Hugging Face's logo

PurCL
/
codeart-26m-mfc-4f-100c

Transformers
PyTorch
codeart
Inference Endpoints
Model card Files Community
sheepy928 commited on
Commit
8e8b37f
•
1 Parent(s): b2c9d6f

Upload 11 files

Browse files
Files changed (11) hide show
  1. config.json +234 -0
  2. optimizer.pt +3 -0
  3. pytorch_model.bin +3 -0
  4. rng_state_0.pth +3 -0
  5. rng_state_1.pth +3 -0
  6. scheduler.pt +3 -0
  7. special_tokens_map.json +7 -0
  8. tokenizer.json +0 -0
  9. tokenizer_config.json +5 -0
  10. trainer_state.json +2936 -0
  11. training_args.bin +3 -0
config.json ADDED
@@ -0,0 +1,234 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "../../save/marinda-default-ep0.4-26m/checkpoint-25000",
3
+ "architectures": [
4
+ "MarindaForMultipleSequenceClassification"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "bos_token_id": 1,
8
+ "classifier_dropout": null,
9
+ "eos_token_id": 2,
10
+ "ep_add_linear_projection": true,
11
+ "hidden_act": "gelu",
12
+ "hidden_dropout_prob": 0.1,
13
+ "hidden_size": 768,
14
+ "id2label": {
15
+ "0": "zbot",
16
+ "1": "turla",
17
+ "2": "mokead",
18
+ "3": "tspy",
19
+ "4": "ponystealer",
20
+ "5": "ulise",
21
+ "6": "sofacy",
22
+ "7": "zxshell",
23
+ "8": "cozer",
24
+ "9": "travnet",
25
+ "10": "admoke",
26
+ "11": "barys",
27
+ "12": "dynamer",
28
+ "13": "apost",
29
+ "14": "enfal",
30
+ "15": "midie",
31
+ "16": "kazy",
32
+ "17": "deepscan",
33
+ "18": "ursu",
34
+ "19": "fugrafa",
35
+ "20": "skipper",
36
+ "21": "korplug",
37
+ "22": "sensode",
38
+ "23": "gulpix",
39
+ "24": "fareit",
40
+ "25": "startpage",
41
+ "26": "winnti",
42
+ "27": "vercuser",
43
+ "28": "rafdy",
44
+ "29": "denes",
45
+ "30": "trojano",
46
+ "31": "virtool",
47
+ "32": "protux",
48
+ "33": "cometer",
49
+ "34": "strictor",
50
+ "35": "r002c0pg821",
51
+ "36": "casdet",
52
+ "37": "daekom",
53
+ "38": "fragtor",
54
+ "39": "stealer",
55
+ "40": "linux",
56
+ "41": "brambul",
57
+ "42": "cryp",
58
+ "43": "dllhijacker",
59
+ "44": "elpq",
60
+ "45": "dialer",
61
+ "46": "meciv",
62
+ "47": "based",
63
+ "48": "filerepmalware",
64
+ "49": "scar",
65
+ "50": "sednit",
66
+ "51": "plugx",
67
+ "52": "msil",
68
+ "53": "genome",
69
+ "54": "mikey",
70
+ "55": "doina",
71
+ "56": "cerbu",
72
+ "57": "liewar",
73
+ "58": "tzeebot",
74
+ "59": "nettool",
75
+ "60": "autohotkey",
76
+ "61": "zusy",
77
+ "62": "spnr",
78
+ "63": "bkdr",
79
+ "64": "graftor",
80
+ "65": "keylogger",
81
+ "66": "bifrose",
82
+ "67": "poison",
83
+ "68": "msilperseus",
84
+ "69": "dangeroussig",
85
+ "70": "abki",
86
+ "71": "scadprv",
87
+ "72": "rogue",
88
+ "73": "symmi",
89
+ "74": "fraudload",
90
+ "75": "salgorea",
91
+ "76": "bulz",
92
+ "77": "flawedammyy",
93
+ "78": "autoit",
94
+ "79": "kelihos",
95
+ "80": "lazy",
96
+ "81": "babar",
97
+ "82": "mint",
98
+ "83": "miniduke",
99
+ "84": "timer",
100
+ "85": "cosmicduke",
101
+ "86": "kryptk",
102
+ "87": "indiqi",
103
+ "88": "zoxpng",
104
+ "89": "zxproxy",
105
+ "90": "rootkit",
106
+ "91": "androm",
107
+ "92": "razy",
108
+ "93": "r002c0dg921",
109
+ "94": "jaik",
110
+ "95": "farfli",
111
+ "96": "oceanlotus",
112
+ "97": "pepex",
113
+ "98": "strongpity",
114
+ "99": "agentb"
115
+ },
116
+ "initializer_range": 0.02,
117
+ "intermediate_size": 3072,
118
+ "label2id": {
119
+ "abki": 70,
120
+ "admoke": 10,
121
+ "agentb": 99,
122
+ "androm": 91,
123
+ "apost": 13,
124
+ "autohotkey": 60,
125
+ "autoit": 78,
126
+ "babar": 81,
127
+ "barys": 11,
128
+ "based": 47,
129
+ "bifrose": 66,
130
+ "bkdr": 63,
131
+ "brambul": 41,
132
+ "bulz": 76,
133
+ "casdet": 36,
134
+ "cerbu": 56,
135
+ "cometer": 33,
136
+ "cosmicduke": 85,
137
+ "cozer": 8,
138
+ "cryp": 42,
139
+ "daekom": 37,
140
+ "dangeroussig": 69,
141
+ "deepscan": 17,
142
+ "denes": 29,
143
+ "dialer": 45,
144
+ "dllhijacker": 43,
145
+ "doina": 55,
146
+ "dynamer": 12,
147
+ "elpq": 44,
148
+ "enfal": 14,
149
+ "fareit": 24,
150
+ "farfli": 95,
151
+ "filerepmalware": 48,
152
+ "flawedammyy": 77,
153
+ "fragtor": 38,
154
+ "fraudload": 74,
155
+ "fugrafa": 19,
156
+ "genome": 53,
157
+ "graftor": 64,
158
+ "gulpix": 23,
159
+ "indiqi": 87,
160
+ "jaik": 94,
161
+ "kazy": 16,
162
+ "kelihos": 79,
163
+ "keylogger": 65,
164
+ "korplug": 21,
165
+ "kryptk": 86,
166
+ "lazy": 80,
167
+ "liewar": 57,
168
+ "linux": 40,
169
+ "meciv": 46,
170
+ "midie": 15,
171
+ "mikey": 54,
172
+ "miniduke": 83,
173
+ "mint": 82,
174
+ "mokead": 2,
175
+ "msil": 52,
176
+ "msilperseus": 68,
177
+ "nettool": 59,
178
+ "oceanlotus": 96,
179
+ "pepex": 97,
180
+ "plugx": 51,
181
+ "poison": 67,
182
+ "ponystealer": 4,
183
+ "protux": 32,
184
+ "r002c0dg921": 93,
185
+ "r002c0pg821": 35,
186
+ "rafdy": 28,
187
+ "razy": 92,
188
+ "rogue": 72,
189
+ "rootkit": 90,
190
+ "salgorea": 75,
191
+ "scadprv": 71,
192
+ "scar": 49,
193
+ "sednit": 50,
194
+ "sensode": 22,
195
+ "skipper": 20,
196
+ "sofacy": 6,
197
+ "spnr": 62,
198
+ "startpage": 25,
199
+ "stealer": 39,
200
+ "strictor": 34,
201
+ "strongpity": 98,
202
+ "symmi": 73,
203
+ "timer": 84,
204
+ "travnet": 9,
205
+ "trojano": 30,
206
+ "tspy": 3,
207
+ "turla": 1,
208
+ "tzeebot": 58,
209
+ "ulise": 5,
210
+ "ursu": 18,
211
+ "vercuser": 27,
212
+ "virtool": 31,
213
+ "winnti": 26,
214
+ "zbot": 0,
215
+ "zoxpng": 88,
216
+ "zusy": 61,
217
+ "zxproxy": 89,
218
+ "zxshell": 7
219
+ },
220
+ "layer_norm_eps": 1e-12,
221
+ "max_position_embeddings": 600,
222
+ "max_relative_position_embeddings": 8,
223
+ "model_type": "marinda",
224
+ "num_attention_heads": 12,
225
+ "num_hidden_layers": 12,
226
+ "pad_token_id": 3,
227
+ "position_embedding_type": "mixed",
228
+ "problem_type": "multi_label_classification",
229
+ "torch_dtype": "float32",
230
+ "transformers_version": "4.30.1",
231
+ "type_vocab_size": 2,
232
+ "use_cache": true,
233
+ "vocab_size": 30523
234
+ }
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4baa2681de9ed801f68ef84e1a1f510935113cd307703df7a6c752e420834e3d
3
+ size 877160773
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3caee88eaa818acbe54e3a11b8e3b62a3b22bd8a3abf6fc5f16764cc054605fb
3
+ size 438591269
rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8e49c5fa2ff2e6b19747cf6819d0c558005d892ce38c960c71b8b6b419d3b9bd
3
+ size 15607
rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:72b3c5a225678de132712ec35e4a6c68bf5f542441898d7b9764b92c28b0fe56
3
+ size 15607
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aa02074245a22f9a5c1b00682ce0efbeb904eafd9fe84ac07fa2b0d3e3bfcff1
3
+ size 627
special_tokens_map.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "cls_token": "[CLS]",
3
+ "mask_token": "[MASK]",
4
+ "pad_token": "[PAD]",
5
+ "sep_token": "[SEP]",
6
+ "unk_token": "[UNK]"
7
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "clean_up_tokenization_spaces": true,
3
+ "model_max_length": 512,
4
+ "tokenizer_class": "MarindaTokenizer"
5
+ }
trainer_state.json ADDED
@@ -0,0 +1,2936 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.4501160092807424,
3
+ "best_model_checkpoint": "../save/marinda-default-ep0.4-26m-mfc-4f-100c/checkpoint-4000",
4
+ "epoch": 9.029345372460497,
5
+ "global_step": 4000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.02,
12
+ "learning_rate": 9.97742663656885e-05,
13
+ "loss": 0.5909,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.05,
18
+ "learning_rate": 9.954853273137697e-05,
19
+ "loss": 0.3154,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.07,
24
+ "learning_rate": 9.932279909706546e-05,
25
+ "loss": 0.1643,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.09,
30
+ "learning_rate": 9.909706546275395e-05,
31
+ "loss": 0.1148,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.11,
36
+ "learning_rate": 9.887133182844243e-05,
37
+ "loss": 0.0938,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 0.14,
42
+ "learning_rate": 9.864559819413092e-05,
43
+ "loss": 0.0826,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 0.16,
48
+ "learning_rate": 9.841986455981941e-05,
49
+ "loss": 0.083,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 0.18,
54
+ "learning_rate": 9.81941309255079e-05,
55
+ "loss": 0.085,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 0.2,
60
+ "learning_rate": 9.79683972911964e-05,
61
+ "loss": 0.0836,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 0.23,
66
+ "learning_rate": 9.774266365688489e-05,
67
+ "loss": 0.0777,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 0.23,
72
+ "eval_accuracy": 0.9827956989247312,
73
+ "eval_f1": 0.0,
74
+ "eval_loss": 0.08297247439622879,
75
+ "eval_precision": 0.0,
76
+ "eval_recall": 0.0,
77
+ "eval_roc_auc_score": 0.6634815201039388,
78
+ "eval_runtime": 5.3932,
79
+ "eval_samples_per_second": 34.488,
80
+ "eval_steps_per_second": 2.225,
81
+ "step": 100
82
+ },
83
+ {
84
+ "epoch": 0.25,
85
+ "learning_rate": 9.751693002257338e-05,
86
+ "loss": 0.0904,
87
+ "step": 110
88
+ },
89
+ {
90
+ "epoch": 0.27,
91
+ "learning_rate": 9.729119638826185e-05,
92
+ "loss": 0.0802,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.29,
97
+ "learning_rate": 9.706546275395035e-05,
98
+ "loss": 0.0818,
99
+ "step": 130
100
+ },
101
+ {
102
+ "epoch": 0.32,
103
+ "learning_rate": 9.683972911963884e-05,
104
+ "loss": 0.0799,
105
+ "step": 140
106
+ },
107
+ {
108
+ "epoch": 0.34,
109
+ "learning_rate": 9.661399548532731e-05,
110
+ "loss": 0.0897,
111
+ "step": 150
112
+ },
113
+ {
114
+ "epoch": 0.36,
115
