Upload 5 files
Browse files- checkpoint-1073/config.json +464 -0
- checkpoint-1073/rng_state.pth +3 -0
- checkpoint-1073/scheduler.pt +3 -0
- checkpoint-1073/trainer_state.json +1635 -0
- checkpoint-1073/training_args.bin +3 -0
checkpoint-1073/config.json
ADDED
|
@@ -0,0 +1,464 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "microsoft/swinv2-tiny-patch4-window16-256",
|
| 3 |
+
"_num_labels": 200,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"Swinv2ForImageClassification"
|
| 6 |
+
],
|
| 7 |
+
"attention_probs_dropout_prob": 0.0,
|
| 8 |
+
"depths": [
|
| 9 |
+
2,
|
| 10 |
+
2,
|
| 11 |
+
6,
|
| 12 |
+
2
|
| 13 |
+
],
|
| 14 |
+
"drop_path_rate": 0.1,
|
| 15 |
+
"embed_dim": 96,
|
| 16 |
+
"encoder_stride": 32,
|
| 17 |
+
"hidden_act": "gelu",
|
| 18 |
+
"hidden_dropout_prob": 0.0,
|
| 19 |
+
"hidden_size": 768,
|
| 20 |
+
"id2label": {
|
| 21 |
+
"0": "acadian_flycatcher",
|
| 22 |
+
"1": "american_crow",
|
| 23 |
+
"2": "american_goldfinch",
|
| 24 |
+
"3": "american_pipit",
|
| 25 |
+
"4": "american_redstart",
|
| 26 |
+
"5": "american_three_toed_woodpecker",
|
| 27 |
+
"6": "anna_hummingbird",
|
| 28 |
+
"7": "artic_tern",
|
| 29 |
+
"8": "baird_sparrow",
|
| 30 |
+
"9": "baltimore_oriole",
|
| 31 |
+
"10": "bank_swallow",
|
| 32 |
+
"11": "barn_swallow",
|
| 33 |
+
"12": "bay_breasted_warbler",
|
| 34 |
+
"13": "belted_kingfisher",
|
| 35 |
+
"14": "bewick_wren",
|
| 36 |
+
"15": "black_and_white_warbler",
|
| 37 |
+
"16": "black_billed_cuckoo",
|
| 38 |
+
"17": "black_capped_vireo",
|
| 39 |
+
"18": "black_footed_albatross",
|
| 40 |
+
"19": "black_tern",
|
| 41 |
+
"20": "black_throated_blue_warbler",
|
| 42 |
+
"21": "black_throated_sparrow",
|
| 43 |
+
"22": "blue_grosbeak",
|
| 44 |
+
"23": "blue_headed_vireo",
|
| 45 |
+
"24": "blue_jay",
|
| 46 |
+
"25": "blue_winged_warbler",
|
| 47 |
+
"26": "boat_tailed_grackle",
|
| 48 |
+
"27": "bobolink",
|
| 49 |
+
"28": "bohemian_waxwing",
|
| 50 |
+
"29": "brandt_cormorant",
|
| 51 |
+
"30": "brewer_blackbird",
|
| 52 |
+
"31": "brewer_sparrow",
|
| 53 |
+
"32": "bronzed_cowbird",
|
| 54 |
+
"33": "brown_creeper",
|
| 55 |
+
"34": "brown_pelican",
|
| 56 |
+
"35": "brown_thrasher",
|
| 57 |
+
"36": "cactus_wren",
|
| 58 |
+
"37": "california_gull",
|
| 59 |
+
"38": "canada_warbler",
|
| 60 |
+
"39": "cape_glossy_starling",
|
| 61 |
+
"40": "cape_may_warbler",
|
| 62 |
+
"41": "cardinal",
|
| 63 |
+
"42": "carolina_wren",
|
| 64 |
+
"43": "caspian_tern",
|
| 65 |
+
"44": "cedar_waxwing",
|
| 66 |
+
"45": "cerulean_warbler",
|
| 67 |
+
"46": "chestnut_sided_warbler",
|
| 68 |
+
"47": "chipping_sparrow",
|
| 69 |
+
"48": "chuck_will_widow",
|
| 70 |
+
"49": "clark_nutcracker",
|
| 71 |
+
"50": "clay_colored_sparrow",
|
| 72 |
+
"51": "cliff_swallow",
|
| 73 |
+
"52": "common_raven",
|
| 74 |
+
"53": "common_tern",
|
| 75 |
+
"54": "common_yellowthroat",
|
| 76 |
+
"55": "crested_auklet",
|
| 77 |
+
"56": "dark_eyed_junco",
|
| 78 |
+
"57": "downy_woodpecker",
|
| 79 |
+
"58": "eared_grebe",
|
| 80 |
+
"59": "eastern_towhee",
|
| 81 |
+
"60": "elegant_tern",
|
| 82 |
+
"61": "european_goldfinch",
|
| 83 |
+
"62": "evening_grosbeak",
|
| 84 |
+
"63": "field_sparrow",
|
| 85 |
+
"64": "fish_crow",
|
| 86 |
+
"65": "florida_jay",
|
| 87 |
+
"66": "forsters_tern",
|
| 88 |
+
"67": "fox_sparrow",
|
| 89 |
+
"68": "frigatebird",
|
| 90 |
+
"69": "gadwall",
|
| 91 |
+
"70": "geococcyx",
|
| 92 |
+
"71": "glaucous_winged_gull",
|
| 93 |
+
"72": "golden_winged_warbler",
|
| 94 |
+
"73": "grasshopper_sparrow",
|
| 95 |
+
"74": "gray_catbird",
|
| 96 |
+
"75": "gray_crowned_rosy_finch",
|
| 97 |
+
"76": "gray_kingbird",
|
| 98 |
+
"77": "great_crested_flycatcher",
|
| 99 |
+
"78": "great_grey_shrike",
|
| 100 |
+
"79": "green_jay",
|
| 101 |
+
"80": "green_kingfisher",
|
| 102 |
+
"81": "green_tailed_towhee",
|
| 103 |
+
"82": "green_violetear",
|
| 104 |
+
"83": "groove_billed_ani",
|
| 105 |
+
"84": "harris_sparrow",
|
| 106 |
+
"85": "heermann_gull",
|
| 107 |
+
"86": "henslow_sparrow",
|
| 108 |
+
"87": "herring_gull",
|
| 109 |
+
"88": "hooded_merganser",
|
| 110 |
+
"89": "hooded_oriole",
|
| 111 |
+
"90": "hooded_warbler",
|
| 112 |
+
"91": "horned_grebe",
|
| 113 |
+
"92": "horned_lark",
|
| 114 |
+
"93": "horned_puffin",
|
| 115 |
+
"94": "house_sparrow",
|
| 116 |
+
"95": "house_wren",
|
| 117 |
+
"96": "indigo_bunting",
|
| 118 |
+
"97": "ivory_gull",
|
| 119 |
+
"98": "kentucky_warbler",
|
| 120 |
+
"99": "laysan_albatross",
|
| 121 |
+
"100": "lazuli_bunting",
|
| 122 |
+
"101": "le_conte_sparrow",
|
| 123 |
+
"102": "least_auklet",
|
| 124 |
+
"103": "least_flycatcher",
|
| 125 |
+
"104": "least_tern",
|
| 126 |
+
"105": "lincoln_sparrow",
|
| 127 |
+
"106": "loggerhead_shrike",
|
| 128 |
+
"107": "long_tailed_jaeger",
|
| 129 |
+
"108": "louisiana_waterthrush",
|
| 130 |
+
"109": "magnolia_warbler",
|
| 131 |
+
"110": "mallard",
|
| 132 |
+
"111": "mangrove_cuckoo",
|
| 133 |
+
"112": "marsh_wren",
|
| 134 |
+
"113": "mockingbird",
|
| 135 |
+
"114": "mourning_warbler",
|
| 136 |
+
"115": "myrtle_warbler",
|
| 137 |
+
"116": "nashville_warbler",
|
| 138 |
+
"117": "nelson_sharp_tailed_sparrow",
|
| 139 |
+
"118": "nighthawk",
|
| 140 |
+
"119": "northern_flicker",
|
| 141 |
+
"120": "northern_fulmar",
|
| 142 |
+
"121": "northern_waterthrush",
|
| 143 |
+
"122": "olive_sided_flycatcher",
|
| 144 |
+
"123": "orange_crowned_warbler",
|
| 145 |
+
"124": "orchard_oriole",
|
| 146 |
+
"125": "ovenbird",
|
| 147 |
+
"126": "pacific_loon",
|
| 148 |
+
"127": "painted_bunting",
|
| 149 |
+
"128": "palm_warbler",
|
| 150 |
+
"129": "parakeet_auklet",
|
| 151 |
+
"130": "pelagic_cormorant",
|
| 152 |
+
"131": "philadelphia_vireo",
|
| 153 |
+
"132": "pied_billed_grebe",
|
| 154 |
+
"133": "pied_kingfisher",
|
| 155 |
+
"134": "pigeon_guillemot",
|
| 156 |
+
"135": "pileated_woodpecker",
|
| 157 |
+
"136": "pine_grosbeak",
|
| 158 |
+
"137": "pine_warbler",
|
| 159 |
+
"138": "pomarine_jaeger",
|
| 160 |
+
"139": "prairie_warbler",
|
| 161 |
+
"140": "prothonotary_warbler",
|
| 162 |
+
"141": "purple_finch",
|
| 163 |
+
"142": "red_bellied_woodpecker",
|
| 164 |
+
"143": "red_breasted_merganser",
|
| 165 |
+
"144": "red_cockaded_woodpecker",
|
| 166 |
+
"145": "red_eyed_vireo",
|
| 167 |
+
"146": "red_faced_cormorant",
|
| 168 |
+
"147": "red_headed_woodpecker",
|
| 169 |
+
"148": "red_legged_kittiwake",
|
| 170 |
+
"149": "red_winged_blackbird",
|
| 171 |
+
"150": "rhinoceros_auklet",
|
| 172 |
+
"151": "ring_billed_gull",
|
| 173 |
+
"152": "ringed_kingfisher",
|
| 174 |
+
"153": "rock_wren",
|
| 175 |
+
"154": "rose_breasted_grosbeak",
|
| 176 |
+
"155": "ruby_throated_hummingbird",
|
| 177 |
+
"156": "rufous_hummingbird",
|
| 178 |
+
"157": "rusty_blackbird",
|
| 179 |
+
"158": "sage_thrasher",
|
| 180 |
+
"159": "savannah_sparrow",
|
| 181 |
+
"160": "sayornis",
|
| 182 |
+
"161": "scarlet_tanager",
|
| 183 |
+
"162": "scissor_tailed_flycatcher",
|
| 184 |
+
"163": "scott_oriole",
|
| 185 |
+
"164": "seaside_sparrow",
|
| 186 |
+
"165": "shiny_cowbird",
|
| 187 |
+
"166": "slaty_backed_gull",
|
| 188 |
+
"167": "song_sparrow",
|
| 189 |
+
"168": "sooty_albatross",
|
| 190 |
+
"169": "spotted_catbird",
|
| 191 |
+
"170": "summer_tanager",
|
| 192 |
+
"171": "swainson_warbler",
|
| 193 |
+
"172": "tennessee_warbler",
|
| 194 |
+
"173": "tree_sparrow",
|
| 195 |
+
"174": "tree_swallow",
|
| 196 |
+
"175": "tropical_kingbird",
|
| 197 |
+
"176": "vermilion_flycatcher",
|
| 198 |
+
"177": "vesper_sparrow",
|
| 199 |
+
"178": "warbling_vireo",
|
| 200 |
+
"179": "western_grebe",
|
| 201 |
+
"180": "western_gull",
|
| 202 |
+
"181": "western_meadowlark",
|
| 203 |
+
"182": "western_wood_pewee",
|
| 204 |
+
"183": "whip_poor_will",
|
| 205 |
+
"184": "white_breasted_kingfisher",
|
| 206 |
+
"185": "white_breasted_nuthatch",
|
| 207 |
+
"186": "white_crowned_sparrow",
|
| 208 |
+
"187": "white_eyed_vireo",
|
| 209 |
+
"188": "white_necked_raven",
|
| 210 |
+
"189": "white_pelican",
|
| 211 |
+
"190": "white_throated_sparrow",
|
| 212 |
+
"191": "wilson_warbler",
|
| 213 |
+
"192": "winter_wren",
|
| 214 |
+
"193": "worm_eating_warbler",
|
| 215 |
+
"194": "yellow_bellied_flycatcher",
|
| 216 |
+
"195": "yellow_billed_cuckoo",
|
| 217 |
+
"196": "yellow_breasted_chat",
|
| 218 |
+
"197": "yellow_headed_blackbird",
|
| 219 |
+
"198": "yellow_throated_vireo",
|
| 220 |
+
"199": "yellow_warbler"
|
| 221 |
+
},
|
| 222 |
+
"image_size": 256,
|
| 223 |
+
"initializer_range": 0.02,
|
| 224 |
+
"label2id": {
|
| 225 |
+
"acadian_flycatcher": 0,
|
| 226 |
+
"american_crow": 1,
|
| 227 |
+
"american_goldfinch": 2,
|
| 228 |
+
"american_pipit": 3,
|
| 229 |
+
"american_redstart": 4,
|
| 230 |
+
"american_three_toed_woodpecker": 5,
|
| 231 |
+
"anna_hummingbird": 6,
|
| 232 |
+
"artic_tern": 7,
|
| 233 |
+
"baird_sparrow": 8,
|
| 234 |
+
"baltimore_oriole": 9,
|
| 235 |
+
"bank_swallow": 10,
|
| 236 |
+
"barn_swallow": 11,
|
| 237 |
+
"bay_breasted_warbler": 12,
|
| 238 |
+
"belted_kingfisher": 13,
|
| 239 |
+
"bewick_wren": 14,
|
| 240 |
+
"black_and_white_warbler": 15,
|
| 241 |
+
"black_billed_cuckoo": 16,
|
| 242 |
+
"black_capped_vireo": 17,
|
| 243 |
+
"black_footed_albatross": 18,
|
| 244 |
+
"black_tern": 19,
|
| 245 |
+
"black_throated_blue_warbler": 20,
|
| 246 |
+
"black_throated_sparrow": 21,
|
| 247 |
+
"blue_grosbeak": 22,
|
| 248 |
+
"blue_headed_vireo": 23,
|
| 249 |
+
"blue_jay": 24,
|
| 250 |
+
"blue_winged_warbler": 25,
|
| 251 |
+
"boat_tailed_grackle": 26,
|
| 252 |
