Training in progress, step 3500
Browse files
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": 105
|
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": 105
|
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 sha256:5991bed8b2dc214418740beee8998ea0710a0ebceb870b39a42a3ee9437f38e7
|
3 |
+
size 31314308
|
runs/Mar04_14-45-46_d8c8da50ec4e/events.out.tfevents.1709563547.d8c8da50ec4e.409.20
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5bdb0635ce49e845022293403cf512b57a359d24be58eeb20280203ebb5402ed
|
3 |
+
size 56829
|
runs/Mar04_14-46-55_d8c8da50ec4e/events.out.tfevents.1709563615.d8c8da50ec4e.409.21
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9c09cffcf49d611c6dedf25da90f73c9b13c75f7247fcfbb61935f7338745a2c
|
3 |
+
size 429544
|
tokenizer.json
CHANGED
@@ -101,200 +101,198 @@
|
|
101 |
"[CLS]": 2,
|
102 |
"[PAD]": 3,
|
103 |
"+": 4,
|
104 |
-
"
|
105 |
-
"
|
106 |
-
"
|
107 |
-
"
|
108 |
-
"
|
109 |
-
"
|
110 |
-
"
|
111 |
-
"
|
112 |
-
"
|
113 |
-
"
|
114 |
-
"
|
115 |
-
"
|
116 |
-
"
|
117 |
-
"
|
118 |
-
"
|
119 |
-
"
|
120 |
-
"
|
121 |
-
"
|
122 |
-
"
|
123 |
-
"
|
124 |
-
"
|
125 |
-
"
|
126 |
-
"
|
127 |
-
"
|
128 |
-
"
|
129 |
-
"
|
130 |
-
"
|
131 |
-
"
|
132 |
-
"
|
133 |
-
"
|
134 |
-
"
|
135 |
-
"
|
136 |
-
"
|
137 |
-
"
|
138 |
-
"
|
139 |
-
"
|
140 |
-
"
|
141 |
-
"
|
142 |
-
"
|
143 |
-
"
|
144 |
-
"
|
145 |
-
"
|
146 |
-
"
|
147 |
-
"
|
148 |
-
"
|
149 |
-
"
|
150 |
-
"
|
151 |
-
"
|
152 |
-
"
|
153 |
-
"
|
154 |
-
"
|
155 |
-
"
|
156 |
"53": 57,
|
157 |
-
"
|
158 |
-
"
|
159 |
-
"
|
160 |
-
"
|
161 |
-
"
|
162 |
-
"
|
163 |
-
"
|
164 |
-
"
|
165 |
-
"
|
166 |
-
"
|
167 |
-
"
|
168 |
-
"
|
169 |
-
"
|
170 |
-
"
|
171 |
-
"
|
172 |
-
"
|
173 |
-
"
|
174 |
-
"
|
175 |
-
"
|
176 |
-
"
|
177 |
-
"
|
178 |
-
"
|
179 |
-
"
|
180 |
-
"
|
181 |
-
"
|
182 |
-
"
|
183 |
-
"
|
184 |
-
"
|
185 |
-
"
|
186 |
-
"
|
187 |
-
"
|
188 |
-
"
|
189 |
-
"
|
190 |
-
"
|
191 |
-
"
|
192 |
-
"
|
193 |
-
"
|
194 |
"55": 95,
|
195 |
-
"
|
196 |
-
"
|
197 |
-
"
|
198 |
"82": 99,
|
199 |
-
"
|
200 |
-
"
|
201 |
-
"
|
202 |
-
"
|
203 |
-
"
|
204 |
-
"98": 105,
|
205 |
-
"69": 106
|
206 |
},
|
207 |
"merges": [
|
208 |
-
"
|
209 |
-
"2
|
|
|
|
|
|
|
|
|
210 |
"1 2",
|
211 |
-
"2 0",
|
212 |
-
"2 5",
|
213 |
-
"3 5",
|
214 |
-
"5 1",
|
215 |
-
"8 6",
|
216 |
-
"2 2",
|
217 |
"4 4",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
218 |
"8 5",
|
219 |
-
"
|
220 |
-
"
|
221 |
-
"9 3",
|
222 |
-
"2 6",
|
223 |
-
"2 4",
|
224 |
-
"1 0",
|
225 |
-
"6 0",
|
226 |
-
"1 4",
|
227 |
-
"3 2",
|
228 |
-
"6 8",
|
229 |
-
"2 1",
|
230 |
-
"9 4",
