MariaFGI commited on
Commit
34f9928
·
verified ·
1 Parent(s): 2b40857

Training in progress, epoch 6

Browse files
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:22546df02e97ac5048243011464bec7f276eecffbc48b82b1441ad405036953f
3
  size 598898116
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81ac3f88bdb451e4316877bdf137e199de70b195697d9835e23b0052e2b2bace
3
  size 598898116
run-0/checkpoint-5724/config.json CHANGED
@@ -21,313 +21,313 @@
21
  "hidden_activation": "gelu",
22
  "hidden_size": 768,
23
  "id2label": {
24
- "0": "LABEL_0",
25
- "1": "LABEL_1",
26
- "2": "LABEL_2",
27
- "3": "LABEL_3",
28
- "4": "LABEL_4",
29
- "5": "LABEL_5",
30
- "6": "LABEL_6",
31
- "7": "LABEL_7",
32
- "8": "LABEL_8",
33
- "9": "LABEL_9",
34
- "10": "LABEL_10",
35
- "11": "LABEL_11",
36
- "12": "LABEL_12",
37
- "13": "LABEL_13",
38
- "14": "LABEL_14",
39
- "15": "LABEL_15",
40
- "16": "LABEL_16",
41
- "17": "LABEL_17",
42
- "18": "LABEL_18",
43
- "19": "LABEL_19",
44
- "20": "LABEL_20",
45
- "21": "LABEL_21",
46
- "22": "LABEL_22",
47
- "23": "LABEL_23",
48
- "24": "LABEL_24",
49
- "25": "LABEL_25",
50
- "26": "LABEL_26",
51
- "27": "LABEL_27",
52
- "28": "LABEL_28",
53
- "29": "LABEL_29",
54
- "30": "LABEL_30",
55
- "31": "LABEL_31",
56
- "32": "LABEL_32",
57
- "33": "LABEL_33",
58
- "34": "LABEL_34",
59
- "35": "LABEL_35",
60
- "36": "LABEL_36",
61
- "37": "LABEL_37",
62
- "38": "LABEL_38",
63
- "39": "LABEL_39",
64
- "40": "LABEL_40",
65
- "41": "LABEL_41",
66
- "42": "LABEL_42",
67
- "43": "LABEL_43",
68
- "44": "LABEL_44",
69
- "45": "LABEL_45",
70
- "46": "LABEL_46",
71
- "47": "LABEL_47",
72
- "48": "LABEL_48",
73
- "49": "LABEL_49",
74
- "50": "LABEL_50",
75
- "51": "LABEL_51",
76
- "52": "LABEL_52",
77
- "53": "LABEL_53",
78
- "54": "LABEL_54",
79
- "55": "LABEL_55",
80
- "56": "LABEL_56",
81
- "57": "LABEL_57",
82
- "58": "LABEL_58",
83
- "59": "LABEL_59",
84
- "60": "LABEL_60",
85
- "61": "LABEL_61",
86
- "62": "LABEL_62",
87
- "63": "LABEL_63",
88
- "64": "LABEL_64",
89
- "65": "LABEL_65",
90
- "66": "LABEL_66",
91
- "67": "LABEL_67",
92
- "68": "LABEL_68",
93
- "69": "LABEL_69",
94
- "70": "LABEL_70",
95
- "71": "LABEL_71",
96
- "72": "LABEL_72",
97
- "73": "LABEL_73",
98
- "74": "LABEL_74",
99
- "75": "LABEL_75",
100
- "76": "LABEL_76",
101
- "77": "LABEL_77",
102
- "78": "LABEL_78",
103
- "79": "LABEL_79",
104
- "80": "LABEL_80",
105
- "81": "LABEL_81",
106
- "82": "LABEL_82",
107
- "83": "LABEL_83",
108
- "84": "LABEL_84",
109
- "85": "LABEL_85",
110
- "86": "LABEL_86",
111
- "87": "LABEL_87",
112
- "88": "LABEL_88",
113
- "89": "LABEL_89",
