ "content": "[UNK]",
10
+ "lstrip": false,
11
+ "normalized": true,
12
+ "rstrip": false,
13
+ "single_word": false
14
+ }
15
+ }
spm.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:13c8d666d62a7bc4ac8f040aab68e942c861f93303156cc28f5c7e885d86d6e3
3
+ size 4305025
tokenizer_config.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "[PAD]",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "[CLS]",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "[SEP]",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "[UNK]",
29
+ "lstrip": false,
30
+ "normalized": true,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ },
35
+ "250101": {
36
+ "content": "[MASK]",
37
+ "lstrip": false,
38
+ "normalized": false,
39
+ "rstrip": false,
40
+ "single_word": false,
41
+ "special": true
42
+ }
43
+ },
44
+ "bos_token": "[CLS]",
45
+ "clean_up_tokenization_spaces": true,
46
+ "cls_token": "[CLS]",
47
+ "do_lower_case": false,
48
+ "eos_token": "[SEP]",
49
+ "mask_token": "[MASK]",
50
+ "model_max_length": 1000000000000000019884624838656,
51
+ "pad_token": "[PAD]",
52
+ "sep_token": "[SEP]",
53
+ "sp_model_kwargs": {},
54
+ "split_by_punct": false,
55
+ "tokenizer_class": "DebertaV2Tokenizer",
56
+ "unk_token": "[UNK]",
57
+ "vocab_type": "spm"
58
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "train_loss": 0.6582008492790169,
4
+ "train_runtime": 15084.1881,
5
+ "train_samples": 73546,
6
+ "train_samples_per_second": 14.627,
7
+ "train_steps_per_second": 1.829
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,360 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 3.0,
5
+ "eval_steps": 500,
6
+ "global_step": 27582,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.05,
13
+ "learning_rate": 4.909361177579581e-05,
14
+ "loss": 1.1223,
15
+ "step": 500
16
+ },
17
+ {
18
+ "epoch": 0.11,
19
+ "learning_rate": 4.818722355159162e-05,
20
+ "loss": 1.035,
21
+ "step": 1000
22
+ },
23
+ {
24
+ "epoch": 0.16,
25
+ "learning_rate": 4.728083532738743e-05,
26
+ "loss": 1.0383,
27
+ "step": 1500
28
+ },
29
+ {
30
+ "epoch": 0.22,
31
+ "learning_rate": 4.637444710318324e-05,
32
+ "loss": 0.9976,
33
+ "step": 2000
34
+ },
35
+ {
36
+ "epoch": 0.27,
37
+ "learning_rate": 4.546805887897905e-05,
38
+ "loss": 0.9862,
39
+ "step": 2500
40
+ },
41
+ {
42
+ "epoch": 0.33,
43
+ "learning_rate": 4.4561670654774854e-05,
44
+ "loss": 0.9828,
45
+ "step": 3000
46
+ },
47
+ {
48
+ "epoch": 0.38,
49
+ "learning_rate": 4.3655282430570665e-05,
50
+ "loss": 0.991,
51
+ "step": 3500
52
+ },
53
+ {
54
+ "epoch": 0.44,
55
+ "learning_rate": 4.2748894206366476e-05,
56
+ "loss": 0.9492,
57
+ "step": 4000
58
+ },
59
+ {
60
+ "epoch": 0.49,
61
+ "learning_rate": 4.184250598216228e-05,
62
+ "loss": 0.9492,
63
+ "step": 4500
64
+ },
65
+ {
66
+ "epoch": 0.54,
67
+ "learning_rate": 4.093611775795809e-05,
68
+ "loss": 0.9483,
69
+ "step": 5000
70
+ },
71
+ {
72
+ "epoch": 0.6,
73
+ "learning_rate": 4.00297295337539e-05,
74
+ "loss": 0.9157,
75
+ "step": 5500
76
+ },
77
+ {
78
+ "epoch": 0.65,
79
+ "learning_rate": 3.912334130954971e-05,
