Upload folder using huggingface_hub
Browse files- model-00001-of-00013.safetensors +1 -1
- model-00002-of-00013.safetensors +1 -1
- model-00003-of-00013.safetensors +1 -1
- model-00004-of-00013.safetensors +1 -1
- model-00005-of-00013.safetensors +1 -1
- model-00006-of-00013.safetensors +1 -1
- model-00007-of-00013.safetensors +1 -1
- model-00008-of-00013.safetensors +1 -1
- model-00009-of-00013.safetensors +1 -1
- model-00010-of-00013.safetensors +1 -1
- model-00011-of-00013.safetensors +1 -1
- model-00012-of-00013.safetensors +1 -1
- model-00013-of-00013.safetensors +1 -1
- trainer_state.json +226 -226
- training_args.bin +1 -1
model-00001-of-00013.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4997184968
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eb6ec89beb1a3a15c3c323a78468a011a90c4d419a710ded43d64b96710ebcb2
|
| 3 |
size 4997184968
|
model-00002-of-00013.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4997741608
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:97e488192375326f97a9d4de1661b5f2853ced9541b05bc9c6bd6fa171162987
|
| 3 |
size 4997741608
|
model-00003-of-00013.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4997742208
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8ee1d363d5ff4cb23949928657ea12ac91b6a8573bfd7c67209d8226028c3212
|
| 3 |
size 4997742208
|
model-00004-of-00013.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4997743184
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:40ecd2225d04967348e873332b3cb64359bdf5c9788e9a125e97c2afc6f882f3
|
| 3 |
size 4997743184
|
model-00005-of-00013.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4997743184
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f5ac201b1d37785ad89ec4b990f9ca824eb2e152daf0800465436e872436d389
|
| 3 |
size 4997743184
|
model-00006-of-00013.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4997743184
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dfd04f2ab3f0732fafb9223fe3e2edab05b433aa9a176ab5e509e7e48d7f08c3
|
| 3 |
size 4997743184
|
model-00007-of-00013.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4997743184
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5d49269020f5db6a3a2867c0a983dc6aee5f2bcd861fd199c5f756b3a21e14f0
|
| 3 |
size 4997743184
|
model-00008-of-00013.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4997743184
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd04590d86a5115cc651443381bdbb8b13aaa6543361ff4aacea69fb3d6bb106
|
| 3 |
size 4997743184
|
model-00009-of-00013.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4997743184
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8d1b44e9a3b53b1e8d17806d67cd92a1c243f5a9b0a5f4067b06c3e056d7fa92
|
| 3 |
size 4997743184
|
model-00010-of-00013.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4997743184
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eed5e6c59693fb4545ef1a751ae89f1d259d017cc1c0f4c16a1724298c5ebbbe
|
| 3 |
size 4997743184
|
model-00011-of-00013.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4997743184
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f15933c6130e6814eb07617d25211181c67a1da13e6b620ed4b67f6c1c78925f
|
| 3 |
size 4997743184
|
model-00012-of-00013.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4997743184
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:63fca5cc1b8a5644c2a7238112ed7c1396ae81b26286dd4ca7d7fa8d7049dd24
|
| 3 |
size 4997743184
|
model-00013-of-00013.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1094220288
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:486407d221e32bbb0ebeeb7e39c521de9120456af2d0a722eca139320d7e1113
|
| 3 |
size 1094220288
|
trainer_state.json
CHANGED
|
@@ -11,800 +11,800 @@
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.017543859649122806,
|
| 14 |
-
"grad_norm":
|
| 15 |
"learning_rate": 0.0,
|
| 16 |
"loss": 0.8483,
|
| 17 |
"step": 1
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.03508771929824561,
|
| 21 |
-
"grad_norm":
|
| 22 |
"learning_rate": 3.3333333333333333e-06,
|
| 23 |
"loss": 0.8763,
|
| 24 |
"step": 2
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.05263157894736842,
|
| 28 |
-
"grad_norm":
|
| 29 |
"learning_rate": 6.666666666666667e-06,
|
| 30 |
-
"loss": 1.
|
| 31 |
"step": 3
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.07017543859649122,
|
| 35 |
-
"grad_norm": 1.
|
| 36 |
"learning_rate": 1e-05,
|
| 37 |
-
"loss": 1.
|
| 38 |
"step": 4
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.08771929824561403,
|
| 42 |
-
"grad_norm":
|
| 43 |
"learning_rate": 1.3333333333333333e-05,
|
| 44 |
-
"loss": 0.
|
| 45 |
"step": 5
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.10526315789473684,
|
| 49 |
-
"grad_norm": 1.
|
| 50 |
"learning_rate": 1.6666666666666667e-05,
|
| 51 |
-
"loss": 0.
|
| 52 |
"step": 6
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 0.12280701754385964,
|
| 56 |
-
"grad_norm": 2.
|
| 57 |
"learning_rate": 2e-05,
|
| 58 |
-
"loss": 0.
|
| 59 |
"step": 7
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 0.14035087719298245,
|
| 63 |
-
"grad_norm":
|
| 64 |
"learning_rate": 2.3333333333333336e-05,
|
| 65 |
-
"loss": 0.
|
| 66 |
"step": 8
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 0.15789473684210525,
|
| 70 |
-
"grad_norm":
|
| 71 |
"learning_rate": 2.6666666666666667e-05,
|
| 72 |
-
"loss": 0.
|
| 73 |
"step": 9
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 0.17543859649122806,
|
| 77 |
-
"grad_norm": 1.
|
| 78 |
"learning_rate": 3e-05,
|
| 79 |
-
"loss": 0.
|
| 80 |
"step": 10
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.19298245614035087,
|
| 84 |
-
"grad_norm": 1.
|
| 85 |
"learning_rate": 3.3333333333333335e-05,
|
| 86 |
-
"loss":
|
| 87 |
"step": 11
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.21052631578947367,
|
| 91 |
-
"grad_norm": 1.
|
| 92 |
"learning_rate": 3.6666666666666666e-05,
|
| 93 |
-
"loss": 0.
|
| 94 |
"step": 12
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 0.22807017543859648,
|
| 98 |
-
"grad_norm":
|
| 99 |
"learning_rate": 4e-05,
|
| 100 |
-
"loss": 0.
|
| 101 |
"step": 13
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 0.24561403508771928,
|
| 105 |
-
"grad_norm": 1.
|
| 106 |
"learning_rate": 4.3333333333333334e-05,
|
| 107 |
-
"loss": 0.
|
| 108 |
"step": 14
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 0.2631578947368421,
|
| 112 |
-
"grad_norm": 1.
|
| 113 |
"learning_rate": 4.666666666666667e-05,
|
| 114 |
-
"loss": 0.
|
| 115 |
"step": 15
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 0.2807017543859649,
|
| 119 |
-
"grad_norm":
|
| 120 |
"learning_rate": 5e-05,
|
| 121 |
-
"loss": 0.
|
| 122 |
"step": 16
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.2982456140350877,
|
| 126 |
-
"grad_norm": 1.
|
| 127 |
"learning_rate": 4.999830770009406e-05,
|
| 128 |
-
"loss": 1.
|
| 129 |
"step": 17
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.3157894736842105,
|
| 133 |
-
"grad_norm":
|
| 134 |
"learning_rate": 4.9993231029486544e-05,
|
| 135 |
-
"loss": 0.
|
| 136 |
"step": 18
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.3333333333333333,
|
| 140 |
-
"grad_norm":
|
| 141 |
"learning_rate": 4.99847706754774e-05,
|
| 142 |
-
"loss": 0.
|
| 143 |
"step": 19
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 0.3508771929824561,
|
| 147 |
-
"grad_norm": 0.
|
| 148 |
"learning_rate": 4.997292778346312e-05,
|
| 149 |
-
"loss": 0.