114
- "90": "LABEL_90",
115
- "91": "LABEL_91",
116
- "92": "LABEL_92",
117
- "93": "LABEL_93",
118
- "94": "LABEL_94",
119
- "95": "LABEL_95",
120
- "96": "LABEL_96",
121
- "97": "LABEL_97",
122
- "98": "LABEL_98",
123
- "99": "LABEL_99",
124
- "100": "LABEL_100",
125
- "101": "LABEL_101",
126
- "102": "LABEL_102",
127
- "103": "LABEL_103",
128
- "104": "LABEL_104",
129
- "105": "LABEL_105",
130
- "106": "LABEL_106",
131
- "107": "LABEL_107",
132
- "108": "LABEL_108",
133
- "109": "LABEL_109",
134
- "110": "LABEL_110",
135
- "111": "LABEL_111",
136
- "112": "LABEL_112",
137
- "113": "LABEL_113",
138
- "114": "LABEL_114",
139
- "115": "LABEL_115",
140
- "116": "LABEL_116",
141
- "117": "LABEL_117",
142
- "118": "LABEL_118",
143
- "119": "LABEL_119",
144
- "120": "LABEL_120",
145
- "121": "LABEL_121",
146
- "122": "LABEL_122",
147
- "123": "LABEL_123",
148
- "124": "LABEL_124",
149
- "125": "LABEL_125",
150
- "126": "LABEL_126",
151
- "127": "LABEL_127",
152
- "128": "LABEL_128",
153
- "129": "LABEL_129",
154
- "130": "LABEL_130",
155
- "131": "LABEL_131",
156
- "132": "LABEL_132",
157
- "133": "LABEL_133",
158
- "134": "LABEL_134",
159
- "135": "LABEL_135",
160
- "136": "LABEL_136",
161
- "137": "LABEL_137",
162
- "138": "LABEL_138",
163
- "139": "LABEL_139",
164
- "140": "LABEL_140",
165
- "141": "LABEL_141",
166
- "142": "LABEL_142",
167
- "143": "LABEL_143",
168
- "144": "LABEL_144",
169
- "145": "LABEL_145",
170
- "146": "LABEL_146",
171
- "147": "LABEL_147",
172
- "148": "LABEL_148",
173
- "149": "LABEL_149",
174
- "150": "LABEL_150"
175
  },
176
  "initializer_cutoff_factor": 2.0,
177
  "initializer_range": 0.02,
178
  "intermediate_size": 1152,
179
  "label2id": {
180
- "LABEL_0": 0,
181
- "LABEL_1": 1,
182
- "LABEL_10": 10,
183
- "LABEL_100": 100,
184
- "LABEL_101": 101,
185
- "LABEL_102": 102,
186
- "LABEL_103": 103,
187
- "LABEL_104": 104,
188
- "LABEL_105": 105,
189
- "LABEL_106": 106,
190
- "LABEL_107": 107,
191
- "LABEL_108": 108,
192
- "LABEL_109": 109,
193
- "LABEL_11": 11,
194
- "LABEL_110": 110,
195
- "LABEL_111": 111,
196
- "LABEL_112": 112,
197
- "LABEL_113": 113,
198
- "LABEL_114": 114,
199
- "LABEL_115": 115,
200
- "LABEL_116": 116,
201
- "LABEL_117": 117,
202
- "LABEL_118": 118,
203
- "LABEL_119": 119,
204
- "LABEL_12": 12,
205
- "LABEL_120": 120,
206
- "LABEL_121": 121,
207
- "LABEL_122": 122,
208
- "LABEL_123": 123,
209
- "LABEL_124": 124,
210
- "LABEL_125": 125,
211
- "LABEL_126": 126,
212
- "LABEL_127": 127,
213
- "LABEL_128": 128,
214
- "LABEL_129": 129,
215
- "LABEL_13": 13,
216
- "LABEL_130": 130,
217
