sjrhuschlee commited on
Commit
6bb6b05
1 Parent(s): f4dc3e0

Upload folder using huggingface_hub

Browse files
added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "[MASK]": 128000
3
+ }
config.json ADDED
@@ -0,0 +1,47 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "microsoft/deberta-v3-large",
3
+ "architectures": [
4
+ "DebertaV2ForSequenceClassification"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "hidden_act": "gelu",
8
+ "hidden_dropout_prob": 0.1,
9
+ "hidden_size": 1024,
10
+ "id2label": {
11
+ "0": "contradiction",
12
+ "1": "no_support",
13
+ "2": "partial_support",
14
+ "3": "full_support"
15
+ },
16
+ "initializer_range": 0.02,
17
+ "intermediate_size": 4096,
18
+ "label2id": {
19
+ "contradiction": 0,
20
+ "full_support": 3,
21
+ "no_support": 1,
22
+ "partial_support": 2
23
+ },
24
+ "layer_norm_eps": 1e-07,
25
+ "max_position_embeddings": 512,
26
+ "max_relative_positions": -1,
27
+ "model_type": "deberta-v2",
28
+ "norm_rel_ebd": "layer_norm",
29
+ "num_attention_heads": 16,
30
+ "num_hidden_layers": 24,
31
+ "pad_token_id": 0,
32
+ "pooler_dropout": 0,
33
+ "pooler_hidden_act": "gelu",
34
+ "pooler_hidden_size": 1024,
35
+ "pos_att_type": [
36
+ "p2c",
37
+ "c2p"
38
+ ],
39
+ "position_biased_input": false,
40
+ "position_buckets": 256,
41
+ "relative_attention": true,
42
+ "share_att_key": true,
43
+ "torch_dtype": "float32",
44
+ "transformers_version": "4.31.0.dev0",
45
+ "type_vocab_size": 0,
46
+ "vocab_size": 128100
47
+ }
eval.jsonl ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {"eval_loss": 0.8360257744789124, "eval_accuracy": 0.71900826446281, "eval_matthews_correlation": 0.5055263088346037, "eval_runtime": 5.3991, "eval_samples_per_second": 44.822, "eval_steps_per_second": 2.963, "eval_samples": 242, "eval_set": "osu"}
2
+ {"eval_loss": 0.6920120120048523, "eval_accuracy": 0.8610421836228288, "eval_matthews_correlation": 0.6476113398062635, "eval_runtime": 3.2008, "eval_samples_per_second": 125.907, "eval_steps_per_second": 8.123, "eval_samples": 403, "eval_set": "deepset_hall"}
3
+ {"eval_loss": 1.7084012031555176, "eval_accuracy": 0.7117117117117117, "eval_matthews_correlation": 0.5285613156736579, "eval_runtime": 1.9454, "eval_samples_per_second": 57.059, "eval_steps_per_second": 3.598, "eval_samples": 111, "eval_set": "deepset_flower"}
4
+ {"eval_loss": 1.042230486869812, "eval_accuracy": 0.609, "eval_matthews_correlation": 0.42100205280404024, "eval_runtime": 5.1712, "eval_samples_per_second": 193.38, "eval_steps_per_second": 12.183, "eval_samples": 1000, "eval_set": "anli"}
logs.jsonl ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {"eval_accuracy": 0.71900826446281, "eval_loss": 0.8360257744789124, "eval_matthews_correlation": 0.5055263088346037, "eval_runtime": 5.3991, "eval_samples_per_second": 44.822, "eval_steps_per_second": 2.963, "step": 0}
2
