Training in progress, step 4000
Browse files- model.safetensors +1 -1
- runs/Jul09_00-03-18_LAPTOP-1GID9RGH/events.out.tfevents.1720458199.LAPTOP-1GID9RGH.8944.8 +2 -2
- tmp-checkpoint-4000/config.json +248 -0
- tmp-checkpoint-4000/model.safetensors +3 -0
- tmp-checkpoint-4000/optimizer.pt +3 -0
- tmp-checkpoint-4000/preprocessor_config.json +9 -0
- tmp-checkpoint-4000/rng_state.pth +3 -0
- tmp-checkpoint-4000/scheduler.pt +3 -0
- tmp-checkpoint-4000/trainer_state.json +1131 -0
- tmp-checkpoint-4000/training_args.bin +3 -0
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 378386248
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d660e2f4ba7425de092e46febcb6dd5a9162f97de6a8f80031a80fd3e1292646
|
| 3 |
size 378386248
|
runs/Jul09_00-03-18_LAPTOP-1GID9RGH/events.out.tfevents.1720458199.LAPTOP-1GID9RGH.8944.8
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b7189cb4e8869ac61b42f06df636e6771a58cc2120ef2a34efa12e05e1747366
|
| 3 |
+
size 52929
|
tmp-checkpoint-4000/config.json
ADDED
|
@@ -0,0 +1,248 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "facebook/hubert-base-ls960",
|
| 3 |
+
"activation_dropout": 0.1,
|
| 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": false,
|
| 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": false,
|
| 42 |
+
"eos_token_id": 2,
|
| 43 |
+
"feat_extract_activation": "gelu",
|
| 44 |
+
"feat_extract_dropout": 0.0,
|
| 45 |
+
"feat_extract_norm": "group",
|
| 46 |
+
"feat_proj_dropout": 0.1,
|
| 47 |
+
"feat_proj_layer_norm": true,
|
| 48 |
+
"final_dropout": 0.1,
|
| 49 |
+
"gradient_checkpointing": false,
|
| 50 |
+
"hidden_act": "gelu",
|
| 51 |
+
"hidden_dropout": 0.1,
|
| 52 |
+
"hidden_dropout_prob": 0.1,
|
| 53 |
+
"hidden_size": 768,
|
| 54 |
+
"id2label": {
|
| 55 |
+
"0": "0000101",
|
| 56 |
+
"1": "0000110",
|
| 57 |
+
"10": "0010010",
|
| 58 |
+
"11": "0010011",
|
| 59 |
+
"12": "0010101",
|
| 60 |
+
"13": "0010110",
|
| 61 |
+
"14": "0010111",
|
| 62 |
+
"15": "0011001",
|
| 63 |
+
"16": "0011010",
|
| 64 |
+
"17": "0011011",
|
| 65 |
+
"18": "0011101",
|
| 66 |
+
"19": "0011110",
|
| 67 |
+
"2": "0000111",
|
| 68 |
+
"20": "0011111",
|
| 69 |
+
"21": "0100001",
|
| 70 |
+
"22": "0100010",
|
| 71 |
+
"23": "0100011",
|
| 72 |
+
"24": "0100101",
|
| 73 |
+
"25": "0100110",
|
| 74 |
+
"26": "0100111",
|
| 75 |
+
"27": "0101001",
|
| 76 |
+
"28": "0101010",
|
| 77 |
+
"29": "0101011",
|
| 78 |
+
"3": "0001001",
|
| 79 |
+
"30": "0101101",
|
| 80 |
+
"31": "0101110",
|
| 81 |
+
"32": "0101111",
|
| 82 |
+
"33": "0110001",
|
| 83 |
+
"34": "0110010",
|
| 84 |
+
"35": "0110011",
|
| 85 |
+
"36": "0110101",
|
| 86 |
+
"37": "0110110",
|
| 87 |
+
"38": "0110111",
|
| 88 |
+
"39": "0111001",
|
| 89 |
+
"4": "0001010",
|
| 90 |
+
"40": "0111010",
|
| 91 |
+
"41": "0111011",
|
| 92 |
+
"42": "0111101",
|
| 93 |
+
"43": "0111110",
|
| 94 |
+
"44": "0111111",
|
| 95 |
+
"45": "1000001",
|
| 96 |
+
"46": "1000010",
|
| 97 |
+
"47": "1000011",
|
| 98 |
+
"48": "1000101",
|
| 99 |
+
"49": "1000110",
|
| 100 |
+
"5": "0001011",
|
| 101 |
+
"50": "1000111",
|
| 102 |
+
"51": "1001001",
|
| 103 |
+
"52": "1001010",
|
| 104 |
+
"53": "1001011",
|
| 105 |
+
"54": "1001101",
|
| 106 |
+
"55": "1001110",
|
| 107 |
+
"56": "1001111",
|
| 108 |
+
"57": "1010001",
|
| 109 |
+
"58": "1010010",
|
| 110 |
+
"59": "1010011",
|
| 111 |
+
"6": "0001101",
|
| 112 |
+
"60": "1010101",
|
| 113 |
+
"61": "1010110",
|
| 114 |
+
"62": "1010111",
|
| 115 |
+
"63": "1011001",
|
| 116 |
+
"64": "1011010",
|
| 117 |
+
"65": "1011011",
|
| 118 |
+
"66": "1011101",
|
| 119 |
+
"67": "1011110",
|
| 120 |
+
"68": "1011111",
|
| 121 |
+
"69": "1100001",
|
| 122 |
+
"7": "0001110",
|
| 123 |
+
"70": "1100010",
|
| 124 |
+
"71": "1100011",
|
| 125 |
+
"72": "1100101",
|
| 126 |
+
"73": "1100110",
|
| 127 |
+
"74": "1100111",
|
| 128 |
+
"75": "1101001",
|
| 129 |
+
"76": "1101010",
|
| 130 |
+
"77": "1101011",
|
| 131 |
+
"78": "1101101",
|
| 132 |
+
"79": "1101110",
|
| 133 |
+
"8": "0001111",
|
| 134 |
+
"80": "1101111",
|
| 135 |
+
"81": "1110001",
|
| 136 |
+
"82": "1110010",
|
| 137 |
+
"83": "1110011",
|
| 138 |
+
"9": "0010001"
|
| 139 |
+
},
|
| 140 |
+
"initializer_range": 0.02,
|
| 141 |
+
"intermediate_size": 3072,
|
| 142 |
+
"label2id": {
|
| 143 |
+
"0000101": "0",
|
| 144 |
+
"0000110": "1",
|
| 145 |
+
"0000111": "2",
|
| 146 |
+
"0001001": "3",
|
| 147 |
+
"0001010": "4",
|
| 148 |
+
"0001011": "5",
|
| 149 |
+
"0001101": "6",
|
| 150 |
+
"0001110": "7",
|
| 151 |
+
"0001111": "8",
|
| 152 |
+
"0010001": "9",
|
| 153 |
+
"0010010": "10",
|
| 154 |
+
"0010011": "11",
|
| 155 |
+
"0010101": "12",
|
| 156 |
+
"0010110": "13",
|
| 157 |
+
"0010111": "14",
|
| 158 |
+
"0011001": "15",
|
| 159 |
+
"0011010": "16",
|
| 160 |
+
"0011011": "17",
|
| 161 |
+
"0011101": "18",
|
| 162 |
+
"0011110": "19",
|
| 163 |
+
"0011111": "20",
|
| 164 |
+
"0100001": "21",
|
| 165 |
+
"0100010": "22",
|
| 166 |
+
"0100011": "23",
|
| 167 |
+
"0100101": "24",
|
| 168 |
+
"0100110": "25",
|
| 169 |
+
"0100111": "26",
|
| 170 |
+
"0101001": "27",
|
| 171 |
+
"0101010": "28",
|
| 172 |
+
"0101011": "29",
|
| 173 |
+
"0101101": "30",
|
| 174 |
+
"0101110": "31",
|
| 175 |
+
"0101111": "32",
|
| 176 |
+
"0110001": "33",
|
| 177 |
+
"0110010": "34",
|
| 178 |
+
"0110011": "35",
|
| 179 |
+
"0110101": "36",
|
| 180 |
+
"0110110": "37",
|
| 181 |
+
"0110111": "38",
|
| 182 |
+
"0111001": "39",
|
| 183 |
+
"0111010": "40",
|
| 184 |
+
"0111011": "41",
|
| 185 |
+
"0111101": "42",
|
| 186 |
+
"0111110": "43",
|
| 187 |
+
"0111111": "44",
|
| 188 |
+
"1000001": "45",
|
| 189 |
+
"1000010": "46",
|
| 190 |
+
"1000011": "47",
|
| 191 |
+
"1000101": "48",
|
| 192 |
+
"1000110": "49",
|
| 193 |
+
"1000111": "50",
|
| 194 |
+
"1001001": "51",
|
| 195 |
+
"1001010": "52",
|
| 196 |
+
"1001011": "53",
|
| 197 |
+
"1001101": "54",
|
| 198 |
+
"1001110": "55",
|
| 199 |
+
"1001111": "56",
|
| 200 |
+
"1010001": "57",
|
| 201 |
+
"1010010": "58",
|
| 202 |
+
"1010011": "59",
|
| 203 |
+
"1010101": "60",
|
| 204 |
+
"1010110": "61",
|
| 205 |
+
"1010111": "62",
|
| 206 |
+
"1011001": "63",
|
| 207 |
+
"1011010": "64",
|
| 208 |
+
"1011011": "65",
|
| 209 |
+
"1011101": "66",
|
| 210 |
+
"1011110": "67",
|
| 211 |
+
"1011111": "68",
|
| 212 |
+
"1100001": "69",
|
| 213 |
+
"1100010": "70",
|
| 214 |
+
"1100011": "71",
|
| 215 |
+
"1100101": "72",
|
