Upload folder using huggingface_hub
Browse files- Internal-Run-EN/SpanBased-DeBERTa/checkpoint-5788/config.json +70 -86
- Internal-Run-EN/SpanBased-DeBERTa/checkpoint-5788/model.safetensors +2 -2
- Internal-Run-EN/SpanBased-DeBERTa/checkpoint-5788/optimizer.pt +2 -2
- Internal-Run-EN/SpanBased-DeBERTa/checkpoint-5788/trainer_state.json +231 -231
- Internal-Run-EN/SpanBased-DeBERTa/config.json +70 -86
- Internal-Run-EN/SpanBased-DeBERTa/pytorch_model.bin +2 -2
Internal-Run-EN/SpanBased-DeBERTa/checkpoint-5788/config.json
CHANGED
|
@@ -24,49 +24,41 @@
|
|
| 24 |
"11": "CREDITCARDCVV",
|
| 25 |
"12": "CREDITCARDISSUER",
|
| 26 |
"13": "CREDITCARDNUMBER",
|
| 27 |
-
"14": "
|
| 28 |
-
"15": "
|
| 29 |
-
"16": "
|
| 30 |
-
"17": "
|
| 31 |
-
"18": "
|
| 32 |
-
"19": "
|
| 33 |
-
"20": "
|
| 34 |
-
"21": "
|
| 35 |
-
"22": "
|
| 36 |
-
"23": "
|
| 37 |
-
"24": "
|
| 38 |
-
"25": "
|
| 39 |
-
"26": "
|
| 40 |
-
"27": "
|
| 41 |
-
"28": "
|
| 42 |
-
"29": "
|
| 43 |
-
"30": "
|
| 44 |
-
"31": "
|
| 45 |
-
"32": "
|
| 46 |
-
"33": "
|
| 47 |
-
"34": "
|
| 48 |
-
"35": "
|
| 49 |
-
"36": "
|
| 50 |
-
"37": "
|
| 51 |
-
"38": "
|
| 52 |
-
"39": "
|
| 53 |
-
"40": "
|
| 54 |
-
"41": "
|
| 55 |
-
"42": "
|
| 56 |
-
"43": "
|
| 57 |
-
"44": "
|
| 58 |
-
"45": "
|
| 59 |
-
"46": "
|
| 60 |
-
"47": "
|
| 61 |
-
"48": "
|
| 62 |
-
"49": "STREET",
|
| 63 |
-
"50": "TIME",
|
| 64 |
-
"51": "URL",
|
| 65 |
-
"52": "USERAGENT",
|
| 66 |
-
"53": "USERNAME",
|
| 67 |
-
"54": "VEHICLEVIN",
|
| 68 |
-
"55": "VEHICLEVRM",
|
| 69 |
-
"56": "ZIPCODE"
|
| 70 |
},
|
| 71 |
"initializer_range": 0.02,
|
| 72 |
"intermediate_size": 3072,
|
|
@@ -84,50 +76,42 @@
|
|
| 84 |
"CREDITCARDCVV": 11,
|
| 85 |
"CREDITCARDISSUER": 12,
|
| 86 |
"CREDITCARDNUMBER": 13,
|
| 87 |
-
"
|
| 88 |
-
"
|
| 89 |
-
"
|
| 90 |
-
"
|
| 91 |
-
"
|
| 92 |
-
"
|
| 93 |
-
"
|
| 94 |
-
"
|
| 95 |
-
"
|
| 96 |
-
"
|
| 97 |
-
"
|
| 98 |
-
"
|
| 99 |
-
"
|
| 100 |
-
"
|
| 101 |
-
"
|
| 102 |
-
"IPV6": 29,
|
| 103 |
-
"JOBAREA": 30,
|
| 104 |
-
"JOBTITLE": 31,
|
| 105 |
-
"JOBTYPE": 32,
|
| 106 |
-
"LASTNAME": 33,
|
| 107 |
-
"LITECOINADDRESS": 34,
|
| 108 |
-
"MAC": 35,
|
| 109 |
-
"MASKEDNUMBER": 36,
|
| 110 |
-
"MIDDLENAME": 37,
|
| 111 |
-
"NEARBYGPSCOORDINATE": 38,
|
| 112 |
"O": 0,
|
| 113 |
-
"ORDINALDIRECTION":
|
| 114 |
-
"
|
| 115 |
-
"
|
| 116 |
-
"
|
| 117 |
-
"
|
| 118 |
-
"
|
| 119 |
-
"
|
| 120 |
-
"
|
| 121 |
-
"
|
| 122 |
-
"
|
| 123 |
-
"
|
| 124 |
-
"
|
| 125 |
-
"
|
| 126 |
-
"
|
| 127 |
-
"
|
| 128 |
-
"
|
| 129 |
-
"
|
| 130 |
-
"
|
|
|
|
|
|
|
| 131 |
},
|
| 132 |
"layer_norm_eps": 1e-07,
|
| 133 |
"legacy": true,
|
|
|
|
| 24 |
"11": "CREDITCARDCVV",
|
| 25 |
"12": "CREDITCARDISSUER",
|
| 26 |
"13": "CREDITCARDNUMBER",
|
| 27 |
+
"14": "DATE",
|
| 28 |
+
"15": "DOB",
|
| 29 |
+
"16": "DRIVERLICENSENUM",
|
| 30 |
+
"17": "EMAIL",
|
| 31 |
+
"18": "ETHEREUMADDRESS",
|
| 32 |
+
"19": "EYECOLOR",
|
| 33 |
+
"20": "GENDER",
|
| 34 |
+
"21": "HEIGHT",
|
| 35 |
+
"22": "IBAN",
|
| 36 |
+
"23": "IPADDRESS",
|
| 37 |
+
"24": "JOBAREA",
|
| 38 |
+
"25": "JOBTYPE",
|
| 39 |
+
"26": "LITECOINADDRESS",
|
| 40 |
+
"27": "MASKEDNUMBER",
|
| 41 |
+
"28": "NEARBYGPSCOORDINATE",
|
| 42 |
+
"29": "ORDINALDIRECTION",
|
| 43 |
+
"30": "PASSPORTNUM",
|
| 44 |
+
"31": "PASSWORD",
|
| 45 |
+
"32": "PERSONNAME",
|
| 46 |
+
"33": "PHONEIMEI",
|
| 47 |
+
"34": "PHONENUMBER",
|
| 48 |
+
"35": "PIN",
|
| 49 |
+
"36": "PREFIX",
|
| 50 |
+
"37": "SECONDARYADDRESS",
|
| 51 |
+
"38": "SSN/CCCD",
|
| 52 |
+
"39": "STATE",
|
| 53 |
+
"40": "STREET",
|
| 54 |
+
"41": "TAXNUM",
|
| 55 |
+
"42": "TIME",
|
| 56 |
+
"43": "URL",
|
| 57 |
+
"44": "USERAGENT",
|
| 58 |
+
"45": "USERNAME",
|
| 59 |
+
"46": "VEHICLEVIN",
|
| 60 |
+
"47": "VEHICLEVRM",
|
| 61 |
+
"48": "ZIPCODE"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 62 |
},
|
| 63 |
"initializer_range": 0.02,
|
| 64 |
"intermediate_size": 3072,
|
|
|
|
| 76 |
"CREDITCARDCVV": 11,
|
| 77 |
"CREDITCARDISSUER": 12,
|
| 78 |
"CREDITCARDNUMBER": 13,
|
| 79 |
+
"DATE": 14,
|
| 80 |
+
"DOB": 15,
|
| 81 |
+
"DRIVERLICENSENUM": 16,
|
| 82 |
+
"EMAIL": 17,
|
| 83 |
+
"ETHEREUMADDRESS": 18,
|
| 84 |
+
"EYECOLOR": 19,
|
| 85 |
+
"GENDER": 20,
|
| 86 |
+
"HEIGHT": 21,
|
| 87 |
+
"IBAN": 22,
|
| 88 |
+
"IPADDRESS": 23,
|
| 89 |
+
"JOBAREA": 24,
|
| 90 |
+
"JOBTYPE": 25,
|
| 91 |
+
"LITECOINADDRESS": 26,
|
| 92 |
+
"MASKEDNUMBER": 27,
|
| 93 |
+
"NEARBYGPSCOORDINATE": 28,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 94 |
"O": 0,
|
| 95 |
+
"ORDINALDIRECTION": 29,
|
| 96 |
+
"PASSPORTNUM": 30,
|
| 97 |
+
"PASSWORD": 31,
|
| 98 |
+
"PERSONNAME": 32,
|
| 99 |
+
"PHONEIMEI": 33,
|
| 100 |
+
"PHONENUMBER": 34,
|
| 101 |
+
"PIN": 35,
|
| 102 |
+
"PREFIX": 36,
|
| 103 |
+
"SECONDARYADDRESS": 37,
|
| 104 |
+
"SSN/CCCD": 38,
|
| 105 |
+
"STATE": 39,
|
| 106 |
+
"STREET": 40,
|
| 107 |
+
"TAXNUM": 41,
|
| 108 |
+
"TIME": 42,
|
| 109 |
+
"URL": 43,
|
| 110 |
+
"USERAGENT": 44,
|
| 111 |
+
"USERNAME": 45,
|
| 112 |
+
"VEHICLEVIN": 46,
|
| 113 |
+
"VEHICLEVRM": 47,
|
| 114 |
+
"ZIPCODE": 48
|
| 115 |
},
|
| 116 |
"layer_norm_eps": 1e-07,
|
| 117 |
"legacy": true,
|
Internal-Run-EN/SpanBased-DeBERTa/checkpoint-5788/model.safetensors
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:e5842025b0d391577502c80552fa064362704776cb0961f384ac659661a0e99e
|
| 3 |
+
size 741479708
|
Internal-Run-EN/SpanBased-DeBERTa/checkpoint-5788/optimizer.pt
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:d8c5fa4a44f96749e3be01156c46ade10f70c69422a2725a268962295b802c47
|
| 3 |
+
size 1483083147
|
Internal-Run-EN/SpanBased-DeBERTa/checkpoint-5788/trainer_state.json
CHANGED
|
@@ -11,807 +11,807 @@
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.017277125086385625,
|
| 14 |
-
"grad_norm": 0.
|
| 15 |
"learning_rate": 1.9830684174153422e-05,
|
| 16 |
-
"loss": 0.
|
| 17 |
"step": 50
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.03455425017277125,
|
| 21 |
-
"grad_norm": 0.
|
| 22 |
"learning_rate": 1.9657912923289565e-05,
|
| 23 |
-
"loss": 0.