|
231 |
-
"3 8",
|
232 |
-
"5 4",
|
233 |
-
"1 7",
|
234 |
"5 8",
|
235 |
-
"
|
236 |
-
"7 0",
|
237 |
-
"5 2",
|
238 |
-
"4 7",
|
239 |
-
"3 1",
|
240 |
"8 7",
|
|
|
|
|
|
|
241 |
"8 1",
|
|
|
|
|
|
|
|
|
|
|
|
|
242 |
"9 7",
|
243 |
-
"
|
244 |
-
"
|
|
|
245 |
"7 8",
|
246 |
-
"
|
247 |
-
"
|
|
|
|
|
|
|
|
|
248 |
"5 3",
|
249 |
-
"5 0",
|
250 |
-
"7 3",
|
251 |
-
"1 8",
|
252 |
-
"4 6",
|
253 |
"8 0",
|
254 |
-
"
|
255 |
-
"
|
256 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
257 |
"9 6",
|
258 |
-
"
|
259 |
-
"
|
260 |
-
"
|
261 |
-
"
|
262 |
-
"8 3",
|
263 |
-
"1 6",
|
264 |
-
"2 3",
|
265 |
-
"8 9",
|
266 |
-
"4 2",
|
267 |
-
"3 0",
|
268 |
-
"9 9",
|
269 |
-
"7 9",
|
270 |
"7 7",
|
271 |
-
"1
|
272 |
-
"7 1",
|
273 |
-
"6 6",
|
274 |
-
"9 1",
|
275 |
-
"4 0",
|
276 |
-
"6 2",
|
277 |
-
"1 9",
|
278 |
"4 1",
|
279 |
-
"5
|
|
|
|
|
|
|
280 |
"4 3",
|
281 |
-
"2
|
282 |
"1 3",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
283 |
"4 9",
|
284 |
-
"
|
285 |
-
"
|
|
|
286 |
"5 5",
|
287 |
-
"
|
288 |
-
"3
|
289 |
-
"7 4",
|
290 |
-
"8 2",
|
291 |
-
"3 4",
|
292 |
"9 0",
|
293 |
-
"
|
294 |
-
"
|
295 |
-
"6
|
296 |
-
"
|
297 |
-
"6
|
|
|
298 |
]
|
299 |
}
|
300 |
}
|
|
|
101 |
"[CLS]": 2,
|
102 |
"[PAD]": 3,
|
103 |
"+": 4,
|
104 |
+
"0": 5,
|
105 |
+
"1": 6,
|
106 |
+
"2": 7,
|
107 |
+
"3": 8,
|
108 |
+
"4": 9,
|
109 |
+
"5": 10,
|
110 |
+
"6": 11,
|
111 |
+
"7": 12,
|
112 |
+
"8": 13,
|
113 |
+
"9": 14,
|
114 |
+
"46": 15,
|
115 |
+
"23": 16,
|
116 |
+
"32": 17,
|
117 |
+
"59": 18,
|
118 |
+
"63": 19,
|
119 |
+
"18": 20,
|
120 |
+
"12": 21,
|
121 |
+
"44": 22,
|
122 |
+
"72": 23,
|
123 |
+
"91": 24,
|
124 |
+
"95": 25,
|
125 |
+
"27": 26,
|
126 |
+
"48": 27,
|
127 |
+
"57": 28,
|
128 |
+
"61": 29,
|
129 |
+
"85": 30,
|
130 |
+
"20": 31,
|
131 |
+
"37": 32,
|
132 |
+
"58": 33,
|
133 |
+
"86": 34,
|
134 |
+
"87": 35,
|
135 |
+
"14": 36,
|
136 |
+
"28": 37,
|
137 |
+
"71": 38,
|
138 |
+
"81": 39,
|
139 |
+
"89": 40,
|
140 |
+
"98": 41,
|
141 |
+
"24": 42,
|
142 |
+
"31": 43,
|
143 |
+
"93": 44,
|
144 |
+
"94": 45,
|
145 |
+
"97": 46,
|
146 |
+
"33": 47,
|
147 |
+
"36": 48,
|
148 |
+
"50": 49,
|
149 |
+
"78": 50,
|
150 |
+
"83": 51,
|
151 |
+
"84": 52,
|
152 |
+
"10": 53,
|
153 |
+
"16": 54,
|
154 |
+
"35": 55,
|
155 |
+
"39": 56,
|
156 |
"53": 57,
|
157 |
+
"80": 58,
|
158 |
+
"19": 59,
|
159 |
+
"29": 60,
|
160 |
+
"38": 61,
|
161 |
+
"47": 62,
|
162 |
+
"54": 63,
|
163 |
+
"66": 64,
|
164 |
+
"74": 65,
|
165 |
+
"11": 66,
|
166 |
+
"34": 67,
|
167 |
+
"96": 68,
|
168 |
+
"22": 69,
|
169 |
+
"56": 70,
|
170 |
+
"73": 71,
|
171 |
+