+ {"eval_accuracy": 0.8610421836228288, "eval_loss": 0.6920120120048523, "eval_matthews_correlation": 0.6476113398062635, "eval_runtime": 3.2008, "eval_samples_per_second": 125.907, "eval_steps_per_second": 8.123, "step": 0}
3
+ {"eval_accuracy": 0.7117117117117117, "eval_loss": 1.7084012031555176, "eval_matthews_correlation": 0.5285613156736579, "eval_runtime": 1.9454, "eval_samples_per_second": 57.059, "eval_steps_per_second": 3.598, "step": 0}
4
+ {"eval_accuracy": 0.609, "eval_loss": 1.042230486869812, "eval_matthews_correlation": 0.42100205280404024, "eval_runtime": 5.1712, "eval_samples_per_second": 193.38, "eval_steps_per_second": 12.183, "step": 0}
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:49fb25159a6ed29b438816cacb763de617457c2a74af7a5d7d0ff5a17ebf9ff3
3
+ size 1740399989
special_tokens_map.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
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": "[UNK]"
9
+ }
spm.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c679fbf93643d19aab7ee10c0b99e460bdbc02fedf34b92b05af343b4af586fd
3
+ size 2464616
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "[CLS]",
3
+ "clean_up_tokenization_spaces": true,
4
+ "cls_token": "[CLS]",
5
+ "do_lower_case": false,
6
+ "eos_token": "[SEP]",
7
+ "mask_token": "[MASK]",
8
+ "model_max_length": 1000000000000000019884624838656,
9
+ "pad_token": "[PAD]",
10
+ "sep_token": "[SEP]",
11
+ "sp_model_kwargs": {},
12
+ "split_by_punct": false,
13
+ "tokenizer_class": "DebertaV2Tokenizer",
14
+ "unk_token": "[UNK]",
15
+ "vocab_type": "spm"
16
+ }
trainer_state.json ADDED
@@ -0,0 +1,466 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.71900826446281,
3
+ "best_model_checkpoint": "models/single_label/deberta-v3-large/exp5/checkpoint-1500",
4
+ "epoch": 0.07913166189678593,
5
+ "global_step": 1500,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.01,
12
+ "learning_rate": 3.1645569620253166e-08,
13
+ "loss": 1.4101,
14
+ "step": 100
15
+ },
16
+ {
17
+ "epoch": 0.01,
18
+ "learning_rate": 6.329113924050633e-08,
19
+ "loss": 1.4037,
20
+ "step": 200
21
+ },
22
+ {
23
+ "epoch": 0.01,
24
+ "eval_osu_accuracy": 0.5289256198347108,
25
+ "eval_osu_loss": 1.3296300172805786,
26
+ "eval_osu_matthews_correlation": 0.08430984949814109,
27
+ "eval_osu_runtime": 4.6863,
28
+ "eval_osu_samples_per_second": 51.64,
29
+ "eval_osu_steps_per_second": 3.414,
30
+ "step": 250
31
+ },
32
+ {
33
+ "epoch": 0.01,
34
+ "eval_deepset_1_accuracy": 0.42016806722689076,
35
+ "eval_deepset_1_loss": 1.3601003885269165,
36
+ "eval_deepset_1_matthews_correlation": 0.007340018855649166,
37
+ "eval_deepset_1_runtime": 1.0598,
38
+ "eval_deepset_1_samples_per_second": 112.29,
39
+ "eval_deepset_1_steps_per_second": 7.549,
40
+ "step": 250
41
+ },
42
+ {
43
+ "epoch": 0.01,
44
+ "eval_deepset_2_accuracy": 0.5128205128205128,
45
+ "eval_deepset_2_loss": 1.341286540031433,
