Hugging Face's logo

ChavyvAkvar
/
pathfox-baseline-3-epoch

Safetensors
path_attn
Model card Files
xet
 Community
ChavyvAkvar commited on
Commit
eb8812d
·
verified ·
1 Parent(s): 5802493

Upload folder using huggingface_hub

Browse files
Files changed (29) hide show
  1. checkpoint-14000/config.json +34 -0
  2. checkpoint-14000/generation_config.json +6 -0
  3. checkpoint-14000/model.safetensors +3 -0
  4. checkpoint-14000/optimizer.pt +3 -0
  5. checkpoint-14000/rng_state.pth +3 -0
  6. checkpoint-14000/scheduler.pt +3 -0
  7. checkpoint-14000/special_tokens_map.json +28 -0
  8. checkpoint-14000/tokenizer.json +1085 -0
  9. checkpoint-14000/tokenizer_config.json +64 -0
  10. checkpoint-14000/trainer_state.json +1994 -0
  11. checkpoint-14000/training_args.bin +3 -0
  12. checkpoint-15627/config.json +34 -0
  13. checkpoint-15627/generation_config.json +6 -0
  14. checkpoint-15627/model.safetensors +3 -0
  15. checkpoint-15627/optimizer.pt +3 -0
  16. checkpoint-15627/rng_state.pth +3 -0
  17. checkpoint-15627/scheduler.pt +3 -0
  18. checkpoint-15627/special_tokens_map.json +28 -0
  19. checkpoint-15627/tokenizer.json +1085 -0
  20. checkpoint-15627/tokenizer_config.json +64 -0
  21. checkpoint-15627/trainer_state.json +2218 -0
  22. checkpoint-15627/training_args.bin +3 -0
  23. config.json +34 -0
  24. generation_config.json +6 -0
  25. model.safetensors +3 -0
  26. special_tokens_map.json +28 -0
  27. tokenizer.json +1085 -0
  28. tokenizer_config.json +64 -0
  29. training_args.bin +3 -0
checkpoint-14000/config.json ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "PaTHAttentionForCausalLM"
4
+ ],
5
+ "bos_token_id": 0,
6
+ "dtype": "float32",
7
+ "elementwise_affine": true,
8
+ "eos_token_id": 1,
9
+ "fuse_cross_entropy": true,
10
+ "fuse_linear_cross_entropy": false,
11
+ "fuse_norm": true,
12
+ "fuse_swiglu": true,
13
+ "head_dim": 64,
14
+ "hidden_act": "swish",
15
+ "hidden_ratio": 4,
16
+ "hidden_size": 256,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 768,
19
+ "max_position_embeddings": 2048,
20
+ "model_type": "path_attn",
21
+ "norm_eps": 1e-05,
22
+ "num_heads": 4,
23
+ "num_hidden_layers": 32,
24
+ "num_kv_heads": 4,
25
+ "rope_theta": 10000.0,
26
+ "tie_word_embeddings": true,
27
+ "transformers_version": "4.57.1",
28
+ "use_cache": true,
29
+ "use_forget_gate": true,
30
+ "use_l2warp": false,
31
+ "use_low_rank_w": true,
32
+ "use_w_shortconv": true,
33
+ "vocab_size": 256
34
+ }
checkpoint-14000/generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 0,
4
+ "eos_token_id": 1,
5
+ "transformers_version": "4.57.1"
6
+ }
checkpoint-14000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e219ececba6269fac76a52c2bce066da8aaa1b4d65b75fd169d1160fec44178a
3
+ size 111892816
checkpoint-14000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e17da616a4be5ed590846db0027d86dd411e407ada37d6742624dce2c115b7f7
3
+ size 224116841
checkpoint-14000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4a9f217e852f439efa6bd32fde98d6867f11aa6ea13ddc021ba10af6a0b0934
3
+ size 14645
checkpoint-14000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7594d76cf2fb1944d4728057a7d6e4e912118c3926f3837095d76332de70f535
3
+ size 1465
checkpoint-14000/special_tokens_map.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<think>",
4
+ "</think>"
5
+ ],
6
+ "bos_token": {
7
+ "content": "<|startoftext|>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false
12
+ },
13
+ "eos_token": {
14
+ "content": "<|endoftext|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false
19
+ },
20
+ "pad_token": "<|pad|>",
21
+ "unk_token": {
22
+ "content": "<|unknown|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false
27
+ }
28
+ }
checkpoint-14000/tokenizer.json ADDED
@@ -0,0 +1,1085 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "1.0",
3
+ "truncation": null,
4
+ "padding": null,
5
+ "added_tokens": [
6
+ {
7
+ "id": 0,
8
+ "content": "<|startoftext|>",
9
+ "single_word": false,
10
+ "lstrip": false,
11
+ "rstrip": false,
12
+ "normalized": false,
13
+ "special": true
14
+ },
15
+ {
16
+ "id": 1,
17
+ "content": "<|endoftext|>",
18
+ "single_word": false,
19
+ "lstrip": false,
20
+ "rstrip": false,
21
+ "normalized": false,
22
+ "special": true
23
+ },
24
+ {
25
+ "id": 2,
26
+ "content": "<|unknown|>",
27
+ "single_word": false,
28
+ "lstrip": false,
29
+ "rstrip": false,
30
+ "normalized": false,
31
+ "special": true
32
+ },
33
+ {
34
+ "id": 3,
35
+ "content": "<|pad|>",
36
+ "single_word": false,
37
+ "lstrip": false,
38
+ "rstrip": false,
39
+ "normalized": false,
40
+ "special": true
41
+ },
42
+ {
43
+ "id": 4,
44
+ "content": "<think>",
45
+ "single_word": false,
46
+ "lstrip": false,
47
+ "rstrip": false,
48
+ "normalized": false,
49
+ "special": true
50
+ },
51
+ {
52
+ "id": 5,
53
+ "content": "</think>",
54
+ "single_word": false,
55
+ "lstrip": false,
56
+ "rstrip": false,
57
+ "normalized": false,
58
+ "special": true
59
+ }
60
+ ],
61
+ "normalizer": null,
62
+ "pre_tokenizer": {
63
+ "type": "ByteLevel",
64
+ "add_prefix_space": false,
65
+ "trim_offsets": true,
66
+ "use_regex": true
67
+ },
68
+ "post_processor": null,
69
+ "decoder": {
70
+ "type": "ByteLevel",
71
+ "add_prefix_space": true,
72
+ "trim_offsets": true,
73
+ "use_regex": true
74
+ },
75
+ "model": {
76
+ "type": "BPE",
77
+ "dropout": null,
78
+ "unk_token": null,
79
+ "continuing_subword_prefix": null,
80
+ "end_of_word_suffix": null,
81
+ "fuse_unk": false,
82
+ "byte_fallback": false,
83
+ "ignore_merges": false,
84
+ "vocab": {
85
+ "<|startoftext|>": 0,
86
+ "<|endoftext|>": 1,
87
+ "<|unknown|>": 2,
88
+ "<|pad|>": 3,
89
+ "<think>": 4,
90
+ "</think>": 5,
91
+ "%": 6,
92
+ "&": 7,
93
+ "'": 8,
94
+ "(": 9,
95
+ ")": 10,
96
+ ",": 11,
97
+ "-": 12,
98
+ ".": 13,
99
+ "0": 14,
100
+ "1": 15,
101
+ "2": 16,
102
+ "3": 17,
103
+ "4": 18,
104
+ "5": 19,
105
+ "6": 20,
106
+ "7": 21,
107
+ "8": 22,
108
+ "9": 23,
109
+ ":": 24,
110
+ ">": 25,
111
+ "A": 26,
112
+ "C": 27,
113
+ "D": 28,
114
+ "E": 29,
115
+ "F": 30,
116
+ "G": 31,
117
+ "H": 32,
118
+ "I": 33,
119
+ "L": 34,
120
+ "M": 35,
121
+ "N": 36,
122
+ "O": 37,
123
+ "P": 38,
124
+ "R": 39,
125
+ "S": 40,
126
+ "T": 41,
127
+ "V": 42,
128
+ "[": 43,
129
+ "]": 44,
130
+ "a": 45,
131
+ "b": 46,
132
+ "c": 47,
133
+ "d": 48,
134
+ "e": 49,
135
+ "f": 50,
136
+ "g": 51,
137
+ "h": 52,
138
+ "i": 53,
139
+ "k": 54,
140
+ "l": 55,
141
+ "m": 56,
142
+ "n": 57,
143
+ "o": 58,
144
+ "p": 59,
145
+ "q": 60,
146
+ "r": 61,
147
+ "s": 62,
148
+ "t": 63,
149
+ "u": 64,
150
+ "v": 65,
151
+ "x": 66,
152
+ "y": 67,
153
+ "|": 68,
154
+ "Ċ": 69,
155
+ "Ġ": 70,
156
+ "00": 71,
157
+ "11": 72,
158
+ "->": 73,
159
+ "Ġ|": 74,
160
+ ":[": 75,
161
+ "]->": 76,
162
+ "10": 77,
163
+ "Ġ1": 78,
164
+ "000": 79,
165
+ "ĠS": 80,
166
+ "',": 81,
167
+ "Ġ'": 82,
168
+ "te": 83,
169
+ "100": 84,
170
+ "ĊĠ": 85,
171
+ "110": 86,
172
+ "tep": 87,
173
+ "re": 88,
174
+ "111": 89,
175
+ "0000": 90,
176
+ "011": 91,
177
+ "101": 92,
178
+ "001": 93,
179
+ "ĠStep": 94,
180
+ "010": 95,
181
+ "in": 96,
182
+ "im": 97,
183
+ "ĠSim": 98,
184
+ "1111": 99,
185
+ "Ġ[": 100,
186
+ "or": 101,
187
+ "ĠI": 102,
188
+ "ho": 103,
189
+ "od": 104,
190
+ "hood": 105,
191
+ "Ġin": 106,
192
+ "Ġ11": 107,
193
+ "Ġ10": 108,
194
+ "ne": 109,
195
+ "Ġne": 110,
196
+ "']": 111,
197
+ "bor": 112,
198
+ "dx": 113,
199
+ "gh": 114,
200
+ "ire": 115,
201
+ "igh": 116,
202
+ "qu": 117,
203
+ "Ġre": 118,
204
+ "Ġ['": 119,
205
+ "ĠIdx": 120,
206
+ "Ġneigh": 121,
207
+ "borhood": 122,