| 216 |
+
"1100110": "73",
|
| 217 |
+
"1100111": "74",
|
| 218 |
+
"1101001": "75",
|
| 219 |
+
"1101010": "76",
|
| 220 |
+
"1101011": "77",
|
| 221 |
+
"1101101": "78",
|
| 222 |
+
"1101110": "79",
|
| 223 |
+
"1101111": "80",
|
| 224 |
+
"1110001": "81",
|
| 225 |
+
"1110010": "82",
|
| 226 |
+
"1110011": "83"
|
| 227 |
+
},
|
| 228 |
+
"layer_norm_eps": 1e-05,
|
| 229 |
+
"layerdrop": 0.1,
|
| 230 |
+
"mask_feature_length": 10,
|
| 231 |
+
"mask_feature_min_masks": 0,
|
| 232 |
+
"mask_feature_prob": 0.0,
|
| 233 |
+
"mask_time_length": 10,
|
| 234 |
+
"mask_time_min_masks": 2,
|
| 235 |
+
"mask_time_prob": 0.05,
|
| 236 |
+
"model_type": "hubert",
|
| 237 |
+
"num_attention_heads": 12,
|
| 238 |
+
"num_conv_pos_embedding_groups": 16,
|
| 239 |
+
"num_conv_pos_embeddings": 128,
|
| 240 |
+
"num_feat_extract_layers": 7,
|
| 241 |
+
"num_hidden_layers": 12,
|
| 242 |
+
"pad_token_id": 0,
|
| 243 |
+
"tokenizer_class": "Wav2Vec2CTCTokenizer",
|
| 244 |
+
"torch_dtype": "float32",
|
| 245 |
+
"transformers_version": "4.38.2",
|
| 246 |
+
"use_weighted_layer_sum": false,
|
| 247 |
+
"vocab_size": 32
|
| 248 |
+
}
|
tmp-checkpoint-4000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d660e2f4ba7425de092e46febcb6dd5a9162f97de6a8f80031a80fd3e1292646
|
| 3 |
+
size 378386248
|
tmp-checkpoint-4000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e30f374e8863f3de830e11606fa008e5faf5f24d67567c2d185d1aaef9d61b59
|
| 3 |
+
size 723289978
|
tmp-checkpoint-4000/preprocessor_config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"feature_extractor_type": "Wav2Vec2FeatureExtractor",
|
| 4 |
+
"feature_size": 1,
|
| 5 |
+
"padding_side": "right",
|
| 6 |
+
"padding_value": 0.0,
|
| 7 |
+
"return_attention_mask": false,
|
| 8 |
+
"sampling_rate": 16000
|
| 9 |
+
}
|
tmp-checkpoint-4000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9609321d39eed828389b64228f351e48770c080465c0ca20848a03ddc9dc4817
|
| 3 |
+
size 14308
|
tmp-checkpoint-4000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d3d630a10c540814c311a2966ecad86e893dd58f8d91138ffb7cb4c0fbb47a42
|
| 3 |
+
size 1064
|
tmp-checkpoint-4000/trainer_state.json
ADDED
|
@@ -0,0 +1,1131 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.8784962349040888,
|
| 3 |
+
"best_model_checkpoint": "fydhfzh/hubert-classifier-aug-fold-4\\checkpoint-4000",
|
| 4 |
+
"epoch": 10.78167115902965,
|
| 5 |
+
"eval_steps": 50,
|
| 6 |
+
"global_step": 4000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.13,
|
| 13 |
+
"eval_accuracy": 0.04588394062078273,
|
| 14 |
+
"eval_binary": 0.3221322537112049,
|
| 15 |
+
"eval_f1": 0.014671635517615288,
|
| 16 |
+
"eval_loss": 3.805781364440918,
|
| 17 |
+
"eval_precision": 0.03779948934318612,
|
| 18 |
+
"eval_recall": 0.04588394062078273,
|
| 19 |
+
"eval_runtime": 1.0685,
|
| 20 |
+
"eval_samples_per_second": 693.469,
|
| 21 |
+
"eval_steps_per_second": 22.461,
|
| 22 |
+
"step": 50
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 0.27,
|
| 26 |
+
"eval_accuracy": 0.1106612685560054,
|
| 27 |
+
"eval_binary": 0.374358974358979,
|
| 28 |
+
"eval_f1": 0.03914518880642342,
|
| 29 |
+
"eval_loss": 3.335998058319092,
|
| 30 |
+
"eval_precision": 0.0301847940928463,
|
| 31 |
+
"eval_recall": 0.1106612685560054,
|
| 32 |
+
"eval_runtime": 1.0612,
|
| 33 |
+
"eval_samples_per_second": 698.296,
|
| 34 |
+
"eval_steps_per_second": 22.617,
|
| 35 |
+
"step": 100
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 0.4,
|
| 39 |
+
"eval_accuracy": 0.23346828609986506,
|
| 40 |
+
"eval_binary": 0.4604588394062126,
|
| 41 |
+
"eval_f1": 0.14009724606429286,
|
| 42 |
+
"eval_loss": 3.007000684738159,
|
| 43 |
+
"eval_precision": 0.12454766261554863,
|
| 44 |
+
"eval_recall": 0.23346828609986506,
|
| 45 |
+
"eval_runtime": 1.0735,
|
| 46 |
+
"eval_samples_per_second": 690.246,
|
| 47 |
+
"eval_steps_per_second": 22.356,
|
| 48 |
+
"step": 150
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 0.54,
|
| 52 |
+
"eval_accuracy": 0.3468286099865047,
|
| 53 |
+
"eval_binary": 0.5406207827260503,
|
| 54 |
+
"eval_f1": 0.25241630871467685,
|
| 55 |
+
"eval_loss": 2.556727647781372,
|
| 56 |
+
"eval_precision": 0.24171104417236225,
|
| 57 |
+
"eval_recall": 0.3468286099865047,
|
| 58 |
+
"eval_runtime": 1.0985,
|
| 59 |
+
"eval_samples_per_second": 674.535,
|
| 60 |
+
"eval_steps_per_second": 21.847,
|
| 61 |
+
"step": 200
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.67,
|
| 65 |
+
"eval_accuracy": 0.42645074224021595,
|
| 66 |
+
"eval_binary": 0.5983805668016237,
|
| 67 |
+
"eval_f1": 0.35188011958617477,
|
| 68 |
+
"eval_loss": 2.177004337310791,
|
| 69 |
+
"eval_precision": 0.3602976489469129,
|
| 70 |
+
"eval_recall": 0.42645074224021595,
|
| 71 |
+
"eval_runtime": 1.1241,
|
| 72 |
+
"eval_samples_per_second": 659.219,
|
| 73 |
+
"eval_steps_per_second": 21.351,
|
| 74 |
+
"step": 250
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.81,
|
| 78 |
+
"eval_accuracy": 0.4966261808367072,
|
| 79 |
+
"eval_binary": 0.647098515519572,
|
| 80 |
+
"eval_f1": 0.43339387931632517,
|
| 81 |
+
"eval_loss": 1.8818293809890747,
|
| 82 |
+
"eval_precision": 0.4889656579714102,
|
| 83 |
+
"eval_recall": 0.4966261808367072,
|
| 84 |
+
"eval_runtime": 1.1323,
|
| 85 |
+
"eval_samples_per_second": 654.399,
|
| 86 |
+
"eval_steps_per_second": 21.195,
|
| 87 |
+
"step": 300
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.94,
|
| 91 |
+
"eval_accuracy": 0.5506072874493927,
|
| 92 |
+
"eval_binary": 0.6852901484480469,
|
| 93 |
+
"eval_f1": 0.5044747772286964,
|
| 94 |
+
"eval_loss": 1.6562384366989136,
|
| 95 |
+
"eval_precision": 0.5479471666625184,
|
| 96 |
+
"eval_recall": 0.5506072874493927,
|
| 97 |
+
"eval_runtime": 1.1248,
|
| 98 |
+
"eval_samples_per_second": 658.769,
|
| 99 |
+
"eval_steps_per_second": 21.337,
|
| 100 |
+
"step": 350
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 1.0,
|
| 104 |
+
"grad_norm": 4.053846836090088,
|
| 105 |
+
"learning_rate": 9.666666666666667e-05,
|
| 106 |
+
"loss": 2.9369,
|
| 107 |
+
"step": 371
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.08,
|
| 111 |
+
"eval_accuracy": 0.6045883940620783,
|
| 112 |
+
"eval_binary": 0.7234817813765205,
|
| 113 |
+
"eval_f1": 0.5598557405312042,
|
| 114 |
+
"eval_loss": 1.4207106828689575,
|
| 115 |
+
"eval_precision": 0.590639979026807,
|
| 116 |
+
"eval_recall": 0.6045883940620783,