+ "learning_rate": 9.63882618510158e-05,
116
+ "loss": 0.0819,
117
+ "step": 160
118
+ },
119
+ {
120
+ "epoch": 0.38,
121
+ "learning_rate": 9.61625282167043e-05,
122
+ "loss": 0.0841,
123
+ "step": 170
124
+ },
125
+ {
126
+ "epoch": 0.41,
127
+ "learning_rate": 9.593679458239279e-05,
128
+ "loss": 0.0861,
129
+ "step": 180
130
+ },
131
+ {
132
+ "epoch": 0.43,
133
+ "learning_rate": 9.571106094808126e-05,
134
+ "loss": 0.0836,
135
+ "step": 190
136
+ },
137
+ {
138
+ "epoch": 0.45,
139
+ "learning_rate": 9.548532731376975e-05,
140
+ "loss": 0.0831,
141
+ "step": 200
142
+ },
143
+ {
144
+ "epoch": 0.45,
145
+ "eval_accuracy": 0.9827956989247312,
146
+ "eval_f1": 0.0,
147
+ "eval_loss": 0.08346579968929291,
148
+ "eval_precision": 0.0,
149
+ "eval_recall": 0.0,
150
+ "eval_roc_auc_score": 0.6618079013949671,
151
+ "eval_runtime": 5.1133,
152
+ "eval_samples_per_second": 36.376,
153
+ "eval_steps_per_second": 2.347,
154
+ "step": 200
155
+ },
156
+ {
157
+ "epoch": 0.47,
158
+ "learning_rate": 9.525959367945825e-05,
159
+ "loss": 0.0912,
160
+ "step": 210
161
+ },
162
+ {
163
+ "epoch": 0.5,
164
+ "learning_rate": 9.503386004514672e-05,
165
+ "loss": 0.0894,
166
+ "step": 220
167
+ },
168
+ {
169
+ "epoch": 0.52,
170
+ "learning_rate": 9.480812641083521e-05,
171
+ "loss": 0.0892,
172
+ "step": 230
173
+ },
174
+ {
175
+ "epoch": 0.54,
176
+ "learning_rate": 9.45823927765237e-05,
177
+ "loss": 0.0834,
178
+ "step": 240
179
+ },
180
+ {
181
+ "epoch": 0.56,
182
+ "learning_rate": 9.43566591422122e-05,
183
+ "loss": 0.0848,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.59,
188
+ "learning_rate": 9.413092550790069e-05,
189
+ "loss": 0.0866,
190
+ "step": 260
191
+ },
192
+ {
193
+ "epoch": 0.61,
194
+ "learning_rate": 9.390519187358918e-05,
195
+ "loss": 0.0843,
196
+ "step": 270
197
+ },
198
+ {
199
+ "epoch": 0.63,
200
+ "learning_rate": 9.367945823927767e-05,
201
+ "loss": 0.0739,
202
+ "step": 280
203
+ },
204
+ {
205
+ "epoch": 0.65,
206
+ "learning_rate": 9.345372460496615e-05,
207
+ "loss": 0.082,
208
+ "step": 290
209
+ },
210
+ {
211
+ "epoch": 0.68,
212
+ "learning_rate": 9.322799097065464e-05,
213
+ "loss": 0.0868,
214
+ "step": 300
215
+ },
216
+ {
217
+ "epoch": 0.68,
218
+ "eval_accuracy": 0.9827956989247312,
219
+ "eval_f1": 0.0,
220
+ "eval_loss": 0.08376671373844147,
221
+ "eval_precision": 0.0,
222
+ "eval_recall": 0.0,
223
+ "eval_roc_auc_score": 0.6519909908369803,
224
+ "eval_runtime": 5.0804,
225
+ "eval_samples_per_second": 36.612,
226
+ "eval_steps_per_second": 2.362,
227
+ "step": 300
228
+ },
229
+ {
230
+ "epoch": 0.7,
231
+ "learning_rate": 9.300225733634313e-05,
232
+ "loss": 0.0759,
233
+ "step": 310
234
+ },
235
+ {
236
+ "epoch": 0.72,
237
+ "learning_rate": 9.27765237020316e-05,
238
+ "loss": 0.0879,
239
+ "step": 320
240
+ },
241
+ {
242
+ "epoch": 0.74,
243
+ "learning_rate": 9.25507900677201e-05,
244
+ "loss": 0.0796,
245
+ "step": 330
246
+ },
247
+ {
248
+ "epoch": 0.77,
249
+ "learning_rate": 9.232505643340859e-05,
250
+ "loss": 0.0842,
251
+ "step": 340
252
+ },
253
+ {
254
+ "epoch": 0.79,
255
+ "learning_rate": 9.209932279909706e-05,
256
+ "loss": 0.0855,
257
+ "step": 350
258
+ },
259
+ {
260
+ "epoch": 0.81,
261
+ "learning_rate": 9.187358916478555e-05,
262
+ "loss": 0.0847,
263
+ "step": 360
264
+ },
265
+ {
266
+ "epoch": 0.84,
267
+ "learning_rate": 9.164785553047405e-05,
268
+ "loss": 0.0785,
269
+ "step": 370
270
+ },
271
+ {
272
+ "epoch": 0.86,
273
+ "learning_rate": 9.142212189616254e-05,
274
+ "loss": 0.0853,
275
+ "step": 380
276
+ },
277
+ {
278
+ "epoch": 0.88,
279
+ "learning_rate": 9.119638826185101e-05,
280
+ "loss": 0.0864,
281
+ "step": 390
282
+ },
283
+ {
284
+ "epoch": 0.9,
285
+ "learning_rate": 9.09706546275395e-05,
286
+ "loss": 0.0854,
287
+ "step": 400
288
+ },
289
+ {
290
+ "epoch": 0.9,
291
+ "eval_accuracy": 0.9827956989247312,
292
+ "eval_f1": 0.0,
293
+ "eval_loss": 0.08326727151870728,
294
+ "eval_precision": 0.0,
295
+ "eval_recall": 0.0,
296
+ "eval_roc_auc_score": 0.6522894727844638,
297
+ "eval_runtime": 5.1454,
298
+ "eval_samples_per_second": 36.149,
299
+ "eval_steps_per_second": 2.332,
300
+ "step": 400
301
+ },
302
+ {
303
+ "epoch": 0.93,
304
+ "learning_rate": 9.0744920993228e-05,
305
+ "loss": 0.0794,
306
+ "step": 410
307
+ },
308
+ {
309
+ "epoch": 0.95,
310
+ "learning_rate": 9.051918735891649e-05,
311
+ "loss": 0.0856,
312
+ "step": 420
313
+ },
314
+ {
315
+ "epoch": 0.97,
316
+ "learning_rate": 9.029345372460498e-05,
317
+ "loss": 0.0835,
318
+ "step": 430
319
+ },
320
+ {
321
+ "epoch": 0.99,
322
+ "learning_rate": 9.006772009029347e-05,
323
+ "loss": 0.0814,
324
+ "step": 440
325
+ },
326
+ {
327
+ "epoch": 1.02,
328
+ "learning_rate": 8.984198645598195e-05,
329
+ "loss": 0.0874,
330
+ "step": 450
331
+ },
332
+ {
333
+ "epoch": 1.04,
334
+ "learning_rate": 8.961625282167044e-05,
335
+ "loss": 0.0821,
336
+ "step": 460
337
+ },
338
+ {
339
+ "epoch": 1.06,
340
+ "learning_rate": 8.939051918735893e-05,
341
+ "loss": 0.0872,
342
+ "step": 470
343
+ },
344
+ {
345
+ "epoch": 1.08,
346
+ "learning_rate": 8.91647855530474e-05,
347
+ "loss": 0.0811,
348
+ "step": 480
349
+ },
350
+ {
351
+ "epoch": 1.11,
352
+ "learning_rate": 8.89390519187359e-05,
353
+ "loss": 0.0845,
354
+ "step": 490
355
+ },
356
+ {
357
+ "epoch": 1.13,
358
+ "learning_rate": 8.871331828442439e-05,
359
+ "loss": 0.0786,
360
+ "step": 500
361
+ },
362
+ {
363
+ "epoch": 1.13,
364
+ "eval_accuracy": 0.9833870967741936,
365
+ "eval_f1": 0.06646525679758308,
366
+ "eval_loss": 0.08231215178966522,
367
+ "eval_precision": 1.0,
368
+ "eval_recall": 0.034375,
369
+ "eval_roc_auc_score": 0.6500255573030636,
370
+ "eval_runtime": 5.105,
371
+ "eval_samples_per_second": 36.435,
372
+ "eval_steps_per_second": 2.351,
373
+ "step": 500
374
+ },
375
+ {
376
+ "epoch": 1.15,
377
+ "learning_rate": 8.848758465011288e-05,
378
+ "loss": 0.0816,
379
+ "step": 510
380
+ },
381
+ {
382
+ "epoch": 1.17,
383
+ "learning_rate": 8.826185101580135e-05,
384
+ "loss": 0.0815,
385
+ "step": 520
386
+ },
387
+ {
388
+ "epoch": 1.2,
389
+ "learning_rate": 8.803611738148985e-05,
390
+ "loss": 0.0833,
391
+ "step": 530
392
+ },
393
+ {
394
+ "epoch": 1.22,
395
+ "learning_rate": 8.781038374717834e-05,
396
+ "loss": 0.0811,
397
+ "step": 540
398
+ },
399
+ {
400
+ "epoch": 1.24,
401
+ "learning_rate": 8.758465011286681e-05,
402
+ "loss": 0.084,
403
+ "step": 550
404
+ },
405
+ {
406
+ "epoch": 1.26,
407
+ "learning_rate": 8.73589164785553e-05,
408
+ "loss": 0.0772,
409
+ "step": 560
410
+ },
411
+ {
412
+ "epoch": 1.29,
413
+ "learning_rate": 8.71331828442438e-05,
414
+ "loss": 0.0799,
415
+ "step": 570
416
+ },
417
+ {
418
+ "epoch": 1.31,
419
+ "learning_rate": 8.690744920993227e-05,
420
+ "loss": 0.0891,
421
+ "step": 580
422
+ },
423
+ {
424
+ "epoch": 1.33,
425
+ "learning_rate": 8.668171557562076e-05,
426
+ "loss": 0.083,
427
+ "step": 590
428
+ },
429
+ {
430
+ "epoch": 1.35,
431
+ "learning_rate": 8.645598194130925e-05,
432
+ "loss": 0.0808,
433
+ "step": 600
434
+ },
435
+ {
436
+ "epoch": 1.35,
437
+ "eval_accuracy": 0.9833870967741936,
438
+ "eval_f1": 0.06646525679758308,
439
+ "eval_loss": 0.08112546801567078,
440
+ "eval_precision": 1.0,
441
+ "eval_recall": 0.034375,
442
+ "eval_roc_auc_score": 0.6772842587527352,
443
+ "eval_runtime": 5.3759,
444
+ "eval_samples_per_second": 34.599,
445
+ "eval_steps_per_second": 2.232,
446
+ "step": 600
447
+ },
448
+ {
449
+ "epoch": 1.38,
450
+ "learning_rate": 8.623024830699775e-05,
451
+ "loss": 0.0802,
452
+ "step": 610
453
+ },
454
+ {
455
+ "epoch": 1.4,
456
+ "learning_rate": 8.600451467268624e-05,
457
+ "loss": 0.0831,
458
+ "step": 620
459
+ },
460
+ {
461
+ "epoch": 1.42,
462
+ "learning_rate": 8.577878103837473e-05,
463
+ "loss": 0.0825,
464
+ "step": 630
465
+ },
466
+ {
467
+ "epoch": 1.44,
468
+ "learning_rate": 8.555304740406322e-05,
469
+ "loss": 0.0838,
470
+ "step": 640
471
+ },
472
+ {
473
+ "epoch": 1.47,
474
+ "learning_rate": 8.53273137697517e-05,
475
+ "loss": 0.0796,
476
+ "step": 650
477
+ },
478
+ {
479
+ "epoch": 1.49,
480
+ "learning_rate": 8.510158013544019e-05,
481
+ "loss": 0.0845,
482
+ "step": 660
483
+ },
484
+ {
485
+ "epoch": 1.51,
486
+ "learning_rate": 8.487584650112868e-05,
487
+ "loss": 0.0782,
488
+ "step": 670
489
+ },
490
+ {
491
+ "epoch": 1.53,
492
+ "learning_rate": 8.465011286681715e-05,
493
+ "loss": 0.0804,
494
+ "step": 680
495
+ },
496
+ {
497
+ "epoch": 1.56,
498
+ "learning_rate": 8.442437923250564e-05,
499
+ "loss": 0.0756,
500
+ "step": 690
501
+ },
502
+ {
503
+ "epoch": 1.58,
504
+ "learning_rate": 8.419864559819414e-05,
505
+ "loss": 0.0788,
506
+ "step": 700
507
+ },
508
+ {
509
+ "epoch": 1.58,
510
+ "eval_accuracy": 0.9833870967741936,
511
+ "eval_f1": 0.06646525679758308,
512
+ "eval_loss": 0.08039283752441406,
513
+ "eval_precision": 1.0,
514
+ "eval_recall": 0.034375,
515
+ "eval_roc_auc_score": 0.6918820090262583,
516
+ "eval_runtime": 5.2446,
517
+ "eval_samples_per_second": 35.465,
518
+ "eval_steps_per_second": 2.288,
519
+ "step": 700
520
+ },
521
+ {
522
+ "epoch": 1.6,
523
+ "learning_rate": 8.397291196388263e-05,
524
+ "loss": 0.0787,
525
+ "step": 710
526
+ },
527
+ {
528
+ "epoch": 1.63,
529
+ "learning_rate": 8.37471783295711e-05,
530
+ "loss": 0.0868,
531
+ "step": 720
532
+ },
533
+ {
534
+ "epoch": 1.65,
535
+ "learning_rate": 8.35214446952596e-05,
536
+ "loss": 0.0743,
537
+ "step": 730