+
"bobolink": 27,
|
| 253 |
+
"bohemian_waxwing": 28,
|
| 254 |
+
"brandt_cormorant": 29,
|
| 255 |
+
"brewer_blackbird": 30,
|
| 256 |
+
"brewer_sparrow": 31,
|
| 257 |
+
"bronzed_cowbird": 32,
|
| 258 |
+
"brown_creeper": 33,
|
| 259 |
+
"brown_pelican": 34,
|
| 260 |
+
"brown_thrasher": 35,
|
| 261 |
+
"cactus_wren": 36,
|
| 262 |
+
"california_gull": 37,
|
| 263 |
+
"canada_warbler": 38,
|
| 264 |
+
"cape_glossy_starling": 39,
|
| 265 |
+
"cape_may_warbler": 40,
|
| 266 |
+
"cardinal": 41,
|
| 267 |
+
"carolina_wren": 42,
|
| 268 |
+
"caspian_tern": 43,
|
| 269 |
+
"cedar_waxwing": 44,
|
| 270 |
+
"cerulean_warbler": 45,
|
| 271 |
+
"chestnut_sided_warbler": 46,
|
| 272 |
+
"chipping_sparrow": 47,
|
| 273 |
+
"chuck_will_widow": 48,
|
| 274 |
+
"clark_nutcracker": 49,
|
| 275 |
+
"clay_colored_sparrow": 50,
|
| 276 |
+
"cliff_swallow": 51,
|
| 277 |
+
"common_raven": 52,
|
| 278 |
+
"common_tern": 53,
|
| 279 |
+
"common_yellowthroat": 54,
|
| 280 |
+
"crested_auklet": 55,
|
| 281 |
+
"dark_eyed_junco": 56,
|
| 282 |
+
"downy_woodpecker": 57,
|
| 283 |
+
"eared_grebe": 58,
|
| 284 |
+
"eastern_towhee": 59,
|
| 285 |
+
"elegant_tern": 60,
|
| 286 |
+
"european_goldfinch": 61,
|
| 287 |
+
"evening_grosbeak": 62,
|
| 288 |
+
"field_sparrow": 63,
|
| 289 |
+
"fish_crow": 64,
|
| 290 |
+
"florida_jay": 65,
|
| 291 |
+
"forsters_tern": 66,
|
| 292 |
+
"fox_sparrow": 67,
|
| 293 |
+
"frigatebird": 68,
|
| 294 |
+
"gadwall": 69,
|
| 295 |
+
"geococcyx": 70,
|
| 296 |
+
"glaucous_winged_gull": 71,
|
| 297 |
+
"golden_winged_warbler": 72,
|
| 298 |
+
"grasshopper_sparrow": 73,
|
| 299 |
+
"gray_catbird": 74,
|
| 300 |
+
"gray_crowned_rosy_finch": 75,
|
| 301 |
+
"gray_kingbird": 76,
|
| 302 |
+
"great_crested_flycatcher": 77,
|
| 303 |
+
"great_grey_shrike": 78,
|
| 304 |
+
"green_jay": 79,
|
| 305 |
+
"green_kingfisher": 80,
|
| 306 |
+
"green_tailed_towhee": 81,
|
| 307 |
+
"green_violetear": 82,
|
| 308 |
+
"groove_billed_ani": 83,
|
| 309 |
+
"harris_sparrow": 84,
|
| 310 |
+
"heermann_gull": 85,
|
| 311 |
+
"henslow_sparrow": 86,
|
| 312 |
+
"herring_gull": 87,
|
| 313 |
+
"hooded_merganser": 88,
|
| 314 |
+
"hooded_oriole": 89,
|
| 315 |
+
"hooded_warbler": 90,
|
| 316 |
+
"horned_grebe": 91,
|
| 317 |
+
"horned_lark": 92,
|
| 318 |
+
"horned_puffin": 93,
|
| 319 |
+
"house_sparrow": 94,
|
| 320 |
+
"house_wren": 95,
|
| 321 |
+
"indigo_bunting": 96,
|
| 322 |
+
"ivory_gull": 97,
|
| 323 |
+
"kentucky_warbler": 98,
|
| 324 |
+
"laysan_albatross": 99,
|
| 325 |
+
"lazuli_bunting": 100,
|
| 326 |
+
"le_conte_sparrow": 101,
|
| 327 |
+
"least_auklet": 102,
|
| 328 |
+
"least_flycatcher": 103,
|
| 329 |
+
"least_tern": 104,
|
| 330 |
+
"lincoln_sparrow": 105,
|
| 331 |
+
"loggerhead_shrike": 106,
|
| 332 |
+
"long_tailed_jaeger": 107,
|
| 333 |
+
"louisiana_waterthrush": 108,
|
| 334 |
+
"magnolia_warbler": 109,
|
| 335 |
+
"mallard": 110,
|
| 336 |
+
"mangrove_cuckoo": 111,
|
| 337 |
+
"marsh_wren": 112,
|
| 338 |
+
"mockingbird": 113,
|
| 339 |
+
"mourning_warbler": 114,
|
| 340 |
+
"myrtle_warbler": 115,
|
| 341 |
+
"nashville_warbler": 116,
|
| 342 |
+
"nelson_sharp_tailed_sparrow": 117,
|
| 343 |
+
"nighthawk": 118,
|
| 344 |
+
"northern_flicker": 119,
|
| 345 |
+
"northern_fulmar": 120,
|
| 346 |
+
"northern_waterthrush": 121,
|
| 347 |
+
"olive_sided_flycatcher": 122,
|
| 348 |
+
"orange_crowned_warbler": 123,
|
| 349 |
+
"orchard_oriole": 124,
|
| 350 |
+
"ovenbird": 125,
|
| 351 |
+
"pacific_loon": 126,
|
| 352 |
+
"painted_bunting": 127,
|
| 353 |
+
"palm_warbler": 128,
|
| 354 |
+
"parakeet_auklet": 129,
|
| 355 |
+
"pelagic_cormorant": 130,
|
| 356 |
+
"philadelphia_vireo": 131,
|
| 357 |
+
"pied_billed_grebe": 132,
|
| 358 |
+
"pied_kingfisher": 133,
|
| 359 |
+
"pigeon_guillemot": 134,
|
| 360 |
+
"pileated_woodpecker": 135,
|
| 361 |
+
"pine_grosbeak": 136,
|
| 362 |
+
"pine_warbler": 137,
|
| 363 |
+
"pomarine_jaeger": 138,
|
| 364 |
+
"prairie_warbler": 139,
|
| 365 |
+
"prothonotary_warbler": 140,
|
| 366 |
+
"purple_finch": 141,
|
| 367 |
+
"red_bellied_woodpecker": 142,
|
| 368 |
+
"red_breasted_merganser": 143,
|
| 369 |
+
"red_cockaded_woodpecker": 144,
|
| 370 |
+
"red_eyed_vireo": 145,
|
| 371 |
+
"red_faced_cormorant": 146,
|
| 372 |
+
"red_headed_woodpecker": 147,
|
| 373 |
+
"red_legged_kittiwake": 148,
|
| 374 |
+
"red_winged_blackbird": 149,
|
| 375 |
+
"rhinoceros_auklet": 150,
|
| 376 |
+
"ring_billed_gull": 151,
|
| 377 |
+
"ringed_kingfisher": 152,
|
| 378 |
+
"rock_wren": 153,
|
| 379 |
+
"rose_breasted_grosbeak": 154,
|
| 380 |
+
"ruby_throated_hummingbird": 155,
|
| 381 |
+
"rufous_hummingbird": 156,
|
| 382 |
+
"rusty_blackbird": 157,
|
| 383 |
+
"sage_thrasher": 158,
|
| 384 |
+
"savannah_sparrow": 159,
|
| 385 |
+
"sayornis": 160,
|
| 386 |
+
"scarlet_tanager": 161,
|
| 387 |
+
"scissor_tailed_flycatcher": 162,
|
| 388 |
+
"scott_oriole": 163,
|
| 389 |
+
"seaside_sparrow": 164,
|
| 390 |
+
"shiny_cowbird": 165,
|
| 391 |
+
"slaty_backed_gull": 166,
|
| 392 |
+
"song_sparrow": 167,
|
| 393 |
+
"sooty_albatross": 168,
|
| 394 |
+
"spotted_catbird": 169,
|
| 395 |
+
"summer_tanager": 170,
|
| 396 |
+
"swainson_warbler": 171,
|
| 397 |
+
"tennessee_warbler": 172,
|
| 398 |
+
"tree_sparrow": 173,
|
| 399 |
+
"tree_swallow": 174,
|
| 400 |
+
"tropical_kingbird": 175,
|
| 401 |
+
"vermilion_flycatcher": 176,
|
| 402 |
+
"vesper_sparrow": 177,
|
| 403 |
+
"warbling_vireo": 178,
|
| 404 |
+
"western_grebe": 179,
|
| 405 |
+
"western_gull": 180,
|
| 406 |
+
"western_meadowlark": 181,
|
| 407 |
+
"western_wood_pewee": 182,
|
| 408 |
+
"whip_poor_will": 183,
|
| 409 |
+
"white_breasted_kingfisher": 184,
|
| 410 |
+
"white_breasted_nuthatch": 185,
|
| 411 |
+
"white_crowned_sparrow": 186,
|
| 412 |
+
"white_eyed_vireo": 187,
|
| 413 |
+
"white_necked_raven": 188,
|
| 414 |
+
"white_pelican": 189,
|
| 415 |
+
"white_throated_sparrow": 190,
|
| 416 |
+
"wilson_warbler": 191,
|
| 417 |
+
"winter_wren": 192,
|
| 418 |
+
"worm_eating_warbler": 193,
|
| 419 |
+
"yellow_bellied_flycatcher": 194,
|
| 420 |
+
"yellow_billed_cuckoo": 195,
|
| 421 |
+
"yellow_breasted_chat": 196,
|
| 422 |
+
"yellow_headed_blackbird": 197,
|
| 423 |
+
"yellow_throated_vireo": 198,
|
| 424 |
+
"yellow_warbler": 199
|
| 425 |
+
},
|
| 426 |
+
"layer_norm_eps": 1e-05,
|
| 427 |
+
"mlp_ratio": 4.0,
|
| 428 |
+
"model_type": "swinv2",
|
| 429 |
+
"num_channels": 3,
|
| 430 |
+
"num_heads": [
|
| 431 |
+
3,
|
| 432 |
+
6,
|
| 433 |
+
12,
|
| 434 |
+
24
|
| 435 |
+
],
|
| 436 |
+
"num_layers": 4,
|
| 437 |
+
"out_features": [
|
| 438 |
+
"stage4"
|
| 439 |
+
],
|
| 440 |
+
"out_indices": [
|
| 441 |
+
4
|
| 442 |
+
],
|
| 443 |
+
"patch_size": 4,
|
| 444 |
+
"path_norm": true,
|
| 445 |
+
"pretrained_window_sizes": [
|
| 446 |
+
0,
|
| 447 |
+
0,
|
| 448 |
+
0,
|
| 449 |
+
0
|
| 450 |
+
],
|
| 451 |
+
"problem_type": "single_label_classification",
|
| 452 |
+
"qkv_bias": true,
|
| 453 |
+
"stage_names": [
|
| 454 |
+
"stem",
|
| 455 |
+
"stage1",
|
| 456 |
+
"stage2",
|
| 457 |
+
"stage3",
|
| 458 |
+
"stage4"
|
| 459 |
+
],
|
| 460 |
+
"torch_dtype": "float32",
|
| 461 |
+
"transformers_version": "4.48.0",
|
| 462 |
+
"use_absolute_embeddings": false,
|
| 463 |
+
"window_size": 16
|
| 464 |
+
}
|
checkpoint-1073/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cf6032f9bf2e544ad0fa1c23268b7718baca4cfd7a23359d0be576a4a12ccd5d
|
| 3 |
+
size 14244
|
checkpoint-1073/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f4912bd4aeceb08125a644060a027fa0b78f1334d81957eeaf0166b5262c59f5
|
| 3 |
+
size 1064
|
checkpoint-1073/trainer_state.json
ADDED
|
@@ -0,0 +1,1635 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.5218645334243774,
|
| 3 |
+
"best_model_checkpoint": "Model-SwinV2-Tiny-\\checkpoint-1073",
|
| 4 |
+
"epoch": 28.980132450331126,
|
| 5 |
+
"eval_steps": 7,
|
| 6 |
+
"global_step": 1073,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.18543046357615894,
|
| 13 |
+
"grad_norm": 2.4086875915527344,
|
| 14 |
+
"learning_rate": 9.45945945945946e-07,
|
| 15 |
+
"loss": 5.3322,
|
| 16 |
+
"step": 7
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.3708609271523179,
|
| 20 |
+
"grad_norm": 2.1939220428466797,
|
| 21 |
+
"learning_rate": 1.891891891891892e-06,
|
| 22 |
+
"loss": 5.3175,
|
| 23 |
+
"step": 14
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.5562913907284768,
|
| 27 |
+
"grad_norm": 2.2086760997772217,
|
| 28 |
+
"learning_rate": 2.837837837837838e-06,
|
| 29 |
+
"loss": 5.3225,
|
| 30 |
+
"step": 21
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.7417218543046358,
|
| 34 |
+
"grad_norm": 2.065668821334839,
|
| 35 |
+
"learning_rate": 3.783783783783784e-06,
|
| 36 |
+
"loss": 5.3296,
|
| 37 |
+
"step": 28
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.9271523178807947,
|
| 41 |
+
"grad_norm": 2.3158862590789795,
|
| 42 |
+
"learning_rate": 4.72972972972973e-06,
|
| 43 |
+
"loss": 5.322,
|
| 44 |
+
"step": 35
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.9801324503311258,
|
| 48 |
+
"eval_accuracy": 0.003278688524590164,
|
| 49 |
+
"eval_f1_macro": 0.0016666666666666666,
|
| 50 |
+
"eval_f1_micro": 0.003278688524590164,
|
| 51 |
+
"eval_f1_weighted": 0.0013661202185792348,
|
| 52 |
+
"eval_loss": 5.303685188293457,
|
| 53 |
+
"eval_precision_macro": 0.0010971403137865806,
|
| 54 |
+
"eval_precision_micro": 0.003278688524590164,
|
| 55 |
+
"eval_precision_weighted": 0.0008992953391693285,
|
| 56 |
+
"eval_recall_macro": 0.004,
|
| 57 |
+
"eval_recall_micro": 0.003278688524590164,
|