208
+ "ires": 123,
209
+ "quires": 124,
210
+ "Ġrequires": 125,
211
+ "Ġneighborhood": 126,
212
+ "Step": 127,
213
+ "ut": 128,
214
+ "Ġ3": 129,
215
+ "Ġ2": 130,
216
+ "Ġ4": 131,
217
+ "Ġ5": 132,
218
+ "Ġ6": 133,
219
+ "Ġ9": 134,
220
+ "Ġ8": 135,
221
+ "Ġ7": 136,
222
+ "Ġ0": 137,
223
+ "ca": 138,
224
+ "Ġ12": 139,
225
+ "Ġ13": 140,
226
+ "1010": 141,
227
+ "Ġ14": 142,
228
+ "Out": 143,
229
+ "put": 144,
230
+ "ĠOut": 145,
231
+ "ĠSca": 146,
232
+ "ĠOutput": 147,
233
+ "ĠScan": 148,
234
+ "Ġ011": 149,
235
+ "ing": 150,
236
+ "ti": 151,
237
+ "1100": 152,
238
+ "00000000": 153,
239
+ "Ġ15": 154,
240
+ "Ġ000": 155,
241
+ "on": 156,
242
+ "Ġ110": 157,
243
+ "ul": 158,
244
+ "Ġ100": 159,
245
+ "at": 160,
246
+ "tion": 161,
247
+ "Ġ010": 162,
248
+ "1011": 163,
249
+ "ĠC": 164,
250
+ "Ġ00": 165,
251
+ "00011": 166,
252
+ "50": 167,
253
+ "er": 168,
254
+ "10000": 169,
255
+ "nd": 170,
256
+ "Ġ16": 171,
257
+ "ule": 172,
258
+ "ĠR": 173,
259
+ "ed": 174,
260
+ "li": 175,
261
+ "Ġ001": 176,
262
+ "Ġ101": 177,
263
+ "ate": 178,
264
+ "ĠRule": 179,
265
+ "1000": 180,
266
+ "fi": 181,
267
+ "10100": 182,
268
+ "as": 183,
269
+ "Ġ17": 184,
270
+ "ĠM": 185,
271
+ "Ġ1111": 186,
272
+ "Ver": 187,
273
+ "ch": 188,
274
+ "es": 189,
275
+ "ifi": 190,
276
+ "mp": 191,
277
+ "ting": 192,
278
+ "cation": 193,
279
+ "atch": 194,
280
+ "Verifi": 195,
281
+ "Verification": 196,
282
+ "..": 197,
283
+ "11011": 198,
284
+ "41": 199,
285
+ "and": 200,
286
+ "date": 201,
287
+ "idate": 202,
288
+ "ĠCand": 203,
289
+ "ĠCandidate": 204,
290
+ "58": 205,
291
+ "ating": 206,
292
+ "10011": 207,
293
+ "ction": 208,
294
+ "111000": 209,
295
+ "000000": 210,
296
+ "10110": 211,
297
+ "Ġ00000000": 212,
298
+ "is": 213,
299
+ "match": 214,
300
+ "ĠMis": 215,
301
+ "ĠMismatch": 216,
302
+ "111111": 217,
303
+ "Tas": 218,
304
+ "Task": 219,
305
+ "Ġ18": 220,
306
+ "Gre": 221,
307
+ "Pr": 222,
308
+ "ĠL": 223,
309
+ "ĠGre": 224,
310
+ "ĠPr": 225,
311
+ "edy": 226,
312
+ "ĠGreedy": 227,
313
+ "At": 228,
314
+ "Ex": 229,
315
+ "Tes": 230,
316
+ "ĠTes": 231,
317
+ "temp": 232,
318
+ "Attemp": 233,
319
+ "ĠTesting": 234,
320
+ "Attempt": 235,
321
+ "100000": 236,
322
+ "Ġs": 237,
323
+ "Ġ0000": 238,
324
+ "uction": 239,
325
+ "ĠSteps": 240,
326
+ "xt": 241,
327
+ "101111": 242,
328
+ "al": 243,
329
+ "cul": 244,
330
+ "eli": 245,
331
+ "ior": 246,
332
+ "keli": 247,
333
+ "ver": 248,
334
+ "cal": 249,
335
+ "culating": 250,
336
+ "iors": 251,
337
+ "kelihood": 252,
338
+ "Ġ19": 253,
339
+ "11111111": 254,
340
+ ").": 255
341
+ },
342
+ "merges": [
343
+ [
344
+ "0",
345
+ "0"
346
+ ],
347
+ [
348
+ "1",
349
+ "1"
350
+ ],
351
+ [
352
+ "-",
353
+ ">"
354
+ ],
355
+ [
356
+ "Ġ",
357
+ "|"
358
+ ],
359
+ [
360
+ ":",
361
+ "["
362
+ ],
363
+ [
364
+ "]",
365
+ "->"
366
+ ],
367
+ [
368
+ "1",
369
+ "0"
370
+ ],
371
+ [
372
+ "Ġ",
373
+ "1"
374
+ ],
375
+ [
376
+ "00",
377
+ "0"
378
+ ],
379
+ [
380
+ "Ġ",
381
+ "S"
382
+ ],
383
+ [
384
+ "'",
385
+ ","
386
+ ],
387
+ [
388
+ "Ġ",
389
+ "'"
390
+ ],
391
+ [
392
+ "t",
393
+ "e"
394
+ ],
395
+ [
396
+ "1",
397
+ "00"
398
+ ],
399
+ [
400
+ "Ċ",
401
+ "Ġ"
402
+ ],
403
+ [
404
+ "11",
405
+ "0"
406
+ ],
407
+ [
408
+ "te",
409
+ "p"
410
+ ],
411
+ [
412
+ "r",
413
+ "e"
414
+ ],
415
+ [
416
+ "11",
417
+ "1"
418
+ ],
419
+ [
420
+ "00",
421
+ "00"
422
+ ],
423
+ [
424
+ "0",
425
+ "11"
426
+ ],
427
+ [
428
+ "10",
429
+ "1"
430
+ ],
431
+ [
432
+ "00",
433
+ "1"
434
+ ],
435
+ [
436
+ "ĠS",
437
+ "tep"
438
+ ],
439
+ [
440
+ "0",
441
+ "10"
442
+ ],
443
+ [
444
+ "i",
445
+ "n"
446
+ ],
447
+ [
448
+ "i",
449
+ "m"
450
+ ],
451
+ [
452
+ "ĠS",
453
+ "im"
454
+ ],
455
+ [
456
+ "11",
457
+ "11"
458
+ ],
459
+ [
460
+ "Ġ",
461
+ "["
462
+ ],
463
+ [
464
+ "o",
465
+ "r"
466
+ ],
467
+ [
468
+ "Ġ",
469
+ "I"
470
+ ],
471
+ [
472
+ "h",
473
+ "o"
474
+ ],
475
+ [
476
+ "o",
477
+ "d"
478
+ ],
479
+ [
480
+ "ho",
481
+ "od"
482
+ ],
483
+ [
484
+ "Ġ",
485
+ "in"
486
+ ],
487
+ [
488
+ "Ġ",
489
+ "11"
490
+ ],
491
+ [
492
+ "Ġ",
493
+ "10"
494
+ ],
495
+ [
496
+ "n",
497
+ "e"
498
+ ],
499
+ [
500
+ "Ġ",
501
+ "ne"
502
+ ],
503
+ [
504
+ "'",
505
+ "]"
506
+ ],
507
+ [
508
+ "b",
509
+ "or"
510
+ ],
511
+ [
512
+ "d",
513
+ "x"
514
+ ],
515
+ [
516
+ "g",
517
+ "h"
518
+ ],
519
+ [
520
+ "i",
521
+ "re"
522
+ ],
523
+ [
524
+ "i",
525
+ "gh"
526
+ ],
527
+ [
528
+ "q",
529
+ "u"
530
+ ],
531
+ [
532
+ "Ġ",
533
+ "re"
534
+ ],
535
+ [
536
+ "Ġ[",
537
+ "'"
538
+ ],
539
+ [
540
+ "ĠI",
541
+ "dx"
542
+ ],
543
+ [
544
+ "Ġne",
545
+ "igh"
546
+ ],
547
+ [
548
+ "bor",
549
+ "hood"
550
+ ],
551
+ [
552
+ "ire",
553
+ "s"
554
+ ],
555
+ [
556
+ "qu",
557
+ "ires"
558
+ ],
559
+ [
560
+ "Ġre",
561
+ "quires"
562
+ ],
563
+ [
564
+ "Ġneigh",
565
+ "borhood"
566
+ ],
567
+ [
568
+ "S",
569
+ "tep"
570
+ ],
571
+ [
572
+ "u",
573
+ "t"
574
+ ],
575
+ [
576
+ "Ġ",
577
+ "3"
578
+ ],
579
+ [
580
+ "Ġ",
581
+ "2"
582
+ ],
583
+ [
584
+ "Ġ",
585
+ "4"
586
+ ],
587
+ [
588
+ "Ġ",
589
+ "5"
590
+ ],
591
+ [
592
+ "Ġ",
593
+ "6"
594
+ ],
595
+ [
596
+ "Ġ",
597
+ "9"
598
+ ],
599
+ [
600
+ "Ġ",
601
+ "8"
602
+ ],
603
+ [
604
+ "Ġ",
605
+ "7"
606
+ ],
607
+ [
608
+ "Ġ",
609
+ "0"
610
+ ],
611
+ [
612
+ "c",
613
+ "a"
614
+ ],
615
+ [
616
+ "Ġ1",
617
+ "2"
618
+ ],
619
+ [
620
+ "Ġ1",
621
+ "3"
622
+ ],
623
+ [
624
+ "10",
625
+ "10"
626
+ ],
627
+ [
628
+ "Ġ1",
629
+ "4"
630
+ ],
631
+ [
632
+ "O",
633
+ "ut"
634
+ ],
635
+ [
636
+ "p",
637
+ "ut"
638
+ ],
639
+ [
640
+ "Ġ",
641
+ "Out"
642
+ ],
643
+ [
644
+ "ĠS",
645
+ "ca"
646
+ ],
647
+ [
648
+ "ĠOut",
649
+ "put"
650
+ ],
651
+ [
652
+ "ĠSca",
653
+ "n"
654
+ ],
655
+ [
656
+ "Ġ",
657
+ "011"
658
+ ],
659
+ [
660
+ "in",
661
+ "g"
662
+ ],
663
+ [
664
+ "t",
665
+ "i"
666
+ ],
667
+ [
668
+ "11",
669
+ "00"
670
+ ],
671
+ [
672
+ "0000",
673
+ "0000"
674
+ ],
675
+ [
676
+ "Ġ1",
677
+ "5"
678
+ ],
679
+ [
680
+ "Ġ",
681
+ "000"
682
+ ],
683
+ [
684
+ "o",
685
+ "n"
686
+ ],
687
+ [
688
+ "Ġ",
689
+ "110"
690
+ ],
691
+ [
692
+ "u",
693
+ "l"
694
+ ],
695
+ [
696
+ "Ġ1",
697
+ "00"
698
+ ],
699
+ [
700
+ "a",
701
+ "t"
702
+ ],
703
+ [
704
+ "ti",
705
+ "on"
706
+ ],
707
+ [
708
+ "Ġ",
709
+ "010"
710
+ ],
711
+ [
712
+ "10",
713
+ "11"
714
+ ],
715
+ [
716
+ "Ġ",
717
+ "C"
718
+ ],
719
+ [
720
+ "Ġ",
721
+ "00"
722
+ ],
723
+ [
724
+ "000",
725
+ "11"
726
+ ],
727
+ [
728
+ "5",
729
+ "0"
730
+ ],
731
+ [
732
+ "e",
733
+ "r"
734
+ ],
735
+ [
736
+ "100",
737
+ "00"
738
+ ],
739
+ [
740
+ "n",
741
+ "d"
742
+ ],
743
+ [
744
+ "Ġ1",
745
+ "6"
746
+ ],
747
+ [
748
+ "ul",
749
+ "e"
750
+ ],
751
+ [
752
+ "Ġ",
753
+ "R"
754
+ ],
755
+ [
756
+ "e",
757
+ "d"
758
+ ],
759
+ [
760
+ "l",
761
+ "i"
762
+ ],
763
+ [
764
+ "Ġ",
765
+ "001"
766
+ ],
767
+ [
768
+ "Ġ",
769
+ "101"
770
+ ],
771
+ [
772
+ "a",
773
+ "te"
774
+ ],
775
+ [
776
+ "ĠR",
777
+ "ule"
778
+ ],
779
+ [
780
+ "1",
781
+ "000"
782
+ ],
783
+ [
784
+ "f",
785
+ "i"
786
+ ],
787
+ [
788
+ "10",
789
+ "100"
790
+ ],
791
+ [
792
+ "a",
793
+ "s"
794
+ ],
795
+ [
796
+ "Ġ1",
797
+ "7"
798
+ ],
799
+ [
800
+ "Ġ",
801
+ "M"
802
+ ],
803
+ [
804
+ "Ġ",
805
+ "1111"
806
+ ],
807
+ [
808
+ "V",
809
+ "er"
810
+ ],
811
+ [
812
+ "c",
813
+ "h"
814
+ ],
815
+ [
816
+ "e",
817
+ "s"
818
+ ],
819
+ [
820
+ "i",
821
+ "fi"
822
+ ],
823
+ [
824
+ "m",
825
+ "p"
826
+ ],
827
+ [
828
+ "t",
829
+ "ing"
830
+ ],
831
+ [
832
+ "ca",
833
+ "tion"
834
+ ],
835
+ [
836
+ "at",
837
+ "ch"
838
+ ],
839
+ [
840
+ "Ver",
841
+ "ifi"
842
+ ],
843
+ [
844
+ "Verifi",
845
+ "cation"
846
+ ],
847
+ [
848
+ ".",
849
+ "."
