Upload folder using huggingface_hub
Browse files- config.json +269 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +2883 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,269 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "facebook/hubert-large-ll60k",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"apply_spec_augment": true,
|
| 5 |
+
"architectures": [
|
| 6 |
+
"HubertForSequenceClassification"
|
| 7 |
+
],
|
| 8 |
+
"attention_dropout": 0.1,
|
| 9 |
+
"bos_token_id": 1,
|
| 10 |
+
"classifier_proj_size": 256,
|
| 11 |
+
"conv_bias": true,
|
| 12 |
+
"conv_dim": [
|
| 13 |
+
512,
|
| 14 |
+
512,
|
| 15 |
+
512,
|
| 16 |
+
512,
|
| 17 |
+
512,
|
| 18 |
+
512,
|
| 19 |
+
512
|
| 20 |
+
],
|
| 21 |
+
"conv_kernel": [
|
| 22 |
+
10,
|
| 23 |
+
3,
|
| 24 |
+
3,
|
| 25 |
+
3,
|
| 26 |
+
3,
|
| 27 |
+
2,
|
| 28 |
+
2
|
| 29 |
+
],
|
| 30 |
+
"conv_stride": [
|
| 31 |
+
5,
|
| 32 |
+
2,
|
| 33 |
+
2,
|
| 34 |
+
2,
|
| 35 |
+
2,
|
| 36 |
+
2,
|
| 37 |
+
2
|
| 38 |
+
],
|
| 39 |
+
"ctc_loss_reduction": "sum",
|
| 40 |
+
"ctc_zero_infinity": false,
|
| 41 |
+
"do_stable_layer_norm": true,
|
| 42 |
+
"eos_token_id": 2,
|
| 43 |
+
"feat_extract_activation": "gelu",
|
| 44 |
+
"feat_extract_dropout": 0.0,
|
| 45 |
+
"feat_extract_norm": "layer",
|
| 46 |
+
"feat_proj_dropout": 0.1,
|
| 47 |
+
"feat_proj_layer_norm": true,
|
| 48 |
+
"final_dropout": 0.0,
|
| 49 |
+
"gradient_checkpointing": false,
|
| 50 |
+
"hidden_act": "gelu",
|
| 51 |
+
"hidden_dropout": 0.1,
|
| 52 |
+
"hidden_size": 1024,
|
| 53 |
+
"id2label": {
|
| 54 |
+
"0": "calendar_set",
|
| 55 |
+
"1": "wavs/audio_volume_up",
|
| 56 |
+
"10": "qa_stock",
|
| 57 |
+
"11": "play_music",
|
| 58 |
+
"12": "recommendation_events",
|
| 59 |
+
"13": "qa_definition",
|
| 60 |
+
"14": "alarm_remove",
|
| 61 |
+
"15": "play_podcasts",
|
| 62 |
+
"16": "social_query",
|
| 63 |
+
"17": "email_addcontact",
|
| 64 |
+
"18": "news_query",
|
| 65 |
+
"19": "calendar_query",
|
| 66 |
+
"2": "iot_hue_lightup",
|
| 67 |
+
"20": "music_likeness",
|
| 68 |
+
"21": "general_quirky",
|
| 69 |
+
"22": "qa_factoid",
|
| 70 |
+
"23": "takeaway_order",
|
| 71 |
+
"24": "play_wavs/audiobook",
|
| 72 |
+
"25": "iot_cleaning",
|
| 73 |
+
"26": "general_greet",
|
| 74 |
+
"27": "transport_query",
|
| 75 |
+
"28": "transport_taxi",
|
| 76 |
+
"29": "email_sendemail",
|
| 77 |
+
"3": "weather_query",
|
| 78 |
+
"30": "general_joke",
|
| 79 |
+
"31": "qa_maths",
|
| 80 |
+
"32": "social_post",
|
| 81 |
+
"33": "transport_ticket",
|
| 82 |
+
"34": "cooking_recipe",
|
| 83 |
+
"35": "music_settings",
|
| 84 |
+
"36": "calendar_remove",
|
| 85 |
+
"37": "iot_wemo_on",
|
| 86 |
+
"38": "iot_hue_lightchange",
|
| 87 |
+
"39": "play_radio",
|
| 88 |
+
"4": "iot_coffee",
|
| 89 |
+
"40": "email_querycontact",
|
| 90 |
+
"41": "transport_traffic",
|
| 91 |
+
"42": "qa_currency",
|
| 92 |
+
"43": "datetime_query",
|
| 93 |
+
"44": "iot_hue_lightoff",
|
| 94 |
+
"45": "takeaway_query",
|
| 95 |
+
"46": "lists_createoradd",
|
| 96 |
+
"47": "music_query",
|
| 97 |
+
"48": "recommendation_locations",
|
| 98 |
+
"49": "lists_query",
|
| 99 |
+
"5": "wavs/audio_volume_mute",
|
| 100 |
+
"50": "recommendation_movies",
|
| 101 |
+
"51": "iot_wemo_off",
|
| 102 |
+
"52": "iot_hue_lighton",
|
| 103 |
+
"53": "sendemail",
|
| 104 |
+
"54": "wavs/audio_volume_down",
|
| 105 |
+
"55": "play_game",
|
| 106 |
+
"56": "music",
|
| 107 |
+
"57": "datetime_convert",
|
| 108 |
+
"58": "iot_hue_lightdim",
|
| 109 |
+
"59": "query",
|
| 110 |
+
"6": "lists_remove",
|
| 111 |
+
"60": "createoradd",
|
| 112 |
+
"61": "music_dislikeness",
|
| 113 |
+
"62": "podcasts",
|
| 114 |
+
"63": "joke",
|
| 115 |
+
"64": "set",
|
| 116 |
+
"65": "hue_lightup",
|
| 117 |
+
"66": "factoid",
|
| 118 |
+
"67": "wavs/audio_volume_other",
|
| 119 |
+
"68": "hue_lightoff",
|
| 120 |
+
"69": "quirky",
|
| 121 |
+
"7": "email_query",
|
| 122 |
+
"70": "querycontact",
|
| 123 |
+
"71": "radio",
|
| 124 |
+
"72": "addcontact",
|
| 125 |
+
"73": "greet",
|
| 126 |
+
"74": "ticket",
|
| 127 |
+
"75": "traffic",
|
| 128 |
+
"76": "cooking_query",
|
| 129 |
+
"77": "remove",
|
| 130 |
+
"78": "currency",
|
| 131 |
+
"79": "coffee",
|
| 132 |
+
"8": "alarm_set",
|
| 133 |
+
"80": "game",
|
| 134 |
+
"81": "wemo_on",
|
| 135 |
+
"82": "definition",
|
| 136 |
+
"83": "events",
|
| 137 |
+
"84": "post",
|
| 138 |
+
"85": "hue_lightdim",
|
| 139 |
+
"86": "convert",
|
| 140 |
+
"87": "wemo_off",
|
| 141 |
+
"88": "cleaning",
|
| 142 |
+
"89": "settings",
|
| 143 |
+
"9": "alarm_query",
|
| 144 |
+
"90": "volume_other"
|
| 145 |
+
},
|
| 146 |
+
"initializer_range": 0.02,
|
| 147 |
+
"intermediate_size": 4096,
|
| 148 |
+
"label2id": {
|
| 149 |
+
"addcontact": "72",
|
| 150 |
+
"alarm_query": "9",
|
| 151 |
+
"alarm_remove": "14",
|
| 152 |
+
"alarm_set": "8",
|
| 153 |
+
"calendar_query": "19",
|
| 154 |
+
"calendar_remove": "36",
|
| 155 |
+
"calendar_set": "0",
|
| 156 |
+
"cleaning": "88",
|
| 157 |
+
"coffee": "79",
|
| 158 |
+
"convert": "86",
|
| 159 |
+
"cooking_query": "76",
|
| 160 |
+
"cooking_recipe": "34",
|
| 161 |
+
"createoradd": "60",
|
| 162 |
+
"currency": "78",
|
| 163 |
+
"datetime_convert": "57",
|
| 164 |
+
"datetime_query": "43",
|
| 165 |
+
"definition": "82",
|
| 166 |
+
"email_addcontact": "17",
|
| 167 |
+
"email_query": "7",
|
| 168 |
+
"email_querycontact": "40",
|
| 169 |
+
"email_sendemail": "29",
|
| 170 |
+
"events": "83",
|
| 171 |
+
"factoid": "66",
|
| 172 |
+
"game": "80",
|
| 173 |
+
"general_greet": "26",
|
| 174 |
+
"general_joke": "30",
|
| 175 |
+
"general_quirky": "21",
|
| 176 |
+
"greet": "73",
|
| 177 |
+
"hue_lightdim": "85",
|
| 178 |
+
"hue_lightoff": "68",
|
| 179 |
+
"hue_lightup": "65",
|
| 180 |
+
"iot_cleaning": "25",
|
| 181 |
+
"iot_coffee": "4",
|
| 182 |
+
"iot_hue_lightchange": "38",
|
| 183 |
+
"iot_hue_lightdim": "58",
|
| 184 |
+
"iot_hue_lightoff": "44",
|
| 185 |
+
"iot_hue_lighton": "52",
|
| 186 |
+
"iot_hue_lightup": "2",
|
| 187 |
+
"iot_wemo_off": "51",
|
| 188 |
+
"iot_wemo_on": "37",
|
| 189 |
+
"joke": "63",
|
| 190 |
+
"lists_createoradd": "46",
|
| 191 |
+
"lists_query": "49",
|
| 192 |
+
"lists_remove": "6",
|
| 193 |
+
"music": "56",
|
| 194 |
+
"music_dislikeness": "61",
|
| 195 |
+
"music_likeness": "20",
|
| 196 |
+
"music_query": "47",
|
| 197 |
+
"music_settings": "35",
|
| 198 |
+
"news_query": "18",
|
| 199 |
+
"play_game": "55",
|
| 200 |
+
"play_music": "11",
|
| 201 |
+
"play_podcasts": "15",
|
| 202 |
+
"play_radio": "39",
|
| 203 |
+
"play_wavs/audiobook": "24",
|
| 204 |
+
"podcasts": "62",
|
| 205 |
+
"post": "84",
|
| 206 |
+
"qa_currency": "42",
|
| 207 |
+
"qa_definition": "13",
|
| 208 |
+
"qa_factoid": "22",
|
| 209 |
+
"qa_maths": "31",
|
| 210 |
+
"qa_stock": "10",
|
| 211 |
+
"query": "59",
|
| 212 |
+
"querycontact": "70",
|
| 213 |
+
"quirky": "69",
|
| 214 |
+
"radio": "71",
|
| 215 |
+
"recommendation_events": "12",
|
| 216 |
+
"recommendation_locations": "48",
|
| 217 |
+
"recommendation_movies": "50",
|
| 218 |
+
"remove": "77",
|
| 219 |
+
"sendemail": "53",
|
| 220 |
+
"set": "64",
|
| 221 |
+
"settings": "89",
|
| 222 |
+
"social_post": "32",
|
| 223 |
+
"social_query": "16",
|
| 224 |
+
"takeaway_order": "23",
|
| 225 |
+
"takeaway_query": "45",
|
| 226 |
+
"ticket": "74",
|
| 227 |
+
"traffic": "75",
|
| 228 |
+
"transport_query": "27",
|
| 229 |
+
"transport_taxi": "28",
|
| 230 |
+
"transport_ticket": "33",
|
| 231 |
+
"transport_traffic": "41",
|
| 232 |
+
"volume_other": "90",
|
| 233 |
+
"wavs/audio_volume_down": "54",
|
| 234 |
+
"wavs/audio_volume_mute": "5",
|
| 235 |
+
"wavs/audio_volume_other": "67",
|
| 236 |
+
"wavs/audio_volume_up": "1",
|
| 237 |
+
"weather_query": "3",
|
| 238 |
+
"wemo_off": "87",
|
| 239 |
+
"wemo_on": "81"
|
| 240 |
+
},
|
| 241 |
+
"layer_norm_eps": 1e-05,
|
| 242 |
+
"layerdrop": 0.1,
|
| 243 |
+
"mask_channel_length": 10,
|
| 244 |
+
"mask_channel_min_space": 1,
|
| 245 |
+
"mask_channel_other": 0.0,
|
| 246 |
+
"mask_channel_prob": 0.0,
|
| 247 |
+
"mask_channel_selection": "static",
|
| 248 |
+
"mask_feature_length": 10,
|
| 249 |
+
"mask_feature_min_masks": 0,
|
| 250 |
+
"mask_feature_prob": 0.0,
|
| 251 |
+
"mask_time_length": 10,
|
| 252 |
+
"mask_time_min_masks": 2,
|
| 253 |
+
"mask_time_min_space": 1,
|
| 254 |
+
"mask_time_other": 0.0,
|
| 255 |
+
"mask_time_prob": 0.075,
|
| 256 |
+
"mask_time_selection": "static",
|
| 257 |
+
"model_type": "hubert",
|
| 258 |
+
"num_attention_heads": 16,
|
| 259 |
+
"num_conv_pos_embedding_groups": 16,
|
| 260 |
+
"num_conv_pos_embeddings": 128,
|
| 261 |
+
"num_feat_extract_layers": 7,
|
| 262 |
+
"num_hidden_layers": 24,
|
| 263 |
+
"pad_token_id": 0,
|
| 264 |
+
"tokenizer_class": "Wav2Vec2CTCTokenizer",
|
| 265 |
+
"torch_dtype": "float32",
|
| 266 |
+
"transformers_version": "4.45.2",
|
| 267 |
+
"use_weighted_layer_sum": false,
|
| 268 |
+
"vocab_size": 32
|
| 269 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:94c5bf935da29df4273858bb64655d3e483c1565367faa48e0454fc328d456cb
|
| 3 |
+
size 1262949964
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:29ed984b50ee2d8d070a53dc90f36912258aaadefb6ebf1b36c4e527a1f5e574
|
| 3 |
+
size 2526152656
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d724277b411545cf366b67774cfe033bc7ad010ad2913a328760fe568fdb6821
|
| 3 |
+
size 14308
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:417c32b40ce266eae8ab814d0af12409a7770b12ed2c5afd1b1ffbf9fa27b42f
|
| 3 |
+
size 1064
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2883 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.7934407364787112,
|
| 3 |
+
"best_model_checkpoint": "results/facebook/hubert-large-ll60k/42/_retain/checkpoint-30000",
|
| 4 |
+
"epoch": 75.80543272267846,
|
| 5 |
+
"eval_steps": 400,
|
| 6 |
+
"global_step": 30000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.2526847757422615,
|
| 13 |
+
"grad_norm": 7.531186580657959,
|
| 14 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 15 |
+
"loss": 4.4765,
|
| 16 |
+
"step": 100
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.505369551484523,
|
| 20 |
+
"grad_norm": 1.3038771152496338,
|
| 21 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 22 |
+
"loss": 4.1328,
|
| 23 |
+
"step": 200
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.7580543272267846,
|
| 27 |
+
"grad_norm": 1.8434199094772339,
|
| 28 |
+
"learning_rate": 5e-05,
|
| 29 |
+
"loss": 3.8858,
|
| 30 |
+
"step": 300
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 1.010739102969046,
|
| 34 |
+
"grad_norm": 4.255322456359863,
|
| 35 |
+
"learning_rate": 6.666666666666667e-05,
|
| 36 |
+
"loss": 3.8141,
|
| 37 |
+
"step": 400
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 1.010739102969046,
|
| 41 |
+
"eval_accuracy": 0.08400460299194476,
|
| 42 |
+
"eval_f1_macro": 0.008669601975968978,
|
| 43 |
+
"eval_loss": 3.7715396881103516,
|
| 44 |
+
"eval_runtime": 108.6457,
|
| 45 |
+
"eval_samples_per_second": 79.985,
|
| 46 |
+
"eval_steps_per_second": 2.504,
|
| 47 |
+
"step": 400
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 1.2634238787113077,
|
| 51 |
+
"grad_norm": 5.987393379211426,
|
| 52 |
+
"learning_rate": 8.333333333333333e-05,
|
| 53 |
+
"loss": 3.7488,
|
| 54 |
+
"step": 500
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 1.5161086544535691,
|
| 58 |
+
"grad_norm": 4.6245269775390625,
|
| 59 |
+
"learning_rate": 0.0001,
|
| 60 |
+
"loss": 3.6051,
|
| 61 |
+
"step": 600
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 1.7687934301958306,
|
| 65 |
+
"grad_norm": 7.047222137451172,
|
| 66 |
+
"learning_rate": 0.00011666666666666667,
|
| 67 |
+
"loss": 3.2059,
|
| 68 |
+
"step": 700
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 2.021478205938092,
|
| 72 |
+
"grad_norm": 11.504085540771484,
|
| 73 |
+
"learning_rate": 0.00013333333333333334,
|
| 74 |
+
"loss": 2.7291,
|
| 75 |
+
"step": 800
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 2.021478205938092,
|
| 79 |
+
"eval_accuracy": 0.42681242807825087,
|
| 80 |
+
"eval_f1_macro": 0.17578343384188652,
|
| 81 |
+
"eval_loss": 2.323087215423584,
|
| 82 |
+
"eval_runtime": 108.6246,
|
| 83 |
+
"eval_samples_per_second": 80.0,
|
| 84 |
+
"eval_steps_per_second": 2.504,
|
| 85 |
+
"step": 800
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 2.274162981680354,
|
| 89 |
+
"grad_norm": 9.302848815917969,
|
| 90 |
+
"learning_rate": 0.00015,
|
| 91 |
+
"loss": 2.2862,
|
| 92 |
+
"step": 900
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 2.5268477574226154,
|
| 96 |
+
"grad_norm": 16.832901000976562,
|
| 97 |
+
"learning_rate": 0.00016666666666666666,
|
| 98 |
+
"loss": 1.9558,
|
| 99 |
+
"step": 1000
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 2.779532533164877,
|
| 103 |
+
"grad_norm": 18.594715118408203,
|
| 104 |
+
"learning_rate": 0.00018333333333333334,
|
| 105 |
+
"loss": 1.6921,
|
| 106 |
+
"step": 1100
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 3.0322173089071383,
|
| 110 |
+
"grad_norm": 21.0137882232666,
|
| 111 |
+
"learning_rate": 0.0002,
|
| 112 |
+
"loss": 1.4799,
|
| 113 |
+
"step": 1200
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 3.0322173089071383,
|
| 117 |
+
"eval_accuracy": 0.68204833141542,
|
| 118 |
+
"eval_f1_macro": 0.4656587521593648,
|
| 119 |
+
"eval_loss": 1.2915173768997192,