| 58 |
+
"eval_recall_weighted": 0.003278688524590164,
|
| 59 |
+
"eval_runtime": 38.1558,
|
| 60 |
+
"eval_samples_per_second": 31.974,
|
| 61 |
+
"eval_steps_per_second": 0.524,
|
| 62 |
+
"step": 37
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 1.1324503311258278,
|
| 66 |
+
"grad_norm": 2.483227252960205,
|
| 67 |
+
"learning_rate": 5.675675675675676e-06,
|
| 68 |
+
"loss": 5.5043,
|
| 69 |
+
"step": 42
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 1.3178807947019868,
|
| 73 |
+
"grad_norm": 2.05220627784729,
|
| 74 |
+
"learning_rate": 6.621621621621622e-06,
|
| 75 |
+
"loss": 5.3074,
|
| 76 |
+
"step": 49
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 1.5033112582781456,
|
| 80 |
+
"grad_norm": 2.017322301864624,
|
| 81 |
+
"learning_rate": 7.567567567567568e-06,
|
| 82 |
+
"loss": 5.3025,
|
| 83 |
+
"step": 56
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 1.6887417218543046,
|
| 87 |
+
"grad_norm": 2.014375925064087,
|
| 88 |
+
"learning_rate": 8.513513513513514e-06,
|
| 89 |
+
"loss": 5.2928,
|
| 90 |
+
"step": 63
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.8741721854304636,
|
| 94 |
+
"grad_norm": 2.1785197257995605,
|
| 95 |
+
"learning_rate": 9.45945945945946e-06,
|
| 96 |
+
"loss": 5.2972,
|
| 97 |
+
"step": 70
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 1.980132450331126,
|
| 101 |
+
"eval_accuracy": 0.00819672131147541,
|
| 102 |
+
"eval_f1_macro": 0.004857079414838036,
|
| 103 |
+
"eval_f1_micro": 0.00819672131147541,
|
| 104 |
+
"eval_f1_weighted": 0.004755061925073231,
|
| 105 |
+
"eval_loss": 5.265906810760498,
|
| 106 |
+
"eval_precision_macro": 0.003902958152958153,
|
| 107 |
+
"eval_precision_micro": 0.00819672131147541,
|
| 108 |
+
"eval_precision_weighted": 0.003934248811297992,
|
| 109 |
+
"eval_recall_macro": 0.008630952380952382,
|
| 110 |
+
"eval_recall_micro": 0.00819672131147541,
|
| 111 |
+
"eval_recall_weighted": 0.00819672131147541,
|
| 112 |
+
"eval_runtime": 38.2173,
|
| 113 |
+
"eval_samples_per_second": 31.923,
|
| 114 |
+
"eval_steps_per_second": 0.523,
|
| 115 |
+
"step": 74
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 2.0794701986754967,
|
| 119 |
+
"grad_norm": 2.0930473804473877,
|
| 120 |
+
"learning_rate": 1.0405405405405407e-05,
|
| 121 |
+
"loss": 5.472,
|
| 122 |
+
"step": 77
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 2.2649006622516556,
|
| 126 |
+
"grad_norm": 2.3473684787750244,
|
| 127 |
+
"learning_rate": 1.1351351351351352e-05,
|
| 128 |
+
"loss": 5.2621,
|
| 129 |
+
"step": 84
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 2.4503311258278146,
|
| 133 |
+
"grad_norm": 2.268230676651001,
|
| 134 |
+
"learning_rate": 1.2297297297297299e-05,
|
| 135 |
+
"loss": 5.2555,
|
| 136 |
+
"step": 91
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 2.6357615894039736,
|
| 140 |
+
"grad_norm": 2.3846263885498047,
|
| 141 |
+
"learning_rate": 1.3243243243243244e-05,
|
| 142 |
+
"loss": 5.2442,
|
| 143 |
+
"step": 98
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 2.821192052980132,
|
| 147 |
+
"grad_norm": 2.452064037322998,
|
| 148 |
+
"learning_rate": 1.4189189189189189e-05,
|
| 149 |
+
"loss": 5.2228,
|
| 150 |
+
"step": 105
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 2.980132450331126,
|
| 154 |
+
"eval_accuracy": 0.030327868852459017,
|
| 155 |
+
"eval_f1_macro": 0.01780828862483212,
|
| 156 |
+
"eval_f1_micro": 0.030327868852459017,
|
| 157 |
+
"eval_f1_weighted": 0.018862280549754456,
|
| 158 |
+
"eval_loss": 5.171996593475342,
|
| 159 |
+
"eval_precision_macro": 0.01519444224935711,
|
| 160 |
+
"eval_precision_micro": 0.030327868852459017,
|
| 161 |
+
"eval_precision_weighted": 0.01609051329492379,
|
| 162 |
+
"eval_recall_macro": 0.028630952380952382,
|
| 163 |
+
"eval_recall_micro": 0.030327868852459017,
|
| 164 |
+
"eval_recall_weighted": 0.030327868852459017,
|
| 165 |
+
"eval_runtime": 40.0831,
|
| 166 |
+
"eval_samples_per_second": 30.437,
|
| 167 |
+
"eval_steps_per_second": 0.499,
|
| 168 |
+
"step": 111
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 3.0264900662251657,
|
| 172 |
+
"grad_norm": 3.297703981399536,
|
| 173 |
+
"learning_rate": 1.5135135135135136e-05,
|
| 174 |
+
"loss": 5.4031,
|
| 175 |
+
"step": 112
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 3.2119205298013247,
|
| 179 |
+
"grad_norm": 3.164050817489624,
|
| 180 |
+
"learning_rate": 1.6081081081081083e-05,
|
| 181 |
+
"loss": 5.1527,
|
| 182 |
+
"step": 119
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.3973509933774833,
|
| 186 |
+
"grad_norm": 3.7540459632873535,
|
| 187 |
+
"learning_rate": 1.7027027027027028e-05,
|
| 188 |
+
"loss": 5.1426,
|
| 189 |
+
"step": 126
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 3.5827814569536423,
|
| 193 |
+
"grad_norm": 4.779909610748291,
|
| 194 |
+
"learning_rate": 1.7972972972972973e-05,
|
| 195 |
+
"loss": 5.0713,
|
| 196 |
+
"step": 133
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 3.7682119205298013,
|
| 200 |
+
"grad_norm": 5.369205951690674,
|
| 201 |
+
"learning_rate": 1.891891891891892e-05,
|
| 202 |
+
"loss": 5.0499,
|
| 203 |
+
"step": 140
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 3.9536423841059603,
|
| 207 |
+
"grad_norm": 5.1240763664245605,
|
| 208 |
+
"learning_rate": 1.9864864864864866e-05,
|
| 209 |
+
"loss": 4.9613,
|
| 210 |
+
"step": 147
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 3.980132450331126,
|
| 214 |
+
"eval_accuracy": 0.08524590163934426,
|
| 215 |
+
"eval_f1_macro": 0.051001666629235665,
|
| 216 |
+
"eval_f1_micro": 0.08524590163934426,
|
| 217 |
+
"eval_f1_weighted": 0.054366012035255526,
|
| 218 |
+
"eval_loss": 4.813209056854248,
|
| 219 |
+
"eval_precision_macro": 0.05762040187775336,
|
| 220 |
+
"eval_precision_micro": 0.08524590163934426,
|
| 221 |
+
"eval_precision_weighted": 0.06070733267814678,
|
| 222 |
+
"eval_recall_macro": 0.07859523809523809,
|
| 223 |
+
"eval_recall_micro": 0.08524590163934426,
|
| 224 |
+
"eval_recall_weighted": 0.08524590163934426,
|
| 225 |
+
"eval_runtime": 58.6633,
|
| 226 |
+
"eval_samples_per_second": 20.797,
|
| 227 |
+
"eval_steps_per_second": 0.341,
|
| 228 |
+
"step": 148
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 4.158940397350993,
|
| 232 |
+
"grad_norm": 5.685731887817383,
|
| 233 |
+
"learning_rate": 2.0810810810810815e-05,
|
| 234 |
+
"loss": 5.0257,
|
| 235 |
+
"step": 154
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 4.344370860927152,
|
| 239 |
+
"grad_norm": 6.418041706085205,
|
| 240 |
+
"learning_rate": 2.1756756756756756e-05,
|
| 241 |
+
"loss": 4.7466,
|
| 242 |
+
"step": 161
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 4.529801324503311,
|
| 246 |
+
"grad_norm": 7.51582670211792,
|
| 247 |
+
"learning_rate": 2.2702702702702705e-05,
|
| 248 |
+
"loss": 4.6731,
|
| 249 |
+
"step": 168
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 4.71523178807947,
|
| 253 |
+
"grad_norm": 9.745444297790527,
|
| 254 |
+
"learning_rate": 2.364864864864865e-05,
|
| 255 |
+
"loss": 4.504,
|
| 256 |
+
"step": 175
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 4.900662251655629,
|
| 260 |
+
"grad_norm": 9.422097206115723,
|
| 261 |
+
"learning_rate": 2.4594594594594598e-05,
|
| 262 |
+
"loss": 4.3707,
|
| 263 |
+
"step": 182
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 4.9801324503311255,
|
| 267 |
+
"eval_accuracy": 0.21065573770491802,
|
| 268 |
+
"eval_f1_macro": 0.1669333623208855,
|
| 269 |
+
"eval_f1_micro": 0.21065573770491802,
|
| 270 |
+
"eval_f1_weighted": 0.1759151675061853,
|
| 271 |
+
"eval_loss": 3.8381166458129883,
|
| 272 |
+
"eval_precision_macro": 0.20365623267847385,
|
| 273 |
+
"eval_precision_micro": 0.21065573770491802,
|
| 274 |
+
"eval_precision_weighted": 0.20953591971339453,
|
| 275 |
+
"eval_recall_macro": 0.19770833333333332,
|
| 276 |
+
"eval_recall_micro": 0.21065573770491802,
|
| 277 |
+
"eval_recall_weighted": 0.21065573770491802,
|
| 278 |
+
"eval_runtime": 44.1331,
|
| 279 |
+
"eval_samples_per_second": 27.644,
|
| 280 |
+
"eval_steps_per_second": 0.453,
|
| 281 |
+
"step": 185
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 5.105960264900662,
|
| 285 |
+
"grad_norm": 9.648882865905762,
|
| 286 |
+
"learning_rate": 2.5540540540540543e-05,
|
| 287 |
+
"loss": 4.2878,
|
| 288 |
+
"step": 189
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 5.291390728476821,
|
| 292 |
+
"grad_norm": 11.289594650268555,
|
| 293 |
+
"learning_rate": 2.6486486486486488e-05,
|
| 294 |
+
"loss": 3.8794,
|
| 295 |
+
"step": 196
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 5.47682119205298,
|
| 299 |
+
"grad_norm": 10.757856369018555,
|
| 300 |
+
"learning_rate": 2.7297297297297298e-05,
|
| 301 |
+
"loss": 3.7505,
|
| 302 |
+
"step": 203
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 5.662251655629139,
|
| 306 |
+
"grad_norm": 16.969234466552734,
|
| 307 |
+
"learning_rate": 2.8243243243243246e-05,
|
| 308 |
+
"loss": 3.7323,
|
| 309 |
+
"step": 210
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 5.847682119205298,
|
| 313 |
+
"grad_norm": 11.33627700805664,
|
| 314 |
+
"learning_rate": 2.918918918918919e-05,
|
| 315 |
+
"loss": 3.5056,
|
| 316 |
+
"step": 217
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 5.9801324503311255,
|
| 320 |
+
"eval_accuracy": 0.37868852459016394,
|
| 321 |
+
"eval_f1_macro": 0.32624656613429687,
|
| 322 |
+
"eval_f1_micro": 0.37868852459016394,
|
| 323 |
+
"eval_f1_weighted": 0.3377910002445998,
|
| 324 |
+