538
+ },
539
+ {
540
+ "epoch": 1.67,
541
+ "learning_rate": 8.329571106094809e-05,
542
+ "loss": 0.0756,
543
+ "step": 740
544
+ },
545
+ {
546
+ "epoch": 1.69,
547
+ "learning_rate": 8.306997742663656e-05,
548
+ "loss": 0.0831,
549
+ "step": 750
550
+ },
551
+ {
552
+ "epoch": 1.72,
553
+ "learning_rate": 8.284424379232505e-05,
554
+ "loss": 0.0736,
555
+ "step": 760
556
+ },
557
+ {
558
+ "epoch": 1.74,
559
+ "learning_rate": 8.261851015801354e-05,
560
+ "loss": 0.0723,
561
+ "step": 770
562
+ },
563
+ {
564
+ "epoch": 1.76,
565
+ "learning_rate": 8.239277652370204e-05,
566
+ "loss": 0.0815,
567
+ "step": 780
568
+ },
569
+ {
570
+ "epoch": 1.78,
571
+ "learning_rate": 8.216704288939053e-05,
572
+ "loss": 0.0798,
573
+ "step": 790
574
+ },
575
+ {
576
+ "epoch": 1.81,
577
+ "learning_rate": 8.194130925507902e-05,
578
+ "loss": 0.073,
579
+ "step": 800
580
+ },
581
+ {
582
+ "epoch": 1.81,
583
+ "eval_accuracy": 0.9833870967741936,
584
+ "eval_f1": 0.06646525679758308,
585
+ "eval_loss": 0.0777764692902565,
586
+ "eval_precision": 1.0,
587
+ "eval_recall": 0.034375,
588
+ "eval_roc_auc_score": 0.7327331783369804,
589
+ "eval_runtime": 5.0281,
590
+ "eval_samples_per_second": 36.992,
591
+ "eval_steps_per_second": 2.387,
592
+ "step": 800
593
+ },
594
+ {
595
+ "epoch": 1.83,
596
+ "learning_rate": 8.17155756207675e-05,
597
+ "loss": 0.078,
598
+ "step": 810
599
+ },
600
+ {
601
+ "epoch": 1.85,
602
+ "learning_rate": 8.148984198645599e-05,
603
+ "loss": 0.0786,
604
+ "step": 820
605
+ },
606
+ {
607
+ "epoch": 1.87,
608
+ "learning_rate": 8.126410835214448e-05,
609
+ "loss": 0.0748,
610
+ "step": 830
611
+ },
612
+ {
613
+ "epoch": 1.9,
614
+ "learning_rate": 8.103837471783297e-05,
615
+ "loss": 0.0704,
616
+ "step": 840
617
+ },
618
+ {
619
+ "epoch": 1.92,
620
+ "learning_rate": 8.081264108352144e-05,
621
+ "loss": 0.0752,
622
+ "step": 850
623
+ },
624
+ {
625
+ "epoch": 1.94,
626
+ "learning_rate": 8.058690744920994e-05,
627
+ "loss": 0.0778,
628
+ "step": 860
629
+ },
630
+ {
631
+ "epoch": 1.96,
632
+ "learning_rate": 8.036117381489843e-05,
633
+ "loss": 0.0763,
634
+ "step": 870
635
+ },
636
+ {
637
+ "epoch": 1.99,
638
+ "learning_rate": 8.01354401805869e-05,
639
+ "loss": 0.0763,
640
+ "step": 880
641
+ },
642
+ {
643
+ "epoch": 2.01,
644
+ "learning_rate": 7.99097065462754e-05,
645
+ "loss": 0.0697,
646
+ "step": 890
647
+ },
648
+ {
649
+ "epoch": 2.03,
650
+ "learning_rate": 7.968397291196389e-05,
651
+ "loss": 0.075,
652
+ "step": 900
653
+ },
654
+ {
655
+ "epoch": 2.03,
656
+ "eval_accuracy": 0.9834408602150537,
657
+ "eval_f1": 0.08333333333333333,
658
+ "eval_loss": 0.07575903832912445,
659
+ "eval_precision": 0.875,
660
+ "eval_recall": 0.04375,
661
+ "eval_roc_auc_score": 0.7530620897155361,
662
+ "eval_runtime": 5.2667,
663
+ "eval_samples_per_second": 35.316,
664
+ "eval_steps_per_second": 2.278,
665
+ "step": 900
666
+ },
667
+ {
668
+ "epoch": 2.05,
669
+ "learning_rate": 7.945823927765236e-05,
670
+ "loss": 0.0775,
671
+ "step": 910
672
+ },
673
+ {
674
+ "epoch": 2.08,
675
+ "learning_rate": 7.923250564334085e-05,
676
+ "loss": 0.0703,
677
+ "step": 920
678
+ },
679
+ {
680
+ "epoch": 2.1,
681
+ "learning_rate": 7.900677200902934e-05,
682
+ "loss": 0.0761,
683
+ "step": 930
684
+ },
685
+ {
686
+ "epoch": 2.12,
687
+ "learning_rate": 7.878103837471784e-05,
688
+ "loss": 0.0778,
689
+ "step": 940
690
+ },
691
+ {
692
+ "epoch": 2.14,
693
+ "learning_rate": 7.855530474040633e-05,
694
+ "loss": 0.073,
695
+ "step": 950
696
+ },
697
+ {
698
+ "epoch": 2.17,
699
+ "learning_rate": 7.832957110609482e-05,
700
+ "loss": 0.0764,
701
+ "step": 960
702
+ },
703
+ {
704
+ "epoch": 2.19,
705
+ "learning_rate": 7.810383747178331e-05,
706
+ "loss": 0.0738,
707
+ "step": 970
708
+ },
709
+ {
710
+ "epoch": 2.21,
711
+ "learning_rate": 7.787810383747179e-05,
712
+ "loss": 0.0689,
713
+ "step": 980
714
+ },
715
+ {
716
+ "epoch": 2.23,
717
+ "learning_rate": 7.765237020316028e-05,
718
+ "loss": 0.0743,
719
+ "step": 990
720
+ },
721
+ {
722
+ "epoch": 2.26,
723
+ "learning_rate": 7.742663656884877e-05,
724
+ "loss": 0.0713,
725
+ "step": 1000
726
+ },
727
+ {
728
+ "epoch": 2.26,
729
+ "eval_accuracy": 0.9833870967741936,
730
+ "eval_f1": 0.06646525679758308,
731
+ "eval_loss": 0.07375543564558029,
732
+ "eval_precision": 1.0,
733
+ "eval_recall": 0.034375,
734
+ "eval_roc_auc_score": 0.7666500273522976,
735
+ "eval_runtime": 5.2333,
736
+ "eval_samples_per_second": 35.541,
737
+ "eval_steps_per_second": 2.293,
738
+ "step": 1000
739
+ },
740
+ {
741
+ "epoch": 2.28,
742
+ "learning_rate": 7.720090293453724e-05,
743
+ "loss": 0.0743,
744
+ "step": 1010
745
+ },
746
+ {
747
+ "epoch": 2.3,
748
+ "learning_rate": 7.697516930022574e-05,
749
+ "loss": 0.0748,
750
+ "step": 1020
751
+ },
752
+ {
753
+ "epoch": 2.33,
754
+ "learning_rate": 7.674943566591423e-05,
755
+ "loss": 0.0747,
756
+ "step": 1030
757
+ },
758
+ {
759
+ "epoch": 2.35,
760
+ "learning_rate": 7.652370203160272e-05,
761
+ "loss": 0.0705,
762
+ "step": 1040
763
+ },
764
+ {
765
+ "epoch": 2.37,
766
+ "learning_rate": 7.62979683972912e-05,
767
+ "loss": 0.0761,
768
+ "step": 1050
769
+ },
770
+ {
771
+ "epoch": 2.39,
772
+ "learning_rate": 7.607223476297969e-05,
773
+ "loss": 0.0688,
774
+ "step": 1060
775
+ },
776
+ {
777
+ "epoch": 2.42,
778
+ "learning_rate": 7.584650112866818e-05,
779
+ "loss": 0.068,
780
+ "step": 1070
781
+ },
782
+ {
783
+ "epoch": 2.44,
784
+ "learning_rate": 7.562076749435665e-05,
785
+ "loss": 0.0714,
786
+ "step": 1080
787
+ },
788
+ {
789
+ "epoch": 2.46,
790
+ "learning_rate": 7.539503386004514e-05,
791
+ "loss": 0.0712,
792
+ "step": 1090
793
+ },
794
+ {
795
+ "epoch": 2.48,
796
+ "learning_rate": 7.516930022573364e-05,
797
+ "loss": 0.0684,
798
+ "step": 1100
799
+ },
800
+ {
801
+ "epoch": 2.48,
802
+ "eval_accuracy": 0.9834408602150537,
803
+ "eval_f1": 0.08333333333333333,
804
+ "eval_loss": 0.07070032507181168,
805
+ "eval_precision": 0.875,
806
+ "eval_recall": 0.04375,
807
+ "eval_roc_auc_score": 0.8081299576039388,
808
+ "eval_runtime": 5.305,
809
+ "eval_samples_per_second": 35.061,
810
+ "eval_steps_per_second": 2.262,
811
+ "step": 1100
812
+ },
813
+ {
814
+ "epoch": 2.51,
815
+ "learning_rate": 7.494356659142213e-05,
816
+ "loss": 0.0702,
817
+ "step": 1110
818
+ },
819
+ {
820
+ "epoch": 2.53,
821
+ "learning_rate": 7.471783295711062e-05,
822
+ "loss": 0.0683,
823
+ "step": 1120
824
+ },
825
+ {
826
+ "epoch": 2.55,
827
+ "learning_rate": 7.449209932279911e-05,
828
+ "loss": 0.0666,
829
+ "step": 1130
830
+ },
831
+ {
832
+ "epoch": 2.57,
833
+ "learning_rate": 7.42663656884876e-05,
834
+ "loss": 0.0665,
835
+ "step": 1140
836
+ },
837
+ {
838
+ "epoch": 2.6,
839
+ "learning_rate": 7.404063205417608e-05,
840
+ "loss": 0.0658,
841
+ "step": 1150
842
+ },
843
+ {
844
+ "epoch": 2.62,
845
+ "learning_rate": 7.381489841986457e-05,
846
+ "loss": 0.0746,
847
+ "step": 1160
848
+ },
849
+ {
850
+ "epoch": 2.64,
851
+ "learning_rate": 7.358916478555306e-05,
852
+ "loss": 0.0705,
853
+ "step": 1170
854
+ },
855
+ {
856
+ "epoch": 2.66,
857
+ "learning_rate": 7.336343115124154e-05,
858
+ "loss": 0.07,
859
+ "step": 1180
860
+ },
861
+ {
862
+ "epoch": 2.69,
863
+ "learning_rate": 7.313769751693003e-05,
864
+ "loss": 0.0718,
865
+ "step": 1190
866
+ },
867
+ {
868
+ "epoch": 2.71,
869
+ "learning_rate": 7.291196388261852e-05,
870
+ "loss": 0.0703,
871
+ "step": 1200
872
+ },
873
+ {
874
+ "epoch": 2.71,
875
+ "eval_accuracy": 0.9835483870967742,
876
+ "eval_f1": 0.08383233532934131,
877
+ "eval_loss": 0.06844215840101242,
878
+ "eval_precision": 1.0,
879
+ "eval_recall": 0.04375,
880
+ "eval_roc_auc_score": 0.8417177242888403,
881
+ "eval_runtime": 5.1331,
882
+ "eval_samples_per_second": 36.235,
883
+ "eval_steps_per_second": 2.338,
884
+ "step": 1200
885
+ },
886
+ {
887
+ "epoch": 2.73,
888
+ "learning_rate": 7.2686230248307e-05,
889
+ "loss": 0.0688,
890
+ "step": 1210
891
+ },
892
+ {
893
+ "epoch": 2.75,
894
+ "learning_rate": 7.246049661399549e-05,
895
+ "loss": 0.0708,
896
+ "step": 1220
897
+ },
898
+ {
899
+ "epoch": 2.78,
900
+ "learning_rate": 7.223476297968398e-05,
901
+ "loss": 0.0665,
902
+ "step": 1230
903
+ },
904
+ {
905
+ "epoch": 2.8,
906
+ "learning_rate": 7.200902934537245e-05,
907
+ "loss": 0.0685,
908
+ "step": 1240
909
+ },
910
+ {
911
+ "epoch": 2.82,
912
+ "learning_rate": 7.178329571106094e-05,
913
+ "loss": 0.0691,
914
+ "step": 1250
915
+ },
916
+ {
917
+ "epoch": 2.84,
918
+ "learning_rate": 7.155756207674944e-05,
919
+ "loss": 0.0636,
920
+ "step": 1260
921
+ },
922
+ {
923
+ "epoch": 2.87,
924
+ "learning_rate": 7.133182844243793e-05,
925
+ "loss": 0.0676,
926
+ "step": 1270
927
+ },
928
+ {
929
+ "epoch": 2.89,
930
+ "learning_rate": 7.110609480812642e-05,
931
+ "loss": 0.0615,
932
+ "step": 1280
933
+ },
934
+ {
935
+ "epoch": 2.91,
936
+ "learning_rate": 7.088036117381491e-05,
937
+ "loss": 0.0645,
938
+ "step": 1290
939
+ },
940
+ {
941
+ "epoch": 2.93,
942
+ "learning_rate": 7.06546275395034e-05,
943
+ "loss": 0.0685,
944
+ "step": 1300
945
+ },
946
+ {
947
+ "epoch": 2.93,
948
+ "eval_accuracy": 0.9839247311827957,
949
+ "eval_f1": 0.138328530259366,
950
+ "eval_loss": 0.06600068509578705,
951
+ "eval_precision": 0.8888888888888888,
952
+ "eval_recall": 0.075,
953
+ "eval_roc_auc_score": 0.8572521198030635,
954
+ "eval_runtime": 5.1885,
955
+ "eval_samples_per_second": 35.848,