|
| 120 |
+
"eval_runtime": 108.6127,
|
| 121 |
+
"eval_samples_per_second": 80.009,
|
| 122 |
+
"eval_steps_per_second": 2.504,
|
| 123 |
+
"step": 1200
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 3.2849020846493997,
|
| 127 |
+
"grad_norm": 10.087828636169434,
|
| 128 |
+
"learning_rate": 0.00021666666666666668,
|
| 129 |
+
"loss": 1.264,
|
| 130 |
+
"step": 1300
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 3.537586860391661,
|
| 134 |
+
"grad_norm": 8.730626106262207,
|
| 135 |
+
"learning_rate": 0.00023333333333333333,
|
| 136 |
+
"loss": 1.1697,
|
| 137 |
+
"step": 1400
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 3.790271636133923,
|
| 141 |
+
"grad_norm": 19.08763885498047,
|
| 142 |
+
"learning_rate": 0.00025,
|
| 143 |
+
"loss": 1.0774,
|
| 144 |
+
"step": 1500
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 4.042956411876184,
|
| 148 |
+
"grad_norm": 12.934535026550293,
|
| 149 |
+
"learning_rate": 0.0002666666666666667,
|
| 150 |
+
"loss": 1.0037,
|
| 151 |
+
"step": 1600
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 4.042956411876184,
|
| 155 |
+
"eval_accuracy": 0.7395857307249712,
|
| 156 |
+
"eval_f1_macro": 0.5527706813137244,
|
| 157 |
+
"eval_loss": 1.1226997375488281,
|
| 158 |
+
"eval_runtime": 108.4906,
|
| 159 |
+
"eval_samples_per_second": 80.099,
|
| 160 |
+
"eval_steps_per_second": 2.507,
|
| 161 |
+
"step": 1600
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 4.295641187618446,
|
| 165 |
+
"grad_norm": 11.546894073486328,
|
| 166 |
+
"learning_rate": 0.00028333333333333335,
|
| 167 |
+
"loss": 0.8737,
|
| 168 |
+
"step": 1700
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 4.548325963360708,
|
| 172 |
+
"grad_norm": 10.242780685424805,
|
| 173 |
+
"learning_rate": 0.0003,
|
| 174 |
+
"loss": 0.8296,
|
| 175 |
+
"step": 1800
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 4.801010739102969,
|
| 179 |
+
"grad_norm": 9.696259498596191,
|
| 180 |
+
"learning_rate": 0.00031666666666666665,
|
| 181 |
+
"loss": 0.8139,
|
| 182 |
+
"step": 1900
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 5.053695514845231,
|
| 186 |
+
"grad_norm": 6.417578220367432,
|
| 187 |
+
"learning_rate": 0.0003333333333333333,
|
| 188 |
+
"loss": 0.7768,
|
| 189 |
+
"step": 2000
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 5.053695514845231,
|
| 193 |
+
"eval_accuracy": 0.7643268124280782,
|
| 194 |
+
"eval_f1_macro": 0.5816116111187802,
|
| 195 |
+
"eval_loss": 1.0890038013458252,
|
| 196 |
+
"eval_runtime": 108.7404,
|
| 197 |
+
"eval_samples_per_second": 79.915,
|
| 198 |
+
"eval_steps_per_second": 2.501,
|
| 199 |
+
"step": 2000
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 5.306380290587492,
|
| 203 |
+
"grad_norm": 9.01593017578125,
|
| 204 |
+
"learning_rate": 0.00035,
|
| 205 |
+
"loss": 0.7057,
|
| 206 |
+
"step": 2100
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 5.559065066329754,
|
| 210 |
+
"grad_norm": 6.170841217041016,
|
| 211 |
+
"learning_rate": 0.00036666666666666667,
|
| 212 |
+
"loss": 0.6655,
|
| 213 |
+
"step": 2200
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 5.8117498420720155,
|
| 217 |
+
"grad_norm": 9.565030097961426,
|
| 218 |
+
"learning_rate": 0.00038333333333333334,
|
| 219 |
+
"loss": 0.6767,
|
| 220 |
+
"step": 2300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 6.0644346178142765,
|
| 224 |
+
"grad_norm": 11.446185111999512,
|
| 225 |
+
"learning_rate": 0.0004,
|
| 226 |
+
"loss": 0.6207,
|
| 227 |
+
"step": 2400
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 6.0644346178142765,
|
| 231 |
+
"eval_accuracy": 0.756156501726122,
|
| 232 |
+
"eval_f1_macro": 0.5962013864268475,
|
| 233 |
+
"eval_loss": 1.1590509414672852,
|
| 234 |
+
"eval_runtime": 108.3732,
|
| 235 |
+
"eval_samples_per_second": 80.186,
|
| 236 |
+
"eval_steps_per_second": 2.51,
|
| 237 |
+
"step": 2400
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 6.317119393556538,
|
| 241 |
+
"grad_norm": 6.382347583770752,
|
| 242 |
+
"learning_rate": 0.0004166666666666667,
|
| 243 |
+
"loss": 0.5489,
|
| 244 |
+
"step": 2500
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 6.569804169298799,
|
| 248 |
+
"grad_norm": 7.796228408813477,
|
| 249 |
+
"learning_rate": 0.00043333333333333337,
|
| 250 |
+
"loss": 0.5615,
|
| 251 |
+
"step": 2600
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 6.822488945041061,
|
| 255 |
+
"grad_norm": 10.157086372375488,
|
| 256 |
+
"learning_rate": 0.00045000000000000004,
|
| 257 |
+
"loss": 0.61,
|
| 258 |
+
"step": 2700
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 7.075173720783323,
|
| 262 |
+
"grad_norm": 9.361109733581543,
|
| 263 |
+
"learning_rate": 0.00046666666666666666,
|
| 264 |
+
"loss": 0.5457,
|
| 265 |
+
"step": 2800
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 7.075173720783323,
|
| 269 |
+
"eval_accuracy": 0.75385500575374,
|
| 270 |
+
"eval_f1_macro": 0.594201593201007,
|
| 271 |
+
"eval_loss": 1.2054548263549805,
|
| 272 |
+
"eval_runtime": 108.935,
|
| 273 |
+
"eval_samples_per_second": 79.772,
|
| 274 |
+
"eval_steps_per_second": 2.497,
|
| 275 |
+
"step": 2800
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 7.327858496525584,
|
| 279 |
+
"grad_norm": 4.478394031524658,
|
| 280 |
+
"learning_rate": 0.00048333333333333334,
|
| 281 |
+
"loss": 0.5006,
|
| 282 |
+
"step": 2900
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 7.580543272267846,
|
| 286 |
+
"grad_norm": 7.6330366134643555,
|
| 287 |
+
"learning_rate": 0.0005,
|
| 288 |
+
"loss": 0.5456,
|
| 289 |
+
"step": 3000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 7.833228048010107,
|
| 293 |
+
"grad_norm": 6.51686429977417,
|
| 294 |
+
"learning_rate": 0.0004981481481481482,
|
| 295 |
+
"loss": 0.5245,
|
| 296 |
+
"step": 3100
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 8.085912823752368,
|
| 300 |
+
"grad_norm": 6.197965621948242,
|
| 301 |
+
"learning_rate": 0.0004962962962962963,
|
| 302 |
+
"loss": 0.4905,
|
| 303 |
+
"step": 3200
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 8.085912823752368,
|
| 307 |
+
"eval_accuracy": 0.7558112773302647,
|
| 308 |
+
"eval_f1_macro": 0.591751644264711,
|
| 309 |
+
"eval_loss": 1.168681263923645,
|
| 310 |
+
"eval_runtime": 108.6923,
|
| 311 |
+
"eval_samples_per_second": 79.95,
|
| 312 |
+
"eval_steps_per_second": 2.502,
|
| 313 |
+
"step": 3200
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 8.33859759949463,
|
| 317 |
+
"grad_norm": 5.975191116333008,
|
| 318 |
+
"learning_rate": 0.0004944444444444445,
|
| 319 |
+
"loss": 0.4324,
|
| 320 |
+
"step": 3300
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 8.591282375236892,
|
| 324 |
+
"grad_norm": 6.450153827667236,
|
| 325 |
+
"learning_rate": 0.0004925925925925925,
|
| 326 |
+
"loss": 0.4385,
|
| 327 |
+
"step": 3400
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 8.843967150979154,
|
| 331 |
+
"grad_norm": 3.6089916229248047,
|
| 332 |
+
"learning_rate": 0.0004907407407407408,
|
| 333 |
+
"loss": 0.4312,
|
| 334 |
+
"step": 3500
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 9.096651926721416,
|
| 338 |
+
"grad_norm": 5.325081825256348,
|
| 339 |
+
"learning_rate": 0.0004888888888888889,
|
| 340 |
+
"loss": 0.3977,
|
| 341 |
+
"step": 3600
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 9.096651926721416,
|
| 345 |
+
"eval_accuracy": 0.7486766398158803,
|
| 346 |
+
"eval_f1_macro": 0.5505823611427755,
|
| 347 |
+
"eval_loss": 1.3269578218460083,
|
| 348 |
+
"eval_runtime": 108.9968,
|
| 349 |
+
"eval_samples_per_second": 79.727,
|
| 350 |
+
"eval_steps_per_second": 2.495,
|
| 351 |
+
"step": 3600
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 9.349336702463676,
|
| 355 |
+
"grad_norm": 6.121856212615967,
|
| 356 |
+
"learning_rate": 0.00048703703703703707,
|
| 357 |
+
"loss": 0.3766,
|
| 358 |
+
"step": 3700
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 9.602021478205938,
|
| 362 |
+
"grad_norm": 3.0020909309387207,
|
| 363 |
+
"learning_rate": 0.0004851851851851852,
|
| 364 |
+
"loss": 0.3605,
|
| 365 |
+
"step": 3800
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 9.8547062539482,
|
| 369 |
+
"grad_norm": 3.841557502746582,
|
| 370 |
+
"learning_rate": 0.00048333333333333334,
|
| 371 |
+
"loss": 0.3802,
|
| 372 |
+
"step": 3900
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 10.107391029690461,
|
| 376 |
+
"grad_norm": 7.770158767700195,
|
| 377 |
+
"learning_rate": 0.00048148148148148144,
|
| 378 |
+
"loss": 0.3404,
|
| 379 |
+
"step": 4000
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 10.107391029690461,
|
| 383 |
+
"eval_accuracy": 0.7620253164556962,
|
| 384 |
+
"eval_f1_macro": 0.5823323463909736,
|
| 385 |
+
"eval_loss": 1.2837411165237427,
|
| 386 |
+
"eval_runtime": 108.7289,
|
| 387 |
+
"eval_samples_per_second": 79.924,
|
| 388 |
+
"eval_steps_per_second": 2.502,
|
| 389 |
+
"step": 4000
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 10.360075805432723,
|
| 393 |
+
"grad_norm": 4.366064071655273,
|
| 394 |
+
"learning_rate": 0.00047962962962962965,
|
| 395 |
+
"loss": 0.314,
|
| 396 |
+
"step": 4100
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 10.612760581174983,
|
| 400 |
+
"grad_norm": 9.094232559204102,
|
| 401 |
+
"learning_rate": 0.0004777777777777778,
|
| 402 |
+
"loss": 0.3174,
|
| 403 |
+
"step": 4200
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 10.865445356917245,
|
| 407 |
+
"grad_norm": 4.532145023345947,
|
| 408 |
+
"learning_rate": 0.0004759259259259259,
|
| 409 |
+
"loss": 0.3083,
|
| 410 |
+
"step": 4300
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 11.118130132659507,
|
| 414 |
+
"grad_norm": 2.8124845027923584,
|
| 415 |
+
"learning_rate": 0.0004740740740740741,
|
| 416 |
+
"loss": 0.28,
|
| 417 |
+
"step": 4400
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 11.118130132659507,
|
| 421 |
+
"eval_accuracy": 0.7566168009205984,
|
| 422 |
+
"eval_f1_macro": 0.5792237564159676,
|
| 423 |
+
"eval_loss": 1.4293328523635864,
|
| 424 |
+
"eval_runtime": 108.8172,
|
| 425 |
+
"eval_samples_per_second": 79.859,
|
| 426 |
+
"eval_steps_per_second": 2.5,
|
| 427 |
+
"step": 4400
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 11.37081490840177,
|
| 431 |
+
"grad_norm": 3.8208773136138916,
|
| 432 |
+
"learning_rate": 0.00047222222222222224,
|
| 433 |
+
"loss": 0.2614,
|
| 434 |
+
"step": 4500
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 11.623499684144031,
|
| 438 |
+
"grad_norm": 4.572866916656494,
|
| 439 |
+
"learning_rate": 0.0004703703703703704,
|
| 440 |
+
"loss": 0.2696,
|
| 441 |
+
"step": 4600
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 11.876184459886291,
|
| 445 |
+
"grad_norm": 3.239572763442993,
|
| 446 |
+
"learning_rate": 0.0004685185185185185,
|
| 447 |
+
"loss": 0.2691,
|
| 448 |
+
"step": 4700
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 12.128869235628553,
|
| 452 |
+
"grad_norm": 4.106109619140625,
|
| 453 |
+
"learning_rate": 0.00046666666666666666,
|
| 454 |
+
"loss": 0.2564,
|
| 455 |
+
"step": 4800
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 12.128869235628553,
|
| 459 |
+
"eval_accuracy": 0.7440736478711162,
|
| 460 |
+
"eval_f1_macro": 0.5907392176160607,
|
| 461 |
+
"eval_loss": 1.45040762424469,
|
| 462 |
+
"eval_runtime": 108.4161,
|
| 463 |
+
"eval_samples_per_second": 80.154,
|
| 464 |
+
"eval_steps_per_second": 2.509,
|
| 465 |
+
"step": 4800
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 12.381554011370815,
|
| 469 |
+
"grad_norm": 2.1584527492523193,
|
| 470 |
+
"learning_rate": 0.0004648148148148148,
|
| 471 |
+
"loss": 0.2412,
|
| 472 |
+
"step": 4900
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 12.634238787113077,
|
| 476 |
+
"grad_norm": 5.035179138183594,
|
| 477 |
+
"learning_rate": 0.000462962962962963,
|
| 478 |
+
"loss": 0.2345,
|
| 479 |
+
"step": 5000
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 12.886923562855339,
|
| 483 |
+
"grad_norm": 4.4065752029418945,
|
| 484 |
+
"learning_rate": 0.00046111111111111114,
|
| 485 |
+
"loss": 0.2373,
|
| 486 |
+
"step": 5100
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 13.139608338597599,
|
| 490 |
+
"grad_norm": 3.795588254928589,
|
| 491 |
+
"learning_rate": 0.00045925925925925925,
|
| 492 |
+
"loss": 0.2166,
|
| 493 |
+
"step": 5200
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 13.139608338597599,
|
| 497 |
+
"eval_accuracy": 0.7540851553509781,
|
| 498 |
+
"eval_f1_macro": 0.5753080086949651,
|
| 499 |
+
"eval_loss": 1.4374622106552124,
|
| 500 |
+
"eval_runtime": 108.3257,
|
| 501 |
+
"eval_samples_per_second": 80.221,
|
| 502 |
+
"eval_steps_per_second": 2.511,
|
| 503 |
+
"step": 5200
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 13.39229311433986,
|
| 507 |
+
"grad_norm": 3.1705026626586914,
|
| 508 |
+
"learning_rate": 0.00045740740740740746,
|
| 509 |
+
"loss": 0.2114,
|
| 510 |
+
"step": 5300
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 13.644977890082123,
|
| 514 |
+
"grad_norm": 5.109796047210693,
|
| 515 |
+
"learning_rate": 0.00045555555555555556,
|
| 516 |
+
"loss": 0.2207,
|
| 517 |
+
"step": 5400
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 13.897662665824384,
|
| 521 |
+
"grad_norm": 5.884572982788086,
|
| 522 |
+
"learning_rate": 0.0004537037037037037,
|