|
| 150 |
"step": 20
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 0.3684210526315789,
|
| 154 |
-
"grad_norm": 0.
|
| 155 |
"learning_rate": 4.995770395678171e-05,
|
| 156 |
-
"loss": 0.
|
| 157 |
"step": 21
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 0.38596491228070173,
|
| 161 |
-
"grad_norm": 0.
|
| 162 |
"learning_rate": 4.993910125649561e-05,
|
| 163 |
-
"loss": 0.
|
| 164 |
"step": 22
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.40350877192982454,
|
| 168 |
-
"grad_norm": 0.
|
| 169 |
"learning_rate": 4.9917122201112656e-05,
|
| 170 |
-
"loss": 0.
|
| 171 |
"step": 23
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.42105263157894735,
|
| 175 |
-
"grad_norm": 0.
|
| 176 |
"learning_rate": 4.989176976624511e-05,
|
| 177 |
-
"loss": 0.
|
| 178 |
"step": 24
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 0.43859649122807015,
|
| 182 |
-
"grad_norm": 0.
|
| 183 |
"learning_rate": 4.9863047384206835e-05,
|
| 184 |
-
"loss": 0.
|
| 185 |
"step": 25
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 0.45614035087719296,
|
| 189 |
-
"grad_norm": 0.
|
| 190 |
"learning_rate": 4.983095894354858e-05,
|
| 191 |
-
"loss": 0.
|
| 192 |
"step": 26
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 0.47368421052631576,
|
| 196 |
-
"grad_norm": 0.
|
| 197 |
"learning_rate": 4.979550878853154e-05,
|
| 198 |
-
"loss": 0.
|
| 199 |
"step": 27
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.49122807017543857,
|
| 203 |
-
"grad_norm": 0.
|
| 204 |
"learning_rate": 4.975670171853926e-05,
|
| 205 |
-
"loss": 0.
|
| 206 |
"step": 28
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.5087719298245614,
|
| 210 |
-
"grad_norm": 0.
|
| 211 |
"learning_rate": 4.971454298742779e-05,
|
| 212 |
-
"loss": 0.
|
| 213 |
"step": 29
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.5263157894736842,
|
| 217 |
-
"grad_norm": 0.
|
| 218 |
"learning_rate": 4.966903830281449e-05,
|
| 219 |
-
"loss": 0.
|
| 220 |
"step": 30
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 0.543859649122807,
|
| 224 |
-
"grad_norm": 0.
|
| 225 |
"learning_rate": 4.962019382530521e-05,
|
| 226 |
-
"loss": 0.
|
| 227 |
"step": 31
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 0.5614035087719298,
|
| 231 |
-
"grad_norm": 0.
|
| 232 |
"learning_rate": 4.9568016167660334e-05,
|
| 233 |
-
"loss": 0.
|
| 234 |
"step": 32
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.5789473684210527,
|
| 238 |
-
"grad_norm": 0.
|
| 239 |
"learning_rate": 4.951251239389948e-05,
|
| 240 |
-
"loss": 0.
|
| 241 |
"step": 33
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 0.5964912280701754,
|
| 245 |
-
"grad_norm": 0.
|
| 246 |
"learning_rate": 4.9453690018345144e-05,
|
| 247 |
-
"loss": 0.
|
| 248 |
"step": 34
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 0.6140350877192983,
|
| 252 |
-
"grad_norm": 0.
|
| 253 |
"learning_rate": 4.939155700460536e-05,
|
| 254 |
-
"loss": 0.
|
| 255 |
"step": 35
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.631578947368421,
|
| 259 |
-
"grad_norm": 0.
|
| 260 |
"learning_rate": 4.9326121764495596e-05,
|
| 261 |
-
"loss": 0.
|
| 262 |
"step": 36
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 0.6491228070175439,
|
| 266 |
-
"grad_norm": 0.
|
| 267 |
"learning_rate": 4.925739315689991e-05,
|
| 268 |
-
"loss": 0.
|
| 269 |
"step": 37
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 0.6666666666666666,
|
| 273 |
-
"grad_norm": 0.
|
| 274 |
"learning_rate": 4.9185380486571595e-05,
|
| 275 |
-
"loss": 0.
|
| 276 |
"step": 38
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 0.6842105263157895,
|
| 280 |
-
"grad_norm": 0.
|
| 281 |
"learning_rate": 4.9110093502873476e-05,
|
| 282 |
-
"loss": 0.
|
| 283 |
"step": 39
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 0.7017543859649122,
|
| 287 |
-
"grad_norm": 0.
|
| 288 |
"learning_rate": 4.9031542398457974e-05,
|
| 289 |
-
"loss": 0.
|
| 290 |
"step": 40
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 0.7192982456140351,
|
| 294 |
-
"grad_norm": 0.
|
| 295 |
"learning_rate": 4.894973780788722e-05,
|
| 296 |
-
"loss": 0.
|
| 297 |
"step": 41
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.7368421052631579,
|
| 301 |
-
"grad_norm": 0.
|
| 302 |
"learning_rate": 4.88646908061933e-05,
|
| 303 |
-
"loss": 0.
|
| 304 |
"step": 42
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 0.7543859649122807,
|
| 308 |
-
"grad_norm": 0.
|
| 309 |
"learning_rate": 4.877641290737884e-05,
|
| 310 |
-
"loss": 0.
|
| 311 |
"step": 43
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 0.7719298245614035,
|
| 315 |
-
"grad_norm": 0.
|
| 316 |
"learning_rate": 4.868491606285823e-05,
|
| 317 |
-
"loss": 0.
|
| 318 |
"step": 44
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 0.7894736842105263,
|
| 322 |
-
"grad_norm": 0.
|
| 323 |
"learning_rate": 4.859021265983959e-05,
|
| 324 |
-
"loss": 0.
|
| 325 |
"step": 45
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.8070175438596491,
|
| 329 |
-
"grad_norm": 0.
|
| 330 |
"learning_rate": 4.849231551964771e-05,
|
| 331 |
-
"loss": 0.
|
| 332 |
"step": 46
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 0.8245614035087719,
|
| 336 |
-
"grad_norm": 0.
|
| 337 |
"learning_rate": 4.839123789598829e-05,
|
| 338 |
-
"loss": 0.
|
| 339 |
"step": 47
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.8421052631578947,
|
| 343 |
-
"grad_norm": 0.
|
| 344 |
"learning_rate": 4.828699347315356e-05,
|
| 345 |
-
"loss": 0.
|
| 346 |
"step": 48
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 0.8596491228070176,
|
| 350 |
-
"grad_norm": 0.
|
| 351 |
"learning_rate": 4.817959636416969e-05,
|
| 352 |
-
"loss": 0.
|
| 353 |
"step": 49
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 0.8771929824561403,
|
| 357 |
-
"grad_norm": 0.
|
| 358 |
"learning_rate": 4.806906110888606e-05,
|
| 359 |
-
"loss": 0.
|
| 360 |
"step": 50
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 0.8947368421052632,
|
| 364 |
-
"grad_norm": 0.
|
| 365 |
"learning_rate": 4.7955402672006854e-05,
|
| 366 |
-
"loss": 0.
|
| 367 |
"step": 51
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 0.9122807017543859,
|
| 371 |
-
"grad_norm": 0.
|
| 372 |
"learning_rate": 4.783863644106502e-05,
|
| 373 |
-
"loss": 0.
|
| 374 |
"step": 52
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 0.9298245614035088,
|
| 378 |
-
"grad_norm": 0.
|
| 379 |
"learning_rate": 4.771877822433911e-05,
|
| 380 |
-
"loss": 0.
|
| 381 |
"step": 53
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 0.9473684210526315,
|
| 385 |
-
"grad_norm": 0.
|
| 386 |
"learning_rate": 4.759584424871302e-05,
|
| 387 |
-
"loss": 0.