80
+ "loss": 0.9092,
81
+ "step": 6000
82
+ },
83
+ {
84
+ "epoch": 0.71,
85
+ "learning_rate": 3.821695308534552e-05,
86
+ "loss": 0.8941,
87
+ "step": 6500
88
+ },
89
+ {
90
+ "epoch": 0.76,
91
+ "learning_rate": 3.731056486114133e-05,
92
+ "loss": 0.9126,
93
+ "step": 7000
94
+ },
95
+ {
96
+ "epoch": 0.82,
97
+ "learning_rate": 3.640417663693714e-05,
98
+ "loss": 0.8869,
99
+ "step": 7500
100
+ },
101
+ {
102
+ "epoch": 0.87,
103
+ "learning_rate": 3.549778841273294e-05,
104
+ "loss": 0.8928,
105
+ "step": 8000
106
+ },
107
+ {
108
+ "epoch": 0.92,
109
+ "learning_rate": 3.459140018852875e-05,
110
+ "loss": 0.8764,
111
+ "step": 8500
112
+ },
113
+ {
114
+ "epoch": 0.98,
115
+ "learning_rate": 3.368501196432456e-05,
116
+ "loss": 0.8788,
117
+ "step": 9000
118
+ },
119
+ {
120
+ "epoch": 1.03,
121
+ "learning_rate": 3.277862374012037e-05,
122
+ "loss": 0.7333,
123
+ "step": 9500
124
+ },
125
+ {
126
+ "epoch": 1.09,
127
+ "learning_rate": 3.187223551591618e-05,
128
+ "loss": 0.6604,
129
+ "step": 10000
130
+ },
131
+ {
132
+ "epoch": 1.14,
133
+ "learning_rate": 3.0965847291711983e-05,
134
+ "loss": 0.6689,
135
+ "step": 10500
136
+ },
137
+ {
138
+ "epoch": 1.2,
139
+ "learning_rate": 3.0059459067507794e-05,
140
+ "loss": 0.6555,
141
+ "step": 11000
142
+ },
143
+ {
144
+ "epoch": 1.25,
145
+ "learning_rate": 2.9153070843303605e-05,
146
+ "loss": 0.658,
147
+ "step": 11500
148
+ },
149
+ {
150
+ "epoch": 1.31,
151
+ "learning_rate": 2.8246682619099413e-05,
152
+ "loss": 0.6627,
153
+ "step": 12000
154
+ },
155
+ {
156
+ "epoch": 1.36,
157
+ "learning_rate": 2.734029439489522e-05,
158
+ "loss": 0.6626,
159
+ "step": 12500
160
+ },
161
+ {
162
+ "epoch": 1.41,
163
+ "learning_rate": 2.643390617069103e-05,
164
+ "loss": 0.6381,
165
+ "step": 13000
166
+ },
167
+ {
168
+ "epoch": 1.47,
169
+ "learning_rate": 2.552751794648684e-05,
170
+ "loss": 0.6319,
171
+ "step": 13500
172
+ },
173
+ {
174
+ "epoch": 1.52,
175
+ "learning_rate": 2.462112972228265e-05,
176
+ "loss": 0.6766,
177
+ "step": 14000
178
+ },
179
+ {
180
+ "epoch": 1.58,
181
+ "learning_rate": 2.3714741498078457e-05,
182
+ "loss": 0.6428,
183
+ "step": 14500
184
+ },
185
+ {
186
+ "epoch": 1.63,
187
+ "learning_rate": 2.2808353273874268e-05,
188
+ "loss": 0.622,
189
+ "step": 15000
190
+ },
191
+ {
192
+ "epoch": 1.69,
193
+ "learning_rate": 2.1901965049670075e-05,
194
+ "loss": 0.6313,
195
+ "step": 15500
196
+ },
197
+ {
198
+ "epoch": 1.74,
199
+ "learning_rate": 2.0995576825465886e-05,
200
+ "loss": 0.6466,
201
+ "step": 16000
202
+ },
203
+ {
204
+ "epoch": 1.79,
205
+ "learning_rate": 2.0089188601261694e-05,
206
+ "loss": 0.6203,
207
+ "step": 16500
208
+ },
209
+ {
210
+ "epoch": 1.85,
211
+ "learning_rate": 1.91828003770575e-05,
212
+ "loss": 0.6103,
213
+ "step": 17000
214
+ },
215
+ {
216
+ "epoch": 1.9,
217
+ "learning_rate": 1.8276412152853312e-05,
218
+ "loss": 0.6085,
219
+ "step": 17500
220
+ },
221
+ {
222
+ "epoch": 1.96,
223
+ "learning_rate": 1.737002392864912e-05,
224
+ "loss": 0.6039,
225
+ "step": 18000
226
+ },
227