- "LABEL_131": 131,
218
- "LABEL_132": 132,
219
- "LABEL_133": 133,
220
- "LABEL_134": 134,
221
- "LABEL_135": 135,
222
- "LABEL_136": 136,
223
- "LABEL_137": 137,
224
- "LABEL_138": 138,
225
- "LABEL_139": 139,
226
- "LABEL_14": 14,
227
- "LABEL_140": 140,
228
- "LABEL_141": 141,
229
- "LABEL_142": 142,
230
- "LABEL_143": 143,
231
- "LABEL_144": 144,
232
- "LABEL_145": 145,
233
- "LABEL_146": 146,
234
- "LABEL_147": 147,
235
- "LABEL_148": 148,
236
- "LABEL_149": 149,
237
- "LABEL_15": 15,
238
- "LABEL_150": 150,
239
- "LABEL_16": 16,
240
- "LABEL_17": 17,
241
- "LABEL_18": 18,
242
- "LABEL_19": 19,
243
- "LABEL_2": 2,
244
- "LABEL_20": 20,
245
- "LABEL_21": 21,
246
- "LABEL_22": 22,
247
- "LABEL_23": 23,
248
- "LABEL_24": 24,
249
- "LABEL_25": 25,
250
- "LABEL_26": 26,
251
- "LABEL_27": 27,
252
- "LABEL_28": 28,
253
- "LABEL_29": 29,
254
- "LABEL_3": 3,
255
- "LABEL_30": 30,
256
- "LABEL_31": 31,
257
- "LABEL_32": 32,
258
- "LABEL_33": 33,
259
- "LABEL_34": 34,
260
- "LABEL_35": 35,
261
- "LABEL_36": 36,
262
- "LABEL_37": 37,
263
- "LABEL_38": 38,
264
- "LABEL_39": 39,
265
- "LABEL_4": 4,
266
- "LABEL_40": 40,
267
- "LABEL_41": 41,
268
- "LABEL_42": 42,
269
- "LABEL_43": 43,
270
- "LABEL_44": 44,
271
- "LABEL_45": 45,
272
- "LABEL_46": 46,
273
- "LABEL_47": 47,
274
- "LABEL_48": 48,
275
- "LABEL_49": 49,
276
- "LABEL_5": 5,
277
- "LABEL_50": 50,
278
- "LABEL_51": 51,
279
- "LABEL_52": 52,
280
- "LABEL_53": 53,
281
- "LABEL_54": 54,
282
- "LABEL_55": 55,
283
- "LABEL_56": 56,
284
- "LABEL_57": 57,
285
- "LABEL_58": 58,
286
- "LABEL_59": 59,
287
- "LABEL_6": 6,
288
- "LABEL_60": 60,
289
- "LABEL_61": 61,
290
- "LABEL_62": 62,
291
- "LABEL_63": 63,
292
- "LABEL_64": 64,
293
- "LABEL_65": 65,
294
- "LABEL_66": 66,
295
- "LABEL_67": 67,
296
- "LABEL_68": 68,
297
- "LABEL_69": 69,
298
- "LABEL_7": 7,
299
- "LABEL_70": 70,
300
- "LABEL_71": 71,
301
- "LABEL_72": 72,
302
- "LABEL_73": 73,
303
- "LABEL_74": 74,
304
- "LABEL_75": 75,
305
- "LABEL_76": 76,
306
- "LABEL_77": 77,
307
- "LABEL_78": 78,
308
- "LABEL_79": 79,
309
- "LABEL_8": 8,
310
- "LABEL_80": 80,
311
- "LABEL_81": 81,
312
- "LABEL_82": 82,
313
- "LABEL_83": 83,
314
- "LABEL_84": 84,
315
- "LABEL_85": 85,
316
- "LABEL_86": 86,
317
- "LABEL_87": 87,
318
- "LABEL_88": 88,
319
- "LABEL_89": 89,
320
- "LABEL_9": 9,
321
- "LABEL_90": 90,
322
- "LABEL_91": 91,
323
- "LABEL_92": 92,
324
- "LABEL_93": 93,
325
- "LABEL_94": 94,
326
- "LABEL_95": 95,
327
- "LABEL_96": 96,
328
- "LABEL_97": 97,
329
- "LABEL_98": 98,
330
- "LABEL_99": 99
331
  },
332
  "layer_norm_eps": 1e-05,
333
  "local_attention": 128,
 