|
| 388 |
"step": 54
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 0.9649122807017544,
|
| 392 |
-
"grad_norm": 0.
|
| 393 |
"learning_rate": 4.7469851157479177e-05,
|
| 394 |
-
"loss": 0.
|
| 395 |
"step": 55
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 0.9824561403508771,
|
| 399 |
-
"grad_norm": 0.
|
| 400 |
"learning_rate": 4.734081600808531e-05,
|
| 401 |
-
"loss": 0.
|
| 402 |
"step": 56
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 1.0,
|
| 406 |
-
"grad_norm": 0.
|
| 407 |
"learning_rate": 4.7208756269825104e-05,
|
| 408 |
-
"loss": 0.
|
| 409 |
"step": 57
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 1.0175438596491229,
|
| 413 |
-
"grad_norm": 0.
|
| 414 |
"learning_rate": 4.707368982147318e-05,
|
| 415 |
-
"loss": 0.
|
| 416 |
"step": 58
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 1.0350877192982457,
|
| 420 |
-
"grad_norm": 0.
|
| 421 |
"learning_rate": 4.693563494886455e-05,
|
| 422 |
-
"loss": 0.
|
| 423 |
"step": 59
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 1.0526315789473684,
|
| 427 |
-
"grad_norm": 0.
|
| 428 |
"learning_rate": 4.679461034241906e-05,
|
| 429 |
-
"loss": 0.
|
| 430 |
"step": 60
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 1.0701754385964912,
|
| 434 |
-
"grad_norm":
|
| 435 |
"learning_rate": 4.665063509461097e-05,
|
| 436 |
-
"loss": 0.
|
| 437 |
"step": 61
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 1.087719298245614,
|
| 441 |
-
"grad_norm": 0.
|
| 442 |
"learning_rate": 4.650372869738414e-05,
|
| 443 |
-
"loss": 0.
|
| 444 |
"step": 62
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 1.1052631578947367,
|
| 448 |
-
"grad_norm": 1.
|
| 449 |
"learning_rate": 4.6353911039513145e-05,
|
| 450 |
-
"loss": 0.
|
| 451 |
"step": 63
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 1.1228070175438596,
|
| 455 |
-
"grad_norm": 0.
|
| 456 |
"learning_rate": 4.620120240391065e-05,
|
| 457 |
-
"loss": 0.
|
| 458 |
"step": 64
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 1.1403508771929824,
|
| 462 |
-
"grad_norm": 0.
|
| 463 |
"learning_rate": 4.604562346488144e-05,
|
| 464 |
-
"loss": 0.
|
| 465 |
"step": 65
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 1.1578947368421053,
|
| 469 |
-
"grad_norm": 0.
|
| 470 |
"learning_rate": 4.588719528532342e-05,
|
| 471 |
-
"loss": 0.
|
| 472 |
"step": 66
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 1.1754385964912282,
|
| 476 |
-
"grad_norm": 0.
|
| 477 |
"learning_rate": 4.572593931387604e-05,
|
| 478 |
-
"loss": 0.
|
| 479 |
"step": 67
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 1.1929824561403508,
|
| 483 |
-
"grad_norm": 0.
|
| 484 |
"learning_rate": 4.556187738201656e-05,
|
| 485 |
-
"loss": 0.
|
| 486 |
"step": 68
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 1.2105263157894737,
|
| 490 |
-
"grad_norm": 0.
|
| 491 |
"learning_rate": 4.539503170110431e-05,
|
| 492 |
-
"loss": 0.
|
| 493 |
"step": 69
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 1.2280701754385965,
|
| 497 |
-
"grad_norm": 0.
|
| 498 |
"learning_rate": 4.522542485937369e-05,
|
| 499 |
-
"loss": 0.
|
| 500 |
"step": 70
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 1.2456140350877192,
|
| 504 |
-
"grad_norm": 0.
|
| 505 |
"learning_rate": 4.50530798188761e-05,
|
| 506 |
-
"loss": 0.
|
| 507 |
"step": 71
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 1.263157894736842,
|
| 511 |
-
"grad_norm": 1.
|
| 512 |
"learning_rate": 4.48780199123712e-05,
|
| 513 |
-
"loss": 0.
|
| 514 |
"step": 72
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 1.280701754385965,
|
| 518 |
-
"grad_norm": 0.
|
| 519 |
"learning_rate": 4.4700268840168045e-05,
|
| 520 |
-
"loss": 0.
|
| 521 |
"step": 73
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 1.2982456140350878,
|
| 525 |
-
"grad_norm": 0.
|
| 526 |
"learning_rate": 4.4519850666916484e-05,
|
| 527 |
-
"loss": 0.
|
| 528 |
"step": 74
|
| 529 |
},
|
| 530 |
{
|
| 531 |
"epoch": 1.3157894736842106,
|
| 532 |
-
"grad_norm": 0.
|
| 533 |
"learning_rate": 4.43367898183491e-05,
|
| 534 |
-
"loss": 0.
|
| 535 |
"step": 75
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"epoch": 1.3333333333333333,
|
| 539 |
-
"grad_norm": 0.
|
| 540 |
"learning_rate": 4.415111107797445e-05,
|
| 541 |
-
"loss": 0.
|
| 542 |
"step": 76
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 1.3508771929824561,
|
| 546 |
-
"grad_norm": 0.
|
| 547 |
"learning_rate": 4.396283958372173e-05,
|
| 548 |
-
"loss": 0.
|
| 549 |
"step": 77
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 1.368421052631579,
|
| 553 |
-
"grad_norm": 0.
|
| 554 |
"learning_rate": 4.377200082453749e-05,
|
| 555 |
-
"loss": 0.
|
| 556 |
"step": 78
|
| 557 |
},
|
| 558 |
{
|
| 559 |
"epoch": 1.3859649122807016,
|
| 560 |
-
"grad_norm": 0.
|
| 561 |
"learning_rate": 4.357862063693486e-05,
|
| 562 |
-
"loss": 0.
|
| 563 |
"step": 79
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 1.4035087719298245,
|
| 567 |
-
"grad_norm": 0.
|
| 568 |
"learning_rate": 4.3382725201495723e-05,
|
| 569 |
-
"loss": 0.
|
| 570 |
"step": 80
|
| 571 |
},
|
| 572 |
{
|
| 573 |
"epoch": 1.4210526315789473,
|
| 574 |
-
"grad_norm": 0.
|
| 575 |
"learning_rate": 4.318434103932622e-05,
|
| 576 |
-
"loss": 0.
|
| 577 |
"step": 81
|
| 578 |
},
|
| 579 |
{
|
| 580 |
"epoch": 1.4385964912280702,
|
| 581 |
-
"grad_norm": 0.
|
| 582 |
"learning_rate": 4.2983495008466276e-05,
|
| 583 |
-
"loss": 0.
|
| 584 |
"step": 82
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 1.456140350877193,
|
| 588 |
-
"grad_norm": 0.
|
| 589 |
"learning_rate": 4.278021430025343e-05,
|
| 590 |
-
"loss": 0.
|
| 591 |
"step": 83
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 1.4736842105263157,
|
| 595 |
-
"grad_norm": 0.
|
| 596 |
"learning_rate": 4.257452643564155e-05,
|
| 597 |
-
"loss": 0.
|
| 598 |
"step": 84
|
| 599 |
},
|
| 600 |
{
|
| 601 |
"epoch": 1.4912280701754386,
|
| 602 |
-
"grad_norm": 0.
|
| 603 |
"learning_rate": 4.2366459261474933e-05,
|
| 604 |
-
"loss": 0.
|
| 605 |
"step": 85
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"epoch": 1.5087719298245614,
|
| 609 |
-
"grad_norm": 0.
|
| 610 |
"learning_rate": 4.215604094671835e-05,
|
| 611 |
-
"loss": 0.
