Training in progress, step 200
Browse files- config.json +2 -2
- model.safetensors +2 -2
- tokenizer.json +52 -51
config.json
CHANGED
|
@@ -40,7 +40,7 @@
|
|
| 40 |
"tie_word_embeddings": true,
|
| 41 |
"type_vocab_size": 2,
|
| 42 |
"use_cache": true,
|
| 43 |
-
"vocab_size":
|
| 44 |
},
|
| 45 |
"decoder_start_token_id": 2,
|
| 46 |
"dtype": "float32",
|
|
@@ -82,7 +82,7 @@
|
|
| 82 |
"tie_word_embeddings": true,
|
| 83 |
"type_vocab_size": 2,
|
| 84 |
"use_cache": true,
|
| 85 |
-
"vocab_size":
|
| 86 |
},
|
| 87 |
"eos_token_id": 0,
|
| 88 |
"is_encoder_decoder": true,
|
|
|
|
| 40 |
"tie_word_embeddings": true,
|
| 41 |
"type_vocab_size": 2,
|
| 42 |
"use_cache": true,
|
| 43 |
+
"vocab_size": 53
|
| 44 |
},
|
| 45 |
"decoder_start_token_id": 2,
|
| 46 |
"dtype": "float32",
|
|
|
|
| 82 |
"tie_word_embeddings": true,
|
| 83 |
"type_vocab_size": 2,
|
| 84 |
"use_cache": true,
|
| 85 |
+
"vocab_size": 53
|
| 86 |
},
|
| 87 |
"eos_token_id": 0,
|
| 88 |
"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:d95ae3ec874591902e206da158e0e0039100edff3acc797f818f86819dca4e57
|
| 3 |
+
size 31207604
|
tokenizer.json
CHANGED
|
@@ -113,42 +113,43 @@
|
|
| 113 |
"7": 13,
|
| 114 |
"8": 14,
|
| 115 |
"9": 15,
|
| 116 |
-
"
|
| 117 |
-
"
|
| 118 |
-
"
|
| 119 |
-
"
|
| 120 |
-
"
|
| 121 |
"97": 21,
|
| 122 |
-
"
|
| 123 |
-
"
|
| 124 |
-
"
|
| 125 |
-
"
|
| 126 |
-
"
|
| 127 |
-
"
|
| 128 |
-
"
|
| 129 |
-
"
|
| 130 |
-
"
|
| 131 |
"92": 31,
|
| 132 |
-
"
|
| 133 |
"91": 33,
|
| 134 |
-
"
|
| 135 |
-
"
|
| 136 |
-
"
|
| 137 |
-
"
|
| 138 |
-
"
|
| 139 |
"88": 39,
|
| 140 |
-
"
|
| 141 |
-
"
|
| 142 |
-
"
|
| 143 |
"86": 43,
|
| 144 |
-
"
|
| 145 |
-
"
|
| 146 |
-
"
|
| 147 |
"25": 47,
|
| 148 |
-
"
|
| 149 |
-
"
|
| 150 |
-
"
|
| 151 |
-
"
|
|
|
|
| 152 |
},
|
| 153 |
"merges": [
|
| 154 |
[
|
|
@@ -167,14 +168,14 @@
|
|
| 167 |
"9",
|
| 168 |
"8"
|
| 169 |
],
|
| 170 |
-
[
|
| 171 |
-
"1",
|
| 172 |
-
"2"
|
| 173 |
-
],
|
| 174 |
[
|
| 175 |
"9",
|
| 176 |
"7"
|
| 177 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 178 |
[
|
| 179 |
"1",
|
| 180 |
"3"
|
|
@@ -207,22 +208,22 @@
|
|
| 207 |
"9",
|
| 208 |
"3"
|
| 209 |
],
|
| 210 |
-
[
|
| 211 |
-
"1",
|
| 212 |
-
"7"
|
| 213 |
-
],
|
| 214 |
[
|
| 215 |
"9",
|
| 216 |
"2"
|
| 217 |
],
|
| 218 |
[
|
| 219 |
"1",
|
| 220 |
-
"
|
| 221 |
],
|
| 222 |
[
|
| 223 |
"9",
|
| 224 |
"1"
|
| 225 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 226 |
[
|
| 227 |
"1",
|
| 228 |
"9"
|
|
@@ -239,14 +240,14 @@
|
|
| 239 |
"8",
|
| 240 |
"9"
|
| 241 |
],
|
| 242 |
-
[
|
| 243 |
-
"2",
|
| 244 |
-
"1"
|
| 245 |
-
],
|
| 246 |
[
|
| 247 |
"8",
|
| 248 |
"8"
|
| 249 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 250 |
[
|
| 251 |
"2",
|
| 252 |
"2"
|
|
@@ -255,14 +256,14 @@
|
|
| 255 |
