Training in progress, step 240
Browse files- config.json +2 -2
- model.safetensors +2 -2
- tokenizer.json +32 -37
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": 51
|
| 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": 51
|
| 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 sha256:2abf59bf8cac9d70ea8d57cf9818a9a623399ddfae3f391d43a8553f55e8aa2e
|
| 3 |
+
size 31203500
|
tokenizer.json
CHANGED
|
@@ -115,26 +115,26 @@
|
|
| 115 |
"9": 15,
|
| 116 |
"99": 16,
|
| 117 |
"10": 17,
|
| 118 |
-
"
|
| 119 |
-
"
|
| 120 |
-
"
|
| 121 |
-
"
|
| 122 |
-
"
|
| 123 |
-
"
|
| 124 |
"14": 24,
|
| 125 |
"95": 25,
|
| 126 |
"15": 26,
|
| 127 |
"94": 27,
|
| 128 |
-
"
|
| 129 |
-
"
|
| 130 |
"17": 30,
|
| 131 |
"92": 31,
|
| 132 |
-
"
|
| 133 |
-
"
|
| 134 |
"19": 34,
|
| 135 |
"90": 35,
|
| 136 |
-
"
|
| 137 |
-
"
|
| 138 |
"21": 38,
|
| 139 |
"88": 39,
|
| 140 |
"87": 40,
|
|
@@ -147,8 +147,7 @@
|
|
| 147 |
"84": 47,
|
| 148 |
"26": 48,
|
| 149 |
"83": 49,
|
| 150 |
-
"
|
| 151 |
-
"82": 51
|
| 152 |
},
|
| 153 |
"merges": [
|
| 154 |
[
|
|
@@ -159,30 +158,30 @@
|
|
| 159 |
"1",
|
| 160 |
"0"
|
| 161 |
],
|
| 162 |
-
[
|
| 163 |
-
"9",
|
| 164 |
-
"8"
|
| 165 |
-
],
|
| 166 |
[
|
| 167 |
"1",
|
| 168 |
"1"
|
| 169 |
],
|
| 170 |
[
|
| 171 |
"9",
|
| 172 |
-
"
|
| 173 |
],
|
| 174 |
[
|
| 175 |
"1",
|
| 176 |
"2"
|
| 177 |
],
|
| 178 |
[
|
| 179 |
-
"
|
| 180 |
-
"
|
| 181 |
],
|
| 182 |
[
|
| 183 |
"9",
|
| 184 |
"6"
|
| 185 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 186 |
[
|
| 187 |
"1",
|
| 188 |
"4"
|
|
@@ -199,30 +198,30 @@
|
|
| 199 |
"9",
|
| 200 |
"4"
|
| 201 |
],
|
| 202 |
-
[
|
| 203 |
-
"1",
|
| 204 |
-
"6"
|
| 205 |
-
],
|
| 206 |
[
|
| 207 |
"9",
|
| 208 |
"3"
|
| 209 |
],
|
| 210 |
[
|
| 211 |
"1",
|
| 212 |
-
"
|
| 213 |
],
|
| 214 |
[
|
| 215 |
-
"
|
| 216 |
-
"
|
| 217 |
],
|
| 218 |
[
|
| 219 |
"9",
|
| 220 |
-
"
|
| 221 |
],
|
| 222 |
[
|
| 223 |
"1",
|
| 224 |
"8"
|
| 225 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 226 |
[
|
| 227 |
"1",
|
| 228 |
"9"
|
|
@@ -231,14 +230,14 @@
|
|
| 231 |
"9",
|
| 232 |
"0"
|
| 233 |
],
|
| 234 |
-
[
|
| 235 |
-
"8",
|
| 236 |
-
"9"
|
| 237 |
-
],
|
| 238 |
[
|
| 239 |
"2",
|
| 240 |
"0"
|
| 241 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 242 |