|
| 612 |
"step": 86
|
| 613 |
},
|
| 614 |
{
|
| 615 |
"epoch": 1.526315789473684,
|
| 616 |
-
"grad_norm": 0.
|
| 617 |
"learning_rate": 4.194329997864331e-05,
|
| 618 |
-
"loss": 0.
|
| 619 |
"step": 87
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"epoch": 1.543859649122807,
|
| 623 |
-
"grad_norm": 0.
|
| 624 |
"learning_rate": 4.172826515897146e-05,
|
| 625 |
-
"loss": 0.
|
| 626 |
"step": 88
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 1.5614035087719298,
|
| 630 |
-
"grad_norm": 0.
|
| 631 |
"learning_rate": 4.1510965599975196e-05,
|
| 632 |
-
"loss": 0.
|
| 633 |
"step": 89
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 1.5789473684210527,
|
| 637 |
-
"grad_norm": 0.
|
| 638 |
"learning_rate": 4.129143072053638e-05,
|
| 639 |
-
"loss": 0.
|
| 640 |
"step": 90
|
| 641 |
},
|
| 642 |
{
|
| 643 |
"epoch": 1.5964912280701755,
|
| 644 |
-
"grad_norm": 0.
|
| 645 |
"learning_rate": 4.1069690242163484e-05,
|
| 646 |
-
"loss": 0.
|
| 647 |
"step": 91
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"epoch": 1.6140350877192984,
|
| 651 |
-
"grad_norm": 0.
|
| 652 |
"learning_rate": 4.0845774184967754e-05,
|
| 653 |
-
"loss": 0.
|
| 654 |
"step": 92
|
| 655 |
},
|
| 656 |
{
|
| 657 |
"epoch": 1.631578947368421,
|
| 658 |
-
"grad_norm": 0.
|
| 659 |
"learning_rate": 4.0619712863599e-05,
|
| 660 |
-
"loss": 0.
|
| 661 |
"step": 93
|
| 662 |
},
|
| 663 |
{
|
| 664 |
"epoch": 1.6491228070175439,
|
| 665 |
-
"grad_norm": 0.
|
| 666 |
"learning_rate": 4.039153688314145e-05,
|
| 667 |
-
"loss": 0.
|
| 668 |
"step": 94
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 1.6666666666666665,
|
| 672 |
-
"grad_norm": 0.
|
| 673 |
"learning_rate": 4.0161277134970345e-05,
|
| 674 |
-
"loss": 0.
|
| 675 |
"step": 95
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 1.6842105263157894,
|
| 679 |
-
"grad_norm": 0.
|
| 680 |
"learning_rate": 3.9928964792569655e-05,
|
| 681 |
-
"loss": 0.
|
| 682 |
"step": 96
|
| 683 |
},
|
| 684 |
{
|
| 685 |
"epoch": 1.7017543859649122,
|
| 686 |
-
"grad_norm": 0.
|
| 687 |
"learning_rate": 3.969463130731183e-05,
|
| 688 |
-
"loss": 0.
|
| 689 |
"step": 97
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"epoch": 1.719298245614035,
|
| 693 |
-
"grad_norm": 0.
|
| 694 |
"learning_rate": 3.945830840419966e-05,
|
| 695 |
-
"loss": 0.
|
| 696 |
"step": 98
|
| 697 |
},
|
| 698 |
{
|
| 699 |
"epoch": 1.736842105263158,
|
| 700 |
-
"grad_norm": 0.
|
| 701 |
"learning_rate": 3.9220028077571295e-05,
|
| 702 |
-
"loss": 0.
|
| 703 |
"step": 99
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"epoch": 1.7543859649122808,
|
| 707 |
-
"grad_norm": 0.
|
| 708 |
"learning_rate": 3.897982258676867e-05,
|
| 709 |
-
"loss": 0.
|
| 710 |
"step": 100
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 1.7719298245614035,
|
| 714 |
-
"grad_norm": 0.
|
| 715 |
"learning_rate": 3.873772445177015e-05,
|
| 716 |
-
"loss": 0.
|
| 717 |
"step": 101
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 1.7894736842105263,
|
| 721 |
-
"grad_norm": 0.
|
| 722 |
"learning_rate": 3.8493766448787825e-05,
|
| 723 |
-
"loss": 0.
|
| 724 |
"step": 102
|
| 725 |
},
|
| 726 |
{
|
| 727 |
"epoch": 1.807017543859649,
|
| 728 |
-
"grad_norm": 0.
|
| 729 |
"learning_rate": 3.824798160583012e-05,
|
| 730 |
-
"loss": 0.
|
| 731 |
"step": 103
|
| 732 |
},
|
| 733 |
{
|
| 734 |
"epoch": 1.8245614035087718,
|
| 735 |
-
"grad_norm": 0.
|
| 736 |
"learning_rate": 3.8000403198230387e-05,
|
| 737 |
-
"loss": 0.
|
| 738 |
"step": 104
|
| 739 |
},
|
| 740 |
{
|
| 741 |
"epoch": 1.8421052631578947,
|
| 742 |
-
"grad_norm": 0.
|
| 743 |
"learning_rate": 3.775106474414188e-05,
|
| 744 |
-
"loss": 0.
|
| 745 |
"step": 105
|
| 746 |
},
|
| 747 |
{
|
| 748 |
"epoch": 1.8596491228070176,
|
| 749 |
-
"grad_norm": 0.
|
| 750 |
"learning_rate": 3.7500000000000003e-05,
|
| 751 |
-
"loss": 0.
|
| 752 |
"step": 106
|
| 753 |
},
|
| 754 |
{
|
| 755 |
"epoch": 1.8771929824561404,
|
| 756 |
-
"grad_norm": 0.
|
| 757 |
"learning_rate": 3.7247242955952175e-05,
|
| 758 |
-
"loss": 0.
|
| 759 |
"step": 107
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 1.8947368421052633,
|
| 763 |
-
"grad_norm": 0.
|
| 764 |
"learning_rate": 3.699282783125616e-05,
|
| 765 |
-
"loss": 0.
|
| 766 |
"step": 108
|
| 767 |
},
|
| 768 |
{
|
| 769 |
"epoch": 1.912280701754386,
|
| 770 |
-
"grad_norm": 0.
|
| 771 |
"learning_rate": 3.673678906964727e-05,
|
| 772 |
-
"loss": 0.
|
| 773 |
"step": 109
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"epoch": 1.9298245614035088,
|
| 777 |
-
"grad_norm": 0.
|
| 778 |
"learning_rate": 3.6479161334675296e-05,
|
| 779 |
-
"loss": 0.
|
| 780 |
"step": 110
|
| 781 |
},
|
| 782 |
{
|
| 783 |
"epoch": 1.9473684210526314,
|
| 784 |
-
"grad_norm": 0.
|
| 785 |
"learning_rate": 3.621997950501156e-05,
|
| 786 |
-
"loss": 0.
|
| 787 |
"step": 111
|
| 788 |
},
|
| 789 |
{
|
| 790 |
"epoch": 1.9649122807017543,
|
| 791 |
-
"grad_norm": 0.
|
| 792 |
"learning_rate": 3.5959278669726935e-05,
|
| 793 |
-
"loss": 0.
|
| 794 |
"step": 112
|
| 795 |
},
|
| 796 |
{
|
| 797 |
"epoch": 1.9824561403508771,
|
| 798 |
-
"grad_norm": 0.
|
| 799 |
"learning_rate": 3.569709412354136e-05,
|
| 800 |
-
"loss": 0.
|
| 801 |
"step": 113
|
| 802 |
},
|
| 803 |
{
|
| 804 |
"epoch": 2.0,
|
| 805 |
-
"grad_norm": 0.
|
| 806 |
"learning_rate": 3.543346136204545e-05,
|
| 807 |
-
"loss": 0.