21
  "hidden_activation": "gelu",
22
  "hidden_size": 768,
23
  "id2label": {
24
+ "0": "restaurant_reviews",
25
+ "1": "nutrition_info",
26
+ "2": "account_blocked",
27
+ "3": "oil_change_how",
28
+ "4": "time",
29
+ "5": "weather",
30
+ "6": "redeem_rewards",
31
+ "7": "interest_rate",
32
+ "8": "gas_type",
33
+ "9": "accept_reservations",
34
+ "10": "smart_home",
35
+ "11": "user_name",
36
+ "12": "report_lost_card",
37
+ "13": "repeat",
38
+ "14": "whisper_mode",
39
+ "15": "what_are_your_hobbies",
40
+ "16": "order",
41
+ "17": "jump_start",
42
+ "18": "schedule_meeting",
43
+ "19": "meeting_schedule",
44
+ "20": "freeze_account",
45
+ "21": "what_song",
46
+ "22": "meaning_of_life",
47
+ "23": "restaurant_reservation",
48
+ "24": "traffic",
49
+ "25": "make_call",
50
+ "26": "text",
51
+ "27": "bill_balance",
52
+ "28": "improve_credit_score",
53
+ "29": "change_language",
54
+ "30": "no",
55
+ "31": "measurement_conversion",
56
+ "32": "timer",
57
+ "33": "flip_coin",
58
+ "34": "do_you_have_pets",
59
+ "35": "balance",
60
+ "36": "tell_joke",
61
+ "37": "last_maintenance",
62
+ "38": "exchange_rate",
63
+ "39": "uber",
64
+ "40": "car_rental",
65
+ "41": "credit_limit",
66
+ "42": "oos",
67
+ "43": "shopping_list",
68
+ "44": "expiration_date",
69
+ "45": "routing",
70
+ "46": "meal_suggestion",
71
+ "47": "tire_change",
72
+ "48": "todo_list",
73
+ "49": "card_declined",
74
+ "50": "rewards_balance",
75
+ "51": "change_accent",
76
+ "52": "vaccines",
77
+ "53": "reminder_update",
78
+ "54": "food_last",
79
+ "55": "change_ai_name",
80
+ "56": "bill_due",
81
+ "57": "who_do_you_work_for",
82
+ "58": "share_location",
83
+ "59": "international_visa",
84
+ "60": "calendar",
85
+ "61": "translate",
86
+ "62": "carry_on",
87
+ "63": "book_flight",
88
+ "64": "insurance_change",
89
+ "65": "todo_list_update",
90
+ "66": "timezone",
91
+ "67": "cancel_reservation",
92
+ "68": "transactions",
93
+ "69": "credit_score",
94
+ "70": "report_fraud",
95
+ "71": "spending_history",
96
+ "72": "directions",
97
+ "73": "spelling",
98
+ "74": "insurance",
99
+ "75": "what_is_your_name",
100
+ "76": "reminder",
101
+ "77": "where_are_you_from",
102
+ "78": "distance",
103
+ "79": "payday",
104
+ "80": "flight_status",
105
+ "81": "find_phone",
106
+ "82": "greeting",
107
+ "83": "alarm",
108
+ "84": "order_status",
109
+ "85": "confirm_reservation",
110
+ "86": "cook_time",
111
+ "87": "damaged_card",
112
+ "88": "reset_settings",
113
+ "89": "pin_change",
114
+ "90": "replacement_card_duration",
115
+ "91": "new_card",
116
+ "92": "roll_dice",
117
+ "93": "income",
118
+ "94": "taxes",
119
+ "95": "date",
120
+ "96": "who_made_you",
121
+ "97": "pto_request",
122
+ "98": "tire_pressure",
123
+ "99": "how_old_are_you",
124
+ "100": "rollover_401k",
125
+ "101": "pto_request_status",
126
+ "102": "how_busy",
127
+ "103": "application_status",
128
+ "104": "recipe",
129
+ "105": "calendar_update",
130
+ "106": "play_music",
131
+ "107": "yes",
132
+ "108": "direct_deposit",
133
+ "109": "credit_limit_change",
134
+ "110": "gas",
135
+ "111": "pay_bill",
136
+ "112": "ingredients_list",
137
+ "113": "lost_luggage",
138
+ "114": "goodbye",
139
+ "115": "what_can_i_ask_you",
140
+ "116": "book_hotel",
141
+ "117": "are_you_a_bot",
142
+ "118": "next_song",
143
+ "119": "change_speed",
144
+ "120": "plug_type",
145
+ "121": "maybe",
146
+ "122": "w2",
147
+ "123": "oil_change_when",
148
+ "124": "thank_you",
