Training in progress, step 2862
Browse files- model.safetensors +1 -1
- run-0/checkpoint-1590/config.json +331 -0
- run-0/checkpoint-1590/model.safetensors +3 -0
- run-0/checkpoint-1590/optimizer.pt +3 -0
- run-0/checkpoint-1590/rng_state.pth +3 -0
- run-0/checkpoint-1590/scheduler.pt +3 -0
- run-0/checkpoint-1590/special_tokens_map.json +7 -0
- run-0/checkpoint-1590/tokenizer.json +0 -0
- run-0/checkpoint-1590/tokenizer_config.json +55 -0
- run-0/checkpoint-1590/trainer_state.json +108 -0
- run-0/checkpoint-1590/training_args.bin +3 -0
- run-0/checkpoint-1590/vocab.txt +0 -0
- run-1/checkpoint-3180/config.json +331 -0
- run-1/checkpoint-3180/model.safetensors +3 -0
- run-1/checkpoint-3180/optimizer.pt +3 -0
- run-1/checkpoint-3180/rng_state.pth +3 -0
- run-1/checkpoint-3180/scheduler.pt +3 -0
- run-1/checkpoint-3180/special_tokens_map.json +7 -0
- run-1/checkpoint-3180/tokenizer.json +0 -0
- run-1/checkpoint-3180/tokenizer_config.json +55 -0
- run-1/checkpoint-3180/trainer_state.json +188 -0
- run-1/checkpoint-3180/training_args.bin +3 -0
- run-1/checkpoint-3180/vocab.txt +0 -0
- run-2/checkpoint-1590/config.json +331 -0
- run-2/checkpoint-1590/model.safetensors +3 -0
- run-2/checkpoint-1590/optimizer.pt +3 -0
- run-2/checkpoint-1590/rng_state.pth +3 -0
- run-2/checkpoint-1590/scheduler.pt +3 -0
- run-2/checkpoint-1590/special_tokens_map.json +7 -0
- run-2/checkpoint-1590/tokenizer.json +0 -0
- run-2/checkpoint-1590/tokenizer_config.json +55 -0
- run-2/checkpoint-1590/trainer_state.json +108 -0
- run-2/checkpoint-1590/training_args.bin +3 -0
- run-2/checkpoint-1590/vocab.txt +0 -0
- run-3/checkpoint-2862/config.json +331 -0
- run-3/checkpoint-2862/model.safetensors +3 -0
- run-3/checkpoint-2862/optimizer.pt +3 -0
- run-3/checkpoint-2862/rng_state.pth +3 -0
- run-3/checkpoint-2862/scheduler.pt +3 -0
- run-3/checkpoint-2862/special_tokens_map.json +7 -0
- run-3/checkpoint-2862/tokenizer.json +0 -0
- run-3/checkpoint-2862/tokenizer_config.json +55 -0
- run-3/checkpoint-2862/trainer_state.json +172 -0
- run-3/checkpoint-2862/training_args.bin +3 -0
- run-3/checkpoint-2862/vocab.txt +0 -0
- runs/Oct23_23-31-51_p16/events.out.tfevents.1729698645.p16.148475.2 +3 -0
- runs/Oct23_23-31-51_p16/events.out.tfevents.1729698905.p16.148475.3 +3 -0
- runs/Oct23_23-31-51_p16/events.out.tfevents.1729699275.p16.148475.4 +3 -0
- runs/Oct23_23-31-51_p16/events.out.tfevents.1729699541.p16.148475.5 +3 -0
- training_args.bin +1 -1
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 268290900
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9a7bec75e776cd487733ff1e96487bc2505714d839c2eeecdc042accc0f1b3a9
|
| 3 |
size 268290900
|
run-0/checkpoint-1590/config.json
ADDED
|
@@ -0,0 +1,331 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "distilbert-base-uncased",
|
| 3 |
+
"activation": "gelu",
|
| 4 |
+
"architectures": [
|
| 5 |
+
"DistilBertForSequenceClassification"
|
| 6 |
+
],
|
| 7 |
+
"attention_dropout": 0.1,
|
| 8 |
+
"dim": 768,
|
| 9 |
+
"dropout": 0.1,
|
| 10 |
+
"hidden_dim": 3072,
|
| 11 |
+
"id2label": {
|
| 12 |
+
"0": "restaurant_reviews",
|
| 13 |
+
"1": "nutrition_info",
|
| 14 |
+
"2": "account_blocked",
|
| 15 |
+
"3": "oil_change_how",
|
| 16 |
+
"4": "time",
|
| 17 |
+
"5": "weather",
|
| 18 |
+
"6": "redeem_rewards",
|
| 19 |
+
"7": "interest_rate",
|
| 20 |
+
"8": "gas_type",
|
| 21 |
+
"9": "accept_reservations",
|
| 22 |
+
"10": "smart_home",
|
| 23 |
+
"11": "user_name",
|
| 24 |
+
"12": "report_lost_card",
|
| 25 |
+
"13": "repeat",
|
| 26 |
+
"14": "whisper_mode",
|
| 27 |
+
"15": "what_are_your_hobbies",
|
| 28 |
+
"16": "order",
|
| 29 |
+
"17": "jump_start",
|
| 30 |
+
"18": "schedule_meeting",
|
| 31 |
+
"19": "meeting_schedule",
|
| 32 |
+
"20": "freeze_account",
|
| 33 |
+
"21": "what_song",
|
| 34 |
+
"22": "meaning_of_life",
|
| 35 |
+
"23": "restaurant_reservation",
|
| 36 |
+
"24": "traffic",
|
| 37 |
+
"25": "make_call",
|
| 38 |
+
"26": "text",
|
| 39 |
+
"27": "bill_balance",
|
| 40 |
+
"28": "improve_credit_score",
|
| 41 |
+
"29": "change_language",
|
| 42 |
+
"30": "no",
|
| 43 |
+
"31": "measurement_conversion",
|
| 44 |
+
"32": "timer",
|
| 45 |
+
"33": "flip_coin",
|
| 46 |
+
"34": "do_you_have_pets",
|
| 47 |
+
"35": "balance",
|
| 48 |
+
"36": "tell_joke",
|
| 49 |
+
"37": "last_maintenance",
|
| 50 |
+
"38": "exchange_rate",
|
| 51 |
+
"39": "uber",
|
| 52 |
+
"40": "car_rental",
|
| 53 |
+
"41": "credit_limit",
|
| 54 |
+
"42": "oos",
|
| 55 |
+
"43": "shopping_list",
|
| 56 |
+
"44": "expiration_date",
|
| 57 |
+
"45": "routing",
|
| 58 |
+
"46": "meal_suggestion",
|
| 59 |
+
"47": "tire_change",
|
| 60 |
+
"48": "todo_list",
|
| 61 |
+
"49": "card_declined",
|
| 62 |
+
"50": "rewards_balance",
|
| 63 |
+
"51": "change_accent",
|
| 64 |
+
"52": "vaccines",
|
| 65 |
+
"53": "reminder_update",
|
| 66 |
+
"54": "food_last",
|
| 67 |
+
"55": "change_ai_name",
|
| 68 |
+
"56": "bill_due",
|
| 69 |
+
"57": "who_do_you_work_for",
|
| 70 |
+
"58": "share_location",
|
| 71 |
+
"59": "international_visa",
|
| 72 |
+
"60": "calendar",
|
| 73 |
+
"61": "translate",
|
| 74 |
+
"62": "carry_on",
|
| 75 |
+
"63": "book_flight",
|
| 76 |
+
"64": "insurance_change",
|
| 77 |
+
"65": "todo_list_update",
|
| 78 |
+
"66": "timezone",
|
| 79 |
+
"67": "cancel_reservation",
|
| 80 |
+
"68": "transactions",
|
| 81 |
+
"69": "credit_score",
|
| 82 |
+
"70": "report_fraud",
|
| 83 |
+
"71": "spending_history",
|
| 84 |
+
"72": "directions",
|
| 85 |
+
"73": "spelling",
|
| 86 |
+
"74": "insurance",
|
| 87 |
+
"75": "what_is_your_name",
|
| 88 |
+
"76": "reminder",
|
| 89 |
+
"77": "where_are_you_from",
|
| 90 |
+
"78": "distance",
|
| 91 |
+
"79": "payday",
|
| 92 |
+
"80": "flight_status",
|
| 93 |
+
"81": "find_phone",
|
| 94 |
+
"82": "greeting",
|
| 95 |
+
"83": "alarm",
|
| 96 |
+
"84": "order_status",
|
| 97 |
+
"85": "confirm_reservation",
|
| 98 |
+
"86": "cook_time",
|
| 99 |
+
"87": "damaged_card",
|
| 100 |
+
"88": "reset_settings",
|
| 101 |
+
"89": "pin_change",
|
| 102 |
+
"90": "replacement_card_duration",
|
| 103 |
+
"91": "new_card",
|
| 104 |
+
"92": "roll_dice",
|
| 105 |
+
"93": "income",
|
| 106 |
+
"94": "taxes",
|
| 107 |
+
"95": "date",
|
| 108 |
+
"96": "who_made_you",
|
| 109 |
+
"97": "pto_request",
|
| 110 |
+
"98": "tire_pressure",
|
| 111 |
+
"99": "how_old_are_you",
|
| 112 |
+
"100": "rollover_401k",
|
| 113 |
+
"101": "pto_request_status",
|
| 114 |
+
"102": "how_busy",
|
| 115 |
+
"103": "application_status",
|
| 116 |
+
"104": "recipe",
|
| 117 |
+
"105": "calendar_update",
|
| 118 |
+
"106": "play_music",
|
| 119 |
+
"107": "yes",
|
| 120 |
+
"108": "direct_deposit",
|
| 121 |
+
"109": "credit_limit_change",
|
| 122 |
+
"110": "gas",
|
| 123 |
+
"111": "pay_bill",
|
| 124 |
+
"112": "ingredients_list",
|
| 125 |
+
"113": "lost_luggage",
|
| 126 |
+
"114": "goodbye",
|
| 127 |
+
"115": "what_can_i_ask_you",
|
| 128 |
+
"116": "book_hotel",
|
| 129 |
+
"117": "are_you_a_bot",
|
| 130 |
+
"118": "next_song",
|
| 131 |
+
"119": "change_speed",
|
| 132 |
+
"120": "plug_type",
|
| 133 |
+
"121": "maybe",
|
| 134 |
+
"122": "w2",
|
| 135 |
+
"123": "oil_change_when",
|
| 136 |
+
"124": "thank_you",
|
| 137 |
+
"125": "shopping_list_update",
|
| 138 |
+
"126": "pto_balance",
|
| 139 |
+
"127": "order_checks",
|
| 140 |
+
"128": "travel_alert",
|
| 141 |
+
"129": "fun_fact",
|
| 142 |
+
"130": "sync_device",
|
| 143 |
+
"131": "schedule_maintenance",
|
| 144 |
+
"132": "apr",
|
| 145 |
+
"133": "transfer",
|
| 146 |
+
"134": "ingredient_substitution",
|
| 147 |
+
"135": "calories",
|
| 148 |
+
"136": "current_location",
|
| 149 |
+
"137": "international_fees",
|
| 150 |
+
"138": "calculator",
|
| 151 |
+
"139": "definition",
|
| 152 |
+
"140": "next_holiday",
|
| 153 |
+
"141": "update_playlist",
|
| 154 |
+
"142": "mpg",
|
| 155 |
+
"143": "min_payment",
|
| 156 |
+
"144": "change_user_name",
|
| 157 |
+
"145": "restaurant_suggestion",
|
| 158 |
+
"146": "travel_notification",
|
| 159 |
+
"147": "cancel",
|
| 160 |
+
"148": "pto_used",
|
| 161 |
+
"149": "travel_suggestion",
|
| 162 |
+
"150": "change_volume"
|
| 163 |
+
},
|
| 164 |
+
"initializer_range": 0.02,
|
| 165 |
+
"label2id": {
|
| 166 |
+
"accept_reservations": 9,
|
| 167 |
+
"account_blocked": 2,
|
| 168 |
+
"alarm": 83,
|
| 169 |
+
"application_status": 103,
|
| 170 |
+
"apr": 132,
|
| 171 |
+
"are_you_a_bot": 117,
|
| 172 |
+
"balance": 35,
|
| 173 |
+
"bill_balance": 27,
|
| 174 |
+
"bill_due": 56,
|
| 175 |
+
"book_flight": 63,
|
| 176 |
+
"book_hotel": 116,
|
| 177 |
+
"calculator": 138,
|
| 178 |
+
"calendar": 60,
|
| 179 |
+
"calendar_update": 105,
|
| 180 |
+
"calories": 135,
|
| 181 |
+
"cancel": 147,
|
| 182 |
+
"cancel_reservation": 67,
|
| 183 |
+
"car_rental": 40,
|
| 184 |
+
"card_declined": 49,
|
| 185 |
+
"carry_on": 62,
|
| 186 |
+
"change_accent": 51,
|
| 187 |
+
"change_ai_name": 55,
|
| 188 |
+
"change_language": 29,
|
| 189 |
+
"change_speed": 119,
|
| 190 |
+
"change_user_name": 144,
|
| 191 |
+
"change_volume": 150,
|
| 192 |
+
"confirm_reservation": 85,
|
| 193 |
+
"cook_time": 86,
|
| 194 |
+
"credit_limit": 41,
|
| 195 |
+
"credit_limit_change": 109,