850
+ ],
851
+ [
852
+ "110",
853
+ "11"
854
+ ],
855
+ [
856
+ "4",
857
+ "1"
858
+ ],
859
+ [
860
+ "a",
861
+ "nd"
862
+ ],
863
+ [
864
+ "d",
865
+ "ate"
866
+ ],
867
+ [
868
+ "i",
869
+ "date"
870
+ ],
871
+ [
872
+ "ĠC",
873
+ "and"
874
+ ],
875
+ [
876
+ "ĠCand",
877
+ "idate"
878
+ ],
879
+ [
880
+ "5",
881
+ "8"
882
+ ],
883
+ [
884
+ "at",
885
+ "ing"
886
+ ],
887
+ [
888
+ "100",
889
+ "11"
890
+ ],
891
+ [
892
+ "c",
893
+ "tion"
894
+ ],
895
+ [
896
+ "111",
897
+ "000"
898
+ ],
899
+ [
900
+ "0000",
901
+ "00"
902
+ ],
903
+ [
904
+ "10",
905
+ "110"
906
+ ],
907
+ [
908
+ "Ġ",
909
+ "00000000"
910
+ ],
911
+ [
912
+ "i",
913
+ "s"
914
+ ],
915
+ [
916
+ "m",
917
+ "atch"
918
+ ],
919
+ [
920
+ "ĠM",
921
+ "is"
922
+ ],
923
+ [
924
+ "ĠMis",
925
+ "match"
926
+ ],
927
+ [
928
+ "1111",
929
+ "11"
930
+ ],
931
+ [
932
+ "T",
933
+ "as"
934
+ ],
935
+ [
936
+ "Tas",
937
+ "k"
938
+ ],
939
+ [
940
+ "Ġ1",
941
+ "8"
942
+ ],
943
+ [
944
+ "G",
945
+ "re"
946
+ ],
947
+ [
948
+ "P",
949
+ "r"
950
+ ],
951
+ [
952
+ "Ġ",
953
+ "L"
954
+ ],
955
+ [
956
+ "Ġ",
957
+ "Gre"
958
+ ],
959
+ [
960
+ "Ġ",
961
+ "Pr"
962
+ ],
963
+ [
964
+ "ed",
965
+ "y"
966
+ ],
967
+ [
968
+ "ĠGre",
969
+ "edy"
970
+ ],
971
+ [
972
+ "A",
973
+ "t"
974
+ ],
975
+ [
976
+ "E",
977
+ "x"
978
+ ],
979
+ [
980
+ "T",
981
+ "es"
982
+ ],
983
+ [
984
+ "Ġ",
985
+ "Tes"
986
+ ],
987
+ [
988
+ "te",
989
+ "mp"
990
+ ],
991
+ [
992
+ "At",
993
+ "temp"
994
+ ],
995
+ [
996
+ "ĠTes",
997
+ "ting"
998
+ ],
999
+ [
1000
+ "Attemp",
1001
+ "t"
1002
+ ],
1003
+ [
1004
+ "100",
1005
+ "000"
1006
+ ],
1007
+ [
1008
+ "Ġ",
1009
+ "s"
1010
+ ],
1011
+ [
1012
+ "Ġ",
1013
+ "0000"
1014
+ ],
1015
+ [
1016
+ "u",
1017
+ "ction"
1018
+ ],
1019
+ [
1020
+ "ĠStep",
1021
+ "s"
1022
+ ],
1023
+ [
1024
+ "x",
1025
+ "t"
1026
+ ],
1027
+ [
1028
+ "10",
1029
+ "1111"
1030
+ ],
1031
+ [
1032
+ "a",
1033
+ "l"
1034
+ ],
1035
+ [
1036
+ "c",
1037
+ "ul"
1038
+ ],
1039
+ [
1040
+ "e",
1041
+ "li"
1042
+ ],
1043
+ [
1044
+ "i",
1045
+ "or"
1046
+ ],
1047
+ [
1048
+ "k",
1049
+ "eli"
1050
+ ],
1051
+ [
1052
+ "v",
1053
+ "er"
1054
+ ],
1055
+ [
1056
+ "ca",
1057
+ "l"
1058
+ ],
1059
+ [
1060
+ "cul",
1061
+ "ating"
1062
+ ],
1063
+ [
1064
+ "ior",
1065
+ "s"
1066
+ ],
1067
+ [
1068
+ "keli",
1069
+ "hood"
1070
+ ],
1071
+ [
1072
+ "Ġ1",
1073
+ "9"
1074
+ ],
1075
+ [
1076
+ "1111",
1077
+ "1111"
1078
+ ],
1079
+ [
1080
+ ")",
1081
+ "."
1082
+ ]
1083
+ ]
1084
+ }
1085
+ }
checkpoint-14000/tokenizer_config.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<|startoftext|>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<|endoftext|>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "<|unknown|>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "<|pad|>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ },
35
+ "4": {
36
+ "content": "<think>",
37
+ "lstrip": false,
38
+ "normalized": false,
39
+ "rstrip": false,
40
+ "single_word": false,
41
+ "special": true
42
+ },
43
+ "5": {
44
+ "content": "</think>",
45
+ "lstrip": false,
46
+ "normalized": false,
47
+ "rstrip": false,
48
+ "single_word": false,
49
+ "special": true
50
+ }
51
+ },
52
+ "additional_special_tokens": [
53
+ "<think>",
54
+ "</think>"
55
+ ],
56
+ "bos_token": "<|startoftext|>",
57
+ "clean_up_tokenization_spaces": false,
58
+ "eos_token": "<|endoftext|>",
59
+ "extra_special_tokens": {},
60
+ "model_max_length": 1000000000000000019884624838656,
61
+ "pad_token": "<|pad|>",
62
+ "tokenizer_class": "PreTrainedTokenizerFast",
63
+ "unk_token": "<|unknown|>"
64
+ }
checkpoint-14000/trainer_state.json ADDED
@@ -0,0 +1,1994 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 2.6877219928962273,
6
+ "eval_steps": 500,
7
+ "global_step": 14000,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.009599692809830085,
14
+ "grad_norm": 7.351896286010742,
15
+ "learning_rate": 1.47e-05,
16
+ "loss": 19.9984,
17
+ "step": 50
18
+ },
19
+ {
20
+ "epoch": 0.01919938561966017,
21
+ "grad_norm": 41.9318733215332,
22
+ "learning_rate": 2.97e-05,
23
+ "loss": 15.992,
24
+ "step": 100
25
+ },
26
+ {
27
+ "epoch": 0.028799078429490255,
28
+ "grad_norm": 41.05648422241211,
29
+ "learning_rate": 4.4699999999999996e-05,
30
+ "loss": 12.7699,
31
+ "step": 150
32
+ },
33
+ {
34
+ "epoch": 0.03839877123932034,
35
+ "grad_norm": 30.36406135559082,
36
+ "learning_rate": 5.97e-05,
37
+ "loss": 10.156,
38
+ "step": 200
39
+ },
40
+ {
41
+ "epoch": 0.047998464049150424,
42
+ "grad_norm": 26.620258331298828,
43
+ "learning_rate": 7.47e-05,
44
+ "loss": 7.7426,
45
+ "step": 250
46
+ },
47
+ {
48
+ "epoch": 0.05759815685898051,
49
+ "grad_norm": 17.737770080566406,
50
+ "learning_rate": 8.969999999999998e-05,
51
+ "loss": 5.743,
52
+ "step": 300
53
+ },
54
+ {
55
+ "epoch": 0.0671978496688106,
56
+ "grad_norm": 16.273975372314453,
57
+ "learning_rate": 0.00010469999999999998,
58
+ "loss": 4.1183,
59
+ "step": 350
60
+ },
61
+ {
62
+ "epoch": 0.07679754247864068,
63
+ "grad_norm": 7.628762722015381,
64
+ "learning_rate": 0.0001197,
65
+ "loss": 3.0495,
66
+ "step": 400
67
+ },
68
+ {
69
+ "epoch": 0.08639723528847076,
70
+ "grad_norm": 11.860695838928223,
71
+ "learning_rate": 0.0001347,
72
+ "loss": 2.3835,
73
+ "step": 450
74
+ },
75
+ {
76
+ "epoch": 0.09599692809830085,
77
+ "grad_norm": 9.723978042602539,
78
+ "learning_rate": 0.00014969999999999998,
79
+ "loss": 2.0257,
80
+ "step": 500
81
+ },
82
+ {
83
+ "epoch": 0.10559662090813093,
84
+ "grad_norm": 8.171798706054688,
85
+ "learning_rate": 0.0001647,
86
+ "loss": 1.7721,
87
+ "step": 550
88
+ },
89
+ {
90
+ "epoch": 0.11519631371796102,
91
+ "grad_norm": 5.052175521850586,
92
+ "learning_rate": 0.00017969999999999998,
93
+ "loss": 1.6081,
94
+ "step": 600
95
+ },
96
+ {
97
+ "epoch": 0.12479600652779112,
98
+ "grad_norm": 6.628243446350098,
99
+ "learning_rate": 0.0001947,
100
+ "loss": 1.4632,
101
+ "step": 650
102
+ },
103
+ {
104
+ "epoch": 0.1343956993376212,
105
+ "grad_norm": 5.753493309020996,
106
+ "learning_rate": 0.00020969999999999997,
107
+ "loss": 1.3574,
108
+ "step": 700
109
+ },
110
+ {
111
+ "epoch": 0.14399539214745127,
112
+ "grad_norm": 3.7359797954559326,
113
+ "learning_rate": 0.0002247,
114
+ "loss": 1.2551,
115
+ "step": 750
116
+ },
117
+ {
118
+ "epoch": 0.15359508495728136,
119
+ "grad_norm": 8.183115005493164,
120
+ "learning_rate": 0.0002397,
121
+ "loss": 1.1362,
122
+ "step": 800
123
+ },
124
+ {
125
+ "epoch": 0.16319477776711144,
126
+ "grad_norm": 5.23441219329834,
127
+ "learning_rate": 0.00025469999999999996,
128
+ "loss": 1.0032,
129
+ "step": 850
130
+ },
131
+ {
132
+ "epoch": 0.17279447057694153,
133
+ "grad_norm": 3.2111752033233643,
134
+ "learning_rate": 0.0002697,
135
+ "loss": 0.9582,
136
+ "step": 900
137
+ },
138
+ {
139
+ "epoch": 0.1823941633867716,
140
+ "grad_norm": 3.7051522731781006,
141
+ "learning_rate": 0.0002847,
142
+ "loss": 0.8247,
143
+ "step": 950
144
+ },
145
+ {
146
+ "epoch": 0.1919938561966017,
147
+ "grad_norm": 2.5947470664978027,
148
+ "learning_rate": 0.00029969999999999997,
149
+ "loss": 0.68,
150
+ "step": 1000
151
+ },
152
+ {
153
+ "epoch": 0.20159354900643178,
154
+ "grad_norm": 2.0631327629089355,
155
+ "learning_rate": 0.00029899500922950706,
156
+ "loss": 0.5732,
157
+ "step": 1050
158
+ },
159
+ {
160
+ "epoch": 0.21119324181626187,
161
+ "grad_norm": 1.6905947923660278,
162
+ "learning_rate": 0.0002979695084432898,
163
+ "loss": 0.5127,
164
+ "step": 1100
165
+ },
166
+ {
167
+ "epoch": 0.22079293462609195,
168
+ "grad_norm": 1.6837217807769775,
169
+ "learning_rate": 0.0002969440076570725,
170
+ "loss": 0.4481,
171
+ "step": 1150
172
+ },
173
+ {
174
+ "epoch": 0.23039262743592204,
175
+ "grad_norm": 1.147798776626587,
176
+ "learning_rate": 0.00029591850687085523,
177
+ "loss": 0.4135,
178
+ "step": 1200
179
+ },
180
+ {
181
+ "epoch": 0.23999232024575212,
182
+ "grad_norm": 2.4957432746887207,
183
+ "learning_rate": 0.00029489300608463796,
184
+ "loss": 0.3848,
185
+ "step": 1250
186
+ },
187
+ {
188
+ "epoch": 0.24959201305558223,
189
+ "grad_norm": 1.402390480041504,
190
+ "learning_rate": 0.0002938675052984207,
191
+ "loss": 0.3413,
192
+ "step": 1300
193
+ },
194
+ {
195
+ "epoch": 0.2591917058654123,
196
+ "grad_norm": 2.157576084136963,
197
+ "learning_rate": 0.0002928420045122034,
198