"eval_loss": 2.865111827850342,
|
| 325 |
+
"eval_precision_macro": 0.37830690588089,
|
| 326 |
+
"eval_precision_micro": 0.37868852459016394,
|
| 327 |
+
"eval_precision_weighted": 0.38336785853552136,
|
| 328 |
+
"eval_recall_macro": 0.3615535714285714,
|
| 329 |
+
"eval_recall_micro": 0.37868852459016394,
|
| 330 |
+
"eval_recall_weighted": 0.37868852459016394,
|
| 331 |
+
"eval_runtime": 48.2126,
|
| 332 |
+
"eval_samples_per_second": 25.305,
|
| 333 |
+
"eval_steps_per_second": 0.415,
|
| 334 |
+
"step": 222
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 6.052980132450331,
|
| 338 |
+
"grad_norm": 11.259369850158691,
|
| 339 |
+
"learning_rate": 3e-05,
|
| 340 |
+
"loss": 3.6028,
|
| 341 |
+
"step": 224
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 6.23841059602649,
|
| 345 |
+
"grad_norm": 13.352005004882812,
|
| 346 |
+
"learning_rate": 3.0945945945945946e-05,
|
| 347 |
+
"loss": 3.1603,
|
| 348 |
+
"step": 231
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 6.423841059602649,
|
| 352 |
+
"grad_norm": 12.193765640258789,
|
| 353 |
+
"learning_rate": 3.1891891891891894e-05,
|
| 354 |
+
"loss": 3.0664,
|
| 355 |
+
"step": 238
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 6.609271523178808,
|
| 359 |
+
"grad_norm": 12.009160995483398,
|
| 360 |
+
"learning_rate": 3.2837837837837836e-05,
|
| 361 |
+
"loss": 2.8827,
|
| 362 |
+
"step": 245
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 6.7947019867549665,
|
| 366 |
+
"grad_norm": 11.383013725280762,
|
| 367 |
+
"learning_rate": 3.3783783783783784e-05,
|
| 368 |
+
"loss": 2.8152,
|
| 369 |
+
"step": 252
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 6.9801324503311255,
|
| 373 |
+
"grad_norm": 14.499237060546875,
|
| 374 |
+
"learning_rate": 3.472972972972973e-05,
|
| 375 |
+
"loss": 2.8677,
|
| 376 |
+
"step": 259
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 6.9801324503311255,
|
| 380 |
+
"eval_accuracy": 0.5237704918032787,
|
| 381 |
+
"eval_f1_macro": 0.4860099386723001,
|
| 382 |
+
"eval_f1_micro": 0.5237704918032787,
|
| 383 |
+
"eval_f1_weighted": 0.497629485231393,
|
| 384 |
+
"eval_loss": 2.090391159057617,
|
| 385 |
+
"eval_precision_macro": 0.5416790735494296,
|
| 386 |
+
"eval_precision_micro": 0.5237704918032787,
|
| 387 |
+
"eval_precision_weighted": 0.5489188118847942,
|
| 388 |
+
"eval_recall_macro": 0.5090833333333333,
|
| 389 |
+
"eval_recall_micro": 0.5237704918032787,
|
| 390 |
+
"eval_recall_weighted": 0.5237704918032787,
|
| 391 |
+
"eval_runtime": 45.6118,
|
| 392 |
+
"eval_samples_per_second": 26.747,
|
| 393 |
+
"eval_steps_per_second": 0.438,
|
| 394 |
+
"step": 259
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 7.185430463576159,
|
| 398 |
+
"grad_norm": 12.168959617614746,
|
| 399 |
+
"learning_rate": 3.567567567567568e-05,
|
| 400 |
+
"loss": 2.7083,
|
| 401 |
+
"step": 266
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 7.370860927152318,
|
| 405 |
+
"grad_norm": 11.984298706054688,
|
| 406 |
+
"learning_rate": 3.662162162162162e-05,
|
| 407 |
+
"loss": 2.552,
|
| 408 |
+
"step": 273
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 7.556291390728477,
|
| 412 |
+
"grad_norm": 12.778799057006836,
|
| 413 |
+
"learning_rate": 3.756756756756757e-05,
|
| 414 |
+
"loss": 2.4453,
|
| 415 |
+
"step": 280
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 7.741721854304636,
|
| 419 |
+
"grad_norm": 12.07343864440918,
|
| 420 |
+
"learning_rate": 3.851351351351351e-05,
|
| 421 |
+
"loss": 2.3233,
|
| 422 |
+
"step": 287
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 7.927152317880795,
|
| 426 |
+
"grad_norm": 11.423672676086426,
|
| 427 |
+
"learning_rate": 3.945945945945946e-05,
|
| 428 |
+
"loss": 2.2578,
|
| 429 |
+
"step": 294
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 7.9801324503311255,
|
| 433 |
+
"eval_accuracy": 0.6139344262295082,
|
| 434 |
+
"eval_f1_macro": 0.5881810404081942,
|
| 435 |
+
"eval_f1_micro": 0.6139344262295082,
|
| 436 |
+
"eval_f1_weighted": 0.596493094032159,
|
| 437 |
+
"eval_loss": 1.5912988185882568,
|
| 438 |
+
"eval_precision_macro": 0.6479628814812638,
|
| 439 |
+
"eval_precision_micro": 0.6139344262295082,
|
| 440 |
+
"eval_precision_weighted": 0.6507304591369202,
|
| 441 |
+
"eval_recall_macro": 0.6029464285714285,
|
| 442 |
+
"eval_recall_micro": 0.6139344262295082,
|
| 443 |
+
"eval_recall_weighted": 0.6139344262295082,
|
| 444 |
+
"eval_runtime": 43.3133,
|
| 445 |
+
"eval_samples_per_second": 28.167,
|
| 446 |
+
"eval_steps_per_second": 0.462,
|
| 447 |
+
"step": 296
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 8.132450331125828,
|
| 451 |
+
"grad_norm": 11.603616714477539,
|
| 452 |
+
"learning_rate": 4.04054054054054e-05,
|
| 453 |
+
"loss": 2.1519,
|
| 454 |
+
"step": 301
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 8.317880794701987,
|
| 458 |
+
"grad_norm": 13.292801856994629,
|
| 459 |
+
"learning_rate": 4.1216216216216216e-05,
|
| 460 |
+
"loss": 2.1126,
|
| 461 |
+
"step": 308
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 8.503311258278146,
|
| 465 |
+
"grad_norm": 10.869479179382324,
|
| 466 |
+
"learning_rate": 4.2162162162162164e-05,
|
| 467 |
+
"loss": 2.0236,
|
| 468 |
+
"step": 315
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 8.688741721854305,
|
| 472 |
+
"grad_norm": 12.716577529907227,
|
| 473 |
+
"learning_rate": 4.3108108108108106e-05,
|
| 474 |
+
"loss": 1.887,
|
| 475 |
+
"step": 322
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 8.874172185430464,
|
| 479 |
+
"grad_norm": 11.295136451721191,
|
| 480 |
+
"learning_rate": 4.4054054054054054e-05,
|
| 481 |
+
"loss": 1.876,
|
| 482 |
+
"step": 329
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 8.980132450331126,
|
| 486 |
+
"eval_accuracy": 0.6721311475409836,
|
| 487 |
+
"eval_f1_macro": 0.6580130012428606,
|
| 488 |
+
"eval_f1_micro": 0.6721311475409836,
|
| 489 |
+
"eval_f1_weighted": 0.6639418969334906,
|
| 490 |
+
"eval_loss": 1.2565070390701294,
|
| 491 |
+
"eval_precision_macro": 0.7232300878945616,
|
| 492 |
+
"eval_precision_micro": 0.6721311475409836,
|
| 493 |
+
"eval_precision_weighted": 0.7223922360953206,
|
| 494 |
+
"eval_recall_macro": 0.6628571428571429,
|
| 495 |
+
"eval_recall_micro": 0.6721311475409836,
|
| 496 |
+
"eval_recall_weighted": 0.6721311475409836,
|
| 497 |
+
"eval_runtime": 40.0516,
|
| 498 |
+
"eval_samples_per_second": 30.461,
|
| 499 |
+
"eval_steps_per_second": 0.499,
|
| 500 |
+
"step": 333
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 9.079470198675496,
|
| 504 |
+
"grad_norm": 11.199146270751953,
|
| 505 |
+
"learning_rate": 4.5e-05,
|
| 506 |
+
"loss": 1.8298,
|
| 507 |
+
"step": 336
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 9.264900662251655,
|
| 511 |
+
"grad_norm": 11.203076362609863,
|
| 512 |
+
"learning_rate": 4.594594594594595e-05,
|
| 513 |
+
"loss": 1.709,
|
| 514 |
+
"step": 343
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 9.450331125827814,
|
| 518 |
+
"grad_norm": 10.677241325378418,
|
| 519 |
+
"learning_rate": 4.689189189189189e-05,
|
| 520 |
+
"loss": 1.6401,
|
| 521 |
+
"step": 350
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 9.635761589403973,
|
| 525 |
+
"grad_norm": 10.3362398147583,
|
| 526 |
+
"learning_rate": 4.783783783783784e-05,
|
| 527 |
+
"loss": 1.6023,
|
| 528 |
+
"step": 357
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 9.821192052980132,
|
| 532 |
+
"grad_norm": 10.059170722961426,
|
| 533 |
+
"learning_rate": 4.878378378378379e-05,
|
| 534 |
+
"loss": 1.6111,
|
| 535 |
+
"step": 364
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 9.980132450331126,
|
| 539 |
+
"eval_accuracy": 0.7163934426229508,
|
| 540 |
+
"eval_f1_macro": 0.6956014124577357,
|
| 541 |
+
"eval_f1_micro": 0.7163934426229508,
|
| 542 |
+
"eval_f1_weighted": 0.7006354276645078,
|
| 543 |
+
"eval_loss": 1.0965081453323364,
|
| 544 |
+
"eval_precision_macro": 0.7575524683649684,
|
| 545 |
+
"eval_precision_micro": 0.7163934426229508,
|
| 546 |
+
"eval_precision_weighted": 0.7569573720905689,
|
| 547 |
+
"eval_recall_macro": 0.7082083333333332,
|
| 548 |
+
"eval_recall_micro": 0.7163934426229508,
|
| 549 |
+
"eval_recall_weighted": 0.7163934426229508,
|
| 550 |
+
"eval_runtime": 46.3112,
|
| 551 |
+
"eval_samples_per_second": 26.344,
|
| 552 |
+
"eval_steps_per_second": 0.432,
|
| 553 |
+
"step": 370
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 10.026490066225165,
|
| 557 |
+
"grad_norm": 10.692099571228027,
|
| 558 |
+
"learning_rate": 4.972972972972974e-05,
|
| 559 |
+
"loss": 1.5381,
|
| 560 |
+
"step": 371
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 10.211920529801324,
|
| 564 |
+
"grad_norm": 9.687626838684082,
|
| 565 |
+
"learning_rate": 4.9924924924924924e-05,
|
| 566 |
+
"loss": 1.4227,
|
| 567 |
+
"step": 378
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 10.397350993377483,
|
| 571 |
+
"grad_norm": 9.885163307189941,
|
| 572 |
+
"learning_rate": 4.9819819819819824e-05,
|
| 573 |
+
"loss": 1.4277,
|
| 574 |
+
"step": 385
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 10.582781456953642,
|
| 578 |
+
"grad_norm": 19.606826782226562,
|
| 579 |
+
"learning_rate": 4.971471471471472e-05,
|
| 580 |
+
"loss": 1.3542,
|
| 581 |
+
"step": 392
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 10.7682119205298,
|
| 585 |
+
"grad_norm": 10.894911766052246,
|
| 586 |
+
"learning_rate": 4.960960960960962e-05,
|
| 587 |
+
"loss": 1.4136,
|
| 588 |
+
"step": 399
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 10.95364238410596,