| 523 |
+
"loss": 0.2135,
|
| 524 |
+
"step": 5500
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 14.150347441566646,
|
| 528 |
+
"grad_norm": 4.0508246421813965,
|
| 529 |
+
"learning_rate": 0.00045185185185185183,
|
| 530 |
+
"loss": 0.2029,
|
| 531 |
+
"step": 5600
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 14.150347441566646,
|
| 535 |
+
"eval_accuracy": 0.7579976985040277,
|
| 536 |
+
"eval_f1_macro": 0.5636077065338464,
|
| 537 |
+
"eval_loss": 1.3845158815383911,
|
| 538 |
+
"eval_runtime": 108.3008,
|
| 539 |
+
"eval_samples_per_second": 80.239,
|
| 540 |
+
"eval_steps_per_second": 2.512,
|
| 541 |
+
"step": 5600
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 14.403032217308906,
|
| 545 |
+
"grad_norm": 3.048769235610962,
|
| 546 |
+
"learning_rate": 0.00045000000000000004,
|
| 547 |
+
"loss": 0.1842,
|
| 548 |
+
"step": 5700
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 14.655716993051168,
|
| 552 |
+
"grad_norm": 5.531657695770264,
|
| 553 |
+
"learning_rate": 0.00044814814814814815,
|
| 554 |
+
"loss": 0.1969,
|
| 555 |
+
"step": 5800
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 14.90840176879343,
|
| 559 |
+
"grad_norm": 4.874760150909424,
|
| 560 |
+
"learning_rate": 0.0004462962962962963,
|
| 561 |
+
"loss": 0.187,
|
| 562 |
+
"step": 5900
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 15.161086544535692,
|
| 566 |
+
"grad_norm": 4.56043815612793,
|
| 567 |
+
"learning_rate": 0.0004444444444444444,
|
| 568 |
+
"loss": 0.1781,
|
| 569 |
+
"step": 6000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 15.161086544535692,
|
| 573 |
+
"eval_accuracy": 0.7592635212888378,
|
| 574 |
+
"eval_f1_macro": 0.6140090678141539,
|
| 575 |
+
"eval_loss": 1.4697922468185425,
|
| 576 |
+
"eval_runtime": 108.1742,
|
| 577 |
+
"eval_samples_per_second": 80.333,
|
| 578 |
+
"eval_steps_per_second": 2.514,
|
| 579 |
+
"step": 6000
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 15.413771320277954,
|
| 583 |
+
"grad_norm": 2.3174920082092285,
|
| 584 |
+
"learning_rate": 0.0004425925925925926,
|
| 585 |
+
"loss": 0.1646,
|
| 586 |
+
"step": 6100
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 15.666456096020214,
|
| 590 |
+
"grad_norm": 6.532171726226807,
|
| 591 |
+
"learning_rate": 0.0004407407407407408,
|
| 592 |
+
"loss": 0.176,
|
| 593 |
+
"step": 6200
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 15.919140871762476,
|
| 597 |
+
"grad_norm": 8.349685668945312,
|
| 598 |
+
"learning_rate": 0.0004388888888888889,
|
| 599 |
+
"loss": 0.1819,
|
| 600 |
+
"step": 6300
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 16.171825647504736,
|
| 604 |
+
"grad_norm": 2.4073634147644043,
|
| 605 |
+
"learning_rate": 0.00043703703703703705,
|
| 606 |
+
"loss": 0.1559,
|
| 607 |
+
"step": 6400
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 16.171825647504736,
|
| 611 |
+
"eval_accuracy": 0.7581127733026467,
|
| 612 |
+
"eval_f1_macro": 0.6005969862926049,
|
| 613 |
+
"eval_loss": 1.6600337028503418,
|
| 614 |
+
"eval_runtime": 108.333,
|
| 615 |
+
"eval_samples_per_second": 80.216,
|
| 616 |
+
"eval_steps_per_second": 2.511,
|
| 617 |
+
"step": 6400
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 16.424510423247,
|
| 621 |
+
"grad_norm": 5.727371692657471,
|
| 622 |
+
"learning_rate": 0.0004351851851851852,
|
| 623 |
+
"loss": 0.1574,
|
| 624 |
+
"step": 6500
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 16.67719519898926,
|
| 628 |
+
"grad_norm": 5.15314245223999,
|
| 629 |
+
"learning_rate": 0.00043333333333333337,
|
| 630 |
+
"loss": 0.162,
|
| 631 |
+
"step": 6600
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 16.929879974731524,
|
| 635 |
+
"grad_norm": 4.066269397735596,
|
| 636 |
+
"learning_rate": 0.00043148148148148147,
|
| 637 |
+
"loss": 0.1541,
|
| 638 |
+
"step": 6700
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 17.182564750473784,
|
| 642 |
+
"grad_norm": 2.442532777786255,
|
| 643 |
+
"learning_rate": 0.00042962962962962963,
|
| 644 |
+
"loss": 0.1449,
|
| 645 |
+
"step": 6800
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 17.182564750473784,
|
| 649 |
+
"eval_accuracy": 0.7632911392405063,
|
| 650 |
+
"eval_f1_macro": 0.5902398414509258,
|
| 651 |
+
"eval_loss": 1.5675625801086426,
|
| 652 |
+
"eval_runtime": 108.2685,
|
| 653 |
+
"eval_samples_per_second": 80.263,
|
| 654 |
+
"eval_steps_per_second": 2.512,
|
| 655 |
+
"step": 6800
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 17.435249526216044,
|
| 659 |
+
"grad_norm": 2.888688564300537,
|
| 660 |
+
"learning_rate": 0.0004277777777777778,
|
| 661 |
+
"loss": 0.1602,
|
| 662 |
+
"step": 6900
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 17.687934301958308,
|
| 666 |
+
"grad_norm": 1.7030471563339233,
|
| 667 |
+
"learning_rate": 0.00042592592592592595,
|
| 668 |
+
"loss": 0.141,
|
| 669 |
+
"step": 7000
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 17.940619077700568,
|
| 673 |
+
"grad_norm": 3.838587760925293,
|
| 674 |
+
"learning_rate": 0.00042407407407407406,
|
| 675 |
+
"loss": 0.14,
|
| 676 |
+
"step": 7100
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 18.19330385344283,
|
| 680 |
+
"grad_norm": 2.5156667232513428,
|
| 681 |
+
"learning_rate": 0.0004222222222222222,
|
| 682 |
+
"loss": 0.144,
|
| 683 |
+
"step": 7200
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 18.19330385344283,
|
| 687 |
+
"eval_accuracy": 0.7498273878020714,
|
| 688 |
+
"eval_f1_macro": 0.5654132069159379,
|
| 689 |
+
"eval_loss": 1.4789820909500122,
|
| 690 |
+
"eval_runtime": 108.261,
|
| 691 |
+
"eval_samples_per_second": 80.269,
|
| 692 |
+
"eval_steps_per_second": 2.512,
|
| 693 |
+
"step": 7200
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 18.44598862918509,
|
| 697 |
+
"grad_norm": 4.657927513122559,
|
| 698 |
+
"learning_rate": 0.00042037037037037043,
|
| 699 |
+
"loss": 0.1385,
|
| 700 |
+
"step": 7300
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 18.69867340492735,
|
| 704 |
+
"grad_norm": 5.224093914031982,
|
| 705 |
+
"learning_rate": 0.00041851851851851853,
|
| 706 |
+
"loss": 0.1333,
|
| 707 |
+
"step": 7400
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 18.951358180669615,
|
| 711 |
+
"grad_norm": 2.436795949935913,
|
| 712 |
+
"learning_rate": 0.0004166666666666667,
|
| 713 |
+
"loss": 0.1333,
|
| 714 |
+
"step": 7500
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 19.204042956411875,
|
| 718 |
+
"grad_norm": 5.1690354347229,
|
| 719 |
+
"learning_rate": 0.0004148148148148148,
|
| 720 |
+
"loss": 0.1136,
|
| 721 |
+
"step": 7600
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 19.204042956411875,
|
| 725 |
+
"eval_accuracy": 0.7498273878020714,
|
| 726 |
+
"eval_f1_macro": 0.5531386036990584,
|
| 727 |
+
"eval_loss": 1.6063206195831299,
|
| 728 |
+
"eval_runtime": 108.3182,
|
| 729 |
+
"eval_samples_per_second": 80.227,
|
| 730 |
+
"eval_steps_per_second": 2.511,
|
| 731 |
+
"step": 7600
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 19.45672773215414,
|
| 735 |
+
"grad_norm": 1.983140230178833,
|
| 736 |
+
"learning_rate": 0.00041296296296296296,
|
| 737 |
+
"loss": 0.1265,
|
| 738 |
+
"step": 7700
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 19.7094125078964,
|
| 742 |
+
"grad_norm": 2.0450029373168945,
|
| 743 |
+
"learning_rate": 0.0004111111111111111,
|
| 744 |
+
"loss": 0.1221,
|
| 745 |
+
"step": 7800
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 19.96209728363866,
|
| 749 |
+
"grad_norm": 6.7355804443359375,
|
| 750 |
+
"learning_rate": 0.0004092592592592593,
|
| 751 |
+
"loss": 0.1369,
|
| 752 |
+
"step": 7900
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 20.214782059380923,
|
| 756 |
+
"grad_norm": 3.7778196334838867,
|
| 757 |
+
"learning_rate": 0.0004074074074074074,
|
| 758 |
+
"loss": 0.1082,
|
| 759 |
+
"step": 8000
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 20.214782059380923,
|
| 763 |
+
"eval_accuracy": 0.7598388952819333,
|
| 764 |
+
"eval_f1_macro": 0.5650987069075204,
|
| 765 |
+
"eval_loss": 1.6282719373703003,
|
| 766 |
+
"eval_runtime": 108.2582,
|
| 767 |
+
"eval_samples_per_second": 80.271,
|
| 768 |
+
"eval_steps_per_second": 2.513,
|
| 769 |
+
"step": 8000
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 20.467466835123183,
|
| 773 |
+
"grad_norm": 3.0025575160980225,
|
| 774 |
+
"learning_rate": 0.00040555555555555554,
|
| 775 |
+
"loss": 0.1108,
|
| 776 |
+
"step": 8100
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 20.720151610865447,
|
| 780 |
+
"grad_norm": 3.365360736846924,
|
| 781 |
+
"learning_rate": 0.00040370370370370375,
|
| 782 |
+
"loss": 0.1188,
|
| 783 |
+
"step": 8200
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 20.972836386607707,
|
| 787 |
+
"grad_norm": 1.7243456840515137,
|
| 788 |
+
"learning_rate": 0.00040185185185185186,
|
| 789 |
+
"loss": 0.1152,
|
| 790 |
+
"step": 8300
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 21.225521162349967,
|
| 794 |
+
"grad_norm": 2.3120861053466797,
|
| 795 |
+
"learning_rate": 0.0004,
|
| 796 |
+
"loss": 0.109,
|
| 797 |
+
"step": 8400
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 21.225521162349967,
|
| 801 |
+
"eval_accuracy": 0.7635212888377445,
|
| 802 |
+
"eval_f1_macro": 0.569418631056331,
|
| 803 |
+
"eval_loss": 1.752591848373413,
|
| 804 |
+
"eval_runtime": 108.3807,
|
| 805 |
+
"eval_samples_per_second": 80.18,
|
| 806 |
+
"eval_steps_per_second": 2.51,
|
| 807 |
+
"step": 8400
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 21.47820593809223,
|
| 811 |
+
"grad_norm": 3.04634165763855,
|
| 812 |
+
"learning_rate": 0.0003981481481481481,
|
| 813 |
+
"loss": 0.1219,
|
| 814 |
+
"step": 8500
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 21.73089071383449,
|
| 818 |
+
"grad_norm": 5.413991928100586,
|
| 819 |
+
"learning_rate": 0.00039629629629629634,
|
| 820 |
+
"loss": 0.1122,
|
| 821 |
+
"step": 8600
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 21.983575489576754,
|
| 825 |
+
"grad_norm": 3.94024395942688,
|
| 826 |
+
"learning_rate": 0.00039444444444444444,
|
| 827 |
+
"loss": 0.1023,
|
| 828 |
+
"step": 8700
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 22.236260265319014,
|
| 832 |
+
"grad_norm": 4.625360488891602,
|
| 833 |
+
"learning_rate": 0.0003925925925925926,
|
| 834 |
+
"loss": 0.1173,
|
| 835 |
+
"step": 8800
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 22.236260265319014,
|
| 839 |
+
"eval_accuracy": 0.7597238204833141,
|
| 840 |
+
"eval_f1_macro": 0.5837145389837423,
|
| 841 |
+
"eval_loss": 1.5915985107421875,
|
| 842 |
+
"eval_runtime": 108.3288,
|
| 843 |
+
"eval_samples_per_second": 80.219,
|
| 844 |
+
"eval_steps_per_second": 2.511,
|
| 845 |
+
"step": 8800
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 22.488945041061275,
|
| 849 |
+
"grad_norm": 4.926342010498047,
|
| 850 |
+
"learning_rate": 0.0003907407407407407,
|
| 851 |
+
"loss": 0.1059,
|
| 852 |
+
"step": 8900
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 22.74162981680354,
|
| 856 |
+
"grad_norm": 3.0098073482513428,
|
| 857 |
+
"learning_rate": 0.0003888888888888889,
|
| 858 |
+
"loss": 0.1021,
|
| 859 |
+
"step": 9000
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 22.9943145925458,
|
| 863 |
+
"grad_norm": 1.6922105550765991,
|
| 864 |
+
"learning_rate": 0.000387037037037037,
|
| 865 |
+
"loss": 0.1046,
|
| 866 |
+
"step": 9100
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 23.246999368288062,
|
| 870 |
+
"grad_norm": 4.048877239227295,
|
| 871 |
+
"learning_rate": 0.0003851851851851852,
|
| 872 |
+
"loss": 0.094,
|
| 873 |
+
"step": 9200
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 23.246999368288062,
|
| 877 |
+
"eval_accuracy": 0.7614499424626007,
|
| 878 |
+
"eval_f1_macro": 0.575035765001433,
|
| 879 |
+
"eval_loss": 1.631859302520752,
|
| 880 |
+
"eval_runtime": 108.7827,
|
| 881 |
+
"eval_samples_per_second": 79.884,
|
| 882 |
+
"eval_steps_per_second": 2.5,
|
| 883 |
+
"step": 9200
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 23.499684144030322,
|
| 887 |
+
"grad_norm": 3.0817179679870605,
|
| 888 |
+
"learning_rate": 0.00038333333333333334,
|
| 889 |
+
"loss": 0.1134,
|
| 890 |
+
"step": 9300
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 23.752368919772582,
|
| 894 |
+
"grad_norm": 0.7254201769828796,
|
| 895 |
+
"learning_rate": 0.0003814814814814815,
|
| 896 |
+
"loss": 0.0916,
|
| 897 |
+
"step": 9400
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 24.005053695514846,
|
| 901 |
+
"grad_norm": 1.1015946865081787,
|
| 902 |
+
"learning_rate": 0.00037962962962962966,
|
| 903 |
+
"loss": 0.1014,
|
| 904 |
+
"step": 9500
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 24.257738471257106,
|
| 908 |
+
"grad_norm": 1.6841648817062378,
|
| 909 |
+
"learning_rate": 0.00037777777777777777,
|
| 910 |
+
"loss": 0.0819,
|
| 911 |
+
"step": 9600
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 24.257738471257106,
|
| 915 |
+
"eval_accuracy": 0.7601841196777905,
|
| 916 |
+
"eval_f1_macro": 0.5741548033552556,
|
| 917 |
+
"eval_loss": 1.7696462869644165,
|
| 918 |
+
"eval_runtime": 108.7945,
|
| 919 |
+
"eval_samples_per_second": 79.875,
|
| 920 |
+
"eval_steps_per_second": 2.5,
|
| 921 |
+
"step": 9600
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 24.51042324699937,