|
| 808 |
"step": 114
|
| 809 |
}
|
| 810 |
],
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.017543859649122806,
|
| 14 |
+
"grad_norm": 2.410881066137832,
|
| 15 |
"learning_rate": 0.0,
|
| 16 |
"loss": 0.8483,
|
| 17 |
"step": 1
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.03508771929824561,
|
| 21 |
+
"grad_norm": 2.665228254124179,
|
| 22 |
"learning_rate": 3.3333333333333333e-06,
|
| 23 |
"loss": 0.8763,
|
| 24 |
"step": 2
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.05263157894736842,
|
| 28 |
+
"grad_norm": 2.3728587057049992,
|
| 29 |
"learning_rate": 6.666666666666667e-06,
|
| 30 |
+
"loss": 1.019,
|
| 31 |
"step": 3
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.07017543859649122,
|
| 35 |
+
"grad_norm": 1.9987528382054482,
|
| 36 |
"learning_rate": 1e-05,
|
| 37 |
+
"loss": 1.0451,
|
| 38 |
"step": 4
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.08771929824561403,
|
| 42 |
+
"grad_norm": 1.2387448314677751,
|
| 43 |
"learning_rate": 1.3333333333333333e-05,
|
| 44 |
+
"loss": 0.8298,
|
| 45 |
"step": 5
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.10526315789473684,
|
| 49 |
+
"grad_norm": 1.4006709843865188,
|
| 50 |
"learning_rate": 1.6666666666666667e-05,
|
| 51 |
+
"loss": 0.9668,
|
| 52 |
"step": 6
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 0.12280701754385964,
|
| 56 |
+
"grad_norm": 2.1941304727738995,
|
| 57 |
"learning_rate": 2e-05,
|
| 58 |
+
"loss": 0.9639,
|
| 59 |
"step": 7
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 0.14035087719298245,
|
| 63 |
+
"grad_norm": 2.0965952504080145,
|
| 64 |
"learning_rate": 2.3333333333333336e-05,
|
| 65 |
+
"loss": 0.906,
|
| 66 |
"step": 8
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 0.15789473684210525,
|
| 70 |
+
"grad_norm": 2.0779003780741325,
|
| 71 |
"learning_rate": 2.6666666666666667e-05,
|
| 72 |
+
"loss": 0.9331,
|
| 73 |
"step": 9
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 0.17543859649122806,
|
| 77 |
+
"grad_norm": 1.914852860029578,
|
| 78 |
"learning_rate": 3e-05,
|
| 79 |
+
"loss": 0.9615,
|
| 80 |
"step": 10
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.19298245614035087,
|
| 84 |
+
"grad_norm": 1.5453059456697873,
|
| 85 |
"learning_rate": 3.3333333333333335e-05,
|
| 86 |
+
"loss": 0.9976,
|
| 87 |
"step": 11
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.21052631578947367,
|
| 91 |
+
"grad_norm": 1.2149278389301041,
|
| 92 |
"learning_rate": 3.6666666666666666e-05,
|
| 93 |
+
"loss": 0.8731,
|
| 94 |
"step": 12
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 0.22807017543859648,
|
| 98 |
+
"grad_norm": 9.574453131193065,
|
| 99 |
"learning_rate": 4e-05,
|
| 100 |
+
"loss": 0.9048,
|
| 101 |
"step": 13
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 0.24561403508771928,
|
| 105 |
+
"grad_norm": 1.7542365583306754,
|
| 106 |
"learning_rate": 4.3333333333333334e-05,
|
| 107 |
+
"loss": 0.9045,
|
| 108 |
"step": 14
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 0.2631578947368421,
|
| 112 |
+
"grad_norm": 1.5687648722942888,
|
| 113 |
"learning_rate": 4.666666666666667e-05,
|
| 114 |
+
"loss": 0.9642,
|
| 115 |
"step": 15
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 0.2807017543859649,
|
| 119 |
+
"grad_norm": 1.1629052446813282,
|
| 120 |
"learning_rate": 5e-05,
|
| 121 |
+
"loss": 0.9157,
|
| 122 |
"step": 16
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.2982456140350877,
|
| 126 |
+
"grad_norm": 1.0662859779498828,
|
| 127 |
"learning_rate": 4.999830770009406e-05,
|
| 128 |
+
"loss": 1.0076,
|
| 129 |
"step": 17
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.3157894736842105,
|
| 133 |
+
"grad_norm": 1.1540729762939772,
|
| 134 |
"learning_rate": 4.9993231029486544e-05,
|
| 135 |
+
"loss": 0.8898,
|
| 136 |
"step": 18
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.3333333333333333,
|
| 140 |
+
"grad_norm": 2.284686909029557,
|
| 141 |
"learning_rate": 4.99847706754774e-05,
|
| 142 |
+
"loss": 0.9111,
|
| 143 |
"step": 19
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 0.3508771929824561,
|
| 147 |
+
"grad_norm": 0.9299877447474001,
|
| 148 |
"learning_rate": 4.997292778346312e-05,
|
| 149 |
+
"loss": 0.9567,
|
| 150 |
"step": 20
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 0.3684210526315789,
|
| 154 |
+
"grad_norm": 0.7891834706045853,
|
| 155 |
"learning_rate": 4.995770395678171e-05,
|
| 156 |
+
"loss": 0.8907,
|
| 157 |
"step": 21
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 0.38596491228070173,
|
| 161 |
+
"grad_norm": 0.7738739475483105,
|
| 162 |
"learning_rate": 4.993910125649561e-05,
|
| 163 |
+
"loss": 0.7582,
|
| 164 |
"step": 22
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.40350877192982454,
|
| 168 |
+
"grad_norm": 0.7564543923604894,
|
| 169 |
"learning_rate": 4.9917122201112656e-05,
|
| 170 |
+
"loss": 0.893,
|
| 171 |
"step": 23
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.42105263157894735,
|
| 175 |
+
"grad_norm": 0.7725302637179225,
|
| 176 |
"learning_rate": 4.989176976624511e-05,
|
| 177 |
+
"loss": 0.794,
|
| 178 |
"step": 24
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 0.43859649122807015,
|
| 182 |
+
"grad_norm": 0.7154589746662706,
|
| 183 |
"learning_rate": 4.9863047384206835e-05,
|
| 184 |
+
"loss": 0.927,
|
| 185 |
"step": 25
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 0.45614035087719296,
|
| 189 |
+
"grad_norm": 0.748200539385192,
|
| 190 |
"learning_rate": 4.983095894354858e-05,
|
| 191 |
+
"loss": 0.8946,
|
| 192 |
"step": 26
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 0.47368421052631576,
|
| 196 |
+
"grad_norm": 0.6877177188589038,
|
| 197 |
"learning_rate": 4.979550878853154e-05,
|
| 198 |
+
"loss": 0.8108,
|
| 199 |
"step": 27
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.49122807017543857,
|
| 203 |
+
"grad_norm": 0.6905557124467382,
|
| 204 |
"learning_rate": 4.975670171853926e-05,
|
| 205 |
+
"loss": 0.8808,
|
| 206 |
"step": 28
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.5087719298245614,
|
| 210 |
+
"grad_norm": 0.6261501879052016,
|
| 211 |
"learning_rate": 4.971454298742779e-05,
|
| 212 |
+
"loss": 0.902,
|
| 213 |