|
| 592 |
+
"grad_norm": 11.28426456451416,
|
| 593 |
+
"learning_rate": 4.95045045045045e-05,
|
| 594 |
+
"loss": 1.4096,
|
| 595 |
+
"step": 406
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 10.980132450331126,
|
| 599 |
+
"eval_accuracy": 0.7442622950819672,
|
| 600 |
+
"eval_f1_macro": 0.7309345112923151,
|
| 601 |
+
"eval_f1_micro": 0.7442622950819672,
|
| 602 |
+
"eval_f1_weighted": 0.7361964062069708,
|
| 603 |
+
"eval_loss": 0.9353717565536499,
|
| 604 |
+
"eval_precision_macro": 0.785672709235209,
|
| 605 |
+
"eval_precision_micro": 0.7442622950819672,
|
| 606 |
+
"eval_precision_weighted": 0.7851744168854825,
|
| 607 |
+
"eval_recall_macro": 0.735404761904762,
|
| 608 |
+
"eval_recall_micro": 0.7442622950819672,
|
| 609 |
+
"eval_recall_weighted": 0.7442622950819672,
|
| 610 |
+
"eval_runtime": 43.3522,
|
| 611 |
+
"eval_samples_per_second": 28.142,
|
| 612 |
+
"eval_steps_per_second": 0.461,
|
| 613 |
+
"step": 407
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 11.158940397350994,
|
| 617 |
+
"grad_norm": 8.755209922790527,
|
| 618 |
+
"learning_rate": 4.93993993993994e-05,
|
| 619 |
+
"loss": 1.2942,
|
| 620 |
+
"step": 413
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 11.344370860927153,
|
| 624 |
+
"grad_norm": 9.342432975769043,
|
| 625 |
+
"learning_rate": 4.9294294294294296e-05,
|
| 626 |
+
"loss": 1.221,
|
| 627 |
+
"step": 420
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 11.52980132450331,
|
| 631 |
+
"grad_norm": 10.622260093688965,
|
| 632 |
+
"learning_rate": 4.9189189189189196e-05,
|
| 633 |
+
"loss": 1.2387,
|
| 634 |
+
"step": 427
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 11.71523178807947,
|
| 638 |
+
"grad_norm": 9.346344947814941,
|
| 639 |
+
"learning_rate": 4.908408408408409e-05,
|
| 640 |
+
"loss": 1.2545,
|
| 641 |
+
"step": 434
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 11.900662251655628,
|
| 645 |
+
"grad_norm": 9.809884071350098,
|
| 646 |
+
"learning_rate": 4.8978978978978976e-05,
|
| 647 |
+
"loss": 1.2697,
|
| 648 |
+
"step": 441
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 11.980132450331126,
|
| 652 |
+
"eval_accuracy": 0.7836065573770492,
|
| 653 |
+
"eval_f1_macro": 0.7754873092164115,
|
| 654 |
+
"eval_f1_micro": 0.7836065573770492,
|
| 655 |
+
"eval_f1_weighted": 0.7788164207329435,
|
| 656 |
+
"eval_loss": 0.8078306913375854,
|
| 657 |
+
"eval_precision_macro": 0.8157236513486512,
|
| 658 |
+
"eval_precision_micro": 0.7836065573770492,
|
| 659 |
+
"eval_precision_weighted": 0.8160415404267863,
|
| 660 |
+
"eval_recall_macro": 0.7780952380952381,
|
| 661 |
+
"eval_recall_micro": 0.7836065573770492,
|
| 662 |
+
"eval_recall_weighted": 0.7836065573770492,
|
| 663 |
+
"eval_runtime": 53.3885,
|
| 664 |
+
"eval_samples_per_second": 22.851,
|
| 665 |
+
"eval_steps_per_second": 0.375,
|
| 666 |
+
"step": 444
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 12.105960264900663,
|
| 670 |
+
"grad_norm": 9.783758163452148,
|
| 671 |
+
"learning_rate": 4.8873873873873876e-05,
|
| 672 |
+
"loss": 1.2426,
|
| 673 |
+
"step": 448
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 12.291390728476822,
|
| 677 |
+
"grad_norm": 8.787826538085938,
|
| 678 |
+
"learning_rate": 4.876876876876877e-05,
|
| 679 |
+
"loss": 1.1275,
|
| 680 |
+
"step": 455
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 12.47682119205298,
|
| 684 |
+
"grad_norm": 10.826144218444824,
|
| 685 |
+
"learning_rate": 4.866366366366367e-05,
|
| 686 |
+
"loss": 1.1154,
|
| 687 |
+
"step": 462
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 12.66225165562914,
|
| 691 |
+
"grad_norm": 9.54553508758545,
|
| 692 |
+
"learning_rate": 4.855855855855856e-05,
|
| 693 |
+
"loss": 1.089,
|
| 694 |
+
"step": 469
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 12.847682119205299,
|
| 698 |
+
"grad_norm": 10.958932876586914,
|
| 699 |
+
"learning_rate": 4.8453453453453455e-05,
|
| 700 |
+
"loss": 1.0683,
|
| 701 |
+
"step": 476
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 12.980132450331126,
|
| 705 |
+
"eval_accuracy": 0.7877049180327869,
|
| 706 |
+
"eval_f1_macro": 0.7809517814057289,
|
| 707 |
+
"eval_f1_micro": 0.7877049180327869,
|
| 708 |
+
"eval_f1_weighted": 0.7842625439420977,
|
| 709 |
+
"eval_loss": 0.7565202116966248,
|
| 710 |
+
"eval_precision_macro": 0.8255538701494585,
|
| 711 |
+
"eval_precision_micro": 0.7877049180327869,
|
| 712 |
+
"eval_precision_weighted": 0.8263731434910316,
|
| 713 |
+
"eval_recall_macro": 0.7832797619047619,
|
| 714 |
+
"eval_recall_micro": 0.7877049180327869,
|
| 715 |
+
"eval_recall_weighted": 0.7877049180327869,
|
| 716 |
+
"eval_runtime": 44.7029,
|
| 717 |
+
"eval_samples_per_second": 27.291,
|
| 718 |
+
"eval_steps_per_second": 0.447,
|
| 719 |
+
"step": 481
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 13.052980132450331,
|
| 723 |
+
"grad_norm": 10.9450044631958,
|
| 724 |
+
"learning_rate": 4.834834834834835e-05,
|
| 725 |
+
"loss": 1.1861,
|
| 726 |
+
"step": 483
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 13.23841059602649,
|
| 730 |
+
"grad_norm": 10.095162391662598,
|
| 731 |
+
"learning_rate": 4.824324324324325e-05,
|
| 732 |
+
"loss": 1.013,
|
| 733 |
+
"step": 490
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 13.42384105960265,
|
| 737 |
+
"grad_norm": 9.601603507995605,
|
| 738 |
+
"learning_rate": 4.813813813813814e-05,
|
| 739 |
+
"loss": 0.9864,
|
| 740 |
+
"step": 497
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 13.609271523178808,
|
| 744 |
+
"grad_norm": 8.540664672851562,
|
| 745 |
+
"learning_rate": 4.8033033033033034e-05,
|
| 746 |
+
"loss": 0.8722,
|
| 747 |
+
"step": 504
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 13.794701986754967,
|
| 751 |
+
"grad_norm": 10.721132278442383,
|
| 752 |
+
"learning_rate": 4.792792792792793e-05,
|
| 753 |
+
"loss": 1.0975,
|
| 754 |
+
"step": 511
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 13.980132450331126,
|
| 758 |
+
"grad_norm": 11.830833435058594,
|
| 759 |
+
"learning_rate": 4.782282282282283e-05,
|
| 760 |
+
"loss": 1.0312,
|
| 761 |
+
"step": 518
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 13.980132450331126,
|
| 765 |
+
"eval_accuracy": 0.7975409836065573,
|
| 766 |
+
"eval_f1_macro": 0.7928224644292756,
|
| 767 |
+
"eval_f1_micro": 0.7975409836065573,
|
| 768 |
+
"eval_f1_weighted": 0.7953112733063121,
|
| 769 |
+
"eval_loss": 0.6965898275375366,
|
| 770 |
+
"eval_precision_macro": 0.8293739316239317,
|
| 771 |
+
"eval_precision_micro": 0.7975409836065573,
|
| 772 |
+
"eval_precision_weighted": 0.8288230894788272,
|
| 773 |
+
"eval_recall_macro": 0.7924821428571428,
|
| 774 |
+
"eval_recall_micro": 0.7975409836065573,
|
| 775 |
+
"eval_recall_weighted": 0.7975409836065573,
|
| 776 |
+
"eval_runtime": 40.1775,
|
| 777 |
+
"eval_samples_per_second": 30.365,
|
| 778 |
+
"eval_steps_per_second": 0.498,
|
| 779 |
+
"step": 518
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 14.185430463576159,
|
| 783 |
+
"grad_norm": 10.47096061706543,
|
| 784 |
+
"learning_rate": 4.771771771771772e-05,
|
| 785 |
+
"loss": 1.0098,
|
| 786 |
+
"step": 525
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 14.370860927152318,
|
| 790 |
+
"grad_norm": 9.135430335998535,
|
| 791 |
+
"learning_rate": 4.761261261261262e-05,
|
| 792 |
+
"loss": 1.0335,
|
| 793 |
+
"step": 532
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 14.556291390728477,
|
| 797 |
+
"grad_norm": 9.590068817138672,
|
| 798 |
+
"learning_rate": 4.7507507507507506e-05,
|
| 799 |
+
"loss": 0.8974,
|
| 800 |
+
"step": 539
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 14.741721854304636,
|
| 804 |
+
"grad_norm": 11.321711540222168,
|
| 805 |
+
"learning_rate": 4.7402402402402406e-05,
|
| 806 |
+
"loss": 0.9296,
|
| 807 |
+
"step": 546
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 14.927152317880795,
|
| 811 |
+
"grad_norm": 11.742105484008789,
|
| 812 |
+
"learning_rate": 4.72972972972973e-05,
|
| 813 |
+
"loss": 0.9903,
|
| 814 |
+
"step": 553
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 14.980132450331126,
|
| 818 |
+
"eval_accuracy": 0.8016393442622951,
|
| 819 |
+
"eval_f1_macro": 0.7979719305748716,
|
| 820 |
+
"eval_f1_micro": 0.8016393442622951,
|
| 821 |
+
"eval_f1_weighted": 0.8006367112297681,
|
| 822 |
+
"eval_loss": 0.6716923117637634,
|
| 823 |
+
"eval_precision_macro": 0.8302031995781997,
|
| 824 |
+
"eval_precision_micro": 0.8016393442622951,
|
| 825 |
+
"eval_precision_weighted": 0.8311227843195056,
|
| 826 |
+
"eval_recall_macro": 0.797970238095238,
|
| 827 |
+
"eval_recall_micro": 0.8016393442622951,
|
| 828 |
+
"eval_recall_weighted": 0.8016393442622951,
|
| 829 |
+
"eval_runtime": 63.6088,
|
| 830 |
+
"eval_samples_per_second": 19.18,
|
| 831 |
+
"eval_steps_per_second": 0.314,
|
| 832 |
+
"step": 555
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 15.132450331125828,
|
| 836 |
+
"grad_norm": 9.648506164550781,
|
| 837 |
+
"learning_rate": 4.71921921921922e-05,
|
| 838 |
+
"loss": 0.9712,
|
| 839 |
+
"step": 560
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 15.317880794701987,
|
| 843 |
+
"grad_norm": 8.26517391204834,
|
| 844 |
+
"learning_rate": 4.708708708708709e-05,
|
| 845 |
+
"loss": 0.8787,
|
| 846 |
+
"step": 567
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 15.503311258278146,
|
| 850 |
+
"grad_norm": 8.81796646118164,
|
| 851 |
+
"learning_rate": 4.698198198198198e-05,
|
| 852 |
+
"loss": 0.9198,
|
| 853 |
+
"step": 574