149
+ "125": "shopping_list_update",
150
+ "126": "pto_balance",
151
+ "127": "order_checks",
152
+ "128": "travel_alert",
153
+ "129": "fun_fact",
154
+ "130": "sync_device",
155
+ "131": "schedule_maintenance",
156
+ "132": "apr",
157
+ "133": "transfer",
158
+ "134": "ingredient_substitution",
159
+ "135": "calories",
160
+ "136": "current_location",
161
+ "137": "international_fees",
162
+ "138": "calculator",
163
+ "139": "definition",
164
+ "140": "next_holiday",
165
+ "141": "update_playlist",
166
+ "142": "mpg",
167
+ "143": "min_payment",
168
+ "144": "change_user_name",
169
+ "145": "restaurant_suggestion",
170
+ "146": "travel_notification",
171
+ "147": "cancel",
172
+ "148": "pto_used",
173
+ "149": "travel_suggestion",
174
+ "150": "change_volume"
175
  },
176
  "initializer_cutoff_factor": 2.0,
177
  "initializer_range": 0.02,
178
  "intermediate_size": 1152,
179
  "label2id": {
180
+ "accept_reservations": 9,
181
+ "account_blocked": 2,
182
+ "alarm": 83,
183
+ "application_status": 103,
184
+ "apr": 132,
185
+ "are_you_a_bot": 117,
186
+ "balance": 35,
187
+ "bill_balance": 27,
188
+ "bill_due": 56,
189
+ "book_flight": 63,
190
+ "book_hotel": 116,
191
+ "calculator": 138,
192
+ "calendar": 60,
193
+ "calendar_update": 105,
194
+ "calories": 135,
195
+ "cancel": 147,
196
+ "cancel_reservation": 67,
197
+ "car_rental": 40,
198
+ "card_declined": 49,
199
+ "carry_on": 62,
200
+ "change_accent": 51,
201
+ "change_ai_name": 55,
202
+ "change_language": 29,
203
+ "change_speed": 119,
204
+ "change_user_name": 144,
205
+ "change_volume": 150,
206
+ "confirm_reservation": 85,
207
+ "cook_time": 86,
208
+ "credit_limit": 41,
209
+ "credit_limit_change": 109,
210
+ "credit_score": 69,
211
+ "current_location": 136,
212
+ "damaged_card": 87,
213
+ "date": 95,
214
+ "definition": 139,
215
+ "direct_deposit": 108,
216
+ "directions": 72,
217
+ "distance": 78,
218
+ "do_you_have_pets": 34,
219
+ "exchange_rate": 38,
220
+ "expiration_date": 44,
221
+ "find_phone": 81,
222
+ "flight_status": 80,
223
+ "flip_coin": 33,
224
+ "food_last": 54,
225
+ "freeze_account": 20,
226
+ "fun_fact": 129,
227
+ "gas": 110,
228
+ "gas_type": 8,
229
+ "goodbye": 114,
230
+ "greeting": 82,
231
+ "how_busy": 102,
232
+ "how_old_are_you": 99,
233
+ "improve_credit_score": 28,
234
+ "income": 93,
235
+ "ingredient_substitution": 134,
236
+ "ingredients_list": 112,
237
+ "insurance": 74,
238
+ "insurance_change": 64,
239
+ "interest_rate": 7,
240
+ "international_fees": 137,
241
+ "international_visa": 59,
242
+ "jump_start": 17,
243
+ "last_maintenance": 37,
244
+ "lost_luggage": 113,
245
+ "make_call": 25,
246
+ "maybe": 121,
247
+ "meal_suggestion": 46,
248
+ "meaning_of_life": 22,
249
+ "measurement_conversion": 31,
250
+ "meeting_schedule": 19,
251
+ "min_payment": 143,
252
+ "mpg": 142,
253
+ "new_card": 91,
254
+ "next_holiday": 140,
255
+ "next_song": 118,
256
+ "no": 30,
257
+ "nutrition_info": 1,
258
+ "oil_change_how": 3,
259
+ "oil_change_when": 123,
260
+ "oos": 42,
261
+ "order": 16,
262
+ "order_checks": 127,
263
+ "order_status": 84,
264
+ "pay_bill": 111,
265
+ "payday": 79,
266
+ "pin_change": 89,
267
+ "play_music": 106,
268
+ "plug_type": 120,
269
+ "pto_balance": 126,
270
+ "pto_request": 97,
271
+ "pto_request_status": 101,
272
+ "pto_used": 148,
273
+ "recipe": 104,
274
+ "redeem_rewards": 6,
275
+ "reminder": 76,
276
+ "reminder_update": 53,
277
+ "repeat": 13,
278