+ "loss": 0.3,
199
+ "step": 1350
200
+ },
201
+ {
202
+ "epoch": 0.2687913986752424,
203
+ "grad_norm": 1.074357032775879,
204
+ "learning_rate": 0.0002918165037259862,
205
+ "loss": 0.2714,
206
+ "step": 1400
207
+ },
208
+ {
209
+ "epoch": 0.27839109148507246,
210
+ "grad_norm": 0.9258896112442017,
211
+ "learning_rate": 0.0002907910029397689,
212
+ "loss": 0.2665,
213
+ "step": 1450
214
+ },
215
+ {
216
+ "epoch": 0.28799078429490255,
217
+ "grad_norm": 1.9370087385177612,
218
+ "learning_rate": 0.0002897655021535516,
219
+ "loss": 0.2627,
220
+ "step": 1500
221
+ },
222
+ {
223
+ "epoch": 0.29759047710473263,
224
+ "grad_norm": 1.3033190965652466,
225
+ "learning_rate": 0.00028874000136733436,
226
+ "loss": 0.2345,
227
+ "step": 1550
228
+ },
229
+ {
230
+ "epoch": 0.3071901699145627,
231
+ "grad_norm": 1.0974582433700562,
232
+ "learning_rate": 0.0002877145005811171,
233
+ "loss": 0.2414,
234
+ "step": 1600
235
+ },
236
+ {
237
+ "epoch": 0.3167898627243928,
238
+ "grad_norm": 0.8440233469009399,
239
+ "learning_rate": 0.0002866889997948998,
240
+ "loss": 0.2372,
241
+ "step": 1650
242
+ },
243
+ {
244
+ "epoch": 0.3263895555342229,
245
+ "grad_norm": 1.1249887943267822,
246
+ "learning_rate": 0.00028566349900868254,
247
+ "loss": 0.2265,
248
+ "step": 1700
249
+ },
250
+ {
251
+ "epoch": 0.33598924834405297,
252
+ "grad_norm": 0.7356661558151245,
253
+ "learning_rate": 0.00028463799822246526,
254
+ "loss": 0.2182,
255
+ "step": 1750
256
+ },
257
+ {
258
+ "epoch": 0.34558894115388306,
259
+ "grad_norm": 0.7215508818626404,
260
+ "learning_rate": 0.00028361249743624804,
261
+ "loss": 0.2077,
262
+ "step": 1800
263
+ },
264
+ {
265
+ "epoch": 0.35518863396371314,
266
+ "grad_norm": 0.8716081976890564,
267
+ "learning_rate": 0.0002825869966500307,
268
+ "loss": 0.206,
269
+ "step": 1850
270
+ },
271
+ {
272
+ "epoch": 0.3647883267735432,
273
+ "grad_norm": 0.5157195329666138,
274
+ "learning_rate": 0.00028156149586381344,
275
+ "loss": 0.2003,
276
+ "step": 1900
277
+ },
278
+ {
279
+ "epoch": 0.3743880195833733,
280
+ "grad_norm": 0.47487232089042664,
281
+ "learning_rate": 0.0002805359950775962,
282
+ "loss": 0.1955,
283
+ "step": 1950
284
+ },
285
+ {
286
+ "epoch": 0.3839877123932034,
287
+ "grad_norm": 0.5665110349655151,
288
+ "learning_rate": 0.00027951049429137894,
289
+ "loss": 0.1884,
290
+ "step": 2000
291
+ },
292
+ {
293
+ "epoch": 0.3935874052030335,
294
+ "grad_norm": 0.772770345211029,
295
+ "learning_rate": 0.00027848499350516167,
296
+ "loss": 0.1893,
297
+ "step": 2050
298
+ },
299
+ {
300
+ "epoch": 0.40318709801286357,
301
+ "grad_norm": 0.5221526622772217,
302
+ "learning_rate": 0.0002774594927189444,
303
+ "loss": 0.1826,
304
+ "step": 2100
305
+ },
306
+ {
307
+ "epoch": 0.41278679082269365,
308
+ "grad_norm": 0.5089606642723083,
309
+ "learning_rate": 0.0002764339919327271,
310
+ "loss": 0.1707,
311
+ "step": 2150
312
+ },
313
+ {
314
+ "epoch": 0.42238648363252373,
315
+ "grad_norm": 0.42394664883613586,
316
+ "learning_rate": 0.00027540849114650985,
317
+ "loss": 0.1672,
318
+ "step": 2200
319
+ },
320
+ {
321
+ "epoch": 0.4319861764423538,
322
+ "grad_norm": 0.40413081645965576,
323
+ "learning_rate": 0.00027438299036029257,
324
+ "loss": 0.1596,
325
+ "step": 2250
326
+ },
327
+ {
328
+ "epoch": 0.4415858692521839,
329
+ "grad_norm": 0.3184243142604828,
330
+ "learning_rate": 0.0002733574895740753,
331
+ "loss": 0.153,
332
+ "step": 2300
333
+ },
334
+ {
335
+ "epoch": 0.451185562062014,
336
+ "grad_norm": 0.43532413244247437,
337
+ "learning_rate": 0.0002723319887878581,
338
+ "loss": 0.1512,
339
+ "step": 2350
340
+ },
341
+ {
342
+ "epoch": 0.4607852548718441,
343
+ "grad_norm": 0.7987850904464722,
344
+ "learning_rate": 0.0002713064880016408,
345
+ "loss": 0.1395,
346
+ "step": 2400
347
+ },
348
+ {
349
+ "epoch": 0.47038494768167416,
350
+ "grad_norm": 0.37470191717147827,
351
+ "learning_rate": 0.00027028098721542347,
352
+ "loss": 0.1426,
353
+ "step": 2450
354
+ },
355
+ {
356
+ "epoch": 0.47998464049150424,
357
+ "grad_norm": 0.32686635851860046,
358
+ "learning_rate": 0.00026925548642920625,
359
+ "loss": 0.1364,
360
+ "step": 2500
361
+ },
362
+ {
363
+ "epoch": 0.4895843333013344,
364
+ "grad_norm": 0.6824387907981873,
365
+ "learning_rate": 0.000268229985642989,
366
+ "loss": 0.1414,
367
+ "step": 2550
368
+ },
369
+ {
370
+ "epoch": 0.49918402611116447,
371
+ "grad_norm": 0.3873808979988098,
372
+ "learning_rate": 0.0002672044848567717,
373
+ "loss": 0.1413,
374
+ "step": 2600
375
+ },
376
+ {
377
+ "epoch": 0.5087837189209945,
378
+ "grad_norm": 0.30228692293167114,
379
+ "learning_rate": 0.0002661789840705544,
380
+ "loss": 0.1355,
381
+ "step": 2650
382
+ },
383
+ {
384
+ "epoch": 0.5183834117308246,
385
+ "grad_norm": 0.2715003788471222,
386
+ "learning_rate": 0.00026515348328433715,
387
+ "loss": 0.1302,
388
+ "step": 2700
389
+ },
390
+ {
391
+ "epoch": 0.5279831045406547,
392
+ "grad_norm": 0.35757285356521606,
393
+ "learning_rate": 0.00026412798249811993,
394
+ "loss": 0.1317,
395
+ "step": 2750
396
+ },
397
+ {
398
+ "epoch": 0.5375827973504848,
399
+ "grad_norm": 0.42059722542762756,
400
+ "learning_rate": 0.0002631024817119026,
401
+ "loss": 0.1289,
402
+ "step": 2800
403
+ },
404
+ {
405
+ "epoch": 0.5471824901603148,
406
+ "grad_norm": 0.680482029914856,
407
+ "learning_rate": 0.0002620769809256853,
408
+ "loss": 0.1312,
409
+ "step": 2850
410
+ },
411
+ {
412
+ "epoch": 0.5567821829701449,
413
+ "grad_norm": 0.7014706134796143,
414
+ "learning_rate": 0.0002610514801394681,
415
+ "loss": 0.1343,
416
+ "step": 2900
417
+ },
418
+ {
419
+ "epoch": 0.566381875779975,
420
+ "grad_norm": 0.4586208164691925,
421
+ "learning_rate": 0.00026002597935325083,
422
+ "loss": 0.1266,
423
+ "step": 2950
424
+ },
425
+ {
426
+ "epoch": 0.5759815685898051,
427
+ "grad_norm": 0.37353283166885376,
428
+ "learning_rate": 0.00025900047856703356,
429
+ "loss": 0.1328,
430
+ "step": 3000
431
+ },
432
+ {
433
+ "epoch": 0.5855812613996352,
434
+ "grad_norm": 0.30957943201065063,
435
+ "learning_rate": 0.0002579749777808163,
436
+ "loss": 0.1218,
437
+ "step": 3050
438
+ },
439
+ {
440
+ "epoch": 0.5951809542094653,
441
+ "grad_norm": 0.35607343912124634,
442
+ "learning_rate": 0.000256949476994599,
443
+ "loss": 0.1235,
444
+ "step": 3100
445
+ },
446
+ {
447
+ "epoch": 0.6047806470192953,
448
+ "grad_norm": 0.3144589066505432,
449
+ "learning_rate": 0.00025592397620838173,
450
+ "loss": 0.1248,
451
+ "step": 3150
452
+ },
453
+ {
454
+ "epoch": 0.6143803398291254,
455
+ "grad_norm": 0.5591529607772827,
456
+ "learning_rate": 0.00025489847542216446,
457
+ "loss": 0.125,
458
+ "step": 3200
459
+ },
460
+ {
461
+ "epoch": 0.6239800326389555,
462
+ "grad_norm": 0.4985520839691162,
463
+ "learning_rate": 0.0002538729746359472,
464
+ "loss": 0.1328,
465
+ "step": 3250
466
+ },
467
+ {
468
+ "epoch": 0.6335797254487856,
469
+ "grad_norm": 0.3131736218929291,
470
+ "learning_rate": 0.00025284747384972996,
471
+ "loss": 0.1376,
472
+ "step": 3300
473
+ },
474
+ {
475
+ "epoch": 0.6431794182586157,
476
+ "grad_norm": 1.0179088115692139,
477
+ "learning_rate": 0.0002518219730635127,
478
+ "loss": 0.1393,
479
+ "step": 3350
480
+ },
481
+ {
482
+ "epoch": 0.6527791110684458,
483
+ "grad_norm": 0.40214893221855164,
484
+ "learning_rate": 0.00025079647227729536,
485
+ "loss": 0.1383,
486
+ "step": 3400
487
+ },
488
+ {
489
+ "epoch": 0.6623788038782759,
490
+ "grad_norm": 0.41477128863334656,
491
+ "learning_rate": 0.00024977097149107814,
492
+ "loss": 0.1252,
493
+ "step": 3450
494
+ },
495
+ {
496
+ "epoch": 0.6719784966881059,
497
+ "grad_norm": 0.3146192133426666,
498
+ "learning_rate": 0.00024874547070486086,
499
+ "loss": 0.1283,
500
+ "step": 3500
501
+ },
502
+ {
503
+ "epoch": 0.681578189497936,
504
+ "grad_norm": 0.34901753067970276,
505
+ "learning_rate": 0.0002477199699186436,
506
+ "loss": 0.1185,
507
+ "step": 3550
508
+ },
509
+ {
510
+ "epoch": 0.6911778823077661,
511
+ "grad_norm": 0.4564303159713745,
512
+ "learning_rate": 0.0002466944691324263,
513
+ "loss": 0.1176,
514
+ "step": 3600
515
+ },
516
+ {
517
+ "epoch": 0.7007775751175962,
518
+ "grad_norm": 0.20335577428340912,
519
+ "learning_rate": 0.00024566896834620904,
520
+ "loss": 0.1184,
521
+ "step": 3650
522
+ },
523
+ {
524