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 15.688741721854305,
|
| 857 |
+
"grad_norm": 8.624659538269043,
|
| 858 |
+
"learning_rate": 4.687687687687688e-05,
|
| 859 |
+
"loss": 0.8703,
|
| 860 |
+
"step": 581
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 15.874172185430464,
|
| 864 |
+
"grad_norm": 8.055384635925293,
|
| 865 |
+
"learning_rate": 4.677177177177177e-05,
|
| 866 |
+
"loss": 0.9453,
|
| 867 |
+
"step": 588
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 15.980132450331126,
|
| 871 |
+
"eval_accuracy": 0.8213114754098361,
|
| 872 |
+
"eval_f1_macro": 0.8200054048548631,
|
| 873 |
+
"eval_f1_micro": 0.8213114754098361,
|
| 874 |
+
"eval_f1_weighted": 0.8235741781663722,
|
| 875 |
+
"eval_loss": 0.65358966588974,
|
| 876 |
+
"eval_precision_macro": 0.8519878316128316,
|
| 877 |
+
"eval_precision_micro": 0.8213114754098361,
|
| 878 |
+
"eval_precision_weighted": 0.8550255686936015,
|
| 879 |
+
"eval_recall_macro": 0.8179345238095237,
|
| 880 |
+
"eval_recall_micro": 0.8213114754098361,
|
| 881 |
+
"eval_recall_weighted": 0.8213114754098361,
|
| 882 |
+
"eval_runtime": 61.357,
|
| 883 |
+
"eval_samples_per_second": 19.884,
|
| 884 |
+
"eval_steps_per_second": 0.326,
|
| 885 |
+
"step": 592
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 16.079470198675498,
|
| 889 |
+
"grad_norm": 7.529909133911133,
|
| 890 |
+
"learning_rate": 4.666666666666667e-05,
|
| 891 |
+
"loss": 0.981,
|
| 892 |
+
"step": 595
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 16.264900662251655,
|
| 896 |
+
"grad_norm": 7.517402172088623,
|
| 897 |
+
"learning_rate": 4.6561561561561565e-05,
|
| 898 |
+
"loss": 0.7995,
|
| 899 |
+
"step": 602
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 16.450331125827816,
|
| 903 |
+
"grad_norm": 8.271308898925781,
|
| 904 |
+
"learning_rate": 4.645645645645646e-05,
|
| 905 |
+
"loss": 0.7973,
|
| 906 |
+
"step": 609
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 16.635761589403973,
|
| 910 |
+
"grad_norm": 8.204742431640625,
|
| 911 |
+
"learning_rate": 4.635135135135135e-05,
|
| 912 |
+
"loss": 0.841,
|
| 913 |
+
"step": 616
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 16.821192052980134,
|
| 917 |
+
"grad_norm": 8.602618217468262,
|
| 918 |
+
"learning_rate": 4.624624624624625e-05,
|
| 919 |
+
"loss": 0.939,
|
| 920 |
+
"step": 623
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 16.980132450331126,
|
| 924 |
+
"eval_accuracy": 0.8221311475409836,
|
| 925 |
+
"eval_f1_macro": 0.8163740850652615,
|
| 926 |
+
"eval_f1_micro": 0.8221311475409836,
|
| 927 |
+
"eval_f1_weighted": 0.8187325264470876,
|
| 928 |
+
"eval_loss": 0.6268297433853149,
|
| 929 |
+
"eval_precision_macro": 0.844480339105339,
|
| 930 |
+
"eval_precision_micro": 0.8221311475409836,
|
| 931 |
+
"eval_precision_weighted": 0.8438404832872046,
|
| 932 |
+
"eval_recall_macro": 0.8187619047619047,
|
| 933 |
+
"eval_recall_micro": 0.8221311475409836,
|
| 934 |
+
"eval_recall_weighted": 0.8221311475409836,
|
| 935 |
+
"eval_runtime": 61.1021,
|
| 936 |
+
"eval_samples_per_second": 19.967,
|
| 937 |
+
"eval_steps_per_second": 0.327,
|
| 938 |
+
"step": 629
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 17.026490066225165,
|
| 942 |
+
"grad_norm": 8.355627059936523,
|
| 943 |
+
"learning_rate": 4.6141141141141144e-05,
|
| 944 |
+
"loss": 0.7965,
|
| 945 |
+
"step": 630
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 17.211920529801326,
|
| 949 |
+
"grad_norm": 8.196239471435547,
|
| 950 |
+
"learning_rate": 4.603603603603604e-05,
|
| 951 |
+
"loss": 0.7299,
|
| 952 |
+
"step": 637
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 17.397350993377483,
|
| 956 |
+
"grad_norm": 5.987618923187256,
|
| 957 |
+
"learning_rate": 4.593093093093093e-05,
|
| 958 |
+
"loss": 0.7109,
|
| 959 |
+
"step": 644
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 17.582781456953644,
|
| 963 |
+
"grad_norm": 7.704870223999023,
|
| 964 |
+
"learning_rate": 4.582582582582583e-05,
|
| 965 |
+
"loss": 0.7835,
|
| 966 |
+
"step": 651
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 17.7682119205298,
|
| 970 |
+
"grad_norm": 9.14775276184082,
|
| 971 |
+
"learning_rate": 4.572072072072072e-05,
|
| 972 |
+
"loss": 0.7746,
|
| 973 |
+
"step": 658
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 17.95364238410596,
|
| 977 |
+
"grad_norm": 9.578638076782227,
|
| 978 |
+
"learning_rate": 4.561561561561562e-05,
|
| 979 |
+
"loss": 0.7787,
|
| 980 |
+
"step": 665
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 17.980132450331126,
|
| 984 |
+
"eval_accuracy": 0.8254098360655737,
|
| 985 |
+
"eval_f1_macro": 0.8200774953460711,
|
| 986 |
+
"eval_f1_micro": 0.8254098360655737,
|
| 987 |
+
"eval_f1_weighted": 0.8212272647936633,
|
| 988 |
+
"eval_loss": 0.6105802655220032,
|
| 989 |
+
"eval_precision_macro": 0.8466414141414141,
|
| 990 |
+
"eval_precision_micro": 0.8254098360655737,
|
| 991 |
+
"eval_precision_weighted": 0.846929452605682,
|
| 992 |
+
"eval_recall_macro": 0.824232142857143,
|
| 993 |
+
"eval_recall_micro": 0.8254098360655737,
|
| 994 |
+
"eval_recall_weighted": 0.8254098360655737,
|
| 995 |
+
"eval_runtime": 61.552,
|
| 996 |
+
"eval_samples_per_second": 19.821,
|
| 997 |
+
"eval_steps_per_second": 0.325,
|
| 998 |
+
"step": 666
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 18.158940397350992,
|
| 1002 |
+
"grad_norm": 10.692727088928223,
|
| 1003 |
+
"learning_rate": 4.551051051051051e-05,
|
| 1004 |
+
"loss": 0.7673,
|
| 1005 |
+
"step": 672
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 18.344370860927153,
|
| 1009 |
+
"grad_norm": 8.184776306152344,
|
| 1010 |
+
"learning_rate": 4.540540540540541e-05,
|
| 1011 |
+
"loss": 0.6994,
|
| 1012 |
+
"step": 679
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 18.52980132450331,
|
| 1016 |
+
"grad_norm": 6.153806686401367,
|
| 1017 |
+
"learning_rate": 4.53003003003003e-05,
|
| 1018 |
+
"loss": 0.7311,
|
| 1019 |
+
"step": 686
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 18.71523178807947,
|
| 1023 |
+
"grad_norm": 8.01244831085205,
|
| 1024 |
+
"learning_rate": 4.5195195195195196e-05,
|
| 1025 |
+
"loss": 0.69,
|
| 1026 |
+
"step": 693
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 18.90066225165563,
|
| 1030 |
+
"grad_norm": 9.361739158630371,
|
| 1031 |
+
"learning_rate": 4.5090090090090095e-05,
|
| 1032 |
+
"loss": 0.7366,
|
| 1033 |
+
"step": 700
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 18.980132450331126,
|
| 1037 |
+
"eval_accuracy": 0.830327868852459,
|
| 1038 |
+
"eval_f1_macro": 0.8230471515069039,
|
| 1039 |
+
"eval_f1_micro": 0.830327868852459,
|
| 1040 |
+
"eval_f1_weighted": 0.8255601272328421,
|
| 1041 |
+
"eval_loss": 0.5895594358444214,
|
| 1042 |
+
"eval_precision_macro": 0.8522619047619048,
|
| 1043 |
+
"eval_precision_micro": 0.830327868852459,
|
| 1044 |
+
"eval_precision_weighted": 0.8527185792349726,
|
| 1045 |
+
"eval_recall_macro": 0.8273750000000001,
|
| 1046 |
+
"eval_recall_micro": 0.830327868852459,
|
| 1047 |
+
"eval_recall_weighted": 0.830327868852459,
|
| 1048 |
+
"eval_runtime": 47.5144,
|
| 1049 |
+
"eval_samples_per_second": 25.676,
|
| 1050 |
+
"eval_steps_per_second": 0.421,
|
| 1051 |
+
"step": 703
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 19.105960264900663,
|
| 1055 |
+
"grad_norm": 9.264578819274902,
|
| 1056 |
+
"learning_rate": 4.498498498498498e-05,
|
| 1057 |
+
"loss": 0.6839,
|
| 1058 |
+
"step": 707
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 19.29139072847682,
|
| 1062 |
+
"grad_norm": 11.78283405303955,
|
| 1063 |
+
"learning_rate": 4.487987987987988e-05,
|
| 1064 |
+
"loss": 0.7013,
|
| 1065 |
+
"step": 714
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 19.47682119205298,
|
| 1069 |
+
"grad_norm": 7.552002906799316,
|
| 1070 |
+
"learning_rate": 4.4774774774774775e-05,
|
| 1071 |
+
"loss": 0.7158,
|
| 1072 |
+
"step": 721
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 19.662251655629138,
|
| 1076 |
+
"grad_norm": 8.90753173828125,
|
| 1077 |
+
"learning_rate": 4.4669669669669675e-05,
|
| 1078 |
+
"loss": 0.7336,
|
| 1079 |
+
"step": 728
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 19.8476821192053,
|
| 1083 |
+
"grad_norm": 9.122941970825195,
|
| 1084 |
+
"learning_rate": 4.456456456456457e-05,
|
| 1085 |
+
"loss": 0.7401,
|
| 1086 |
+
"step": 735
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 19.980132450331126,
|
| 1090 |
+
"eval_accuracy": 0.8245901639344262,
|
| 1091 |
+
"eval_f1_macro": 0.8191573312652259,
|
| 1092 |
+
"eval_f1_micro": 0.8245901639344262,
|
| 1093 |
+
"eval_f1_weighted": 0.8222225393243513,
|
| 1094 |
+
"eval_loss": 0.6023704409599304,
|
| 1095 |
+
"eval_precision_macro": 0.8524332611832612,
|
| 1096 |
+
"eval_precision_micro": 0.8245901639344262,
|
| 1097 |
+
"eval_precision_weighted": 0.8533740212428738,
|
| 1098 |
+
"eval_recall_macro": 0.8203273809523809,
|
| 1099 |
+
"eval_recall_micro": 0.8245901639344262,
|
| 1100 |
+
"eval_recall_weighted": 0.8245901639344262,
|
| 1101 |
+
"eval_runtime": 43.4261,
|
| 1102 |
+
"eval_samples_per_second": 28.094,
|
| 1103 |
+
"eval_steps_per_second": 0.461,
|
| 1104 |
+
"step": 740
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 20.05298013245033,
|
| 1108 |
+
"grad_norm": 7.7251386642456055,
|
| 1109 |
+
"learning_rate": 4.445945945945946e-05,
|
| 1110 |
+
"loss": 0.8092,
|
| 1111 |
+
"step": 742
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 20.23841059602649,
|
| 1115 |
+
"grad_norm": 7.6273112297058105,