|
| 196 |
+
"credit_score": 69,
|
| 197 |
+
"current_location": 136,
|
| 198 |
+
"damaged_card": 87,
|
| 199 |
+
"date": 95,
|
| 200 |
+
"definition": 139,
|
| 201 |
+
"direct_deposit": 108,
|
| 202 |
+
"directions": 72,
|
| 203 |
+
"distance": 78,
|
| 204 |
+
"do_you_have_pets": 34,
|
| 205 |
+
"exchange_rate": 38,
|
| 206 |
+
"expiration_date": 44,
|
| 207 |
+
"find_phone": 81,
|
| 208 |
+
"flight_status": 80,
|
| 209 |
+
"flip_coin": 33,
|
| 210 |
+
"food_last": 54,
|
| 211 |
+
"freeze_account": 20,
|
| 212 |
+
"fun_fact": 129,
|
| 213 |
+
"gas": 110,
|
| 214 |
+
"gas_type": 8,
|
| 215 |
+
"goodbye": 114,
|
| 216 |
+
"greeting": 82,
|
| 217 |
+
"how_busy": 102,
|
| 218 |
+
"how_old_are_you": 99,
|
| 219 |
+
"improve_credit_score": 28,
|
| 220 |
+
"income": 93,
|
| 221 |
+
"ingredient_substitution": 134,
|
| 222 |
+
"ingredients_list": 112,
|
| 223 |
+
"insurance": 74,
|
| 224 |
+
"insurance_change": 64,
|
| 225 |
+
"interest_rate": 7,
|
| 226 |
+
"international_fees": 137,
|
| 227 |
+
"international_visa": 59,
|
| 228 |
+
"jump_start": 17,
|
| 229 |
+
"last_maintenance": 37,
|
| 230 |
+
"lost_luggage": 113,
|
| 231 |
+
"make_call": 25,
|
| 232 |
+
"maybe": 121,
|
| 233 |
+
"meal_suggestion": 46,
|
| 234 |
+
"meaning_of_life": 22,
|
| 235 |
+
"measurement_conversion": 31,
|
| 236 |
+
"meeting_schedule": 19,
|
| 237 |
+
"min_payment": 143,
|
| 238 |
+
"mpg": 142,
|
| 239 |
+
"new_card": 91,
|
| 240 |
+
"next_holiday": 140,
|
| 241 |
+
"next_song": 118,
|
| 242 |
+
"no": 30,
|
| 243 |
+
"nutrition_info": 1,
|
| 244 |
+
"oil_change_how": 3,
|
| 245 |
+
"oil_change_when": 123,
|
| 246 |
+
"oos": 42,
|
| 247 |
+
"order": 16,
|
| 248 |
+
"order_checks": 127,
|
| 249 |
+
"order_status": 84,
|
| 250 |
+
"pay_bill": 111,
|
| 251 |
+
"payday": 79,
|
| 252 |
+
"pin_change": 89,
|
| 253 |
+
"play_music": 106,
|
| 254 |
+
"plug_type": 120,
|
| 255 |
+
"pto_balance": 126,
|
| 256 |
+
"pto_request": 97,
|
| 257 |
+
"pto_request_status": 101,
|
| 258 |
+
"pto_used": 148,
|
| 259 |
+
"recipe": 104,
|
| 260 |
+
"redeem_rewards": 6,
|
| 261 |
+
"reminder": 76,
|
| 262 |
+
"reminder_update": 53,
|
| 263 |
+
"repeat": 13,
|
| 264 |
+
"replacement_card_duration": 90,
|
| 265 |
+
"report_fraud": 70,
|
| 266 |
+
"report_lost_card": 12,
|
| 267 |
+
"reset_settings": 88,
|
| 268 |
+
"restaurant_reservation": 23,
|
| 269 |
+
"restaurant_reviews": 0,
|
| 270 |
+
"restaurant_suggestion": 145,
|
| 271 |
+
"rewards_balance": 50,
|
| 272 |
+
"roll_dice": 92,
|
| 273 |
+
"rollover_401k": 100,
|
| 274 |
+
"routing": 45,
|
| 275 |
+
"schedule_maintenance": 131,
|
| 276 |
+
"schedule_meeting": 18,
|
| 277 |
+
"share_location": 58,
|
| 278 |
+
"shopping_list": 43,
|
| 279 |
+
"shopping_list_update": 125,
|
| 280 |
+
"smart_home": 10,
|
| 281 |
+
"spelling": 73,
|
| 282 |
+
"spending_history": 71,
|
| 283 |
+
"sync_device": 130,
|
| 284 |
+
"taxes": 94,
|
| 285 |
+
"tell_joke": 36,
|
| 286 |
+
"text": 26,
|
| 287 |
+
"thank_you": 124,
|
| 288 |
+
"time": 4,
|
| 289 |
+
"timer": 32,
|
| 290 |
+
"timezone": 66,
|
| 291 |
+
"tire_change": 47,
|
| 292 |
+
"tire_pressure": 98,
|
| 293 |
+
"todo_list": 48,
|
| 294 |
+
"todo_list_update": 65,
|
| 295 |
+
"traffic": 24,
|
| 296 |
+
"transactions": 68,
|
| 297 |
+
"transfer": 133,
|
| 298 |
+
"translate": 61,
|
| 299 |
+
"travel_alert": 128,
|
| 300 |
+
"travel_notification": 146,
|
| 301 |
+
"travel_suggestion": 149,
|
| 302 |
+
"uber": 39,
|
| 303 |
+
"update_playlist": 141,
|
| 304 |
+
"user_name": 11,
|
| 305 |
+
"vaccines": 52,
|
| 306 |
+
"w2": 122,
|
| 307 |
+
"weather": 5,
|
| 308 |
+
"what_are_your_hobbies": 15,
|
| 309 |
+
"what_can_i_ask_you": 115,
|
| 310 |
+
"what_is_your_name": 75,
|
| 311 |
+
"what_song": 21,
|
| 312 |
+
"where_are_you_from": 77,
|
| 313 |
+
"whisper_mode": 14,
|
| 314 |
+
"who_do_you_work_for": 57,
|
| 315 |
+
"who_made_you": 96,
|
| 316 |
+
"yes": 107
|
| 317 |
+
},
|
| 318 |
+
"max_position_embeddings": 512,
|
| 319 |
+
"model_type": "distilbert",
|
| 320 |
+
"n_heads": 12,
|
| 321 |
+
"n_layers": 6,
|
| 322 |
+
"pad_token_id": 0,
|
| 323 |
+
"problem_type": "single_label_classification",
|
| 324 |
+
"qa_dropout": 0.1,
|
| 325 |
+
"seq_classif_dropout": 0.2,
|
| 326 |
+
"sinusoidal_pos_embds": false,
|
| 327 |
+
"tie_weights_": true,
|
| 328 |
+
"torch_dtype": "float32",
|
| 329 |
+
"transformers_version": "4.45.2",
|
| 330 |
+
"vocab_size": 30522
|
| 331 |
+
}
|
run-0/checkpoint-1590/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a75f712ffe2cccb83a37473554292b57f169534081cb726f946cae2d6e382fe5
|
| 3 |
+
size 268290900
|
run-0/checkpoint-1590/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ffd56295e7c4f4bdb7d729a3751d539d79d0c53193448db3b760963dbdbddd32
|
| 3 |
+
size 536643898
|
run-0/checkpoint-1590/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a66c0cd02823b9fa3c17cb8050d4f36f19ab8409fe337d5d545a289b3ca30007
|
| 3 |
+
size 14244
|
run-0/checkpoint-1590/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ee1082c0efb09baac545ba26829a8ad0ace46851b360d7c70f5f11088ecb04f2
|
| 3 |
+
size 1064
|
run-0/checkpoint-1590/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 |
+
}
|
run-0/checkpoint-1590/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
run-0/checkpoint-1590/tokenizer_config.json
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "[PAD]",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"100": {
|
| 12 |
+
"content": "[UNK]",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"101": {
|
| 20 |
+
"content": "[CLS]",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"102": {
|
| 28 |
+
"content": "[SEP]",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"103": {
|
| 36 |
+
"content": "[MASK]",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"clean_up_tokenization_spaces": false,
|
| 45 |
+
"cls_token": "[CLS]",
|
| 46 |
+
"do_lower_case": true,
|
| 47 |
+
"mask_token": "[MASK]",
|
| 48 |
+
"model_max_length": 512,
|
| 49 |
+
"pad_token": "[PAD]",
|
| 50 |
+
"sep_token": "[SEP]",
|
| 51 |
+
"strip_accents": null,
|
| 52 |
+
"tokenize_chinese_chars": true,
|
| 53 |
+
"tokenizer_class": "DistilBertTokenizer",
|
| 54 |
+
"unk_token": "[UNK]"
|
| 55 |
+
}
|
run-0/checkpoint-1590/trainer_state.json
ADDED
|
@@ -0,0 +1,108 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1590,
|
| 7 |
+
"is_hyper_param_search": true,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.9968553459119497,
|
| 13 |
+
"grad_norm": 0.6873843669891357,
|
| 14 |
+
"learning_rate": 1.6012578616352204e-05,
|
| 15 |
+
"loss": 0.3856,
|
| 16 |
+
"step": 317
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 1.0,
|
| 20 |
+
"eval_accuracy": 0.5548387096774193,
|
| 21 |
+
"eval_loss": 0.1993625909090042,
|
| 22 |
+
"eval_runtime": 1.6757,
|
| 23 |
+
"eval_samples_per_second": 1850.019,
|
| 24 |
+
"eval_steps_per_second": 38.791,
|
| 25 |
+
"step": 318
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 1.9937106918238994,
|
| 29 |
+
"grad_norm": 0.48621249198913574,
|
| 30 |
+
"learning_rate": 1.2025157232704403e-05,
|
| 31 |
+
"loss": 0.1708,
|
| 32 |
+
"step": 634
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 2.0,
|
| 36 |
+
"eval_accuracy": 0.7961290322580645,
|
| 37 |
+
"eval_loss": 0.1054580882191658,
|
| 38 |
+
"eval_runtime": 1.6956,
|
| 39 |
+
"eval_samples_per_second": 1828.253,
|
| 40 |
+
"eval_steps_per_second": 38.334,
|
| 41 |
+
"step": 636
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 2.990566037735849,
|
| 45 |
+
"grad_norm": 0.469368577003479,
|
| 46 |
+
"learning_rate": 8.037735849056606e-06,
|
| 47 |
+
"loss": 0.1136,
|
| 48 |
+
"step": 951
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 3.0,
|
| 52 |
+
"eval_accuracy": 0.8574193548387097,
|
| 53 |
+
"eval_loss": 0.07675347477197647,
|
| 54 |
+
"eval_runtime": 1.7225,
|
| 55 |
+
"eval_samples_per_second": 1799.687,
|
| 56 |
+
"eval_steps_per_second": 37.735,
|
| 57 |
+
"step": 954
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 3.9874213836477987,
|
| 61 |
+
"grad_norm": 0.4127233624458313,
|
| 62 |
+
"learning_rate": 4.0503144654088055e-06,
|
| 63 |
+
"loss": 0.0922,
|
| 64 |
+
"step": 1268
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 4.0,
|
| 68 |
+
"eval_accuracy": 0.8716129032258064,
|
| 69 |
+
"eval_loss": 0.0650419071316719,
|
| 70 |
+
"eval_runtime": 1.7329,
|
| 71 |
+
"eval_samples_per_second": 1788.887,
|
| 72 |
+
"eval_steps_per_second": 37.509,
|
| 73 |
+
"step": 1272
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 4.984276729559748,
|
| 77 |
+
"grad_norm": 0.3342190086841583,
|
| 78 |
+
"learning_rate": 6.289308176100629e-08,
|
| 79 |
+
"loss": 0.0835,
|
| 80 |
+
"step": 1585
|
| 81 |
+
}
|
| 82 |
+
],
|
| 83 |
+
"logging_steps": 317,
|
| 84 |
+
"max_steps": 1590,
|
| 85 |
+
"num_input_tokens_seen": 0,
|
| 86 |
+
"num_train_epochs": 5,
|
| 87 |
+
"save_steps": 1000000000.0,
|
| 88 |
+
"stateful_callbacks": {
|
| 89 |
+
"TrainerControl": {
|
| 90 |
+
"args": {
|
| 91 |
+
"should_epoch_stop": false,
|
| 92 |
+
"should_evaluate": false,
|
| 93 |
+
"should_log": false,
|
| 94 |
+
"should_save": true,
|
| 95 |
+
"should_training_stop": true
|
| 96 |
+
},
|
| 97 |
+
"attributes": {}
|
| 98 |
+
}
|
| 99 |