956
+ "eval_steps_per_second": 2.313,
957
+ "step": 1300
958
+ },
959
+ {
960
+ "epoch": 2.96,
961
+ "learning_rate": 7.042889390519188e-05,
962
+ "loss": 0.0615,
963
+ "step": 1310
964
+ },
965
+ {
966
+ "epoch": 2.98,
967
+ "learning_rate": 7.020316027088037e-05,
968
+ "loss": 0.0686,
969
+ "step": 1320
970
+ },
971
+ {
972
+ "epoch": 3.0,
973
+ "learning_rate": 6.997742663656886e-05,
974
+ "loss": 0.0643,
975
+ "step": 1330
976
+ },
977
+ {
978
+ "epoch": 3.02,
979
+ "learning_rate": 6.975169300225734e-05,
980
+ "loss": 0.062,
981
+ "step": 1340
982
+ },
983
+ {
984
+ "epoch": 3.05,
985
+ "learning_rate": 6.952595936794583e-05,
986
+ "loss": 0.0603,
987
+ "step": 1350
988
+ },
989
+ {
990
+ "epoch": 3.07,
991
+ "learning_rate": 6.930022573363432e-05,
992
+ "loss": 0.061,
993
+ "step": 1360
994
+ },
995
+ {
996
+ "epoch": 3.09,
997
+ "learning_rate": 6.907449209932281e-05,
998
+ "loss": 0.0654,
999
+ "step": 1370
1000
+ },
1001
+ {
1002
+ "epoch": 3.12,
1003
+ "learning_rate": 6.884875846501129e-05,
1004
+ "loss": 0.0661,
1005
+ "step": 1380
1006
+ },
1007
+ {
1008
+ "epoch": 3.14,
1009
+ "learning_rate": 6.862302483069978e-05,
1010
+ "loss": 0.0641,
1011
+ "step": 1390
1012
+ },
1013
+ {
1014
+ "epoch": 3.16,
1015
+ "learning_rate": 6.839729119638827e-05,
1016
+ "loss": 0.0622,
1017
+ "step": 1400
1018
+ },
1019
+ {
1020
+ "epoch": 3.16,
1021
+ "eval_accuracy": 0.9839784946236559,
1022
+ "eval_f1": 0.14367816091954022,
1023
+ "eval_loss": 0.06484588235616684,
1024
+ "eval_precision": 0.8928571428571429,
1025
+ "eval_recall": 0.078125,
1026
+ "eval_roc_auc_score": 0.85113221416849,
1027
+ "eval_runtime": 5.2041,
1028
+ "eval_samples_per_second": 35.741,
1029
+ "eval_steps_per_second": 2.306,
1030
+ "step": 1400
1031
+ },
1032
+ {
1033
+ "epoch": 3.18,
1034
+ "learning_rate": 6.817155756207674e-05,
1035
+ "loss": 0.0588,
1036
+ "step": 1410
1037
+ },
1038
+ {
1039
+ "epoch": 3.21,
1040
+ "learning_rate": 6.794582392776524e-05,
1041
+ "loss": 0.0666,
1042
+ "step": 1420
1043
+ },
1044
+ {
1045
+ "epoch": 3.23,
1046
+ "learning_rate": 6.772009029345373e-05,
1047
+ "loss": 0.0618,
1048
+ "step": 1430
1049
+ },
1050
+ {
1051
+ "epoch": 3.25,
1052
+ "learning_rate": 6.749435665914222e-05,
1053
+ "loss": 0.0625,
1054
+ "step": 1440
1055
+ },
1056
+ {
1057
+ "epoch": 3.27,
1058
+ "learning_rate": 6.726862302483071e-05,
1059
+ "loss": 0.0605,
1060
+ "step": 1450
1061
+ },
1062
+ {
1063
+ "epoch": 3.3,
1064
+ "learning_rate": 6.70428893905192e-05,
1065
+ "loss": 0.0631,
1066
+ "step": 1460
1067
+ },
1068
+ {
1069
+ "epoch": 3.32,
1070
+ "learning_rate": 6.681715575620769e-05,
1071
+ "loss": 0.0583,
1072
+ "step": 1470
1073
+ },
1074
+ {
1075
+ "epoch": 3.34,
1076
+ "learning_rate": 6.659142212189617e-05,
1077
+ "loss": 0.0624,
1078
+ "step": 1480
1079
+ },
1080
+ {
1081
+ "epoch": 3.36,
1082
+ "learning_rate": 6.636568848758466e-05,
1083
+ "loss": 0.0649,
1084
+ "step": 1490
1085
+ },
1086
+ {
1087
+ "epoch": 3.39,
1088
+ "learning_rate": 6.613995485327315e-05,
1089
+ "loss": 0.0588,
1090
+ "step": 1500
1091
+ },
1092
+ {
1093
+ "epoch": 3.39,
1094
+ "eval_accuracy": 0.9840860215053764,
1095
+ "eval_f1": 0.14942528735632185,
1096
+ "eval_loss": 0.06257249414920807,
1097
+ "eval_precision": 0.9285714285714286,
1098
+ "eval_recall": 0.08125,
1099
+ "eval_roc_auc_score": 0.8762311952954048,
1100
+ "eval_runtime": 5.2626,
1101
+ "eval_samples_per_second": 35.344,
1102
+ "eval_steps_per_second": 2.28,
1103
+ "step": 1500
1104
+ },
1105
+ {
1106
+ "epoch": 3.41,
1107
+ "learning_rate": 6.591422121896163e-05,
1108
+ "loss": 0.0619,
1109
+ "step": 1510
1110
+ },
1111
+ {
1112
+ "epoch": 3.43,
1113
+ "learning_rate": 6.568848758465012e-05,
1114
+ "loss": 0.0587,
1115
+ "step": 1520
1116
+ },
1117
+ {
1118
+ "epoch": 3.45,
1119
+ "learning_rate": 6.546275395033861e-05,
1120
+ "loss": 0.0609,
1121
+ "step": 1530
1122
+ },
1123
+ {
1124
+ "epoch": 3.48,
1125
+ "learning_rate": 6.523702031602708e-05,
1126
+ "loss": 0.0595,
1127
+ "step": 1540
1128
+ },
1129
+ {
1130
+ "epoch": 3.5,
1131
+ "learning_rate": 6.501128668171558e-05,
1132
+ "loss": 0.056,
1133
+ "step": 1550
1134
+ },
1135
+ {
1136
+ "epoch": 3.52,
1137
+ "learning_rate": 6.478555304740407e-05,
1138
+ "loss": 0.06,
1139
+ "step": 1560
1140
+ },
1141
+ {
1142
+ "epoch": 3.54,
1143
+ "learning_rate": 6.455981941309256e-05,
1144
+ "loss": 0.0599,
1145
+ "step": 1570
1146
+ },
1147
+ {
1148
+ "epoch": 3.57,
1149
+ "learning_rate": 6.433408577878103e-05,
1150
+ "loss": 0.0627,
1151
+ "step": 1580
1152
+ },
1153
+ {
1154
+ "epoch": 3.59,
1155
+ "learning_rate": 6.410835214446953e-05,
1156
+ "loss": 0.0625,
1157
+ "step": 1590
1158
+ },
1159
+ {
1160
+ "epoch": 3.61,
1161
+ "learning_rate": 6.388261851015802e-05,
1162
+ "loss": 0.0628,
1163
+ "step": 1600
1164
+ },
1165
+ {
1166
+ "epoch": 3.61,
1167
+ "eval_accuracy": 0.984516129032258,
1168
+ "eval_f1": 0.21311475409836067,
1169
+ "eval_loss": 0.06195955350995064,
1170
+ "eval_precision": 0.8478260869565217,
1171
+ "eval_recall": 0.121875,
1172
+ "eval_roc_auc_score": 0.8683221074945295,
1173
+ "eval_runtime": 5.1808,
1174
+ "eval_samples_per_second": 35.902,
1175
+ "eval_steps_per_second": 2.316,
1176
+ "step": 1600
1177
+ },
1178
+ {
1179
+ "epoch": 3.63,
1180
+ "learning_rate": 6.365688487584651e-05,
1181
+ "loss": 0.0646,
1182
+ "step": 1610
1183
+ },
1184
+ {
1185
+ "epoch": 3.66,
1186
+ "learning_rate": 6.3431151241535e-05,
1187
+ "loss": 0.0644,
1188
+ "step": 1620
1189
+ },
1190
+ {
1191
+ "epoch": 3.68,
1192
+ "learning_rate": 6.320541760722349e-05,
1193
+ "loss": 0.067,
1194
+ "step": 1630
1195
+ },
1196
+ {
1197
+ "epoch": 3.7,
1198
+ "learning_rate": 6.297968397291197e-05,
1199
+ "loss": 0.0617,
1200
+ "step": 1640
1201
+ },
1202
+ {
1203
+ "epoch": 3.72,
1204
+ "learning_rate": 6.275395033860046e-05,
1205
+ "loss": 0.0619,
1206
+ "step": 1650
1207
+ },
1208
+ {
1209
+ "epoch": 3.75,
1210
+ "learning_rate": 6.252821670428895e-05,
1211
+ "loss": 0.0591,
1212
+ "step": 1660
1213
+ },
1214
+ {
1215
+ "epoch": 3.77,
1216
+ "learning_rate": 6.230248306997743e-05,
1217
+ "loss": 0.0601,
1218
+ "step": 1670
1219
+ },
1220
+ {
1221
+ "epoch": 3.79,
1222
+ "learning_rate": 6.207674943566592e-05,
1223
+ "loss": 0.057,
1224
+ "step": 1680
1225
+ },
1226
+ {
1227
+ "epoch": 3.81,
1228
+ "learning_rate": 6.185101580135441e-05,
1229
+ "loss": 0.0631,
1230
+ "step": 1690
1231
+ },
1232
+ {
1233
+ "epoch": 3.84,
1234
+ "learning_rate": 6.16252821670429e-05,
1235
+ "loss": 0.0593,
1236
+ "step": 1700
1237
+ },
1238
+ {
1239
+ "epoch": 3.84,
1240
+ "eval_accuracy": 0.9848924731182795,
1241
+ "eval_f1": 0.24258760107816713,
1242
+ "eval_loss": 0.060088034719228745,
1243
+ "eval_precision": 0.8823529411764706,
1244
+ "eval_recall": 0.140625,
1245
+ "eval_roc_auc_score": 0.8836091356673962,
1246
+ "eval_runtime": 5.2773,
1247
+ "eval_samples_per_second": 35.245,
1248
+ "eval_steps_per_second": 2.274,
1249
+ "step": 1700
1250
+ },
1251
+ {
1252
+ "epoch": 3.86,
1253
+ "learning_rate": 6.139954853273138e-05,
1254
+ "loss": 0.0624,
1255
+ "step": 1710
1256
+ },
1257
+ {
1258
+ "epoch": 3.88,
1259
+ "learning_rate": 6.117381489841987e-05,
1260
+ "loss": 0.0592,
1261
+ "step": 1720
1262
+ },
1263
+ {
1264
+ "epoch": 3.91,
1265
+ "learning_rate": 6.094808126410836e-05,
1266
+ "loss": 0.0615,
1267
+ "step": 1730
1268
+ },
1269
+ {
1270
+ "epoch": 3.93,
1271
+ "learning_rate": 6.072234762979684e-05,
1272
+ "loss": 0.0567,
1273
+ "step": 1740
1274
+ },
1275
+ {
1276
+ "epoch": 3.95,
1277
+ "learning_rate": 6.049661399548533e-05,
1278
+ "loss": 0.0586,
1279
+ "step": 1750
1280
+ },
1281
+ {
1282
+ "epoch": 3.97,
1283
+ "learning_rate": 6.027088036117382e-05,
1284
+ "loss": 0.059,
1285
+ "step": 1760
1286
+ },
1287
+ {
1288
+ "epoch": 4.0,
1289
+ "learning_rate": 6.00451467268623e-05,
1290
+ "loss": 0.0536,
1291
+ "step": 1770
1292
+ },
1293
+ {
1294
+ "epoch": 4.02,
1295
+ "learning_rate": 5.981941309255079e-05,
1296
+ "loss": 0.0558,
1297
+ "step": 1780
1298
+ },
1299
+ {
1300
+ "epoch": 4.04,
1301
+ "learning_rate": 5.959367945823928e-05,
1302
+ "loss": 0.0552,
1303
+ "step": 1790
1304
+ },
1305
+ {
1306
+ "epoch": 4.06,
1307
+ "learning_rate": 5.936794582392777e-05,
1308
+ "loss": 0.0556,
1309
+ "step": 1800
1310
+ },
1311
+ {
1312
+ "epoch": 4.06,
1313
+ "eval_accuracy": 0.9850537634408603,
1314
+ "eval_f1": 0.2606382978723405,
1315
+ "eval_loss": 0.05858895182609558,
1316
+ "eval_precision": 0.875,
1317
+ "eval_recall": 0.153125,
1318
+ "eval_roc_auc_score": 0.8913255607221007,
1319
+ "eval_runtime": 5.2674,
1320
+ "eval_samples_per_second": 35.312,
1321
+ "eval_steps_per_second": 2.278,
1322
+ "step": 1800
1323
+ },
1324
+ {
1325
+ "epoch": 4.09,
1326
+ "learning_rate": 5.914221218961625e-05,
1327
+ "loss": 0.0566,
1328
+ "step": 1810
1329
+ },
1330
+ {
1331
+ "epoch": 4.11,
1332
+ "learning_rate": 5.891647855530474e-05,
1333
+ "loss": 0.0498,
1334
+ "step": 1820
1335
+ },
1336
+ {
1337
+ "epoch": 4.13,
1338
+ "learning_rate": 5.869074492099323e-05,
1339
+ "loss": 0.0561,
1340
+ "step": 1830
1341
+ },
1342
+ {
1343
+ "epoch": 4.15,
1344
+ "learning_rate": 5.8465011286681716e-05,
1345
+ "loss": 0.0582,
1346
+ "step": 1840
1347
+ },
1348
+ {
1349
+ "epoch": 4.18,
1350
+ "learning_rate": 5.823927765237021e-05,
1351
+ "loss": 0.055,
1352
+ "step": 1850
1353
+ },
1354
+ {
1355
+ "epoch": 4.2,
1356
+ "learning_rate": 5.80135440180587e-05,
1357
+ "loss": 0.0523,