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.051831375259156875,
|
| 28 |
-
"grad_norm": 0.
|
| 29 |
"learning_rate": 1.948514167242571e-05,
|
| 30 |
-
"loss": 0.
|
| 31 |
"step": 150
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.0691085003455425,
|
| 35 |
-
"grad_norm": 0.
|
| 36 |
"learning_rate": 1.9312370421561853e-05,
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 200
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.08638562543192813,
|
| 42 |
-
"grad_norm": 0.
|
| 43 |
"learning_rate": 1.9139599170698e-05,
|
| 44 |
-
"loss": 0.
|
| 45 |
"step": 250
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.10366275051831375,
|
| 49 |
-
"grad_norm": 0.
|
| 50 |
"learning_rate": 1.896682791983414e-05,
|
| 51 |
-
"loss": 0.
|
| 52 |
"step": 300
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 0.12093987560469938,
|
| 56 |
-
"grad_norm": 0.
|
| 57 |
"learning_rate": 1.8794056668970284e-05,
|
| 58 |
-
"loss": 0.
|
| 59 |
"step": 350
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 0.138217000691085,
|
| 63 |
-
"grad_norm": 0.
|
| 64 |
"learning_rate": 1.862128541810643e-05,
|
| 65 |
-
"loss": 0.
|
| 66 |
"step": 400
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 0.15549412577747063,
|
| 70 |
-
"grad_norm": 0.
|
| 71 |
"learning_rate": 1.8448514167242572e-05,
|
| 72 |
-
"loss": 0.
|
| 73 |
"step": 450
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 0.17277125086385625,
|
| 77 |
-
"grad_norm": 0.
|
| 78 |
"learning_rate": 1.8275742916378715e-05,
|
| 79 |
-
"loss": 0.
|
| 80 |
"step": 500
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.19004837595024188,
|
| 84 |
-
"grad_norm": 0.
|
| 85 |
"learning_rate": 1.810297166551486e-05,
|
| 86 |
-
"loss": 0.
|
| 87 |
"step": 550
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.2073255010366275,
|
| 91 |
-
"grad_norm": 0.
|
| 92 |
"learning_rate": 1.7930200414651003e-05,
|
| 93 |
-
"loss": 0.
|
| 94 |
"step": 600
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 0.22460262612301313,
|
| 98 |
-
"grad_norm": 0.
|
| 99 |
"learning_rate": 1.775742916378715e-05,
|
| 100 |
-
"loss": 0.
|
| 101 |
"step": 650
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 0.24187975120939875,
|
| 105 |
-
"grad_norm": 0.
|
| 106 |
"learning_rate": 1.758465791292329e-05,
|
| 107 |
-
"loss": 0.
|
| 108 |
"step": 700
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 0.2591568762957844,
|
| 112 |
-
"grad_norm": 0.
|
| 113 |
"learning_rate": 1.7411886662059434e-05,
|
| 114 |
-
"loss": 0.
|
| 115 |
"step": 750
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 0.27643400138217,
|
| 119 |
-
"grad_norm": 0.
|
| 120 |
"learning_rate": 1.723911541119558e-05,
|
| 121 |
-
"loss": 0.
|
| 122 |
"step": 800
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.29371112646855563,
|
| 126 |
-
"grad_norm": 0.
|
| 127 |
"learning_rate": 1.7066344160331722e-05,
|
| 128 |
-
"loss": 0.
|
| 129 |
"step": 850
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.31098825155494125,
|
| 133 |
-
"grad_norm": 0.
|
| 134 |
"learning_rate": 1.6893572909467865e-05,
|
| 135 |
-
"loss": 0.
|
| 136 |
"step": 900
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.3282653766413269,
|
| 140 |
-
"grad_norm": 0.
|
| 141 |
"learning_rate": 1.6720801658604007e-05,
|
| 142 |
-
"loss": 0.
|
| 143 |
"step": 950
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 0.3455425017277125,
|
| 147 |
-
"grad_norm": 0.
|
| 148 |
"learning_rate": 1.6548030407740153e-05,
|
| 149 |
-
"loss": 0.
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 0.36281962681409813,
|
| 154 |
-
"grad_norm": 0.
|
| 155 |
"learning_rate": 1.63752591568763e-05,
|
| 156 |
-
"loss": 0.
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 0.38009675190048375,
|
| 161 |
-
"grad_norm": 0.
|
| 162 |
"learning_rate": 1.620248790601244e-05,
|
| 163 |
-
"loss": 0.
|
| 164 |
"step": 1100
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.3973738769868694,
|
| 168 |
-
"grad_norm": 0.
|
| 169 |
"learning_rate": 1.6029716655148584e-05,
|
| 170 |
-
"loss": 0.
|
| 171 |
"step": 1150
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.414651002073255,
|
| 175 |
-
"grad_norm": 0.
|
| 176 |
"learning_rate": 1.585694540428473e-05,
|
| 177 |
-
"loss": 0.
|
| 178 |
"step": 1200
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 0.43192812715964063,
|
| 182 |
-
"grad_norm": 0.
|
| 183 |
"learning_rate": 1.5684174153420872e-05,
|
| 184 |
-
"loss": 0.
|
| 185 |
"step": 1250
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 0.44920525224602625,
|
| 189 |
-
"grad_norm": 0.
|
| 190 |
"learning_rate": 1.5511402902557015e-05,
|
| 191 |
-
"loss": 0.
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 0.4664823773324119,
|
| 196 |
-
"grad_norm": 0.
|
| 197 |
"learning_rate": 1.5338631651693157e-05,
|
| 198 |
-
"loss": 0.
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.4837595024187975,
|
| 203 |
-
"grad_norm": 0.
|
| 204 |
"learning_rate": 1.5165860400829305e-05,
|
| 205 |
-
"loss": 0.
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.5010366275051832,
|
| 210 |
-
"grad_norm": 0.
|
| 211 |
"learning_rate": 1.4993089149965447e-05,
|
| 212 |
-
"loss": 0.
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.5183137525915688,
|
| 217 |
-
"grad_norm": 0.
|
| 218 |
"learning_rate": 1.482031789910159e-05,
|
| 219 |
-
"loss": 0.
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 0.5355908776779544,
|
| 224 |
-
"grad_norm": 0.
|
| 225 |
"learning_rate": 1.4647546648237734e-05,
|
| 226 |
-
"loss": 0.
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 0.55286800276434,
|
| 231 |
-
"grad_norm": 0.
|
| 232 |
"learning_rate": 1.447477539737388e-05,
|
| 233 |
-
"loss": 0.
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.5701451278507257,
|
| 238 |
-
"grad_norm": 0.
|
| 239 |
"learning_rate": 1.4302004146510022e-05,
|
| 240 |
-
"loss": 0.
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 0.5874222529371113,
|
| 245 |
-
"grad_norm": 0.
|
| 246 |
"learning_rate": 1.4129232895646165e-05,
|
| 247 |
-
"loss": 0.
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 0.6046993780234969,
|
| 252 |
-
"grad_norm": 0.
|
| 253 |
"learning_rate": 1.3956461644782309e-05,
|
| 254 |
-
"loss": 0.
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.6219765031098825,
|
| 259 |
-
"grad_norm": 0.
|
| 260 |
"learning_rate": 1.3783690393918455e-05,
|
| 261 |
-
"loss": 0.
|
| 262 |
"step": 1800
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 0.6392536281962682,
|
| 266 |
-
"grad_norm": 0.
|
| 267 |
"learning_rate": 1.3610919143054597e-05,
|
| 268 |
-
"loss": 0.
|
| 269 |
"step": 1850
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 0.6565307532826538,
|
| 273 |
-
"grad_norm": 0.
|
| 274 |
"learning_rate": 1.343814789219074e-05,
|
| 275 |
-
"loss": 0.
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 0.6738078783690394,
|
| 280 |
-
"grad_norm": 0.
|
| 281 |
"learning_rate": 1.3265376641326884e-05,
|
| 282 |
-
"loss": 0.
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 0.691085003455425,
|
| 287 |
-
"grad_norm": 0.
|
| 288 |
"learning_rate": 1.309260539046303e-05,
|
| 289 |
-
"loss": 0.
|
| 290 |
"step": 2000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 0.7083621285418107,
|
| 294 |
-
"grad_norm": 0.
|
| 295 |
"learning_rate": 1.2919834139599172e-05,
|
| 296 |
-
"loss": 0.
|
| 297 |
"step": 2050
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.7256392536281963,
|
| 301 |
-
"grad_norm": 0.
|
| 302 |
"learning_rate": 1.2747062888735315e-05,
|
| 303 |
-
"loss": 0.
|
| 304 |
"step": 2100
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 0.7429163787145819,
|
| 308 |
-
"grad_norm": 0.
|
| 309 |
"learning_rate": 1.2574291637871459e-05,
|
| 310 |
-
"loss": 0.
|
| 311 |
"step": 2150
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 0.7601935038009675,
|
| 315 |
-
"grad_norm": 0.
|
| 316 |
"learning_rate": 1.2401520387007605e-05,
|
| 317 |
-
"loss": 0.
|
| 318 |
"step": 2200
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 0.7774706288873532,
|
| 322 |
-
"grad_norm": 0.
|
| 323 |
"learning_rate": 1.2228749136143747e-05,
|
| 324 |
-
"loss": 0.
|
| 325 |
"step": 2250
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.7947477539737388,
|
| 329 |
-
"grad_norm": 0.
|
| 330 |
"learning_rate": 1.205597788527989e-05,
|
| 331 |
-
"loss": 0.