+
},
|
| 100 |
+
"total_flos": 413495819944032.0,
|
| 101 |
+
"train_batch_size": 48,
|
| 102 |
+
"trial_name": null,
|
| 103 |
+
"trial_params": {
|
| 104 |
+
"alpha": 0.48126961318325356,
|
| 105 |
+
"num_train_epochs": 5,
|
| 106 |
+
"temperature": 20
|
| 107 |
+
}
|
| 108 |
+
}
|
run-0/checkpoint-1590/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:48c6d02ce55c1cad47514b732e15f02ecf493b5aa4d1c1200f5099c610e116c6
|
| 3 |
+
size 5304
|
run-0/checkpoint-1590/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
run-1/checkpoint-3180/config.json
ADDED
|
@@ -0,0 +1,331 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "distilbert-base-uncased",
|
| 3 |
+
"activation": "gelu",
|
| 4 |
+
"architectures": [
|
| 5 |
+
"DistilBertForSequenceClassification"
|
| 6 |
+
],
|
| 7 |
+
"attention_dropout": 0.1,
|
| 8 |
+
"dim": 768,
|
| 9 |
+
"dropout": 0.1,
|
| 10 |
+
"hidden_dim": 3072,
|
| 11 |
+
"id2label": {
|
| 12 |
+
"0": "restaurant_reviews",
|
| 13 |
+
"1": "nutrition_info",
|
| 14 |
+
"2": "account_blocked",
|
| 15 |
+
"3": "oil_change_how",
|
| 16 |
+
"4": "time",
|
| 17 |
+
"5": "weather",
|
| 18 |
+
"6": "redeem_rewards",
|
| 19 |
+
"7": "interest_rate",
|
| 20 |
+
"8": "gas_type",
|
| 21 |
+
"9": "accept_reservations",
|
| 22 |
+
"10": "smart_home",
|
| 23 |
+
"11": "user_name",
|
| 24 |
+
"12": "report_lost_card",
|
| 25 |
+
"13": "repeat",
|
| 26 |
+
"14": "whisper_mode",
|
| 27 |
+
"15": "what_are_your_hobbies",
|
| 28 |
+
"16": "order",
|
| 29 |
+
"17": "jump_start",
|
| 30 |
+
"18": "schedule_meeting",
|
| 31 |
+
"19": "meeting_schedule",
|
| 32 |
+
"20": "freeze_account",
|
| 33 |
+
"21": "what_song",
|
| 34 |
+
"22": "meaning_of_life",
|
| 35 |
+
"23": "restaurant_reservation",
|
| 36 |
+
"24": "traffic",
|
| 37 |
+
"25": "make_call",
|
| 38 |
+
"26": "text",
|
| 39 |
+
"27": "bill_balance",
|
| 40 |
+
"28": "improve_credit_score",
|
| 41 |
+
"29": "change_language",
|
| 42 |
+
"30": "no",
|
| 43 |
+
"31": "measurement_conversion",
|
| 44 |
+
"32": "timer",
|
| 45 |
+
"33": "flip_coin",
|
| 46 |
+
"34": "do_you_have_pets",
|
| 47 |
+
"35": "balance",
|
| 48 |
+
"36": "tell_joke",
|
| 49 |
+
"37": "last_maintenance",
|
| 50 |
+
"38": "exchange_rate",
|
| 51 |
+
"39": "uber",
|
| 52 |
+
"40": "car_rental",
|
| 53 |
+
"41": "credit_limit",
|
| 54 |
+
"42": "oos",
|
| 55 |
+
"43": "shopping_list",
|
| 56 |
+
"44": "expiration_date",
|
| 57 |
+
"45": "routing",
|
| 58 |
+
"46": "meal_suggestion",
|
| 59 |
+
"47": "tire_change",
|
| 60 |
+
"48": "todo_list",
|
| 61 |
+
"49": "card_declined",
|
| 62 |
+
"50": "rewards_balance",
|
| 63 |
+
"51": "change_accent",
|
| 64 |
+
"52": "vaccines",
|
| 65 |
+
"53": "reminder_update",
|
| 66 |
+
"54": "food_last",
|
| 67 |
+
"55": "change_ai_name",
|
| 68 |
+
"56": "bill_due",
|
| 69 |
+
"57": "who_do_you_work_for",
|
| 70 |
+
"58": "share_location",
|
| 71 |
+
"59": "international_visa",
|
| 72 |
+
"60": "calendar",
|
| 73 |
+
"61": "translate",
|
| 74 |
+
"62": "carry_on",
|
| 75 |
+
"63": "book_flight",
|
| 76 |
+
"64": "insurance_change",
|
| 77 |
+
"65": "todo_list_update",
|
| 78 |
+
"66": "timezone",
|
| 79 |
+
"67": "cancel_reservation",
|
| 80 |
+
"68": "transactions",
|
| 81 |
+
"69": "credit_score",
|
| 82 |
+
"70": "report_fraud",
|
| 83 |
+
"71": "spending_history",
|
| 84 |
+
"72": "directions",
|
| 85 |
+
"73": "spelling",
|
| 86 |
+
"74": "insurance",
|
| 87 |
+
"75": "what_is_your_name",
|
| 88 |
+
"76": "reminder",
|
| 89 |
+
"77": "where_are_you_from",
|
| 90 |
+
"78": "distance",
|
| 91 |
+
"79": "payday",
|
| 92 |
+
"80": "flight_status",
|
| 93 |
+
"81": "find_phone",
|
| 94 |
+
"82": "greeting",
|
| 95 |
+
"83": "alarm",
|
| 96 |
+
"84": "order_status",
|
| 97 |
+
"85": "confirm_reservation",
|
| 98 |
+
"86": "cook_time",
|
| 99 |
+
"87": "damaged_card",
|
| 100 |
+
"88": "reset_settings",
|
| 101 |
+
"89": "pin_change",
|
| 102 |
+
"90": "replacement_card_duration",
|
| 103 |
+
"91": "new_card",
|
| 104 |
+
"92": "roll_dice",
|
| 105 |
+
"93": "income",
|
| 106 |
+
"94": "taxes",
|
| 107 |
+
"95": "date",
|
| 108 |
+
"96": "who_made_you",
|
| 109 |
+
"97": "pto_request",
|
| 110 |
+
"98": "tire_pressure",
|
| 111 |
+
"99": "how_old_are_you",
|
| 112 |
+
"100": "rollover_401k",
|
| 113 |
+
"101": "pto_request_status",
|
| 114 |
+
"102": "how_busy",
|
| 115 |
+
"103": "application_status",
|
| 116 |
+
"104": "recipe",
|
| 117 |
+
"105": "calendar_update",
|
| 118 |
+
"106": "play_music",
|
| 119 |
+
"107": "yes",
|
| 120 |
+
"108": "direct_deposit",
|
| 121 |
+
"109": "credit_limit_change",
|
| 122 |
+
"110": "gas",
|
| 123 |
+
"111": "pay_bill",
|
| 124 |
+
"112": "ingredients_list",
|
| 125 |
+
"113": "lost_luggage",
|
| 126 |
+
"114": "goodbye",
|
| 127 |
+
"115": "what_can_i_ask_you",
|
| 128 |
+
"116": "book_hotel",
|
| 129 |
+
"117": "are_you_a_bot",
|
| 130 |
+
"118": "next_song",
|
| 131 |
+
"119": "change_speed",
|
| 132 |
+
"120": "plug_type",
|
| 133 |
+
"121": "maybe",
|
| 134 |
+
"122": "w2",
|
| 135 |
+
"123": "oil_change_when",
|
| 136 |
+
"124": "thank_you",
|
| 137 |
+
"125": "shopping_list_update",
|
| 138 |
+
"126": "pto_balance",
|
| 139 |
+
"127": "order_checks",
|
| 140 |
+
"128": "travel_alert",
|
| 141 |
+
"129": "fun_fact",
|
| 142 |
+
"130": "sync_device",
|
| 143 |
+
"131": "schedule_maintenance",
|
| 144 |
+
"132": "apr",
|
| 145 |
+
"133": "transfer",
|
| 146 |
+
"134": "ingredient_substitution",
|
| 147 |
+
"135": "calories",
|
| 148 |
+
"136": "current_location",
|
| 149 |
+
"137": "international_fees",
|
| 150 |
+
"138": "calculator",
|
| 151 |
+
"139": "definition",
|
| 152 |
+
"140": "next_holiday",
|
| 153 |
+
"141": "update_playlist",
|
| 154 |
+
"142": "mpg",
|
| 155 |
+
"143": "min_payment",
|
| 156 |
+
"144": "change_user_name",
|
| 157 |
+
"145": "restaurant_suggestion",
|
| 158 |
+
"146": "travel_notification",
|
| 159 |
+
"147": "cancel",
|
| 160 |
+
"148": "pto_used",
|
| 161 |
+
"149": "travel_suggestion",
|
| 162 |
+
"150": "change_volume"
|
| 163 |
+
},
|
| 164 |
+
"initializer_range": 0.02,
|
| 165 |
+
"label2id": {
|
| 166 |
+
"accept_reservations": 9,
|
| 167 |
+
"account_blocked": 2,
|
| 168 |
+
"alarm": 83,
|
| 169 |
+
"application_status": 103,
|
| 170 |
+
"apr": 132,
|
| 171 |
+
"are_you_a_bot": 117,
|
| 172 |
+
"balance": 35,
|
| 173 |
+
"bill_balance": 27,
|
| 174 |
+
"bill_due": 56,
|
| 175 |
+
"book_flight": 63,
|
| 176 |
+
"book_hotel": 116,
|
| 177 |
+
"calculator": 138,
|
| 178 |
+
"calendar": 60,
|
| 179 |
+
"calendar_update": 105,
|
| 180 |
+
"calories": 135,
|
| 181 |
+
"cancel": 147,
|
| 182 |
+
"cancel_reservation": 67,
|
| 183 |
+
"car_rental": 40,
|
| 184 |
+
"card_declined": 49,
|
| 185 |
+
"carry_on": 62,
|
| 186 |
+
"change_accent": 51,
|
| 187 |
+
"change_ai_name": 55,
|
| 188 |
+
"change_language": 29,
|
| 189 |
+
"change_speed": 119,
|
| 190 |
+
"change_user_name": 144,
|
| 191 |
+
"change_volume": 150,
|
| 192 |
+
"confirm_reservation": 85,
|
| 193 |
+
"cook_time": 86,
|
| 194 |
+
"credit_limit": 41,
|
| 195 |
+
"credit_limit_change": 109,
|
| 196 |
+
"credit_score": 69,
|
| 197 |
+
"current_location": 136,
|
| 198 |
+
"damaged_card": 87,
|
| 199 |
+
"date": 95,
|
| 200 |
+
"definition": 139,
|
| 201 |
+
"direct_deposit": 108,
|
| 202 |
+
"directions": 72,
|
| 203 |
+
"distance": 78,
|
| 204 |
+
"do_you_have_pets": 34,
|
| 205 |
+
"exchange_rate": 38,
|
| 206 |
+
"expiration_date": 44,
|
| 207 |
+
"find_phone": 81,
|
| 208 |
+
"flight_status": 80,
|
| 209 |
+
"flip_coin": 33,
|
| 210 |
+
"food_last": 54,
|
| 211 |
+
"freeze_account": 20,
|
| 212 |
+
"fun_fact": 129,
|
| 213 |
+
"gas": 110,
|
| 214 |
+
"gas_type": 8,
|
| 215 |
+
"goodbye": 114,
|
| 216 |
+
"greeting": 82,
|
| 217 |
+
"how_busy": 102,
|
| 218 |
+
"how_old_are_you": 99,
|
| 219 |
+
"improve_credit_score": 28,
|
| 220 |
+
"income": 93,
|
| 221 |
+
"ingredient_substitution": 134,
|
| 222 |
+
"ingredients_list": 112,
|
| 223 |
+
"insurance": 74,
|
| 224 |
+
"insurance_change": 64,
|
| 225 |
+
"interest_rate": 7,
|
| 226 |
+
"international_fees": 137,
|
| 227 |
+
"international_visa": 59,
|
| 228 |
+
"jump_start": 17,
|
| 229 |
+
"last_maintenance": 37,
|
| 230 |
+
"lost_luggage": 113,
|
| 231 |
+
"make_call": 25,
|
| 232 |
+
"maybe": 121,
|
| 233 |
+
"meal_suggestion": 46,
|
| 234 |
+
"meaning_of_life": 22,
|
| 235 |
+
"measurement_conversion": 31,
|
| 236 |
+
"meeting_schedule": 19,
|
| 237 |
+
"min_payment": 143,
|
| 238 |
+
"mpg": 142,
|
| 239 |
+
"new_card": 91,
|
| 240 |
+
"next_holiday": 140,
|
| 241 |
+
"next_song": 118,
|
| 242 |
+
"no": 30,
|
| 243 |
+
"nutrition_info": 1,
|
| 244 |
+
"oil_change_how": 3,
|
| 245 |
+
"oil_change_when": 123,
|
| 246 |
+
"oos": 42,
|
| 247 |
+
"order": 16,
|
| 248 |
+
"order_checks": 127,
|
| 249 |
+
"order_status": 84,
|
| 250 |
+
"pay_bill": 111,
|
| 251 |
+
"payday": 79,
|
| 252 |
+
"pin_change": 89,
|
| 253 |
+
"play_music": 106,
|
| 254 |
+
"plug_type": 120,
|
| 255 |
+
"pto_balance": 126,
|
| 256 |
+
"pto_request": 97,
|
| 257 |
+
"pto_request_status": 101,
|
| 258 |
+
"pto_used": 148,
|
| 259 |
+
"recipe": 104,
|
| 260 |
+
"redeem_rewards": 6,
|
| 261 |
+
"reminder": 76,
|
| 262 |
+
"reminder_update": 53,
|
| 263 |
+
"repeat": 13,