|
| 1116 |
+
"learning_rate": 4.4354354354354354e-05,
|
| 1117 |
+
"loss": 0.7029,
|
| 1118 |
+
"step": 749
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 20.423841059602648,
|
| 1122 |
+
"grad_norm": 8.434123039245605,
|
| 1123 |
+
"learning_rate": 4.4249249249249254e-05,
|
| 1124 |
+
"loss": 0.6458,
|
| 1125 |
+
"step": 756
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 20.60927152317881,
|
| 1129 |
+
"grad_norm": 8.381632804870605,
|
| 1130 |
+
"learning_rate": 4.414414414414415e-05,
|
| 1131 |
+
"loss": 0.6677,
|
| 1132 |
+
"step": 763
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 20.794701986754966,
|
| 1136 |
+
"grad_norm": 7.361513614654541,
|
| 1137 |
+
"learning_rate": 4.403903903903904e-05,
|
| 1138 |
+
"loss": 0.6974,
|
| 1139 |
+
"step": 770
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 20.980132450331126,
|
| 1143 |
+
"grad_norm": 6.694199085235596,
|
| 1144 |
+
"learning_rate": 4.393393393393393e-05,
|
| 1145 |
+
"loss": 0.6855,
|
| 1146 |
+
"step": 777
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 20.980132450331126,
|
| 1150 |
+
"eval_accuracy": 0.8352459016393443,
|
| 1151 |
+
"eval_f1_macro": 0.8301821885974363,
|
| 1152 |
+
"eval_f1_micro": 0.8352459016393443,
|
| 1153 |
+
"eval_f1_weighted": 0.8337427464021765,
|
| 1154 |
+
"eval_loss": 0.5506462454795837,
|
| 1155 |
+
"eval_precision_macro": 0.8520505050505051,
|
| 1156 |
+
"eval_precision_micro": 0.8352459016393443,
|
| 1157 |
+
"eval_precision_weighted": 0.8550262283254086,
|
| 1158 |
+
"eval_recall_macro": 0.8311190476190475,
|
| 1159 |
+
"eval_recall_micro": 0.8352459016393443,
|
| 1160 |
+
"eval_recall_weighted": 0.8352459016393443,
|
| 1161 |
+
"eval_runtime": 42.4103,
|
| 1162 |
+
"eval_samples_per_second": 28.767,
|
| 1163 |
+
"eval_steps_per_second": 0.472,
|
| 1164 |
+
"step": 777
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 21.185430463576157,
|
| 1168 |
+
"grad_norm": 7.033413887023926,
|
| 1169 |
+
"learning_rate": 4.382882882882883e-05,
|
| 1170 |
+
"loss": 0.6146,
|
| 1171 |
+
"step": 784
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 21.370860927152318,
|
| 1175 |
+
"grad_norm": 6.893985271453857,
|
| 1176 |
+
"learning_rate": 4.3723723723723726e-05,
|
| 1177 |
+
"loss": 0.6297,
|
| 1178 |
+
"step": 791
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 21.556291390728475,
|
| 1182 |
+
"grad_norm": 8.514236450195312,
|
| 1183 |
+
"learning_rate": 4.3618618618618626e-05,
|
| 1184 |
+
"loss": 0.5797,
|
| 1185 |
+
"step": 798
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 21.741721854304636,
|
| 1189 |
+
"grad_norm": 6.57477331161499,
|
| 1190 |
+
"learning_rate": 4.351351351351351e-05,
|
| 1191 |
+
"loss": 0.7013,
|
| 1192 |
+
"step": 805
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 21.927152317880793,
|
| 1196 |
+
"grad_norm": 8.258438110351562,
|
| 1197 |
+
"learning_rate": 4.340840840840841e-05,
|
| 1198 |
+
"loss": 0.6179,
|
| 1199 |
+
"step": 812
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 21.980132450331126,
|
| 1203 |
+
"eval_accuracy": 0.8434426229508196,
|
| 1204 |
+
"eval_f1_macro": 0.8351991206677739,
|
| 1205 |
+
"eval_f1_micro": 0.8434426229508196,
|
| 1206 |
+
"eval_f1_weighted": 0.839283469657334,
|
| 1207 |
+
"eval_loss": 0.5583724975585938,
|
| 1208 |
+
"eval_precision_macro": 0.8586598401598402,
|
| 1209 |
+
"eval_precision_micro": 0.8434426229508196,
|
| 1210 |
+
"eval_precision_weighted": 0.8618578962021584,
|
| 1211 |
+
"eval_recall_macro": 0.8398928571428571,
|
| 1212 |
+
"eval_recall_micro": 0.8434426229508196,
|
| 1213 |
+
"eval_recall_weighted": 0.8434426229508196,
|
| 1214 |
+
"eval_runtime": 42.4207,
|
| 1215 |
+
"eval_samples_per_second": 28.76,
|
| 1216 |
+
"eval_steps_per_second": 0.471,
|
| 1217 |
+
"step": 814
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 22.132450331125828,
|
| 1221 |
+
"grad_norm": 6.932776927947998,
|
| 1222 |
+
"learning_rate": 4.3303303303303305e-05,
|
| 1223 |
+
"loss": 0.6852,
|
| 1224 |
+
"step": 819
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 22.31788079470199,
|
| 1228 |
+
"grad_norm": 7.310856342315674,
|
| 1229 |
+
"learning_rate": 4.31981981981982e-05,
|
| 1230 |
+
"loss": 0.6273,
|
| 1231 |
+
"step": 826
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 22.503311258278146,
|
| 1235 |
+
"grad_norm": 7.708891868591309,
|
| 1236 |
+
"learning_rate": 4.30930930930931e-05,
|
| 1237 |
+
"loss": 0.6736,
|
| 1238 |
+
"step": 833
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 22.688741721854306,
|
| 1242 |
+
"grad_norm": 7.841245651245117,
|
| 1243 |
+
"learning_rate": 4.2987987987987985e-05,
|
| 1244 |
+
"loss": 0.6196,
|
| 1245 |
+
"step": 840
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 22.874172185430464,
|
| 1249 |
+
"grad_norm": 8.23265266418457,
|
| 1250 |
+
"learning_rate": 4.2882882882882885e-05,
|
| 1251 |
+
"loss": 0.7047,
|
| 1252 |
+
"step": 847
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 22.980132450331126,
|
| 1256 |
+
"eval_accuracy": 0.8377049180327869,
|
| 1257 |
+
"eval_f1_macro": 0.8308334231128349,
|
| 1258 |
+
"eval_f1_micro": 0.8377049180327869,
|
| 1259 |
+
"eval_f1_weighted": 0.8344787582769744,
|
| 1260 |
+
"eval_loss": 0.5581976175308228,
|
| 1261 |
+
"eval_precision_macro": 0.8549415584415584,
|
| 1262 |
+
"eval_precision_micro": 0.8377049180327869,
|
| 1263 |
+
"eval_precision_weighted": 0.8558833652686111,
|
| 1264 |
+
"eval_recall_macro": 0.832327380952381,
|
| 1265 |
+
"eval_recall_micro": 0.8377049180327869,
|
| 1266 |
+
"eval_recall_weighted": 0.8377049180327869,
|
| 1267 |
+
"eval_runtime": 40.4504,
|
| 1268 |
+
"eval_samples_per_second": 30.16,
|
| 1269 |
+
"eval_steps_per_second": 0.494,
|
| 1270 |
+
"step": 851
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 23.079470198675498,
|
| 1274 |
+
"grad_norm": 6.447187900543213,
|
| 1275 |
+
"learning_rate": 4.277777777777778e-05,
|
| 1276 |
+
"loss": 0.6207,
|
| 1277 |
+
"step": 854
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 23.264900662251655,
|
| 1281 |
+
"grad_norm": 7.463189601898193,
|
| 1282 |
+
"learning_rate": 4.267267267267268e-05,
|
| 1283 |
+
"loss": 0.5505,
|
| 1284 |
+
"step": 861
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 23.450331125827816,
|
| 1288 |
+
"grad_norm": 6.670192718505859,
|
| 1289 |
+
"learning_rate": 4.256756756756757e-05,
|
| 1290 |
+
"loss": 0.5429,
|
| 1291 |
+
"step": 868
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 23.635761589403973,
|
| 1295 |
+
"grad_norm": 7.076443672180176,
|
| 1296 |
+
"learning_rate": 4.2462462462462464e-05,
|
| 1297 |
+
"loss": 0.562,
|
| 1298 |
+
"step": 875
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 23.821192052980134,
|
| 1302 |
+
"grad_norm": 7.10822057723999,
|
| 1303 |
+
"learning_rate": 4.235735735735736e-05,
|
| 1304 |
+
"loss": 0.6327,
|
| 1305 |
+
"step": 882
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 23.980132450331126,
|
| 1309 |
+
"eval_accuracy": 0.8475409836065574,
|
| 1310 |
+
"eval_f1_macro": 0.8420048204719258,
|
| 1311 |
+
"eval_f1_micro": 0.8475409836065574,
|
| 1312 |
+
"eval_f1_weighted": 0.8443291816274991,
|
| 1313 |
+
"eval_loss": 0.536198079586029,
|
| 1314 |
+
"eval_precision_macro": 0.86352886002886,
|
| 1315 |
+
"eval_precision_micro": 0.8475409836065574,
|
| 1316 |
+
"eval_precision_weighted": 0.8657139013081634,
|
| 1317 |
+
"eval_recall_macro": 0.8454404761904761,
|
| 1318 |
+
"eval_recall_micro": 0.8475409836065574,
|
| 1319 |
+
"eval_recall_weighted": 0.8475409836065574,
|
| 1320 |
+
"eval_runtime": 41.4502,
|
| 1321 |
+
"eval_samples_per_second": 29.433,
|
| 1322 |
+
"eval_steps_per_second": 0.483,
|
| 1323 |
+
"step": 888
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 24.026490066225165,
|
| 1327 |
+
"grad_norm": 7.926750183105469,
|
| 1328 |
+
"learning_rate": 4.225225225225226e-05,
|
| 1329 |
+
"loss": 0.6482,
|
| 1330 |
+
"step": 889
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 24.211920529801326,
|
| 1334 |
+
"grad_norm": 6.154850006103516,
|
| 1335 |
+
"learning_rate": 4.214714714714715e-05,
|
| 1336 |
+
"loss": 0.5294,
|
| 1337 |
+
"step": 896
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 24.397350993377483,
|
| 1341 |
+
"grad_norm": 7.406444549560547,
|
| 1342 |
+
"learning_rate": 4.204204204204204e-05,
|
| 1343 |
+
"loss": 0.5871,
|
| 1344 |
+
"step": 903
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 24.582781456953644,
|
| 1348 |
+
"grad_norm": 7.258051872253418,
|
| 1349 |
+
"learning_rate": 4.1936936936936936e-05,
|
| 1350 |
+
"loss": 0.582,
|
| 1351 |
+
"step": 910
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 24.7682119205298,
|
| 1355 |
+
"grad_norm": 7.2824859619140625,
|
| 1356 |
+
"learning_rate": 4.1831831831831836e-05,
|
| 1357 |
+
"loss": 0.569,
|
| 1358 |
+
"step": 917
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 24.95364238410596,
|
| 1362 |
+
"grad_norm": 7.095022678375244,
|
| 1363 |
+
"learning_rate": 4.172672672672673e-05,
|
| 1364 |
+
"loss": 0.6194,
|
| 1365 |
+
"step": 924
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 24.980132450331126,
|
| 1369 |
+
"eval_accuracy": 0.8418032786885246,
|
| 1370 |
+
"eval_f1_macro": 0.8358616677440207,
|
| 1371 |
+
"eval_f1_micro": 0.8418032786885246,
|
| 1372 |
+
"eval_f1_weighted": 0.8394348088694761,
|
| 1373 |
+
"eval_loss": 0.53657066822052,
|
| 1374 |
+
"eval_precision_macro": 0.8616504606504606,
|
| 1375 |
+
"eval_precision_micro": 0.8418032786885246,
|
| 1376 |
+
"eval_precision_weighted": 0.8629244162031047,