|
| 925 |
+
"grad_norm": 1.9033466577529907,
|
| 926 |
+
"learning_rate": 0.00037592592592592593,
|
| 927 |
+
"loss": 0.1013,
|
| 928 |
+
"step": 9700
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 24.76310802274163,
|
| 932 |
+
"grad_norm": 3.788855791091919,
|
| 933 |
+
"learning_rate": 0.0003740740740740741,
|
| 934 |
+
"loss": 0.1057,
|
| 935 |
+
"step": 9800
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 25.01579279848389,
|
| 939 |
+
"grad_norm": 1.6506637334823608,
|
| 940 |
+
"learning_rate": 0.00037222222222222225,
|
| 941 |
+
"loss": 0.0979,
|
| 942 |
+
"step": 9900
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 25.268477574226154,
|
| 946 |
+
"grad_norm": 4.115843296051025,
|
| 947 |
+
"learning_rate": 0.00037037037037037035,
|
| 948 |
+
"loss": 0.0978,
|
| 949 |
+
"step": 10000
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 25.268477574226154,
|
| 953 |
+
"eval_accuracy": 0.7659378596087457,
|
| 954 |
+
"eval_f1_macro": 0.5667775496828193,
|
| 955 |
+
"eval_loss": 1.633931040763855,
|
| 956 |
+
"eval_runtime": 108.6433,
|
| 957 |
+
"eval_samples_per_second": 79.987,
|
| 958 |
+
"eval_steps_per_second": 2.504,
|
| 959 |
+
"step": 10000
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 25.521162349968414,
|
| 963 |
+
"grad_norm": 3.1308891773223877,
|
| 964 |
+
"learning_rate": 0.0003685185185185185,
|
| 965 |
+
"loss": 0.0804,
|
| 966 |
+
"step": 10100
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 25.773847125710677,
|
| 970 |
+
"grad_norm": 3.1427507400512695,
|
| 971 |
+
"learning_rate": 0.00036666666666666667,
|
| 972 |
+
"loss": 0.0861,
|
| 973 |
+
"step": 10200
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 26.026531901452937,
|
| 977 |
+
"grad_norm": 2.7073075771331787,
|
| 978 |
+
"learning_rate": 0.00036481481481481483,
|
| 979 |
+
"loss": 0.0846,
|
| 980 |
+
"step": 10300
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 26.279216677195198,
|
| 984 |
+
"grad_norm": 0.8588293194770813,
|
| 985 |
+
"learning_rate": 0.000362962962962963,
|
| 986 |
+
"loss": 0.0854,
|
| 987 |
+
"step": 10400
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 26.279216677195198,
|
| 991 |
+
"eval_accuracy": 0.7668584579976985,
|
| 992 |
+
"eval_f1_macro": 0.566287765137334,
|
| 993 |
+
"eval_loss": 1.6085623502731323,
|
| 994 |
+
"eval_runtime": 108.9617,
|
| 995 |
+
"eval_samples_per_second": 79.753,
|
| 996 |
+
"eval_steps_per_second": 2.496,
|
| 997 |
+
"step": 10400
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 26.53190145293746,
|
| 1001 |
+
"grad_norm": 2.541748285293579,
|
| 1002 |
+
"learning_rate": 0.0003611111111111111,
|
| 1003 |
+
"loss": 0.0828,
|
| 1004 |
+
"step": 10500
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 26.78458622867972,
|
| 1008 |
+
"grad_norm": 4.645153999328613,
|
| 1009 |
+
"learning_rate": 0.0003592592592592593,
|
| 1010 |
+
"loss": 0.0943,
|
| 1011 |
+
"step": 10600
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 27.037271004421985,
|
| 1015 |
+
"grad_norm": 2.1111841201782227,
|
| 1016 |
+
"learning_rate": 0.0003574074074074074,
|
| 1017 |
+
"loss": 0.0757,
|
| 1018 |
+
"step": 10700
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 27.289955780164245,
|
| 1022 |
+
"grad_norm": 1.5534864664077759,
|
| 1023 |
+
"learning_rate": 0.00035555555555555557,
|
| 1024 |
+
"loss": 0.0818,
|
| 1025 |
+
"step": 10800
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 27.289955780164245,
|
| 1029 |
+
"eval_accuracy": 0.768584579976985,
|
| 1030 |
+
"eval_f1_macro": 0.5967195650070578,
|
| 1031 |
+
"eval_loss": 1.6560544967651367,
|
| 1032 |
+
"eval_runtime": 108.8685,
|
| 1033 |
+
"eval_samples_per_second": 79.821,
|
| 1034 |
+
"eval_steps_per_second": 2.498,
|
| 1035 |
+
"step": 10800
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 27.542640555906505,
|
| 1039 |
+
"grad_norm": 1.9038456678390503,
|
| 1040 |
+
"learning_rate": 0.0003537037037037037,
|
| 1041 |
+
"loss": 0.081,
|
| 1042 |
+
"step": 10900
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 27.79532533164877,
|
| 1046 |
+
"grad_norm": 1.55673348903656,
|
| 1047 |
+
"learning_rate": 0.0003518518518518519,
|
| 1048 |
+
"loss": 0.0828,
|
| 1049 |
+
"step": 11000
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 28.04801010739103,
|
| 1053 |
+
"grad_norm": 5.855372905731201,
|
| 1054 |
+
"learning_rate": 0.00035,
|
| 1055 |
+
"loss": 0.0862,
|
| 1056 |
+
"step": 11100
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 28.300694883133293,
|
| 1060 |
+
"grad_norm": 3.111088991165161,
|
| 1061 |
+
"learning_rate": 0.00034814814814814816,
|
| 1062 |
+
"loss": 0.0725,
|
| 1063 |
+
"step": 11200
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 28.300694883133293,
|
| 1067 |
+
"eval_accuracy": 0.7542002301495973,
|
| 1068 |
+
"eval_f1_macro": 0.5740863312977772,
|
| 1069 |
+
"eval_loss": 1.7682907581329346,
|
| 1070 |
+
"eval_runtime": 144.2676,
|
| 1071 |
+
"eval_samples_per_second": 60.235,
|
| 1072 |
+
"eval_steps_per_second": 1.885,
|
| 1073 |
+
"step": 11200
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 28.553379658875553,
|
| 1077 |
+
"grad_norm": 1.5844389200210571,
|
| 1078 |
+
"learning_rate": 0.00034629629629629626,
|
| 1079 |
+
"loss": 0.0771,
|
| 1080 |
+
"step": 11300
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 28.806064434617813,
|
| 1084 |
+
"grad_norm": 3.93422269821167,
|
| 1085 |
+
"learning_rate": 0.0003444444444444445,
|
| 1086 |
+
"loss": 0.0863,
|
| 1087 |
+
"step": 11400
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 29.058749210360077,
|
| 1091 |
+
"grad_norm": 1.9465683698654175,
|
| 1092 |
+
"learning_rate": 0.00034259259259259263,
|
| 1093 |
+
"loss": 0.0777,
|
| 1094 |
+
"step": 11500
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 29.311433986102337,
|
| 1098 |
+
"grad_norm": 1.951295018196106,
|
| 1099 |
+
"learning_rate": 0.00034074074074074074,
|
| 1100 |
+
"loss": 0.0672,
|
| 1101 |
+
"step": 11600
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 29.311433986102337,
|
| 1105 |
+
"eval_accuracy": 0.7616800920598389,
|
| 1106 |
+
"eval_f1_macro": 0.5831099181645485,
|
| 1107 |
+
"eval_loss": 1.8726670742034912,
|
| 1108 |
+
"eval_runtime": 143.7286,
|
| 1109 |
+
"eval_samples_per_second": 60.461,
|
| 1110 |
+
"eval_steps_per_second": 1.892,
|
| 1111 |
+
"step": 11600
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 29.5641187618446,
|
| 1115 |
+
"grad_norm": 2.5593600273132324,
|
| 1116 |
+
"learning_rate": 0.0003388888888888889,
|
| 1117 |
+
"loss": 0.0707,
|
| 1118 |
+
"step": 11700
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 29.81680353758686,
|
| 1122 |
+
"grad_norm": 1.8622692823410034,
|
| 1123 |
+
"learning_rate": 0.00033703703703703706,
|
| 1124 |
+
"loss": 0.0709,
|
| 1125 |
+
"step": 11800
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 30.06948831332912,
|
| 1129 |
+
"grad_norm": 2.1602845191955566,
|
| 1130 |
+
"learning_rate": 0.0003351851851851852,
|
| 1131 |
+
"loss": 0.0653,
|
| 1132 |
+
"step": 11900
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 30.322173089071384,
|
| 1136 |
+
"grad_norm": 3.438899278640747,
|
| 1137 |
+
"learning_rate": 0.0003333333333333333,
|
| 1138 |
+
"loss": 0.076,
|
| 1139 |
+
"step": 12000
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 30.322173089071384,
|
| 1143 |
+
"eval_accuracy": 0.7635212888377445,
|
| 1144 |
+
"eval_f1_macro": 0.608710633839459,
|
| 1145 |
+
"eval_loss": 1.6385372877120972,
|
| 1146 |
+
"eval_runtime": 143.5758,
|
| 1147 |
+
"eval_samples_per_second": 60.526,
|
| 1148 |
+
"eval_steps_per_second": 1.894,
|
| 1149 |
+
"step": 12000
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 30.574857864813644,
|
| 1153 |
+
"grad_norm": 4.342224597930908,
|
| 1154 |
+
"learning_rate": 0.0003314814814814815,
|
| 1155 |
+
"loss": 0.0682,
|
| 1156 |
+
"step": 12100
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 30.827542640555908,
|
| 1160 |
+
"grad_norm": 3.186768054962158,
|
| 1161 |
+
"learning_rate": 0.0003296296296296296,
|
| 1162 |
+
"loss": 0.0705,
|
| 1163 |
+
"step": 12200
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 31.080227416298168,
|
| 1167 |
+
"grad_norm": 1.2158719301223755,
|
| 1168 |
+
"learning_rate": 0.0003277777777777778,
|
| 1169 |
+
"loss": 0.0665,
|
| 1170 |
+
"step": 12300
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 31.33291219204043,
|
| 1174 |
+
"grad_norm": 1.0089023113250732,
|
| 1175 |
+
"learning_rate": 0.00032592592592592596,
|
| 1176 |
+
"loss": 0.0702,
|
| 1177 |
+
"step": 12400
|
| 1178 |
+
},
|
| 1179 |
+
{
|
| 1180 |
+
"epoch": 31.33291219204043,
|
| 1181 |
+
"eval_accuracy": 0.7721518987341772,
|
| 1182 |
+
"eval_f1_macro": 0.5988474647551651,
|
| 1183 |
+
"eval_loss": 1.7596935033798218,
|
| 1184 |
+
"eval_runtime": 145.7714,
|
| 1185 |
+
"eval_samples_per_second": 59.614,
|
| 1186 |
+
"eval_steps_per_second": 1.866,
|
| 1187 |
+
"step": 12400
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 31.585596967782692,
|
| 1191 |
+
"grad_norm": 1.9418379068374634,
|
| 1192 |
+
"learning_rate": 0.00032407407407407406,
|
| 1193 |
+
"loss": 0.0674,
|
| 1194 |
+
"step": 12500
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 31.838281743524952,
|
| 1198 |
+
"grad_norm": 2.2405946254730225,
|
| 1199 |
+
"learning_rate": 0.0003222222222222222,
|
| 1200 |
+
"loss": 0.0645,
|
| 1201 |
+
"step": 12600
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 32.09096651926721,
|
| 1205 |
+
"grad_norm": 1.296802043914795,
|
| 1206 |
+
"learning_rate": 0.0003203703703703704,
|
| 1207 |
+
"loss": 0.0654,
|
| 1208 |
+
"step": 12700
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 32.34365129500947,
|
| 1212 |
+
"grad_norm": 5.712021827697754,
|
| 1213 |
+
"learning_rate": 0.00031851851851851854,
|
| 1214 |
+
"loss": 0.0644,
|
| 1215 |
+
"step": 12800
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 32.34365129500947,
|
| 1219 |
+
"eval_accuracy": 0.7686996547756041,
|
| 1220 |
+
"eval_f1_macro": 0.5840964451710614,
|
| 1221 |
+
"eval_loss": 1.795427918434143,
|
| 1222 |
+
"eval_runtime": 137.5518,
|
| 1223 |
+
"eval_samples_per_second": 63.176,
|
| 1224 |
+
"eval_steps_per_second": 1.977,
|
| 1225 |
+
"step": 12800
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 32.59633607075174,
|
| 1229 |
+
"grad_norm": 1.1909292936325073,
|
| 1230 |
+
"learning_rate": 0.00031666666666666665,
|
| 1231 |
+
"loss": 0.0638,
|
| 1232 |
+
"step": 12900
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 32.849020846494,
|
| 1236 |
+
"grad_norm": 4.957464694976807,
|
| 1237 |
+
"learning_rate": 0.0003148148148148148,
|
| 1238 |
+
"loss": 0.0566,
|
| 1239 |
+
"step": 13000
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 33.10170562223626,
|
| 1243 |
+
"grad_norm": 0.9219366908073425,
|
| 1244 |
+
"learning_rate": 0.00031296296296296297,
|
| 1245 |
+
"loss": 0.0622,
|
| 1246 |
+
"step": 13100
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 33.35439039797852,
|
| 1250 |
+
"grad_norm": 0.8782810568809509,
|
| 1251 |
+
"learning_rate": 0.0003111111111111111,
|
| 1252 |
+
"loss": 0.0521,
|
| 1253 |
+
"step": 13200
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 33.35439039797852,
|
| 1257 |
+
"eval_accuracy": 0.7619102416570771,
|
| 1258 |
+
"eval_f1_macro": 0.6192131512606447,
|
| 1259 |
+
"eval_loss": 1.901371955871582,
|
| 1260 |
+
"eval_runtime": 134.6686,
|
| 1261 |
+
"eval_samples_per_second": 64.529,
|
| 1262 |
+
"eval_steps_per_second": 2.02,
|
| 1263 |
+
"step": 13200
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 33.60707517372078,
|
| 1267 |
+
"grad_norm": 2.130397319793701,
|
| 1268 |
+
"learning_rate": 0.00030925925925925923,
|
| 1269 |
+
"loss": 0.0686,
|
| 1270 |
+
"step": 13300
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 33.85975994946305,
|
| 1274 |
+
"grad_norm": 1.2521294355392456,
|
| 1275 |
+
"learning_rate": 0.0003074074074074074,
|
| 1276 |
+
"loss": 0.058,
|
| 1277 |
+
"step": 13400
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 34.11244472520531,
|
| 1281 |
+
"grad_norm": 2.1233527660369873,
|
| 1282 |
+
"learning_rate": 0.0003055555555555556,
|
| 1283 |
+
"loss": 0.0528,
|
| 1284 |
+
"step": 13500
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 34.36512950094757,
|
| 1288 |
+
"grad_norm": 2.557140827178955,
|
| 1289 |
+
"learning_rate": 0.0003037037037037037,
|
| 1290 |
+
"loss": 0.0489,
|
| 1291 |
+
"step": 13600
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 34.36512950094757,
|
| 1295 |
+
"eval_accuracy": 0.768584579976985,
|
| 1296 |
+
"eval_f1_macro": 0.6008906251293721,
|
| 1297 |
+
"eval_loss": 1.8431710004806519,
|
| 1298 |
+
"eval_runtime": 109.9242,
|
| 1299 |
+
"eval_samples_per_second": 79.054,
|
| 1300 |
+
"eval_steps_per_second": 2.474,
|
| 1301 |
+
"step": 13600
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 34.61781427668983,
|
| 1305 |
+
"grad_norm": 1.0921064615249634,
|
| 1306 |
+
"learning_rate": 0.00030185185185185187,
|
| 1307 |
+
"loss": 0.0561,
|
| 1308 |
+
"step": 13700
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 34.87049905243209,
|
| 1312 |
+
"grad_norm": 2.349817991256714,
|
| 1313 |
+
"learning_rate": 0.0003,
|
| 1314 |
+
"loss": 0.0565,
|
| 1315 |
+
"step": 13800
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 35.123183828174355,
|
| 1319 |
+