+ {
228
+ "epoch": 2.01,
229
+ "learning_rate": 1.646363570444493e-05,
230
+ "loss": 0.5586,
231
+ "step": 18500
232
+ },
233
+ {
234
+ "epoch": 2.07,
235
+ "learning_rate": 1.5557247480240738e-05,
236
+ "loss": 0.4023,
237
+ "step": 19000
238
+ },
239
+ {
240
+ "epoch": 2.12,
241
+ "learning_rate": 1.4650859256036547e-05,
242
+ "loss": 0.39,
243
+ "step": 19500
244
+ },
245
+ {
246
+ "epoch": 2.18,
247
+ "learning_rate": 1.3744471031832355e-05,
248
+ "loss": 0.3777,
249
+ "step": 20000
250
+ },
251
+ {
252
+ "epoch": 2.23,
253
+ "learning_rate": 1.2838082807628162e-05,
254
+ "loss": 0.3958,
255
+ "step": 20500
256
+ },
257
+ {
258
+ "epoch": 2.28,
259
+ "learning_rate": 1.1931694583423973e-05,
260
+ "loss": 0.4169,
261
+ "step": 21000
262
+ },
263
+ {
264
+ "epoch": 2.34,
265
+ "learning_rate": 1.1025306359219782e-05,
266
+ "loss": 0.3802,
267
+ "step": 21500
268
+ },
269
+ {
270
+ "epoch": 2.39,
271
+ "learning_rate": 1.011891813501559e-05,
272
+ "loss": 0.3793,
273
+ "step": 22000
274
+ },
275
+ {
276
+ "epoch": 2.45,
277
+ "learning_rate": 9.212529910811399e-06,
278
+ "loss": 0.4118,
279
+ "step": 22500
280
+ },
281
+ {
282
+ "epoch": 2.5,
283
+ "learning_rate": 8.306141686607208e-06,
284
+ "loss": 0.3776,
285
+ "step": 23000
286
+ },
287
+ {
288
+ "epoch": 2.56,
289
+ "learning_rate": 7.399753462403016e-06,
290
+ "loss": 0.3892,
291
+ "step": 23500
292
+ },
293
+ {
294
+ "epoch": 2.61,
295
+ "learning_rate": 6.493365238198826e-06,
296
+ "loss": 0.3759,
297
+ "step": 24000
298
+ },
299
+ {
300
+ "epoch": 2.66,
301
+ "learning_rate": 5.586977013994634e-06,
302
+ "loss": 0.3663,
303
+ "step": 24500
304
+ },
305
+ {
306
+ "epoch": 2.72,
307
+ "learning_rate": 4.680588789790443e-06,
308
+ "loss": 0.3836,
309
+ "step": 25000
310
+ },
311
+ {
312
+ "epoch": 2.77,
313
+ "learning_rate": 3.7742005655862523e-06,
314
+ "loss": 0.3731,
315
+ "step": 25500
316
+ },
317
+ {
318
+ "epoch": 2.83,
319
+ "learning_rate": 2.8678123413820606e-06,
320
+ "loss": 0.3657,
321
+ "step": 26000
322
+ },
323
+ {
324
+ "epoch": 2.88,
325
+ "learning_rate": 1.96142411717787e-06,
326
+ "loss": 0.3635,
327
+ "step": 26500
328
+ },
329
+ {
330
+ "epoch": 2.94,
331
+ "learning_rate": 1.0550358929736786e-06,
332
+ "loss": 0.3885,
333
+ "step": 27000
334
+ },
335
+ {
336
+ "epoch": 2.99,
337
+ "learning_rate": 1.4864766876948735e-07,
338
+ "loss": 0.3573,
339
+ "step": 27500
340
+ },
341
+ {
342
+ "epoch": 3.0,
343
+ "step": 27582,
344
+ "total_flos": 2.836008195999293e+16,
345
+ "train_loss": 0.6582008492790169,
346
+ "train_runtime": 15084.1881,
347
+ "train_samples_per_second": 14.627,
348
+ "train_steps_per_second": 1.829
349
+ }
350
+ ],
351
+ "logging_steps": 500,
352
+ "max_steps": 27582,
353
+ "num_input_tokens_seen": 0,
354
+ "num_train_epochs": 3,
355
+ "save_steps": 500,
356
+ "total_flos": 2.836008195999293e+16,
357
+ "train_batch_size": 8,
358
+ "trial_name": null,
359
+ "trial_params": null
360
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5aa6359a4275e54941850b46e99f67e3b63904f5cac60221b891f449b5277bbf
3
+ size 4728