Training in progress, step 240
Browse files- config.json +2 -2
- model.safetensors +2 -2
- runs/Mar03_17-56-25_87c5064f18ba/events.out.tfevents.1741024586.87c5064f18ba.406.1 +3 -0
- tokenizer.json +10 -140
- training_args.bin +1 -1
config.json
CHANGED
|
@@ -78,7 +78,7 @@
|
|
| 78 |
"typical_p": 1.0,
|
| 79 |
"use_bfloat16": false,
|
| 80 |
"use_cache": true,
|
| 81 |
-
"vocab_size":
|
| 82 |
},
|
| 83 |
"decoder_start_token_id": 2,
|
| 84 |
"encoder": {
|
|
@@ -157,7 +157,7 @@
|
|
| 157 |
"typical_p": 1.0,
|
| 158 |
"use_bfloat16": false,
|
| 159 |
"use_cache": true,
|
| 160 |
-
"vocab_size":
|
| 161 |
},
|
| 162 |
"eos_token_id": 0,
|
| 163 |
"is_encoder_decoder": true,
|
|
|
|
| 78 |
"typical_p": 1.0,
|
| 79 |
"use_bfloat16": false,
|
| 80 |
"use_cache": true,
|
| 81 |
+
"vocab_size": 25
|
| 82 |
},
|
| 83 |
"decoder_start_token_id": 2,
|
| 84 |
"encoder": {
|
|
|
|
| 157 |
"typical_p": 1.0,
|
| 158 |
"use_bfloat16": false,
|
| 159 |
"use_cache": true,
|
| 160 |
+
"vocab_size": 25
|
| 161 |
},
|
| 162 |
"eos_token_id": 0,
|
| 163 |
"is_encoder_decoder": true,
|
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 sha256:69faef0b5e4d9ffca24550e7c0876e0b4914a085780635e7ffcf1b1a652a9724
|
| 3 |
+
size 31150148
|
runs/Mar03_17-56-25_87c5064f18ba/events.out.tfevents.1741024586.87c5064f18ba.406.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3154d699cbf52fb40d26559e21e4f18b9153cfacbbf21a70868bb8b69bc29fd
|
| 3 |
+
size 28368
|
tokenizer.json
CHANGED
|
@@ -113,95 +113,41 @@
|
|
| 113 |
"7": 13,
|
| 114 |
"8": 14,
|
| 115 |
"9": 15,
|
| 116 |
-
"
|
| 117 |
"11": 17,
|
| 118 |
-
"
|
| 119 |
-
"
|
| 120 |
-
"
|
| 121 |
-
"
|
| 122 |
-
"
|
| 123 |
-
"
|
| 124 |
-
"
|
| 125 |
-
"95": 25,
|
| 126 |
-
"15": 26,
|
| 127 |
-
"94": 27,
|
| 128 |
-
"93": 28,
|
| 129 |
-
"16": 29,
|
| 130 |
-
"17": 30,
|
| 131 |
-
"92": 31,
|
| 132 |
-
"91": 32,
|
| 133 |
-
"18": 33,
|
| 134 |
-
"19": 34,
|
| 135 |
-
"90": 35,
|
| 136 |
-
"89": 36,
|
| 137 |
-
"20": 37,
|
| 138 |
-
"21": 38,
|
| 139 |
-
"88": 39,
|
| 140 |
-
"22": 40,
|
| 141 |
-
"87": 41,
|
| 142 |
-
"86": 42,
|
| 143 |
-
"23": 43,
|
| 144 |
-
"85": 44,
|
| 145 |
-
"24": 45,
|
| 146 |
-
"25": 46,
|
| 147 |
-
"84": 47,
|
| 148 |
-
"26": 48,
|
| 149 |
-
"83": 49,
|
| 150 |
-
"82": 50
|
| 151 |
},
|
| 152 |
"merges": [
|
| 153 |
-
[
|
| 154 |
-
"9",
|
| 155 |
-
"9"
|
| 156 |
-
],
|
| 157 |
[
|
| 158 |
"1",
|
| 159 |
-
"
|
| 160 |
-
],
|
| 161 |
-
[
|
| 162 |
-
"9",
|
| 163 |
-
"8"
|
| 164 |
],
|
| 165 |
[
|
| 166 |
"1",
|
| 167 |
-
"
|
| 168 |
],
|
| 169 |
[
|
| 170 |
"1",
|
| 171 |
"2"
|
| 172 |
],
|
| 173 |
-
[
|
| 174 |
-
"9",
|
| 175 |
-
"7"
|
| 176 |
-
],
|
| 177 |
[
|
| 178 |
"1",
|
| 179 |
"3"
|
| 180 |
],
|
| 181 |
-
[
|
| 182 |
-
"9",
|
| 183 |
-
"6"
|
| 184 |
-
],
|
| 185 |
[
|
| 186 |
"1",
|
| 187 |
"4"
|
| 188 |
],
|
| 189 |
-
[
|
| 190 |
-
"9",
|
| 191 |
-
"5"
|
| 192 |
-
],
|
| 193 |