|
| 117 |
+
"eval_runtime": 1.1338,
|
| 118 |
+
"eval_samples_per_second": 653.569,
|
| 119 |
+
"eval_steps_per_second": 21.168,
|
| 120 |
+
"step": 400
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 1.21,
|
| 124 |
+
"eval_accuracy": 0.6531713900134952,
|
| 125 |
+
"eval_binary": 0.7587044534412956,
|
| 126 |
+
"eval_f1": 0.6214387578818519,
|
| 127 |
+
"eval_loss": 1.256138801574707,
|
| 128 |
+
"eval_precision": 0.6601188715535435,
|
| 129 |
+
"eval_recall": 0.6531713900134952,
|
| 130 |
+
"eval_runtime": 1.1142,
|
| 131 |
+
"eval_samples_per_second": 665.023,
|
| 132 |
+
"eval_steps_per_second": 21.539,
|
| 133 |
+
"step": 450
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 1.35,
|
| 137 |
+
"eval_accuracy": 0.6653171390013495,
|
| 138 |
+
"eval_binary": 0.7655870445344135,
|
| 139 |
+
"eval_f1": 0.6326851465486375,
|
| 140 |
+
"eval_loss": 1.1726747751235962,
|
| 141 |
+
"eval_precision": 0.6850642147547112,
|
| 142 |
+
"eval_recall": 0.6653171390013495,
|
| 143 |
+
"eval_runtime": 1.176,
|
| 144 |
+
"eval_samples_per_second": 630.094,
|
| 145 |
+
"eval_steps_per_second": 20.408,
|
| 146 |
+
"step": 500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 1.48,
|
| 150 |
+
"eval_accuracy": 0.7300944669365722,
|
| 151 |
+
"eval_binary": 0.809986504723346,
|
| 152 |
+
"eval_f1": 0.7136995231626676,
|
| 153 |
+
"eval_loss": 1.0430372953414917,
|
| 154 |
+
"eval_precision": 0.7479252809718396,
|
| 155 |
+
"eval_recall": 0.7300944669365722,
|
| 156 |
+
"eval_runtime": 1.1554,
|
| 157 |
+
"eval_samples_per_second": 641.328,
|
| 158 |
+
"eval_steps_per_second": 20.772,
|
| 159 |
+
"step": 550
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 1.62,
|
| 163 |
+
"eval_accuracy": 0.728744939271255,
|
| 164 |
+
"eval_binary": 0.809986504723346,
|
| 165 |
+
"eval_f1": 0.7074703954108517,
|
| 166 |
+
"eval_loss": 0.9479284286499023,
|
| 167 |
+
"eval_precision": 0.740343928828029,
|
| 168 |
+
"eval_recall": 0.728744939271255,
|
| 169 |
+
"eval_runtime": 1.1389,
|
| 170 |
+
"eval_samples_per_second": 650.651,
|
| 171 |
+
"eval_steps_per_second": 21.074,
|
| 172 |
+
"step": 600
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.75,
|
| 176 |
+
"eval_accuracy": 0.766531713900135,
|
| 177 |
+
"eval_binary": 0.8377867746288786,
|
| 178 |
+
"eval_f1": 0.7547522208317685,
|
| 179 |
+
"eval_loss": 0.8898987770080566,
|
| 180 |
+
"eval_precision": 0.7879262475416322,
|
| 181 |
+
"eval_recall": 0.766531713900135,
|
| 182 |
+
"eval_runtime": 1.156,
|
| 183 |
+
"eval_samples_per_second": 641.001,
|
| 184 |
+
"eval_steps_per_second": 20.761,
|
| 185 |
+
"step": 650
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.89,
|
| 189 |
+
"eval_accuracy": 0.815114709851552,
|
| 190 |
+
"eval_binary": 0.8699055330634268,
|
| 191 |
+
"eval_f1": 0.8043569052236664,
|
| 192 |
+
"eval_loss": 0.7485046982765198,
|
| 193 |
+
"eval_precision": 0.8233385560308638,
|
| 194 |
+
"eval_recall": 0.815114709851552,
|
| 195 |
+
"eval_runtime": 1.1399,
|
| 196 |
+
"eval_samples_per_second": 650.034,
|
| 197 |
+
"eval_steps_per_second": 21.054,
|
| 198 |
+
"step": 700
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 2.0,
|
| 202 |
+
"grad_norm": 4.031270980834961,
|
| 203 |
+
"learning_rate": 9.333333333333334e-05,
|
| 204 |
+
"loss": 1.3588,
|
| 205 |
+
"step": 742
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 2.02,
|
| 209 |
+
"eval_accuracy": 0.7921727395411606,
|
| 210 |
+
"eval_binary": 0.8553306342780018,
|
| 211 |
+
"eval_f1": 0.7800833014875206,
|
| 212 |
+
"eval_loss": 0.7457680702209473,
|
| 213 |
+
"eval_precision": 0.8196362229763039,
|
| 214 |
+
"eval_recall": 0.7921727395411606,
|
| 215 |
+
"eval_runtime": 1.1503,
|
| 216 |
+
"eval_samples_per_second": 644.196,
|
| 217 |
+
"eval_steps_per_second": 20.865,
|
| 218 |
+
"step": 750
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 2.16,
|
| 222 |
+
"eval_accuracy": 0.8137651821862348,
|
| 223 |
+
"eval_binary": 0.8686909581646418,
|
| 224 |
+
"eval_f1": 0.8106549167288675,
|
| 225 |
+
"eval_loss": 0.7069482803344727,
|
| 226 |
+
"eval_precision": 0.8318749701340793,
|
| 227 |
+
"eval_recall": 0.8137651821862348,
|
| 228 |
+
"eval_runtime": 1.1687,
|
| 229 |
+
"eval_samples_per_second": 634.021,
|
| 230 |
+
"eval_steps_per_second": 20.535,
|
| 231 |
+
"step": 800
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 2.29,
|
| 235 |
+
"eval_accuracy": 0.8137651821862348,
|
| 236 |
+
"eval_binary": 0.8685560053981101,
|
| 237 |
+
"eval_f1": 0.8088124908295262,
|
| 238 |
+
"eval_loss": 0.6896880269050598,
|
| 239 |
+
"eval_precision": 0.8376693376217073,
|
| 240 |
+
"eval_recall": 0.8137651821862348,
|
| 241 |
+
"eval_runtime": 1.1497,
|
| 242 |
+
"eval_samples_per_second": 644.512,
|
| 243 |
+
"eval_steps_per_second": 20.875,
|
| 244 |
+
"step": 850
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 2.43,
|
| 248 |
+
"eval_accuracy": 0.8070175438596491,
|
| 249 |
+
"eval_binary": 0.8650472334682855,
|
| 250 |
+
"eval_f1": 0.8034708035750392,
|
| 251 |
+
"eval_loss": 0.6972086429595947,
|
| 252 |
+
"eval_precision": 0.8367843338997185,
|
| 253 |
+
"eval_recall": 0.8070175438596491,
|
| 254 |
+
"eval_runtime": 1.1633,
|
| 255 |
+
"eval_samples_per_second": 636.983,
|
| 256 |
+
"eval_steps_per_second": 20.631,
|
| 257 |
+
"step": 900
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 2.56,
|
| 261 |
+
"eval_accuracy": 0.8164642375168691,
|
| 262 |
+
"eval_binary": 0.8704453441295538,
|
| 263 |
+
"eval_f1": 0.813993268748201,
|
| 264 |
+
"eval_loss": 0.6781038641929626,
|
| 265 |
+
"eval_precision": 0.8366608032093864,
|
| 266 |
+
"eval_recall": 0.8164642375168691,
|
| 267 |
+
"eval_runtime": 1.1823,
|
| 268 |
+
"eval_samples_per_second": 626.741,
|
| 269 |
+
"eval_steps_per_second": 20.299,
|
| 270 |
+
"step": 950
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 2.7,
|
| 274 |
+
"eval_accuracy": 0.815114709851552,
|
| 275 |
+
"eval_binary": 0.8703103913630216,
|
| 276 |
+
"eval_f1": 0.8100265350585438,
|
| 277 |
+
"eval_loss": 0.6554349660873413,
|
| 278 |
+
"eval_precision": 0.8380368300914861,
|
| 279 |
+
"eval_recall": 0.815114709851552,
|
| 280 |
+
"eval_runtime": 1.1543,
|
| 281 |
+
"eval_samples_per_second": 641.937,
|
| 282 |
+
"eval_steps_per_second": 20.791,
|
| 283 |
+
"step": 1000
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 2.83,
|
| 287 |
+