|
| 264 |
+
"replacement_card_duration": 90,
|
| 265 |
+
"report_fraud": 70,
|
| 266 |
+
"report_lost_card": 12,
|
| 267 |
+
"reset_settings": 88,
|
| 268 |
+
"restaurant_reservation": 23,
|
| 269 |
+
"restaurant_reviews": 0,
|
| 270 |
+
"restaurant_suggestion": 145,
|
| 271 |
+
"rewards_balance": 50,
|
| 272 |
+
"roll_dice": 92,
|
| 273 |
+
"rollover_401k": 100,
|
| 274 |
+
"routing": 45,
|
| 275 |
+
"schedule_maintenance": 131,
|
| 276 |
+
"schedule_meeting": 18,
|
| 277 |
+
"share_location": 58,
|
| 278 |
+
"shopping_list": 43,
|
| 279 |
+
"shopping_list_update": 125,
|
| 280 |
+
"smart_home": 10,
|
| 281 |
+
"spelling": 73,
|
| 282 |
+
"spending_history": 71,
|
| 283 |
+
"sync_device": 130,
|
| 284 |
+
"taxes": 94,
|
| 285 |
+
"tell_joke": 36,
|
| 286 |
+
"text": 26,
|
| 287 |
+
"thank_you": 124,
|
| 288 |
+
"time": 4,
|
| 289 |
+
"timer": 32,
|
| 290 |
+
"timezone": 66,
|
| 291 |
+
"tire_change": 47,
|
| 292 |
+
"tire_pressure": 98,
|
| 293 |
+
"todo_list": 48,
|
| 294 |
+
"todo_list_update": 65,
|
| 295 |
+
"traffic": 24,
|
| 296 |
+
"transactions": 68,
|
| 297 |
+
"transfer": 133,
|
| 298 |
+
"translate": 61,
|
| 299 |
+
"travel_alert": 128,
|
| 300 |
+
"travel_notification": 146,
|
| 301 |
+
"travel_suggestion": 149,
|
| 302 |
+
"uber": 39,
|
| 303 |
+
"update_playlist": 141,
|
| 304 |
+
"user_name": 11,
|
| 305 |
+
"vaccines": 52,
|
| 306 |
+
"w2": 122,
|
| 307 |
+
"weather": 5,
|
| 308 |
+
"what_are_your_hobbies": 15,
|
| 309 |
+
"what_can_i_ask_you": 115,
|
| 310 |
+
"what_is_your_name": 75,
|
| 311 |
+
"what_song": 21,
|
| 312 |
+
"where_are_you_from": 77,
|
| 313 |
+
"whisper_mode": 14,
|
| 314 |
+
"who_do_you_work_for": 57,
|
| 315 |
+
"who_made_you": 96,
|
| 316 |
+
"yes": 107
|
| 317 |
+
},
|
| 318 |
+
"max_position_embeddings": 512,
|
| 319 |
+
"model_type": "distilbert",
|
| 320 |
+
"n_heads": 12,
|
| 321 |
+
"n_layers": 6,
|
| 322 |
+
"pad_token_id": 0,
|
| 323 |
+
"problem_type": "single_label_classification",
|
| 324 |
+
"qa_dropout": 0.1,
|
| 325 |
+
"seq_classif_dropout": 0.2,
|
| 326 |
+
"sinusoidal_pos_embds": false,
|
| 327 |
+
"tie_weights_": true,
|
| 328 |
+
"torch_dtype": "float32",
|
| 329 |
+
"transformers_version": "4.45.2",
|
| 330 |
+
"vocab_size": 30522
|
| 331 |
+
}
|
run-1/checkpoint-3180/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d8608f317b79583c57a72cf842cf0d0cb5b1d2b40dfe81b50b53c6652404325c
|
| 3 |
+
size 268290900
|
run-1/checkpoint-3180/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3e856598db54a66dd6f2b35e5e02a0859c10727a86202fb4a3aa04d1d2faa2be
|
| 3 |
+
size 536643898
|
run-1/checkpoint-3180/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:acb5062e318cbe7ae055df28b0df8c2fb1c196d1f8ef47a1f6f1de04dd8779ef
|
| 3 |
+
size 14244
|
run-1/checkpoint-3180/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:33fadc11bb5c6bd0a5ea603f5e48cedcd72384fa2714656ecf6f8da629f7ae05
|
| 3 |
+
size 1064
|
run-1/checkpoint-3180/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 |
+
}
|
run-1/checkpoint-3180/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
run-1/checkpoint-3180/tokenizer_config.json
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "[PAD]",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"100": {
|
| 12 |
+
"content": "[UNK]",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"101": {
|
| 20 |
+
"content": "[CLS]",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"102": {
|
| 28 |
+
"content": "[SEP]",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"103": {
|
| 36 |
+
"content": "[MASK]",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"clean_up_tokenization_spaces": false,
|
| 45 |
+
"cls_token": "[CLS]",
|
| 46 |
+
"do_lower_case": true,
|
| 47 |
+
"mask_token": "[MASK]",
|
| 48 |
+
"model_max_length": 512,
|
| 49 |
+
"pad_token": "[PAD]",
|
| 50 |
+
"sep_token": "[SEP]",
|
| 51 |
+
"strip_accents": null,
|
| 52 |
+
"tokenize_chinese_chars": true,
|
| 53 |
+
"tokenizer_class": "DistilBertTokenizer",
|
| 54 |
+
"unk_token": "[UNK]"
|
| 55 |
+
}
|
run-1/checkpoint-3180/trainer_state.json
ADDED
|
@@ -0,0 +1,188 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 3180,
|
| 7 |
+
"is_hyper_param_search": true,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.9968553459119497,
|
| 13 |
+
"grad_norm": 0.731029212474823,
|
| 14 |
+
"learning_rate": 1.8006289308176103e-05,
|
| 15 |
+
"loss": 0.408,
|
| 16 |
+
"step": 317
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 1.0,
|
| 20 |
+
"eval_accuracy": 0.5906451612903226,
|
| 21 |
+
"eval_loss": 0.20307187736034393,
|
| 22 |
+
"eval_runtime": 1.6907,
|
| 23 |
+
"eval_samples_per_second": 1833.602,
|
| 24 |
+
"eval_steps_per_second": 38.446,
|
| 25 |
+
"step": 318
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 1.9937106918238994,
|
| 29 |
+
"grad_norm": 0.4595111310482025,
|
| 30 |
+
"learning_rate": 1.6012578616352204e-05,
|
| 31 |
+
"loss": 0.1697,
|
| 32 |
+
"step": 634
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 2.0,
|
| 36 |
+
"eval_accuracy": 0.8306451612903226,
|
| 37 |
+
"eval_loss": 0.09759204834699631,
|
| 38 |
+
"eval_runtime": 1.7036,
|
| 39 |
+
"eval_samples_per_second": 1819.686,
|
| 40 |
+
"eval_steps_per_second": 38.155,
|
| 41 |
+
"step": 636
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 2.990566037735849,
|
| 45 |
+
"grad_norm": 0.43162059783935547,
|
| 46 |
+
"learning_rate": 1.4018867924528304e-05,
|
| 47 |
+
"loss": 0.104,
|
| 48 |
+
"step": 951
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 3.0,
|
| 52 |
+
"eval_accuracy": 0.88,
|
| 53 |
+
"eval_loss": 0.06371938437223434,
|
| 54 |
+
"eval_runtime": 1.7221,
|
| 55 |
+
"eval_samples_per_second": 1800.179,
|
| 56 |
+
"eval_steps_per_second": 37.746,
|
| 57 |
+
"step": 954
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 3.9874213836477987,
|
| 61 |
+
"grad_norm": 0.3561622202396393,
|
| 62 |
+
"learning_rate": 1.2025157232704403e-05,
|
| 63 |
+
"loss": 0.0773,
|
| 64 |
+
"step": 1268
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 4.0,
|
| 68 |
+
"eval_accuracy": 0.8977419354838709,
|
| 69 |
+
"eval_loss": 0.04856124520301819,
|
| 70 |
+
"eval_runtime": 1.7777,
|
| 71 |
+
"eval_samples_per_second": 1743.857,
|
| 72 |
+
"eval_steps_per_second": 36.565,
|
| 73 |
+
"step": 1272
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 4.984276729559748,
|
| 77 |
+
"grad_norm": 0.2814004123210907,
|
| 78 |
+
"learning_rate": 1.0031446540880504e-05,
|
| 79 |
+
"loss": 0.0631,
|
| 80 |
+
"step": 1585
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 5.0,
|
| 84 |
+
"eval_accuracy": 0.9067741935483871,
|
| 85 |
+
"eval_loss": 0.0395587794482708,
|
| 86 |
+
"eval_runtime": 1.7657,
|
| 87 |
+
"eval_samples_per_second": 1755.721,
|
| 88 |
+
"eval_steps_per_second": 36.814,
|
| 89 |
+
"step": 1590
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 5.981132075471698,
|
| 93 |
+
"grad_norm": 0.39958587288856506,
|
| 94 |
+
"learning_rate": 8.037735849056606e-06,
|
| 95 |
+
"loss": 0.0544,
|
| 96 |
+
"step": 1902
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 6.0,
|
| 100 |
+
"eval_accuracy": 0.9183870967741935,
|
| 101 |
+
"eval_loss": 0.03417970985174179,
|
| 102 |
+
"eval_runtime": 1.8,
|
| 103 |
+
"eval_samples_per_second": 1722.246,
|
| 104 |
+
"eval_steps_per_second": 36.112,
|
| 105 |
+
"step": 1908
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 6.977987421383648,
|
| 109 |
+
"grad_norm": 0.27226564288139343,
|
| 110 |
+
"learning_rate": 6.044025157232704e-06,
|
| 111 |
+
"loss": 0.0493,
|
| 112 |
+
"step": 2219
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 7.0,
|
| 116 |
+
"eval_accuracy": 0.9248387096774193,
|
| 117 |
+
"eval_loss": 0.031123189255595207,
|
| 118 |
+
"eval_runtime": 1.7804,
|
| 119 |
+
"eval_samples_per_second": 1741.143,
|
| 120 |
+
"eval_steps_per_second": 36.508,
|
| 121 |
+
"step": 2226
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 7.9748427672955975,
|
| 125 |
+
"grad_norm": 0.2539806067943573,
|
| 126 |
+
"learning_rate": 4.0503144654088055e-06,
|
| 127 |
+
"loss": 0.0459,
|
| 128 |
+
"step": 2536
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 8.0,
|
| 132 |
+
"eval_accuracy": 0.9258064516129032,
|
| 133 |
+
"eval_loss": 0.029144473373889923,
|
| 134 |
+
"eval_runtime": 1.8058,
|
| 135 |
+
"eval_samples_per_second": 1716.732,
|
| 136 |
+
"eval_steps_per_second": 35.996,
|
| 137 |
+
"step": 2544
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 8.971698113207546,
|
| 141 |
+
"grad_norm": 0.24147309362888336,
|
| 142 |
+
"learning_rate": 2.056603773584906e-06,
|
| 143 |
+
"loss": 0.0438,
|
| 144 |
+
"step": 2853
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 9.0,
|
| 148 |
+
"eval_accuracy": 0.9270967741935484,
|
| 149 |
+
"eval_loss": 0.028016844764351845,
|
| 150 |
+
"eval_runtime": 1.7889,
|
| 151 |
+
"eval_samples_per_second": 1732.923,
|
| 152 |
+
"eval_steps_per_second": 36.335,
|
| 153 |
+
"step": 2862
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 9.968553459119496,
|
| 157 |
+
"grad_norm": 0.20841307938098907,
|
| 158 |
+
"learning_rate": 6.289308176100629e-08,
|
| 159 |
+
"loss": 0.0426,
|
| 160 |
+
"step": 3170
|
| 161 |
+
}
|
| 162 |