[
|
| 194 |
"1",
|
| 195 |
"5"
|
| 196 |
],
|
| 197 |
-
[
|
| 198 |
-
"9",
|
| 199 |
-
"4"
|
| 200 |
-
],
|
| 201 |
-
[
|
| 202 |
-
"9",
|
| 203 |
-
"3"
|
| 204 |
-
],
|
| 205 |
[
|
| 206 |
"1",
|
| 207 |
"6"
|
|
@@ -210,85 +156,9 @@
|
|
| 210 |
"1",
|
| 211 |
"7"
|
| 212 |
],
|
| 213 |
-
[
|
| 214 |
-
"9",
|
| 215 |
-
"2"
|
| 216 |
-
],
|
| 217 |
-
[
|
| 218 |
-
"9",
|
| 219 |
-
"1"
|
| 220 |
-
],
|
| 221 |
[
|
| 222 |
"1",
|
| 223 |
"8"
|
| 224 |
-
],
|
| 225 |
-
[
|
| 226 |
-
"1",
|
| 227 |
-
"9"
|
| 228 |
-
],
|
| 229 |
-
[
|
| 230 |
-
"9",
|
| 231 |
-
"0"
|
| 232 |
-
],
|
| 233 |
-
[
|
| 234 |
-
"8",
|
| 235 |
-
"9"
|
| 236 |
-
],
|
| 237 |
-
[
|
| 238 |
-
"2",
|
| 239 |
-
"0"
|
| 240 |
-
],
|
| 241 |
-
[
|
| 242 |
-
"2",
|
| 243 |
-
"1"
|
| 244 |
-
],
|
| 245 |
-
[
|
| 246 |
-
"8",
|
| 247 |
-
"8"
|
| 248 |
-
],
|
| 249 |
-
[
|
| 250 |
-
"2",
|
| 251 |
-
"2"
|
| 252 |
-
],
|
| 253 |
-
[
|
| 254 |
-
"8",
|
| 255 |
-
"7"
|
| 256 |
-
],
|
| 257 |
-
[
|
| 258 |
-
"8",
|
| 259 |
-
"6"
|
| 260 |
-
],
|
| 261 |
-
[
|
| 262 |
-
"2",
|
| 263 |
-
"3"
|
| 264 |
-
],
|
| 265 |
-
[
|
| 266 |
-
"8",
|
| 267 |
-
"5"
|
| 268 |
-
],
|
| 269 |
-
[
|
| 270 |
-
"2",
|
| 271 |
-
"4"
|
| 272 |
-
],
|
| 273 |
-
[
|
| 274 |
-
"2",
|
| 275 |
-
"5"
|
| 276 |
-
],
|
| 277 |
-
[
|
| 278 |
-
"8",
|
| 279 |
-
"4"
|
| 280 |
-
],
|
| 281 |
-
[
|
| 282 |
-
"2",
|
| 283 |
-
"6"
|
| 284 |
-
],
|
| 285 |
-
[
|
| 286 |
-
"8",
|
| 287 |
-
"3"
|
| 288 |
-
],
|
| 289 |
-
[
|
| 290 |
-
"8",
|
| 291 |
-
"2"
|
| 292 |
]
|
| 293 |
]
|
| 294 |
}
|
|
|
|
| 113 |
"7": 13,
|
| 114 |
"8": 14,
|
| 115 |
"9": 15,
|
| 116 |
+
"10": 16,
|
| 117 |
"11": 17,
|
| 118 |
+
"12": 18,
|
| 119 |
+
"13": 19,
|
| 120 |
+
"14": 20,
|
| 121 |
+
"15": 21,
|
| 122 |
+
"16": 22,
|
| 123 |
+
"17": 23,
|
| 124 |
+
"18": 24
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 125 |
},
|
| 126 |
"merges": [
|
|
|
|
|
|
|
|
|
|
|
|
|
| 127 |
[
|
| 128 |
"1",
|
| 129 |
+
"0"
|
|
|
|
|
|
|
|
|
|
|
|
|
| 130 |
],
|
| 131 |
[
|
| 132 |
"1",
|
| 133 |
+
"1"
|
| 134 |
],
|
| 135 |
[
|
| 136 |
"1",
|
| 137 |
"2"
|
| 138 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 139 |
[
|
| 140 |
"1",
|
| 141 |
"3"
|
| 142 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 143 |
[
|
| 144 |
"1",
|
| 145 |
"4"
|
| 146 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 147 |
[
|
| 148 |
"1",
|
| 149 |
"5"
|
| 150 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 151 |
[
|
| 152 |
"1",
|
| 153 |
"6"
|
|
|
|
| 156 |
"1",
|
| 157 |
"7"
|
| 158 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 159 |
[
|
| 160 |
"1",
|
| 161 |
"8"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 162 |
]
|
| 163 |
]
|
| 164 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5432
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9f08896176750d8bc479c04685ff68f257c2dffd50f192e82840e66514dc739a
|
| 3 |
size 5432
|