"eval_accuracy": 0.8218623481781376,
|
| 288 |
+
"eval_binary": 0.8759784075573539,
|
| 289 |
+
"eval_f1": 0.8182895287255976,
|
| 290 |
+
"eval_loss": 0.6537468433380127,
|
| 291 |
+
"eval_precision": 0.8421630941099861,
|
| 292 |
+
"eval_recall": 0.8218623481781376,
|
| 293 |
+
"eval_runtime": 1.1679,
|
| 294 |
+
"eval_samples_per_second": 634.474,
|
| 295 |
+
"eval_steps_per_second": 20.55,
|
| 296 |
+
"step": 1050
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.96,
|
| 300 |
+
"eval_accuracy": 0.8407557354925776,
|
| 301 |
+
"eval_binary": 0.8882591093117402,
|
| 302 |
+
"eval_f1": 0.8327839441497319,
|
| 303 |
+
"eval_loss": 0.6294392347335815,
|
| 304 |
+
"eval_precision": 0.8517352283784768,
|
| 305 |
+
"eval_recall": 0.8407557354925776,
|
| 306 |
+
"eval_runtime": 1.1945,
|
| 307 |
+
"eval_samples_per_second": 620.327,
|
| 308 |
+
"eval_steps_per_second": 20.092,
|
| 309 |
+
"step": 1100
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 3.0,
|
| 313 |
+
"grad_norm": 4.313427925109863,
|
| 314 |
+
"learning_rate": 9e-05,
|
| 315 |
+
"loss": 0.8613,
|
| 316 |
+
"step": 1113
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 3.1,
|
| 320 |
+
"eval_accuracy": 0.8340080971659919,
|
| 321 |
+
"eval_binary": 0.8840755735492567,
|
| 322 |
+
"eval_f1": 0.8305322924524646,
|
| 323 |
+
"eval_loss": 0.5901665687561035,
|
| 324 |
+
"eval_precision": 0.849580143203625,
|
| 325 |
+
"eval_recall": 0.8340080971659919,
|
| 326 |
+
"eval_runtime": 1.1609,
|
| 327 |
+
"eval_samples_per_second": 638.313,
|
| 328 |
+
"eval_steps_per_second": 20.674,
|
| 329 |
+
"step": 1150
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 3.23,
|
| 333 |
+
"eval_accuracy": 0.8286099865047234,
|
| 334 |
+
"eval_binary": 0.8800269905533059,
|
| 335 |
+
"eval_f1": 0.8244801946311869,
|
| 336 |
+
"eval_loss": 0.6372344493865967,
|
| 337 |
+
"eval_precision": 0.8429155377535945,
|
| 338 |
+
"eval_recall": 0.8286099865047234,
|
| 339 |
+
"eval_runtime": 1.1722,
|
| 340 |
+
"eval_samples_per_second": 632.156,
|
| 341 |
+
"eval_steps_per_second": 20.475,
|
| 342 |
+
"step": 1200
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 3.37,
|
| 346 |
+
"eval_accuracy": 0.8502024291497976,
|
| 347 |
+
"eval_binary": 0.8960863697705795,
|
| 348 |
+
"eval_f1": 0.8459444607208269,
|
| 349 |
+
"eval_loss": 0.5361122488975525,
|
| 350 |
+
"eval_precision": 0.8694864046584694,
|
| 351 |
+
"eval_recall": 0.8502024291497976,
|
| 352 |
+
"eval_runtime": 1.1813,
|
| 353 |
+
"eval_samples_per_second": 627.265,
|
| 354 |
+
"eval_steps_per_second": 20.316,
|
| 355 |
+
"step": 1250
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 3.5,
|
| 359 |
+
"eval_accuracy": 0.8690958164642375,
|
| 360 |
+
"eval_binary": 0.9082321187584337,
|
| 361 |
+
"eval_f1": 0.8673805683180668,
|
| 362 |
+
"eval_loss": 0.5066567659378052,
|
| 363 |
+
"eval_precision": 0.8832905286346582,
|
| 364 |
+
"eval_recall": 0.8690958164642375,
|
| 365 |
+
"eval_runtime": 1.1611,
|
| 366 |
+
"eval_samples_per_second": 638.193,
|
| 367 |
+
"eval_steps_per_second": 20.67,
|
| 368 |
+
"step": 1300
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 3.64,
|
| 372 |
+
"eval_accuracy": 0.8690958164642375,
|
| 373 |
+
"eval_binary": 0.908636977058029,
|
| 374 |
+
"eval_f1": 0.8662532335354076,
|
| 375 |
+
"eval_loss": 0.4976394772529602,
|
| 376 |
+
"eval_precision": 0.8768591769603915,
|
| 377 |
+
"eval_recall": 0.8690958164642375,
|
| 378 |
+
"eval_runtime": 1.1871,
|
| 379 |
+
"eval_samples_per_second": 624.2,
|
| 380 |
+
"eval_steps_per_second": 20.217,
|
| 381 |
+
"step": 1350
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 3.77,
|
| 385 |
+
"eval_accuracy": 0.863697705802969,
|
| 386 |
+
"eval_binary": 0.9044534412955457,
|
| 387 |
+
"eval_f1": 0.858932558800567,
|
| 388 |
+
"eval_loss": 0.5320582985877991,
|
| 389 |
+
"eval_precision": 0.8738656979547668,
|
| 390 |
+
"eval_recall": 0.863697705802969,
|
| 391 |
+
"eval_runtime": 1.1707,
|
| 392 |
+
"eval_samples_per_second": 632.936,
|
| 393 |
+
"eval_steps_per_second": 20.5,
|
| 394 |
+
"step": 1400
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 3.91,
|
| 398 |
+
"eval_accuracy": 0.8529014844804319,
|
| 399 |
+
"eval_binary": 0.8963562753036428,
|
| 400 |
+
"eval_f1": 0.8490350520713121,
|
| 401 |
+
"eval_loss": 0.6004767417907715,
|
| 402 |
+
"eval_precision": 0.8682368800789853,
|
| 403 |
+
"eval_recall": 0.8529014844804319,
|
| 404 |
+
"eval_runtime": 1.1646,
|
| 405 |
+
"eval_samples_per_second": 636.271,
|
| 406 |
+
"eval_steps_per_second": 20.608,
|
| 407 |
+
"step": 1450
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 4.0,
|
| 411 |
+
"grad_norm": 6.45872163772583,
|
| 412 |
+
"learning_rate": 8.666666666666667e-05,
|
| 413 |
+
"loss": 0.6427,
|
| 414 |
+
"step": 1484
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 4.04,
|
| 418 |
+
"eval_accuracy": 0.8609986504723347,
|
| 419 |
+
"eval_binary": 0.9025641025641017,
|
| 420 |
+
"eval_f1": 0.8585759542711945,
|
| 421 |
+
"eval_loss": 0.5946248173713684,
|
| 422 |
+
"eval_precision": 0.8751688934053783,
|
| 423 |
+
"eval_recall": 0.8609986504723347,
|
| 424 |
+
"eval_runtime": 1.1421,
|
| 425 |
+
"eval_samples_per_second": 648.789,
|
| 426 |
+
"eval_steps_per_second": 21.013,
|
| 427 |
+
"step": 1500
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 4.18,
|
| 431 |
+
"eval_accuracy": 0.8690958164642375,
|
| 432 |
+
"eval_binary": 0.9079622132253704,
|
| 433 |
+
"eval_f1": 0.8669378239477148,
|
| 434 |
+
"eval_loss": 0.5153417587280273,
|
| 435 |
+
"eval_precision": 0.8819396228303111,
|
| 436 |
+
"eval_recall": 0.8690958164642375,
|
| 437 |
+
"eval_runtime": 1.1485,
|
| 438 |
+
"eval_samples_per_second": 645.203,
|
| 439 |
+
"eval_steps_per_second": 20.897,
|
| 440 |
+
"step": 1550
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 4.31,
|
| 444 |
+
"eval_accuracy": 0.8569500674763832,
|
| 445 |
+
"eval_binary": 0.8986504723346818,
|
| 446 |
+
"eval_f1": 0.855017890388154,
|
| 447 |
+
"eval_loss": 0.569421648979187,
|
| 448 |
+
"eval_precision": 0.8722095081406823,
|
| 449 |
+
"eval_recall": 0.8569500674763832,
|
| 450 |
+
"eval_runtime": 1.1556,
|
| 451 |
+
"eval_samples_per_second": 641.21,
|
| 452 |
+
"eval_steps_per_second": 20.768,
|
| 453 |
+
"step": 1600
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 4.45,
|
| 457 |
+
"eval_accuracy": 0.8717948717948718,
|
| 458 |
+