"grad_norm": 1.187012791633606,
|
| 1320 |
+
"learning_rate": 0.0002981481481481482,
|
| 1321 |
+
"loss": 0.049,
|
| 1322 |
+
"step": 13900
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 35.375868603916615,
|
| 1326 |
+
"grad_norm": 0.04951922222971916,
|
| 1327 |
+
"learning_rate": 0.0002962962962962963,
|
| 1328 |
+
"loss": 0.0447,
|
| 1329 |
+
"step": 14000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 35.375868603916615,
|
| 1333 |
+
"eval_accuracy": 0.769620253164557,
|
| 1334 |
+
"eval_f1_macro": 0.5941725933407251,
|
| 1335 |
+
"eval_loss": 1.8310672044754028,
|
| 1336 |
+
"eval_runtime": 109.8974,
|
| 1337 |
+
"eval_samples_per_second": 79.074,
|
| 1338 |
+
"eval_steps_per_second": 2.475,
|
| 1339 |
+
"step": 14000
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 35.628553379658875,
|
| 1343 |
+
"grad_norm": 1.9226619005203247,
|
| 1344 |
+
"learning_rate": 0.00029444444444444445,
|
| 1345 |
+
"loss": 0.0575,
|
| 1346 |
+
"step": 14100
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 35.881238155401135,
|
| 1350 |
+
"grad_norm": 0.9691677093505859,
|
| 1351 |
+
"learning_rate": 0.00029259259259259256,
|
| 1352 |
+
"loss": 0.0569,
|
| 1353 |
+
"step": 14200
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 36.133922931143395,
|
| 1357 |
+
"grad_norm": 3.869662046432495,
|
| 1358 |
+
"learning_rate": 0.00029074074074074077,
|
| 1359 |
+
"loss": 0.0572,
|
| 1360 |
+
"step": 14300
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 36.38660770688566,
|
| 1364 |
+
"grad_norm": 1.026167869567871,
|
| 1365 |
+
"learning_rate": 0.0002888888888888889,
|
| 1366 |
+
"loss": 0.0506,
|
| 1367 |
+
"step": 14400
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 36.38660770688566,
|
| 1371 |
+
"eval_accuracy": 0.7692750287686997,
|
| 1372 |
+
"eval_f1_macro": 0.5692835911324542,
|
| 1373 |
+
"eval_loss": 1.7118504047393799,
|
| 1374 |
+
"eval_runtime": 110.0615,
|
| 1375 |
+
"eval_samples_per_second": 78.956,
|
| 1376 |
+
"eval_steps_per_second": 2.471,
|
| 1377 |
+
"step": 14400
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 36.63929248262792,
|
| 1381 |
+
"grad_norm": 0.6115164756774902,
|
| 1382 |
+
"learning_rate": 0.00028703703703703703,
|
| 1383 |
+
"loss": 0.0472,
|
| 1384 |
+
"step": 14500
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 36.89197725837018,
|
| 1388 |
+
"grad_norm": 2.26706862449646,
|
| 1389 |
+
"learning_rate": 0.0002851851851851852,
|
| 1390 |
+
"loss": 0.0507,
|
| 1391 |
+
"step": 14600
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 37.14466203411244,
|
| 1395 |
+
"grad_norm": 1.0484291315078735,
|
| 1396 |
+
"learning_rate": 0.00028333333333333335,
|
| 1397 |
+
"loss": 0.0533,
|
| 1398 |
+
"step": 14700
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 37.3973468098547,
|
| 1402 |
+
"grad_norm": 1.587339162826538,
|
| 1403 |
+
"learning_rate": 0.0002814814814814815,
|
| 1404 |
+
"loss": 0.0489,
|
| 1405 |
+
"step": 14800
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"epoch": 37.3973468098547,
|
| 1409 |
+
"eval_accuracy": 0.7665132336018412,
|
| 1410 |
+
"eval_f1_macro": 0.5648640266723668,
|
| 1411 |
+
"eval_loss": 1.8756378889083862,
|
| 1412 |
+
"eval_runtime": 109.745,
|
| 1413 |
+
"eval_samples_per_second": 79.184,
|
| 1414 |
+
"eval_steps_per_second": 2.478,
|
| 1415 |
+
"step": 14800
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 37.65003158559697,
|
| 1419 |
+
"grad_norm": 0.5741543173789978,
|
| 1420 |
+
"learning_rate": 0.0002796296296296296,
|
| 1421 |
+
"loss": 0.0539,
|
| 1422 |
+
"step": 14900
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 37.90271636133923,
|
| 1426 |
+
"grad_norm": 2.3613216876983643,
|
| 1427 |
+
"learning_rate": 0.0002777777777777778,
|
| 1428 |
+
"loss": 0.0601,
|
| 1429 |
+
"step": 15000
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 38.15540113708149,
|
| 1433 |
+
"grad_norm": 2.9896764755249023,
|
| 1434 |
+
"learning_rate": 0.00027592592592592594,
|
| 1435 |
+
"loss": 0.0441,
|
| 1436 |
+
"step": 15100
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 38.40808591282375,
|
| 1440 |
+
"grad_norm": 2.065026044845581,
|
| 1441 |
+
"learning_rate": 0.0002740740740740741,
|
| 1442 |
+
"loss": 0.0523,
|
| 1443 |
+
"step": 15200
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 38.40808591282375,
|
| 1447 |
+
"eval_accuracy": 0.777445339470656,
|
| 1448 |
+
"eval_f1_macro": 0.5952390760705957,
|
| 1449 |
+
"eval_loss": 1.6472355127334595,
|
| 1450 |
+
"eval_runtime": 109.7433,
|
| 1451 |
+
"eval_samples_per_second": 79.185,
|
| 1452 |
+
"eval_steps_per_second": 2.479,
|
| 1453 |
+
"step": 15200
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 38.66077068856601,
|
| 1457 |
+
"grad_norm": 0.533388078212738,
|
| 1458 |
+
"learning_rate": 0.0002722222222222222,
|
| 1459 |
+
"loss": 0.0465,
|
| 1460 |
+
"step": 15300
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 38.91345546430828,
|
| 1464 |
+
"grad_norm": 2.1074326038360596,
|
| 1465 |
+
"learning_rate": 0.00027037037037037036,
|
| 1466 |
+
"loss": 0.0398,
|
| 1467 |
+
"step": 15400
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 39.16614024005054,
|
| 1471 |
+
"grad_norm": 1.6379950046539307,
|
| 1472 |
+
"learning_rate": 0.0002685185185185186,
|
| 1473 |
+
"loss": 0.04,
|
| 1474 |
+
"step": 15500
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 39.4188250157928,
|
| 1478 |
+
"grad_norm": 1.9422637224197388,
|
| 1479 |
+
"learning_rate": 0.0002666666666666667,
|
| 1480 |
+
"loss": 0.0384,
|
| 1481 |
+
"step": 15600
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 39.4188250157928,
|
| 1485 |
+
"eval_accuracy": 0.7736478711162256,
|
| 1486 |
+
"eval_f1_macro": 0.6025598876629095,
|
| 1487 |
+
"eval_loss": 1.8879679441452026,
|
| 1488 |
+
"eval_runtime": 109.8175,
|
| 1489 |
+
"eval_samples_per_second": 79.131,
|
| 1490 |
+
"eval_steps_per_second": 2.477,
|
| 1491 |
+
"step": 15600
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 39.67150979153506,
|
| 1495 |
+
"grad_norm": 1.7022106647491455,
|
| 1496 |
+
"learning_rate": 0.00026481481481481484,
|
| 1497 |
+
"loss": 0.036,
|
| 1498 |
+
"step": 15700
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 39.92419456727732,
|
| 1502 |
+
"grad_norm": 1.5560113191604614,
|
| 1503 |
+
"learning_rate": 0.00026296296296296294,
|
| 1504 |
+
"loss": 0.0419,
|
| 1505 |
+
"step": 15800
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 40.176879343019586,
|
| 1509 |
+
"grad_norm": 3.473234176635742,
|
| 1510 |
+
"learning_rate": 0.00026111111111111116,
|
| 1511 |
+
"loss": 0.0375,
|
| 1512 |
+
"step": 15900
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"epoch": 40.429564118761846,
|
| 1516 |
+
"grad_norm": 1.8860403299331665,
|
| 1517 |
+
"learning_rate": 0.00025925925925925926,
|
| 1518 |
+
"loss": 0.0423,
|
| 1519 |
+
"step": 16000
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 40.429564118761846,
|
| 1523 |
+
"eval_accuracy": 0.775258918296893,
|
| 1524 |
+
"eval_f1_macro": 0.5970358418964591,
|
| 1525 |
+
"eval_loss": 1.8289481401443481,
|
| 1526 |
+
"eval_runtime": 109.6274,
|
| 1527 |
+
"eval_samples_per_second": 79.268,
|
| 1528 |
+
"eval_steps_per_second": 2.481,
|
| 1529 |
+
"step": 16000
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 40.682248894504106,
|
| 1533 |
+
"grad_norm": 3.3260364532470703,
|
| 1534 |
+
"learning_rate": 0.0002574074074074074,
|
| 1535 |
+
"loss": 0.0384,
|
| 1536 |
+
"step": 16100
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 40.934933670246366,
|
| 1540 |
+
"grad_norm": 0.306608110666275,
|
| 1541 |
+
"learning_rate": 0.00025555555555555553,
|
| 1542 |
+
"loss": 0.0485,
|
| 1543 |
+
"step": 16200
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 41.187618445988626,
|
| 1547 |
+
"grad_norm": 1.0209686756134033,
|
| 1548 |
+
"learning_rate": 0.00025370370370370374,
|
| 1549 |
+
"loss": 0.0388,
|
| 1550 |
+
"step": 16300
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 41.44030322173089,
|
| 1554 |
+
"grad_norm": 1.1874313354492188,
|
| 1555 |
+
"learning_rate": 0.00025185185185185185,
|
| 1556 |
+
"loss": 0.0421,
|
| 1557 |
+
"step": 16400
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 41.44030322173089,
|
| 1561 |
+
"eval_accuracy": 0.7693901035673187,
|
| 1562 |
+
"eval_f1_macro": 0.6018061729163716,
|
| 1563 |
+
"eval_loss": 1.6877504587173462,
|
| 1564 |
+
"eval_runtime": 110.175,
|
| 1565 |
+
"eval_samples_per_second": 78.874,
|
| 1566 |
+
"eval_steps_per_second": 2.469,
|
| 1567 |
+
"step": 16400
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 41.69298799747315,
|
| 1571 |
+
"grad_norm": 1.121667742729187,
|
| 1572 |
+
"learning_rate": 0.00025,
|
| 1573 |
+
"loss": 0.0382,
|
| 1574 |
+
"step": 16500
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 41.94567277321541,
|
| 1578 |
+
"grad_norm": 1.1757502555847168,
|
| 1579 |
+
"learning_rate": 0.00024814814814814816,
|
| 1580 |
+
"loss": 0.0397,
|
| 1581 |
+
"step": 16600
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 42.198357548957674,
|
| 1585 |
+
"grad_norm": 9.486854553222656,
|
| 1586 |
+
"learning_rate": 0.00024629629629629627,
|
| 1587 |
+
"loss": 0.0409,
|
| 1588 |
+
"step": 16700
|
| 1589 |
+
},
|
| 1590 |
+
{
|
| 1591 |
+
"epoch": 42.451042324699934,
|
| 1592 |
+
"grad_norm": 0.4830320179462433,
|
| 1593 |
+
"learning_rate": 0.00024444444444444443,
|
| 1594 |
+
"loss": 0.0404,
|
| 1595 |
+
"step": 16800
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 42.451042324699934,
|
| 1599 |
+
"eval_accuracy": 0.7681242807825086,
|
| 1600 |
+
"eval_f1_macro": 0.5853419424167555,
|
| 1601 |
+
"eval_loss": 1.8393748998641968,
|
| 1602 |
+
"eval_runtime": 109.7881,
|
| 1603 |
+
"eval_samples_per_second": 79.152,
|
| 1604 |
+
"eval_steps_per_second": 2.478,
|
| 1605 |
+
"step": 16800
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 42.7037271004422,
|
| 1609 |
+
"grad_norm": 1.6823731660842896,
|
| 1610 |
+
"learning_rate": 0.0002425925925925926,
|
| 1611 |
+
"loss": 0.0398,
|
| 1612 |
+
"step": 16900
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 42.95641187618446,
|
| 1616 |
+
"grad_norm": 6.2725749015808105,
|
| 1617 |
+
"learning_rate": 0.00024074074074074072,
|
| 1618 |
+
"loss": 0.041,
|
| 1619 |
+
"step": 17000
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 43.20909665192672,
|
| 1623 |
+
"grad_norm": 1.0358606576919556,
|
| 1624 |
+
"learning_rate": 0.0002388888888888889,
|
| 1625 |
+
"loss": 0.0359,
|
| 1626 |
+
"step": 17100
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 43.46178142766898,
|
| 1630 |
+
"grad_norm": 0.5175994634628296,
|
| 1631 |
+
"learning_rate": 0.00023703703703703704,
|
| 1632 |
+
"loss": 0.0311,
|
| 1633 |
+
"step": 17200
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 43.46178142766898,
|
| 1637 |
+
"eval_accuracy": 0.7772151898734178,
|
| 1638 |
+
"eval_f1_macro": 0.5963074285064315,
|
| 1639 |
+
"eval_loss": 1.7879575490951538,
|
| 1640 |
+
"eval_runtime": 109.7858,
|
| 1641 |
+
"eval_samples_per_second": 79.154,
|
| 1642 |
+
"eval_steps_per_second": 2.478,
|
| 1643 |
+
"step": 17200
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 43.71446620341124,
|
| 1647 |
+
"grad_norm": 3.763942003250122,
|
| 1648 |
+
"learning_rate": 0.0002351851851851852,
|
| 1649 |
+
"loss": 0.0283,
|
| 1650 |
+
"step": 17300
|
| 1651 |
+
},
|
| 1652 |
+
{
|
| 1653 |
+
"epoch": 43.96715097915351,
|
| 1654 |
+
"grad_norm": 0.7010844349861145,
|
| 1655 |
+
"learning_rate": 0.00023333333333333333,
|
| 1656 |
+
"loss": 0.0392,
|
| 1657 |
+
"step": 17400
|
| 1658 |
+
},
|
| 1659 |
+
{
|
| 1660 |
+
"epoch": 44.21983575489577,
|
| 1661 |
+
"grad_norm": 1.6681573390960693,
|
| 1662 |
+
"learning_rate": 0.0002314814814814815,
|
| 1663 |
+
"loss": 0.0306,
|
| 1664 |
+
"step": 17500
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 44.47252053063803,
|
| 1668 |
+
"grad_norm": 3.1810269355773926,
|
| 1669 |
+
"learning_rate": 0.00022962962962962962,
|
| 1670 |
+
"loss": 0.0421,
|
| 1671 |
+
"step": 17600
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 44.47252053063803,
|
| 1675 |
+
"eval_accuracy": 0.7758342922899885,
|
| 1676 |
+
"eval_f1_macro": 0.601629063986742,
|
| 1677 |
+
"eval_loss": 1.70499587059021,
|
| 1678 |
+
"eval_runtime": 109.8026,
|
| 1679 |
+
"eval_samples_per_second": 79.142,
|
| 1680 |
+
"eval_steps_per_second": 2.477,
|
| 1681 |
+
"step": 17600
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 44.72520530638029,
|
| 1685 |
+
"grad_norm": 1.0086809396743774,
|
| 1686 |
+
"learning_rate": 0.00022777777777777778,
|
| 1687 |
+
"loss": 0.0337,
|
| 1688 |
+
"step": 17700
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 44.97789008212255,
|
| 1692 |
+
"grad_norm": 0.7673502564430237,
|
| 1693 |
+
"learning_rate": 0.00022592592592592591,
|
| 1694 |
+
"loss": 0.0381,
|
| 1695 |
+
"step": 17800
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 45.230574857864816,
|
| 1699 |
+
"grad_norm": 0.9791069626808167,
|
| 1700 |
+
"learning_rate": 0.00022407407407407407,
|
| 1701 |
+
"loss": 0.0263,
|
| 1702 |
+
"step": 17900
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 45.48325963360708,
|
| 1706 |
+
"grad_norm": 2.0957231521606445,
|
| 1707 |
+
"learning_rate": 0.0002222222222222222,
|
| 1708 |
+
"loss": 0.039,
|
| 1709 |
+
"step": 18000
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 45.48325963360708,