46
+ "eval_deepset_2_matthews_correlation": 0.09702062677085851,
47
+ "eval_deepset_2_runtime": 1.5595,
48
+ "eval_deepset_2_samples_per_second": 100.032,
49
+ "eval_deepset_2_steps_per_second": 6.412,
50
+ "step": 250
51
+ },
52
+ {
53
+ "epoch": 0.01,
54
+ "eval_deepset_3_accuracy": 0.5260115606936416,
55
+ "eval_deepset_3_loss": 1.3386189937591553,
56
+ "eval_deepset_3_matthews_correlation": 0.13147932110355937,
57
+ "eval_deepset_3_runtime": 3.0465,
58
+ "eval_deepset_3_samples_per_second": 113.572,
59
+ "eval_deepset_3_steps_per_second": 7.221,
60
+ "step": 250
61
+ },
62
+ {
63
+ "epoch": 0.01,
64
+ "eval_deepset_4_accuracy": 0.46601941747572817,
65
+ "eval_deepset_4_loss": 1.354163646697998,
66
+ "eval_deepset_4_matthews_correlation": 0.04602315745107765,
67
+ "eval_deepset_4_runtime": 1.7129,
68
+ "eval_deepset_4_samples_per_second": 120.262,
69
+ "eval_deepset_4_steps_per_second": 7.589,
70
+ "step": 250
71
+ },
72
+ {
73
+ "epoch": 0.01,
74
+ "eval_anli_accuracy": 0.314,
75
+ "eval_anli_loss": 1.3443490266799927,
76
+ "eval_anli_matthews_correlation": -0.03669996966921473,
77
+ "eval_anli_runtime": 5.1544,
78
+ "eval_anli_samples_per_second": 194.01,
79
+ "eval_anli_steps_per_second": 12.223,
80
+ "step": 250
81
+ },
82
+ {
83
+ "epoch": 0.02,
84
+ "learning_rate": 9.462025316455696e-08,
85
+ "loss": 1.381,
86
+ "step": 300
87
+ },
88
+ {
89
+ "epoch": 0.02,
90
+ "learning_rate": 1.2626582278481014e-07,
91
+ "loss": 1.3016,
92
+ "step": 400
93
+ },
94
+ {
95
+ "epoch": 0.03,
96
+ "learning_rate": 1.5791139240506326e-07,
97
+ "loss": 1.1103,
98
+ "step": 500
99
+ },
100
+ {
101
+ "epoch": 0.03,
102
+ "eval_osu_accuracy": 0.3677685950413223,
103
+ "eval_osu_loss": 1.2251540422439575,
104
+ "eval_osu_matthews_correlation": 0.009389673632694094,
105
+ "eval_osu_runtime": 4.6743,
106
+ "eval_osu_samples_per_second": 51.773,
107
+ "eval_osu_steps_per_second": 3.423,
108
+ "step": 500
109
+ },
110
+ {
111
+ "epoch": 0.03,
112
+ "eval_deepset_1_accuracy": 0.2605042016806723,
113
+ "eval_deepset_1_loss": 1.498146414756775,
114
+ "eval_deepset_1_matthews_correlation": -0.12070085589088507,
115
+ "eval_deepset_1_runtime": 1.0684,
116
+ "eval_deepset_1_samples_per_second": 111.378,
117
+ "eval_deepset_1_steps_per_second": 7.488,
118
+ "step": 500
119
+ },
120
+ {
121
+ "epoch": 0.03,
122
+ "eval_deepset_2_accuracy": 0.3076923076923077,
123
+ "eval_deepset_2_loss": 1.3761013746261597,
124
+ "eval_deepset_2_matthews_correlation": -0.08397491442025497,
125
+ "eval_deepset_2_runtime": 1.5607,
126
+ "eval_deepset_2_samples_per_second": 99.953,
127
+ "eval_deepset_2_steps_per_second": 6.407,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 0.03,
132
+ "eval_deepset_3_accuracy": 0.31213872832369943,
133
+ "eval_deepset_3_loss": 1.4205697774887085,
134
+ "eval_deepset_3_matthews_correlation": -0.06528708661392457,
135