"step": 29
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.5263157894736842,
|
| 217 |
+
"grad_norm": 0.6935833833701227,
|
| 218 |
"learning_rate": 4.966903830281449e-05,
|
| 219 |
+
"loss": 0.7902,
|
| 220 |
"step": 30
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 0.543859649122807,
|
| 224 |
+
"grad_norm": 0.7380295796572786,
|
| 225 |
"learning_rate": 4.962019382530521e-05,
|
| 226 |
+
"loss": 0.8427,
|
| 227 |
"step": 31
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 0.5614035087719298,
|
| 231 |
+
"grad_norm": 0.6686373976763716,
|
| 232 |
"learning_rate": 4.9568016167660334e-05,
|
| 233 |
+
"loss": 0.7824,
|
| 234 |
"step": 32
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.5789473684210527,
|
| 238 |
+
"grad_norm": 0.7284830674570405,
|
| 239 |
"learning_rate": 4.951251239389948e-05,
|
| 240 |
+
"loss": 0.8758,
|
| 241 |
"step": 33
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 0.5964912280701754,
|
| 245 |
+
"grad_norm": 0.7667345279664498,
|
| 246 |
"learning_rate": 4.9453690018345144e-05,
|
| 247 |
+
"loss": 0.9769,
|
| 248 |
"step": 34
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 0.6140350877192983,
|
| 252 |
+
"grad_norm": 0.7186885267427574,
|
| 253 |
"learning_rate": 4.939155700460536e-05,
|
| 254 |
+
"loss": 0.8222,
|
| 255 |
"step": 35
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.631578947368421,
|
| 259 |
+
"grad_norm": 0.7129526253734785,
|
| 260 |
"learning_rate": 4.9326121764495596e-05,
|
| 261 |
+
"loss": 0.8398,
|
| 262 |
"step": 36
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 0.6491228070175439,
|
| 266 |
+
"grad_norm": 0.6763307332151739,
|
| 267 |
"learning_rate": 4.925739315689991e-05,
|
| 268 |
+
"loss": 0.881,
|
| 269 |
"step": 37
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 0.6666666666666666,
|
| 273 |
+
"grad_norm": 0.7596612492978558,
|
| 274 |
"learning_rate": 4.9185380486571595e-05,
|
| 275 |
+
"loss": 0.8883,
|
| 276 |
"step": 38
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 0.6842105263157895,
|
| 280 |
+
"grad_norm": 0.574360173449421,
|
| 281 |
"learning_rate": 4.9110093502873476e-05,
|
| 282 |
+
"loss": 0.9078,
|
| 283 |
"step": 39
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 0.7017543859649122,
|
| 287 |
+
"grad_norm": 0.5045882171460956,
|
| 288 |
"learning_rate": 4.9031542398457974e-05,
|
| 289 |
+
"loss": 0.8722,
|
| 290 |
"step": 40
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 0.7192982456140351,
|
| 294 |
+
"grad_norm": 0.6267476132219085,
|
| 295 |
"learning_rate": 4.894973780788722e-05,
|
| 296 |
+
"loss": 0.883,
|
| 297 |
"step": 41
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.7368421052631579,
|
| 301 |
+
"grad_norm": 0.6021189923969766,
|
| 302 |
"learning_rate": 4.88646908061933e-05,
|
| 303 |
+
"loss": 0.8144,
|
| 304 |
"step": 42
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 0.7543859649122807,
|
| 308 |
+
"grad_norm": 0.6014066609170303,
|
| 309 |
"learning_rate": 4.877641290737884e-05,
|
| 310 |
+
"loss": 0.8496,
|
| 311 |
"step": 43
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 0.7719298245614035,
|
| 315 |
+
"grad_norm": 0.6014121657776419,
|
| 316 |
"learning_rate": 4.868491606285823e-05,
|
| 317 |
+
"loss": 0.768,
|
| 318 |
"step": 44
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 0.7894736842105263,
|
| 322 |
+
"grad_norm": 0.5324268865416073,
|
| 323 |
"learning_rate": 4.859021265983959e-05,
|
| 324 |
+
"loss": 0.8825,
|
| 325 |
"step": 45
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.8070175438596491,
|
| 329 |
+
"grad_norm": 0.6348338184686191,
|
| 330 |
"learning_rate": 4.849231551964771e-05,
|
| 331 |
+
"loss": 0.8587,
|
| 332 |
"step": 46
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 0.8245614035087719,
|
| 336 |
+
"grad_norm": 0.6361465840640576,
|
| 337 |
"learning_rate": 4.839123789598829e-05,
|
| 338 |
+
"loss": 0.868,
|
| 339 |
"step": 47
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.8421052631578947,
|
| 343 |
+
"grad_norm": 0.5806423652725062,
|
| 344 |
"learning_rate": 4.828699347315356e-05,
|
| 345 |
+
"loss": 0.8445,
|
| 346 |
"step": 48
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 0.8596491228070176,
|
| 350 |
+
"grad_norm": 0.6650303211642944,
|
| 351 |
"learning_rate": 4.817959636416969e-05,
|
| 352 |
+
"loss": 0.9439,
|
| 353 |
"step": 49
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 0.8771929824561403,
|
| 357 |
+
"grad_norm": 0.49618932027271617,
|
| 358 |
"learning_rate": 4.806906110888606e-05,
|
| 359 |
+
"loss": 0.676,
|
| 360 |
"step": 50
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 0.8947368421052632,
|
| 364 |
+
"grad_norm": 0.7201128419875511,
|
| 365 |
"learning_rate": 4.7955402672006854e-05,
|
| 366 |
+
"loss": 0.9129,
|
| 367 |
"step": 51
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 0.9122807017543859,
|
| 371 |
+
"grad_norm": 0.7979009332190922,
|
| 372 |
"learning_rate": 4.783863644106502e-05,
|
| 373 |
+
"loss": 0.8844,
|
| 374 |
"step": 52
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 0.9298245614035088,
|
| 378 |
+
"grad_norm": 0.5780879587849941,
|
| 379 |
"learning_rate": 4.771877822433911e-05,
|
| 380 |
+
"loss": 0.835,
|
| 381 |
"step": 53
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 0.9473684210526315,
|
| 385 |
+
"grad_norm": 0.5125413888902927,
|
| 386 |
"learning_rate": 4.759584424871302e-05,
|
| 387 |
+
"loss": 0.7854,
|
| 388 |
"step": 54
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 0.9649122807017544,
|
| 392 |
+
"grad_norm": 0.6493310136858548,
|
| 393 |
"learning_rate": 4.7469851157479177e-05,
|
| 394 |
+
"loss": 0.8246,
|
| 395 |
"step": 55
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 0.9824561403508771,
|
| 399 |
+
"grad_norm": 0.5420670379906682,
|
| 400 |
"learning_rate": 4.734081600808531e-05,
|
| 401 |
+
"loss": 0.8357,
|
| 402 |
"step": 56
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 1.0,
|
| 406 |
+
"grad_norm": 0.8014271486806656,
|
| 407 |
"learning_rate": 4.7208756269825104e-05,
|
| 408 |
+
"loss": 0.7012,
|
| 409 |
"step": 57
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 1.0175438596491229,
|
| 413 |
+