|
| 1713 |
+
"eval_accuracy": 0.7800920598388953,
|
| 1714 |
+
"eval_f1_macro": 0.6025897002203272,
|
| 1715 |
+
"eval_loss": 1.7869104146957397,
|
| 1716 |
+
"eval_runtime": 109.5691,
|
| 1717 |
+
"eval_samples_per_second": 79.311,
|
| 1718 |
+
"eval_steps_per_second": 2.482,
|
| 1719 |
+
"step": 18000
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 45.73594440934934,
|
| 1723 |
+
"grad_norm": 2.0811707973480225,
|
| 1724 |
+
"learning_rate": 0.0002203703703703704,
|
| 1725 |
+
"loss": 0.0284,
|
| 1726 |
+
"step": 18100
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"epoch": 45.9886291850916,
|
| 1730 |
+
"grad_norm": 4.134197235107422,
|
| 1731 |
+
"learning_rate": 0.00021851851851851852,
|
| 1732 |
+
"loss": 0.0347,
|
| 1733 |
+
"step": 18200
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 46.24131396083386,
|
| 1737 |
+
"grad_norm": 0.5542005300521851,
|
| 1738 |
+
"learning_rate": 0.00021666666666666668,
|
| 1739 |
+
"loss": 0.0293,
|
| 1740 |
+
"step": 18300
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 46.493998736576124,
|
| 1744 |
+
"grad_norm": 2.4530539512634277,
|
| 1745 |
+
"learning_rate": 0.00021481481481481482,
|
| 1746 |
+
"loss": 0.0268,
|
| 1747 |
+
"step": 18400
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 46.493998736576124,
|
| 1751 |
+
"eval_accuracy": 0.7783659378596087,
|
| 1752 |
+
"eval_f1_macro": 0.5935536715539459,
|
| 1753 |
+
"eval_loss": 1.9122555255889893,
|
| 1754 |
+
"eval_runtime": 109.4826,
|
| 1755 |
+
"eval_samples_per_second": 79.373,
|
| 1756 |
+
"eval_steps_per_second": 2.484,
|
| 1757 |
+
"step": 18400
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 46.746683512318384,
|
| 1761 |
+
"grad_norm": 0.5376617312431335,
|
| 1762 |
+
"learning_rate": 0.00021296296296296298,
|
| 1763 |
+
"loss": 0.0348,
|
| 1764 |
+
"step": 18500
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 46.999368288060644,
|
| 1768 |
+
"grad_norm": 1.4315904378890991,
|
| 1769 |
+
"learning_rate": 0.0002111111111111111,
|
| 1770 |
+
"loss": 0.036,
|
| 1771 |
+
"step": 18600
|
| 1772 |
+
},
|
| 1773 |
+
{
|
| 1774 |
+
"epoch": 47.252053063802904,
|
| 1775 |
+
"grad_norm": 0.866862952709198,
|
| 1776 |
+
"learning_rate": 0.00020925925925925927,
|
| 1777 |
+
"loss": 0.0273,
|
| 1778 |
+
"step": 18700
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 47.504737839545164,
|
| 1782 |
+
"grad_norm": 2.0620691776275635,
|
| 1783 |
+
"learning_rate": 0.0002074074074074074,
|
| 1784 |
+
"loss": 0.0305,
|
| 1785 |
+
"step": 18800
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 47.504737839545164,
|
| 1789 |
+
"eval_accuracy": 0.7820483314154201,
|
| 1790 |
+
"eval_f1_macro": 0.5937346523730859,
|
| 1791 |
+
"eval_loss": 1.7703049182891846,
|
| 1792 |
+
"eval_runtime": 109.6053,
|
| 1793 |
+
"eval_samples_per_second": 79.284,
|
| 1794 |
+
"eval_steps_per_second": 2.482,
|
| 1795 |
+
"step": 18800
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 47.75742261528743,
|
| 1799 |
+
"grad_norm": 1.7660882472991943,
|
| 1800 |
+
"learning_rate": 0.00020555555555555556,
|
| 1801 |
+
"loss": 0.0248,
|
| 1802 |
+
"step": 18900
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 48.01010739102969,
|
| 1806 |
+
"grad_norm": 0.0962400734424591,
|
| 1807 |
+
"learning_rate": 0.0002037037037037037,
|
| 1808 |
+
"loss": 0.0316,
|
| 1809 |
+
"step": 19000
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 48.26279216677195,
|
| 1813 |
+
"grad_norm": 1.133728265762329,
|
| 1814 |
+
"learning_rate": 0.00020185185185185188,
|
| 1815 |
+
"loss": 0.0308,
|
| 1816 |
+
"step": 19100
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 48.51547694251421,
|
| 1820 |
+
"grad_norm": 0.4573793411254883,
|
| 1821 |
+
"learning_rate": 0.0002,
|
| 1822 |
+
"loss": 0.0279,
|
| 1823 |
+
"step": 19200
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 48.51547694251421,
|
| 1827 |
+
"eval_accuracy": 0.7795166858457998,
|
| 1828 |
+
"eval_f1_macro": 0.5972420776230559,
|
| 1829 |
+
"eval_loss": 1.6281923055648804,
|
| 1830 |
+
"eval_runtime": 109.5117,
|
| 1831 |
+
"eval_samples_per_second": 79.352,
|
| 1832 |
+
"eval_steps_per_second": 2.484,
|
| 1833 |
+
"step": 19200
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 48.76816171825647,
|
| 1837 |
+
"grad_norm": 1.396851658821106,
|
| 1838 |
+
"learning_rate": 0.00019814814814814817,
|
| 1839 |
+
"loss": 0.0216,
|
| 1840 |
+
"step": 19300
|
| 1841 |
+
},
|
| 1842 |
+
{
|
| 1843 |
+
"epoch": 49.02084649399874,
|
| 1844 |
+
"grad_norm": 1.0762335062026978,
|
| 1845 |
+
"learning_rate": 0.0001962962962962963,
|
| 1846 |
+
"loss": 0.0219,
|
| 1847 |
+
"step": 19400
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 49.273531269741,
|
| 1851 |
+
"grad_norm": 5.425212860107422,
|
| 1852 |
+
"learning_rate": 0.00019444444444444446,
|
| 1853 |
+
"loss": 0.0276,
|
| 1854 |
+
"step": 19500
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 49.52621604548326,
|
| 1858 |
+
"grad_norm": 0.31777116656303406,
|
| 1859 |
+
"learning_rate": 0.0001925925925925926,
|
| 1860 |
+
"loss": 0.0238,
|
| 1861 |
+
"step": 19600
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 49.52621604548326,
|
| 1865 |
+
"eval_accuracy": 0.768584579976985,
|
| 1866 |
+
"eval_f1_macro": 0.5921492404881852,
|
| 1867 |
+
"eval_loss": 1.8642910718917847,
|
| 1868 |
+
"eval_runtime": 109.7124,
|
| 1869 |
+
"eval_samples_per_second": 79.207,
|
| 1870 |
+
"eval_steps_per_second": 2.479,
|
| 1871 |
+
"step": 19600
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 49.77890082122552,
|
| 1875 |
+
"grad_norm": 2.751400947570801,
|
| 1876 |
+
"learning_rate": 0.00019074074074074075,
|
| 1877 |
+
"loss": 0.0271,
|
| 1878 |
+
"step": 19700
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 50.03158559696778,
|
| 1882 |
+
"grad_norm": 2.2741167545318604,
|
| 1883 |
+
"learning_rate": 0.00018888888888888888,
|
| 1884 |
+
"loss": 0.0238,
|
| 1885 |
+
"step": 19800
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 50.28427037271005,
|
| 1889 |
+
"grad_norm": 0.2512575387954712,
|
| 1890 |
+
"learning_rate": 0.00018703703703703704,
|
| 1891 |
+
"loss": 0.0288,
|
| 1892 |
+
"step": 19900
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 50.53695514845231,
|
| 1896 |
+
"grad_norm": 1.9498554468154907,
|
| 1897 |
+
"learning_rate": 0.00018518518518518518,
|
| 1898 |
+
"loss": 0.0243,
|
| 1899 |
+
"step": 20000
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 50.53695514845231,
|
| 1903 |
+
"eval_accuracy": 0.770771001150748,
|
| 1904 |
+
"eval_f1_macro": 0.598576194898749,
|
| 1905 |
+
"eval_loss": 1.976339340209961,
|
| 1906 |
+
"eval_runtime": 109.5523,
|
| 1907 |
+
"eval_samples_per_second": 79.323,
|
| 1908 |
+
"eval_steps_per_second": 2.483,
|
| 1909 |
+
"step": 20000
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 50.78963992419457,
|
| 1913 |
+
"grad_norm": 0.7483634948730469,
|
| 1914 |
+
"learning_rate": 0.00018333333333333334,
|
| 1915 |
+
"loss": 0.0253,
|
| 1916 |
+
"step": 20100
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 51.04232469993683,
|
| 1920 |
+
"grad_norm": 1.7610820531845093,
|
| 1921 |
+
"learning_rate": 0.0001814814814814815,
|
| 1922 |
+
"loss": 0.025,
|
| 1923 |
+
"step": 20200
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 51.29500947567909,
|
| 1927 |
+
"grad_norm": 2.2282676696777344,
|
| 1928 |
+
"learning_rate": 0.00017962962962962965,
|
| 1929 |
+
"loss": 0.0205,
|
| 1930 |
+
"step": 20300
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 51.547694251421355,
|
| 1934 |
+
"grad_norm": 1.0908398628234863,
|
| 1935 |
+
"learning_rate": 0.00017777777777777779,
|
| 1936 |
+
"loss": 0.0264,
|
| 1937 |
+
"step": 20400
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 51.547694251421355,
|
| 1941 |
+
"eval_accuracy": 0.7785960874568469,
|
| 1942 |
+
"eval_f1_macro": 0.6159147237572236,
|
| 1943 |
+
"eval_loss": 1.9196436405181885,
|
| 1944 |
+
"eval_runtime": 109.4823,
|
| 1945 |
+
"eval_samples_per_second": 79.374,
|
| 1946 |
+
"eval_steps_per_second": 2.484,
|
| 1947 |
+
"step": 20400
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 51.800379027163615,
|
| 1951 |
+
"grad_norm": 3.9686079025268555,
|
| 1952 |
+
"learning_rate": 0.00017592592592592595,
|
| 1953 |
+
"loss": 0.0215,
|
| 1954 |
+
"step": 20500
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 52.053063802905875,
|
| 1958 |
+
"grad_norm": 1.9715676307678223,
|
| 1959 |
+
"learning_rate": 0.00017407407407407408,
|
| 1960 |
+
"loss": 0.0247,
|
| 1961 |
+
"step": 20600
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 52.305748578648135,
|
| 1965 |
+
"grad_norm": 0.6495607495307922,
|
| 1966 |
+
"learning_rate": 0.00017222222222222224,
|
| 1967 |
+
"loss": 0.0249,
|
| 1968 |
+
"step": 20700
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 52.558433354390395,
|
| 1972 |
+
"grad_norm": 1.0535669326782227,
|
| 1973 |
+
"learning_rate": 0.00017037037037037037,
|
| 1974 |
+
"loss": 0.0236,
|
| 1975 |
+
"step": 20800
|
| 1976 |
+
},
|
| 1977 |
+
{
|
| 1978 |
+
"epoch": 52.558433354390395,
|
| 1979 |
+
"eval_accuracy": 0.7753739930955121,
|
| 1980 |
+
"eval_f1_macro": 0.5985595705467979,
|
| 1981 |
+
"eval_loss": 1.9067500829696655,
|
| 1982 |
+
"eval_runtime": 109.3981,
|
| 1983 |
+
"eval_samples_per_second": 79.435,
|
| 1984 |
+
"eval_steps_per_second": 2.486,
|
| 1985 |
+
"step": 20800
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 52.81111813013266,
|
| 1989 |
+
"grad_norm": 1.502255916595459,
|
| 1990 |
+
"learning_rate": 0.00016851851851851853,
|
| 1991 |
+
"loss": 0.018,
|
| 1992 |
+
"step": 20900
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 53.06380290587492,
|
| 1996 |
+
"grad_norm": 0.7225372791290283,
|
| 1997 |
+
"learning_rate": 0.00016666666666666666,
|
| 1998 |
+
"loss": 0.0238,
|
| 1999 |
+
"step": 21000
|
| 2000 |
+
},
|
| 2001 |
+
{
|
| 2002 |
+
"epoch": 53.31648768161718,
|
| 2003 |
+
"grad_norm": 1.2900406122207642,
|
| 2004 |
+
"learning_rate": 0.0001648148148148148,
|
| 2005 |
+
"loss": 0.0203,
|
| 2006 |
+
"step": 21100
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 53.56917245735944,
|
| 2010 |
+
"grad_norm": 1.0755679607391357,
|
| 2011 |
+
"learning_rate": 0.00016296296296296298,
|
| 2012 |
+
"loss": 0.0243,
|
| 2013 |
+
"step": 21200
|
| 2014 |
+
},
|
| 2015 |
+
{
|
| 2016 |
+
"epoch": 53.56917245735944,
|
| 2017 |
+
"eval_accuracy": 0.7788262370540852,
|
| 2018 |
+
"eval_f1_macro": 0.6106899497765403,
|
| 2019 |
+
"eval_loss": 1.8558076620101929,
|
| 2020 |
+
"eval_runtime": 109.5516,
|
| 2021 |
+
"eval_samples_per_second": 79.323,
|
| 2022 |
+
"eval_steps_per_second": 2.483,
|
| 2023 |
+
"step": 21200
|
| 2024 |
+
},
|
| 2025 |
+
{
|
| 2026 |
+
"epoch": 53.8218572331017,
|
| 2027 |
+
"grad_norm": 0.06415720283985138,
|
| 2028 |
+
"learning_rate": 0.0001611111111111111,
|
| 2029 |
+
"loss": 0.0205,
|
| 2030 |
+
"step": 21300
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 54.07454200884397,
|
| 2034 |
+
"grad_norm": 0.032455410808324814,
|
| 2035 |
+
"learning_rate": 0.00015925925925925927,
|
| 2036 |
+
"loss": 0.0221,
|
| 2037 |
+
"step": 21400
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 54.32722678458623,
|
| 2041 |
+
"grad_norm": 0.04551871493458748,
|
| 2042 |
+
"learning_rate": 0.0001574074074074074,
|
| 2043 |
+
"loss": 0.0173,
|
| 2044 |
+
"step": 21500
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"epoch": 54.57991156032849,
|
| 2048 |
+
"grad_norm": 1.3965240716934204,
|
| 2049 |
+
"learning_rate": 0.00015555555555555556,
|
| 2050 |
+
"loss": 0.0168,
|
| 2051 |
+
"step": 21600
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 54.57991156032849,
|
| 2055 |
+
"eval_accuracy": 0.7788262370540852,
|
| 2056 |
+
"eval_f1_macro": 0.5821705462809209,
|
| 2057 |
+
"eval_loss": 1.8485394716262817,
|
| 2058 |
+
"eval_runtime": 110.0893,
|
| 2059 |
+
"eval_samples_per_second": 78.936,
|
| 2060 |
+
"eval_steps_per_second": 2.471,
|
| 2061 |
+
"step": 21600
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 54.83259633607075,
|
| 2065 |
+
"grad_norm": 0.8048573136329651,
|
| 2066 |
+
"learning_rate": 0.0001537037037037037,
|
| 2067 |
+
"loss": 0.0212,
|
| 2068 |
+
"step": 21700
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 55.08528111181301,
|
| 2072 |
+
"grad_norm": 3.6472041606903076,
|
| 2073 |
+
"learning_rate": 0.00015185185185185185,
|
| 2074 |
+
"loss": 0.0229,
|
| 2075 |
+
"step": 21800
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 55.33796588755528,
|
| 2079 |
+
"grad_norm": 1.6580862998962402,
|
| 2080 |
+
"learning_rate": 0.00015,
|
| 2081 |
+
"loss": 0.0185,
|
| 2082 |
+
"step": 21900
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 55.59065066329754,
|
| 2086 |
+
"grad_norm": 0.0511002242565155,
|
| 2087 |
+
"learning_rate": 0.00014814814814814815,
|
| 2088 |
+
"loss": 0.019,
|
| 2089 |
+
"step": 22000
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 55.59065066329754,
|
| 2093 |
+
"eval_accuracy": 0.7772151898734178,
|
| 2094 |
+
"eval_f1_macro": 0.6042884739793327,
|
| 2095 |
+
"eval_loss": 1.8060083389282227,
|
| 2096 |
+
"eval_runtime": 109.6887,
|
| 2097 |
+
"eval_samples_per_second": 79.224,
|
| 2098 |
+
"eval_steps_per_second": 2.48,
|
| 2099 |
+
"step": 22000
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 55.8433354390398,
|
| 2103 |
+