1358
+ "step": 1860
1359
+ },
1360
+ {
1361
+ "epoch": 4.22,
1362
+ "learning_rate": 5.7787810383747176e-05,
1363
+ "loss": 0.0593,
1364
+ "step": 1870
1365
+ },
1366
+ {
1367
+ "epoch": 4.24,
1368
+ "learning_rate": 5.7562076749435666e-05,
1369
+ "loss": 0.0534,
1370
+ "step": 1880
1371
+ },
1372
+ {
1373
+ "epoch": 4.27,
1374
+ "learning_rate": 5.733634311512416e-05,
1375
+ "loss": 0.057,
1376
+ "step": 1890
1377
+ },
1378
+ {
1379
+ "epoch": 4.29,
1380
+ "learning_rate": 5.711060948081265e-05,
1381
+ "loss": 0.0583,
1382
+ "step": 1900
1383
+ },
1384
+ {
1385
+ "epoch": 4.29,
1386
+ "eval_accuracy": 0.984516129032258,
1387
+ "eval_f1": 0.19553072625698326,
1388
+ "eval_loss": 0.057477883994579315,
1389
+ "eval_precision": 0.9210526315789473,
1390
+ "eval_recall": 0.109375,
1391
+ "eval_roc_auc_score": 0.9017182371444201,
1392
+ "eval_runtime": 5.1921,
1393
+ "eval_samples_per_second": 35.824,
1394
+ "eval_steps_per_second": 2.311,
1395
+ "step": 1900
1396
+ },
1397
+ {
1398
+ "epoch": 4.31,
1399
+ "learning_rate": 5.688487584650113e-05,
1400
+ "loss": 0.0634,
1401
+ "step": 1910
1402
+ },
1403
+ {
1404
+ "epoch": 4.33,
1405
+ "learning_rate": 5.665914221218962e-05,
1406
+ "loss": 0.0534,
1407
+ "step": 1920
1408
+ },
1409
+ {
1410
+ "epoch": 4.36,
1411
+ "learning_rate": 5.6433408577878114e-05,
1412
+ "loss": 0.054,
1413
+ "step": 1930
1414
+ },
1415
+ {
1416
+ "epoch": 4.38,
1417
+ "learning_rate": 5.620767494356659e-05,
1418
+ "loss": 0.0538,
1419
+ "step": 1940
1420
+ },
1421
+ {
1422
+ "epoch": 4.4,
1423
+ "learning_rate": 5.598194130925508e-05,
1424
+ "loss": 0.0586,
1425
+ "step": 1950
1426
+ },
1427
+ {
1428
+ "epoch": 4.42,
1429
+ "learning_rate": 5.575620767494357e-05,
1430
+ "loss": 0.0509,
1431
+ "step": 1960
1432
+ },
1433
+ {
1434
+ "epoch": 4.45,
1435
+ "learning_rate": 5.553047404063205e-05,
1436
+ "loss": 0.0511,
1437
+ "step": 1970
1438
+ },
1439
+ {
1440
+ "epoch": 4.47,
1441
+ "learning_rate": 5.530474040632054e-05,
1442
+ "loss": 0.0556,
1443
+ "step": 1980
1444
+ },
1445
+ {
1446
+ "epoch": 4.49,
1447
+ "learning_rate": 5.507900677200903e-05,
1448
+ "loss": 0.053,
1449
+ "step": 1990
1450
+ },
1451
+ {
1452
+ "epoch": 4.51,
1453
+ "learning_rate": 5.485327313769752e-05,
1454
+ "loss": 0.0559,
1455
+ "step": 2000
1456
+ },
1457
+ {
1458
+ "epoch": 4.51,
1459
+ "eval_accuracy": 0.9845698924731183,
1460
+ "eval_f1": 0.22641509433962267,
1461
+ "eval_loss": 0.056463927030563354,
1462
+ "eval_precision": 0.8235294117647058,
1463
+ "eval_recall": 0.13125,
1464
+ "eval_roc_auc_score": 0.9055733725382933,
1465
+ "eval_runtime": 5.1748,
1466
+ "eval_samples_per_second": 35.944,
1467
+ "eval_steps_per_second": 2.319,
1468
+ "step": 2000
1469
+ },
1470
+ {
1471
+ "epoch": 4.54,
1472
+ "learning_rate": 5.462753950338601e-05,
1473
+ "loss": 0.0566,
1474
+ "step": 2010
1475
+ },
1476
+ {
1477
+ "epoch": 4.56,
1478
+ "learning_rate": 5.44018058690745e-05,
1479
+ "loss": 0.0504,
1480
+ "step": 2020
1481
+ },
1482
+ {
1483
+ "epoch": 4.58,
1484
+ "learning_rate": 5.417607223476299e-05,
1485
+ "loss": 0.0538,
1486
+ "step": 2030
1487
+ },
1488
+ {
1489
+ "epoch": 4.6,
1490
+ "learning_rate": 5.3950338600451466e-05,
1491
+ "loss": 0.05,
1492
+ "step": 2040
1493
+ },
1494
+ {
1495
+ "epoch": 4.63,
1496
+ "learning_rate": 5.372460496613996e-05,
1497
+ "loss": 0.0539,
1498
+ "step": 2050
1499
+ },
1500
+ {
1501
+ "epoch": 4.65,
1502
+ "learning_rate": 5.349887133182845e-05,
1503
+ "loss": 0.0557,
1504
+ "step": 2060
1505
+ },
1506
+ {
1507
+ "epoch": 4.67,
1508
+ "learning_rate": 5.3273137697516925e-05,
1509
+ "loss": 0.057,
1510
+ "step": 2070
1511
+ },
1512
+ {
1513
+ "epoch": 4.7,
1514
+ "learning_rate": 5.3047404063205416e-05,
1515
+ "loss": 0.0578,
1516
+ "step": 2080
1517
+ },
1518
+ {
1519
+ "epoch": 4.72,
1520
+ "learning_rate": 5.282167042889391e-05,
1521
+ "loss": 0.0574,
1522
+ "step": 2090
1523
+ },
1524
+ {
1525
+ "epoch": 4.74,
1526
+ "learning_rate": 5.259593679458239e-05,
1527
+ "loss": 0.0502,
1528
+ "step": 2100
1529
+ },
1530
+ {
1531
+ "epoch": 4.74,
1532
+ "eval_accuracy": 0.9849462365591398,
1533
+ "eval_f1": 0.2553191489361702,
1534
+ "eval_loss": 0.0548568032681942,
1535
+ "eval_precision": 0.8571428571428571,
1536
+ "eval_recall": 0.15,
1537
+ "eval_roc_auc_score": 0.9057971485229759,
1538
+ "eval_runtime": 5.1983,
1539
+ "eval_samples_per_second": 35.781,
1540
+ "eval_steps_per_second": 2.308,
1541
+ "step": 2100
1542
+ },
1543
+ {
1544
+ "epoch": 4.76,
1545
+ "learning_rate": 5.237020316027088e-05,
1546
+ "loss": 0.0487,
1547
+ "step": 2110
1548
+ },
1549
+ {
1550
+ "epoch": 4.79,
1551
+ "learning_rate": 5.214446952595937e-05,
1552
+ "loss": 0.0524,
1553
+ "step": 2120
1554
+ },
1555
+ {
1556
+ "epoch": 4.81,
1557
+ "learning_rate": 5.1918735891647864e-05,
1558
+ "loss": 0.0517,
1559
+ "step": 2130
1560
+ },
1561
+ {
1562
+ "epoch": 4.83,
1563
+ "learning_rate": 5.169300225733634e-05,
1564
+ "loss": 0.0538,
1565
+ "step": 2140
1566
+ },
1567
+ {
1568
+ "epoch": 4.85,
1569
+ "learning_rate": 5.146726862302483e-05,
1570
+ "loss": 0.0601,
1571
+ "step": 2150
1572
+ },
1573
+ {
1574
+ "epoch": 4.88,
1575
+ "learning_rate": 5.124153498871332e-05,
1576
+ "loss": 0.056,
1577
+ "step": 2160
1578
+ },
1579
+ {
1580
+ "epoch": 4.9,
1581
+ "learning_rate": 5.101580135440181e-05,
1582
+ "loss": 0.0548,
1583
+ "step": 2170
1584
+ },
1585
+ {
1586
+ "epoch": 4.92,
1587
+ "learning_rate": 5.07900677200903e-05,
1588
+ "loss": 0.0561,
1589
+ "step": 2180
1590
+ },
1591
+ {
1592
+ "epoch": 4.94,
1593
+ "learning_rate": 5.056433408577879e-05,
1594
+ "loss": 0.0521,
1595
+ "step": 2190
1596
+ },
1597
+ {
1598
+ "epoch": 4.97,
1599
+ "learning_rate": 5.0338600451467266e-05,
1600
+ "loss": 0.058,
1601
+ "step": 2200
1602
+ },
1603
+ {
1604
+ "epoch": 4.97,
1605
+ "eval_accuracy": 0.985268817204301,
1606
+ "eval_f1": 0.2751322751322751,
1607
+ "eval_loss": 0.0537453219294548,
1608
+ "eval_precision": 0.896551724137931,
1609
+ "eval_recall": 0.1625,
1610
+ "eval_roc_auc_score": 0.9138139359956235,
1611
+ "eval_runtime": 5.1845,
1612
+ "eval_samples_per_second": 35.876,
1613
+ "eval_steps_per_second": 2.315,
1614
+ "step": 2200
1615
+ },
1616
+ {
1617
+ "epoch": 4.99,
1618
+ "learning_rate": 5.011286681715576e-05,
1619
+ "loss": 0.0533,
1620
+ "step": 2210
1621
+ },
1622
+ {
1623
+ "epoch": 5.01,
1624
+ "learning_rate": 4.988713318284425e-05,
1625
+ "loss": 0.0528,
1626
+ "step": 2220
1627
+ },
1628
+ {
1629
+ "epoch": 5.03,
1630
+ "learning_rate": 4.966139954853273e-05,
1631
+ "loss": 0.0527,
1632
+ "step": 2230
1633
+ },
1634
+ {
1635
+ "epoch": 5.06,
1636
+ "learning_rate": 4.9435665914221216e-05,
1637
+ "loss": 0.0566,
1638
+ "step": 2240
1639
+ },
1640
+ {
1641
+ "epoch": 5.08,
1642
+ "learning_rate": 4.920993227990971e-05,
1643
+ "loss": 0.0485,
1644
+ "step": 2250
1645
+ },
1646
+ {
1647
+ "epoch": 5.1,
1648
+ "learning_rate": 4.89841986455982e-05,
1649
+ "loss": 0.0516,
1650
+ "step": 2260
1651
+ },
1652
+ {
1653
+ "epoch": 5.12,
1654
+ "learning_rate": 4.875846501128669e-05,
1655
+ "loss": 0.0514,
1656
+ "step": 2270
1657
+ },
1658
+ {
1659
+ "epoch": 5.15,
1660
+ "learning_rate": 4.853273137697517e-05,
1661
+ "loss": 0.0476,
1662
+ "step": 2280
1663
+ },
1664
+ {
1665
+ "epoch": 5.17,
1666
+ "learning_rate": 4.830699774266366e-05,
1667
+ "loss": 0.0502,
1668
+ "step": 2290
1669
+ },
1670
+ {
1671
+ "epoch": 5.19,
1672
+ "learning_rate": 4.808126410835215e-05,
1673
+ "loss": 0.0527,
1674
+ "step": 2300
1675
+ },
1676
+ {
1677
+ "epoch": 5.19,
1678
+ "eval_accuracy": 0.9853225806451613,
1679
+ "eval_f1": 0.28720626631853785,
1680
+ "eval_loss": 0.05435943603515625,
1681
+ "eval_precision": 0.873015873015873,
1682
+ "eval_recall": 0.171875,
1683
+ "eval_roc_auc_score": 0.9080290960065646,
1684
+ "eval_runtime": 5.1704,
1685
+ "eval_samples_per_second": 35.974,
1686
+ "eval_steps_per_second": 2.321,
1687
+ "step": 2300
1688
+ },
1689
+ {
1690
+ "epoch": 5.21,
1691
+ "learning_rate": 4.785553047404063e-05,
1692
+ "loss": 0.0516,
1693
+ "step": 2310
1694
+ },
1695
+ {
1696
+ "epoch": 5.24,
1697
+ "learning_rate": 4.762979683972912e-05,
1698
+ "loss": 0.0457,
1699
+ "step": 2320
1700
+ },
1701
+ {
1702
+ "epoch": 5.26,
1703
+ "learning_rate": 4.740406320541761e-05,
1704
+ "loss": 0.0519,
1705
+ "step": 2330
1706
+ },
1707
+ {
1708
+ "epoch": 5.28,
1709
+ "learning_rate": 4.71783295711061e-05,
1710
+ "loss": 0.0523,
1711
+ "step": 2340
1712
+ },
1713
+ {
1714
+ "epoch": 5.3,
1715
+ "learning_rate": 4.695259593679459e-05,
1716
+ "loss": 0.0491,
1717
+ "step": 2350
1718
+ },
1719
+ {
1720
+ "epoch": 5.33,
1721
+ "learning_rate": 4.672686230248307e-05,
1722
+ "loss": 0.0511,
1723
+ "step": 2360
1724
+ },
1725
+ {
1726
+ "epoch": 5.35,
1727
+ "learning_rate": 4.6501128668171564e-05,
1728
+ "loss": 0.0491,
1729
+ "step": 2370
1730
+ },
1731
+ {
1732
+ "epoch": 5.37,
1733
+ "learning_rate": 4.627539503386005e-05,
1734
+ "loss": 0.0499,
1735
+ "step": 2380
1736
+ },
1737
+ {
1738
+ "epoch": 5.4,
1739
+ "learning_rate": 4.604966139954853e-05,
1740
+ "loss": 0.0471,
1741
+ "step": 2390
1742
+ },
1743
+ {
1744
+ "epoch": 5.42,
1745
+ "learning_rate": 4.582392776523702e-05,
1746
+ "loss": 0.0442,
1747
+ "step": 2400
1748
+ },
1749
+ {
1750
+ "epoch": 5.42,
1751
+ "eval_accuracy": 0.9856451612903225,
1752
+ "eval_f1": 0.30649350649350654,
1753
+ "eval_loss": 0.052230026572942734,
1754
+ "eval_precision": 0.9076923076923077,
1755