|
| 1377 |
+
"eval_recall_macro": 0.8374583333333332,
|
| 1378 |
+
"eval_recall_micro": 0.8418032786885246,
|
| 1379 |
+
"eval_recall_weighted": 0.8418032786885246,
|
| 1380 |
+
"eval_runtime": 40.6599,
|
| 1381 |
+
"eval_samples_per_second": 30.005,
|
| 1382 |
+
"eval_steps_per_second": 0.492,
|
| 1383 |
+
"step": 925
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 25.158940397350992,
|
| 1387 |
+
"grad_norm": 8.208196640014648,
|
| 1388 |
+
"learning_rate": 4.162162162162163e-05,
|
| 1389 |
+
"loss": 0.5924,
|
| 1390 |
+
"step": 931
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 25.344370860927153,
|
| 1394 |
+
"grad_norm": 5.883511066436768,
|
| 1395 |
+
"learning_rate": 4.1516516516516515e-05,
|
| 1396 |
+
"loss": 0.5908,
|
| 1397 |
+
"step": 938
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 25.52980132450331,
|
| 1401 |
+
"grad_norm": 7.456212997436523,
|
| 1402 |
+
"learning_rate": 4.1411411411411415e-05,
|
| 1403 |
+
"loss": 0.5531,
|
| 1404 |
+
"step": 945
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 25.71523178807947,
|
| 1408 |
+
"grad_norm": 6.946765422821045,
|
| 1409 |
+
"learning_rate": 4.130630630630631e-05,
|
| 1410 |
+
"loss": 0.6191,
|
| 1411 |
+
"step": 952
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 25.90066225165563,
|
| 1415 |
+
"grad_norm": 5.55208158493042,
|
| 1416 |
+
"learning_rate": 4.12012012012012e-05,
|
| 1417 |
+
"loss": 0.5062,
|
| 1418 |
+
"step": 959
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 25.980132450331126,
|
| 1422 |
+
"eval_accuracy": 0.839344262295082,
|
| 1423 |
+
"eval_f1_macro": 0.8345011108172872,
|
| 1424 |
+
"eval_f1_micro": 0.839344262295082,
|
| 1425 |
+
"eval_f1_weighted": 0.8372446574640403,
|
| 1426 |
+
"eval_loss": 0.5442374348640442,
|
| 1427 |
+
"eval_precision_macro": 0.8636957070707072,
|
| 1428 |
+
"eval_precision_micro": 0.839344262295082,
|
| 1429 |
+
"eval_precision_weighted": 0.8666759219832991,
|
| 1430 |
+
"eval_recall_macro": 0.8376964285714286,
|
| 1431 |
+
"eval_recall_micro": 0.839344262295082,
|
| 1432 |
+
"eval_recall_weighted": 0.839344262295082,
|
| 1433 |
+
"eval_runtime": 40.5992,
|
| 1434 |
+
"eval_samples_per_second": 30.05,
|
| 1435 |
+
"eval_steps_per_second": 0.493,
|
| 1436 |
+
"step": 962
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 26.105960264900663,
|
| 1440 |
+
"grad_norm": 6.749058723449707,
|
| 1441 |
+
"learning_rate": 4.10960960960961e-05,
|
| 1442 |
+
"loss": 0.5717,
|
| 1443 |
+
"step": 966
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 26.29139072847682,
|
| 1447 |
+
"grad_norm": 7.851437568664551,
|
| 1448 |
+
"learning_rate": 4.099099099099099e-05,
|
| 1449 |
+
"loss": 0.594,
|
| 1450 |
+
"step": 973
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 26.47682119205298,
|
| 1454 |
+
"grad_norm": 5.15859842300415,
|
| 1455 |
+
"learning_rate": 4.088588588588589e-05,
|
| 1456 |
+
"loss": 0.5236,
|
| 1457 |
+
"step": 980
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 26.662251655629138,
|
| 1461 |
+
"grad_norm": 6.7099103927612305,
|
| 1462 |
+
"learning_rate": 4.078078078078078e-05,
|
| 1463 |
+
"loss": 0.5225,
|
| 1464 |
+
"step": 987
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 26.8476821192053,
|
| 1468 |
+
"grad_norm": 8.952303886413574,
|
| 1469 |
+
"learning_rate": 4.067567567567568e-05,
|
| 1470 |
+
"loss": 0.536,
|
| 1471 |
+
"step": 994
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 26.980132450331126,
|
| 1475 |
+
"eval_accuracy": 0.8540983606557377,
|
| 1476 |
+
"eval_f1_macro": 0.8494095251154075,
|
| 1477 |
+
"eval_f1_micro": 0.8540983606557377,
|
| 1478 |
+
"eval_f1_weighted": 0.8529669682080483,
|
| 1479 |
+
"eval_loss": 0.5279183983802795,
|
| 1480 |
+
"eval_precision_macro": 0.8678869047619048,
|
| 1481 |
+
"eval_precision_micro": 0.8540983606557377,
|
| 1482 |
+
"eval_precision_weighted": 0.8713430262815508,
|
| 1483 |
+
"eval_recall_macro": 0.851375,
|
| 1484 |
+
"eval_recall_micro": 0.8540983606557377,
|
| 1485 |
+
"eval_recall_weighted": 0.8540983606557377,
|
| 1486 |
+
"eval_runtime": 50.6859,
|
| 1487 |
+
"eval_samples_per_second": 24.07,
|
| 1488 |
+
"eval_steps_per_second": 0.395,
|
| 1489 |
+
"step": 999
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 27.05298013245033,
|
| 1493 |
+
"grad_norm": 6.7562713623046875,
|
| 1494 |
+
"learning_rate": 4.0570570570570574e-05,
|
| 1495 |
+
"loss": 0.556,
|
| 1496 |
+
"step": 1001
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 27.23841059602649,
|
| 1500 |
+
"grad_norm": 8.072733879089355,
|
| 1501 |
+
"learning_rate": 4.046546546546547e-05,
|
| 1502 |
+
"loss": 0.5502,
|
| 1503 |
+
"step": 1008
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 27.423841059602648,
|
| 1507 |
+
"grad_norm": 7.506572246551514,
|
| 1508 |
+
"learning_rate": 4.036036036036036e-05,
|
| 1509 |
+
"loss": 0.5333,
|
| 1510 |
+
"step": 1015
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 27.60927152317881,
|
| 1514 |
+
"grad_norm": 6.526910305023193,
|
| 1515 |
+
"learning_rate": 4.025525525525526e-05,
|
| 1516 |
+
"loss": 0.5411,
|
| 1517 |
+
"step": 1022
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 27.794701986754966,
|
| 1521 |
+
"grad_norm": 7.3132219314575195,
|
| 1522 |
+
"learning_rate": 4.015015015015015e-05,
|
| 1523 |
+
"loss": 0.4913,
|
| 1524 |
+
"step": 1029
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 27.980132450331126,
|
| 1528 |
+
"grad_norm": 6.7607293128967285,
|
| 1529 |
+
"learning_rate": 4.0045045045045046e-05,
|
| 1530 |
+
"loss": 0.4581,
|
| 1531 |
+
"step": 1036
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 27.980132450331126,
|
| 1535 |
+
"eval_accuracy": 0.8450819672131148,
|
| 1536 |
+
"eval_f1_macro": 0.8403489754999043,
|
| 1537 |
+
"eval_f1_micro": 0.8450819672131148,
|
| 1538 |
+
"eval_f1_weighted": 0.8432557118100563,
|
| 1539 |
+
"eval_loss": 0.5285335183143616,
|
| 1540 |
+
"eval_precision_macro": 0.8622770562770563,
|
| 1541 |
+
"eval_precision_micro": 0.8450819672131148,
|
| 1542 |
+
"eval_precision_weighted": 0.8648690653608687,
|
| 1543 |
+
"eval_recall_macro": 0.8421666666666667,
|
| 1544 |
+
"eval_recall_micro": 0.8450819672131148,
|
| 1545 |
+
"eval_recall_weighted": 0.8450819672131148,
|
| 1546 |
+
"eval_runtime": 47.0435,
|
| 1547 |
+
"eval_samples_per_second": 25.933,
|
| 1548 |
+
"eval_steps_per_second": 0.425,
|
| 1549 |
+
"step": 1036
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 28.185430463576157,
|
| 1553 |
+
"grad_norm": 5.862381935119629,
|
| 1554 |
+
"learning_rate": 3.993993993993994e-05,
|
| 1555 |
+
"loss": 0.463,
|
| 1556 |
+
"step": 1043
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 28.370860927152318,
|
| 1560 |
+
"grad_norm": 7.303929328918457,
|
| 1561 |
+
"learning_rate": 3.983483483483484e-05,
|
| 1562 |
+
"loss": 0.4783,
|
| 1563 |
+
"step": 1050
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 28.556291390728475,
|
| 1567 |
+
"grad_norm": 6.46702766418457,
|
| 1568 |
+
"learning_rate": 3.972972972972973e-05,
|
| 1569 |
+
"loss": 0.4826,
|
| 1570 |
+
"step": 1057
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 28.741721854304636,
|
| 1574 |
+
"grad_norm": 7.615893363952637,
|
| 1575 |
+
"learning_rate": 3.962462462462463e-05,
|
| 1576 |
+
"loss": 0.5393,
|
| 1577 |
+
"step": 1064
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 28.927152317880793,
|
| 1581 |
+
"grad_norm": 6.58096170425415,
|
| 1582 |
+
"learning_rate": 3.951951951951952e-05,
|
| 1583 |
+
"loss": 0.5844,
|
| 1584 |
+
"step": 1071
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 28.980132450331126,
|
| 1588 |
+
"eval_accuracy": 0.8377049180327869,
|
| 1589 |
+
"eval_f1_macro": 0.8312956210456212,
|
| 1590 |
+
"eval_f1_micro": 0.8377049180327869,
|
| 1591 |
+
"eval_f1_weighted": 0.8337990174670502,
|
| 1592 |
+
"eval_loss": 0.5218645334243774,
|
| 1593 |
+
"eval_precision_macro": 0.8636145382395383,
|
| 1594 |
+
"eval_precision_micro": 0.8377049180327869,
|
| 1595 |
+
"eval_precision_weighted": 0.8647297329264542,
|
| 1596 |
+
"eval_recall_macro": 0.8355833333333333,
|
| 1597 |
+
"eval_recall_micro": 0.8377049180327869,
|
| 1598 |
+
"eval_recall_weighted": 0.8377049180327869,
|
| 1599 |
+
"eval_runtime": 53.5985,
|
| 1600 |
+
"eval_samples_per_second": 22.762,
|
| 1601 |
+
"eval_steps_per_second": 0.373,
|
| 1602 |
+
"step": 1073
|
| 1603 |
+
}
|
| 1604 |
+
],
|
| 1605 |
+
"logging_steps": 7,
|
| 1606 |
+
"max_steps": 3700,
|
| 1607 |
+
"num_input_tokens_seen": 0,
|
| 1608 |
+
"num_train_epochs": 100,
|
| 1609 |
+
"save_steps": 7,
|
| 1610 |
+
"stateful_callbacks": {
|
| 1611 |
+
"EarlyStoppingCallback": {
|
| 1612 |
+
"args": {
|
| 1613 |
+
"early_stopping_patience": 5,
|
| 1614 |
+
"early_stopping_threshold": 0.01
|
| 1615 |
+
},
|
| 1616 |
+
"attributes": {
|
| 1617 |
+
"early_stopping_patience_counter": 5
|
| 1618 |
+
}
|
| 1619 |
+
},
|
| 1620 |
+
"TrainerControl": {
|
| 1621 |
+
"args": {
|
| 1622 |
+
"should_epoch_stop": false,
|
| 1623 |
+
"should_evaluate": false,
|
| 1624 |
+
"should_log": false,
|
| 1625 |
+
"should_save": true,
|
| 1626 |
+
"should_training_stop": true
|
| 1627 |
+
},
|
| 1628 |
+
"attributes": {}
|
| 1629 |
+
}
|
| 1630 |
+
},
|
| 1631 |
+
"total_flos": 4.523422474750132e+18,
|
| 1632 |
+
"train_batch_size": 32,
|
| 1633 |
+
"trial_name": null,
|
| 1634 |
+
"trial_params": null
|
| 1635 |
+
}
|
checkpoint-1073/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1da0df3ee160ed78a94d79268add1dd382f9eb019c94de90bf1adfdeaace04bb
|
| 3 |
+
size 5304
|