+ "epoch": 0.7103772679274263,
525
+ "grad_norm": 0.16178852319717407,
526
+ "learning_rate": 0.00024464346755999176,
527
+ "loss": 0.1136,
528
+ "step": 3700
529
+ },
530
+ {
531
+ "epoch": 0.7199769607372564,
532
+ "grad_norm": 0.270074725151062,
533
+ "learning_rate": 0.00024361796677377452,
534
+ "loss": 0.116,
535
+ "step": 3750
536
+ },
537
+ {
538
+ "epoch": 0.7295766535470865,
539
+ "grad_norm": 0.6328276991844177,
540
+ "learning_rate": 0.00024259246598755724,
541
+ "loss": 0.1152,
542
+ "step": 3800
543
+ },
544
+ {
545
+ "epoch": 0.7391763463569165,
546
+ "grad_norm": 0.290368914604187,
547
+ "learning_rate": 0.00024156696520133997,
548
+ "loss": 0.1145,
549
+ "step": 3850
550
+ },
551
+ {
552
+ "epoch": 0.7487760391667466,
553
+ "grad_norm": 0.3527914881706238,
554
+ "learning_rate": 0.0002405414644151227,
555
+ "loss": 0.1145,
556
+ "step": 3900
557
+ },
558
+ {
559
+ "epoch": 0.7583757319765767,
560
+ "grad_norm": 0.2402893602848053,
561
+ "learning_rate": 0.00023951596362890542,
562
+ "loss": 0.1159,
563
+ "step": 3950
564
+ },
565
+ {
566
+ "epoch": 0.7679754247864068,
567
+ "grad_norm": 0.19898560643196106,
568
+ "learning_rate": 0.00023849046284268817,
569
+ "loss": 0.1141,
570
+ "step": 4000
571
+ },
572
+ {
573
+ "epoch": 0.7775751175962369,
574
+ "grad_norm": 0.19476300477981567,
575
+ "learning_rate": 0.0002374649620564709,
576
+ "loss": 0.1158,
577
+ "step": 4050
578
+ },
579
+ {
580
+ "epoch": 0.787174810406067,
581
+ "grad_norm": 0.3251754343509674,
582
+ "learning_rate": 0.0002364394612702536,
583
+ "loss": 0.1137,
584
+ "step": 4100
585
+ },
586
+ {
587
+ "epoch": 0.796774503215897,
588
+ "grad_norm": 0.787211537361145,
589
+ "learning_rate": 0.00023541396048403634,
590
+ "loss": 0.1199,
591
+ "step": 4150
592
+ },
593
+ {
594
+ "epoch": 0.8063741960257271,
595
+ "grad_norm": 0.27289190888404846,
596
+ "learning_rate": 0.00023438845969781907,
597
+ "loss": 0.1152,
598
+ "step": 4200
599
+ },
600
+ {
601
+ "epoch": 0.8159738888355572,
602
+ "grad_norm": 0.17066220939159393,
603
+ "learning_rate": 0.00023336295891160182,
604
+ "loss": 0.1098,
605
+ "step": 4250
606
+ },
607
+ {
608
+ "epoch": 0.8255735816453873,
609
+ "grad_norm": 0.5239670276641846,
610
+ "learning_rate": 0.00023233745812538455,
611
+ "loss": 0.109,
612
+ "step": 4300
613
+ },
614
+ {
615
+ "epoch": 0.8351732744552174,
616
+ "grad_norm": 0.5953475832939148,
617
+ "learning_rate": 0.00023131195733916727,
618
+ "loss": 0.1174,
619
+ "step": 4350
620
+ },
621
+ {
622
+ "epoch": 0.8447729672650475,
623
+ "grad_norm": 0.3444459140300751,
624
+ "learning_rate": 0.00023028645655295002,
625
+ "loss": 0.1156,
626
+ "step": 4400
627
+ },
628
+ {
629
+ "epoch": 0.8543726600748776,
630
+ "grad_norm": 0.1797035187482834,
631
+ "learning_rate": 0.00022926095576673272,
632
+ "loss": 0.1117,
633
+ "step": 4450
634
+ },
635
+ {
636
+ "epoch": 0.8639723528847076,
637
+ "grad_norm": 0.1553630381822586,
638
+ "learning_rate": 0.00022823545498051545,
639
+ "loss": 0.1099,
640
+ "step": 4500
641
+ },
642
+ {
643
+ "epoch": 0.8735720456945377,
644
+ "grad_norm": 0.2846544086933136,
645
+ "learning_rate": 0.0002272099541942982,
646
+ "loss": 0.1118,
647
+ "step": 4550
648
+ },
649
+ {
650
+ "epoch": 0.8831717385043678,
651
+ "grad_norm": 0.18632030487060547,
652
+ "learning_rate": 0.00022618445340808093,
653
+ "loss": 0.1108,
654
+ "step": 4600
655
+ },
656
+ {
657
+ "epoch": 0.8927714313141979,
658
+ "grad_norm": 0.3297656178474426,
659
+ "learning_rate": 0.00022515895262186365,
660
+ "loss": 0.1159,
661
+ "step": 4650
662
+ },
663
+ {
664
+ "epoch": 0.902371124124028,
665
+ "grad_norm": 0.5725330710411072,
666
+ "learning_rate": 0.0002241334518356464,
667
+ "loss": 0.1256,
668
+ "step": 4700
669
+ },
670
+ {
671
+ "epoch": 0.9119708169338581,
672
+ "grad_norm": 0.4595456123352051,
673
+ "learning_rate": 0.0002231079510494291,
674
+ "loss": 0.1293,
675
+ "step": 4750
676
+ },
677
+ {
678
+ "epoch": 0.9215705097436881,
679
+ "grad_norm": 0.2964916527271271,
680
+ "learning_rate": 0.00022208245026321185,
681
+ "loss": 0.1123,
682
+ "step": 4800
683
+ },
684
+ {
685
+ "epoch": 0.9311702025535182,
686
+ "grad_norm": 0.18435810506343842,
687
+ "learning_rate": 0.00022105694947699458,
688
+ "loss": 0.1141,
689
+ "step": 4850
690
+ },
691
+ {
692
+ "epoch": 0.9407698953633483,
693
+ "grad_norm": 0.18365661799907684,
694
+ "learning_rate": 0.0002200314486907773,
695
+ "loss": 0.109,
696
+ "step": 4900
697
+ },
698
+ {
699
+ "epoch": 0.9503695881731784,
700
+ "grad_norm": 0.19481471180915833,
701
+ "learning_rate": 0.00021900594790456006,
702
+ "loss": 0.1095,
703
+ "step": 4950
704
+ },
705
+ {
706
+ "epoch": 0.9599692809830085,
707
+ "grad_norm": 0.15485212206840515,
708
+ "learning_rate": 0.00021798044711834278,
709
+ "loss": 0.1076,
710
+ "step": 5000
711
+ },
712
+ {
713
+ "epoch": 0.9695689737928387,
714
+ "grad_norm": 0.15106704831123352,
715
+ "learning_rate": 0.00021695494633212548,
716
+ "loss": 0.1093,
717
+ "step": 5050
718
+ },
719
+ {
720
+ "epoch": 0.9791686666026688,
721
+ "grad_norm": 0.1416679173707962,
722
+ "learning_rate": 0.00021592944554590823,
723
+ "loss": 0.1072,
724
+ "step": 5100
725
+ },
726
+ {
727
+ "epoch": 0.9887683594124989,
728
+ "grad_norm": 0.2634902596473694,
729
+ "learning_rate": 0.00021490394475969096,
730
+ "loss": 0.1106,
731
+ "step": 5150
732
+ },
733
+ {
734
+ "epoch": 0.9983680522223289,
735
+ "grad_norm": 0.2642424404621124,
736
+ "learning_rate": 0.00021387844397347368,
737
+ "loss": 0.1088,
738
+ "step": 5200
739
+ },
740
+ {
741
+ "epoch": 1.0078717481040607,
742
+ "grad_norm": 0.16377754509449005,
743
+ "learning_rate": 0.00021285294318725643,
744
+ "loss": 0.1043,
745
+ "step": 5250
746
+ },
747
+ {
748
+ "epoch": 1.0174714409138907,
749
+ "grad_norm": 0.2642650008201599,
750
+ "learning_rate": 0.00021182744240103916,
751
+ "loss": 0.1055,
752
+ "step": 5300
753
+ },
754
+ {
755
+ "epoch": 1.0270711337237208,
756
+ "grad_norm": 0.14765985310077667,
757
+ "learning_rate": 0.0002108019416148219,
758
+ "loss": 0.1065,
759
+ "step": 5350
760
+ },
761
+ {
762
+ "epoch": 1.036670826533551,
763
+ "grad_norm": 0.1779513657093048,
764
+ "learning_rate": 0.0002097764408286046,
765
+ "loss": 0.1062,
766
+ "step": 5400
767
+ },
768
+ {
769
+ "epoch": 1.046270519343381,
770
+ "grad_norm": 0.7090527415275574,
771
+ "learning_rate": 0.00020875094004238733,
772
+ "loss": 0.1098,
773
+ "step": 5450
774
+ },
775
+ {
776
+ "epoch": 1.055870212153211,
777
+ "grad_norm": 0.5224199891090393,
778
+ "learning_rate": 0.0002077254392561701,
779
+ "loss": 0.138,
780
+ "step": 5500
781
+ },
782
+ {
783
+ "epoch": 1.0654699049630412,
784
+ "grad_norm": 0.2672184109687805,
785
+ "learning_rate": 0.0002066999384699528,
786
+ "loss": 0.1349,
787
+ "step": 5550
788
+ },
789
+ {
790
+ "epoch": 1.0750695977728713,
791
+ "grad_norm": 0.19684670865535736,
792
+ "learning_rate": 0.00020567443768373554,
793
+ "loss": 0.109,
794
+ "step": 5600
795
+ },
796
+ {
797
+ "epoch": 1.0846692905827013,
798
+ "grad_norm": 0.11916083097457886,
799
+ "learning_rate": 0.0002046489368975183,
800
+ "loss": 0.1058,
801
+ "step": 5650
802
+ },
803
+ {
804
+ "epoch": 1.0942689833925314,
805
+ "grad_norm": 0.17690648138523102,
806
+ "learning_rate": 0.000203623436111301,
807
+ "loss": 0.1061,
808
+ "step": 5700
809
+ },
810
+ {
811
+ "epoch": 1.1038686762023615,
812
+ "grad_norm": 0.20282995700836182,
813
+ "learning_rate": 0.0002025979353250837,
814
+ "loss": 0.1063,
815
+ "step": 5750
816
+ },
817
+ {
818
+ "epoch": 1.1134683690121916,
819
+ "grad_norm": 0.17138956487178802,
820
+ "learning_rate": 0.00020157243453886647,
821
+ "loss": 0.1069,
822
+ "step": 5800
823
+ },
824
+ {
825
+ "epoch": 1.1230680618220217,
826
+ "grad_norm": 0.1042153611779213,
827
+ "learning_rate": 0.0002005469337526492,
828
+ "loss": 0.1036,
829
+ "step": 5850
830
+ },
831
+ {
832
+ "epoch": 1.1326677546318518,
833
+ "grad_norm": 0.15628878772258759,
834
+ "learning_rate": 0.00019952143296643194,
835
+ "loss": 0.104,
836
+ "step": 5900
837
+ },
838
+ {
839
+ "epoch": 1.1422674474416818,
840
+ "grad_norm": 0.19753845036029816,
841
+ "learning_rate": 0.00019849593218021467,
842
+ "loss": 0.107,
843
+ "step": 5950
844
+ },
845
+ {
846
+ "epoch": 1.151867140251512,
847
+ "grad_norm": 0.3391304612159729,
848