|
| 332 |
"step": 2300
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 0.8120248790601244,
|
| 336 |
-
"grad_norm": 0.
|
| 337 |
"learning_rate": 1.1883206634416034e-05,
|
| 338 |
-
"loss": 0.
|
| 339 |
"step": 2350
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.82930200414651,
|
| 343 |
-
"grad_norm": 0.
|
| 344 |
"learning_rate": 1.1710435383552178e-05,
|
| 345 |
-
"loss": 0.
|
| 346 |
"step": 2400
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 0.8465791292328957,
|
| 350 |
-
"grad_norm": 0.
|
| 351 |
"learning_rate": 1.1537664132688322e-05,
|
| 352 |
-
"loss": 0.
|
| 353 |
"step": 2450
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 0.8638562543192813,
|
| 357 |
-
"grad_norm": 0.
|
| 358 |
"learning_rate": 1.1364892881824465e-05,
|
| 359 |
-
"loss": 0.
|
| 360 |
"step": 2500
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 0.8811333794056669,
|
| 364 |
-
"grad_norm": 0.
|
| 365 |
"learning_rate": 1.1192121630960609e-05,
|
| 366 |
-
"loss": 0.
|
| 367 |
"step": 2550
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 0.8984105044920525,
|
| 371 |
-
"grad_norm": 0.
|
| 372 |
"learning_rate": 1.1019350380096753e-05,
|
| 373 |
-
"loss": 0.
|
| 374 |
"step": 2600
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 0.9156876295784382,
|
| 378 |
-
"grad_norm": 0.
|
| 379 |
"learning_rate": 1.0846579129232897e-05,
|
| 380 |
-
"loss": 0.
|
| 381 |
"step": 2650
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 0.9329647546648238,
|
| 385 |
-
"grad_norm": 0.
|
| 386 |
"learning_rate": 1.067380787836904e-05,
|
| 387 |
-
"loss": 0.
|
| 388 |
"step": 2700
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 0.9502418797512094,
|
| 392 |
-
"grad_norm": 0.
|
| 393 |
"learning_rate": 1.0501036627505184e-05,
|
| 394 |
-
"loss": 0.
|
| 395 |
"step": 2750
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 0.967519004837595,
|
| 399 |
-
"grad_norm": 0.
|
| 400 |
"learning_rate": 1.0328265376641328e-05,
|
| 401 |
-
"loss": 0.
|
| 402 |
"step": 2800
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 0.9847961299239807,
|
| 406 |
-
"grad_norm": 0.
|
| 407 |
"learning_rate": 1.0155494125777472e-05,
|
| 408 |
-
"loss": 0.
|
| 409 |
"step": 2850
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 1.0020732550103664,
|
| 413 |
-
"grad_norm": 0.
|
| 414 |
"learning_rate": 9.982722874913615e-06,
|
| 415 |
-
"loss": 0.
|
| 416 |
"step": 2900
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 1.0193503800967518,
|
| 420 |
-
"grad_norm": 0.
|
| 421 |
"learning_rate": 9.809951624049759e-06,
|
| 422 |
-
"loss": 0.
|
| 423 |
"step": 2950
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 1.0366275051831375,
|
| 427 |
-
"grad_norm": 0.
|
| 428 |
"learning_rate": 9.637180373185903e-06,
|
| 429 |
-
"loss": 0.
|
| 430 |
"step": 3000
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 1.0539046302695232,
|
| 434 |
-
"grad_norm": 0.
|
| 435 |
"learning_rate": 9.464409122322047e-06,
|
| 436 |
-
"loss": 0.
|
| 437 |
"step": 3050
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 1.0711817553559089,
|
| 441 |
-
"grad_norm": 0.
|
| 442 |
"learning_rate": 9.29163787145819e-06,
|
| 443 |
-
"loss": 0.
|
| 444 |
"step": 3100
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 1.0884588804422943,
|
| 448 |
-
"grad_norm": 0.
|
| 449 |
"learning_rate": 9.118866620594334e-06,
|
| 450 |
-
"loss": 0.
|
| 451 |
"step": 3150
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 1.10573600552868,
|
| 455 |
-
"grad_norm": 0.
|
| 456 |
"learning_rate": 8.946095369730478e-06,
|
| 457 |
-
"loss": 0.
|
| 458 |
"step": 3200
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 1.1230131306150657,
|
| 462 |
-
"grad_norm": 0.
|
| 463 |
"learning_rate": 8.773324118866622e-06,
|
| 464 |
-
"loss": 0.
|
| 465 |
"step": 3250
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 1.1402902557014514,
|
| 469 |
-
"grad_norm": 0.
|
| 470 |
"learning_rate": 8.600552868002765e-06,
|
| 471 |
-
"loss": 0.
|
| 472 |
"step": 3300
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 1.1575673807878368,
|
| 476 |
-
"grad_norm": 0.
|
| 477 |
"learning_rate": 8.427781617138909e-06,
|
| 478 |
-
"loss": 0.
|
| 479 |
"step": 3350
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 1.1748445058742225,
|
| 483 |
-
"grad_norm": 0.
|
| 484 |
"learning_rate": 8.255010366275053e-06,
|
| 485 |
-
"loss": 0.
|
| 486 |
"step": 3400
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 1.1921216309606082,
|
| 490 |
-
"grad_norm": 0.
|
| 491 |
"learning_rate": 8.082239115411197e-06,
|
| 492 |
-
"loss": 0.
|
| 493 |
"step": 3450
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 1.2093987560469939,
|
| 497 |
-
"grad_norm": 0.
|
| 498 |
"learning_rate": 7.90946786454734e-06,
|
| 499 |
-
"loss": 0.
|
| 500 |
"step": 3500
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 1.2266758811333793,
|
| 504 |
-
"grad_norm": 0.
|
| 505 |
"learning_rate": 7.736696613683484e-06,
|
| 506 |
-
"loss": 0.
|
| 507 |
"step": 3550
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 1.243953006219765,
|
| 511 |
-
"grad_norm": 0.
|
| 512 |
"learning_rate": 7.563925362819627e-06,
|
| 513 |
-
"loss": 0.
|
| 514 |
"step": 3600
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 1.2612301313061507,
|
| 518 |
-
"grad_norm": 0.
|
| 519 |
"learning_rate": 7.391154111955771e-06,
|
| 520 |
-
"loss": 0.
|
| 521 |
"step": 3650
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 1.2785072563925364,
|
| 525 |
-
"grad_norm": 0.
|
| 526 |
"learning_rate": 7.218382861091915e-06,
|
| 527 |
-
"loss": 0.
|
| 528 |
"step": 3700
|
| 529 |
},
|
| 530 |
{
|
| 531 |
"epoch": 1.2957843814789218,
|
| 532 |
-
"grad_norm": 0.
|
| 533 |
"learning_rate": 7.045611610228059e-06,
|
| 534 |
-
"loss": 0.
|
| 535 |
"step": 3750
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"epoch": 1.3130615065653075,
|
| 539 |
-
"grad_norm": 0.
|
| 540 |
"learning_rate": 6.872840359364202e-06,
|
| 541 |
-
"loss": 0.
|
| 542 |
"step": 3800
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 1.3303386316516932,
|
| 546 |
-
"grad_norm": 0.
|
| 547 |
"learning_rate": 6.700069108500346e-06,
|
| 548 |
-
"loss": 0.
|
| 549 |
"step": 3850
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 1.3476157567380787,
|
| 553 |
-
"grad_norm": 0.
|
| 554 |
"learning_rate": 6.52729785763649e-06,
|
| 555 |
-
"loss": 0.
|
| 556 |
"step": 3900
|
| 557 |
},
|
| 558 |
{
|
| 559 |
"epoch": 1.3648928818244643,
|
| 560 |
-
"grad_norm": 0.
|
| 561 |
"learning_rate": 6.354526606772633e-06,
|
| 562 |
-
"loss": 0.
|
| 563 |
"step": 3950
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 1.38217000691085,
|
| 567 |
-
"grad_norm": 0.
|
| 568 |
"learning_rate": 6.181755355908777e-06,
|
| 569 |
-
"loss": 0.
|
| 570 |
"step": 4000
|
| 571 |
},
|
| 572 |
{
|
| 573 |
"epoch": 1.3994471319972357,
|
| 574 |
-
"grad_norm": 0.
|
| 575 |
"learning_rate": 6.0089841050449205e-06,
|
| 576 |
-
"loss": 0.
|
| 577 |
"step": 4050
|
| 578 |
},
|
| 579 |
{
|
| 580 |
"epoch": 1.4167242570836214,
|
| 581 |
-
"grad_norm": 0.
|
| 582 |
"learning_rate": 5.836212854181065e-06,
|
| 583 |
-
"loss": 0.
|
| 584 |
"step": 4100
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 1.4340013821700068,
|
| 588 |
-
"grad_norm": 0.
|
| 589 |
"learning_rate": 5.663441603317208e-06,
|
| 590 |
-
"loss": 0.
|
| 591 |
"step": 4150
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 1.4512785072563925,
|
| 595 |
-
"grad_norm": 0.
|
| 596 |
"learning_rate": 5.490670352453352e-06,
|
| 597 |
-
"loss": 0.
|
| 598 |
"step": 4200
|
| 599 |
},
|
| 600 |
{
|
| 601 |
"epoch": 1.4685556323427782,
|
| 602 |
-
"grad_norm": 0.
|
| 603 |
"learning_rate": 5.3178991015894955e-06,
|
| 604 |
-
"loss": 0.
|
| 605 |
"step": 4250
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"epoch": 1.4858327574291637,
|
| 609 |
-
"grad_norm": 0.
|
| 610 |
"learning_rate": 5.14512785072564e-06,
|
| 611 |
-
"loss": 0.
|
| 612 |
"step": 4300
|
| 613 |
},
|
| 614 |
{
|
| 615 |
"epoch": 1.5031098825155493,
|
| 616 |
-
"grad_norm": 0.
|
| 617 |
"learning_rate": 4.972356599861784e-06,
|
| 618 |
-
"loss": 0.