+
],
|
| 163 |
+
"logging_steps": 317,
|
| 164 |
+
"max_steps": 3180,
|
| 165 |
+
"num_input_tokens_seen": 0,
|
| 166 |
+
"num_train_epochs": 10,
|
| 167 |
+
"save_steps": 1000000000.0,
|
| 168 |
+
"stateful_callbacks": {
|
| 169 |
+
"TrainerControl": {
|
| 170 |
+
"args": {
|
| 171 |
+
"should_epoch_stop": false,
|
| 172 |
+
"should_evaluate": false,
|
| 173 |
+
"should_log": false,
|
| 174 |
+
"should_save": true,
|
| 175 |
+
"should_training_stop": true
|
| 176 |
+
},
|
| 177 |
+
"attributes": {}
|
| 178 |
+
}
|
| 179 |
+
},
|
| 180 |
+
"total_flos": 825254092458012.0,
|
| 181 |
+
"train_batch_size": 48,
|
| 182 |
+
"trial_name": null,
|
| 183 |
+
"trial_params": {
|
| 184 |
+
"alpha": 0.47967279863257795,
|
| 185 |
+
"num_train_epochs": 10,
|
| 186 |
+
"temperature": 9
|
| 187 |
+
}
|
| 188 |
+
}
|
run-1/checkpoint-3180/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a25d63abf0cdd55f7279d86dd585a8ef38e8e89522e1ffd1f4a381861fcedd19
|
| 3 |
+
size 5304
|
run-1/checkpoint-3180/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
run-2/checkpoint-1590/config.json
ADDED
|
@@ -0,0 +1,331 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "distilbert-base-uncased",
|
| 3 |
+
"activation": "gelu",
|
| 4 |
+
"architectures": [
|
| 5 |
+
"DistilBertForSequenceClassification"
|
| 6 |
+
],
|
| 7 |
+
"attention_dropout": 0.1,
|
| 8 |
+
"dim": 768,
|
| 9 |
+
"dropout": 0.1,
|
| 10 |
+
"hidden_dim": 3072,
|
| 11 |
+
"id2label": {
|
| 12 |
+
"0": "restaurant_reviews",
|
| 13 |
+
"1": "nutrition_info",
|
| 14 |
+
"2": "account_blocked",
|
| 15 |
+
"3": "oil_change_how",
|
| 16 |
+
"4": "time",
|
| 17 |
+
"5": "weather",
|
| 18 |
+
"6": "redeem_rewards",
|
| 19 |
+
"7": "interest_rate",
|
| 20 |
+
"8": "gas_type",
|
| 21 |
+
"9": "accept_reservations",
|
| 22 |
+
"10": "smart_home",
|
| 23 |
+
"11": "user_name",
|
| 24 |
+
"12": "report_lost_card",
|
| 25 |
+
"13": "repeat",
|
| 26 |
+
"14": "whisper_mode",
|
| 27 |
+
"15": "what_are_your_hobbies",
|
| 28 |
+
"16": "order",
|
| 29 |
+
"17": "jump_start",
|
| 30 |
+
"18": "schedule_meeting",
|
| 31 |
+
"19": "meeting_schedule",
|
| 32 |
+
"20": "freeze_account",
|
| 33 |
+
"21": "what_song",
|
| 34 |
+
"22": "meaning_of_life",
|
| 35 |
+
"23": "restaurant_reservation",
|
| 36 |
+
"24": "traffic",
|
| 37 |
+
"25": "make_call",
|
| 38 |
+
"26": "text",
|
| 39 |
+
"27": "bill_balance",
|
| 40 |
+
"28": "improve_credit_score",
|
| 41 |
+
"29": "change_language",
|
| 42 |
+
"30": "no",
|
| 43 |
+
"31": "measurement_conversion",
|
| 44 |
+
"32": "timer",
|
| 45 |
+
"33": "flip_coin",
|
| 46 |
+
"34": "do_you_have_pets",
|
| 47 |
+
"35": "balance",
|
| 48 |
+
"36": "tell_joke",
|
| 49 |
+
"37": "last_maintenance",
|
| 50 |
+
"38": "exchange_rate",
|
| 51 |
+
"39": "uber",
|
| 52 |
+
"40": "car_rental",
|
| 53 |
+
"41": "credit_limit",
|
| 54 |
+
"42": "oos",
|
| 55 |
+
"43": "shopping_list",
|
| 56 |
+
"44": "expiration_date",
|
| 57 |
+
"45": "routing",
|
| 58 |
+
"46": "meal_suggestion",
|
| 59 |
+
"47": "tire_change",
|
| 60 |
+
"48": "todo_list",
|
| 61 |
+
"49": "card_declined",
|
| 62 |
+
"50": "rewards_balance",
|
| 63 |
+
"51": "change_accent",
|
| 64 |
+
"52": "vaccines",
|
| 65 |
+
"53": "reminder_update",
|
| 66 |
+
"54": "food_last",
|
| 67 |
+
"55": "change_ai_name",
|
| 68 |
+
"56": "bill_due",
|
| 69 |
+
"57": "who_do_you_work_for",
|
| 70 |
+
"58": "share_location",
|
| 71 |
+
"59": "international_visa",
|
| 72 |
+
"60": "calendar",
|
| 73 |
+
"61": "translate",
|
| 74 |
+
"62": "carry_on",
|
| 75 |
+
"63": "book_flight",
|
| 76 |
+
"64": "insurance_change",
|
| 77 |
+
"65": "todo_list_update",
|
| 78 |
+
"66": "timezone",
|
| 79 |
+
"67": "cancel_reservation",
|
| 80 |
+
"68": "transactions",
|
| 81 |
+
"69": "credit_score",
|
| 82 |
+
"70": "report_fraud",
|
| 83 |
+
"71": "spending_history",
|
| 84 |
+
"72": "directions",
|
| 85 |
+
"73": "spelling",
|
| 86 |
+
"74": "insurance",
|
| 87 |
+
"75": "what_is_your_name",
|
| 88 |
+
"76": "reminder",
|
| 89 |
+
"77": "where_are_you_from",
|
| 90 |
+
"78": "distance",
|
| 91 |
+
"79": "payday",
|
| 92 |
+
"80": "flight_status",
|
| 93 |
+
"81": "find_phone",
|
| 94 |
+
"82": "greeting",
|
| 95 |
+
"83": "alarm",
|
| 96 |
+
"84": "order_status",
|
| 97 |
+
"85": "confirm_reservation",
|
| 98 |
+
"86": "cook_time",
|
| 99 |
+
"87": "damaged_card",
|
| 100 |
+
"88": "reset_settings",
|
| 101 |
+
"89": "pin_change",
|
| 102 |
+
"90": "replacement_card_duration",
|
| 103 |
+
"91": "new_card",
|
| 104 |
+
"92": "roll_dice",
|
| 105 |
+
"93": "income",
|
| 106 |
+
"94": "taxes",
|
| 107 |
+
"95": "date",
|
| 108 |
+
"96": "who_made_you",
|
| 109 |
+
"97": "pto_request",
|
| 110 |
+
"98": "tire_pressure",
|
| 111 |
+
"99": "how_old_are_you",
|
| 112 |
+
"100": "rollover_401k",
|
| 113 |
+
"101": "pto_request_status",
|
| 114 |
+
"102": "how_busy",
|
| 115 |
+
"103": "application_status",
|
| 116 |
+
"104": "recipe",
|
| 117 |
+
"105": "calendar_update",
|
| 118 |
+
"106": "play_music",
|
| 119 |
+
"107": "yes",
|
| 120 |
+
"108": "direct_deposit",
|
| 121 |
+
"109": "credit_limit_change",
|
| 122 |
+
"110": "gas",
|
| 123 |
+
"111": "pay_bill",
|
| 124 |
+
"112": "ingredients_list",
|
| 125 |
+
"113": "lost_luggage",
|
| 126 |
+
"114": "goodbye",
|
| 127 |
+
"115": "what_can_i_ask_you",
|
| 128 |
+
"116": "book_hotel",
|
| 129 |
+
"117": "are_you_a_bot",
|
| 130 |
+
"118": "next_song",
|
| 131 |
+
"119": "change_speed",
|
| 132 |
+
"120": "plug_type",
|
| 133 |
+
"121": "maybe",
|
| 134 |
+
"122": "w2",
|
| 135 |
+
"123": "oil_change_when",
|
| 136 |
+
"124": "thank_you",
|
| 137 |
+
"125": "shopping_list_update",
|
| 138 |
+
"126": "pto_balance",
|
| 139 |
+
"127": "order_checks",
|
| 140 |
+
"128": "travel_alert",
|
| 141 |
+
"129": "fun_fact",
|
| 142 |
+
"130": "sync_device",
|
| 143 |
+
"131": "schedule_maintenance",
|
| 144 |
+
"132": "apr",
|
| 145 |
+
"133": "transfer",
|
| 146 |
+
"134": "ingredient_substitution",
|
| 147 |
+
"135": "calories",
|
| 148 |
+
"136": "current_location",
|
| 149 |
+
"137": "international_fees",
|
| 150 |
+
"138": "calculator",
|
| 151 |
+
"139": "definition",
|
| 152 |
+
"140": "next_holiday",
|
| 153 |
+
"141": "update_playlist",
|
| 154 |
+
"142": "mpg",
|
| 155 |
+
"143": "min_payment",
|
| 156 |
+
"144": "change_user_name",
|
| 157 |
+
"145": "restaurant_suggestion",
|
| 158 |
+
"146": "travel_notification",
|
| 159 |
+
"147": "cancel",
|
| 160 |
+
"148": "pto_used",
|
| 161 |
+
"149": "travel_suggestion",
|
| 162 |
+
"150": "change_volume"
|
| 163 |
+
},
|
| 164 |
+
"initializer_range": 0.02,
|
| 165 |
+
"label2id": {
|
| 166 |
+
"accept_reservations": 9,
|
| 167 |
+
"account_blocked": 2,
|
| 168 |
+
"alarm": 83,
|
| 169 |
+
"application_status": 103,
|
| 170 |
+
"apr": 132,
|
| 171 |
+
"are_you_a_bot": 117,
|
| 172 |
+
"balance": 35,
|
| 173 |
+
"bill_balance": 27,
|
| 174 |
+
"bill_due": 56,
|
| 175 |
+
"book_flight": 63,
|
| 176 |
+
"book_hotel": 116,
|
| 177 |
+
"calculator": 138,
|
| 178 |
+
"calendar": 60,
|
| 179 |
+
"calendar_update": 105,
|
| 180 |
+
"calories": 135,
|
| 181 |
+
"cancel": 147,
|
| 182 |
+
"cancel_reservation": 67,
|
| 183 |
+
"car_rental": 40,
|
| 184 |
+
"card_declined": 49,
|
| 185 |
+
"carry_on": 62,
|
| 186 |
+
"change_accent": 51,
|
| 187 |
+
"change_ai_name": 55,
|
| 188 |
+
"change_language": 29,
|
| 189 |
+
"change_speed": 119,
|
| 190 |
+
"change_user_name": 144,
|
| 191 |
+
"change_volume": 150,
|
| 192 |
+
"confirm_reservation": 85,
|
| 193 |
+
"cook_time": 86,
|
| 194 |
+
"credit_limit": 41,
|
| 195 |
+
"credit_limit_change": 109,
|
| 196 |
+
"credit_score": 69,
|
| 197 |
+
"current_location": 136,
|
| 198 |
+
"damaged_card": 87,
|
| 199 |
+
"date": 95,
|
| 200 |
+
"definition": 139,
|
| 201 |
+
"direct_deposit": 108,
|
| 202 |
+
"directions": 72,
|
| 203 |
+
"distance": 78,
|
| 204 |
+
"do_you_have_pets": 34,
|
| 205 |
+
"exchange_rate": 38,
|
| 206 |
+
"expiration_date": 44,
|
| 207 |
+
"find_phone": 81,
|
| 208 |
+
"flight_status": 80,
|
| 209 |
+
"flip_coin": 33,
|
| 210 |
+
"food_last": 54,
|
| 211 |
+
"freeze_account": 20,
|
| 212 |
+
"fun_fact": 129,
|
| 213 |
+
"gas": 110,
|
| 214 |
+
"gas_type": 8,
|
| 215 |
+
"goodbye": 114,
|
| 216 |
+
"greeting": 82,
|
| 217 |
+
"how_busy": 102,
|
| 218 |
+
"how_old_are_you": 99,
|
| 219 |
+
"improve_credit_score": 28,
|
| 220 |
+
"income": 93,
|
| 221 |
+
"ingredient_substitution": 134,
|
| 222 |
+
"ingredients_list": 112,
|
| 223 |
+
"insurance": 74,
|
| 224 |
+
"insurance_change": 64,
|
| 225 |
+
"interest_rate": 7,
|
| 226 |
+
"international_fees": 137,
|
| 227 |
+
"international_visa": 59,
|
| 228 |
+
"jump_start": 17,
|
| 229 |
+
"last_maintenance": 37,
|
| 230 |
+
"lost_luggage": 113,
|
| 231 |
+
"make_call": 25,
|
| 232 |
+
"maybe": 121,
|
| 233 |
+
"meal_suggestion": 46,
|
| 234 |
+
"meaning_of_life": 22,
|
| 235 |
+
"measurement_conversion": 31,
|
| 236 |
+
"meeting_schedule": 19,
|
| 237 |
+
"min_payment": 143,
|
| 238 |
+
"mpg": 142,
|
| 239 |
+
"new_card": 91,
|
| 240 |
+
"next_holiday": 140,
|
| 241 |
+
"next_song": 118,
|
| 242 |
+
"no": 30,
|
| 243 |
+
"nutrition_info": 1,
|
| 244 |
+
"oil_change_how": 3,
|
| 245 |
+
"oil_change_when": 123,
|