"eval_binary": 0.9094466936572189,
|
| 459 |
+
"eval_f1": 0.869835417212902,
|
| 460 |
+
"eval_loss": 0.5046663880348206,
|
| 461 |
+
"eval_precision": 0.8909411654648616,
|
| 462 |
+
"eval_recall": 0.8717948717948718,
|
| 463 |
+
"eval_runtime": 1.1301,
|
| 464 |
+
"eval_samples_per_second": 655.718,
|
| 465 |
+
"eval_steps_per_second": 21.238,
|
| 466 |
+
"step": 1650
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 4.58,
|
| 470 |
+
"eval_accuracy": 0.8717948717948718,
|
| 471 |
+
"eval_binary": 0.9099865047233455,
|
| 472 |
+
"eval_f1": 0.8699350147474851,
|
| 473 |
+
"eval_loss": 0.5270289182662964,
|
| 474 |
+
"eval_precision": 0.8804698038503707,
|
| 475 |
+
"eval_recall": 0.8717948717948718,
|
| 476 |
+
"eval_runtime": 1.1458,
|
| 477 |
+
"eval_samples_per_second": 646.682,
|
| 478 |
+
"eval_steps_per_second": 20.945,
|
| 479 |
+
"step": 1700
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 4.72,
|
| 483 |
+
"eval_accuracy": 0.8569500674763832,
|
| 484 |
+
"eval_binary": 0.9001349527665304,
|
| 485 |
+
"eval_f1": 0.8530294562245342,
|
| 486 |
+
"eval_loss": 0.5278229117393494,
|
| 487 |
+
"eval_precision": 0.8713434076794401,
|
| 488 |
+
"eval_recall": 0.8569500674763832,
|
| 489 |
+
"eval_runtime": 1.1651,
|
| 490 |
+
"eval_samples_per_second": 635.992,
|
| 491 |
+
"eval_steps_per_second": 20.599,
|
| 492 |
+
"step": 1750
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 4.85,
|
| 496 |
+
"eval_accuracy": 0.8771929824561403,
|
| 497 |
+
"eval_binary": 0.9148448043184877,
|
| 498 |
+
"eval_f1": 0.8765031109078509,
|
| 499 |
+
"eval_loss": 0.5112797021865845,
|
| 500 |
+
"eval_precision": 0.8899662303508457,
|
| 501 |
+
"eval_recall": 0.8771929824561403,
|
| 502 |
+
"eval_runtime": 1.161,
|
| 503 |
+
"eval_samples_per_second": 638.269,
|
| 504 |
+
"eval_steps_per_second": 20.673,
|
| 505 |
+
"step": 1800
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 4.99,
|
| 509 |
+
"eval_accuracy": 0.8677462887989204,
|
| 510 |
+
"eval_binary": 0.90715249662618,
|
| 511 |
+
"eval_f1": 0.864758306975439,
|
| 512 |
+
"eval_loss": 0.5557437539100647,
|
| 513 |
+
"eval_precision": 0.8762526459894882,
|
| 514 |
+
"eval_recall": 0.8677462887989204,
|
| 515 |
+
"eval_runtime": 1.2253,
|
| 516 |
+
"eval_samples_per_second": 604.742,
|
| 517 |
+
"eval_steps_per_second": 19.587,
|
| 518 |
+
"step": 1850
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 5.0,
|
| 522 |
+
"grad_norm": 4.927316188812256,
|
| 523 |
+
"learning_rate": 8.333333333333334e-05,
|
| 524 |
+
"loss": 0.5222,
|
| 525 |
+
"step": 1855
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 5.12,
|
| 529 |
+
"eval_accuracy": 0.8717948717948718,
|
| 530 |
+
"eval_binary": 0.9093117408906874,
|
| 531 |
+
"eval_f1": 0.8695718983671247,
|
| 532 |
+
"eval_loss": 0.5396840572357178,
|
| 533 |
+
"eval_precision": 0.8828931646269107,
|
| 534 |
+
"eval_recall": 0.8717948717948718,
|
| 535 |
+
"eval_runtime": 1.1447,
|
| 536 |
+
"eval_samples_per_second": 647.325,
|
| 537 |
+
"eval_steps_per_second": 20.966,
|
| 538 |
+
"step": 1900
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 5.26,
|
| 542 |
+
"eval_accuracy": 0.8785425101214575,
|
| 543 |
+
"eval_binary": 0.9141700404858291,
|
| 544 |
+
"eval_f1": 0.8778235100479747,
|
| 545 |
+
"eval_loss": 0.5459949970245361,
|
| 546 |
+
"eval_precision": 0.8931182365392891,
|
| 547 |
+
"eval_recall": 0.8785425101214575,
|
| 548 |
+
"eval_runtime": 1.1802,
|
| 549 |
+
"eval_samples_per_second": 627.864,
|
| 550 |
+
"eval_steps_per_second": 20.336,
|
| 551 |
+
"step": 1950
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 5.39,
|
| 555 |
+
"eval_accuracy": 0.8717948717948718,
|
| 556 |
+
"eval_binary": 0.909986504723346,
|
| 557 |
+
"eval_f1": 0.8702993779030054,
|
| 558 |
+
"eval_loss": 0.5364383459091187,
|
| 559 |
+
"eval_precision": 0.88322276389078,
|
| 560 |
+
"eval_recall": 0.8717948717948718,
|
| 561 |
+
"eval_runtime": 1.1678,
|
| 562 |
+
"eval_samples_per_second": 634.529,
|
| 563 |
+
"eval_steps_per_second": 20.552,
|
| 564 |
+
"step": 2000
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 5.53,
|
| 568 |
+
"eval_accuracy": 0.8717948717948718,
|
| 569 |
+
"eval_binary": 0.909986504723346,
|
| 570 |
+
"eval_f1": 0.8693471297847463,
|
| 571 |
+
"eval_loss": 0.5403723120689392,
|
| 572 |
+
"eval_precision": 0.8857783354746918,
|
| 573 |
+
"eval_recall": 0.8717948717948718,
|
| 574 |
+
"eval_runtime": 1.2223,
|
| 575 |
+
"eval_samples_per_second": 606.259,
|
| 576 |
+
"eval_steps_per_second": 19.636,
|
| 577 |
+
"step": 2050
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 5.66,
|
| 581 |
+
"eval_accuracy": 0.8690958164642375,
|
| 582 |
+
"eval_binary": 0.9071524966261797,
|
| 583 |
+
"eval_f1": 0.8663898349797059,
|
| 584 |
+
"eval_loss": 0.6065850257873535,
|
| 585 |
+
"eval_precision": 0.8798577528334613,
|
| 586 |
+
"eval_recall": 0.8690958164642375,
|
| 587 |
+
"eval_runtime": 1.1672,
|
| 588 |
+
"eval_samples_per_second": 634.865,
|
| 589 |
+
"eval_steps_per_second": 20.562,
|
| 590 |
+
"step": 2100
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 5.8,
|
| 594 |
+
"eval_accuracy": 0.8569500674763832,
|
| 595 |
+
"eval_binary": 0.8995951417004041,
|
| 596 |
+
"eval_f1": 0.8543825643734608,
|
| 597 |
+
"eval_loss": 0.609086275100708,
|
| 598 |
+
"eval_precision": 0.8699245573739501,
|
| 599 |
+
"eval_recall": 0.8569500674763832,
|
| 600 |
+
"eval_runtime": 1.1615,
|
| 601 |
+
"eval_samples_per_second": 637.949,
|
| 602 |
+
"eval_steps_per_second": 20.662,
|
| 603 |
+
"step": 2150
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 5.93,
|
| 607 |
+
"eval_accuracy": 0.863697705802969,
|
| 608 |
+
"eval_binary": 0.9044534412955451,
|
| 609 |
+
"eval_f1": 0.8615752789891996,
|
| 610 |
+
"eval_loss": 0.5661621689796448,
|
| 611 |
+
"eval_precision": 0.8777716435611175,
|
| 612 |
+
"eval_recall": 0.863697705802969,
|
| 613 |
+
"eval_runtime": 1.1549,
|
| 614 |
+
"eval_samples_per_second": 641.6,
|
| 615 |
+
"eval_steps_per_second": 20.781,
|
| 616 |
+
"step": 2200
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 6.0,
|
| 620 |
+
"grad_norm": 3.95198392868042,
|
| 621 |
+
"learning_rate": 8e-05,
|
| 622 |
+
"loss": 0.4432,
|
| 623 |
+
"step": 2226
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 6.06,
|
| 627 |
+
"eval_accuracy": 0.8623481781376519,
|
| 628 |
+
"eval_binary": 0.9033738191632924,
|