|
| 619 |
"step": 4350
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"epoch": 1.520387007601935,
|
| 623 |
-
"grad_norm": 0.
|
| 624 |
"learning_rate": 4.799585348997927e-06,
|
| 625 |
-
"loss": 0.
|
| 626 |
"step": 4400
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 1.5376641326883207,
|
| 630 |
-
"grad_norm": 0.
|
| 631 |
"learning_rate": 4.6268140981340705e-06,
|
| 632 |
-
"loss": 0.
|
| 633 |
"step": 4450
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 1.5549412577747064,
|
| 637 |
-
"grad_norm": 0.
|
| 638 |
"learning_rate": 4.454042847270215e-06,
|
| 639 |
-
"loss": 0.
|
| 640 |
"step": 4500
|
| 641 |
},
|
| 642 |
{
|
| 643 |
"epoch": 1.572218382861092,
|
| 644 |
-
"grad_norm": 0.
|
| 645 |
"learning_rate": 4.281271596406358e-06,
|
| 646 |
-
"loss": 0.
|
| 647 |
"step": 4550
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"epoch": 1.5894955079474775,
|
| 651 |
-
"grad_norm": 0.
|
| 652 |
"learning_rate": 4.108500345542502e-06,
|
| 653 |
-
"loss": 0.
|
| 654 |
"step": 4600
|
| 655 |
},
|
| 656 |
{
|
| 657 |
"epoch": 1.6067726330338632,
|
| 658 |
-
"grad_norm": 0.
|
| 659 |
"learning_rate": 3.9357290946786455e-06,
|
| 660 |
-
"loss": 0.
|
| 661 |
"step": 4650
|
| 662 |
},
|
| 663 |
{
|
| 664 |
"epoch": 1.6240497581202487,
|
| 665 |
-
"grad_norm": 0.
|
| 666 |
"learning_rate": 3.7629578438147897e-06,
|
| 667 |
-
"loss": 0.
|
| 668 |
"step": 4700
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 1.6413268832066343,
|
| 672 |
-
"grad_norm": 0.
|
| 673 |
"learning_rate": 3.5901865929509334e-06,
|
| 674 |
-
"loss": 0.
|
| 675 |
"step": 4750
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 1.65860400829302,
|
| 679 |
-
"grad_norm": 0.
|
| 680 |
"learning_rate": 3.417415342087077e-06,
|
| 681 |
-
"loss": 0.
|
| 682 |
"step": 4800
|
| 683 |
},
|
| 684 |
{
|
| 685 |
"epoch": 1.6758811333794057,
|
| 686 |
-
"grad_norm": 0.
|
| 687 |
"learning_rate": 3.244644091223221e-06,
|
| 688 |
-
"loss": 0.
|
| 689 |
"step": 4850
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"epoch": 1.6931582584657914,
|
| 693 |
-
"grad_norm": 0.
|
| 694 |
"learning_rate": 3.0718728403593647e-06,
|
| 695 |
-
"loss": 0.
|
| 696 |
"step": 4900
|
| 697 |
},
|
| 698 |
{
|
| 699 |
"epoch": 1.710435383552177,
|
| 700 |
-
"grad_norm": 0.
|
| 701 |
"learning_rate": 2.8991015894955084e-06,
|
| 702 |
-
"loss": 0.
|
| 703 |
"step": 4950
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"epoch": 1.7277125086385625,
|
| 707 |
-
"grad_norm": 0.
|
| 708 |
"learning_rate": 2.726330338631652e-06,
|
| 709 |
-
"loss": 0.
|
| 710 |
"step": 5000
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 1.7449896337249482,
|
| 714 |
-
"grad_norm": 0.
|
| 715 |
"learning_rate": 2.553559087767796e-06,
|
| 716 |
-
"loss": 0.
|
| 717 |
"step": 5050
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 1.7622667588113337,
|
| 721 |
-
"grad_norm": 0.
|
| 722 |
"learning_rate": 2.3807878369039396e-06,
|
| 723 |
-
"loss": 0.
|
| 724 |
"step": 5100
|
| 725 |
},
|
| 726 |
{
|
| 727 |
"epoch": 1.7795438838977193,
|
| 728 |
-
"grad_norm": 0.
|
| 729 |
"learning_rate": 2.208016586040083e-06,
|
| 730 |
-
"loss": 0.
|
| 731 |
"step": 5150
|
| 732 |
},
|
| 733 |
{
|
| 734 |
"epoch": 1.796821008984105,
|
| 735 |
-
"grad_norm": 0.
|
| 736 |
"learning_rate": 2.0352453351762267e-06,
|
| 737 |
-
"loss": 0.
|
| 738 |
"step": 5200
|
| 739 |
},
|
| 740 |
{
|
| 741 |
"epoch": 1.8140981340704907,
|
| 742 |
-
"grad_norm": 0.
|
| 743 |
"learning_rate": 1.8624740843123707e-06,
|
| 744 |
-
"loss": 0.
|
| 745 |
"step": 5250
|
| 746 |
},
|
| 747 |
{
|
| 748 |
"epoch": 1.8313752591568764,
|
| 749 |
-
"grad_norm": 0.
|
| 750 |
"learning_rate": 1.6897028334485144e-06,
|
| 751 |
-
"loss": 0.
|
| 752 |
"step": 5300
|
| 753 |
},
|
| 754 |
{
|
| 755 |
"epoch": 1.848652384243262,
|
| 756 |
-
"grad_norm": 0.
|
| 757 |
"learning_rate": 1.5169315825846582e-06,
|
| 758 |
-
"loss": 0.
|
| 759 |
"step": 5350
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 1.8659295093296475,
|
| 763 |
-
"grad_norm": 0.
|
| 764 |
"learning_rate": 1.344160331720802e-06,
|
| 765 |
-
"loss": 0.
|
| 766 |
"step": 5400
|
| 767 |
},
|
| 768 |
{
|
| 769 |
"epoch": 1.8832066344160332,
|
| 770 |
-
"grad_norm": 0.
|
| 771 |
"learning_rate": 1.1713890808569455e-06,
|
| 772 |
-
"loss": 0.
|
| 773 |
"step": 5450
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"epoch": 1.9004837595024187,
|
| 777 |
-
"grad_norm": 0.
|
| 778 |
"learning_rate": 9.986178299930892e-07,
|
| 779 |
-
"loss": 0.
|
| 780 |
"step": 5500
|
| 781 |
},
|
| 782 |
{
|
| 783 |
"epoch": 1.9177608845888043,
|
| 784 |
-
"grad_norm": 0.
|
| 785 |
"learning_rate": 8.25846579129233e-07,
|
| 786 |
-
"loss": 0.
|
| 787 |
"step": 5550
|
| 788 |
},
|
| 789 |
{
|
| 790 |
"epoch": 1.93503800967519,
|
| 791 |
-
"grad_norm": 0.
|
| 792 |
"learning_rate": 6.530753282653766e-07,
|
| 793 |
-
"loss": 0.
|
| 794 |
"step": 5600
|
| 795 |
},
|
| 796 |
{
|
| 797 |
"epoch": 1.9523151347615757,
|
| 798 |
-
"grad_norm": 0.
|
| 799 |
"learning_rate": 4.803040774015205e-07,
|
| 800 |
-
"loss": 0.
|
| 801 |
"step": 5650
|
| 802 |
},
|
| 803 |
{
|
| 804 |
"epoch": 1.9695922598479614,
|
| 805 |
-
"grad_norm": 0.
|
| 806 |
"learning_rate": 3.0753282653766415e-07,
|
| 807 |
-
"loss": 0.
|
| 808 |
"step": 5700
|
| 809 |
},
|
| 810 |
{
|
| 811 |
"epoch": 1.986869384934347,
|
| 812 |
-
"grad_norm": 0.
|
| 813 |
"learning_rate": 1.3476157567380787e-07,
|
| 814 |
-
"loss": 0.
|
| 815 |
"step": 5750
|
| 816 |
}
|
| 817 |
],
|
|
@@ -832,7 +832,7 @@
|
|
| 832 |
"attributes": {}
|
| 833 |
}
|
| 834 |
},
|
| 835 |
-
"total_flos": 1.