+ "learning_rate": 0.00019747043139399737,
849
+ "loss": 0.1097,
850
+ "step": 6000
851
+ },
852
+ {
853
+ "epoch": 1.161466833061342,
854
+ "grad_norm": 0.16991937160491943,
855
+ "learning_rate": 0.00019644493060778012,
856
+ "loss": 0.1051,
857
+ "step": 6050
858
+ },
859
+ {
860
+ "epoch": 1.171066525871172,
861
+ "grad_norm": 0.2567026913166046,
862
+ "learning_rate": 0.00019541942982156284,
863
+ "loss": 0.1049,
864
+ "step": 6100
865
+ },
866
+ {
867
+ "epoch": 1.1806662186810022,
868
+ "grad_norm": 0.1503693014383316,
869
+ "learning_rate": 0.00019439392903534557,
870
+ "loss": 0.1067,
871
+ "step": 6150
872
+ },
873
+ {
874
+ "epoch": 1.1902659114908323,
875
+ "grad_norm": 0.22622846066951752,
876
+ "learning_rate": 0.00019336842824912832,
877
+ "loss": 0.1059,
878
+ "step": 6200
879
+ },
880
+ {
881
+ "epoch": 1.1998656043006624,
882
+ "grad_norm": 0.15262149274349213,
883
+ "learning_rate": 0.00019234292746291105,
884
+ "loss": 0.1056,
885
+ "step": 6250
886
+ },
887
+ {
888
+ "epoch": 1.2094652971104924,
889
+ "grad_norm": 0.313942015171051,
890
+ "learning_rate": 0.00019131742667669374,
891
+ "loss": 0.1041,
892
+ "step": 6300
893
+ },
894
+ {
895
+ "epoch": 1.2190649899203225,
896
+ "grad_norm": 0.17761795222759247,
897
+ "learning_rate": 0.0001902919258904765,
898
+ "loss": 0.1046,
899
+ "step": 6350
900
+ },
901
+ {
902
+ "epoch": 1.2286646827301526,
903
+ "grad_norm": 0.20730365812778473,
904
+ "learning_rate": 0.00018926642510425922,
905
+ "loss": 0.1041,
906
+ "step": 6400
907
+ },
908
+ {
909
+ "epoch": 1.2382643755399827,
910
+ "grad_norm": 0.15528739988803864,
911
+ "learning_rate": 0.00018824092431804197,
912
+ "loss": 0.1053,
913
+ "step": 6450
914
+ },
915
+ {
916
+ "epoch": 1.2478640683498128,
917
+ "grad_norm": 0.19049355387687683,
918
+ "learning_rate": 0.0001872154235318247,
919
+ "loss": 0.1043,
920
+ "step": 6500
921
+ },
922
+ {
923
+ "epoch": 1.2574637611596429,
924
+ "grad_norm": 0.1925603300333023,
925
+ "learning_rate": 0.00018618992274560742,
926
+ "loss": 0.1043,
927
+ "step": 6550
928
+ },
929
+ {
930
+ "epoch": 1.267063453969473,
931
+ "grad_norm": 0.10907876491546631,
932
+ "learning_rate": 0.00018516442195939018,
933
+ "loss": 0.1013,
934
+ "step": 6600
935
+ },
936
+ {
937
+ "epoch": 1.276663146779303,
938
+ "grad_norm": 0.1089431643486023,
939
+ "learning_rate": 0.00018413892117317287,
940
+ "loss": 0.1028,
941
+ "step": 6650
942
+ },
943
+ {
944
+ "epoch": 1.2862628395891331,
945
+ "grad_norm": 0.13383787870407104,
946
+ "learning_rate": 0.0001831134203869556,
947
+ "loss": 0.1036,
948
+ "step": 6700
949
+ },
950
+ {
951
+ "epoch": 1.2958625323989632,
952
+ "grad_norm": 0.3068512976169586,
953
+ "learning_rate": 0.00018208791960073835,
954
+ "loss": 0.1024,
955
+ "step": 6750
956
+ },
957
+ {
958
+ "epoch": 1.3054622252087933,
959
+ "grad_norm": 0.25988587737083435,
960
+ "learning_rate": 0.00018106241881452108,
961
+ "loss": 0.1053,
962
+ "step": 6800
963
+ },
964
+ {
965
+ "epoch": 1.3150619180186234,
966
+ "grad_norm": 0.16754411160945892,
967
+ "learning_rate": 0.0001800369180283038,
968
+ "loss": 0.1069,
969
+ "step": 6850
970
+ },
971
+ {
972
+ "epoch": 1.3246616108284535,
973
+ "grad_norm": 0.11950711905956268,
974
+ "learning_rate": 0.00017901141724208653,
975
+ "loss": 0.1021,
976
+ "step": 6900
977
+ },
978
+ {
979
+ "epoch": 1.3342613036382835,
980
+ "grad_norm": 0.19078785181045532,
981
+ "learning_rate": 0.00017798591645586925,
982
+ "loss": 0.1047,
983
+ "step": 6950
984
+ },
985
+ {
986
+ "epoch": 1.3438609964481136,
987
+ "grad_norm": 0.1755492091178894,
988
+ "learning_rate": 0.000176960415669652,
989
+ "loss": 0.1059,
990
+ "step": 7000
991
+ },
992
+ {
993
+ "epoch": 1.3534606892579437,
994
+ "grad_norm": 0.15602967143058777,
995
+ "learning_rate": 0.00017593491488343473,
996
+ "loss": 0.1034,
997
+ "step": 7050
998
+ },
999
+ {
1000
+ "epoch": 1.3630603820677738,
1001
+ "grad_norm": 0.2121271789073944,
1002
+ "learning_rate": 0.00017490941409721746,
1003
+ "loss": 0.1033,
1004
+ "step": 7100
1005
+ },
1006
+ {
1007
+ "epoch": 1.3726600748776039,
1008
+ "grad_norm": 0.24380344152450562,
1009
+ "learning_rate": 0.0001738839133110002,
1010
+ "loss": 0.104,
1011
+ "step": 7150
1012
+ },
1013
+ {
1014
+ "epoch": 1.382259767687434,
1015
+ "grad_norm": 0.2051132768392563,
1016
+ "learning_rate": 0.0001728584125247829,
1017
+ "loss": 0.102,
1018
+ "step": 7200
1019
+ },
1020
+ {
1021
+ "epoch": 1.391859460497264,
1022
+ "grad_norm": 0.196175217628479,
1023
+ "learning_rate": 0.00017183291173856563,
1024
+ "loss": 0.1019,
1025
+ "step": 7250
1026
+ },
1027
+ {
1028
+ "epoch": 1.4014591533070941,
1029
+ "grad_norm": 0.11703308671712875,
1030
+ "learning_rate": 0.00017080741095234838,
1031
+ "loss": 0.1066,
1032
+ "step": 7300
1033
+ },
1034
+ {
1035
+ "epoch": 1.4110588461169242,
1036
+ "grad_norm": 0.13491560518741608,
1037
+ "learning_rate": 0.0001697819101661311,
1038
+ "loss": 0.1026,
1039
+ "step": 7350
1040
+ },
1041
+ {
1042
+ "epoch": 1.4206585389267543,
1043
+ "grad_norm": 0.11467725783586502,
1044
+ "learning_rate": 0.00016875640937991386,
1045
+ "loss": 0.1024,
1046
+ "step": 7400
1047
+ },
1048
+ {
1049
+ "epoch": 1.4302582317365844,
1050
+ "grad_norm": 0.23132862150669098,
1051
+ "learning_rate": 0.00016773090859369659,
1052
+ "loss": 0.1041,
1053
+ "step": 7450
1054
+ },
1055
+ {
1056
+ "epoch": 1.4398579245464145,
1057
+ "grad_norm": 0.09554282575845718,
1058
+ "learning_rate": 0.00016670540780747928,
1059
+ "loss": 0.1077,
1060
+ "step": 7500
1061
+ },
1062
+ {
1063
+ "epoch": 1.4494576173562446,
1064
+ "grad_norm": 0.1522522270679474,
1065
+ "learning_rate": 0.00016567990702126204,
1066
+ "loss": 0.1044,
1067
+ "step": 7550
1068
+ },
1069
+ {
1070
+ "epoch": 1.4590573101660747,
1071
+ "grad_norm": 0.13647988438606262,
1072
+ "learning_rate": 0.00016465440623504476,
1073
+ "loss": 0.1038,
1074
+ "step": 7600
1075
+ },
1076
+ {
1077
+ "epoch": 1.4686570029759047,
1078
+ "grad_norm": 0.21043138206005096,
1079
+ "learning_rate": 0.0001636289054488275,
1080
+ "loss": 0.1013,
1081
+ "step": 7650
1082
+ },
1083
+ {
1084
+ "epoch": 1.4782566957857348,
1085
+ "grad_norm": 0.15425056219100952,
1086
+ "learning_rate": 0.00016260340466261024,
1087
+ "loss": 0.1032,
1088
+ "step": 7700
1089
+ },
1090
+ {
1091
+ "epoch": 1.487856388595565,
1092
+ "grad_norm": 0.1538008749485016,
1093
+ "learning_rate": 0.00016157790387639296,
1094
+ "loss": 0.1022,
1095
+ "step": 7750
1096
+ },
1097
+ {
1098
+ "epoch": 1.497456081405395,
1099
+ "grad_norm": 0.09889411181211472,
1100
+ "learning_rate": 0.00016055240309017566,
1101
+ "loss": 0.1034,
1102
+ "step": 7800
1103
+ },
1104
+ {
1105
+ "epoch": 1.507055774215225,
1106
+ "grad_norm": 0.0999131053686142,
1107
+ "learning_rate": 0.00015952690230395841,
1108
+ "loss": 0.1042,
1109
+ "step": 7850
1110
+ },
1111
+ {
1112
+ "epoch": 1.5166554670250552,
1113
+ "grad_norm": 0.10318354517221451,
1114
+ "learning_rate": 0.00015850140151774114,
1115
+ "loss": 0.1026,
1116
+ "step": 7900
1117
+ },
1118
+ {
1119
+ "epoch": 1.5262551598348852,
1120
+ "grad_norm": 0.1949579417705536,
1121
+ "learning_rate": 0.0001574759007315239,
1122
+ "loss": 0.101,
1123
+ "step": 7950
1124
+ },
1125
+ {
1126
+ "epoch": 1.5358548526447153,
1127
+ "grad_norm": 0.17605358362197876,
1128
+ "learning_rate": 0.00015645039994530662,
1129
+ "loss": 0.1052,
1130
+ "step": 8000
1131
+ },
1132
+ {
1133
+ "epoch": 1.5454545454545454,
1134
+ "grad_norm": 0.23278145492076874,
1135
+ "learning_rate": 0.00015542489915908934,
1136
+ "loss": 0.1066,
1137
+ "step": 8050
1138
+ },
1139
+ {
1140
+ "epoch": 1.5550542382643755,
1141
+ "grad_norm": 0.24276228249073029,
1142
+ "learning_rate": 0.0001543993983728721,
1143
+ "loss": 0.1064,
1144
+ "step": 8100
1145
+ },
1146
+ {
1147
+ "epoch": 1.5646539310742056,
1148
+ "grad_norm": 0.16037604212760925,
1149
+ "learning_rate": 0.0001533738975866548,
1150
+ "loss": 0.1023,
1151
+ "step": 8150
1152
+ },
1153
+ {
1154
+ "epoch": 1.5742536238840357,
1155
+ "grad_norm": 0.2868128716945648,
1156
+ "learning_rate": 0.00015234839680043752,
1157
+ "loss": 0.1031,
1158
+ "step": 8200
1159
+ },
1160
+ {
1161
+ "epoch": 1.5838533166938658,
1162
+ "grad_norm": 0.10882575064897537,
1163
+ "learning_rate": 0.00015132289601422027,
1164
+ "loss": 0.1033,
1165