+ "eval_recall": 0.184375,
1756
+ "eval_roc_auc_score": 0.9181778241247265,
1757
+ "eval_runtime": 5.2656,
1758
+ "eval_samples_per_second": 35.324,
1759
+ "eval_steps_per_second": 2.279,
1760
+ "step": 2400
1761
+ },
1762
+ {
1763
+ "epoch": 5.44,
1764
+ "learning_rate": 4.559819413092551e-05,
1765
+ "loss": 0.048,
1766
+ "step": 2410
1767
+ },
1768
+ {
1769
+ "epoch": 5.46,
1770
+ "learning_rate": 4.5372460496614e-05,
1771
+ "loss": 0.0511,
1772
+ "step": 2420
1773
+ },
1774
+ {
1775
+ "epoch": 5.49,
1776
+ "learning_rate": 4.514672686230249e-05,
1777
+ "loss": 0.052,
1778
+ "step": 2430
1779
+ },
1780
+ {
1781
+ "epoch": 5.51,
1782
+ "learning_rate": 4.492099322799097e-05,
1783
+ "loss": 0.0503,
1784
+ "step": 2440
1785
+ },
1786
+ {
1787
+ "epoch": 5.53,
1788
+ "learning_rate": 4.4695259593679463e-05,
1789
+ "loss": 0.0489,
1790
+ "step": 2450
1791
+ },
1792
+ {
1793
+ "epoch": 5.55,
1794
+ "learning_rate": 4.446952595936795e-05,
1795
+ "loss": 0.0502,
1796
+ "step": 2460
1797
+ },
1798
+ {
1799
+ "epoch": 5.58,
1800
+ "learning_rate": 4.424379232505644e-05,
1801
+ "loss": 0.0477,
1802
+ "step": 2470
1803
+ },
1804
+ {
1805
+ "epoch": 5.6,
1806
+ "learning_rate": 4.401805869074492e-05,
1807
+ "loss": 0.0527,
1808
+ "step": 2480
1809
+ },
1810
+ {
1811
+ "epoch": 5.62,
1812
+ "learning_rate": 4.379232505643341e-05,
1813
+ "loss": 0.0481,
1814
+ "step": 2490
1815
+ },
1816
+ {
1817
+ "epoch": 5.64,
1818
+ "learning_rate": 4.35665914221219e-05,
1819
+ "loss": 0.0431,
1820
+ "step": 2500
1821
+ },
1822
+ {
1823
+ "epoch": 5.64,
1824
+ "eval_accuracy": 0.9854838709677419,
1825
+ "eval_f1": 0.296875,
1826
+ "eval_loss": 0.05206151306629181,
1827
+ "eval_precision": 0.890625,
1828
+ "eval_recall": 0.178125,
1829
+ "eval_roc_auc_score": 0.9214854007111598,
1830
+ "eval_runtime": 5.1881,
1831
+ "eval_samples_per_second": 35.851,
1832
+ "eval_steps_per_second": 2.313,
1833
+ "step": 2500
1834
+ },
1835
+ {
1836
+ "epoch": 5.67,
1837
+ "learning_rate": 4.334085778781038e-05,
1838
+ "loss": 0.0488,
1839
+ "step": 2510
1840
+ },
1841
+ {
1842
+ "epoch": 5.69,
1843
+ "learning_rate": 4.311512415349887e-05,
1844
+ "loss": 0.0491,
1845
+ "step": 2520
1846
+ },
1847
+ {
1848
+ "epoch": 5.71,
1849
+ "learning_rate": 4.2889390519187363e-05,
1850
+ "loss": 0.0489,
1851
+ "step": 2530
1852
+ },
1853
+ {
1854
+ "epoch": 5.73,
1855
+ "learning_rate": 4.266365688487585e-05,
1856
+ "loss": 0.0471,
1857
+ "step": 2540
1858
+ },
1859
+ {
1860
+ "epoch": 5.76,
1861
+ "learning_rate": 4.243792325056434e-05,
1862
+ "loss": 0.0477,
1863
+ "step": 2550
1864
+ },
1865
+ {
1866
+ "epoch": 5.78,
1867
+ "learning_rate": 4.221218961625282e-05,
1868
+ "loss": 0.0507,
1869
+ "step": 2560
1870
+ },
1871
+ {
1872
+ "epoch": 5.8,
1873
+ "learning_rate": 4.198645598194131e-05,
1874
+ "loss": 0.0533,
1875
+ "step": 2570
1876
+ },
1877
+ {
1878
+ "epoch": 5.82,
1879
+ "learning_rate": 4.17607223476298e-05,
1880
+ "loss": 0.0536,
1881
+ "step": 2580
1882
+ },
1883
+ {
1884
+ "epoch": 5.85,
1885
+ "learning_rate": 4.153498871331828e-05,
1886
+ "loss": 0.05,
1887
+ "step": 2590
1888
+ },
1889
+ {
1890
+ "epoch": 5.87,
1891
+ "learning_rate": 4.130925507900677e-05,
1892
+ "loss": 0.0482,
1893
+ "step": 2600
1894
+ },
1895
+ {
1896
+ "epoch": 5.87,
1897
+ "eval_accuracy": 0.9855913978494624,
1898
+ "eval_f1": 0.30927835051546393,
1899
+ "eval_loss": 0.051463063806295395,
1900
+ "eval_precision": 0.8823529411764706,
1901
+ "eval_recall": 0.1875,
1902
+ "eval_roc_auc_score": 0.9182954390043764,
1903
+ "eval_runtime": 5.1963,
1904
+ "eval_samples_per_second": 35.794,
1905
+ "eval_steps_per_second": 2.309,
1906
+ "step": 2600
1907
+ },
1908
+ {
1909
+ "epoch": 5.89,
1910
+ "learning_rate": 4.108352144469526e-05,
1911
+ "loss": 0.0468,
1912
+ "step": 2610
1913
+ },
1914
+ {
1915
+ "epoch": 5.91,
1916
+ "learning_rate": 4.085778781038375e-05,
1917
+ "loss": 0.047,
1918
+ "step": 2620
1919
+ },
1920
+ {
1921
+ "epoch": 5.94,
1922
+ "learning_rate": 4.063205417607224e-05,
1923
+ "loss": 0.0485,
1924
+ "step": 2630
1925
+ },
1926
+ {
1927
+ "epoch": 5.96,
1928
+ "learning_rate": 4.040632054176072e-05,
1929
+ "loss": 0.0485,
1930
+ "step": 2640
1931
+ },
1932
+ {
1933
+ "epoch": 5.98,
1934
+ "learning_rate": 4.018058690744921e-05,
1935
+ "loss": 0.0513,
1936
+ "step": 2650
1937
+ },
1938
+ {
1939
+ "epoch": 6.0,
1940
+ "learning_rate": 3.99548532731377e-05,
1941
+ "loss": 0.0521,
1942
+ "step": 2660
1943
+ },
1944
+ {
1945
+ "epoch": 6.03,
1946
+ "learning_rate": 3.972911963882618e-05,
1947
+ "loss": 0.0456,
1948
+ "step": 2670
1949
+ },
1950
+ {
1951
+ "epoch": 6.05,
1952
+ "learning_rate": 3.950338600451467e-05,
1953
+ "loss": 0.0501,
1954
+ "step": 2680
1955
+ },
1956
+ {
1957
+ "epoch": 6.07,
1958
+ "learning_rate": 3.927765237020316e-05,
1959
+ "loss": 0.0459,
1960
+ "step": 2690
1961
+ },
1962
+ {
1963
+ "epoch": 6.09,
1964
+ "learning_rate": 3.9051918735891654e-05,
1965
+ "loss": 0.048,
1966
+ "step": 2700
1967
+ },
1968
+ {
1969
+ "epoch": 6.09,
1970
+ "eval_accuracy": 0.986021505376344,
1971
+ "eval_f1": 0.3434343434343434,
1972
+ "eval_loss": 0.050363171845674515,
1973
+ "eval_precision": 0.8947368421052632,
1974
+ "eval_recall": 0.2125,
1975
+ "eval_roc_auc_score": 0.9214696731400437,
1976
+ "eval_runtime": 5.2157,
1977
+ "eval_samples_per_second": 35.662,
1978
+ "eval_steps_per_second": 2.301,
1979
+ "step": 2700
1980
+ },
1981
+ {
1982
+ "epoch": 6.12,
1983
+ "learning_rate": 3.882618510158014e-05,
1984
+ "loss": 0.0508,
1985
+ "step": 2710
1986
+ },
1987
+ {
1988
+ "epoch": 6.14,
1989
+ "learning_rate": 3.860045146726862e-05,
1990
+ "loss": 0.0496,
1991
+ "step": 2720
1992
+ },
1993
+ {
1994
+ "epoch": 6.16,
1995
+ "learning_rate": 3.837471783295711e-05,
1996
+ "loss": 0.045,
1997
+ "step": 2730
1998
+ },
1999
+ {
2000
+ "epoch": 6.19,
2001
+ "learning_rate": 3.81489841986456e-05,
2002
+ "loss": 0.0438,
2003
+ "step": 2740
2004
+ },
2005
+ {
2006
+ "epoch": 6.21,
2007
+ "learning_rate": 3.792325056433409e-05,
2008
+ "loss": 0.0476,
2009
+ "step": 2750
2010
+ },
2011
+ {
2012
+ "epoch": 6.23,
2013
+ "learning_rate": 3.769751693002257e-05,
2014
+ "loss": 0.0496,
2015
+ "step": 2760
2016
+ },
2017
+ {
2018
+ "epoch": 6.25,
2019
+ "learning_rate": 3.747178329571106e-05,
2020
+ "loss": 0.0458,
2021
+ "step": 2770
2022
+ },
2023
+ {
2024
+ "epoch": 6.28,
2025
+ "learning_rate": 3.7246049661399554e-05,
2026
+ "loss": 0.0454,
2027
+ "step": 2780
2028
+ },
2029
+ {
2030
+ "epoch": 6.3,
2031
+ "learning_rate": 3.702031602708804e-05,
2032
+ "loss": 0.0472,
2033
+ "step": 2790
2034
+ },
2035
+ {
2036
+ "epoch": 6.32,
2037
+ "learning_rate": 3.679458239277653e-05,
2038
+ "loss": 0.0423,
2039
+ "step": 2800
2040
+ },
2041
+ {
2042
+ "epoch": 6.32,
2043
+ "eval_accuracy": 0.9862365591397849,
2044
+ "eval_f1": 0.3694581280788177,
2045
+ "eval_loss": 0.04973728582262993,
2046
+ "eval_precision": 0.872093023255814,
2047
+ "eval_recall": 0.234375,
2048
+ "eval_roc_auc_score": 0.9257974049507658,
2049
+ "eval_runtime": 5.164,
2050
+ "eval_samples_per_second": 36.019,
2051
+ "eval_steps_per_second": 2.324,
2052
+ "step": 2800
2053
+ },
2054
+ {
2055
+ "epoch": 6.34,
2056
+ "learning_rate": 3.656884875846501e-05,
2057
+ "loss": 0.042,
2058
+ "step": 2810
2059
+ },
2060
+ {
2061
+ "epoch": 6.37,
2062
+ "learning_rate": 3.63431151241535e-05,
2063
+ "loss": 0.0479,
2064
+ "step": 2820
2065
+ },
2066
+ {
2067
+ "epoch": 6.39,
2068
+ "learning_rate": 3.611738148984199e-05,
2069
+ "loss": 0.0437,
2070
+ "step": 2830
2071
+ },
2072
+ {
2073
+ "epoch": 6.41,
2074
+ "learning_rate": 3.589164785553047e-05,
2075
+ "loss": 0.0463,
2076
+ "step": 2840
2077
+ },
2078
+ {
2079
+ "epoch": 6.43,
2080
+ "learning_rate": 3.566591422121896e-05,
2081
+ "loss": 0.0457,
2082
+ "step": 2850
2083
+ },
2084
+ {
2085
+ "epoch": 6.46,
2086
+ "learning_rate": 3.5440180586907454e-05,
2087
+ "loss": 0.0473,
2088
+ "step": 2860
2089
+ },
2090
+ {
2091
+ "epoch": 6.48,
2092
+ "learning_rate": 3.521444695259594e-05,
2093
+ "loss": 0.0438,
2094
+ "step": 2870
2095
+ },
2096
+ {
2097
+ "epoch": 6.5,
2098
+ "learning_rate": 3.498871331828443e-05,
2099
+ "loss": 0.0468,
2100
+ "step": 2880
2101
+ },
2102
+ {
2103
+ "epoch": 6.52,
2104
+ "learning_rate": 3.476297968397291e-05,
2105
+ "loss": 0.0445,
2106
+ "step": 2890
2107
+ },
2108
+ {
2109
+ "epoch": 6.55,
2110
+ "learning_rate": 3.4537246049661404e-05,
2111
+ "loss": 0.0433,
2112
+ "step": 2900
2113
+ },
2114
+ {
2115
+ "epoch": 6.55,
2116
+ "eval_accuracy": 0.9862365591397849,
2117
+ "eval_f1": 0.36318407960199006,
2118
+ "eval_loss": 0.04913609102368355,
2119
+ "eval_precision": 0.8902439024390244,
2120
+ "eval_recall": 0.228125,
2121
+ "eval_roc_auc_score": 0.92824193107221,
2122
+ "eval_runtime": 5.1037,
2123
+ "eval_samples_per_second": 36.444,
2124
+ "eval_steps_per_second": 2.351,
2125
+ "step": 2900
2126
+ },
2127
+ {
2128
+ "epoch": 6.57,
2129
+ "learning_rate": 3.431151241534989e-05,
2130
+ "loss": 0.0496,
2131
+ "step": 2910
2132
+ },
2133
+ {
2134
+ "epoch": 6.59,
2135
+ "learning_rate": 3.408577878103837e-05,
2136
+ "loss": 0.0432,
2137
+ "step": 2920
2138
+ },
2139
+ {
2140
+ "epoch": 6.61,
2141
+ "learning_rate": 3.386004514672686e-05,
2142
+ "loss": 0.0438,
2143
+ "step": 2930
2144
+ },
2145
+ {
2146
+ "epoch": 6.64,
2147
+ "learning_rate": 3.3634311512415354e-05,
2148
+ "loss": 0.0447,
2149
+ "step": 2940
2150
+ },
2151
+ {
2152
+ "epoch": 6.66,
2153