| 629 |
+
"eval_f1": 0.8596934053855256,
|
| 630 |
+
"eval_loss": 0.6643653512001038,
|
| 631 |
+
"eval_precision": 0.8785115319528275,
|
| 632 |
+
"eval_recall": 0.8623481781376519,
|
| 633 |
+
"eval_runtime": 1.195,
|
| 634 |
+
"eval_samples_per_second": 620.059,
|
| 635 |
+
"eval_steps_per_second": 20.083,
|
| 636 |
+
"step": 2250
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 6.2,
|
| 640 |
+
"eval_accuracy": 0.8609986504723347,
|
| 641 |
+
"eval_binary": 0.9020242914979746,
|
| 642 |
+
"eval_f1": 0.858697058942677,
|
| 643 |
+
"eval_loss": 0.6462463736534119,
|
| 644 |
+
"eval_precision": 0.874255314336286,
|
| 645 |
+
"eval_recall": 0.8609986504723347,
|
| 646 |
+
"eval_runtime": 1.1742,
|
| 647 |
+
"eval_samples_per_second": 631.066,
|
| 648 |
+
"eval_steps_per_second": 20.439,
|
| 649 |
+
"step": 2300
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 6.33,
|
| 653 |
+
"eval_accuracy": 0.8771929824561403,
|
| 654 |
+
"eval_binary": 0.9130904183535752,
|
| 655 |
+
"eval_f1": 0.8725328911535255,
|
| 656 |
+
"eval_loss": 0.5374723672866821,
|
| 657 |
+
"eval_precision": 0.8823031219792354,
|
| 658 |
+
"eval_recall": 0.8771929824561403,
|
| 659 |
+
"eval_runtime": 1.1706,
|
| 660 |
+
"eval_samples_per_second": 632.998,
|
| 661 |
+
"eval_steps_per_second": 20.502,
|
| 662 |
+
"step": 2350
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 6.47,
|
| 666 |
+
"eval_accuracy": 0.863697705802969,
|
| 667 |
+
"eval_binary": 0.9044534412955454,
|
| 668 |
+
"eval_f1": 0.8571119491194448,
|
| 669 |
+
"eval_loss": 0.6624985337257385,
|
| 670 |
+
"eval_precision": 0.8729715216557322,
|
| 671 |
+
"eval_recall": 0.863697705802969,
|
| 672 |
+
"eval_runtime": 1.1671,
|
| 673 |
+
"eval_samples_per_second": 634.923,
|
| 674 |
+
"eval_steps_per_second": 20.564,
|
| 675 |
+
"step": 2400
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 6.6,
|
| 679 |
+
"eval_accuracy": 0.8704453441295547,
|
| 680 |
+
"eval_binary": 0.9105263157894729,
|
| 681 |
+
"eval_f1": 0.8676961482191342,
|
| 682 |
+
"eval_loss": 0.5533118844032288,
|
| 683 |
+
"eval_precision": 0.880092126347187,
|
| 684 |
+
"eval_recall": 0.8704453441295547,
|
| 685 |
+
"eval_runtime": 1.1466,
|
| 686 |
+
"eval_samples_per_second": 646.275,
|
| 687 |
+
"eval_steps_per_second": 20.932,
|
| 688 |
+
"step": 2450
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 6.74,
|
| 692 |
+
"eval_accuracy": 0.8744939271255061,
|
| 693 |
+
"eval_binary": 0.9122807017543849,
|
| 694 |
+
"eval_f1": 0.871742048083642,
|
| 695 |
+
"eval_loss": 0.6021706461906433,
|
| 696 |
+
"eval_precision": 0.8837557104763583,
|
| 697 |
+
"eval_recall": 0.8744939271255061,
|
| 698 |
+
"eval_runtime": 1.1589,
|
| 699 |
+
"eval_samples_per_second": 639.39,
|
| 700 |
+
"eval_steps_per_second": 20.709,
|
| 701 |
+
"step": 2500
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 6.87,
|
| 705 |
+
"eval_accuracy": 0.8663967611336032,
|
| 706 |
+
"eval_binary": 0.9063427800269895,
|
| 707 |
+
"eval_f1": 0.8625935560840108,
|
| 708 |
+
"eval_loss": 0.6515318751335144,
|
| 709 |
+
"eval_precision": 0.8741236309050073,
|
| 710 |
+
"eval_recall": 0.8663967611336032,
|
| 711 |
+
"eval_runtime": 1.1683,
|
| 712 |
+
"eval_samples_per_second": 634.261,
|
| 713 |
+
"eval_steps_per_second": 20.543,
|
| 714 |
+
"step": 2550
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 7.0,
|
| 718 |
+
"grad_norm": 4.3103928565979,
|
| 719 |
+
"learning_rate": 7.666666666666667e-05,
|
| 720 |
+
"loss": 0.3999,
|
| 721 |
+
"step": 2597
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 7.01,
|
| 725 |
+
"eval_accuracy": 0.8717948717948718,
|
| 726 |
+
"eval_binary": 0.9103913630229407,
|
| 727 |
+
"eval_f1": 0.8700599583697827,
|
| 728 |
+
"eval_loss": 0.5570635795593262,
|
| 729 |
+
"eval_precision": 0.8831367238249829,
|
| 730 |
+
"eval_recall": 0.8717948717948718,
|
| 731 |
+
"eval_runtime": 1.1765,
|
| 732 |
+
"eval_samples_per_second": 629.854,
|
| 733 |
+
"eval_steps_per_second": 20.4,
|
| 734 |
+
"step": 2600
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 7.14,
|
| 738 |
+
"eval_accuracy": 0.8785425101214575,
|
| 739 |
+
"eval_binary": 0.9151147098515511,
|
| 740 |
+
"eval_f1": 0.8766044389904654,
|
| 741 |
+
"eval_loss": 0.5620068311691284,
|
| 742 |
+
"eval_precision": 0.8879959446761065,
|
| 743 |
+
"eval_recall": 0.8785425101214575,
|
| 744 |
+
"eval_runtime": 1.1587,
|
| 745 |
+
"eval_samples_per_second": 639.508,
|
| 746 |
+
"eval_steps_per_second": 20.713,
|
| 747 |
+
"step": 2650
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 7.28,
|
| 751 |
+
"eval_accuracy": 0.8798920377867746,
|
| 752 |
+
"eval_binary": 0.9160593792172727,
|
| 753 |
+
"eval_f1": 0.8778735845274245,
|
| 754 |
+
"eval_loss": 0.5602428317070007,
|
| 755 |
+
"eval_precision": 0.887021746434702,
|
| 756 |
+
"eval_recall": 0.8798920377867746,
|
| 757 |
+
"eval_runtime": 1.1417,
|
| 758 |
+
"eval_samples_per_second": 649.02,
|
| 759 |
+
"eval_steps_per_second": 21.021,
|
| 760 |
+
"step": 2700
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 7.41,
|
| 764 |
+
"eval_accuracy": 0.8758434547908233,
|
| 765 |
+
"eval_binary": 0.9122807017543852,
|
| 766 |
+
"eval_f1": 0.8734723052118281,
|
| 767 |
+
"eval_loss": 0.5291705131530762,
|
| 768 |
+
"eval_precision": 0.884985605937023,
|
| 769 |
+
"eval_recall": 0.8758434547908233,
|
| 770 |
+
"eval_runtime": 1.1639,
|
| 771 |
+
"eval_samples_per_second": 636.647,
|
| 772 |
+
"eval_steps_per_second": 20.62,
|
| 773 |
+
"step": 2750
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 7.55,
|
| 777 |
+
"eval_accuracy": 0.873144399460189,
|
| 778 |
+
"eval_binary": 0.9109311740890679,
|
| 779 |
+
"eval_f1": 0.8728506306960341,
|
| 780 |
+
"eval_loss": 0.6643885374069214,
|
| 781 |
+
"eval_precision": 0.8896590139505118,
|
| 782 |
+
"eval_recall": 0.873144399460189,
|
| 783 |
+
"eval_runtime": 1.157,
|
| 784 |
+
"eval_samples_per_second": 640.431,
|
| 785 |
+
"eval_steps_per_second": 20.743,
|
| 786 |
+
"step": 2800
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 7.68,
|
| 790 |
+
"eval_accuracy": 0.8771929824561403,
|
| 791 |
+
"eval_binary": 0.9132253711201073,
|
| 792 |
+
"eval_f1": 0.8755511463124777,
|
| 793 |
+
"eval_loss": 0.587805449962616,
|
| 794 |
+
"eval_precision": 0.8871340834903588,
|
| 795 |
+
"eval_recall": 0.8771929824561403,
|
| 796 |
+
"eval_runtime": 1.1575,
|
| 797 |
+