"grad_norm": 0.9216029966396811,
|
| 414 |
"learning_rate": 4.707368982147318e-05,
|
| 415 |
+
"loss": 0.7618,
|
| 416 |
"step": 58
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 1.0350877192982457,
|
| 420 |
+
"grad_norm": 0.7324508540649931,
|
| 421 |
"learning_rate": 4.693563494886455e-05,
|
| 422 |
+
"loss": 0.6381,
|
| 423 |
"step": 59
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 1.0526315789473684,
|
| 427 |
+
"grad_norm": 0.9404509582476525,
|
| 428 |
"learning_rate": 4.679461034241906e-05,
|
| 429 |
+
"loss": 0.7367,
|
| 430 |
"step": 60
|
| 431 |
},
|
| 432 |
{
|
| 433 |
"epoch": 1.0701754385964912,
|
| 434 |
+
"grad_norm": 1.1079545865955729,
|
| 435 |
"learning_rate": 4.665063509461097e-05,
|
| 436 |
+
"loss": 0.6252,
|
| 437 |
"step": 61
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 1.087719298245614,
|
| 441 |
+
"grad_norm": 0.8253491984345485,
|
| 442 |
"learning_rate": 4.650372869738414e-05,
|
| 443 |
+
"loss": 0.6622,
|
| 444 |
"step": 62
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 1.1052631578947367,
|
| 448 |
+
"grad_norm": 1.237778560255967,
|
| 449 |
"learning_rate": 4.6353911039513145e-05,
|
| 450 |
+
"loss": 0.7474,
|
| 451 |
"step": 63
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"epoch": 1.1228070175438596,
|
| 455 |
+
"grad_norm": 0.8429832395594149,
|
| 456 |
"learning_rate": 4.620120240391065e-05,
|
| 457 |
+
"loss": 0.6785,
|
| 458 |
"step": 64
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 1.1403508771929824,
|
| 462 |
+
"grad_norm": 0.7067163708835564,
|
| 463 |
"learning_rate": 4.604562346488144e-05,
|
| 464 |
+
"loss": 0.633,
|
| 465 |
"step": 65
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 1.1578947368421053,
|
| 469 |
+
"grad_norm": 0.7599104494469864,
|
| 470 |
"learning_rate": 4.588719528532342e-05,
|
| 471 |
+
"loss": 0.6462,
|
| 472 |
"step": 66
|
| 473 |
},
|
| 474 |
{
|
| 475 |
"epoch": 1.1754385964912282,
|
| 476 |
+
"grad_norm": 0.74600695035756,
|
| 477 |
"learning_rate": 4.572593931387604e-05,
|
| 478 |
+
"loss": 0.6408,
|
| 479 |
"step": 67
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 1.1929824561403508,
|
| 483 |
+
"grad_norm": 0.8901782092713822,
|
| 484 |
"learning_rate": 4.556187738201656e-05,
|
| 485 |
+
"loss": 0.68,
|
| 486 |
"step": 68
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 1.2105263157894737,
|
| 490 |
+
"grad_norm": 0.7884946794083679,
|
| 491 |
"learning_rate": 4.539503170110431e-05,
|
| 492 |
+
"loss": 0.7179,
|
| 493 |
"step": 69
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"epoch": 1.2280701754385965,
|
| 497 |
+
"grad_norm": 0.6924211614290692,
|
| 498 |
"learning_rate": 4.522542485937369e-05,
|
| 499 |
+
"loss": 0.6849,
|
| 500 |
"step": 70
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 1.2456140350877192,
|
| 504 |
+
"grad_norm": 0.6629364599996329,
|
| 505 |
"learning_rate": 4.50530798188761e-05,
|
| 506 |
+
"loss": 0.7385,
|
| 507 |
"step": 71
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 1.263157894736842,
|
| 511 |
+
"grad_norm": 1.155774350961545,
|
| 512 |
"learning_rate": 4.48780199123712e-05,
|
| 513 |
+
"loss": 0.6942,
|
| 514 |
"step": 72
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 1.280701754385965,
|
| 518 |
+
"grad_norm": 0.8390095837968929,
|
| 519 |
"learning_rate": 4.4700268840168045e-05,
|
| 520 |
+
"loss": 0.6789,
|
| 521 |
"step": 73
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 1.2982456140350878,
|
| 525 |
+
"grad_norm": 0.5703687799634394,
|
| 526 |
"learning_rate": 4.4519850666916484e-05,
|
| 527 |
+
"loss": 0.6164,
|
| 528 |
"step": 74
|
| 529 |
},
|
| 530 |
{
|
| 531 |
"epoch": 1.3157894736842106,
|
| 532 |
+
"grad_norm": 0.6583590025049137,
|
| 533 |
"learning_rate": 4.43367898183491e-05,
|
| 534 |
+
"loss": 0.7449,
|
| 535 |
"step": 75
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"epoch": 1.3333333333333333,
|
| 539 |
+
"grad_norm": 0.5701968757608501,
|
| 540 |
"learning_rate": 4.415111107797445e-05,
|
| 541 |
+
"loss": 0.6234,
|
| 542 |
"step": 76
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 1.3508771929824561,
|
| 546 |
+
"grad_norm": 0.5635406203198912,
|
| 547 |
"learning_rate": 4.396283958372173e-05,
|
| 548 |
+
"loss": 0.6082,
|
| 549 |
"step": 77
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 1.368421052631579,
|
| 553 |
+
"grad_norm": 0.7103841851203655,
|
| 554 |
"learning_rate": 4.377200082453749e-05,
|
| 555 |
+
"loss": 0.716,
|
| 556 |
"step": 78
|
| 557 |
},
|
| 558 |
{
|
| 559 |
"epoch": 1.3859649122807016,
|
| 560 |
+
"grad_norm": 0.7190614642338603,
|
| 561 |
"learning_rate": 4.357862063693486e-05,
|
| 562 |
+
"loss": 0.7011,
|
| 563 |
"step": 79
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 1.4035087719298245,
|
| 567 |
+
"grad_norm": 0.6721203312978761,
|
| 568 |
"learning_rate": 4.3382725201495723e-05,
|
| 569 |
+
"loss": 0.683,
|
| 570 |
"step": 80
|
| 571 |
},
|
| 572 |
{
|
| 573 |
"epoch": 1.4210526315789473,
|
| 574 |
+
"grad_norm": 0.8333526214828886,
|
| 575 |
"learning_rate": 4.318434103932622e-05,
|
| 576 |
+
"loss": 0.679,
|
| 577 |
"step": 81
|
| 578 |
},
|
| 579 |
{
|
| 580 |
"epoch": 1.4385964912280702,
|
| 581 |
+
"grad_norm": 0.6238744952578675,
|
| 582 |
"learning_rate": 4.2983495008466276e-05,
|
| 583 |
+
"loss": 0.663,
|
| 584 |
"step": 82
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 1.456140350877193,
|
| 588 |
+
"grad_norm": 0.6288416586111145,
|
| 589 |
"learning_rate": 4.278021430025343e-05,
|
| 590 |
+
"loss": 0.5932,
|
| 591 |
"step": 83
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 1.4736842105263157,
|
| 595 |
+
"grad_norm": 0.5006163307702345,
|
| 596 |
"learning_rate": 4.257452643564155e-05,
|
| 597 |
+
"loss": 0.5148,
|
| 598 |
"step": 84
|
| 599 |
},
|
| 600 |
{
|
| 601 |
"epoch": 1.4912280701754386,
|
| 602 |
+
"grad_norm": 0.6562209743705945,
|
| 603 |
"learning_rate": 4.2366459261474933e-05,
|
| 604 |
+
"loss": 0.6796,
|
| 605 |
"step": 85
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"epoch": 1.5087719298245614,
|
| 609 |
+
"grad_norm": 0.7234775007820414,
|
| 610 |
"learning_rate": 4.215604094671835e-05,
|
| 611 |
+
"loss": 0.6544,
|
| 612 |
"step": 86
|
| 613 |
},
|