+ "replacement_card_duration": 90,
279
+ "report_fraud": 70,
280
+ "report_lost_card": 12,
281
+ "reset_settings": 88,
282
+ "restaurant_reservation": 23,
283
+ "restaurant_reviews": 0,
284
+ "restaurant_suggestion": 145,
285
+ "rewards_balance": 50,
286
+ "roll_dice": 92,
287
+ "rollover_401k": 100,
288
+ "routing": 45,
289
+ "schedule_maintenance": 131,
290
+ "schedule_meeting": 18,
291
+ "share_location": 58,
292
+ "shopping_list": 43,
293
+ "shopping_list_update": 125,
294
+ "smart_home": 10,
295
+ "spelling": 73,
296
+ "spending_history": 71,
297
+ "sync_device": 130,
298
+ "taxes": 94,
299
+ "tell_joke": 36,
300
+ "text": 26,
301
+ "thank_you": 124,
302
+ "time": 4,
303
+ "timer": 32,
304
+ "timezone": 66,
305
+ "tire_change": 47,
306
+ "tire_pressure": 98,
307
+ "todo_list": 48,
308
+ "todo_list_update": 65,
309
+ "traffic": 24,
310
+ "transactions": 68,
311
+ "transfer": 133,
312
+ "translate": 61,
313
+ "travel_alert": 128,
314
+ "travel_notification": 146,
315
+ "travel_suggestion": 149,
316
+ "uber": 39,
317
+ "update_playlist": 141,
318
+ "user_name": 11,
319
+ "vaccines": 52,
320
+ "w2": 122,
321
+ "weather": 5,
322
+ "what_are_your_hobbies": 15,
323
+ "what_can_i_ask_you": 115,
324
+ "what_is_your_name": 75,
325
+ "what_song": 21,
326
+ "where_are_you_from": 77,
327
+ "whisper_mode": 14,
328
+ "who_do_you_work_for": 57,
329
+ "who_made_you": 96,
330
+ "yes": 107
331
  },
332
  "layer_norm_eps": 1e-05,
333
  "local_attention": 128,
run-0/checkpoint-5724/model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:24a006b81a9845a728a3fa67d35c80dee72d4da31a4cbda44dd9977fb07d74d9
3
  size 598898116
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81ac3f88bdb451e4316877bdf137e199de70b195697d9835e23b0052e2b2bace
3
  size 598898116
run-0/checkpoint-5724/optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7d9e22354024db7c032a1b4a4ad71cdab8ca4062c9e574d1c954475436affc15
3
  size 1197886411
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:022d1a6aa461991f7bc9e38c5ac0505359b0feab028c560565db26ffc81ef9f4
3
  size 1197886411
run-0/checkpoint-5724/scaler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bb7b98119882cbdd89bb47ead830327452055697a63bdbf54f9abf37edb33f4e
3
  size 1383
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:507b0db25825069fb54cb54f098269305d212c34ae4cb16af06f08bb514410fc
3
  size 1383
run-0/checkpoint-5724/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:32c8b83ee05ef8c4193664ffb5ae67f72e54f0bd8969acf1bf3f302ab852e755
3
  size 1465
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:842c477c03985d57c436f042185e94f8b68859521c9e0660a688efc82daf99b1
3
  size 1465
run-0/checkpoint-5724/trainer_state.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "best_global_step": 3816,
3
- "best_metric": 0.9651612903225807,
4
  "best_model_checkpoint": "ModernBERT-base-finetuned-distilled-clinc/run-0/checkpoint-3816",
5
  "epoch": 6.0,
6
  "eval_steps": 500,
@@ -10,142 +10,107 @@
10
  "is_world_process_zero": true,
11
  "log_history": [
12
  {
13
- "epoch": 0.5241090146750524,
14
- "grad_norm": 25.577024459838867,
15
- "learning_rate": 1.8837642674120664e-05,
16
- "loss": 7.8275,
17
- "step": 500
18
  },
19
  {
20
  "epoch": 1.0,
21
- "eval_accuracy": 0.9361290322580645,
22
- "eval_loss": 1.787779688835144,
23
- "eval_runtime": 17.5117,
24
- "eval_samples_per_second": 177.025,
25
- "eval_steps_per_second": 11.078,
26
  "step": 954
27
  },
28
  {
29