+ "eval_deepset_3_runtime": 3.0447,
136
+ "eval_deepset_3_samples_per_second": 113.639,
137
+ "eval_deepset_3_steps_per_second": 7.226,
138
+ "step": 500
139
+ },
140
+ {
141
+ "epoch": 0.03,
142
+ "eval_deepset_4_accuracy": 0.25728155339805825,
143
+ "eval_deepset_4_loss": 1.5763508081436157,
144
+ "eval_deepset_4_matthews_correlation": -0.11908261162663966,
145
+ "eval_deepset_4_runtime": 1.7183,
146
+ "eval_deepset_4_samples_per_second": 119.883,
147
+ "eval_deepset_4_steps_per_second": 7.565,
148
+ "step": 500
149
+ },
150
+ {
151
+ "epoch": 0.03,
152
+ "eval_anli_accuracy": 0.33,
153
+ "eval_anli_loss": 1.1667168140411377,
154
+ "eval_anli_matthews_correlation": 0.0012307081400291311,
155
+ "eval_anli_runtime": 5.1504,
156
+ "eval_anli_samples_per_second": 194.161,
157
+ "eval_anli_steps_per_second": 12.232,
158
+ "step": 500
159
+ },
160
+ {
161
+ "epoch": 0.03,
162
+ "learning_rate": 1.8955696202531644e-07,
163
+ "loss": 0.8892,
164
+ "step": 600
165
+ },
166
+ {
167
+ "epoch": 0.04,
168
+ "learning_rate": 2.2120253164556962e-07,
169
+ "loss": 0.7838,
170
+ "step": 700
171
+ },
172
+ {
173
+ "epoch": 0.04,
174
+ "eval_osu_accuracy": 0.6942148760330579,
175
+ "eval_osu_loss": 0.9712508916854858,
176
+ "eval_osu_matthews_correlation": 0.4897642238477718,
177
+ "eval_osu_runtime": 4.6823,
178
+ "eval_osu_samples_per_second": 51.684,
179
+ "eval_osu_steps_per_second": 3.417,
180
+ "step": 750
181
+ },
182
+ {
183
+ "epoch": 0.04,
184
+ "eval_deepset_1_accuracy": 0.6134453781512605,
185
+ "eval_deepset_1_loss": 1.6580817699432373,
186
+ "eval_deepset_1_matthews_correlation": 0.41985911107813456,
187
+ "eval_deepset_1_runtime": 1.0652,
188
+ "eval_deepset_1_samples_per_second": 111.721,
189
+ "eval_deepset_1_steps_per_second": 7.511,
190
+ "step": 750
191
+ },
192
+ {
193
+ "epoch": 0.04,
194
+ "eval_deepset_2_accuracy": 0.7051282051282052,
195
+ "eval_deepset_2_loss": 1.3619518280029297,
196
+ "eval_deepset_2_matthews_correlation": 0.5328568142577198,
197
+ "eval_deepset_2_runtime": 1.5605,
198
+ "eval_deepset_2_samples_per_second": 99.966,
199
+ "eval_deepset_2_steps_per_second": 6.408,
200
+ "step": 750
201
+ },
202
+ {
203
+ "epoch": 0.04,
204
+ "eval_deepset_3_accuracy": 0.5578034682080925,
205
+ "eval_deepset_3_loss": 1.4653511047363281,
206
+ "eval_deepset_3_matthews_correlation": 0.2932617859503659,
207
+ "eval_deepset_3_runtime": 3.0504,
208
+ "eval_deepset_3_samples_per_second": 113.429,
209
+ "eval_deepset_3_steps_per_second": 7.212,
210
+ "step": 750
211
+ },
212
+ {
213
+ "epoch": 0.04,
214
+ "eval_deepset_4_accuracy": 0.529126213592233,
215
+ "eval_deepset_4_loss": 1.7414307594299316,
216
+ "eval_deepset_4_matthews_correlation": 0.3054166536717615,
217
+ "eval_deepset_4_runtime": 1.7235,
218
+ "eval_deepset_4_samples_per_second": 119.522,
219
+ "eval_deepset_4_steps_per_second": 7.543,
220
+ "step": 750
221
+ },
222
+ {
223