|
| 836 |
"train_batch_size": 64,
|
| 837 |
"trial_name": null,
|
| 838 |
"trial_params": null
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.017277125086385625,
|
| 14 |
+
"grad_norm": 0.04233066737651825,
|
| 15 |
"learning_rate": 1.9830684174153422e-05,
|
| 16 |
+
"loss": 0.031055023670196535,
|
| 17 |
"step": 50
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.03455425017277125,
|
| 21 |
+
"grad_norm": 0.02456316538155079,
|
| 22 |
"learning_rate": 1.9657912923289565e-05,
|
| 23 |
+
"loss": 0.0218997859954834,
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.051831375259156875,
|
| 28 |
+
"grad_norm": 0.0479276068508625,
|
| 29 |
"learning_rate": 1.948514167242571e-05,
|
| 30 |
+
"loss": 0.01705536723136902,
|
| 31 |
"step": 150
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.0691085003455425,
|
| 35 |
+
"grad_norm": 0.027037542313337326,
|
| 36 |
"learning_rate": 1.9312370421561853e-05,
|
| 37 |
+
"loss": 0.0129636549949646,
|
| 38 |
"step": 200
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.08638562543192813,
|
| 42 |
+
"grad_norm": 0.017864206805825233,
|
| 43 |
"learning_rate": 1.9139599170698e-05,
|
| 44 |
+
"loss": 0.01030057668685913,
|
| 45 |
"step": 250
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.10366275051831375,
|
| 49 |
+
"grad_norm": 0.019284004345536232,
|
| 50 |
"learning_rate": 1.896682791983414e-05,
|
| 51 |
+
"loss": 0.008808751702308654,
|
| 52 |
"step": 300
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 0.12093987560469938,
|
| 56 |
+
"grad_norm": 0.02335270307958126,
|
| 57 |
"learning_rate": 1.8794056668970284e-05,
|
| 58 |
+
"loss": 0.007309812307357788,
|
| 59 |
"step": 350
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 0.138217000691085,
|
| 63 |
+
"grad_norm": 0.02300521731376648,
|
| 64 |
"learning_rate": 1.862128541810643e-05,
|
| 65 |
+
"loss": 0.006364336013793945,
|
| 66 |
"step": 400
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 0.15549412577747063,
|
| 70 |
+
"grad_norm": 0.019678954035043716,
|
| 71 |
"learning_rate": 1.8448514167242572e-05,
|
| 72 |
+
"loss": 0.005564928650856018,
|
| 73 |
"step": 450
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 0.17277125086385625,
|
| 77 |
+
"grad_norm": 0.018229542300105095,
|
| 78 |
"learning_rate": 1.8275742916378715e-05,
|
| 79 |
+
"loss": 0.004760843515396118,
|
| 80 |
"step": 500
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.19004837595024188,
|
| 84 |
+
"grad_norm": 0.022495370358228683,
|
| 85 |
"learning_rate": 1.810297166551486e-05,
|
| 86 |
+
"loss": 0.004270125031471252,
|
| 87 |
"step": 550
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.2073255010366275,
|
| 91 |
+
"grad_norm": 0.024747125804424286,
|
| 92 |
"learning_rate": 1.7930200414651003e-05,
|
| 93 |
+
"loss": 0.0037263017892837525,
|
| 94 |
"step": 600
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 0.22460262612301313,
|
| 98 |
+
"grad_norm": 0.017567673698067665,
|
| 99 |
"learning_rate": 1.775742916378715e-05,
|
| 100 |
+
"loss": 0.003303785026073456,
|
| 101 |
"step": 650
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 0.24187975120939875,
|
| 105 |
+
"grad_norm": 0.014267382211983204,
|
| 106 |
"learning_rate": 1.758465791292329e-05,
|
| 107 |
+
"loss": 0.0030617621541023255,
|
| 108 |
"step": 700
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 0.2591568762957844,
|
| 112 |
+
"grad_norm": 0.021974245086312294,
|
| 113 |
"learning_rate": 1.7411886662059434e-05,
|
| 114 |
+
"loss": 0.00269344836473465,
|
| 115 |
"step": 750
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 0.27643400138217,
|
| 119 |
+
"grad_norm": 0.018784578889608383,
|
| 120 |
"learning_rate": 1.723911541119558e-05,
|
| 121 |
+
"loss": 0.002522537112236023,
|
| 122 |
"step": 800
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.29371112646855563,
|
| 126 |
+
"grad_norm": 0.013223987072706223,
|
| 127 |
"learning_rate": 1.7066344160331722e-05,
|
| 128 |
+
"loss": 0.002301780879497528,
|
| 129 |
"step": 850
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.31098825155494125,
|
| 133 |
+
"grad_norm": 0.025622107088565826,
|
| 134 |
"learning_rate": 1.6893572909467865e-05,
|
| 135 |
+
"loss": 0.002054130434989929,
|
| 136 |
"step": 900
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.3282653766413269,
|
| 140 |
+
"grad_norm": 0.019053716212511063,
|
| 141 |
"learning_rate": 1.6720801658604007e-05,
|
| 142 |
+
"loss": 0.0018760859966278076,
|
| 143 |
"step": 950
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 0.3455425017277125,
|
| 147 |
+
"grad_norm": 0.01584235392510891,
|
| 148 |
"learning_rate": 1.6548030407740153e-05,
|
| 149 |
+
"loss": 0.0017820017039775848,
|
| 150 |
"step": 1000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 0.36281962681409813,
|
| 154 |
+
"grad_norm": 0.023229464888572693,
|
| 155 |
"learning_rate": 1.63752591568763e-05,
|
| 156 |
+
"loss": 0.0017064496874809264,
|
| 157 |
"step": 1050
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 0.38009675190048375,
|
| 161 |
+
"grad_norm": 0.025416497141122818,
|
| 162 |
"learning_rate": 1.620248790601244e-05,
|
| 163 |
+
"loss": 0.0016316136717796326,
|
| 164 |
"step": 1100
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.3973738769868694,
|
| 168 |
+
"grad_norm": 0.013280889019370079,
|
| 169 |
"learning_rate": 1.6029716655148584e-05,
|
| 170 |
+
"loss": 0.0015296098589897156,
|
| 171 |
"step": 1150
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.414651002073255,
|
| 175 |
+
"grad_norm": 0.012821532785892487,
|
| 176 |
"learning_rate": 1.585694540428473e-05,
|
| 177 |
+
"loss": 0.001470087617635727,
|
| 178 |
"step": 1200
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 0.43192812715964063,
|
| 182 |
+
"grad_norm": 0.011737545020878315,
|
| 183 |
"learning_rate": 1.5684174153420872e-05,
|
| 184 |
+
"loss": 0.0013830630481243133,
|
| 185 |
"step": 1250
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 0.44920525224602625,
|
| 189 |
+
"grad_norm": 0.017931129783391953,
|
| 190 |
"learning_rate": 1.5511402902557015e-05,
|
| 191 |
+
"loss": 0.001326202005147934,
|
| 192 |
"step": 1300
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 0.4664823773324119,
|
| 196 |
+
"grad_norm": 0.01637415960431099,
|
| 197 |
"learning_rate": 1.5338631651693157e-05,
|
| 198 |
+
"loss": 0.001227569431066513,
|
| 199 |
"step": 1350
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.4837595024187975,
|
| 203 |
+
"grad_norm": 0.020321404561400414,
|
| 204 |
"learning_rate": 1.5165860400829305e-05,
|
| 205 |
+
"loss": 0.0012098293006420136,
|
| 206 |
"step": 1400
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.5010366275051832,
|
| 210 |
+
"grad_norm": 0.013214107602834702,
|
| 211 |
"learning_rate": 1.4993089149965447e-05,
|
| 212 |
+
"loss": 0.001184120774269104,
|
| 213 |
"step": 1450
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.5183137525915688,
|
| 217 |
+
"grad_norm": 0.011687219142913818,
|
| 218 |
"learning_rate": 1.482031789910159e-05,
|
| 219 |
+
"loss": 0.0010832401365041733,
|
| 220 |
"step": 1500
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 0.5355908776779544,
|
| 224 |
+
"grad_norm": 0.013312515802681446,
|
| 225 |
"learning_rate": 1.4647546648237734e-05,
|
| 226 |
+
"loss": 0.0010921508818864823,
|
| 227 |
"step": 1550
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 0.55286800276434,
|
| 231 |
+
"grad_norm": 0.02098967134952545,
|
| 232 |
"learning_rate": 1.447477539737388e-05,
|
| 233 |
+
"loss": 0.0010538918524980545,
|
| 234 |
"step": 1600
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.5701451278507257,
|
| 238 |
+
"grad_norm": 0.018116634339094162,
|
| 239 |
"learning_rate": 1.4302004146510022e-05,
|
| 240 |
+
"loss": 0.0009815704822540283,
|
| 241 |
"step": 1650
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 0.5874222529371113,
|
| 245 |
+
"grad_norm": 0.020911073312163353,
|
| 246 |
"learning_rate": 1.4129232895646165e-05,
|
| 247 |
+
"loss": 0.000898132175207138,
|
| 248 |
"step": 1700