+ "learning_rate": 3.3408577878103845e-05,
2154
+ "loss": 0.0448,
2155
+ "step": 2950
2156
+ },
2157
+ {
2158
+ "epoch": 6.68,
2159
+ "learning_rate": 3.318284424379233e-05,
2160
+ "loss": 0.0387,
2161
+ "step": 2960
2162
+ },
2163
+ {
2164
+ "epoch": 6.7,
2165
+ "learning_rate": 3.295711060948081e-05,
2166
+ "loss": 0.0403,
2167
+ "step": 2970
2168
+ },
2169
+ {
2170
+ "epoch": 6.73,
2171
+ "learning_rate": 3.2731376975169304e-05,
2172
+ "loss": 0.0427,
2173
+ "step": 2980
2174
+ },
2175
+ {
2176
+ "epoch": 6.75,
2177
+ "learning_rate": 3.250564334085779e-05,
2178
+ "loss": 0.0483,
2179
+ "step": 2990
2180
+ },
2181
+ {
2182
+ "epoch": 6.77,
2183
+ "learning_rate": 3.227990970654628e-05,
2184
+ "loss": 0.0444,
2185
+ "step": 3000
2186
+ },
2187
+ {
2188
+ "epoch": 6.77,
2189
+ "eval_accuracy": 0.9864516129032258,
2190
+ "eval_f1": 0.3793103448275862,
2191
+ "eval_loss": 0.04895901307463646,
2192
+ "eval_precision": 0.8953488372093024,
2193
+ "eval_recall": 0.240625,
2194
+ "eval_roc_auc_score": 0.9280762787199124,
2195
+ "eval_runtime": 5.2128,
2196
+ "eval_samples_per_second": 35.681,
2197
+ "eval_steps_per_second": 2.302,
2198
+ "step": 3000
2199
+ },
2200
+ {
2201
+ "epoch": 6.79,
2202
+ "learning_rate": 3.205417607223476e-05,
2203
+ "loss": 0.0396,
2204
+ "step": 3010
2205
+ },
2206
+ {
2207
+ "epoch": 6.82,
2208
+ "learning_rate": 3.1828442437923254e-05,
2209
+ "loss": 0.0401,
2210
+ "step": 3020
2211
+ },
2212
+ {
2213
+ "epoch": 6.84,
2214
+ "learning_rate": 3.1602708803611745e-05,
2215
+ "loss": 0.0444,
2216
+ "step": 3030
2217
+ },
2218
+ {
2219
+ "epoch": 6.86,
2220
+ "learning_rate": 3.137697516930023e-05,
2221
+ "loss": 0.0444,
2222
+ "step": 3040
2223
+ },
2224
+ {
2225
+ "epoch": 6.88,
2226
+ "learning_rate": 3.115124153498871e-05,
2227
+ "loss": 0.0439,
2228
+ "step": 3050
2229
+ },
2230
+ {
2231
+ "epoch": 6.91,
2232
+ "learning_rate": 3.0925507900677204e-05,
2233
+ "loss": 0.049,
2234
+ "step": 3060
2235
+ },
2236
+ {
2237
+ "epoch": 6.93,
2238
+ "learning_rate": 3.069977426636569e-05,
2239
+ "loss": 0.0481,
2240
+ "step": 3070
2241
+ },
2242
+ {
2243
+ "epoch": 6.95,
2244
+ "learning_rate": 3.047404063205418e-05,
2245
+ "loss": 0.0461,
2246
+ "step": 3080
2247
+ },
2248
+ {
2249
+ "epoch": 6.98,
2250
+ "learning_rate": 3.0248306997742666e-05,
2251
+ "loss": 0.0515,
2252
+ "step": 3090
2253
+ },
2254
+ {
2255
+ "epoch": 7.0,
2256
+ "learning_rate": 3.002257336343115e-05,
2257
+ "loss": 0.049,
2258
+ "step": 3100
2259
+ },
2260
+ {
2261
+ "epoch": 7.0,
2262
+ "eval_accuracy": 0.9867204301075269,
2263
+ "eval_f1": 0.3990267639902676,
2264
+ "eval_loss": 0.048538196831941605,
2265
+ "eval_precision": 0.9010989010989011,
2266
+ "eval_recall": 0.25625,
2267
+ "eval_roc_auc_score": 0.9276545404814005,
2268
+ "eval_runtime": 5.2996,
2269
+ "eval_samples_per_second": 35.097,
2270
+ "eval_steps_per_second": 2.264,
2271
+ "step": 3100
2272
+ },
2273
+ {
2274
+ "epoch": 7.02,
2275
+ "learning_rate": 2.979683972911964e-05,
2276
+ "loss": 0.0424,
2277
+ "step": 3110
2278
+ },
2279
+ {
2280
+ "epoch": 7.04,
2281
+ "learning_rate": 2.9571106094808125e-05,
2282
+ "loss": 0.0416,
2283
+ "step": 3120
2284
+ },
2285
+ {
2286
+ "epoch": 7.07,
2287
+ "learning_rate": 2.9345372460496616e-05,
2288
+ "loss": 0.0394,
2289
+ "step": 3130
2290
+ },
2291
+ {
2292
+ "epoch": 7.09,
2293
+ "learning_rate": 2.9119638826185104e-05,
2294
+ "loss": 0.042,
2295
+ "step": 3140
2296
+ },
2297
+ {
2298
+ "epoch": 7.11,
2299
+ "learning_rate": 2.8893905191873588e-05,
2300
+ "loss": 0.0424,
2301
+ "step": 3150
2302
+ },
2303
+ {
2304
+ "epoch": 7.13,
2305
+ "learning_rate": 2.866817155756208e-05,
2306
+ "loss": 0.0406,
2307
+ "step": 3160
2308
+ },
2309
+ {
2310
+ "epoch": 7.16,
2311
+ "learning_rate": 2.8442437923250566e-05,
2312
+ "loss": 0.0406,
2313
+ "step": 3170
2314
+ },
2315
+ {
2316
+ "epoch": 7.18,
2317
+ "learning_rate": 2.8216704288939057e-05,
2318
+ "loss": 0.0485,
2319
+ "step": 3180
2320
+ },
2321
+ {
2322
+ "epoch": 7.2,
2323
+ "learning_rate": 2.799097065462754e-05,
2324
+ "loss": 0.0447,
2325
+ "step": 3190
2326
+ },
2327
+ {
2328
+ "epoch": 7.22,
2329
+ "learning_rate": 2.7765237020316025e-05,
2330
+ "loss": 0.0441,
2331
+ "step": 3200
2332
+ },
2333
+ {
2334
+ "epoch": 7.22,
2335
+ "eval_accuracy": 0.986505376344086,
2336
+ "eval_f1": 0.3802469135802469,
2337
+ "eval_loss": 0.04807009920477867,
2338
+ "eval_precision": 0.9058823529411765,
2339
+ "eval_recall": 0.240625,
2340
+ "eval_roc_auc_score": 0.9290772873358861,
2341
+ "eval_runtime": 5.1517,
2342
+ "eval_samples_per_second": 36.104,
2343
+ "eval_steps_per_second": 2.329,
2344
+ "step": 3200
2345
+ },
2346
+ {
2347
+ "epoch": 7.25,
2348
+ "learning_rate": 2.7539503386004516e-05,
2349
+ "loss": 0.0407,
2350
+ "step": 3210
2351
+ },
2352
+ {
2353
+ "epoch": 7.27,
2354
+ "learning_rate": 2.7313769751693004e-05,
2355
+ "loss": 0.0395,
2356
+ "step": 3220
2357
+ },
2358
+ {
2359
+ "epoch": 7.29,
2360
+ "learning_rate": 2.7088036117381494e-05,
2361
+ "loss": 0.0434,
2362
+ "step": 3230
2363
+ },
2364
+ {
2365
+ "epoch": 7.31,
2366
+ "learning_rate": 2.686230248306998e-05,
2367
+ "loss": 0.0407,
2368
+ "step": 3240
2369
+ },
2370
+ {
2371
+ "epoch": 7.34,
2372
+ "learning_rate": 2.6636568848758463e-05,
2373
+ "loss": 0.0377,
2374
+ "step": 3250
2375
+ },
2376
+ {
2377
+ "epoch": 7.36,
2378
+ "learning_rate": 2.6410835214446954e-05,
2379
+ "loss": 0.0423,
2380
+ "step": 3260
2381
+ },
2382
+ {
2383
+ "epoch": 7.38,
2384
+ "learning_rate": 2.618510158013544e-05,
2385
+ "loss": 0.0447,
2386
+ "step": 3270
2387
+ },
2388
+ {
2389
+ "epoch": 7.4,
2390
+ "learning_rate": 2.5959367945823932e-05,
2391
+ "loss": 0.0425,
2392
+ "step": 3280
2393
+ },
2394
+ {
2395
+ "epoch": 7.43,
2396
+ "learning_rate": 2.5733634311512416e-05,
2397
+ "loss": 0.0399,
2398
+ "step": 3290
2399
+ },
2400
+ {
2401
+ "epoch": 7.45,
2402
+ "learning_rate": 2.5507900677200903e-05,
2403
+ "loss": 0.0454,
2404
+ "step": 3300
2405
+ },
2406
+ {
2407
+ "epoch": 7.45,
2408
+ "eval_accuracy": 0.9869354838709677,
2409
+ "eval_f1": 0.42004773269689744,
2410
+ "eval_loss": 0.04823455959558487,
2411
+ "eval_precision": 0.8888888888888888,
2412
+ "eval_recall": 0.275,
2413
+ "eval_roc_auc_score": 0.9287259812636762,
2414
+ "eval_runtime": 5.2168,
2415
+ "eval_samples_per_second": 35.654,
2416
+ "eval_steps_per_second": 2.3,
2417
+ "step": 3300
2418
+ },
2419
+ {
2420
+ "epoch": 7.47,
2421
+ "learning_rate": 2.5282167042889394e-05,
2422
+ "loss": 0.0417,
2423
+ "step": 3310
2424
+ },
2425
+ {
2426
+ "epoch": 7.49,
2427
+ "learning_rate": 2.505643340857788e-05,
2428
+ "loss": 0.0403,
2429
+ "step": 3320
2430
+ },
2431
+ {
2432
+ "epoch": 7.52,
2433
+ "learning_rate": 2.4830699774266366e-05,
2434
+ "loss": 0.0411,
2435
+ "step": 3330
2436
+ },
2437
+ {
2438
+ "epoch": 7.54,
2439
+ "learning_rate": 2.4604966139954853e-05,
2440
+ "loss": 0.038,
2441
+ "step": 3340
2442
+ },
2443
+ {
2444
+ "epoch": 7.56,
2445
+ "learning_rate": 2.4379232505643344e-05,
2446
+ "loss": 0.045,
2447
+ "step": 3350
2448
+ },
2449
+ {
2450
+ "epoch": 7.58,
2451
+ "learning_rate": 2.415349887133183e-05,
2452
+ "loss": 0.044,
2453
+ "step": 3360
2454
+ },
2455
+ {
2456
+ "epoch": 7.61,
2457
+ "learning_rate": 2.3927765237020316e-05,
2458
+ "loss": 0.0397,
2459
+ "step": 3370
2460
+ },
2461
+ {
2462
+ "epoch": 7.63,
2463
+ "learning_rate": 2.3702031602708803e-05,
2464
+ "loss": 0.0453,
2465
+ "step": 3380
2466
+ },
2467
+ {
2468
+ "epoch": 7.65,
2469
+ "learning_rate": 2.3476297968397294e-05,
2470
+ "loss": 0.0455,
2471
+ "step": 3390
2472
+ },
2473
+ {
2474
+ "epoch": 7.67,
2475
+ "learning_rate": 2.3250564334085782e-05,
2476
+ "loss": 0.0405,
2477
+ "step": 3400
2478
+ },
2479
+ {
2480
+ "epoch": 7.67,
2481
+ "eval_accuracy": 0.9869354838709677,
2482
+ "eval_f1": 0.42004773269689744,
2483
+ "eval_loss": 0.04781893268227577,
2484
+ "eval_precision": 0.8888888888888888,
2485
+ "eval_recall": 0.275,
2486
+ "eval_roc_auc_score": 0.9291827646334793,
2487
+ "eval_runtime": 5.1939,
2488
+ "eval_samples_per_second": 35.811,
2489
+ "eval_steps_per_second": 2.31,
2490
+ "step": 3400
2491
+ },
2492
+ {
2493
+ "epoch": 7.7,
2494
+ "learning_rate": 2.3024830699774266e-05,
2495
+ "loss": 0.0404,
2496
+ "step": 3410
2497
+ },
2498
+ {
2499
+ "epoch": 7.72,
2500
+ "learning_rate": 2.2799097065462753e-05,
2501
+ "loss": 0.0404,
2502
+ "step": 3420
2503
+ },
2504
+ {
2505
+ "epoch": 7.74,
2506
+ "learning_rate": 2.2573363431151244e-05,
2507
+ "loss": 0.045,
2508
+ "step": 3430
2509
+ },
2510
+ {
2511
+ "epoch": 7.77,
2512
+ "learning_rate": 2.2347629796839732e-05,
2513
+ "loss": 0.0401,
2514
+ "step": 3440
2515
+ },
2516
+ {
2517
+ "epoch": 7.79,
2518
+ "learning_rate": 2.212189616252822e-05,
2519
+ "loss": 0.0425,
2520
+ "step": 3450
2521
+ },
2522
+ {
2523
+ "epoch": 7.81,
2524
+ "learning_rate": 2.1896162528216703e-05,
2525
+ "loss": 0.0395,
2526
+ "step": 3460
2527
+ },
2528
+ {
2529
+ "epoch": 7.83,
2530
+ "learning_rate": 2.167042889390519e-05,
2531
+ "loss": 0.0474,
2532
+ "step": 3470
2533
+ },
2534
+ {
2535
+ "epoch": 7.86,
2536
+ "learning_rate": 2.1444695259593682e-05,
2537
+ "loss": 0.0423,
2538
+ "step": 3480
2539
+ },
2540
+ {
2541
+ "epoch": 7.88,
2542
+ "learning_rate": 2.121896162528217e-05,
2543
+ "loss": 0.0439,
2544
+ "step": 3490
2545
+ },
2546
+ {
2547
+ "epoch": 7.9,
2548
+ "learning_rate": 2.0993227990970657e-05,
2549
+ "loss": 0.042,
2550
+ "step": 3500