+ "step": 8250
1166
+ },
1167
+ {
1168
+ "epoch": 1.5934530095036958,
1169
+ "grad_norm": 0.45313918590545654,
1170
+ "learning_rate": 0.000150297395228003,
1171
+ "loss": 0.1042,
1172
+ "step": 8300
1173
+ },
1174
+ {
1175
+ "epoch": 1.603052702313526,
1176
+ "grad_norm": 1.096095323562622,
1177
+ "learning_rate": 0.00014927189444178572,
1178
+ "loss": 0.1123,
1179
+ "step": 8350
1180
+ },
1181
+ {
1182
+ "epoch": 1.612652395123356,
1183
+ "grad_norm": 0.15017154812812805,
1184
+ "learning_rate": 0.00014824639365556847,
1185
+ "loss": 0.1158,
1186
+ "step": 8400
1187
+ },
1188
+ {
1189
+ "epoch": 1.622252087933186,
1190
+ "grad_norm": 0.11559096723794937,
1191
+ "learning_rate": 0.00014722089286935117,
1192
+ "loss": 0.1024,
1193
+ "step": 8450
1194
+ },
1195
+ {
1196
+ "epoch": 1.6318517807430162,
1197
+ "grad_norm": 0.13219603896141052,
1198
+ "learning_rate": 0.00014619539208313392,
1199
+ "loss": 0.1017,
1200
+ "step": 8500
1201
+ },
1202
+ {
1203
+ "epoch": 1.6414514735528463,
1204
+ "grad_norm": 0.11071953922510147,
1205
+ "learning_rate": 0.00014516989129691665,
1206
+ "loss": 0.1005,
1207
+ "step": 8550
1208
+ },
1209
+ {
1210
+ "epoch": 1.6510511663626763,
1211
+ "grad_norm": 0.09291780740022659,
1212
+ "learning_rate": 0.00014414439051069937,
1213
+ "loss": 0.1002,
1214
+ "step": 8600
1215
+ },
1216
+ {
1217
+ "epoch": 1.6606508591725064,
1218
+ "grad_norm": 0.24729683995246887,
1219
+ "learning_rate": 0.0001431188897244821,
1220
+ "loss": 0.1001,
1221
+ "step": 8650
1222
+ },
1223
+ {
1224
+ "epoch": 1.6702505519823365,
1225
+ "grad_norm": 0.10102404654026031,
1226
+ "learning_rate": 0.00014209338893826485,
1227
+ "loss": 0.1017,
1228
+ "step": 8700
1229
+ },
1230
+ {
1231
+ "epoch": 1.6798502447921666,
1232
+ "grad_norm": 0.08600980043411255,
1233
+ "learning_rate": 0.00014106788815204758,
1234
+ "loss": 0.0992,
1235
+ "step": 8750
1236
+ },
1237
+ {
1238
+ "epoch": 1.6894499376019967,
1239
+ "grad_norm": 0.11300817131996155,
1240
+ "learning_rate": 0.0001400423873658303,
1241
+ "loss": 0.102,
1242
+ "step": 8800
1243
+ },
1244
+ {
1245
+ "epoch": 1.6990496304118268,
1246
+ "grad_norm": 0.07726065069437027,
1247
+ "learning_rate": 0.00013901688657961303,
1248
+ "loss": 0.101,
1249
+ "step": 8850
1250
+ },
1251
+ {
1252
+ "epoch": 1.7086493232216569,
1253
+ "grad_norm": 0.07588201761245728,
1254
+ "learning_rate": 0.00013799138579339575,
1255
+ "loss": 0.0981,
1256
+ "step": 8900
1257
+ },
1258
+ {
1259
+ "epoch": 1.718249016031487,
1260
+ "grad_norm": 0.12571977078914642,
1261
+ "learning_rate": 0.0001369658850071785,
1262
+ "loss": 0.0998,
1263
+ "step": 8950
1264
+ },
1265
+ {
1266
+ "epoch": 1.727848708841317,
1267
+ "grad_norm": 0.11262161284685135,
1268
+ "learning_rate": 0.00013594038422096123,
1269
+ "loss": 0.1018,
1270
+ "step": 9000
1271
+ },
1272
+ {
1273
+ "epoch": 1.7374484016511471,
1274
+ "grad_norm": 0.07787909358739853,
1275
+ "learning_rate": 0.00013491488343474395,
1276
+ "loss": 0.1008,
1277
+ "step": 9050
1278
+ },
1279
+ {
1280
+ "epoch": 1.7470480944609772,
1281
+ "grad_norm": 0.19743576645851135,
1282
+ "learning_rate": 0.00013388938264852668,
1283
+ "loss": 0.1018,
1284
+ "step": 9100
1285
+ },
1286
+ {
1287
+ "epoch": 1.7566477872708073,
1288
+ "grad_norm": 0.10666341334581375,
1289
+ "learning_rate": 0.00013286388186230943,
1290
+ "loss": 0.1004,
1291
+ "step": 9150
1292
+ },
1293
+ {
1294
+ "epoch": 1.7662474800806374,
1295
+ "grad_norm": 0.12897364795207977,
1296
+ "learning_rate": 0.00013183838107609213,
1297
+ "loss": 0.1005,
1298
+ "step": 9200
1299
+ },
1300
+ {
1301
+ "epoch": 1.7758471728904675,
1302
+ "grad_norm": 0.10816533863544464,
1303
+ "learning_rate": 0.00013081288028987488,
1304
+ "loss": 0.1001,
1305
+ "step": 9250
1306
+ },
1307
+ {
1308
+ "epoch": 1.7854468657002975,
1309
+ "grad_norm": 0.16208283603191376,
1310
+ "learning_rate": 0.0001297873795036576,
1311
+ "loss": 0.1002,
1312
+ "step": 9300
1313
+ },
1314
+ {
1315
+ "epoch": 1.7950465585101276,
1316
+ "grad_norm": 0.26765286922454834,
1317
+ "learning_rate": 0.00012876187871744033,
1318
+ "loss": 0.1014,
1319
+ "step": 9350
1320
+ },
1321
+ {
1322
+ "epoch": 1.8046462513199577,
1323
+ "grad_norm": 0.0876406878232956,
1324
+ "learning_rate": 0.00012773637793122306,
1325
+ "loss": 0.1006,
1326
+ "step": 9400
1327
+ },
1328
+ {
1329
+ "epoch": 1.8142459441297878,
1330
+ "grad_norm": 0.3883680999279022,
1331
+ "learning_rate": 0.0001267108771450058,
1332
+ "loss": 0.0999,
1333
+ "step": 9450
1334
+ },
1335
+ {
1336
+ "epoch": 1.8238456369396179,
1337
+ "grad_norm": 0.17567026615142822,
1338
+ "learning_rate": 0.00012568537635878854,
1339
+ "loss": 0.1001,
1340
+ "step": 9500
1341
+ },
1342
+ {
1343
+ "epoch": 1.833445329749448,
1344
+ "grad_norm": 0.11428207159042358,
1345
+ "learning_rate": 0.00012465987557257126,
1346
+ "loss": 0.1019,
1347
+ "step": 9550
1348
+ },
1349
+ {
1350
+ "epoch": 1.843045022559278,
1351
+ "grad_norm": 0.10391507297754288,
1352
+ "learning_rate": 0.00012363437478635399,
1353
+ "loss": 0.0986,
1354
+ "step": 9600
1355
+ },
1356
+ {
1357
+ "epoch": 1.8526447153691081,
1358
+ "grad_norm": 0.08824704587459564,
1359
+ "learning_rate": 0.00012260887400013674,
1360
+ "loss": 0.0994,
1361
+ "step": 9650
1362
+ },
1363
+ {
1364
+ "epoch": 1.8622444081789382,
1365
+ "grad_norm": 0.10438057780265808,
1366
+ "learning_rate": 0.00012158337321391946,
1367
+ "loss": 0.1006,
1368
+ "step": 9700
1369
+ },
1370
+ {
1371
+ "epoch": 1.8718441009887683,
1372
+ "grad_norm": 0.10627109557390213,
1373
+ "learning_rate": 0.00012055787242770217,
1374
+ "loss": 0.1003,
1375
+ "step": 9750
1376
+ },
1377
+ {
1378
+ "epoch": 1.8814437937985984,
1379
+ "grad_norm": 0.1276552826166153,
1380
+ "learning_rate": 0.00011953237164148491,
1381
+ "loss": 0.1006,
1382
+ "step": 9800
1383
+ },
1384
+ {
1385
+ "epoch": 1.8910434866084285,
1386
+ "grad_norm": 0.1135965958237648,
1387
+ "learning_rate": 0.00011850687085526765,
1388
+ "loss": 0.0994,
1389
+ "step": 9850
1390
+ },
1391
+ {
1392
+ "epoch": 1.9006431794182586,
1393
+ "grad_norm": 0.06545350700616837,
1394
+ "learning_rate": 0.00011748137006905036,
1395
+ "loss": 0.0999,
1396
+ "step": 9900
1397
+ },
1398
+ {
1399
+ "epoch": 1.9102428722280886,
1400
+ "grad_norm": 0.07519547641277313,
1401
+ "learning_rate": 0.0001164558692828331,
1402
+ "loss": 0.0994,
1403
+ "step": 9950
1404
+ },
1405
+ {
1406
+ "epoch": 1.9198425650379187,
1407
+ "grad_norm": 0.06651521474123001,
1408
+ "learning_rate": 0.00011543036849661584,
1409
+ "loss": 0.1022,
1410
+ "step": 10000
1411
+ },
1412
+ {
1413
+ "epoch": 1.9294422578477488,
1414
+ "grad_norm": 0.08063454180955887,
1415
+ "learning_rate": 0.00011440486771039857,
1416
+ "loss": 0.0995,
1417
+ "step": 10050
1418
+ },
1419
+ {
1420
+ "epoch": 1.939041950657579,
1421
+ "grad_norm": 0.09275515377521515,
1422
+ "learning_rate": 0.00011337936692418129,
1423
+ "loss": 0.1018,
1424
+ "step": 10100
1425
+ },
1426
+ {
1427
+ "epoch": 1.948641643467409,
1428
+ "grad_norm": 0.1343824565410614,
1429
+ "learning_rate": 0.00011235386613796403,
1430
+ "loss": 0.1013,
1431
+ "step": 10150
1432
+ },
1433
+ {
1434
+ "epoch": 1.958241336277239,
1435
+ "grad_norm": 0.07984330505132675,
1436
+ "learning_rate": 0.00011132836535174676,
1437
+ "loss": 0.0975,
1438
+ "step": 10200
1439
+ },
1440
+ {
1441
+ "epoch": 1.9678410290870691,
1442
+ "grad_norm": 0.17874419689178467,
1443
+ "learning_rate": 0.0001103028645655295,
1444
+ "loss": 0.0989,
1445
+ "step": 10250
1446
+ },
1447
+ {
1448
+ "epoch": 1.9774407218968992,
1449
+ "grad_norm": 0.10690494626760483,
1450
+ "learning_rate": 0.00010927736377931222,
1451
+ "loss": 0.0994,
1452
+ "step": 10300
1453
+ },
1454
+ {
1455
+ "epoch": 1.9870404147067293,
1456
+ "grad_norm": 0.08402542024850845,
1457
+ "learning_rate": 0.00010825186299309494,
1458
+ "loss": 0.098,
1459
+ "step": 10350
1460
+ },
1461
+ {
1462
+ "epoch": 1.9966401075165594,
1463
+ "grad_norm": 0.08502990752458572,
1464
+ "learning_rate": 0.00010722636220687768,
1465
+ "loss": 0.0992,
1466
+ "step": 10400
1467
+ },
1468
+ {
1469
+ "epoch": 2.0061438033982912,
1470
+ "grad_norm": 0.18513023853302002,
1471
+ "learning_rate": 0.00010620086142066042,
1472
+ "loss": 0.103,
1473
+ "step": 10450
1474
+ },
1475
+ {
1476
+ "epoch": 2.0157434962081213,
1477