| 246 |
+
"oos": 42,
|
| 247 |
+
"order": 16,
|
| 248 |
+
"order_checks": 127,
|
| 249 |
+
"order_status": 84,
|
| 250 |
+
"pay_bill": 111,
|
| 251 |
+
"payday": 79,
|
| 252 |
+
"pin_change": 89,
|
| 253 |
+
"play_music": 106,
|
| 254 |
+
"plug_type": 120,
|
| 255 |
+
"pto_balance": 126,
|
| 256 |
+
"pto_request": 97,
|
| 257 |
+
"pto_request_status": 101,
|
| 258 |
+
"pto_used": 148,
|
| 259 |
+
"recipe": 104,
|
| 260 |
+
"redeem_rewards": 6,
|
| 261 |
+
"reminder": 76,
|
| 262 |
+
"reminder_update": 53,
|
| 263 |
+
"repeat": 13,
|
| 264 |
+
"replacement_card_duration": 90,
|
| 265 |
+
"report_fraud": 70,
|
| 266 |
+
"report_lost_card": 12,
|
| 267 |
+
"reset_settings": 88,
|
| 268 |
+
"restaurant_reservation": 23,
|
| 269 |
+
"restaurant_reviews": 0,
|
| 270 |
+
"restaurant_suggestion": 145,
|
| 271 |
+
"rewards_balance": 50,
|
| 272 |
+
"roll_dice": 92,
|
| 273 |
+
"rollover_401k": 100,
|
| 274 |
+
"routing": 45,
|
| 275 |
+
"schedule_maintenance": 131,
|
| 276 |
+
"schedule_meeting": 18,
|
| 277 |
+
"share_location": 58,
|
| 278 |
+
"shopping_list": 43,
|
| 279 |
+
"shopping_list_update": 125,
|
| 280 |
+
"smart_home": 10,
|
| 281 |
+
"spelling": 73,
|
| 282 |
+
"spending_history": 71,
|
| 283 |
+
"sync_device": 130,
|
| 284 |
+
"taxes": 94,
|
| 285 |
+
"tell_joke": 36,
|
| 286 |
+
"text": 26,
|
| 287 |
+
"thank_you": 124,
|
| 288 |
+
"time": 4,
|
| 289 |
+
"timer": 32,
|
| 290 |
+
"timezone": 66,
|
| 291 |
+
"tire_change": 47,
|
| 292 |
+
"tire_pressure": 98,
|
| 293 |
+
"todo_list": 48,
|
| 294 |
+
"todo_list_update": 65,
|
| 295 |
+
"traffic": 24,
|
| 296 |
+
"transactions": 68,
|
| 297 |
+
"transfer": 133,
|
| 298 |
+
"translate": 61,
|
| 299 |
+
"travel_alert": 128,
|
| 300 |
+
"travel_notification": 146,
|
| 301 |
+
"travel_suggestion": 149,
|
| 302 |
+
"uber": 39,
|
| 303 |
+
"update_playlist": 141,
|
| 304 |
+
"user_name": 11,
|
| 305 |
+
"vaccines": 52,
|
| 306 |
+
"w2": 122,
|
| 307 |
+
"weather": 5,
|
| 308 |
+
"what_are_your_hobbies": 15,
|
| 309 |
+
"what_can_i_ask_you": 115,
|
| 310 |
+
"what_is_your_name": 75,
|
| 311 |
+
"what_song": 21,
|
| 312 |
+
"where_are_you_from": 77,
|
| 313 |
+
"whisper_mode": 14,
|
| 314 |
+
"who_do_you_work_for": 57,
|
| 315 |
+
"who_made_you": 96,
|
| 316 |
+
"yes": 107
|
| 317 |
+
},
|
| 318 |
+
"max_position_embeddings": 512,
|
| 319 |
+
"model_type": "distilbert",
|
| 320 |
+
"n_heads": 12,
|
| 321 |
+
"n_layers": 6,
|
| 322 |
+
"pad_token_id": 0,
|
| 323 |
+
"problem_type": "single_label_classification",
|
| 324 |
+
"qa_dropout": 0.1,
|
| 325 |
+
"seq_classif_dropout": 0.2,
|
| 326 |
+
"sinusoidal_pos_embds": false,
|
| 327 |
+
"tie_weights_": true,
|
| 328 |
+
"torch_dtype": "float32",
|
| 329 |
+
"transformers_version": "4.45.2",
|
| 330 |
+
"vocab_size": 30522
|
| 331 |
+
}
|
run-2/checkpoint-1590/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9faa53c0f7c95831aa4246fd1488c17cb15fa62ccc2048546bed34de2d209359
|
| 3 |
+
size 268290900
|
run-2/checkpoint-1590/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0fad6337d654493f73f3b91c3242c5c8cdc675d99206f7a6c346070b4393677f
|
| 3 |
+
size 536643898
|
run-2/checkpoint-1590/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a66c0cd02823b9fa3c17cb8050d4f36f19ab8409fe337d5d545a289b3ca30007
|
| 3 |
+
size 14244
|
run-2/checkpoint-1590/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ee1082c0efb09baac545ba26829a8ad0ace46851b360d7c70f5f11088ecb04f2
|
| 3 |
+
size 1064
|
run-2/checkpoint-1590/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 |
+
}
|
run-2/checkpoint-1590/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
run-2/checkpoint-1590/tokenizer_config.json
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "[PAD]",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"100": {
|
| 12 |
+
"content": "[UNK]",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"101": {
|
| 20 |
+
"content": "[CLS]",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"102": {
|
| 28 |
+
"content": "[SEP]",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"103": {
|
| 36 |
+
"content": "[MASK]",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"clean_up_tokenization_spaces": false,
|
| 45 |
+
"cls_token": "[CLS]",
|
| 46 |
+
"do_lower_case": true,
|
| 47 |
+
"mask_token": "[MASK]",
|
| 48 |
+
"model_max_length": 512,
|
| 49 |
+
"pad_token": "[PAD]",
|
| 50 |
+
"sep_token": "[SEP]",
|
| 51 |
+
"strip_accents": null,
|
| 52 |
+
"tokenize_chinese_chars": true,
|
| 53 |
+
"tokenizer_class": "DistilBertTokenizer",
|
| 54 |
+
"unk_token": "[UNK]"
|
| 55 |
+
}
|
run-2/checkpoint-1590/trainer_state.json
ADDED
|
@@ -0,0 +1,108 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1590,
|
| 7 |
+
"is_hyper_param_search": true,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.9968553459119497,
|
| 13 |
+
"grad_norm": 0.6950033903121948,
|
| 14 |
+
"learning_rate": 1.6012578616352204e-05,
|
| 15 |
+
"loss": 0.3925,
|
| 16 |
+
"step": 317
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 1.0,
|
| 20 |
+
"eval_accuracy": 0.5609677419354838,
|
| 21 |
+
"eval_loss": 0.20299369096755981,
|
| 22 |
+
"eval_runtime": 1.7311,
|
| 23 |
+
"eval_samples_per_second": 1790.751,
|
| 24 |
+
"eval_steps_per_second": 37.548,
|
| 25 |
+
"step": 318
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 1.9937106918238994,
|
| 29 |
+
"grad_norm": 0.4946860373020172,
|
| 30 |
+
"learning_rate": 1.2025157232704403e-05,
|
| 31 |
+
"loss": 0.1736,
|
| 32 |
+
"step": 634
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 2.0,
|
| 36 |
+
"eval_accuracy": 0.7987096774193548,
|
| 37 |
+
"eval_loss": 0.10688149929046631,
|
| 38 |
+
"eval_runtime": 1.717,
|
| 39 |
+
"eval_samples_per_second": 1805.465,
|
| 40 |
+
"eval_steps_per_second": 37.857,
|
| 41 |
+
"step": 636
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 2.990566037735849,
|
| 45 |
+
"grad_norm": 0.4743373990058899,
|
| 46 |
+
"learning_rate": 8.037735849056606e-06,
|
| 47 |
+
"loss": 0.1151,
|
| 48 |
+
"step": 951
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 3.0,
|
| 52 |
+
"eval_accuracy": 0.8580645161290322,
|
| 53 |
+
"eval_loss": 0.07752583920955658,
|
| 54 |
+
"eval_runtime": 1.7617,
|
| 55 |
+
"eval_samples_per_second": 1759.674,
|
| 56 |
+
"eval_steps_per_second": 36.896,
|
| 57 |
+
"step": 954
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 3.9874213836477987,
|
| 61 |
+
"grad_norm": 0.4146045744419098,
|
| 62 |
+
"learning_rate": 4.0503144654088055e-06,
|
| 63 |
+
"loss": 0.0933,
|
| 64 |
+
"step": 1268
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 4.0,
|
| 68 |
+
"eval_accuracy": 0.8754838709677419,
|
| 69 |
+
"eval_loss": 0.06553368270397186,
|
| 70 |
+
"eval_runtime": 1.7696,
|
| 71 |
+
"eval_samples_per_second": 1751.791,
|
| 72 |
+
"eval_steps_per_second": 36.731,
|
| 73 |
+
"step": 1272
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 4.984276729559748,
|
| 77 |
+
"grad_norm": 0.3400251865386963,
|
| 78 |
+
"learning_rate": 6.289308176100629e-08,
|
| 79 |
+
"loss": 0.0843,
|
| 80 |
+
"step": 1585
|
| 81 |
+
}
|
| 82 |
+
],
|
| 83 |
+
"logging_steps": 317,
|
| 84 |
+
"max_steps": 1590,
|
| 85 |
+
"num_input_tokens_seen": 0,
|
| 86 |
+
"num_train_epochs": 5,
|
| 87 |
+
"save_steps": 1000000000.0,
|
| 88 |
+
"stateful_callbacks": {
|
| 89 |
+
"TrainerControl": {
|
| 90 |
+
"args": {
|
| 91 |
+
"should_epoch_stop": false,
|
| 92 |
+
"should_evaluate": false,
|
| 93 |
+
"should_log": false,
|
| 94 |
+
"should_save": true,
|
| 95 |
+
"should_training_stop": true
|
| 96 |
+
},
|
| 97 |
+
"attributes": {}
|
| 98 |
+
}
|
| 99 |
+
},
|
| 100 |
+
"total_flos": 413495819944032.0,
|
| 101 |
+
"train_batch_size": 48,
|
| 102 |
+
"trial_name": null,
|
| 103 |
+
"trial_params": {
|
| 104 |
+
"alpha": 0.800129291798159,
|
| 105 |
+
"num_train_epochs": 5,
|
| 106 |
+
"temperature": 15
|
| 107 |
+
}
|
| 108 |
+
}
|
run-2/checkpoint-1590/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0e308b8dc4066f1fc7a9f1b91e464e0978b0830cbd2f3213afa3c14ff3bc462a
|
| 3 |
+
size 5304
|
run-2/checkpoint-1590/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
run-3/checkpoint-2862/config.json
ADDED
|
@@ -0,0 +1,331 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "distilbert-base-uncased",
|
| 3 |
+
"activation": "gelu",
|
| 4 |
+
"architectures": [
|
| 5 |
+
"DistilBertForSequenceClassification"
|
| 6 |
+
],
|
| 7 |
+
"attention_dropout": 0.1,
|
| 8 |
+
"dim": 768,
|
| 9 |
+
"dropout": 0.1,
|
| 10 |
+
"hidden_dim": 3072,
|
| 11 |
+
"id2label": {
|
| 12 |
+
"0": "restaurant_reviews",
|
| 13 |
+
"1": "nutrition_info",
|
| 14 |
+
"2": "account_blocked",
|
| 15 |
+
"3": "oil_change_how",
|
| 16 |
+
"4": "time",
|
| 17 |
+
"5": "weather",
|
| 18 |
+
"6": "redeem_rewards",
|
| 19 |
+
"7": "interest_rate",
|
| 20 |
+
"8": "gas_type",
|
| 21 |
+
"9": "accept_reservations",
|
| 22 |
+
"10": "smart_home",
|
| 23 |
+
"11": "user_name",
|
| 24 |
+
"12": "report_lost_card",
|
| 25 |
+
"13": "repeat",
|
| 26 |
+
"14": "whisper_mode",
|
| 27 |
+
"15": "what_are_your_hobbies",
|
| 28 |
+
"16": "order",
|
| 29 |
+
"17": "jump_start",
|
| 30 |
+
"18": "schedule_meeting",
|
| 31 |
+
"19": "meeting_schedule",
|
| 32 |
+
"20": "freeze_account",