"eval_samples_per_second": 640.186,
|
| 798 |
+
"eval_steps_per_second": 20.735,
|
| 799 |
+
"step": 2850
|
| 800 |
+
},
|
| 801 |
+
{
|
| 802 |
+
"epoch": 7.82,
|
| 803 |
+
"eval_accuracy": 0.8677462887989204,
|
| 804 |
+
"eval_binary": 0.9062078272604579,
|
| 805 |
+
"eval_f1": 0.8663130912154735,
|
| 806 |
+
"eval_loss": 0.6276022791862488,
|
| 807 |
+
"eval_precision": 0.8797659493003622,
|
| 808 |
+
"eval_recall": 0.8677462887989204,
|
| 809 |
+
"eval_runtime": 1.1612,
|
| 810 |
+
"eval_samples_per_second": 638.108,
|
| 811 |
+
"eval_steps_per_second": 20.667,
|
| 812 |
+
"step": 2900
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 7.95,
|
| 816 |
+
"eval_accuracy": 0.873144399460189,
|
| 817 |
+
"eval_binary": 0.9124156545209167,
|
| 818 |
+
"eval_f1": 0.8719900033412215,
|
| 819 |
+
"eval_loss": 0.5735969543457031,
|
| 820 |
+
"eval_precision": 0.8796167283009387,
|
| 821 |
+
"eval_recall": 0.873144399460189,
|
| 822 |
+
"eval_runtime": 1.2203,
|
| 823 |
+
"eval_samples_per_second": 607.227,
|
| 824 |
+
"eval_steps_per_second": 19.667,
|
| 825 |
+
"step": 2950
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 8.0,
|
| 829 |
+
"grad_norm": 5.361498832702637,
|
| 830 |
+
"learning_rate": 7.333333333333333e-05,
|
| 831 |
+
"loss": 0.3609,
|
| 832 |
+
"step": 2968
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 8.09,
|
| 836 |
+
"eval_accuracy": 0.8771929824561403,
|
| 837 |
+
"eval_binary": 0.9143049932523605,
|
| 838 |
+
"eval_f1": 0.8768958010180178,
|
| 839 |
+
"eval_loss": 0.6093275547027588,
|
| 840 |
+
"eval_precision": 0.8875668341255385,
|
| 841 |
+
"eval_recall": 0.8771929824561403,
|
| 842 |
+
"eval_runtime": 1.1465,
|
| 843 |
+
"eval_samples_per_second": 646.287,
|
| 844 |
+
"eval_steps_per_second": 20.932,
|
| 845 |
+
"step": 3000
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 8.22,
|
| 849 |
+
"eval_accuracy": 0.873144399460189,
|
| 850 |
+
"eval_binary": 0.9099865047233455,
|
| 851 |
+
"eval_f1": 0.8728942603104318,
|
| 852 |
+
"eval_loss": 0.5862379670143127,
|
| 853 |
+
"eval_precision": 0.8861477141841515,
|
| 854 |
+
"eval_recall": 0.873144399460189,
|
| 855 |
+
"eval_runtime": 1.1249,
|
| 856 |
+
"eval_samples_per_second": 658.698,
|
| 857 |
+
"eval_steps_per_second": 21.334,
|
| 858 |
+
"step": 3050
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 8.36,
|
| 862 |
+
"eval_accuracy": 0.8839406207827261,
|
| 863 |
+
"eval_binary": 0.9179487179487168,
|
| 864 |
+
"eval_f1": 0.8812727621395834,
|
| 865 |
+
"eval_loss": 0.5736364126205444,
|
| 866 |
+
"eval_precision": 0.8869233630367234,
|
| 867 |
+
"eval_recall": 0.8839406207827261,
|
| 868 |
+
"eval_runtime": 1.1449,
|
| 869 |
+
"eval_samples_per_second": 647.229,
|
| 870 |
+
"eval_steps_per_second": 20.963,
|
| 871 |
+
"step": 3100
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 8.49,
|
| 875 |
+
"eval_accuracy": 0.8798920377867746,
|
| 876 |
+
"eval_binary": 0.915114709851551,
|
| 877 |
+
"eval_f1": 0.8781431747563492,
|
| 878 |
+
"eval_loss": 0.5536712408065796,
|
| 879 |
+
"eval_precision": 0.8903398258661418,
|
| 880 |
+
"eval_recall": 0.8798920377867746,
|
| 881 |
+
"eval_runtime": 1.128,
|
| 882 |
+
"eval_samples_per_second": 656.932,
|
| 883 |
+
"eval_steps_per_second": 21.277,
|
| 884 |
+
"step": 3150
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 8.63,
|
| 888 |
+
"eval_accuracy": 0.8771929824561403,
|
| 889 |
+
"eval_binary": 0.9137651821862336,
|
| 890 |
+
"eval_f1": 0.8748654521289038,
|
| 891 |
+
"eval_loss": 0.5788822770118713,
|
| 892 |
+
"eval_precision": 0.8854517989133374,
|
| 893 |
+
"eval_recall": 0.8771929824561403,
|
| 894 |
+
"eval_runtime": 1.1554,
|
| 895 |
+
"eval_samples_per_second": 641.322,
|
| 896 |
+
"eval_steps_per_second": 20.772,
|
| 897 |
+
"step": 3200
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 8.76,
|
| 901 |
+
"eval_accuracy": 0.8758434547908233,
|
| 902 |
+
"eval_binary": 0.9129554655870437,
|
| 903 |
+
"eval_f1": 0.873432972085747,
|
| 904 |
+
"eval_loss": 0.6053817272186279,
|
| 905 |
+
"eval_precision": 0.8816829856303541,
|
| 906 |
+
"eval_recall": 0.8758434547908233,
|
| 907 |
+
"eval_runtime": 1.1318,
|
| 908 |
+
"eval_samples_per_second": 654.685,
|
| 909 |
+
"eval_steps_per_second": 21.204,
|
| 910 |
+
"step": 3250
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 8.89,
|
| 914 |
+
"eval_accuracy": 0.8866396761133604,
|
| 915 |
+
"eval_binary": 0.9203778677462877,
|
| 916 |
+
"eval_f1": 0.8852200459634372,
|
| 917 |
+
"eval_loss": 0.5305805206298828,
|
| 918 |
+
"eval_precision": 0.8933747127066964,
|
| 919 |
+
"eval_recall": 0.8866396761133604,
|
| 920 |
+
"eval_runtime": 1.1366,
|
| 921 |
+
"eval_samples_per_second": 651.943,
|
| 922 |
+
"eval_steps_per_second": 21.116,
|
| 923 |
+
"step": 3300
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 9.0,
|
| 927 |
+
"grad_norm": 2.9827425479888916,
|
| 928 |
+
"learning_rate": 7.000898472596586e-05,
|
| 929 |
+
"loss": 0.3284,
|
| 930 |
+
"step": 3339
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 9.03,
|
| 934 |
+
"eval_accuracy": 0.8866396761133604,
|
| 935 |
+
"eval_binary": 0.9203778677462882,
|
| 936 |
+
"eval_f1": 0.8851786363349267,
|
| 937 |
+
"eval_loss": 0.5871447920799255,
|
| 938 |
+
"eval_precision": 0.8929871747685514,
|
| 939 |
+
"eval_recall": 0.8866396761133604,
|
| 940 |
+
"eval_runtime": 1.1268,
|
| 941 |
+
"eval_samples_per_second": 657.63,
|
| 942 |
+
"eval_steps_per_second": 21.3,
|
| 943 |
+
"step": 3350
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 9.16,
|
| 947 |
+
"eval_accuracy": 0.8717948717948718,
|
| 948 |
+
"eval_binary": 0.9114709851551945,
|
| 949 |
+
"eval_f1": 0.8692786610310018,
|
| 950 |
+
"eval_loss": 0.6212562918663025,
|
| 951 |
+
"eval_precision": 0.8831649569503822,
|
| 952 |
+
"eval_recall": 0.8717948717948718,
|
| 953 |
+
"eval_runtime": 1.125,
|
| 954 |
+
"eval_samples_per_second": 658.664,
|
| 955 |
+
"eval_steps_per_second": 21.333,
|
| 956 |
+
"step": 3400
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 9.3,
|
| 960 |
+
"eval_accuracy": 0.8893387314439946,
|
| 961 |
+
"eval_binary": 0.921862348178137,
|
| 962 |
+
"eval_f1": 0.8877552195071403,
|
| 963 |
+
"eval_loss": 0.5114206671714783,
|
| 964 |
+
"eval_precision": 0.8954543919118818,
|
| 965 |
+
"eval_recall": 0.8893387314439946,
|
| 966 |
+
"eval_runtime": 1.1299,
|
| 967 |
+
"eval_samples_per_second": 655.79,