+ "epoch": 0.04,
224
+ "eval_anli_accuracy": 0.351,
225
+ "eval_anli_loss": 1.1338268518447876,
226
+ "eval_anli_matthews_correlation": 0.030305510765310642,
227
+ "eval_anli_runtime": 5.1432,
228
+ "eval_anli_samples_per_second": 194.432,
229
+ "eval_anli_steps_per_second": 12.249,
230
+ "step": 750
231
+ },
232
+ {
233
+ "epoch": 0.04,
234
+ "learning_rate": 2.5284810126582275e-07,
235
+ "loss": 0.6807,
236
+ "step": 800
237
+ },
238
+ {
239
+ "epoch": 0.05,
240
+ "learning_rate": 2.844936708860759e-07,
241
+ "loss": 0.502,
242
+ "step": 900
243
+ },
244
+ {
245
+ "epoch": 0.05,
246
+ "learning_rate": 2.991503304270561e-07,
247
+ "loss": 0.4006,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 0.05,
252
+ "eval_osu_accuracy": 0.7024793388429752,
253
+ "eval_osu_loss": 0.7255080938339233,
254
+ "eval_osu_matthews_correlation": 0.4761839486992106,
255
+ "eval_osu_runtime": 4.6909,
256
+ "eval_osu_samples_per_second": 51.589,
257
+ "eval_osu_steps_per_second": 3.411,
258
+ "step": 1000
259
+ },
260
+ {
261
+ "epoch": 0.05,
262
+ "eval_deepset_1_accuracy": 0.6134453781512605,
263
+ "eval_deepset_1_loss": 1.6959413290023804,
264
+ "eval_deepset_1_matthews_correlation": 0.42105052198193255,
265
+ "eval_deepset_1_runtime": 1.0672,
266
+ "eval_deepset_1_samples_per_second": 111.505,
267
+ "eval_deepset_1_steps_per_second": 7.496,
268
+ "step": 1000
269
+ },
270
+ {
271
+ "epoch": 0.05,
272
+ "eval_deepset_2_accuracy": 0.7243589743589743,
273
+ "eval_deepset_2_loss": 1.2153555154800415,
274
+ "eval_deepset_2_matthews_correlation": 0.5466918371684817,
275
+ "eval_deepset_2_runtime": 1.556,
276
+ "eval_deepset_2_samples_per_second": 100.254,
277
+ "eval_deepset_2_steps_per_second": 6.427,
278
+ "step": 1000
279
+ },
280
+ {
281
+ "epoch": 0.05,
282
+ "eval_deepset_3_accuracy": 0.6473988439306358,
283
+ "eval_deepset_3_loss": 1.4247812032699585,
284
+ "eval_deepset_3_matthews_correlation": 0.4017768477834457,
285
+ "eval_deepset_3_runtime": 3.045,
286
+ "eval_deepset_3_samples_per_second": 113.628,
287
+ "eval_deepset_3_steps_per_second": 7.225,
288
+ "step": 1000
289
+ },
290
+ {
291
+ "epoch": 0.05,
292
+ "eval_deepset_4_accuracy": 0.6504854368932039,
293
+ "eval_deepset_4_loss": 1.7334672212600708,
294
+ "eval_deepset_4_matthews_correlation": 0.4508615553005039,
295
+ "eval_deepset_4_runtime": 1.7091,
296
+ "eval_deepset_4_samples_per_second": 120.534,
297
+ "eval_deepset_4_steps_per_second": 7.606,
298
+ "step": 1000
299
+ },
300
+ {
301
+ "epoch": 0.05,
302
+ "eval_anli_accuracy": 0.477,
303
+ "eval_anli_loss": 1.3187198638916016,
304
+ "eval_anli_matthews_correlation": 0.218035345395415,
305
+ "eval_anli_runtime": 5.1567,
306
+ "eval_anli_samples_per_second": 193.922,
307
+ "eval_anli_steps_per_second": 12.217,
308
+ "step": 1000
309
+ },
310
+ {
311
+ "epoch": 0.06,
312
+ "learning_rate": 2.9748431165657797e-07,
313
+ "loss": 0.3456,