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 0.6046993780234969,
|
| 252 |
+
"grad_norm": 0.007671894505620003,
|
| 253 |
"learning_rate": 1.3956461644782309e-05,
|
| 254 |
+
"loss": 0.0009254106879234314,
|
| 255 |
"step": 1750
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.6219765031098825,
|
| 259 |
+
"grad_norm": 0.015347246080636978,
|
| 260 |
"learning_rate": 1.3783690393918455e-05,
|
| 261 |
+
"loss": 0.0009475262463092804,
|
| 262 |
"step": 1800
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 0.6392536281962682,
|
| 266 |
+
"grad_norm": 0.016122082248330116,
|
| 267 |
"learning_rate": 1.3610919143054597e-05,
|
| 268 |
+
"loss": 0.0008720143884420394,
|
| 269 |
"step": 1850
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 0.6565307532826538,
|
| 273 |
+
"grad_norm": 0.01304946281015873,
|
| 274 |
"learning_rate": 1.343814789219074e-05,
|
| 275 |
+
"loss": 0.0009039641171693801,
|
| 276 |
"step": 1900
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 0.6738078783690394,
|
| 280 |
+
"grad_norm": 0.014165911823511124,
|
| 281 |
"learning_rate": 1.3265376641326884e-05,
|
| 282 |
+
"loss": 0.0008714392036199569,
|
| 283 |
"step": 1950
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 0.691085003455425,
|
| 287 |
+
"grad_norm": 0.018862200900912285,
|
| 288 |
"learning_rate": 1.309260539046303e-05,
|
| 289 |
+
"loss": 0.0008268249034881592,
|
| 290 |
"step": 2000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 0.7083621285418107,
|
| 294 |
+
"grad_norm": 0.015998877584934235,
|
| 295 |
"learning_rate": 1.2919834139599172e-05,
|
| 296 |
+
"loss": 0.0007983983308076859,
|
| 297 |
"step": 2050
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.7256392536281963,
|
| 301 |
+
"grad_norm": 0.010671521537005901,
|
| 302 |
"learning_rate": 1.2747062888735315e-05,
|
| 303 |
+
"loss": 0.0007587751001119614,
|
| 304 |
"step": 2100
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 0.7429163787145819,
|
| 308 |
+
"grad_norm": 0.013473653234541416,
|
| 309 |
"learning_rate": 1.2574291637871459e-05,
|
| 310 |
+
"loss": 0.0008197207748889923,
|
| 311 |
"step": 2150
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 0.7601935038009675,
|
| 315 |
+
"grad_norm": 0.015992162749171257,
|
| 316 |
"learning_rate": 1.2401520387007605e-05,
|
| 317 |
+
"loss": 0.0007695946097373963,
|
| 318 |
"step": 2200
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 0.7774706288873532,
|
| 322 |
+
"grad_norm": 0.016132045537233353,
|
| 323 |
"learning_rate": 1.2228749136143747e-05,
|
| 324 |
+
"loss": 0.0007311414182186127,
|
| 325 |
"step": 2250
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.7947477539737388,
|
| 329 |
+
"grad_norm": 0.017123883590102196,
|
| 330 |
"learning_rate": 1.205597788527989e-05,
|
| 331 |
+
"loss": 0.0007468417286872863,
|
| 332 |
"step": 2300
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 0.8120248790601244,
|
| 336 |
+
"grad_norm": 0.014997811056673527,
|
| 337 |
"learning_rate": 1.1883206634416034e-05,
|
| 338 |
+
"loss": 0.000710819736123085,
|
| 339 |
"step": 2350
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.82930200414651,
|
| 343 |
+
"grad_norm": 0.015333279967308044,
|
| 344 |
"learning_rate": 1.1710435383552178e-05,
|
| 345 |
+
"loss": 0.0007229457050561905,
|
| 346 |
"step": 2400
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 0.8465791292328957,
|
| 350 |
+
"grad_norm": 0.01576118730008602,
|
| 351 |
"learning_rate": 1.1537664132688322e-05,
|
| 352 |
+
"loss": 0.0007234162837266922,
|
| 353 |
"step": 2450
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 0.8638562543192813,
|
| 357 |
+
"grad_norm": 0.011966513469815254,
|
| 358 |
"learning_rate": 1.1364892881824465e-05,
|
| 359 |
+
"loss": 0.0006654725223779679,
|
| 360 |
"step": 2500
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 0.8811333794056669,
|
| 364 |
+
"grad_norm": 0.011741974391043186,
|
| 365 |
"learning_rate": 1.1192121630960609e-05,
|
| 366 |
+
"loss": 0.0007037873566150666,
|
| 367 |
"step": 2550
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 0.8984105044920525,
|
| 371 |
+
"grad_norm": 0.009139756672084332,
|
| 372 |
"learning_rate": 1.1019350380096753e-05,
|
| 373 |
+
"loss": 0.0006298752129077911,
|
| 374 |
"step": 2600
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 0.9156876295784382,
|
| 378 |
+
"grad_norm": 0.00690425792708993,
|
| 379 |
"learning_rate": 1.0846579129232897e-05,
|
| 380 |
+
"loss": 0.0006032006442546845,
|
| 381 |
"step": 2650
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 0.9329647546648238,
|
| 385 |
+
"grad_norm": 0.015512525103986263,
|
| 386 |
"learning_rate": 1.067380787836904e-05,
|
| 387 |
+
"loss": 0.0006475722044706344,
|
| 388 |
"step": 2700
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 0.9502418797512094,
|
| 392 |
+
"grad_norm": 0.012796995230019093,
|
| 393 |
"learning_rate": 1.0501036627505184e-05,
|
| 394 |
+
"loss": 0.0005854096636176109,
|
| 395 |
"step": 2750
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 0.967519004837595,
|
| 399 |
+
"grad_norm": 0.01241610199213028,
|
| 400 |
"learning_rate": 1.0328265376641328e-05,
|
| 401 |
+
"loss": 0.000655377209186554,
|
| 402 |
"step": 2800
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 0.9847961299239807,
|
| 406 |
+
"grad_norm": 0.008658281527459621,
|
| 407 |
"learning_rate": 1.0155494125777472e-05,
|
| 408 |
+
"loss": 0.0006553089618682861,
|
| 409 |
"step": 2850
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 1.0020732550103664,
|
| 413 |
+
"grad_norm": 0.012649709358811378,
|
| 414 |
"learning_rate": 9.982722874913615e-06,
|
| 415 |
+
"loss": 0.0005679767578840256,
|
| 416 |
"step": 2900
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 1.0193503800967518,
|
| 420 |
+
"grad_norm": 0.0128778126090765,
|
| 421 |
"learning_rate": 9.809951624049759e-06,
|
| 422 |
+
"loss": 0.0005776724964380264,
|
| 423 |
"step": 2950
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 1.0366275051831375,
|
| 427 |
+
"grad_norm": 0.008131473325192928,
|
| 428 |
"learning_rate": 9.637180373185903e-06,
|
| 429 |
+
"loss": 0.0004936759918928147,
|
| 430 |
"step": 3000
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 1.0539046302695232,
|
| 434 |
+
"grad_norm": 0.010393351316452026,
|
| 435 |
"learning_rate": 9.464409122322047e-06,
|
| 436 |
+
"loss": 0.0004966352134943008,
|
| 437 |
"step": 3050
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 1.0711817553559089,
|
| 441 |
+
"grad_norm": 0.00899389199912548,
|
| 442 |
"learning_rate": 9.29163787145819e-06,
|
| 443 |
+
"loss": 0.0005515943840146064,
|
| 444 |
"step": 3100
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 1.0884588804422943,
|
| 448 |
+
"grad_norm": 0.008057264611124992,
|
| 449 |
"learning_rate": 9.118866620594334e-06,
|
| 450 |
+
"loss": 0.0005295329168438912,
|
| 451 |
"step": 3150
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 1.10573600552868,
|
| 455 |
+
"grad_norm": 0.00806473195552826,
|
| 456 |
"learning_rate": 8.946095369730478e-06,
|
| 457 |
+
"loss": 0.0005242252722382545,
|
| 458 |
"step": 3200
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 1.1230131306150657,
|
| 462 |
+
"grad_norm": 0.005094733089208603,
|
| 463 |
"learning_rate": 8.773324118866622e-06,
|
| 464 |
+
"loss": 0.000492001548409462,
|
| 465 |
"step": 3250
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 1.1402902557014514,
|
| 469 |
+
"grad_norm": 0.010061942972242832,
|
| 470 |
"learning_rate": 8.600552868002765e-06,
|
| 471 |
+
"loss": 0.0005547917261719703,
|
| 472 |
"step": 3300
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 1.1575673807878368,
|
| 476 |
+
"grad_norm": 0.01032029464840889,
|
| 477 |
"learning_rate": 8.427781617138909e-06,
|
| 478 |
+
"loss": 0.0004815058037638664,
|
| 479 |
"step": 3350
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 1.1748445058742225,
|
| 483 |
+
"grad_norm": 0.008551975712180138,
|
| 484 |
"learning_rate": 8.255010366275053e-06,
|
| 485 |
+
"loss": 0.0005315033346414566,
|
| 486 |
"step": 3400