"grad_norm": 0.7249517440795898,
|
| 2104 |
+
"learning_rate": 0.00014629629629629628,
|
| 2105 |
+
"loss": 0.022,
|
| 2106 |
+
"step": 22100
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 56.09602021478206,
|
| 2110 |
+
"grad_norm": 0.0043795122765004635,
|
| 2111 |
+
"learning_rate": 0.00014444444444444444,
|
| 2112 |
+
"loss": 0.0156,
|
| 2113 |
+
"step": 22200
|
| 2114 |
+
},
|
| 2115 |
+
{
|
| 2116 |
+
"epoch": 56.34870499052432,
|
| 2117 |
+
"grad_norm": 1.0969189405441284,
|
| 2118 |
+
"learning_rate": 0.0001425925925925926,
|
| 2119 |
+
"loss": 0.0186,
|
| 2120 |
+
"step": 22300
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 56.601389766266585,
|
| 2124 |
+
"grad_norm": 0.018035294488072395,
|
| 2125 |
+
"learning_rate": 0.00014074074074074076,
|
| 2126 |
+
"loss": 0.0193,
|
| 2127 |
+
"step": 22400
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 56.601389766266585,
|
| 2131 |
+
"eval_accuracy": 0.7785960874568469,
|
| 2132 |
+
"eval_f1_macro": 0.6042077712818668,
|
| 2133 |
+
"eval_loss": 1.7848457098007202,
|
| 2134 |
+
"eval_runtime": 109.4669,
|
| 2135 |
+
"eval_samples_per_second": 79.385,
|
| 2136 |
+
"eval_steps_per_second": 2.485,
|
| 2137 |
+
"step": 22400
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 56.854074542008846,
|
| 2141 |
+
"grad_norm": 1.3541306257247925,
|
| 2142 |
+
"learning_rate": 0.0001388888888888889,
|
| 2143 |
+
"loss": 0.014,
|
| 2144 |
+
"step": 22500
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 57.106759317751106,
|
| 2148 |
+
"grad_norm": 0.32108399271965027,
|
| 2149 |
+
"learning_rate": 0.00013703703703703705,
|
| 2150 |
+
"loss": 0.0147,
|
| 2151 |
+
"step": 22600
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 57.359444093493366,
|
| 2155 |
+
"grad_norm": 1.5764517784118652,
|
| 2156 |
+
"learning_rate": 0.00013518518518518518,
|
| 2157 |
+
"loss": 0.0121,
|
| 2158 |
+
"step": 22700
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 57.612128869235626,
|
| 2162 |
+
"grad_norm": 0.08151420950889587,
|
| 2163 |
+
"learning_rate": 0.00013333333333333334,
|
| 2164 |
+
"loss": 0.0167,
|
| 2165 |
+
"step": 22800
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 57.612128869235626,
|
| 2169 |
+
"eval_accuracy": 0.7812428078250863,
|
| 2170 |
+
"eval_f1_macro": 0.5890612998790621,
|
| 2171 |
+
"eval_loss": 1.9067610502243042,
|
| 2172 |
+
"eval_runtime": 109.8252,
|
| 2173 |
+
"eval_samples_per_second": 79.126,
|
| 2174 |
+
"eval_steps_per_second": 2.477,
|
| 2175 |
+
"step": 22800
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 57.86481364497789,
|
| 2179 |
+
"grad_norm": 0.2571534514427185,
|
| 2180 |
+
"learning_rate": 0.00013148148148148147,
|
| 2181 |
+
"loss": 0.0175,
|
| 2182 |
+
"step": 22900
|
| 2183 |
+
},
|
| 2184 |
+
{
|
| 2185 |
+
"epoch": 58.11749842072015,
|
| 2186 |
+
"grad_norm": 1.1256097555160522,
|
| 2187 |
+
"learning_rate": 0.00012962962962962963,
|
| 2188 |
+
"loss": 0.0185,
|
| 2189 |
+
"step": 23000
|
| 2190 |
+
},
|
| 2191 |
+
{
|
| 2192 |
+
"epoch": 58.37018319646241,
|
| 2193 |
+
"grad_norm": 0.036026712507009506,
|
| 2194 |
+
"learning_rate": 0.00012777777777777776,
|
| 2195 |
+
"loss": 0.0149,
|
| 2196 |
+
"step": 23100
|
| 2197 |
+
},
|
| 2198 |
+
{
|
| 2199 |
+
"epoch": 58.62286797220467,
|
| 2200 |
+
"grad_norm": 0.8259644508361816,
|
| 2201 |
+
"learning_rate": 0.00012592592592592592,
|
| 2202 |
+
"loss": 0.0103,
|
| 2203 |
+
"step": 23200
|
| 2204 |
+
},
|
| 2205 |
+
{
|
| 2206 |
+
"epoch": 58.62286797220467,
|
| 2207 |
+
"eval_accuracy": 0.7819332566168009,
|
| 2208 |
+
"eval_f1_macro": 0.5807141259596321,
|
| 2209 |
+
"eval_loss": 1.945181965827942,
|
| 2210 |
+
"eval_runtime": 109.5353,
|
| 2211 |
+
"eval_samples_per_second": 79.335,
|
| 2212 |
+
"eval_steps_per_second": 2.483,
|
| 2213 |
+
"step": 23200
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 58.875552747946934,
|
| 2217 |
+
"grad_norm": 0.3167104423046112,
|
| 2218 |
+
"learning_rate": 0.00012407407407407408,
|
| 2219 |
+
"loss": 0.0111,
|
| 2220 |
+
"step": 23300
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 59.1282375236892,
|
| 2224 |
+
"grad_norm": 0.19305191934108734,
|
| 2225 |
+
"learning_rate": 0.00012222222222222221,
|
| 2226 |
+
"loss": 0.0198,
|
| 2227 |
+
"step": 23400
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 59.38092229943146,
|
| 2231 |
+
"grad_norm": 0.9000421762466431,
|
| 2232 |
+
"learning_rate": 0.00012037037037037036,
|
| 2233 |
+
"loss": 0.0175,
|
| 2234 |
+
"step": 23500
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 59.63360707517372,
|
| 2238 |
+
"grad_norm": 0.16270488500595093,
|
| 2239 |
+
"learning_rate": 0.00011851851851851852,
|
| 2240 |
+
"loss": 0.0134,
|
| 2241 |
+
"step": 23600
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 59.63360707517372,
|
| 2245 |
+
"eval_accuracy": 0.7856156501726121,
|
| 2246 |
+
"eval_f1_macro": 0.6008830980384848,
|
| 2247 |
+
"eval_loss": 1.8185406923294067,
|
| 2248 |
+
"eval_runtime": 110.5509,
|
| 2249 |
+
"eval_samples_per_second": 78.606,
|
| 2250 |
+
"eval_steps_per_second": 2.46,
|
| 2251 |
+
"step": 23600
|
| 2252 |
+
},
|
| 2253 |
+
{
|
| 2254 |
+
"epoch": 59.88629185091598,
|
| 2255 |
+
"grad_norm": 0.15302228927612305,
|
| 2256 |
+
"learning_rate": 0.00011666666666666667,
|
| 2257 |
+
"loss": 0.0114,
|
| 2258 |
+
"step": 23700
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 60.13897662665824,
|
| 2262 |
+
"grad_norm": 0.07428427040576935,
|
| 2263 |
+
"learning_rate": 0.00011481481481481481,
|
| 2264 |
+
"loss": 0.0115,
|
| 2265 |
+
"step": 23800
|
| 2266 |
+
},
|
| 2267 |
+
{
|
| 2268 |
+
"epoch": 60.39166140240051,
|
| 2269 |
+
"grad_norm": 0.025211408734321594,
|
| 2270 |
+
"learning_rate": 0.00011296296296296296,
|
| 2271 |
+
"loss": 0.0129,
|
| 2272 |
+
"step": 23900
|
| 2273 |
+
},
|
| 2274 |
+
{
|
| 2275 |
+
"epoch": 60.64434617814277,
|
| 2276 |
+
"grad_norm": 0.12629875540733337,
|
| 2277 |
+
"learning_rate": 0.0001111111111111111,
|
| 2278 |
+
"loss": 0.0105,
|
| 2279 |
+
"step": 24000
|
| 2280 |
+
},
|
| 2281 |
+
{
|
| 2282 |
+
"epoch": 60.64434617814277,
|
| 2283 |
+
"eval_accuracy": 0.7771001150747986,
|
| 2284 |
+
"eval_f1_macro": 0.5844044138241861,
|
| 2285 |
+
"eval_loss": 1.9464218616485596,
|
| 2286 |
+
"eval_runtime": 109.5176,
|
| 2287 |
+
"eval_samples_per_second": 79.348,
|
| 2288 |
+
"eval_steps_per_second": 2.484,
|
| 2289 |
+
"step": 24000
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 60.89703095388503,
|
| 2293 |
+
"grad_norm": 0.08720462769269943,
|
| 2294 |
+
"learning_rate": 0.00010925925925925926,
|
| 2295 |
+
"loss": 0.0104,
|
| 2296 |
+
"step": 24100
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 61.14971572962729,
|
| 2300 |
+
"grad_norm": 2.3958306312561035,
|
| 2301 |
+
"learning_rate": 0.00010740740740740741,
|
| 2302 |
+
"loss": 0.011,
|
| 2303 |
+
"step": 24200
|
| 2304 |
+
},
|
| 2305 |
+
{
|
| 2306 |
+
"epoch": 61.40240050536955,
|
| 2307 |
+
"grad_norm": 0.0953073650598526,
|
| 2308 |
+
"learning_rate": 0.00010555555555555555,
|
| 2309 |
+
"loss": 0.0115,
|
| 2310 |
+
"step": 24300
|
| 2311 |
+
},
|
| 2312 |
+
{
|
| 2313 |
+
"epoch": 61.655085281111816,
|
| 2314 |
+
"grad_norm": 0.013684218749403954,
|
| 2315 |
+
"learning_rate": 0.0001037037037037037,
|
| 2316 |
+
"loss": 0.0154,
|
| 2317 |
+
"step": 24400
|
| 2318 |
+
},
|
| 2319 |
+
{
|
| 2320 |
+
"epoch": 61.655085281111816,
|
| 2321 |
+
"eval_accuracy": 0.7811277330264672,
|
| 2322 |
+
"eval_f1_macro": 0.5880759609293783,
|
| 2323 |
+
"eval_loss": 1.9170596599578857,
|
| 2324 |
+
"eval_runtime": 109.5265,
|
| 2325 |
+
"eval_samples_per_second": 79.342,
|
| 2326 |
+
"eval_steps_per_second": 2.483,
|
| 2327 |
+
"step": 24400
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 61.907770056854076,
|
| 2331 |
+
"grad_norm": 0.02879270352423191,
|
| 2332 |
+
"learning_rate": 0.00010185185185185185,
|
| 2333 |
+
"loss": 0.0133,
|
| 2334 |
+
"step": 24500
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 62.160454832596336,
|
| 2338 |
+
"grad_norm": 2.672358512878418,
|
| 2339 |
+
"learning_rate": 0.0001,
|
| 2340 |
+
"loss": 0.014,
|
| 2341 |
+
"step": 24600
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 62.4131396083386,
|
| 2345 |
+
"grad_norm": 0.22067667543888092,
|
| 2346 |
+
"learning_rate": 9.814814814814815e-05,
|
| 2347 |
+
"loss": 0.0152,
|
| 2348 |
+
"step": 24700
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 62.66582438408086,
|
| 2352 |
+
"grad_norm": 1.4633599519729614,
|
| 2353 |
+
"learning_rate": 9.62962962962963e-05,
|
| 2354 |
+
"loss": 0.0132,
|
| 2355 |
+
"step": 24800
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 62.66582438408086,
|
| 2359 |
+
"eval_accuracy": 0.7836593785960875,
|
| 2360 |
+
"eval_f1_macro": 0.5992199483923372,
|
| 2361 |
+
"eval_loss": 1.8896235227584839,
|
| 2362 |
+
"eval_runtime": 109.6805,
|
| 2363 |
+
"eval_samples_per_second": 79.23,
|
| 2364 |
+
"eval_steps_per_second": 2.48,
|
| 2365 |
+
"step": 24800
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 62.918509159823124,
|
| 2369 |
+
"grad_norm": 1.5682035684585571,
|
| 2370 |
+
"learning_rate": 9.444444444444444e-05,
|
| 2371 |
+
"loss": 0.0103,
|
| 2372 |
+
"step": 24900
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 63.171193935565384,
|
| 2376 |
+
"grad_norm": 0.010575806722044945,
|
| 2377 |
+
"learning_rate": 9.259259259259259e-05,
|
| 2378 |
+
"loss": 0.0099,
|
| 2379 |
+
"step": 25000
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 63.423878711307644,
|
| 2383 |
+
"grad_norm": 1.3311010599136353,
|
| 2384 |
+
"learning_rate": 9.074074074074075e-05,
|
| 2385 |
+
"loss": 0.0082,
|
| 2386 |
+
"step": 25100
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 63.676563487049904,
|
| 2390 |
+
"grad_norm": 0.13537296652793884,
|
| 2391 |
+
"learning_rate": 8.888888888888889e-05,
|
| 2392 |
+
"loss": 0.0059,
|
| 2393 |
+
"step": 25200
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 63.676563487049904,
|
| 2397 |
+
"eval_accuracy": 0.785385500575374,
|
| 2398 |
+
"eval_f1_macro": 0.5910071350304901,
|
| 2399 |
+
"eval_loss": 1.93367600440979,
|
| 2400 |
+
"eval_runtime": 109.8909,
|
| 2401 |
+
"eval_samples_per_second": 79.078,
|
| 2402 |
+
"eval_steps_per_second": 2.475,
|
| 2403 |
+
"step": 25200
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 63.929248262792164,
|
| 2407 |
+
"grad_norm": 0.13171835243701935,
|
| 2408 |
+
"learning_rate": 8.703703703703704e-05,
|
| 2409 |
+
"loss": 0.0099,
|
| 2410 |
+
"step": 25300
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 64.18193303853442,
|
| 2414 |
+
"grad_norm": 0.03450850397348404,
|
| 2415 |
+
"learning_rate": 8.518518518518518e-05,
|
| 2416 |
+
"loss": 0.0125,
|
| 2417 |
+
"step": 25400
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 64.43461781427669,
|
| 2421 |
+
"grad_norm": 2.559318780899048,
|
| 2422 |
+
"learning_rate": 8.333333333333333e-05,
|
| 2423 |
+
"loss": 0.0134,
|
| 2424 |
+
"step": 25500
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 64.68730259001894,
|
| 2428 |
+
"grad_norm": 0.4207484722137451,
|
| 2429 |
+
"learning_rate": 8.148148148148149e-05,
|
| 2430 |
+
"loss": 0.0082,
|
| 2431 |
+
"step": 25600
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 64.68730259001894,
|
| 2435 |
+
"eval_accuracy": 0.7844649021864212,
|
| 2436 |
+
"eval_f1_macro": 0.5981216790852985,
|
| 2437 |
+
"eval_loss": 1.9173579216003418,
|
| 2438 |
+
"eval_runtime": 109.5317,
|
| 2439 |
+
"eval_samples_per_second": 79.338,
|
| 2440 |
+
"eval_steps_per_second": 2.483,
|
| 2441 |
+
"step": 25600
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 64.93998736576121,
|
| 2445 |
+
"grad_norm": 0.02727402001619339,
|
| 2446 |
+
"learning_rate": 7.962962962962964e-05,
|
| 2447 |
+
"loss": 0.0079,
|
| 2448 |
+
"step": 25700
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 65.19267214150348,
|
| 2452 |
+
"grad_norm": 0.055505283176898956,
|
| 2453 |
+
"learning_rate": 7.777777777777778e-05,
|
| 2454 |
+
"loss": 0.0088,
|
| 2455 |
+
"step": 25800
|
| 2456 |
+
},
|
| 2457 |
+
{
|
| 2458 |
+
"epoch": 65.44535691724573,
|
| 2459 |
+
"grad_norm": 0.0938887745141983,
|
| 2460 |
+
"learning_rate": 7.592592592592593e-05,
|
| 2461 |
+
"loss": 0.0081,
|
| 2462 |
+
"step": 25900
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 65.698041692988,
|
| 2466 |
+
"grad_norm": 0.06274864077568054,
|
| 2467 |
+
"learning_rate": 7.407407407407407e-05,
|
| 2468 |
+
"loss": 0.01,
|
| 2469 |
+
"step": 26000
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 65.698041692988,
|
| 2473 |
+
"eval_accuracy": 0.7873417721518987,
|
| 2474 |
+
"eval_f1_macro": 0.5952116194605911,
|
| 2475 |
+
"eval_loss": 1.9076104164123535,
|
| 2476 |
+
"eval_runtime": 109.6326,
|
| 2477 |
+
"eval_samples_per_second": 79.265,
|
| 2478 |
+
"eval_steps_per_second": 2.481,
|
| 2479 |
+
"step": 26000
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 65.95072646873025,
|
| 2483 |
+
"grad_norm": 0.395120769739151,
|
| 2484 |
+
"learning_rate": 7.222222222222222e-05,
|
| 2485 |
+
"loss": 0.013,
|
| 2486 |
+
"step": 26100
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 66.20341124447252,
|
| 2490 |
+
"grad_norm": 0.0025830611120909452,
|
| 2491 |
+
"learning_rate": 7.037037037037038e-05,
|
| 2492 |
+
"loss": 0.0072,
|
| 2493 |
+
"step": 26200