[
|
| 243 |
"2",
|
| 244 |
"1"
|
|
@@ -287,10 +286,6 @@
|
|
| 287 |
"8",
|
| 288 |
"3"
|
| 289 |
],
|
| 290 |
-
[
|
| 291 |
-
"2",
|
| 292 |
-
"7"
|
| 293 |
-
],
|
| 294 |
[
|
| 295 |
"8",
|
| 296 |
"2"
|
|
|
|
| 115 |
"9": 15,
|
| 116 |
"99": 16,
|
| 117 |
"10": 17,
|
| 118 |
+
"11": 18,
|
| 119 |
+
"98": 19,
|
| 120 |
+
"12": 20,
|
| 121 |
+
"97": 21,
|
| 122 |
+
"96": 22,
|
| 123 |
+
"13": 23,
|
| 124 |
"14": 24,
|
| 125 |
"95": 25,
|
| 126 |
"15": 26,
|
| 127 |
"94": 27,
|
| 128 |
+
"93": 28,
|
| 129 |
+
"16": 29,
|
| 130 |
"17": 30,
|
| 131 |
"92": 31,
|
| 132 |
+
"18": 32,
|
| 133 |
+
"91": 33,
|
| 134 |
"19": 34,
|
| 135 |
"90": 35,
|
| 136 |
+
"20": 36,
|
| 137 |
+
"89": 37,
|
| 138 |
"21": 38,
|
| 139 |
"88": 39,
|
| 140 |
"87": 40,
|
|
|
|
| 147 |
"84": 47,
|
| 148 |
"26": 48,
|
| 149 |
"83": 49,
|
| 150 |
+
"82": 50
|
|
|
|
| 151 |
},
|
| 152 |
"merges": [
|
| 153 |
[
|
|
|
|
| 158 |
"1",
|
| 159 |
"0"
|
| 160 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 161 |
[
|
| 162 |
"1",
|
| 163 |
"1"
|
| 164 |
],
|
| 165 |
[
|
| 166 |
"9",
|
| 167 |
+
"8"
|
| 168 |
],
|
| 169 |
[
|
| 170 |
"1",
|
| 171 |
"2"
|
| 172 |
],
|
| 173 |
[
|
| 174 |
+
"9",
|
| 175 |
+
"7"
|
| 176 |
],
|
| 177 |
[
|
| 178 |
"9",
|
| 179 |
"6"
|
| 180 |
],
|
| 181 |
+
[
|
| 182 |
+
"1",
|
| 183 |
+
"3"
|
| 184 |
+
],
|
| 185 |
[
|
| 186 |
"1",
|
| 187 |
"4"
|
|
|
|
| 198 |
"9",
|
| 199 |
"4"
|
| 200 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 201 |
[
|
| 202 |
"9",
|
| 203 |
"3"
|
| 204 |
],
|
| 205 |
[
|
| 206 |
"1",
|
| 207 |
+
"6"
|
| 208 |
],
|
| 209 |
[
|
| 210 |
+
"1",
|
| 211 |
+
"7"
|
| 212 |
],
|
| 213 |
[
|
| 214 |
"9",
|
| 215 |
+
"2"
|
| 216 |
],
|
| 217 |
[
|
| 218 |
"1",
|
| 219 |
"8"
|
| 220 |
],
|
| 221 |
+
[
|
| 222 |
+
"9",
|
| 223 |
+
"1"
|
| 224 |
+
],
|
| 225 |
[
|
| 226 |
"1",
|
| 227 |
"9"
|
|
|
|
| 230 |
"9",
|
| 231 |
"0"
|
| 232 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 233 |
[
|
| 234 |
"2",
|
| 235 |
"0"
|
| 236 |
],
|
| 237 |
+
[
|
| 238 |
+
"8",
|
| 239 |
+
"9"
|
| 240 |
+
],
|
| 241 |
[
|
| 242 |
"2",
|
| 243 |
"1"
|
|
|
|
| 286 |
"8",
|
| 287 |
"3"
|
| 288 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
| 289 |
[
|
| 290 |
"8",
|
| 291 |
"2"
|