314
+ "step": 1100
315
+ },
316
+ {
317
+ "epoch": 0.06,
318
+ "learning_rate": 2.9581829288609986e-07,
319
+ "loss": 0.3098,
320
+ "step": 1200
321
+ },
322
+ {
323
+ "epoch": 0.07,
324
+ "eval_osu_accuracy": 0.6900826446280992,
325
+ "eval_osu_loss": 0.794772207736969,
326
+ "eval_osu_matthews_correlation": 0.4632054751499522,
327
+ "eval_osu_runtime": 4.6759,
328
+ "eval_osu_samples_per_second": 51.755,
329
+ "eval_osu_steps_per_second": 3.422,
330
+ "step": 1250
331
+ },
332
+ {
333
+ "epoch": 0.07,
334
+ "eval_deepset_1_accuracy": 0.5882352941176471,
335
+ "eval_deepset_1_loss": 1.9199906587600708,
336
+ "eval_deepset_1_matthews_correlation": 0.3787987593817238,
337
+ "eval_deepset_1_runtime": 1.0581,
338
+ "eval_deepset_1_samples_per_second": 112.466,
339
+ "eval_deepset_1_steps_per_second": 7.561,
340
+ "step": 1250
341
+ },
342
+ {
343
+ "epoch": 0.07,
344
+ "eval_deepset_2_accuracy": 0.6923076923076923,
345
+ "eval_deepset_2_loss": 1.3611226081848145,
346
+ "eval_deepset_2_matthews_correlation": 0.5007307553834116,
347
+ "eval_deepset_2_runtime": 1.5603,
348
+ "eval_deepset_2_samples_per_second": 99.978,
349
+ "eval_deepset_2_steps_per_second": 6.409,
350
+ "step": 1250
351
+ },
352
+ {
353
+ "epoch": 0.07,
354
+ "eval_deepset_3_accuracy": 0.630057803468208,
355
+ "eval_deepset_3_loss": 1.5691180229187012,
356
+ "eval_deepset_3_matthews_correlation": 0.38258183358832687,
357
+ "eval_deepset_3_runtime": 3.052,
358
+ "eval_deepset_3_samples_per_second": 113.37,
359
+ "eval_deepset_3_steps_per_second": 7.208,
360
+ "step": 1250
361
+ },
362
+ {
363
+ "epoch": 0.07,
364
+ "eval_deepset_4_accuracy": 0.5922330097087378,
365
+ "eval_deepset_4_loss": 1.946365237236023,
366
+ "eval_deepset_4_matthews_correlation": 0.36618144302865463,
367
+ "eval_deepset_4_runtime": 1.7085,
368
+ "eval_deepset_4_samples_per_second": 120.574,
369
+ "eval_deepset_4_steps_per_second": 7.609,
370
+ "step": 1250
371
+ },
372
+ {
373
+ "epoch": 0.07,
374
+ "eval_anli_accuracy": 0.577,
375
+ "eval_anli_loss": 1.1495600938796997,
376
+ "eval_anli_matthews_correlation": 0.36758926218287485,
377
+ "eval_anli_runtime": 5.1325,
378
+ "eval_anli_samples_per_second": 194.837,
379
+ "eval_anli_steps_per_second": 12.275,
380
+ "step": 1250
381
+ },
382
+ {
383
+ "epoch": 0.07,
384
+ "learning_rate": 2.941522741156217e-07,
385
+ "loss": 0.2936,
386
+ "step": 1300
387
+ },
388
+ {
389
+ "epoch": 0.07,
390
+ "learning_rate": 2.9248625534514355e-07,
391
+ "loss": 0.275,
392
+ "step": 1400
393
+ },
394
+ {
395
+ "epoch": 0.08,
396
+ "learning_rate": 2.9083689676237014e-07,
397
+ "loss": 0.2711,
398
+ "step": 1500
399
+ },
400
+ {
401
+ "epoch": 0.08,
402
+ "eval_osu_accuracy": 0.71900826446281,
403
+ "eval_osu_loss": 0.8360257744789124,
404
+ "eval_osu_matthews_correlation": 0.5055263088346037,
405
+ "eval_osu_runtime": 4.6825,
406
+ "eval_osu_samples_per_second": 51.682,