|
| 33 |
+
"21": "what_song",
|
| 34 |
+
"22": "meaning_of_life",
|
| 35 |
+
"23": "restaurant_reservation",
|
| 36 |
+
"24": "traffic",
|
| 37 |
+
"25": "make_call",
|
| 38 |
+
"26": "text",
|
| 39 |
+
"27": "bill_balance",
|
| 40 |
+
"28": "improve_credit_score",
|
| 41 |
+
"29": "change_language",
|
| 42 |
+
"30": "no",
|
| 43 |
+
"31": "measurement_conversion",
|
| 44 |
+
"32": "timer",
|
| 45 |
+
"33": "flip_coin",
|
| 46 |
+
"34": "do_you_have_pets",
|
| 47 |
+
"35": "balance",
|
| 48 |
+
"36": "tell_joke",
|
| 49 |
+
"37": "last_maintenance",
|
| 50 |
+
"38": "exchange_rate",
|
| 51 |
+
"39": "uber",
|
| 52 |
+
"40": "car_rental",
|
| 53 |
+
"41": "credit_limit",
|
| 54 |
+
"42": "oos",
|
| 55 |
+
"43": "shopping_list",
|
| 56 |
+
"44": "expiration_date",
|
| 57 |
+
"45": "routing",
|
| 58 |
+
"46": "meal_suggestion",
|
| 59 |
+
"47": "tire_change",
|
| 60 |
+
"48": "todo_list",
|
| 61 |
+
"49": "card_declined",
|
| 62 |
+
"50": "rewards_balance",
|
| 63 |
+
"51": "change_accent",
|
| 64 |
+
"52": "vaccines",
|
| 65 |
+
"53": "reminder_update",
|
| 66 |
+
"54": "food_last",
|
| 67 |
+
"55": "change_ai_name",
|
| 68 |
+
"56": "bill_due",
|
| 69 |
+
"57": "who_do_you_work_for",
|
| 70 |
+
"58": "share_location",
|
| 71 |
+
"59": "international_visa",
|
| 72 |
+
"60": "calendar",
|
| 73 |
+
"61": "translate",
|
| 74 |
+
"62": "carry_on",
|
| 75 |
+
"63": "book_flight",
|
| 76 |
+
"64": "insurance_change",
|
| 77 |
+
"65": "todo_list_update",
|
| 78 |
+
"66": "timezone",
|
| 79 |
+
"67": "cancel_reservation",
|
| 80 |
+
"68": "transactions",
|
| 81 |
+
"69": "credit_score",
|
| 82 |
+
"70": "report_fraud",
|
| 83 |
+
"71": "spending_history",
|
| 84 |
+
"72": "directions",
|
| 85 |
+
"73": "spelling",
|
| 86 |
+
"74": "insurance",
|
| 87 |
+
"75": "what_is_your_name",
|
| 88 |
+
"76": "reminder",
|
| 89 |
+
"77": "where_are_you_from",
|
| 90 |
+
"78": "distance",
|
| 91 |
+
"79": "payday",
|
| 92 |
+
"80": "flight_status",
|
| 93 |
+
"81": "find_phone",
|
| 94 |
+
"82": "greeting",
|
| 95 |
+
"83": "alarm",
|
| 96 |
+
"84": "order_status",
|
| 97 |
+
"85": "confirm_reservation",
|
| 98 |
+
"86": "cook_time",
|
| 99 |
+
"87": "damaged_card",
|
| 100 |
+
"88": "reset_settings",
|
| 101 |
+
"89": "pin_change",
|
| 102 |
+
"90": "replacement_card_duration",
|
| 103 |
+
"91": "new_card",
|
| 104 |
+
"92": "roll_dice",
|
| 105 |
+
"93": "income",
|
| 106 |
+
"94": "taxes",
|
| 107 |
+
"95": "date",
|
| 108 |
+
"96": "who_made_you",
|
| 109 |
+
"97": "pto_request",
|
| 110 |
+
"98": "tire_pressure",
|
| 111 |
+
"99": "how_old_are_you",
|
| 112 |
+
"100": "rollover_401k",
|
| 113 |
+
"101": "pto_request_status",
|
| 114 |
+
"102": "how_busy",
|
| 115 |
+
"103": "application_status",
|
| 116 |
+
"104": "recipe",
|
| 117 |
+
"105": "calendar_update",
|
| 118 |
+
"106": "play_music",
|
| 119 |
+
"107": "yes",
|
| 120 |
+
"108": "direct_deposit",
|
| 121 |
+
"109": "credit_limit_change",
|
| 122 |
+
"110": "gas",
|
| 123 |
+
"111": "pay_bill",
|
| 124 |
+
"112": "ingredients_list",
|
| 125 |
+
"113": "lost_luggage",
|
| 126 |
+
"114": "goodbye",
|
| 127 |
+
"115": "what_can_i_ask_you",
|
| 128 |
+
"116": "book_hotel",
|
| 129 |
+
"117": "are_you_a_bot",
|
| 130 |
+
"118": "next_song",
|
| 131 |
+
"119": "change_speed",
|
| 132 |
+
"120": "plug_type",
|
| 133 |
+
"121": "maybe",
|
| 134 |
+
"122": "w2",
|
| 135 |
+
"123": "oil_change_when",
|
| 136 |
+
"124": "thank_you",
|
| 137 |
+
"125": "shopping_list_update",
|
| 138 |
+
"126": "pto_balance",
|
| 139 |
+
"127": "order_checks",
|
| 140 |
+
"128": "travel_alert",
|
| 141 |
+
"129": "fun_fact",
|
| 142 |
+
"130": "sync_device",
|
| 143 |
+
"131": "schedule_maintenance",
|
| 144 |
+
"132": "apr",
|
| 145 |
+
"133": "transfer",
|
| 146 |
+
"134": "ingredient_substitution",
|
| 147 |
+
"135": "calories",
|
| 148 |
+
"136": "current_location",
|
| 149 |
+
"137": "international_fees",
|
| 150 |
+
"138": "calculator",
|
| 151 |
+
"139": "definition",
|
| 152 |
+
"140": "next_holiday",
|
| 153 |
+
"141": "update_playlist",
|
| 154 |
+
"142": "mpg",
|
| 155 |
+
"143": "min_payment",
|
| 156 |
+
"144": "change_user_name",
|
| 157 |
+
"145": "restaurant_suggestion",
|
| 158 |
+
"146": "travel_notification",
|
| 159 |
+
"147": "cancel",
|
| 160 |
+
"148": "pto_used",
|
| 161 |
+
"149": "travel_suggestion",
|
| 162 |
+
"150": "change_volume"
|
| 163 |
+
},
|
| 164 |
+
"initializer_range": 0.02,
|
| 165 |
+
"label2id": {
|
| 166 |
+
"accept_reservations": 9,
|
| 167 |
+
"account_blocked": 2,
|
| 168 |
+
"alarm": 83,
|
| 169 |
+
"application_status": 103,
|
| 170 |
+
"apr": 132,
|
| 171 |
+
"are_you_a_bot": 117,
|
| 172 |
+
"balance": 35,
|
| 173 |
+
"bill_balance": 27,
|
| 174 |
+
"bill_due": 56,
|
| 175 |
+
"book_flight": 63,
|
| 176 |
+
"book_hotel": 116,
|
| 177 |
+
"calculator": 138,
|
| 178 |
+
"calendar": 60,
|
| 179 |
+
"calendar_update": 105,
|
| 180 |
+
"calories": 135,
|
| 181 |
+
"cancel": 147,
|
| 182 |
+
"cancel_reservation": 67,
|
| 183 |
+
"car_rental": 40,
|
| 184 |
+
"card_declined": 49,
|
| 185 |
+
"carry_on": 62,
|
| 186 |
+
"change_accent": 51,
|
| 187 |
+
"change_ai_name": 55,
|
| 188 |
+
"change_language": 29,
|
| 189 |
+
"change_speed": 119,
|
| 190 |
+
"change_user_name": 144,
|
| 191 |
+
"change_volume": 150,
|
| 192 |
+
"confirm_reservation": 85,
|
| 193 |
+
"cook_time": 86,
|
| 194 |
+
"credit_limit": 41,
|
| 195 |
+
"credit_limit_change": 109,
|
| 196 |
+
"credit_score": 69,
|
| 197 |
+
"current_location": 136,
|
| 198 |
+
"damaged_card": 87,
|
| 199 |
+
"date": 95,
|
| 200 |
+
"definition": 139,
|
| 201 |
+
"direct_deposit": 108,
|
| 202 |
+
"directions": 72,
|
| 203 |
+
"distance": 78,
|
| 204 |
+
"do_you_have_pets": 34,
|
| 205 |
+
"exchange_rate": 38,
|
| 206 |
+
"expiration_date": 44,
|
| 207 |
+
"find_phone": 81,
|
| 208 |
+
"flight_status": 80,
|
| 209 |
+
"flip_coin": 33,
|
| 210 |
+
"food_last": 54,
|
| 211 |
+
"freeze_account": 20,
|
| 212 |
+
"fun_fact": 129,
|
| 213 |
+
"gas": 110,
|
| 214 |
+
"gas_type": 8,
|
| 215 |
+
"goodbye": 114,
|
| 216 |
+
"greeting": 82,
|
| 217 |
+
"how_busy": 102,
|
| 218 |
+
"how_old_are_you": 99,
|
| 219 |
+
"improve_credit_score": 28,
|
| 220 |
+
"income": 93,
|
| 221 |
+
"ingredient_substitution": 134,
|
| 222 |
+
"ingredients_list": 112,
|
| 223 |
+
"insurance": 74,
|
| 224 |
+
"insurance_change": 64,
|
| 225 |
+
"interest_rate": 7,
|
| 226 |
+
"international_fees": 137,
|
| 227 |
+
"international_visa": 59,
|
| 228 |
+
"jump_start": 17,
|
| 229 |
+
"last_maintenance": 37,
|
| 230 |
+
"lost_luggage": 113,
|
| 231 |
+
"make_call": 25,
|
| 232 |
+
"maybe": 121,
|
| 233 |
+
"meal_suggestion": 46,
|
| 234 |
+
"meaning_of_life": 22,
|
| 235 |
+
"measurement_conversion": 31,
|
| 236 |
+
"meeting_schedule": 19,
|
| 237 |
+
"min_payment": 143,
|
| 238 |
+
"mpg": 142,
|
| 239 |
+
"new_card": 91,
|
| 240 |
+
"next_holiday": 140,
|
| 241 |
+
"next_song": 118,
|
| 242 |
+
"no": 30,
|
| 243 |
+
"nutrition_info": 1,
|
| 244 |
+
"oil_change_how": 3,
|
| 245 |
+
"oil_change_when": 123,
|
| 246 |
+
"oos": 42,
|
| 247 |
+
"order": 16,
|
| 248 |
+
"order_checks": 127,
|
| 249 |
+
"order_status": 84,
|
| 250 |
+
"pay_bill": 111,
|
| 251 |
+
"payday": 79,
|
| 252 |
+
"pin_change": 89,
|
| 253 |
+
"play_music": 106,
|
| 254 |
+
"plug_type": 120,
|
| 255 |
+
"pto_balance": 126,
|
| 256 |
+
"pto_request": 97,
|
| 257 |
+
"pto_request_status": 101,
|
| 258 |
+
"pto_used": 148,
|
| 259 |
+
"recipe": 104,
|
| 260 |
+
"redeem_rewards": 6,
|
| 261 |
+
"reminder": 76,
|
| 262 |
+
"reminder_update": 53,
|
| 263 |
+
"repeat": 13,
|
| 264 |
+
"replacement_card_duration": 90,
|
| 265 |
+
"report_fraud": 70,
|
| 266 |
+
"report_lost_card": 12,
|
| 267 |
+
"reset_settings": 88,
|
| 268 |
+
"restaurant_reservation": 23,
|
| 269 |
+
"restaurant_reviews": 0,
|
| 270 |
+
"restaurant_suggestion": 145,
|
| 271 |
+
"rewards_balance": 50,
|
| 272 |
+
"roll_dice": 92,
|
| 273 |
+
"rollover_401k": 100,
|
| 274 |
+
"routing": 45,
|
| 275 |
+
"schedule_maintenance": 131,
|
| 276 |
+
"schedule_meeting": 18,
|
| 277 |
+
"share_location": 58,
|
| 278 |
+
"shopping_list": 43,
|
| 279 |
+
"shopping_list_update": 125,
|
| 280 |
+
"smart_home": 10,
|
| 281 |
+
"spelling": 73,
|
| 282 |
+
"spending_history": 71,
|
| 283 |
+
"sync_device": 130,
|
| 284 |
+
"taxes": 94,
|
| 285 |
+
"tell_joke": 36,
|
| 286 |
+
"text": 26,
|
| 287 |
+
"thank_you": 124,
|
| 288 |
+
"time": 4,
|
| 289 |
+
"timer": 32,
|
| 290 |
+
"timezone": 66,
|
| 291 |
+
"tire_change": 47,
|
| 292 |
+
"tire_pressure": 98,
|
| 293 |
+
"todo_list": 48,
|
| 294 |
+
"todo_list_update": 65,
|
| 295 |
+
"traffic": 24,
|
| 296 |
+
"transactions": 68,
|
| 297 |
+
"transfer": 133,
|
| 298 |
+
"translate": 61,
|
| 299 |
+
"travel_alert": 128,
|
| 300 |
+
"travel_notification": 146,
|
| 301 |
+
"travel_suggestion": 149,
|
| 302 |
+
"uber": 39,
|
| 303 |
+
"update_playlist": 141,
|
| 304 |
+
"user_name": 11,
|
| 305 |
+
"vaccines": 52,
|
| 306 |
+
"w2": 122,
|
| 307 |
+
"weather": 5,
|
| 308 |
+
"what_are_your_hobbies": 15,
|
| 309 |
+
"what_can_i_ask_you": 115,
|
| 310 |
+
"what_is_your_name": 75,
|
| 311 |
+
"what_song": 21,
|
| 312 |
+
"where_are_you_from": 77,
|