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 1.1921216309606082,
|
| 490 |
+
"grad_norm": 0.02174876257777214,
|
| 491 |
"learning_rate": 8.082239115411197e-06,
|
| 492 |
+
"loss": 0.00048085026443004606,
|
| 493 |
"step": 3450
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 1.2093987560469939,
|
| 497 |
+
"grad_norm": 0.015282322652637959,
|
| 498 |
"learning_rate": 7.90946786454734e-06,
|
| 499 |
+
"loss": 0.0004923569411039352,
|
| 500 |
"step": 3500
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 1.2266758811333793,
|
| 504 |
+
"grad_norm": 0.008838672190904617,
|
| 505 |
"learning_rate": 7.736696613683484e-06,
|
| 506 |
+
"loss": 0.00046344950795173645,
|
| 507 |
"step": 3550
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 1.243953006219765,
|
| 511 |
+
"grad_norm": 0.005751848686486483,
|
| 512 |
"learning_rate": 7.563925362819627e-06,
|
| 513 |
+
"loss": 0.00047770991921424865,
|
| 514 |
"step": 3600
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 1.2612301313061507,
|
| 518 |
+
"grad_norm": 0.011446288786828518,
|
| 519 |
"learning_rate": 7.391154111955771e-06,
|
| 520 |
+
"loss": 0.0005263676866889,
|
| 521 |
"step": 3650
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 1.2785072563925364,
|
| 525 |
+
"grad_norm": 0.007091975770890713,
|
| 526 |
"learning_rate": 7.218382861091915e-06,
|
| 527 |
+
"loss": 0.0004377379268407822,
|
| 528 |
"step": 3700
|
| 529 |
},
|
| 530 |
{
|
| 531 |
"epoch": 1.2957843814789218,
|
| 532 |
+
"grad_norm": 0.004327921196818352,
|
| 533 |
"learning_rate": 7.045611610228059e-06,
|
| 534 |
+
"loss": 0.0004530222341418266,
|
| 535 |
"step": 3750
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"epoch": 1.3130615065653075,
|
| 539 |
+
"grad_norm": 0.010974051430821419,
|
| 540 |
"learning_rate": 6.872840359364202e-06,
|
| 541 |
+
"loss": 0.00045473653823137285,
|
| 542 |
"step": 3800
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 1.3303386316516932,
|
| 546 |
+
"grad_norm": 0.010680513456463814,
|
| 547 |
"learning_rate": 6.700069108500346e-06,
|
| 548 |
+
"loss": 0.00047681089490652083,
|
| 549 |
"step": 3850
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 1.3476157567380787,
|
| 553 |
+
"grad_norm": 0.007604369893670082,
|
| 554 |
"learning_rate": 6.52729785763649e-06,
|
| 555 |
+
"loss": 0.00046367585659027097,
|
| 556 |
"step": 3900
|
| 557 |
},
|
| 558 |
{
|
| 559 |
"epoch": 1.3648928818244643,
|
| 560 |
+
"grad_norm": 0.009516065008938313,
|
| 561 |
"learning_rate": 6.354526606772633e-06,
|
| 562 |
+
"loss": 0.00047222837805747985,
|
| 563 |
"step": 3950
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 1.38217000691085,
|
| 567 |
+
"grad_norm": 0.005651406012475491,
|
| 568 |
"learning_rate": 6.181755355908777e-06,
|
| 569 |
+
"loss": 0.00048513900488615034,
|
| 570 |
"step": 4000
|
| 571 |
},
|
| 572 |
{
|
| 573 |
"epoch": 1.3994471319972357,
|
| 574 |
+
"grad_norm": 0.00899775605648756,
|
| 575 |
"learning_rate": 6.0089841050449205e-06,
|
| 576 |
+
"loss": 0.00045138660818338394,
|
| 577 |
"step": 4050
|
| 578 |
},
|
| 579 |
{
|
| 580 |
"epoch": 1.4167242570836214,
|
| 581 |
+
"grad_norm": 0.01663408800959587,
|
| 582 |
"learning_rate": 5.836212854181065e-06,
|
| 583 |
+
"loss": 0.0004322423785924911,
|
| 584 |
"step": 4100
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 1.4340013821700068,
|
| 588 |
+
"grad_norm": 0.01465668622404337,
|
| 589 |
"learning_rate": 5.663441603317208e-06,
|
| 590 |
+
"loss": 0.00046591993421316144,
|
| 591 |
"step": 4150
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 1.4512785072563925,
|
| 595 |
+
"grad_norm": 0.0021400833502411842,
|
| 596 |
"learning_rate": 5.490670352453352e-06,
|
| 597 |
+
"loss": 0.0004399510845541954,
|
| 598 |
"step": 4200
|
| 599 |
},
|
| 600 |
{
|
| 601 |
"epoch": 1.4685556323427782,
|
| 602 |
+
"grad_norm": 0.011773742735385895,
|
| 603 |
"learning_rate": 5.3178991015894955e-06,
|
| 604 |
+
"loss": 0.00040802594274282457,
|
| 605 |
"step": 4250
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"epoch": 1.4858327574291637,
|
| 609 |
+
"grad_norm": 0.011613302864134312,
|
| 610 |
"learning_rate": 5.14512785072564e-06,
|
| 611 |
+
"loss": 0.0004606429487466812,
|
| 612 |
"step": 4300
|
| 613 |
},
|
| 614 |
{
|
| 615 |
"epoch": 1.5031098825155493,
|
| 616 |
+
"grad_norm": 0.008227858692407608,
|
| 617 |
"learning_rate": 4.972356599861784e-06,
|
| 618 |
+
"loss": 0.0004804874956607819,
|
| 619 |
"step": 4350
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"epoch": 1.520387007601935,
|
| 623 |
+
"grad_norm": 0.011196565814316273,
|
| 624 |
"learning_rate": 4.799585348997927e-06,
|
| 625 |
+
"loss": 0.00046057090163230895,
|
| 626 |
"step": 4400
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 1.5376641326883207,
|
| 630 |
+
"grad_norm": 0.00936612393707037,
|
| 631 |
"learning_rate": 4.6268140981340705e-06,
|
| 632 |
+
"loss": 0.0004251852631568909,
|
| 633 |
"step": 4450
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 1.5549412577747064,
|
| 637 |
+
"grad_norm": 0.01185729168355465,
|
| 638 |
"learning_rate": 4.454042847270215e-06,
|
| 639 |
+
"loss": 0.0004098330810666084,
|
| 640 |
"step": 4500
|
| 641 |
},
|
| 642 |
{
|
| 643 |
"epoch": 1.572218382861092,
|
| 644 |
+
"grad_norm": 0.013988607563078403,
|
| 645 |
"learning_rate": 4.281271596406358e-06,
|
| 646 |
+
"loss": 0.0004165947064757347,
|
| 647 |
"step": 4550
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"epoch": 1.5894955079474775,
|
| 651 |
+
"grad_norm": 0.014477470889687538,
|
| 652 |
"learning_rate": 4.108500345542502e-06,
|
| 653 |
+
"loss": 0.0004135962948203087,
|
| 654 |
"step": 4600
|
| 655 |
},
|
| 656 |
{
|
| 657 |
"epoch": 1.6067726330338632,
|
| 658 |
+
"grad_norm": 0.009254582226276398,
|
| 659 |
"learning_rate": 3.9357290946786455e-06,
|
| 660 |
+
"loss": 0.0004228399693965912,
|
| 661 |
"step": 4650
|
| 662 |
},
|
| 663 |
{
|
| 664 |
"epoch": 1.6240497581202487,
|
| 665 |
+
"grad_norm": 0.00860146339982748,
|
| 666 |
"learning_rate": 3.7629578438147897e-06,
|
| 667 |
+
"loss": 0.00041457165032625197,
|
| 668 |
"step": 4700
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 1.6413268832066343,
|
| 672 |
+
"grad_norm": 0.004228601232171059,
|
| 673 |
"learning_rate": 3.5901865929509334e-06,
|
| 674 |
+
"loss": 0.00036855176091194153,
|
| 675 |
"step": 4750
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 1.65860400829302,
|
| 679 |
+
"grad_norm": 0.010816280730068684,
|
| 680 |
"learning_rate": 3.417415342087077e-06,
|
| 681 |
+
"loss": 0.00047114014625549317,
|
| 682 |
"step": 4800
|
| 683 |
},
|
| 684 |
{
|
| 685 |
"epoch": 1.6758811333794057,
|
| 686 |
+
"grad_norm": 0.009063015691936016,
|
| 687 |
"learning_rate": 3.244644091223221e-06,
|
| 688 |
+
"loss": 0.00044060099869966504,
|
| 689 |
"step": 4850
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"epoch": 1.6931582584657914,
|
| 693 |
+
"grad_norm": 0.009867575950920582,
|
| 694 |
"learning_rate": 3.0718728403593647e-06,
|
| 695 |
+
"loss": 0.0003931435942649841,
|
| 696 |
"step": 4900
|
| 697 |
},
|
| 698 |
{
|
| 699 |
"epoch": 1.710435383552177,
|
| 700 |
+
"grad_norm": 0.010455707088112831,
|
| 701 |
"learning_rate": 2.8991015894955084e-06,
|
| 702 |
+
"loss": 0.00042293861508369446,
|
| 703 |
"step": 4950
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"epoch": 1.7277125086385625,
|
| 707 |
+
"grad_norm": 0.008534860797226429,
|
| 708 |
"learning_rate": 2.726330338631652e-06,
|
| 709 |
+
"loss": 0.00037612821906805036,
|
| 710 |
"step": 5000
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 1.7449896337249482,
|
| 714 |
+
"grad_norm": 0.009214954450726509,
|
| 715 |
"learning_rate": 2.553559087767796e-06,
|
| 716 |
+
"loss": 0.0004578226804733276,
|
| 717 |
"step": 5050
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 1.7622667588113337,
|
| 721 |
+
"grad_norm": 0.0113722188398242,
|
| 722 |
"learning_rate": 2.3807878369039396e-06,
|
| 723 |
+
"loss": 0.0003856651484966278,
|
| 724 |
"step": 5100
|
| 725 |
},
|
| 726 |
{
|