| 614 |
{
|
| 615 |
"epoch": 1.526315789473684,
|
| 616 |
+
"grad_norm": 0.6290459893449781,
|
| 617 |
"learning_rate": 4.194329997864331e-05,
|
| 618 |
+
"loss": 0.8034,
|
| 619 |
"step": 87
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"epoch": 1.543859649122807,
|
| 623 |
+
"grad_norm": 0.586369579542573,
|
| 624 |
"learning_rate": 4.172826515897146e-05,
|
| 625 |
+
"loss": 0.6832,
|
| 626 |
"step": 88
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 1.5614035087719298,
|
| 630 |
+
"grad_norm": 0.5753053969606661,
|
| 631 |
"learning_rate": 4.1510965599975196e-05,
|
| 632 |
+
"loss": 0.7196,
|
| 633 |
"step": 89
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 1.5789473684210527,
|
| 637 |
+
"grad_norm": 0.6527437493399515,
|
| 638 |
"learning_rate": 4.129143072053638e-05,
|
| 639 |
+
"loss": 0.6268,
|
| 640 |
"step": 90
|
| 641 |
},
|
| 642 |
{
|
| 643 |
"epoch": 1.5964912280701755,
|
| 644 |
+
"grad_norm": 0.5347518989317783,
|
| 645 |
"learning_rate": 4.1069690242163484e-05,
|
| 646 |
+
"loss": 0.66,
|
| 647 |
"step": 91
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"epoch": 1.6140350877192984,
|
| 651 |
+
"grad_norm": 0.5466222360441246,
|
| 652 |
"learning_rate": 4.0845774184967754e-05,
|
| 653 |
+
"loss": 0.6863,
|
| 654 |
"step": 92
|
| 655 |
},
|
| 656 |
{
|
| 657 |
"epoch": 1.631578947368421,
|
| 658 |
+
"grad_norm": 0.5287265061280089,
|
| 659 |
"learning_rate": 4.0619712863599e-05,
|
| 660 |
+
"loss": 0.6054,
|
| 661 |
"step": 93
|
| 662 |
},
|
| 663 |
{
|
| 664 |
"epoch": 1.6491228070175439,
|
| 665 |
+
"grad_norm": 0.5020449411728808,
|
| 666 |
"learning_rate": 4.039153688314145e-05,
|
| 667 |
+
"loss": 0.6106,
|
| 668 |
"step": 94
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 1.6666666666666665,
|
| 672 |
+
"grad_norm": 0.5326278815255991,
|
| 673 |
"learning_rate": 4.0161277134970345e-05,
|
| 674 |
+
"loss": 0.7097,
|
| 675 |
"step": 95
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 1.6842105263157894,
|
| 679 |
+
"grad_norm": 0.5377095024358528,
|
| 680 |
"learning_rate": 3.9928964792569655e-05,
|
| 681 |
+
"loss": 0.6227,
|
| 682 |
"step": 96
|
| 683 |
},
|
| 684 |
{
|
| 685 |
"epoch": 1.7017543859649122,
|
| 686 |
+
"grad_norm": 0.576841931441445,
|
| 687 |
"learning_rate": 3.969463130731183e-05,
|
| 688 |
+
"loss": 0.7862,
|
| 689 |
"step": 97
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"epoch": 1.719298245614035,
|
| 693 |
+
"grad_norm": 0.6066486813440297,
|
| 694 |
"learning_rate": 3.945830840419966e-05,
|
| 695 |
+
"loss": 0.6729,
|
| 696 |
"step": 98
|
| 697 |
},
|
| 698 |
{
|
| 699 |
"epoch": 1.736842105263158,
|
| 700 |
+
"grad_norm": 0.5159053057386966,
|
| 701 |
"learning_rate": 3.9220028077571295e-05,
|
| 702 |
+
"loss": 0.7059,
|
| 703 |
"step": 99
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"epoch": 1.7543859649122808,
|
| 707 |
+
"grad_norm": 0.4950698707861886,
|
| 708 |
"learning_rate": 3.897982258676867e-05,
|
| 709 |
+
"loss": 0.6351,
|
| 710 |
"step": 100
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 1.7719298245614035,
|
| 714 |
+
"grad_norm": 0.5420056949722063,
|
| 715 |
"learning_rate": 3.873772445177015e-05,
|
| 716 |
+
"loss": 0.711,
|
| 717 |
"step": 101
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 1.7894736842105263,
|
| 721 |
+
"grad_norm": 0.4564353605846405,
|
| 722 |
"learning_rate": 3.8493766448787825e-05,
|
| 723 |
+
"loss": 0.7022,
|
| 724 |
"step": 102
|
| 725 |
},
|
| 726 |
{
|
| 727 |
"epoch": 1.807017543859649,
|
| 728 |
+
"grad_norm": 0.5625326599444068,
|
| 729 |
"learning_rate": 3.824798160583012e-05,
|
| 730 |
+
"loss": 0.7015,
|
| 731 |
"step": 103
|
| 732 |
},
|
| 733 |
{
|
| 734 |
"epoch": 1.8245614035087718,
|
| 735 |
+
"grad_norm": 0.48056030105986436,
|
| 736 |
"learning_rate": 3.8000403198230387e-05,
|
| 737 |
+
"loss": 0.5815,
|
| 738 |
"step": 104
|
| 739 |
},
|
| 740 |
{
|
| 741 |
"epoch": 1.8421052631578947,
|
| 742 |
+
"grad_norm": 0.453650733165799,
|
| 743 |
"learning_rate": 3.775106474414188e-05,
|
| 744 |
+
"loss": 0.5505,
|
| 745 |
"step": 105
|
| 746 |
},
|
| 747 |
{
|
| 748 |
"epoch": 1.8596491228070176,
|
| 749 |
+
"grad_norm": 0.9587534224527449,
|
| 750 |
"learning_rate": 3.7500000000000003e-05,
|
| 751 |
+
"loss": 0.7681,
|
| 752 |
"step": 106
|
| 753 |
},
|
| 754 |
{
|
| 755 |
"epoch": 1.8771929824561404,
|
| 756 |
+
"grad_norm": 0.6628313082859163,
|
| 757 |
"learning_rate": 3.7247242955952175e-05,
|
| 758 |
+
"loss": 0.6832,
|
| 759 |
"step": 107
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 1.8947368421052633,
|
| 763 |
+
"grad_norm": 0.5627752640940548,
|
| 764 |
"learning_rate": 3.699282783125616e-05,
|
| 765 |
+
"loss": 0.6747,
|
| 766 |
"step": 108
|
| 767 |
},
|
| 768 |
{
|
| 769 |
"epoch": 1.912280701754386,
|
| 770 |
+
"grad_norm": 0.5379495312193924,
|
| 771 |
"learning_rate": 3.673678906964727e-05,
|
| 772 |
+
"loss": 0.5904,
|
| 773 |
"step": 109
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"epoch": 1.9298245614035088,
|
| 777 |
+
"grad_norm": 0.5290950361826221,
|
| 778 |
"learning_rate": 3.6479161334675296e-05,
|
| 779 |
+
"loss": 0.7161,
|
| 780 |
"step": 110
|
| 781 |
},
|
| 782 |
{
|
| 783 |
"epoch": 1.9473684210526314,
|
| 784 |
+
"grad_norm": 0.6329223546784813,
|
| 785 |
"learning_rate": 3.621997950501156e-05,
|
| 786 |
+
"loss": 0.7186,
|
| 787 |
"step": 111
|
| 788 |
},
|
| 789 |
{
|
| 790 |
"epoch": 1.9649122807017543,
|
| 791 |
+
"grad_norm": 0.722712664381171,
|
| 792 |
"learning_rate": 3.5959278669726935e-05,
|
| 793 |
+
"loss": 0.6872,
|
| 794 |
"step": 112
|
| 795 |
},
|
| 796 |
{
|
| 797 |
"epoch": 1.9824561403508771,
|
| 798 |
+
"grad_norm": 0.5556537835599905,
|
| 799 |
"learning_rate": 3.569709412354136e-05,
|
| 800 |
+
"loss": 0.7653,
|
| 801 |
"step": 113
|
| 802 |
},
|
| 803 |
{
|
| 804 |
"epoch": 2.0,
|
| 805 |
+
"grad_norm": 0.7972949317622412,
|
| 806 |
"learning_rate": 3.543346136204545e-05,
|
| 807 |
+
"loss": 0.5974,
|
| 808 |
"step": 114
|
| 809 |
}
|
| 810 |
],
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8081
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3b09193bae298a0a841c37ce61432d61767b4d633b9680c59539c690bef78161
|
| 3 |
size 8081
|