"75": 72,
|
172 |
+
"77": 73,
|
173 |
+
"21": 74,
|
174 |
+
"41": 75,
|
175 |
+
"65": 76,
|
176 |
+
"67": 77,
|
177 |
+
"68": 78,
|
178 |
+
"45": 79,
|
179 |
+
"43": 80,
|
180 |
+
"92": 81,
|
181 |
+
"13": 82,
|
182 |
+
"26": 83,
|
183 |
+
"62": 84,
|
184 |
+
"60": 85,
|
185 |
+
"64": 86,
|
186 |
+
"88": 87,
|
187 |
+
"15": 88,
|
188 |
+
"17": 89,
|
189 |
+
"40": 90,
|
190 |
+
"49": 91,
|
191 |
+
"52": 92,
|
192 |
+
"79": 93,
|
193 |
+
"25": 94,
|
194 |
"55": 95,
|
195 |
+
"99": 96,
|
196 |
+
"30": 97,
|
197 |
+
"90": 98,
|
198 |
"82": 99,
|
199 |
+
"51": 100,
|
200 |
+
"69": 101,
|
201 |
+
"42": 102,
|
202 |
+
"76": 103,
|
203 |
+
"70": 104
|
|
|
|
|
204 |
},
|
205 |
"merges": [
|
206 |
+
"4 6",
|
207 |
+
"2 3",
|
208 |
+
"3 2",
|
209 |
+
"5 9",
|
210 |
+
"6 3",
|
211 |
+
"1 8",
|
212 |
"1 2",
|
|
|
|
|
|
|
|
|
|
|
|
|
213 |
"4 4",
|
214 |
+
"7 2",
|
215 |
+
"9 1",
|
216 |
+
"9 5",
|
217 |
+
"2 7",
|
218 |
+
"4 8",
|
219 |
+
"5 7",
|
220 |
+
"6 1",
|
221 |
"8 5",
|
222 |
+
"2 0",
|
223 |
+
"3 7",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
224 |
"5 8",
|
225 |
+
"8 6",
|
|
|
|
|
|
|
|
|
226 |
"8 7",
|
227 |
+
"1 4",
|
228 |
+
"2 8",
|
229 |
+
"7 1",
|
230 |
"8 1",
|
231 |
+
"8 9",
|
232 |
+
"9 8",
|
233 |
+
"2 4",
|
234 |
+
"3 1",
|
235 |
+
"9 3",
|
236 |
+
"9 4",
|
237 |
"9 7",
|
238 |
+
"3 3",
|
239 |
+
"3 6",
|
240 |
+
"5 0",
|
241 |
"7 8",
|
242 |
+
"8 3",
|
243 |
+
"8 4",
|
244 |
+
"1 0",
|
245 |
+
"1 6",
|
246 |
+
"3 5",
|
247 |
+
"3 9",
|
248 |
"5 3",
|
|
|
|
|
|
|
|
|
249 |
"8 0",
|
250 |
+
"1 9",
|
251 |
+
"2 9",
|
252 |
+
"3 8",
|
253 |
+
"4 7",
|
254 |
+
"5 4",
|
255 |
+
"6 6",
|
256 |
+
"7 4",
|
257 |
+
"1 1",
|
258 |
+
"3 4",
|
259 |
"9 6",
|
260 |
+
"2 2",
|
261 |
+
"5 6",
|
262 |
+
"7 3",
|
263 |
+
"7 5",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
264 |
"7 7",
|
265 |
+
"2 1",
|
|
|
|
|
|
|
|
|
|
|
|
|
266 |
"4 1",
|
267 |
+
"6 5",
|
268 |
+
"6 7",
|
269 |
+
"6 8",
|
270 |
+
"4 5",
|
271 |
"4 3",
|
272 |
+
"9 2",
|
273 |
"1 3",
|
274 |
+
"2 6",
|
275 |
+
"6 2",
|
276 |
+
"6 0",
|
277 |
+
"6 4",
|
278 |
+
"8 8",
|
279 |
+
"1 5",
|
280 |
+
"1 7",
|
281 |
+
"4 0",
|
282 |
"4 9",
|
283 |
+
"5 2",
|
284 |
+
"7 9",
|
285 |
+
"2 5",
|
286 |
"5 5",
|
287 |
+
"9 9",
|
288 |
+
"3 0",
|
|
|
|
|
|
|
289 |
"9 0",
|
290 |
+
"8 2",
|
291 |
+
"5 1",
|
292 |
+
"6 9",
|
293 |
+
"4 2",
|
294 |
+
"7 6",
|
295 |
+
"7 0"
|
296 |
]
|
297 |
}
|
298 |
}
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5112
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:02b2bd218691a408d5c59002c641453e4f0410c056fa4553b9db65b8f27be893
|
3 |
size 5112
|