| 727 |
"epoch": 1.7795438838977193,
|
| 728 |
+
"grad_norm": 0.008885476738214493,
|
| 729 |
"learning_rate": 2.208016586040083e-06,
|
| 730 |
+
"loss": 0.0004339037090539932,
|
| 731 |
"step": 5150
|
| 732 |
},
|
| 733 |
{
|
| 734 |
"epoch": 1.796821008984105,
|
| 735 |
+
"grad_norm": 0.004335694946348667,
|
| 736 |
"learning_rate": 2.0352453351762267e-06,
|
| 737 |
+
"loss": 0.0003705260530114174,
|
| 738 |
"step": 5200
|
| 739 |
},
|
| 740 |
{
|
| 741 |
"epoch": 1.8140981340704907,
|
| 742 |
+
"grad_norm": 0.012399842031300068,
|
| 743 |
"learning_rate": 1.8624740843123707e-06,
|
| 744 |
+
"loss": 0.00040423810482025145,
|
| 745 |
"step": 5250
|
| 746 |
},
|
| 747 |
{
|
| 748 |
"epoch": 1.8313752591568764,
|
| 749 |
+
"grad_norm": 0.007849883288145065,
|
| 750 |
"learning_rate": 1.6897028334485144e-06,
|
| 751 |
+
"loss": 0.00040913678705692294,
|
| 752 |
"step": 5300
|
| 753 |
},
|
| 754 |
{
|
| 755 |
"epoch": 1.848652384243262,
|
| 756 |
+
"grad_norm": 0.012463894672691822,
|
| 757 |
"learning_rate": 1.5169315825846582e-06,
|
| 758 |
+
"loss": 0.0003825708106160164,
|
| 759 |
"step": 5350
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 1.8659295093296475,
|
| 763 |
+
"grad_norm": 0.006620150990784168,
|
| 764 |
"learning_rate": 1.344160331720802e-06,
|
| 765 |
+
"loss": 0.00039904817938804624,
|
| 766 |
"step": 5400
|
| 767 |
},
|
| 768 |
{
|
| 769 |
"epoch": 1.8832066344160332,
|
| 770 |
+
"grad_norm": 0.007635048590600491,
|
| 771 |
"learning_rate": 1.1713890808569455e-06,
|
| 772 |
+
"loss": 0.0004040185362100601,
|
| 773 |
"step": 5450
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"epoch": 1.9004837595024187,
|
| 777 |
+
"grad_norm": 0.005908569786697626,
|
| 778 |
"learning_rate": 9.986178299930892e-07,
|
| 779 |
+
"loss": 0.0003646549955010414,
|
| 780 |
"step": 5500
|
| 781 |
},
|
| 782 |
{
|
| 783 |
"epoch": 1.9177608845888043,
|
| 784 |
+
"grad_norm": 0.006155763287097216,
|
| 785 |
"learning_rate": 8.25846579129233e-07,
|
| 786 |
+
"loss": 0.0003758473694324493,
|
| 787 |
"step": 5550
|
| 788 |
},
|
| 789 |
{
|
| 790 |
"epoch": 1.93503800967519,
|
| 791 |
+
"grad_norm": 0.010079983621835709,
|
| 792 |
"learning_rate": 6.530753282653766e-07,
|
| 793 |
+
"loss": 0.0003937266021966934,
|
| 794 |
"step": 5600
|
| 795 |
},
|
| 796 |
{
|
| 797 |
"epoch": 1.9523151347615757,
|
| 798 |
+
"grad_norm": 0.013179412111639977,
|
| 799 |
"learning_rate": 4.803040774015205e-07,
|
| 800 |
+
"loss": 0.0003903582692146301,
|
| 801 |
"step": 5650
|
| 802 |
},
|
| 803 |
{
|
| 804 |
"epoch": 1.9695922598479614,
|
| 805 |
+
"grad_norm": 0.004289121367037296,
|
| 806 |
"learning_rate": 3.0753282653766415e-07,
|
| 807 |
+
"loss": 0.00037710186094045637,
|
| 808 |
"step": 5700
|
| 809 |
},
|
| 810 |
{
|
| 811 |
"epoch": 1.986869384934347,
|
| 812 |
+
"grad_norm": 0.0074286069720983505,
|
| 813 |
"learning_rate": 1.3476157567380787e-07,
|
| 814 |
+
"loss": 0.0003793236240744591,
|
| 815 |
"step": 5750
|
| 816 |
}
|
| 817 |
],
|
|
|
|
| 832 |
"attributes": {}
|
| 833 |
}
|
| 834 |
},
|
| 835 |
+
"total_flos": 1.818477805876812e+16,
|
| 836 |
"train_batch_size": 64,
|
| 837 |
"trial_name": null,
|
| 838 |
"trial_params": null
|
Internal-Run-EN/SpanBased-DeBERTa/config.json
CHANGED
|
@@ -24,49 +24,41 @@
|
|
| 24 |
"11": "CREDITCARDCVV",
|
| 25 |
"12": "CREDITCARDISSUER",
|
| 26 |
"13": "CREDITCARDNUMBER",
|
| 27 |
-
"14": "
|
| 28 |
-
"15": "
|
| 29 |
-
"16": "
|
| 30 |
-
"17": "
|
| 31 |
-
"18": "
|
| 32 |
-
"19": "
|
| 33 |
-
"20": "
|
| 34 |
-
"21": "
|
| 35 |
-
"22": "
|
| 36 |
-
"23": "
|
| 37 |
-
"24": "
|
| 38 |
-
"25": "
|
| 39 |
-
"26": "
|
| 40 |
-
"27": "
|
| 41 |
-
"28": "
|
| 42 |
-
"29": "
|
| 43 |
-
"30": "
|
| 44 |
-
"31": "
|
| 45 |
-
"32": "
|
| 46 |
-
"33": "
|
| 47 |
-
"34": "
|
| 48 |
-
"35": "
|
| 49 |
-
"36": "
|
| 50 |
-
"37": "
|
| 51 |
-
"38": "
|
| 52 |
-
"39": "
|
| 53 |
-
"40": "
|
| 54 |
-
"41": "
|
| 55 |
-
"42": "
|
| 56 |
-
"43": "
|
| 57 |
-
"44": "
|
| 58 |
-
"45": "
|
| 59 |
-
"46": "
|
| 60 |
-
"47": "
|
| 61 |
-
"48": "
|
| 62 |
-
"49": "STREET",
|
| 63 |
-
"50": "TIME",
|
| 64 |
-
"51": "URL",
|
| 65 |
-
"52": "USERAGENT",
|
| 66 |
-
"53": "USERNAME",
|
| 67 |
-
"54": "VEHICLEVIN",
|
| 68 |
-
"55": "VEHICLEVRM",
|
| 69 |
-
"56": "ZIPCODE"
|
| 70 |
},
|
| 71 |
"initializer_range": 0.02,
|
| 72 |
"intermediate_size": 3072,
|
|
@@ -84,50 +76,42 @@
|
|
| 84 |
"CREDITCARDCVV": 11,
|
| 85 |
"CREDITCARDISSUER": 12,
|
| 86 |
"CREDITCARDNUMBER": 13,
|
| 87 |
-
"
|
| 88 |
-
"
|
| 89 |
-
"
|
| 90 |
-
"
|
| 91 |
-
"
|
| 92 |
-
"
|
| 93 |
-
"
|
| 94 |
-
"
|
| 95 |
-
"
|
| 96 |
-
"
|
| 97 |
-
"
|
| 98 |
-
"
|
| 99 |
-
"
|
| 100 |
-
"
|
| 101 |
-
"
|
| 102 |
-
"IPV6": 29,
|
| 103 |
-
"JOBAREA": 30,
|
| 104 |
-
"JOBTITLE": 31,
|
| 105 |
-
"JOBTYPE": 32,
|
| 106 |
-
"LASTNAME": 33,
|
| 107 |
-
"LITECOINADDRESS": 34,
|
| 108 |
-
"MAC": 35,
|
| 109 |
-
"MASKEDNUMBER": 36,
|
| 110 |
-
"MIDDLENAME": 37,
|
| 111 |
-
"NEARBYGPSCOORDINATE": 38,
|
| 112 |
"O": 0,
|
| 113 |
-
"ORDINALDIRECTION":
|
| 114 |
-
"
|
| 115 |
-
"
|
| 116 |
-
"
|
| 117 |
-
"
|
| 118 |
-
"
|
| 119 |
-
"
|
| 120 |
-
"
|
| 121 |
-
"
|
| 122 |
-
"
|
| 123 |
-
"
|
| 124 |
-
"
|
| 125 |
-
"
|
| 126 |
-
"
|
| 127 |
-
"
|
| 128 |
-
"
|
| 129 |
-
"
|
| 130 |
-
"
|
|
|
|
|
|
|
| 131 |
},
|
| 132 |
"layer_norm_eps": 1e-07,
|
| 133 |
"legacy": true,
|
|
|
|
| 24 |
"11": "CREDITCARDCVV",
|
| 25 |
"12": "CREDITCARDISSUER",
|
| 26 |
"13": "CREDITCARDNUMBER",
|
| 27 |
+
"14": "DATE",
|
| 28 |
+
"15": "DOB",
|
| 29 |
+
"16": "DRIVERLICENSENUM",
|
| 30 |
+
"17": "EMAIL",
|
| 31 |
+
"18": "ETHEREUMADDRESS",
|
| 32 |
+
"19": "EYECOLOR",
|
| 33 |
+
"20": "GENDER",
|
| 34 |
+
"21": "HEIGHT",
|
| 35 |
+
"22": "IBAN",
|
| 36 |
+
"23": "IPADDRESS",
|
| 37 |
+
"24": "JOBAREA",
|
| 38 |
+
"25": "JOBTYPE",
|
| 39 |
+
"26": "LITECOINADDRESS",
|
| 40 |
+
"27": "MASKEDNUMBER",
|
| 41 |
+
"28": "NEARBYGPSCOORDINATE",
|
| 42 |
+
"29": "ORDINALDIRECTION",
|
| 43 |
+
"30": "PASSPORTNUM",
|
| 44 |
+
"31": "PASSWORD",
|
| 45 |
+
"32": "PERSONNAME",
|
| 46 |
+
"33": "PHONEIMEI",
|
| 47 |
+
"34": "PHONENUMBER",
|
| 48 |
+
"35": "PIN",
|
| 49 |
+
"36": "PREFIX",
|
| 50 |
+
"37": "SECONDARYADDRESS",
|
| 51 |
+
"38": "SSN/CCCD",
|
| 52 |
+
"39": "STATE",
|
| 53 |
+
"40": "STREET",
|
| 54 |
+
"41": "TAXNUM",
|
| 55 |
+
"42": "TIME",
|
| 56 |
+
"43": "URL",
|
| 57 |
+
"44": "USERAGENT",
|
| 58 |
+
"45": "USERNAME",
|
| 59 |
+
"46": "VEHICLEVIN",
|
| 60 |
+
"47": "VEHICLEVRM",
|
| 61 |
+
"48": "ZIPCODE"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 62 |
},
|
| 63 |
"initializer_range": 0.02,
|
| 64 |
"intermediate_size": 3072,
|
|
|
|
| 76 |
"CREDITCARDCVV": 11,
|
| 77 |
"CREDITCARDISSUER": 12,
|
| 78 |
"CREDITCARDNUMBER": 13,
|
| 79 |
+
"DATE": 14,
|
| 80 |
+
"DOB": 15,
|
| 81 |
+
"DRIVERLICENSENUM": 16,
|
| 82 |
+
"EMAIL": 17,
|
| 83 |
+
"ETHEREUMADDRESS": 18,
|
| 84 |
+
"EYECOLOR": 19,
|
| 85 |
+
"GENDER": 20,
|
| 86 |
+
"HEIGHT": 21,
|
| 87 |
+
"IBAN": 22,
|
| 88 |
+
"IPADDRESS": 23,
|
| 89 |
+
"JOBAREA": 24,
|
| 90 |
+
"JOBTYPE": 25,
|
| 91 |
+
"LITECOINADDRESS": 26,
|
| 92 |
+
"MASKEDNUMBER": 27,
|
| 93 |
+
"NEARBYGPSCOORDINATE": 28,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 94 |
"O": 0,
|
| 95 |
+
"ORDINALDIRECTION": 29,
|
| 96 |
+
"PASSPORTNUM": 30,
|
| 97 |
+
"PASSWORD": 31,
|
| 98 |
+
"PERSONNAME": 32,
|
| 99 |
+
"PHONEIMEI": 33,
|
| 100 |
+
"PHONENUMBER": 34,
|
| 101 |
+
"PIN": 35,
|
| 102 |
+
"PREFIX": 36,
|
| 103 |
+
"SECONDARYADDRESS": 37,
|
| 104 |
+
"SSN/CCCD": 38,
|
| 105 |
+
"STATE": 39,
|
| 106 |
+
"STREET": 40,
|
| 107 |
+
"TAXNUM": 41,
|
| 108 |
+
"TIME": 42,
|
| 109 |
+
"URL": 43,
|
| 110 |
+
"USERAGENT": 44,
|
| 111 |
+
"USERNAME": 45,
|
| 112 |
+
"VEHICLEVIN": 46,
|
| 113 |
+
"VEHICLEVRM": 47,
|
| 114 |
+
"ZIPCODE": 48
|
| 115 |
},
|
| 116 |
"layer_norm_eps": 1e-07,
|
| 117 |
"legacy": true,
|
Internal-Run-EN/SpanBased-DeBERTa/pytorch_model.bin
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 sha256:80e44dd45c7c5c2577e71b97e97ee8708117f8091d66901f2518944641572947
|
| 3 |
+
size 741541647
|