|
| 968 |
+
"eval_steps_per_second": 21.24,
|
| 969 |
+
"step": 3450
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 9.43,
|
| 973 |
+
"eval_accuracy": 0.8758434547908233,
|
| 974 |
+
"eval_binary": 0.9137651821862343,
|
| 975 |
+
"eval_f1": 0.8732295909377273,
|
| 976 |
+
"eval_loss": 0.5960059762001038,
|
| 977 |
+
"eval_precision": 0.881588392114708,
|
| 978 |
+
"eval_recall": 0.8758434547908233,
|
| 979 |
+
"eval_runtime": 1.1351,
|
| 980 |
+
"eval_samples_per_second": 652.779,
|
| 981 |
+
"eval_steps_per_second": 21.143,
|
| 982 |
+
"step": 3500
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 9.57,
|
| 986 |
+
"eval_accuracy": 0.8825910931174089,
|
| 987 |
+
"eval_binary": 0.9190283400809708,
|
| 988 |
+
"eval_f1": 0.8811241165340947,
|
| 989 |
+
"eval_loss": 0.6340000629425049,
|
| 990 |
+
"eval_precision": 0.8921271074105084,
|
| 991 |
+
"eval_recall": 0.8825910931174089,
|
| 992 |
+
"eval_runtime": 1.1227,
|
| 993 |
+
"eval_samples_per_second": 660.008,
|
| 994 |
+
"eval_steps_per_second": 21.377,
|
| 995 |
+
"step": 3550
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 9.7,
|
| 999 |
+
"eval_accuracy": 0.8758434547908233,
|
| 1000 |
+
"eval_binary": 0.9126855600539802,
|
| 1001 |
+
"eval_f1": 0.8735923320711677,
|
| 1002 |
+
"eval_loss": 0.606035590171814,
|
| 1003 |
+
"eval_precision": 0.8846766855876167,
|
| 1004 |
+
"eval_recall": 0.8758434547908233,
|
| 1005 |
+
"eval_runtime": 1.1444,
|
| 1006 |
+
"eval_samples_per_second": 647.476,
|
| 1007 |
+
"eval_steps_per_second": 20.971,
|
| 1008 |
+
"step": 3600
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 9.84,
|
| 1012 |
+
"eval_accuracy": 0.8704453441295547,
|
| 1013 |
+
"eval_binary": 0.9099865047233459,
|
| 1014 |
+
"eval_f1": 0.8673800362820931,
|
| 1015 |
+
"eval_loss": 0.6465194821357727,
|
| 1016 |
+
"eval_precision": 0.8800944669365723,
|
| 1017 |
+
"eval_recall": 0.8704453441295547,
|
| 1018 |
+
"eval_runtime": 1.1227,
|
| 1019 |
+
"eval_samples_per_second": 659.991,
|
| 1020 |
+
"eval_steps_per_second": 21.376,
|
| 1021 |
+
"step": 3650
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 9.97,
|
| 1025 |
+
"eval_accuracy": 0.8785425101214575,
|
| 1026 |
+
"eval_binary": 0.9148448043184876,
|
| 1027 |
+
"eval_f1": 0.8767805082692359,
|
| 1028 |
+
"eval_loss": 0.6755142211914062,
|
| 1029 |
+
"eval_precision": 0.8911575117040704,
|
| 1030 |
+
"eval_recall": 0.8785425101214575,
|
| 1031 |
+
"eval_runtime": 1.1379,
|
| 1032 |
+
"eval_samples_per_second": 651.188,
|
| 1033 |
+
"eval_steps_per_second": 21.091,
|
| 1034 |
+
"step": 3700
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 10.0,
|
| 1038 |
+
"grad_norm": 4.129298686981201,
|
| 1039 |
+
"learning_rate": 6.667565139263254e-05,
|
| 1040 |
+
"loss": 0.3062,
|
| 1041 |
+
"step": 3710
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 10.11,
|
| 1045 |
+
"eval_accuracy": 0.8785425101214575,
|
| 1046 |
+
"eval_binary": 0.9161943319838048,
|
| 1047 |
+
"eval_f1": 0.877249561583927,
|
| 1048 |
+
"eval_loss": 0.6168139576911926,
|
| 1049 |
+
"eval_precision": 0.8873360033279062,
|
| 1050 |
+
"eval_recall": 0.8785425101214575,
|
| 1051 |
+
"eval_runtime": 1.1239,
|
| 1052 |
+
"eval_samples_per_second": 659.319,
|
| 1053 |
+
"eval_steps_per_second": 21.354,
|
| 1054 |
+
"step": 3750
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 10.24,
|
| 1058 |
+
"eval_accuracy": 0.8812415654520918,
|
| 1059 |
+
"eval_binary": 0.917543859649122,
|
| 1060 |
+
"eval_f1": 0.8794928490642383,
|
| 1061 |
+
"eval_loss": 0.6084047555923462,
|
| 1062 |
+
"eval_precision": 0.8893881946513525,
|
| 1063 |
+
"eval_recall": 0.8812415654520918,
|
| 1064 |
+
"eval_runtime": 1.1288,
|
| 1065 |
+
"eval_samples_per_second": 656.471,
|
| 1066 |
+
"eval_steps_per_second": 21.262,
|
| 1067 |
+
"step": 3800
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 10.38,
|
| 1071 |
+
"eval_accuracy": 0.8704453441295547,
|
| 1072 |
+
"eval_binary": 0.9106612685560047,
|
| 1073 |
+
"eval_f1": 0.8674074511392077,
|
| 1074 |
+
"eval_loss": 0.733109712600708,
|
| 1075 |
+
"eval_precision": 0.8810352752660445,
|
| 1076 |
+
"eval_recall": 0.8704453441295547,
|
| 1077 |
+
"eval_runtime": 1.1179,
|
| 1078 |
+
"eval_samples_per_second": 662.846,
|
| 1079 |
+
"eval_steps_per_second": 21.469,
|
| 1080 |
+
"step": 3850
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 10.51,
|
| 1084 |
+
"eval_accuracy": 0.8704453441295547,
|
| 1085 |
+
"eval_binary": 0.9085020242914971,
|
| 1086 |
+
"eval_f1": 0.8681938294778445,
|
| 1087 |
+
"eval_loss": 0.6585559248924255,
|
| 1088 |
+
"eval_precision": 0.8790380185117028,
|
| 1089 |
+
"eval_recall": 0.8704453441295547,
|
| 1090 |
+
"eval_runtime": 1.1015,
|
| 1091 |
+
"eval_samples_per_second": 672.696,
|
| 1092 |
+
"eval_steps_per_second": 21.788,
|
| 1093 |
+
"step": 3900
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 10.65,
|
| 1097 |
+
"eval_accuracy": 0.865047233468286,
|
| 1098 |
+
"eval_binary": 0.9035087719298243,
|
| 1099 |
+
"eval_f1": 0.8619257548463832,
|
| 1100 |
+
"eval_loss": 0.7438000440597534,
|
| 1101 |
+
"eval_precision": 0.8722900768447327,
|
| 1102 |
+
"eval_recall": 0.865047233468286,
|
| 1103 |
+
"eval_runtime": 1.1141,
|
| 1104 |
+
"eval_samples_per_second": 665.131,
|
| 1105 |
+
"eval_steps_per_second": 21.543,
|
| 1106 |
+
"step": 3950
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 10.78,
|
| 1110 |
+
"eval_accuracy": 0.8798920377867746,
|
| 1111 |
+
"eval_binary": 0.9157894736842094,
|
| 1112 |
+
"eval_f1": 0.8784962349040888,
|
| 1113 |
+
"eval_loss": 0.5920050740242004,
|
| 1114 |
+
"eval_precision": 0.8895519684993368,
|
| 1115 |
+
"eval_recall": 0.8798920377867746,
|
| 1116 |
+
"eval_runtime": 1.1303,
|
| 1117 |
+
"eval_samples_per_second": 655.59,
|
| 1118 |
+
"eval_steps_per_second": 21.234,
|
| 1119 |
+
"step": 4000
|
| 1120 |
+
}
|
| 1121 |
+
],
|
| 1122 |
+
"logging_steps": 500,
|
| 1123 |
+
"max_steps": 11130,
|
| 1124 |
+
"num_input_tokens_seen": 0,
|
| 1125 |
+
"num_train_epochs": 30,
|
| 1126 |
+
"save_steps": 500,
|
| 1127 |
+
"total_flos": 5.3953021213009e+18,
|
| 1128 |
+
"train_batch_size": 32,
|
| 1129 |
+
"trial_name": null,
|
| 1130 |
+
"trial_params": null
|
| 1131 |
+
}
|
tmp-checkpoint-4000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ad911284276196968ad8f0dfc5b63b04d6d3ebd414b55f47dc985e6b8a9095a5
|
| 3 |
+
size 4984
|