407
+ "eval_osu_steps_per_second": 3.417,
408
+ "step": 1500
409
+ },
410
+ {
411
+ "epoch": 0.08,
412
+ "eval_deepset_1_accuracy": 0.6302521008403361,
413
+ "eval_deepset_1_loss": 2.0135045051574707,
414
+ "eval_deepset_1_matthews_correlation": 0.43404618683091417,
415
+ "eval_deepset_1_runtime": 1.0652,
416
+ "eval_deepset_1_samples_per_second": 111.718,
417
+ "eval_deepset_1_steps_per_second": 7.51,
418
+ "step": 1500
419
+ },
420
+ {
421
+ "epoch": 0.08,
422
+ "eval_deepset_2_accuracy": 0.7115384615384616,
423
+ "eval_deepset_2_loss": 1.3997838497161865,
424
+ "eval_deepset_2_matthews_correlation": 0.5105532533477752,
425
+ "eval_deepset_2_runtime": 1.5611,
426
+ "eval_deepset_2_samples_per_second": 99.928,
427
+ "eval_deepset_2_steps_per_second": 6.406,
428
+ "step": 1500
429
+ },
430
+ {
431
+ "epoch": 0.08,
432
+ "eval_deepset_3_accuracy": 0.6676300578034682,
433
+ "eval_deepset_3_loss": 1.5432208776474,
434
+ "eval_deepset_3_matthews_correlation": 0.42876096695199906,
435
+ "eval_deepset_3_runtime": 3.0533,
436
+ "eval_deepset_3_samples_per_second": 113.322,
437
+ "eval_deepset_3_steps_per_second": 7.205,
438
+ "step": 1500
439
+ },
440
+ {
441
+ "epoch": 0.08,
442
+ "eval_deepset_4_accuracy": 0.6359223300970874,
443
+ "eval_deepset_4_loss": 1.9390015602111816,
444
+ "eval_deepset_4_matthews_correlation": 0.4108775577859797,
445
+ "eval_deepset_4_runtime": 1.7173,
446
+ "eval_deepset_4_samples_per_second": 119.953,
447
+ "eval_deepset_4_steps_per_second": 7.57,
448
+ "step": 1500
449
+ },
450
+ {
451
+ "epoch": 0.08,
452
+ "eval_anli_accuracy": 0.609,
453
+ "eval_anli_loss": 1.042230486869812,
454
+ "eval_anli_matthews_correlation": 0.42100205280404024,
455
+ "eval_anli_runtime": 5.1501,
456
+ "eval_anli_samples_per_second": 194.17,
457
+ "eval_anli_steps_per_second": 12.233,
458
+ "step": 1500
459
+ }
460
+ ],
461
+ "max_steps": 18955,
462
+ "num_train_epochs": 1,
463
+ "total_flos": 2.042867740823347e+16,
464
+ "trial_name": null,
465
+ "trial_params": null
466
+ }
training_cfg.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_name": "/home/ubuntu/code/llm-eval/models/single_label/deberta-v3-large/exp5/checkpoint-1500",
3
+ "lr": 3e-07,
4
+ "epochs": 1,
5
+ "weight_decay": 0.01,
6
+ "train_bs": 16,
7
+ "eval_bs": 16,
8
+ "gradient_accumulation_steps": 4,
9
+ "gradient_checkpointing": true,
10
+ "eval_datasets": [
11
+ {"path": "./data/all_eval/osu_eval.csv", "name": "osu"},
12
+ {"path": "./data/all_eval/deepset_hallucination_eval_only_all_agree.csv", "name": "deepset_hall"},
13
+ {"path": "./data/all_eval/deepset_flower_hallucinations_eval_only_all_agree.csv", "name": "deepset_flower"}
14
+ ],
15
+ "train_dataset_path": "./data/all_training/all_train_v2.csv",
16
+ "exclude_datasets": ["nq"],
17
+ "tf32": true,
18
+ "bf16": false,
19
+ "half_precision": true,
20
+ "optim": "adamw_bnb_8bit"
21
+ }