- "epoch": 1.0482180293501049,
30
- "grad_norm": 15.088406562805176,
31
- "learning_rate": 1.767295597484277e-05,
32
- "loss": 2.5947,
33
- "step": 1000
34
- },
35
- {
36
- "epoch": 1.5723270440251573,
37
- "grad_norm": 24.63665199279785,
38
- "learning_rate": 1.6508269275564874e-05,
39
- "loss": 1.3397,
40
- "step": 1500
41
  },
42
  {
43
  "epoch": 2.0,
44
- "eval_accuracy": 0.9587096774193549,
45
- "eval_loss": 1.0319586992263794,
46
- "eval_runtime": 17.514,
47
- "eval_samples_per_second": 177.001,
48
- "eval_steps_per_second": 11.077,
49
  "step": 1908
50
  },
51
  {
52
- "epoch": 2.0964360587002098,
53
- "grad_norm": 5.869675636291504,
54
- "learning_rate": 1.534358257628698e-05,
55
- "loss": 0.9836,
56
- "step": 2000
57
- },
58
- {
59
- "epoch": 2.620545073375262,
60
- "grad_norm": 3.776374101638794,
61
- "learning_rate": 1.4178895877009086e-05,
62
- "loss": 0.7147,
63
- "step": 2500
64
  },
65
  {
66
  "epoch": 3.0,
67
- "eval_accuracy": 0.9641935483870968,
68
- "eval_loss": 0.8226571679115295,
69
- "eval_runtime": 17.6007,
70
- "eval_samples_per_second": 176.129,
71
- "eval_steps_per_second": 11.022,
72
  "step": 2862
73
  },
74
  {
75
- "epoch": 3.1446540880503147,
76
- "grad_norm": 3.600783586502075,
77
- "learning_rate": 1.3014209177731191e-05,
78
- "loss": 0.6283,
79
- "step": 3000
80
- },
81
- {
82
- "epoch": 3.668763102725367,
83
- "grad_norm": 4.3173909187316895,
84
- "learning_rate": 1.1849522478453297e-05,
85
- "loss": 0.5303,
86
- "step": 3500
87
  },
88
  {
89
  "epoch": 4.0,
90
- "eval_accuracy": 0.9651612903225807,
91
- "eval_loss": 0.738650381565094,
92
- "eval_runtime": 17.5566,
93
- "eval_samples_per_second": 176.571,
94
- "eval_steps_per_second": 11.05,
95
  "step": 3816
96
  },
97
  {
98
- "epoch": 4.1928721174004195,
99
- "grad_norm": 3.643286943435669,
100
- "learning_rate": 1.0684835779175402e-05,
101
- "loss": 0.4893,
102
- "step": 4000
103
- },
104
- {
105
- "epoch": 4.716981132075472,
106
- "grad_norm": 3.2583296298980713,
107
- "learning_rate": 9.520149079897509e-06,
108
- "loss": 0.4425,
109
- "step": 4500
110
  },
111
  {
112
  "epoch": 5.0,
113
- "eval_accuracy": 0.964516129032258,
114
- "eval_loss": 0.6903958320617676,
115
- "eval_runtime": 17.5542,
116
- "eval_samples_per_second": 176.596,
117
- "eval_steps_per_second": 11.051,
118
  "step": 4770
119
  },
120
  {
121
- "epoch": 5.241090146750524,
122
- "grad_norm": 2.9098360538482666,
123
- "learning_rate": 8.355462380619614e-06,
124
- "loss": 0.4084,
125
- "step": 5000
126
- },
127
- {
128
- "epoch": 5.765199161425577,
129
- "grad_norm": 3.1272575855255127,
130
- "learning_rate": 7.190775681341719e-06,
131
- "loss": 0.382,
132
- "step": 5500
133
  },
134
  {
135
  "epoch": 6.0,
136
- "eval_accuracy": 0.9641935483870968,
137
- "eval_loss": 0.6621975302696228,
138
- "eval_runtime": 17.5962,
139
- "eval_samples_per_second": 176.174,
140
- "eval_steps_per_second": 11.025,
141
  "step": 5724
142
  }
143
  ],
144
- "logging_steps": 500,
145
- "max_steps": 8586,
146
  "num_input_tokens_seen": 0,
147
- "num_train_epochs": 9,
148
- "save_steps": 500,
149
  "stateful_callbacks": {
150
  "EarlyStoppingCallback": {
151
  "args": {
@@ -167,12 +132,12 @@
167
  "attributes": {}
168
  }
169
  },
170
- "total_flos": 1116452502549504.0,
171
  "train_batch_size": 16,
172
  "trial_name": null,
173
  "trial_params": {
174
- "alpha": 0.5429119091572833,
175
- "num_train_epochs": 9,
176
- "temperature": 8
177
  }
178
  }
 