"8",
|
| 256 |
"7"
|
| 257 |
],
|
| 258 |
-
[
|
| 259 |
-
"2",
|
| 260 |
-
"3"
|
| 261 |
-
],
|
| 262 |
[
|
| 263 |
"8",
|
| 264 |
"6"
|
| 265 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 266 |
[
|
| 267 |
"8",
|
| 268 |
"5"
|
|
@@ -271,14 +272,14 @@
|
|
| 271 |
"2",
|
| 272 |
"4"
|
| 273 |
],
|
| 274 |
-
[
|
| 275 |
-
"8",
|
| 276 |
-
"4"
|
| 277 |
-
],
|
| 278 |
[
|
| 279 |
"2",
|
| 280 |
"5"
|
| 281 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 282 |
[
|
| 283 |
"2",
|
| 284 |
"6"
|
|
|
|
| 113 |
"7": 13,
|
| 114 |
"8": 14,
|
| 115 |
"9": 15,
|
| 116 |
+
"=": 16,
|
| 117 |
+
"10": 17,
|
| 118 |
+
"99": 18,
|
| 119 |
+
"11": 19,
|
| 120 |
+
"98": 20,
|
| 121 |
"97": 21,
|
| 122 |
+
"12": 22,
|
| 123 |
+
"13": 23,
|
| 124 |
+
"96": 24,
|
| 125 |
+
"95": 25,
|
| 126 |
+
"14": 26,
|
| 127 |
+
"94": 27,
|
| 128 |
+
"15": 28,
|
| 129 |
+
"16": 29,
|
| 130 |
+
"93": 30,
|
| 131 |
"92": 31,
|
| 132 |
+
"17": 32,
|
| 133 |
"91": 33,
|
| 134 |
+
"18": 34,
|
| 135 |
+
"19": 35,
|
| 136 |
+
"90": 36,
|
| 137 |
+
"20": 37,
|
| 138 |
+
"89": 38,
|
| 139 |
"88": 39,
|
| 140 |
+
"21": 40,
|
| 141 |
+
"22": 41,
|
| 142 |
+
"87": 42,
|
| 143 |
"86": 43,
|
| 144 |
+
"23": 44,
|
| 145 |
+
"85": 45,
|
| 146 |
+
"24": 46,
|
| 147 |
"25": 47,
|
| 148 |
+
"84": 48,
|
| 149 |
+
"26": 49,
|
| 150 |
+
"83": 50,
|
| 151 |
+
"27": 51,
|
| 152 |
+
"82": 52
|
| 153 |
},
|
| 154 |
"merges": [
|
| 155 |
[
|
|
|
|
| 168 |
"9",
|
| 169 |
"8"
|
| 170 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 171 |
[
|
| 172 |
"9",
|
| 173 |
"7"
|
| 174 |
],
|
| 175 |
+
[
|
| 176 |
+
"1",
|
| 177 |
+
"2"
|
| 178 |
+
],
|
| 179 |
[
|
| 180 |
"1",
|
| 181 |
"3"
|
|
|
|
| 208 |
"9",
|
| 209 |
"3"
|
| 210 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 211 |
[
|
| 212 |
"9",
|
| 213 |
"2"
|
| 214 |
],
|
| 215 |
[
|
| 216 |
"1",
|
| 217 |
+
"7"
|
| 218 |
],
|
| 219 |
[
|
| 220 |
"9",
|
| 221 |
"1"
|
| 222 |
],
|
| 223 |
+
[
|
| 224 |
+
"1",
|
| 225 |
+
"8"
|
| 226 |
+
],
|
| 227 |
[
|
| 228 |
"1",
|
| 229 |
"9"
|
|
|
|
| 240 |
"8",
|
| 241 |
"9"
|
| 242 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 243 |
[
|
| 244 |
"8",
|
| 245 |
"8"
|
| 246 |
],
|
| 247 |
+
[
|
| 248 |
+
"2",
|
| 249 |
+
"1"
|
| 250 |
+
],
|
| 251 |
[
|
| 252 |
"2",
|
| 253 |
"2"
|
|
|
|
| 256 |
"8",
|
| 257 |
"7"
|
| 258 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 259 |
[
|
| 260 |
"8",
|
| 261 |
"6"
|
| 262 |
],
|
| 263 |
+
[
|
| 264 |
+
"2",
|
| 265 |
+
"3"
|
| 266 |
+
],
|
| 267 |
[
|
| 268 |
"8",
|
| 269 |
"5"
|
|
|
|
| 272 |
"2",
|
| 273 |
"4"
|
| 274 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 275 |
[
|
| 276 |
"2",
|
| 277 |
"5"
|
| 278 |
],
|
| 279 |
+
[
|
| 280 |
+
"8",
|
| 281 |
+
"4"
|
| 282 |
+
],
|
| 283 |
[
|
| 284 |
"2",
|
| 285 |
"6"
|