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 66.45609602021479,
|
| 2497 |
+
"grad_norm": 0.08498067408800125,
|
| 2498 |
+
"learning_rate": 6.851851851851852e-05,
|
| 2499 |
+
"loss": 0.0092,
|
| 2500 |
+
"step": 26300
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 66.70878079595704,
|
| 2504 |
+
"grad_norm": 0.03530678525567055,
|
| 2505 |
+
"learning_rate": 6.666666666666667e-05,
|
| 2506 |
+
"loss": 0.0119,
|
| 2507 |
+
"step": 26400
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 66.70878079595704,
|
| 2511 |
+
"eval_accuracy": 0.7857307249712313,
|
| 2512 |
+
"eval_f1_macro": 0.6071046848722826,
|
| 2513 |
+
"eval_loss": 1.8940664529800415,
|
| 2514 |
+
"eval_runtime": 109.9317,
|
| 2515 |
+
"eval_samples_per_second": 79.049,
|
| 2516 |
+
"eval_steps_per_second": 2.474,
|
| 2517 |
+
"step": 26400
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 66.9614655716993,
|
| 2521 |
+
"grad_norm": 0.031329553574323654,
|
| 2522 |
+
"learning_rate": 6.481481481481482e-05,
|
| 2523 |
+
"loss": 0.0096,
|
| 2524 |
+
"step": 26500
|
| 2525 |
+
},
|
| 2526 |
+
{
|
| 2527 |
+
"epoch": 67.21415034744156,
|
| 2528 |
+
"grad_norm": 0.30799156427383423,
|
| 2529 |
+
"learning_rate": 6.296296296296296e-05,
|
| 2530 |
+
"loss": 0.0095,
|
| 2531 |
+
"step": 26600
|
| 2532 |
+
},
|
| 2533 |
+
{
|
| 2534 |
+
"epoch": 67.46683512318383,
|
| 2535 |
+
"grad_norm": 1.5470929145812988,
|
| 2536 |
+
"learning_rate": 6.111111111111111e-05,
|
| 2537 |
+
"loss": 0.0089,
|
| 2538 |
+
"step": 26700
|
| 2539 |
+
},
|
| 2540 |
+
{
|
| 2541 |
+
"epoch": 67.7195198989261,
|
| 2542 |
+
"grad_norm": 2.5448250770568848,
|
| 2543 |
+
"learning_rate": 5.925925925925926e-05,
|
| 2544 |
+
"loss": 0.0067,
|
| 2545 |
+
"step": 26800
|
| 2546 |
+
},
|
| 2547 |
+
{
|
| 2548 |
+
"epoch": 67.7195198989261,
|
| 2549 |
+
"eval_accuracy": 0.7878020713463751,
|
| 2550 |
+
"eval_f1_macro": 0.5999522395147027,
|
| 2551 |
+
"eval_loss": 1.930634617805481,
|
| 2552 |
+
"eval_runtime": 109.663,
|
| 2553 |
+
"eval_samples_per_second": 79.243,
|
| 2554 |
+
"eval_steps_per_second": 2.48,
|
| 2555 |
+
"step": 26800
|
| 2556 |
+
},
|
| 2557 |
+
{
|
| 2558 |
+
"epoch": 67.97220467466835,
|
| 2559 |
+
"grad_norm": 0.7930640578269958,
|
| 2560 |
+
"learning_rate": 5.7407407407407406e-05,
|
| 2561 |
+
"loss": 0.0069,
|
| 2562 |
+
"step": 26900
|
| 2563 |
+
},
|
| 2564 |
+
{
|
| 2565 |
+
"epoch": 68.22488945041061,
|
| 2566 |
+
"grad_norm": 0.26925787329673767,
|
| 2567 |
+
"learning_rate": 5.555555555555555e-05,
|
| 2568 |
+
"loss": 0.007,
|
| 2569 |
+
"step": 27000
|
| 2570 |
+
},
|
| 2571 |
+
{
|
| 2572 |
+
"epoch": 68.47757422615287,
|
| 2573 |
+
"grad_norm": 0.1341189444065094,
|
| 2574 |
+
"learning_rate": 5.3703703703703704e-05,
|
| 2575 |
+
"loss": 0.0087,
|
| 2576 |
+
"step": 27100
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 68.73025900189513,
|
| 2580 |
+
"grad_norm": 0.02547445334494114,
|
| 2581 |
+
"learning_rate": 5.185185185185185e-05,
|
| 2582 |
+
"loss": 0.0081,
|
| 2583 |
+
"step": 27200
|
| 2584 |
+
},
|
| 2585 |
+
{
|
| 2586 |
+
"epoch": 68.73025900189513,
|
| 2587 |
+
"eval_accuracy": 0.792174913693901,
|
| 2588 |
+
"eval_f1_macro": 0.6008077464320003,
|
| 2589 |
+
"eval_loss": 1.9338207244873047,
|
| 2590 |
+
"eval_runtime": 109.813,
|
| 2591 |
+
"eval_samples_per_second": 79.135,
|
| 2592 |
+
"eval_steps_per_second": 2.477,
|
| 2593 |
+
"step": 27200
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 68.9829437776374,
|
| 2597 |
+
"grad_norm": 0.03730254992842674,
|
| 2598 |
+
"learning_rate": 5e-05,
|
| 2599 |
+
"loss": 0.0065,
|
| 2600 |
+
"step": 27300
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 69.23562855337966,
|
| 2604 |
+
"grad_norm": 0.11829439550638199,
|
| 2605 |
+
"learning_rate": 4.814814814814815e-05,
|
| 2606 |
+
"loss": 0.0058,
|
| 2607 |
+
"step": 27400
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 69.48831332912192,
|
| 2611 |
+
"grad_norm": 1.238792896270752,
|
| 2612 |
+
"learning_rate": 4.6296296296296294e-05,
|
| 2613 |
+
"loss": 0.0047,
|
| 2614 |
+
"step": 27500
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 69.74099810486418,
|
| 2618 |
+
"grad_norm": 0.08776523917913437,
|
| 2619 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 2620 |
+
"loss": 0.0069,
|
| 2621 |
+
"step": 27600
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 69.74099810486418,
|
| 2625 |
+
"eval_accuracy": 0.7914844649021864,
|
| 2626 |
+
"eval_f1_macro": 0.5974683158925355,
|
| 2627 |
+
"eval_loss": 1.9561039209365845,
|
| 2628 |
+
"eval_runtime": 109.6013,
|
| 2629 |
+
"eval_samples_per_second": 79.287,
|
| 2630 |
+
"eval_steps_per_second": 2.482,
|
| 2631 |
+
"step": 27600
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 69.99368288060644,
|
| 2635 |
+
"grad_norm": 0.02279621548950672,
|
| 2636 |
+
"learning_rate": 4.259259259259259e-05,
|
| 2637 |
+
"loss": 0.0077,
|
| 2638 |
+
"step": 27700
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 70.24636765634871,
|
| 2642 |
+
"grad_norm": 1.407116413116455,
|
| 2643 |
+
"learning_rate": 4.0740740740740745e-05,
|
| 2644 |
+
"loss": 0.0083,
|
| 2645 |
+
"step": 27800
|
| 2646 |
+
},
|
| 2647 |
+
{
|
| 2648 |
+
"epoch": 70.49905243209096,
|
| 2649 |
+
"grad_norm": 0.8993849754333496,
|
| 2650 |
+
"learning_rate": 3.888888888888889e-05,
|
| 2651 |
+
"loss": 0.0037,
|
| 2652 |
+
"step": 27900
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"epoch": 70.75173720783323,
|
| 2656 |
+
"grad_norm": 0.01359980646520853,
|
| 2657 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 2658 |
+
"loss": 0.0037,
|
| 2659 |
+
"step": 28000
|
| 2660 |
+
},
|
| 2661 |
+
{
|
| 2662 |
+
"epoch": 70.75173720783323,
|
| 2663 |
+
"eval_accuracy": 0.7919447640966628,
|
| 2664 |
+
"eval_f1_macro": 0.5933817664756982,
|
| 2665 |
+
"eval_loss": 1.9411331415176392,
|
| 2666 |
+
"eval_runtime": 109.976,
|
| 2667 |
+
"eval_samples_per_second": 79.017,
|
| 2668 |
+
"eval_steps_per_second": 2.473,
|
| 2669 |
+
"step": 28000
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 71.00442198357548,
|
| 2673 |
+
"grad_norm": 0.00356766814365983,
|
| 2674 |
+
"learning_rate": 3.518518518518519e-05,
|
| 2675 |
+
"loss": 0.0053,
|
| 2676 |
+
"step": 28100
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 71.25710675931775,
|
| 2680 |
+
"grad_norm": 0.0005093297804705799,
|
| 2681 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 2682 |
+
"loss": 0.0046,
|
| 2683 |
+
"step": 28200
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 71.50979153506002,
|
| 2687 |
+
"grad_norm": 0.007542046718299389,
|
| 2688 |
+
"learning_rate": 3.148148148148148e-05,
|
| 2689 |
+
"loss": 0.0063,
|
| 2690 |
+
"step": 28300
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 71.76247631080227,
|
| 2694 |
+
"grad_norm": 0.008996374905109406,
|
| 2695 |
+
"learning_rate": 2.962962962962963e-05,
|
| 2696 |
+
"loss": 0.0055,
|
| 2697 |
+
"step": 28400
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 71.76247631080227,
|
| 2701 |
+
"eval_accuracy": 0.7947065592635213,
|
| 2702 |
+
"eval_f1_macro": 0.606388230425645,
|
| 2703 |
+
"eval_loss": 1.9448614120483398,
|
| 2704 |
+
"eval_runtime": 109.7807,
|
| 2705 |
+
"eval_samples_per_second": 79.158,
|
| 2706 |
+
"eval_steps_per_second": 2.478,
|
| 2707 |
+
"step": 28400
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 72.01516108654454,
|
| 2711 |
+
"grad_norm": 0.27878421545028687,
|
| 2712 |
+
"learning_rate": 2.7777777777777776e-05,
|
| 2713 |
+
"loss": 0.0046,
|
| 2714 |
+
"step": 28500
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 72.26784586228679,
|
| 2718 |
+
"grad_norm": 0.011219154112040997,
|
| 2719 |
+
"learning_rate": 2.5925925925925925e-05,
|
| 2720 |
+
"loss": 0.0067,
|
| 2721 |
+
"step": 28600
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 72.52053063802906,
|
| 2725 |
+
"grad_norm": 0.0006956435390748084,
|
| 2726 |
+
"learning_rate": 2.4074074074074074e-05,
|
| 2727 |
+
"loss": 0.0063,
|
| 2728 |
+
"step": 28700
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 72.77321541377133,
|
| 2732 |
+
"grad_norm": 0.07210740447044373,
|
| 2733 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 2734 |
+
"loss": 0.0046,
|
| 2735 |
+
"step": 28800
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 72.77321541377133,
|
| 2739 |
+
"eval_accuracy": 0.7941311852704258,
|
| 2740 |
+
"eval_f1_macro": 0.6069777778254928,
|
| 2741 |
+
"eval_loss": 1.9499226808547974,
|
| 2742 |
+
"eval_runtime": 109.6721,
|
| 2743 |
+
"eval_samples_per_second": 79.236,
|
| 2744 |
+
"eval_steps_per_second": 2.48,
|
| 2745 |
+
"step": 28800
|
| 2746 |
+
},
|
| 2747 |
+
{
|
| 2748 |
+
"epoch": 73.02590018951358,
|
| 2749 |
+
"grad_norm": 0.04884390905499458,
|
| 2750 |
+
"learning_rate": 2.0370370370370372e-05,
|
| 2751 |
+
"loss": 0.006,
|
| 2752 |
+
"step": 28900
|
| 2753 |
+
},
|
| 2754 |
+
{
|
| 2755 |
+
"epoch": 73.27858496525585,
|
| 2756 |
+
"grad_norm": 0.06649865210056305,
|
| 2757 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 2758 |
+
"loss": 0.0028,
|
| 2759 |
+
"step": 29000
|
| 2760 |
+
},
|
| 2761 |
+
{
|
| 2762 |
+
"epoch": 73.5312697409981,
|
| 2763 |
+
"grad_norm": 0.0004434076545294374,
|
| 2764 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 2765 |
+
"loss": 0.009,
|
| 2766 |
+
"step": 29100
|
| 2767 |
+
},
|
| 2768 |
+
{
|
| 2769 |
+
"epoch": 73.78395451674037,
|
| 2770 |
+
"grad_norm": 0.15433602035045624,
|
| 2771 |
+
"learning_rate": 1.4814814814814815e-05,
|
| 2772 |
+
"loss": 0.0063,
|
| 2773 |
+
"step": 29200
|
| 2774 |
+
},
|
| 2775 |
+
{
|
| 2776 |
+
"epoch": 73.78395451674037,
|
| 2777 |
+
"eval_accuracy": 0.7918296892980438,
|
| 2778 |
+
"eval_f1_macro": 0.5927699398802871,
|
| 2779 |
+
"eval_loss": 1.9630827903747559,
|
| 2780 |
+
"eval_runtime": 109.0221,
|
| 2781 |
+
"eval_samples_per_second": 79.709,
|
| 2782 |
+
"eval_steps_per_second": 2.495,
|
| 2783 |
+
"step": 29200
|
| 2784 |
+
},
|
| 2785 |
+
{
|
| 2786 |
+
"epoch": 74.03663929248263,
|
| 2787 |
+
"grad_norm": 0.0017094516661018133,
|
| 2788 |
+
"learning_rate": 1.2962962962962962e-05,
|
| 2789 |
+
"loss": 0.0046,
|
| 2790 |
+
"step": 29300
|
| 2791 |
+
},
|
| 2792 |
+
{
|
| 2793 |
+
"epoch": 74.28932406822489,
|
| 2794 |
+
"grad_norm": 1.173220157623291,
|
| 2795 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 2796 |
+
"loss": 0.0037,
|
| 2797 |
+
"step": 29400
|
| 2798 |
+
},
|
| 2799 |
+
{
|
| 2800 |
+
"epoch": 74.54200884396715,
|
| 2801 |
+
"grad_norm": 0.006894304417073727,
|
| 2802 |
+
"learning_rate": 9.259259259259259e-06,
|
| 2803 |
+
"loss": 0.0042,
|
| 2804 |
+
"step": 29500
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 74.7946936197094,
|
| 2808 |
+
"grad_norm": 0.005111688282340765,
|
| 2809 |
+
"learning_rate": 7.4074074074074075e-06,
|
| 2810 |
+
"loss": 0.0052,
|
| 2811 |
+
"step": 29600
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 74.7946936197094,
|
| 2815 |
+
"eval_accuracy": 0.7937859608745684,
|
| 2816 |
+
"eval_f1_macro": 0.5940671707873302,
|
| 2817 |
+
"eval_loss": 1.951813817024231,
|
| 2818 |
+
"eval_runtime": 108.972,
|
| 2819 |
+
"eval_samples_per_second": 79.745,
|
| 2820 |
+
"eval_steps_per_second": 2.496,
|
| 2821 |
+
"step": 29600
|
| 2822 |
+
},
|
| 2823 |
+
{
|
| 2824 |
+
"epoch": 75.04737839545167,
|
| 2825 |
+
"grad_norm": 0.0013843182241544127,
|
| 2826 |
+
"learning_rate": 5.555555555555556e-06,
|
| 2827 |
+
"loss": 0.0057,
|
| 2828 |
+
"step": 29700
|
| 2829 |
+
},
|
| 2830 |
+
{
|
| 2831 |
+
"epoch": 75.30006317119394,
|
| 2832 |
+
"grad_norm": 9.947911894414574e-05,
|
| 2833 |
+
"learning_rate": 3.7037037037037037e-06,
|
| 2834 |
+
"loss": 0.0052,
|
| 2835 |
+
"step": 29800
|
| 2836 |
+
},
|
| 2837 |
+
{
|
| 2838 |
+
"epoch": 75.5527479469362,
|
| 2839 |
+
"grad_norm": 0.3020813763141632,
|
| 2840 |
+
"learning_rate": 1.8518518518518519e-06,
|
| 2841 |
+
"loss": 0.006,
|
| 2842 |
+
"step": 29900
|
| 2843 |
+
},
|
| 2844 |
+
{
|
| 2845 |
+
"epoch": 75.80543272267846,
|
| 2846 |
+
"grad_norm": 1.3610678911209106,
|
| 2847 |
+
"learning_rate": 0.0,
|
| 2848 |
+
"loss": 0.0058,
|
| 2849 |
+
"step": 30000
|
| 2850 |
+
},
|
| 2851 |
+
{
|
| 2852 |
+
"epoch": 75.80543272267846,
|
| 2853 |
+
"eval_accuracy": 0.7934407364787112,
|
| 2854 |
+
"eval_f1_macro": 0.6028809095714917,
|
| 2855 |
+
"eval_loss": 1.9492535591125488,
|
| 2856 |
+
"eval_runtime": 108.7371,
|
| 2857 |
+
"eval_samples_per_second": 79.918,
|
| 2858 |
+
"eval_steps_per_second": 2.501,
|
| 2859 |
+
"step": 30000
|
| 2860 |
+
}
|
| 2861 |
+
],
|
| 2862 |
+
"logging_steps": 100,
|
| 2863 |
+
"max_steps": 30000,
|
| 2864 |
+
"num_input_tokens_seen": 0,
|
| 2865 |
+
"num_train_epochs": 76,
|
| 2866 |
+
"save_steps": 30000,
|
| 2867 |
+
"stateful_callbacks": {
|
| 2868 |
+
"TrainerControl": {
|
| 2869 |
+
"args": {
|
| 2870 |
+
"should_epoch_stop": false,
|
| 2871 |
+
"should_evaluate": false,
|
| 2872 |
+
"should_log": false,
|
| 2873 |
+
"should_save": true,
|
| 2874 |
+
"should_training_stop": true
|
| 2875 |
+
},
|
| 2876 |
+
"attributes": {}
|
| 2877 |
+
}
|
| 2878 |
+
},
|
| 2879 |
+
"total_flos": 5.8164789316384843e+20,
|
| 2880 |
+
"train_batch_size": 32,
|
| 2881 |
+
"trial_name": null,
|
| 2882 |
+
"trial_params": null
|
| 2883 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0b14532a009c641bdd0c8eb184939690a7f953fdf842b11ea7a30fd47ea759a9
|
| 3 |
+
size 5240
|