1
  {
2
  "best_global_step": 3816,
3
+ "best_metric": 0.9638709677419355,
4
  "best_model_checkpoint": "ModernBERT-base-finetuned-distilled-clinc/run-0/checkpoint-3816",
5
  "epoch": 6.0,
6
  "eval_steps": 500,
 
10
  "is_world_process_zero": true,
11
  "log_history": [
12
  {
13
+ "epoch": 0.9989517819706499,
14
+ "grad_norm": 7.180769920349121,
15
+ "learning_rate": 1.7148846960167716e-05,
16
+ "loss": 4.5857,
17
+ "step": 953
18
  },
19
  {
20
  "epoch": 1.0,
21
+ "eval_accuracy": 0.9364516129032258,
22
+ "eval_loss": 1.5901516675949097,
23
+ "eval_runtime": 13.3547,
24
+ "eval_samples_per_second": 232.128,
25
+ "eval_steps_per_second": 14.527,
26
  "step": 954
27
  },
28
  {
29
+ "epoch": 1.9979035639412999,
30
+ "grad_norm": 9.229886054992676,
31
+ "learning_rate": 1.4294699011680145e-05,
32
+ "loss": 1.0911,
33
+ "step": 1906
 
 
 
 
 
 
 
34
  },
35
  {
36
  "epoch": 2.0,
37
+ "eval_accuracy": 0.9570967741935484,
38
+ "eval_loss": 0.8912312388420105,
39
+ "eval_runtime": 13.3096,
40
+ "eval_samples_per_second": 232.915,
41
+ "eval_steps_per_second": 14.576,
42
  "step": 1908
43
  },
44
  {
45
+ "epoch": 2.99685534591195,
46
+ "grad_norm": 5.046926975250244,
47
+ "learning_rate": 1.1440551063192572e-05,
48
+ "loss": 0.6315,
49
+ "step": 2859
 
 
 
 
 
 
 
50
  },
51
  {
52
  "epoch": 3.0,
53
+ "eval_accuracy": 0.9609677419354838,
54
+ "eval_loss": 0.7318851351737976,
55
+ "eval_runtime": 13.4127,
56
+ "eval_samples_per_second": 231.124,
57
+ "eval_steps_per_second": 14.464,
58
  "step": 2862
59
  },
60
  {
61
+ "epoch": 3.9958071278825997,
62
+ "grad_norm": 3.7578282356262207,
63
+ "learning_rate": 8.586403114705001e-06,
64
+ "loss": 0.481,
65
+ "step": 3812
 
 
 
 
 
 
 
66
  },
67
  {
68
  "epoch": 4.0,
69
+ "eval_accuracy": 0.9638709677419355,
70
+ "eval_loss": 0.6467007994651794,
71
+ "eval_runtime": 13.39,
72
+ "eval_samples_per_second": 231.516,
73
+ "eval_steps_per_second": 14.488,
74
  "step": 3816
75
  },
76
  {
77
+ "epoch": 4.99475890985325,
78
+ "grad_norm": 2.5420217514038086,
79
+ "learning_rate": 5.732255166217431e-06,
80
+ "loss": 0.4072,
81
+ "step": 4765
 
 
 
 
 
 
 
82
  },
83
  {
84
  "epoch": 5.0,
85
+ "eval_accuracy": 0.9632258064516129,
86
+ "eval_loss": 0.607825517654419,
87
+ "eval_runtime": 13.2517,
88
+ "eval_samples_per_second": 233.933,
89
+ "eval_steps_per_second": 14.64,
90
  "step": 4770
91
  },
92
  {
93
+ "epoch": 5.9937106918239,
94
+ "grad_norm": 6.932933330535889,
95
+ "learning_rate": 2.8781072177298598e-06,
96
+ "loss": 0.3637,
97
+ "step": 5718
 
 
 
 
 
 
 
98
  },
99
  {
100
  "epoch": 6.0,
101
+ "eval_accuracy": 0.9632258064516129,
102
+ "eval_loss": 0.5857376456260681,
103
+ "eval_runtime": 13.3136,
104
+ "eval_samples_per_second": 232.845,
105
+ "eval_steps_per_second": 14.572,
106
  "step": 5724
107
  }
108
  ],
109
+ "logging_steps": 953,
110
+ "max_steps": 6678,
111
  "num_input_tokens_seen": 0,
112
+ "num_train_epochs": 7,
113
+ "save_steps": 1000000000.0,
114
  "stateful_callbacks": {
115
  "EarlyStoppingCallback": {
116
  "args": {
 
132
  "attributes": {}
133
  }
134
  },
135
+ "total_flos": 1255667955529344.0,
136
  "train_batch_size": 16,
137
  "trial_name": null,
138
  "trial_params": {
139
+ "alpha": 0.5630385610905387,
140
+ "num_train_epochs": 7,
141
+ "temperature": 9
142
  }
143
  }
run-0/checkpoint-5724/training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9214c4155d6ffda26ae1d2fc305125bdc6e4db065907bac5bb45379f08c56ec9
3
- size 5969
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d2c126b5b10d3fa5a3cf6095b18e906cad6909d2dc20a3626092a9a968fbe44
3
+ size 5905