2551
+ },
2552
+ {
2553
+ "epoch": 7.9,
2554
+ "eval_accuracy": 0.9867204301075269,
2555
+ "eval_f1": 0.41330166270783847,
2556
+ "eval_loss": 0.047578372061252594,
2557
+ "eval_precision": 0.8613861386138614,
2558
+ "eval_recall": 0.271875,
2559
+ "eval_roc_auc_score": 0.9298246033916849,
2560
+ "eval_runtime": 5.0799,
2561
+ "eval_samples_per_second": 36.615,
2562
+ "eval_steps_per_second": 2.362,
2563
+ "step": 3500
2564
+ },
2565
+ {
2566
+ "epoch": 7.92,
2567
+ "learning_rate": 2.076749435665914e-05,
2568
+ "loss": 0.0402,
2569
+ "step": 3510
2570
+ },
2571
+ {
2572
+ "epoch": 7.95,
2573
+ "learning_rate": 2.054176072234763e-05,
2574
+ "loss": 0.0417,
2575
+ "step": 3520
2576
+ },
2577
+ {
2578
+ "epoch": 7.97,
2579
+ "learning_rate": 2.031602708803612e-05,
2580
+ "loss": 0.0416,
2581
+ "step": 3530
2582
+ },
2583
+ {
2584
+ "epoch": 7.99,
2585
+ "learning_rate": 2.0090293453724607e-05,
2586
+ "loss": 0.0375,
2587
+ "step": 3540
2588
+ },
2589
+ {
2590
+ "epoch": 8.01,
2591
+ "learning_rate": 1.986455981941309e-05,
2592
+ "loss": 0.0404,
2593
+ "step": 3550
2594
+ },
2595
+ {
2596
+ "epoch": 8.04,
2597
+ "learning_rate": 1.963882618510158e-05,
2598
+ "loss": 0.043,
2599
+ "step": 3560
2600
+ },
2601
+ {
2602
+ "epoch": 8.06,
2603
+ "learning_rate": 1.941309255079007e-05,
2604
+ "loss": 0.0376,
2605
+ "step": 3570
2606
+ },
2607
+ {
2608
+ "epoch": 8.08,
2609
+ "learning_rate": 1.9187358916478557e-05,
2610
+ "loss": 0.0371,
2611
+ "step": 3580
2612
+ },
2613
+ {
2614
+ "epoch": 8.1,
2615
+ "learning_rate": 1.8961625282167044e-05,
2616
+ "loss": 0.0396,
2617
+ "step": 3590
2618
+ },
2619
+ {
2620
+ "epoch": 8.13,
2621
+ "learning_rate": 1.873589164785553e-05,
2622
+ "loss": 0.0364,
2623
+ "step": 3600
2624
+ },
2625
+ {
2626
+ "epoch": 8.13,
2627
+ "eval_accuracy": 0.9872580645161291,
2628
+ "eval_f1": 0.44496487119437944,
2629
+ "eval_loss": 0.04716450721025467,
2630
+ "eval_precision": 0.8878504672897196,
2631
+ "eval_recall": 0.296875,
2632
+ "eval_roc_auc_score": 0.9299637582056892,
2633
+ "eval_runtime": 5.1786,
2634
+ "eval_samples_per_second": 35.917,
2635
+ "eval_steps_per_second": 2.317,
2636
+ "step": 3600
2637
+ },
2638
+ {
2639
+ "epoch": 8.15,
2640
+ "learning_rate": 1.851015801354402e-05,
2641
+ "loss": 0.0431,
2642
+ "step": 3610
2643
+ },
2644
+ {
2645
+ "epoch": 8.17,
2646
+ "learning_rate": 1.8284424379232507e-05,
2647
+ "loss": 0.0364,
2648
+ "step": 3620
2649
+ },
2650
+ {
2651
+ "epoch": 8.19,
2652
+ "learning_rate": 1.8058690744920994e-05,
2653
+ "loss": 0.0404,
2654
+ "step": 3630
2655
+ },
2656
+ {
2657
+ "epoch": 8.22,
2658
+ "learning_rate": 1.783295711060948e-05,
2659
+ "loss": 0.0359,
2660
+ "step": 3640
2661
+ },
2662
+ {
2663
+ "epoch": 8.24,
2664
+ "learning_rate": 1.760722347629797e-05,
2665
+ "loss": 0.0395,
2666
+ "step": 3650
2667
+ },
2668
+ {
2669
+ "epoch": 8.26,
2670
+ "learning_rate": 1.7381489841986457e-05,
2671
+ "loss": 0.0421,
2672
+ "step": 3660
2673
+ },
2674
+ {
2675
+ "epoch": 8.28,
2676
+ "learning_rate": 1.7155756207674944e-05,
2677
+ "loss": 0.0366,
2678
+ "step": 3670
2679
+ },
2680
+ {
2681
+ "epoch": 8.31,
2682
+ "learning_rate": 1.693002257336343e-05,
2683
+ "loss": 0.0354,
2684
+ "step": 3680
2685
+ },
2686
+ {
2687
+ "epoch": 8.33,
2688
+ "learning_rate": 1.6704288939051922e-05,
2689
+ "loss": 0.0365,
2690
+ "step": 3690
2691
+ },
2692
+ {
2693
+ "epoch": 8.35,
2694
+ "learning_rate": 1.6478555304740406e-05,
2695
+ "loss": 0.0425,
2696
+ "step": 3700
2697
+ },
2698
+ {
2699
+ "epoch": 8.35,
2700
+ "eval_accuracy": 0.9871505376344086,
2701
+ "eval_f1": 0.43498817966903064,
2702
+ "eval_loss": 0.04713977128267288,
2703
+ "eval_precision": 0.8932038834951457,
2704
+ "eval_recall": 0.2875,
2705
+ "eval_roc_auc_score": 0.929676046225383,
2706
+ "eval_runtime": 5.3146,
2707
+ "eval_samples_per_second": 34.998,
2708
+ "eval_steps_per_second": 2.258,
2709
+ "step": 3700
2710
+ },
2711
+ {
2712
+ "epoch": 8.37,
2713
+ "learning_rate": 1.6252821670428894e-05,
2714
+ "loss": 0.0367,
2715
+ "step": 3710
2716
+ },
2717
+ {
2718
+ "epoch": 8.4,
2719
+ "learning_rate": 1.602708803611738e-05,
2720
+ "loss": 0.0375,
2721
+ "step": 3720
2722
+ },
2723
+ {
2724
+ "epoch": 8.42,
2725
+ "learning_rate": 1.5801354401805872e-05,
2726
+ "loss": 0.0387,
2727
+ "step": 3730
2728
+ },
2729
+ {
2730
+ "epoch": 8.44,
2731
+ "learning_rate": 1.5575620767494356e-05,
2732
+ "loss": 0.039,
2733
+ "step": 3740
2734
+ },
2735
+ {
2736
+ "epoch": 8.47,
2737
+ "learning_rate": 1.5349887133182844e-05,
2738
+ "loss": 0.0407,
2739
+ "step": 3750
2740
+ },
2741
+ {
2742
+ "epoch": 8.49,
2743
+ "learning_rate": 1.5124153498871333e-05,
2744
+ "loss": 0.04,
2745
+ "step": 3760
2746
+ },
2747
+ {
2748
+ "epoch": 8.51,
2749
+ "learning_rate": 1.489841986455982e-05,
2750
+ "loss": 0.0433,
2751
+ "step": 3770
2752
+ },
2753
+ {
2754
+ "epoch": 8.53,
2755
+ "learning_rate": 1.4672686230248308e-05,
2756
+ "loss": 0.0406,
2757
+ "step": 3780
2758
+ },
2759
+ {
2760
+ "epoch": 8.56,
2761
+ "learning_rate": 1.4446952595936794e-05,
2762
+ "loss": 0.0403,
2763
+ "step": 3790
2764
+ },
2765
+ {
2766
+ "epoch": 8.58,
2767
+ "learning_rate": 1.4221218961625283e-05,
2768
+ "loss": 0.0416,
2769
+ "step": 3800
2770
+ },
2771
+ {
2772
+ "epoch": 8.58,
2773
+ "eval_accuracy": 0.9874731182795699,
2774
+ "eval_f1": 0.45433255269320844,
2775
+ "eval_loss": 0.04680037871003151,
2776
+ "eval_precision": 0.9065420560747663,
2777
+ "eval_recall": 0.303125,
2778
+ "eval_roc_auc_score": 0.9314481332056893,
2779
+ "eval_runtime": 5.2387,
2780
+ "eval_samples_per_second": 35.505,
2781
+ "eval_steps_per_second": 2.291,
2782
+ "step": 3800
2783
+ },
2784
+ {
2785
+ "epoch": 8.6,
2786
+ "learning_rate": 1.399548532731377e-05,
2787
+ "loss": 0.039,
2788
+ "step": 3810
2789
+ },
2790
+ {
2791
+ "epoch": 8.62,
2792
+ "learning_rate": 1.3769751693002258e-05,
2793
+ "loss": 0.0405,
2794
+ "step": 3820
2795
+ },
2796
+ {
2797
+ "epoch": 8.65,
2798
+ "learning_rate": 1.3544018058690747e-05,
2799
+ "loss": 0.0388,
2800
+ "step": 3830
2801
+ },
2802
+ {
2803
+ "epoch": 8.67,
2804
+ "learning_rate": 1.3318284424379231e-05,
2805
+ "loss": 0.039,
2806
+ "step": 3840
2807
+ },
2808
+ {
2809
+ "epoch": 8.69,
2810
+ "learning_rate": 1.309255079006772e-05,
2811
+ "loss": 0.0369,
2812
+ "step": 3850
2813
+ },
2814
+ {
2815
+ "epoch": 8.71,
2816
+ "learning_rate": 1.2866817155756208e-05,
2817
+ "loss": 0.0406,
2818
+ "step": 3860
2819
+ },
2820
+ {
2821
+ "epoch": 8.74,
2822
+ "learning_rate": 1.2641083521444697e-05,
2823
+ "loss": 0.0413,
2824
+ "step": 3870
2825
+ },
2826
+ {
2827
+ "epoch": 8.76,
2828
+ "learning_rate": 1.2415349887133183e-05,
2829
+ "loss": 0.0399,
2830
+ "step": 3880
2831
+ },
2832
+ {
2833
+ "epoch": 8.78,
2834
+ "learning_rate": 1.2189616252821672e-05,
2835
+ "loss": 0.0448,
2836
+ "step": 3890
2837
+ },
2838
+ {
2839
+ "epoch": 8.8,
2840
+ "learning_rate": 1.1963882618510158e-05,
2841
+ "loss": 0.0394,
2842
+ "step": 3900
2843
+ },
2844
+ {
2845
+ "epoch": 8.8,
2846
+ "eval_accuracy": 0.9870430107526882,
2847
+ "eval_f1": 0.4329411764705882,
2848
+ "eval_loss": 0.04629945382475853,
2849
+ "eval_precision": 0.8761904761904762,
2850
+ "eval_recall": 0.2875,
2851
+ "eval_roc_auc_score": 0.9352978836159738,
2852
+ "eval_runtime": 5.2999,
2853
+ "eval_samples_per_second": 35.095,
2854
+ "eval_steps_per_second": 2.264,
2855
+ "step": 3900
2856
+ },
2857
+ {
2858
+ "epoch": 8.83,
2859
+ "learning_rate": 1.1738148984198647e-05,
2860
+ "loss": 0.0346,
2861
+ "step": 3910
2862
+ },
2863
+ {
2864
+ "epoch": 8.85,
2865
+ "learning_rate": 1.1512415349887133e-05,
2866
+ "loss": 0.0399,
2867
+ "step": 3920
2868
+ },
2869
+ {
2870
+ "epoch": 8.87,
2871
+ "learning_rate": 1.1286681715575622e-05,
2872
+ "loss": 0.0396,
2873
+ "step": 3930
2874
+ },
2875
+ {
2876
+ "epoch": 8.89,
2877
+ "learning_rate": 1.106094808126411e-05,
2878
+ "loss": 0.0395,
2879
+ "step": 3940
2880
+ },
2881
+ {
2882
+ "epoch": 8.92,
2883
+ "learning_rate": 1.0835214446952595e-05,
2884
+ "loss": 0.0398,
2885
+ "step": 3950
2886
+ },
2887
+ {
2888
+ "epoch": 8.94,
2889
+ "learning_rate": 1.0609480812641085e-05,
2890
+ "loss": 0.0449,
2891
+ "step": 3960
2892
+ },
2893
+ {
2894
+ "epoch": 8.96,
2895
+ "learning_rate": 1.038374717832957e-05,
2896
+ "loss": 0.0334,
2897
+ "step": 3970
2898
+ },
2899
+ {
2900
+ "epoch": 8.98,
2901
+ "learning_rate": 1.015801354401806e-05,
2902
+ "loss": 0.042,
2903
+ "step": 3980
2904
+ },
2905
+ {
2906
+ "epoch": 9.01,
2907
+ "learning_rate": 9.932279909706545e-06,
2908
+ "loss": 0.037,
2909
+ "step": 3990
2910
+ },
2911
+ {
2912
+ "epoch": 9.03,
2913
+ "learning_rate": 9.706546275395035e-06,
2914
+ "loss": 0.037,
2915
+ "step": 4000
2916
+ },
2917
+ {
2918
+ "epoch": 9.03,
2919
+ "eval_accuracy": 0.9872580645161291,
2920
+ "eval_f1": 0.4501160092807424,
2921
+ "eval_loss": 0.046479061245918274,
2922
+ "eval_precision": 0.8738738738738738,
2923
+ "eval_recall": 0.303125,
2924
+ "eval_roc_auc_score": 0.9332275711159737,
2925
+ "eval_runtime": 5.2391,
2926
+ "eval_samples_per_second": 35.502,
2927
+ "eval_steps_per_second": 2.29,
2928
+ "step": 4000
2929
+ }
2930
+ ],
2931
+ "max_steps": 4430,
2932
+ "num_train_epochs": 10,
2933
+ "total_flos": 3.368919067577549e+16,
2934
+ "trial_name": null,
2935
+ "trial_params": null
2936
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31264859847870bff90ce6bbf0113d57f352be7f018d36918ff338f6bfb55f7e
3
+ size 3963