| 313 |
+
"whisper_mode": 14,
|
| 314 |
+
"who_do_you_work_for": 57,
|
| 315 |
+
"who_made_you": 96,
|
| 316 |
+
"yes": 107
|
| 317 |
+
},
|
| 318 |
+
"max_position_embeddings": 512,
|
| 319 |
+
"model_type": "distilbert",
|
| 320 |
+
"n_heads": 12,
|
| 321 |
+
"n_layers": 6,
|
| 322 |
+
"pad_token_id": 0,
|
| 323 |
+
"problem_type": "single_label_classification",
|
| 324 |
+
"qa_dropout": 0.1,
|
| 325 |
+
"seq_classif_dropout": 0.2,
|
| 326 |
+
"sinusoidal_pos_embds": false,
|
| 327 |
+
"tie_weights_": true,
|
| 328 |
+
"torch_dtype": "float32",
|
| 329 |
+
"transformers_version": "4.45.2",
|
| 330 |
+
"vocab_size": 30522
|
| 331 |
+
}
|
run-3/checkpoint-2862/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9a7bec75e776cd487733ff1e96487bc2505714d839c2eeecdc042accc0f1b3a9
|
| 3 |
+
size 268290900
|
run-3/checkpoint-2862/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c59bbd466a86cf42e2c98a7569a2be10f5bf38a14ecfb954c2c6fccd1a2d2045
|
| 3 |
+
size 536643898
|
run-3/checkpoint-2862/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4b9e71d8b85c2c3bd3c2ff01693d7078f05f0c4c5121e6afa9ab875cbeb7116f
|
| 3 |
+
size 14244
|
run-3/checkpoint-2862/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2a738824491ff8b261b6db2a842e15bf80355d04a07892a16df2c4715b2ee556
|
| 3 |
+
size 1064
|
run-3/checkpoint-2862/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 |
+
}
|
run-3/checkpoint-2862/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
run-3/checkpoint-2862/tokenizer_config.json
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "[PAD]",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"100": {
|
| 12 |
+
"content": "[UNK]",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"101": {
|
| 20 |
+
"content": "[CLS]",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"102": {
|
| 28 |
+
"content": "[SEP]",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"103": {
|
| 36 |
+
"content": "[MASK]",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"clean_up_tokenization_spaces": false,
|
| 45 |
+
"cls_token": "[CLS]",
|
| 46 |
+
"do_lower_case": true,
|
| 47 |
+
"mask_token": "[MASK]",
|
| 48 |
+
"model_max_length": 512,
|
| 49 |
+
"pad_token": "[PAD]",
|
| 50 |
+
"sep_token": "[SEP]",
|
| 51 |
+
"strip_accents": null,
|
| 52 |
+
"tokenize_chinese_chars": true,
|
| 53 |
+
"tokenizer_class": "DistilBertTokenizer",
|
| 54 |
+
"unk_token": "[UNK]"
|
| 55 |
+
}
|
run-3/checkpoint-2862/trainer_state.json
ADDED
|
@@ -0,0 +1,172 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 9.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2862,
|
| 7 |
+
"is_hyper_param_search": true,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.9968553459119497,
|
| 13 |
+
"grad_norm": 0.965503990650177,
|
| 14 |
+
"learning_rate": 1.7784765897973445e-05,
|
| 15 |
+
"loss": 0.8047,
|
| 16 |
+
"step": 317
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 1.0,
|
| 20 |
+
"eval_accuracy": 0.6812903225806451,
|
| 21 |
+
"eval_loss": 0.4058631658554077,
|
| 22 |
+
"eval_runtime": 1.6943,
|
| 23 |
+
"eval_samples_per_second": 1829.632,
|
| 24 |
+
"eval_steps_per_second": 38.363,
|
| 25 |
+
"step": 318
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 1.9937106918238994,
|
| 29 |
+
"grad_norm": 0.7085171341896057,
|
| 30 |
+
"learning_rate": 1.556953179594689e-05,
|
| 31 |
+
"loss": 0.3062,
|
| 32 |
+
"step": 634
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 2.0,
|
| 36 |
+
"eval_accuracy": 0.8503225806451613,
|
| 37 |
+
"eval_loss": 0.13832151889801025,
|
| 38 |
+
"eval_runtime": 1.7583,
|
| 39 |
+
"eval_samples_per_second": 1763.081,
|
| 40 |
+
"eval_steps_per_second": 36.968,
|
| 41 |
+
"step": 636
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 2.990566037735849,
|
| 45 |
+
"grad_norm": 0.619353711605072,
|
| 46 |
+
"learning_rate": 1.3354297693920338e-05,
|
| 47 |
+
"loss": 0.1421,
|
| 48 |
+
"step": 951
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 3.0,
|
| 52 |
+
"eval_accuracy": 0.8958064516129032,
|
| 53 |
+
"eval_loss": 0.07200490683317184,
|
| 54 |
+
"eval_runtime": 1.7884,
|
| 55 |
+
"eval_samples_per_second": 1733.388,
|
| 56 |
+
"eval_steps_per_second": 36.345,
|
| 57 |
+
"step": 954
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 3.9874213836477987,
|
| 61 |
+
"grad_norm": 0.5501407384872437,
|
| 62 |
+
"learning_rate": 1.1139063591893781e-05,
|
| 63 |
+
"loss": 0.0916,
|
| 64 |
+
"step": 1268
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 4.0,
|
| 68 |
+
"eval_accuracy": 0.9125806451612903,
|
| 69 |
+
"eval_loss": 0.05142820626497269,
|
| 70 |
+
"eval_runtime": 1.7858,
|
| 71 |
+
"eval_samples_per_second": 1735.903,
|
| 72 |
+
"eval_steps_per_second": 36.398,
|
| 73 |
+
"step": 1272
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 4.984276729559748,
|
| 77 |
+
"grad_norm": 0.3894004821777344,
|
| 78 |
+
"learning_rate": 8.923829489867226e-06,
|
| 79 |
+
"loss": 0.0717,
|
| 80 |
+
"step": 1585
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 5.0,
|
| 84 |
+
"eval_accuracy": 0.925483870967742,
|
| 85 |
+
"eval_loss": 0.04311361908912659,
|
| 86 |
+
"eval_runtime": 1.7985,
|
| 87 |
+
"eval_samples_per_second": 1723.616,
|
| 88 |
+
"eval_steps_per_second": 36.14,
|
| 89 |
+
"step": 1590
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 5.981132075471698,
|
| 93 |
+
"grad_norm": 0.4861418902873993,
|
| 94 |
+
"learning_rate": 6.708595387840672e-06,
|
| 95 |
+
"loss": 0.0615,
|
| 96 |
+
"step": 1902
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 6.0,
|
| 100 |
+
"eval_accuracy": 0.9261290322580645,
|
| 101 |
+
"eval_loss": 0.03821605071425438,
|
| 102 |
+
"eval_runtime": 1.8068,
|
| 103 |
+
"eval_samples_per_second": 1715.761,
|
| 104 |
+
"eval_steps_per_second": 35.976,
|
| 105 |
+
"step": 1908
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 6.977987421383648,
|
| 109 |
+
"grad_norm": 0.3920305669307709,
|
| 110 |
+
"learning_rate": 4.4933612858141165e-06,
|
| 111 |
+
"loss": 0.0561,
|
| 112 |
+
"step": 2219
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 7.0,
|
| 116 |
+
"eval_accuracy": 0.9325806451612904,
|
| 117 |
+
"eval_loss": 0.035470735281705856,
|
| 118 |
+
"eval_runtime": 1.7961,
|
| 119 |
+
"eval_samples_per_second": 1725.979,
|
| 120 |
+
"eval_steps_per_second": 36.19,
|
| 121 |
+
"step": 2226
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 7.9748427672955975,
|
| 125 |
+
"grad_norm": 0.3401239514350891,
|
| 126 |
+
"learning_rate": 2.2781271837875614e-06,
|
| 127 |
+
"loss": 0.0528,
|
| 128 |
+
"step": 2536
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 8.0,
|
| 132 |
+
"eval_accuracy": 0.9316129032258065,
|
| 133 |
+
"eval_loss": 0.03403536230325699,
|
| 134 |
+
"eval_runtime": 1.8035,
|
| 135 |
+
"eval_samples_per_second": 1718.877,
|
| 136 |
+
"eval_steps_per_second": 36.041,
|
| 137 |
+
"step": 2544
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 8.971698113207546,
|
| 141 |
+
"grad_norm": 0.3334876596927643,
|
| 142 |
+
"learning_rate": 6.289308176100629e-08,
|
| 143 |
+
"loss": 0.0511,
|
| 144 |
+
"step": 2853
|
| 145 |
+
}
|
| 146 |
+
],
|
| 147 |
+
"logging_steps": 317,
|
| 148 |
+
"max_steps": 2862,
|
| 149 |
+
"num_input_tokens_seen": 0,
|
| 150 |
+
"num_train_epochs": 9,
|
| 151 |
+
"save_steps": 1000000000.0,
|
| 152 |
+
"stateful_callbacks": {
|
| 153 |
+
"TrainerControl": {
|
| 154 |
+
"args": {
|
| 155 |
+
"should_epoch_stop": false,
|
| 156 |
+
"should_evaluate": false,
|
| 157 |
+
"should_log": false,
|
| 158 |
+
"should_save": true,
|
| 159 |
+
"should_training_stop": true
|
| 160 |
+
},
|
| 161 |
+
"attributes": {}
|
| 162 |
+
}
|
| 163 |
+
},
|
| 164 |
+
"total_flos": 742838622388212.0,
|
| 165 |
+
"train_batch_size": 48,
|
| 166 |
+
"trial_name": null,
|
| 167 |
+
"trial_params": {
|
| 168 |
+
"alpha": 0.6623158454366724,
|
| 169 |
+
"num_train_epochs": 9,
|
| 170 |
+
"temperature": 2
|
| 171 |
+
}
|
| 172 |
+
}
|
run-3/checkpoint-2862/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:65529af1ef1c2d0831a1f47d1b8ad3ab10ce486dd1675094550361b35d052a83
|
| 3 |
+
size 5304
|
run-3/checkpoint-2862/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
runs/Oct23_23-31-51_p16/events.out.tfevents.1729698645.p16.148475.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fd67f9d56d38103d8868eb20753b9b9b4a6565579d73301aa1663ff1a7f3700f
|
| 3 |
+
size 15504
|
runs/Oct23_23-31-51_p16/events.out.tfevents.1729698905.p16.148475.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:45494df703b54ee98a129bfaf29fef1d1c5b88383843d116b2be396fe31f62fc
|
| 3 |
+
size 18138
|
runs/Oct23_23-31-51_p16/events.out.tfevents.1729699275.p16.148475.4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:da42e4d72b97e3003f12a52546359c72fbbff2818795df5bf75177aa00635771
|
| 3 |
+
size 15467
|
runs/Oct23_23-31-51_p16/events.out.tfevents.1729699541.p16.148475.5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eeecb1e38ccc52525276864e7595d16cb41abe1018f9ec72b49a499647b08bb4
|
| 3 |
+
size 17603
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5304
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:65529af1ef1c2d0831a1f47d1b8ad3ab10ce486dd1675094550361b35d052a83
|
| 3 |
size 5304
|