+ "grad_norm": 0.10088857263326645,
1478
+ "learning_rate": 0.00010517536063444313,
1479
+ "loss": 0.099,
1480
+ "step": 10500
1481
+ },
1482
+ {
1483
+ "epoch": 2.0253431890179514,
1484
+ "grad_norm": 0.08932794630527496,
1485
+ "learning_rate": 0.00010414985984822587,
1486
+ "loss": 0.0986,
1487
+ "step": 10550
1488
+ },
1489
+ {
1490
+ "epoch": 2.0349428818277815,
1491
+ "grad_norm": 0.12779156863689423,
1492
+ "learning_rate": 0.00010312435906200861,
1493
+ "loss": 0.0987,
1494
+ "step": 10600
1495
+ },
1496
+ {
1497
+ "epoch": 2.0445425746376116,
1498
+ "grad_norm": 0.08535610139369965,
1499
+ "learning_rate": 0.00010209885827579132,
1500
+ "loss": 0.0976,
1501
+ "step": 10650
1502
+ },
1503
+ {
1504
+ "epoch": 2.0541422674474417,
1505
+ "grad_norm": 0.13694776594638824,
1506
+ "learning_rate": 0.00010107335748957406,
1507
+ "loss": 0.0992,
1508
+ "step": 10700
1509
+ },
1510
+ {
1511
+ "epoch": 2.0637419602572717,
1512
+ "grad_norm": 0.09219267219305038,
1513
+ "learning_rate": 0.0001000478567033568,
1514
+ "loss": 0.0985,
1515
+ "step": 10750
1516
+ },
1517
+ {
1518
+ "epoch": 2.073341653067102,
1519
+ "grad_norm": 0.20255500078201294,
1520
+ "learning_rate": 9.902235591713954e-05,
1521
+ "loss": 0.0992,
1522
+ "step": 10800
1523
+ },
1524
+ {
1525
+ "epoch": 2.082941345876932,
1526
+ "grad_norm": 0.09075580537319183,
1527
+ "learning_rate": 9.799685513092225e-05,
1528
+ "loss": 0.0993,
1529
+ "step": 10850
1530
+ },
1531
+ {
1532
+ "epoch": 2.092541038686762,
1533
+ "grad_norm": 0.0955098420381546,
1534
+ "learning_rate": 9.697135434470499e-05,
1535
+ "loss": 0.0977,
1536
+ "step": 10900
1537
+ },
1538
+ {
1539
+ "epoch": 2.102140731496592,
1540
+ "grad_norm": 0.14620476961135864,
1541
+ "learning_rate": 9.594585355848773e-05,
1542
+ "loss": 0.0988,
1543
+ "step": 10950
1544
+ },
1545
+ {
1546
+ "epoch": 2.111740424306422,
1547
+ "grad_norm": 0.06722934544086456,
1548
+ "learning_rate": 9.492035277227045e-05,
1549
+ "loss": 0.0994,
1550
+ "step": 11000
1551
+ },
1552
+ {
1553
+ "epoch": 2.1213401171162523,
1554
+ "grad_norm": 0.11390715092420578,
1555
+ "learning_rate": 9.389485198605318e-05,
1556
+ "loss": 0.0996,
1557
+ "step": 11050
1558
+ },
1559
+ {
1560
+ "epoch": 2.1309398099260823,
1561
+ "grad_norm": 0.08618480712175369,
1562
+ "learning_rate": 9.286935119983592e-05,
1563
+ "loss": 0.0985,
1564
+ "step": 11100
1565
+ },
1566
+ {
1567
+ "epoch": 2.1405395027359124,
1568
+ "grad_norm": 0.09730342775583267,
1569
+ "learning_rate": 9.184385041361864e-05,
1570
+ "loss": 0.0987,
1571
+ "step": 11150
1572
+ },
1573
+ {
1574
+ "epoch": 2.1501391955457425,
1575
+ "grad_norm": 0.23508231341838837,
1576
+ "learning_rate": 9.081834962740137e-05,
1577
+ "loss": 0.0965,
1578
+ "step": 11200
1579
+ },
1580
+ {
1581
+ "epoch": 2.1597388883555726,
1582
+ "grad_norm": 0.0964941680431366,
1583
+ "learning_rate": 8.97928488411841e-05,
1584
+ "loss": 0.1017,
1585
+ "step": 11250
1586
+ },
1587
+ {
1588
+ "epoch": 2.1693385811654027,
1589
+ "grad_norm": 0.06535065919160843,
1590
+ "learning_rate": 8.876734805496683e-05,
1591
+ "loss": 0.0986,
1592
+ "step": 11300
1593
+ },
1594
+ {
1595
+ "epoch": 2.1789382739752328,
1596
+ "grad_norm": 0.06535002589225769,
1597
+ "learning_rate": 8.774184726874957e-05,
1598
+ "loss": 0.0988,
1599
+ "step": 11350
1600
+ },
1601
+ {
1602
+ "epoch": 2.188537966785063,
1603
+ "grad_norm": 0.06418288499116898,
1604
+ "learning_rate": 8.67163464825323e-05,
1605
+ "loss": 0.0987,
1606
+ "step": 11400
1607
+ },
1608
+ {
1609
+ "epoch": 2.198137659594893,
1610
+ "grad_norm": 0.1190769225358963,
1611
+ "learning_rate": 8.569084569631502e-05,
1612
+ "loss": 0.0972,
1613
+ "step": 11450
1614
+ },
1615
+ {
1616
+ "epoch": 2.207737352404723,
1617
+ "grad_norm": 0.07921498268842697,
1618
+ "learning_rate": 8.466534491009776e-05,
1619
+ "loss": 0.0975,
1620
+ "step": 11500
1621
+ },
1622
+ {
1623
+ "epoch": 2.217337045214553,
1624
+ "grad_norm": 0.08557629585266113,
1625
+ "learning_rate": 8.36398441238805e-05,
1626
+ "loss": 0.0964,
1627
+ "step": 11550
1628
+ },
1629
+ {
1630
+ "epoch": 2.226936738024383,
1631
+ "grad_norm": 0.09319253265857697,
1632
+ "learning_rate": 8.261434333766321e-05,
1633
+ "loss": 0.0976,
1634
+ "step": 11600
1635
+ },
1636
+ {
1637
+ "epoch": 2.2365364308342133,
1638
+ "grad_norm": 0.08666611462831497,
1639
+ "learning_rate": 8.158884255144595e-05,
1640
+ "loss": 0.0987,
1641
+ "step": 11650
1642
+ },
1643
+ {
1644
+ "epoch": 2.2461361236440434,
1645
+ "grad_norm": 0.05421123653650284,
1646
+ "learning_rate": 8.056334176522869e-05,
1647
+ "loss": 0.0961,
1648
+ "step": 11700
1649
+ },
1650
+ {
1651
+ "epoch": 2.2557358164538734,
1652
+ "grad_norm": 0.1075129583477974,
1653
+ "learning_rate": 7.95378409790114e-05,
1654
+ "loss": 0.0989,
1655
+ "step": 11750
1656
+ },
1657
+ {
1658
+ "epoch": 2.2653355092637035,
1659
+ "grad_norm": 0.10412245243787766,
1660
+ "learning_rate": 7.851234019279414e-05,
1661
+ "loss": 0.0988,
1662
+ "step": 11800
1663
+ },
1664
+ {
1665
+ "epoch": 2.2749352020735336,
1666
+ "grad_norm": 0.10957926511764526,
1667
+ "learning_rate": 7.748683940657688e-05,
1668
+ "loss": 0.0972,
1669
+ "step": 11850
1670
+ },
1671
+ {
1672
+ "epoch": 2.2845348948833637,
1673
+ "grad_norm": 0.06667158007621765,
1674
+ "learning_rate": 7.64613386203596e-05,
1675
+ "loss": 0.0959,
1676
+ "step": 11900
1677
+ },
1678
+ {
1679
+ "epoch": 2.294134587693194,
1680
+ "grad_norm": 0.07828564196825027,
1681
+ "learning_rate": 7.543583783414233e-05,
1682
+ "loss": 0.0968,
1683
+ "step": 11950
1684
+ },
1685
+ {
1686
+ "epoch": 2.303734280503024,
1687
+ "grad_norm": 0.07054735720157623,
1688
+ "learning_rate": 7.441033704792507e-05,
1689
+ "loss": 0.0975,
1690
+ "step": 12000
1691
+ },
1692
+ {
1693
+ "epoch": 2.313333973312854,
1694
+ "grad_norm": 0.06447632610797882,
1695
+ "learning_rate": 7.338483626170779e-05,
1696
+ "loss": 0.0962,
1697
+ "step": 12050
1698
+ },
1699
+ {
1700
+ "epoch": 2.322933666122684,
1701
+ "grad_norm": 0.07028626650571823,
1702
+ "learning_rate": 7.235933547549053e-05,
1703
+ "loss": 0.0961,
1704
+ "step": 12100
1705
+ },
1706
+ {
1707
+ "epoch": 2.332533358932514,
1708
+ "grad_norm": 0.06260377168655396,
1709
+ "learning_rate": 7.133383468927325e-05,
1710
+ "loss": 0.0964,
1711
+ "step": 12150
1712
+ },
1713
+ {
1714
+ "epoch": 2.342133051742344,
1715
+ "grad_norm": 0.07249750196933746,
1716
+ "learning_rate": 7.030833390305598e-05,
1717
+ "loss": 0.0976,
1718
+ "step": 12200
1719
+ },
1720
+ {
1721
+ "epoch": 2.3517327445521743,
1722
+ "grad_norm": 0.08855904638767242,
1723
+ "learning_rate": 6.928283311683872e-05,
1724
+ "loss": 0.0967,
1725
+ "step": 12250
1726
+ },
1727
+ {
1728
+ "epoch": 2.3613324373620044,
1729
+ "grad_norm": 0.051147934049367905,
1730
+ "learning_rate": 6.825733233062144e-05,
1731
+ "loss": 0.0981,
1732
+ "step": 12300
1733
+ },
1734
+ {
1735
+ "epoch": 2.3709321301718345,
1736
+ "grad_norm": 0.08245785534381866,
1737
+ "learning_rate": 6.723183154440417e-05,
1738
+ "loss": 0.0965,
1739
+ "step": 12350
1740
+ },
1741
+ {
1742
+ "epoch": 2.3805318229816645,
1743
+ "grad_norm": 0.08681339770555496,
1744
+ "learning_rate": 6.620633075818691e-05,
1745
+ "loss": 0.0977,
1746
+ "step": 12400
1747
+ },
1748
+ {
1749
+ "epoch": 2.3901315157914946,
1750
+ "grad_norm": 0.05284340679645538,
1751
+ "learning_rate": 6.518082997196963e-05,
1752
+ "loss": 0.0945,
1753
+ "step": 12450
1754
+ },
1755
+ {
1756
+ "epoch": 2.3997312086013247,
1757
+ "grad_norm": 0.05780034512281418,
1758
+ "learning_rate": 6.415532918575237e-05,
1759
+ "loss": 0.0988,
1760
+ "step": 12500
1761
+ },
1762
+ {
1763
+ "epoch": 2.409330901411155,
1764
+ "grad_norm": 0.05405628681182861,
1765
+ "learning_rate": 6.31298283995351e-05,
1766
+ "loss": 0.0991,
1767
+ "step": 12550
1768
+ },
1769
+ {
1770
+ "epoch": 2.418930594220985,
1771
+ "grad_norm": 0.06091487780213356,
1772
+ "learning_rate": 6.210432761331784e-05,
1773
+ "loss": 0.0973,
1774
+ "step": 12600
1775
+ },
1776
+ {
1777
+ "epoch": 2.428530287030815,
1778
+ "grad_norm": 0.09573518484830856,
1779
+ "learning_rate": 6.107882682710056e-05,
1780
+ "loss": 0.0968,
1781
+ "step": 12650
1782
+ },
1783
+ {
1784
+ "epoch": 2.438129979840645,
1785
+ "grad_norm": 0.06391169875860214,
1786
+ "learning_rate": 6.005332604088329e-05,
1787
+ "loss": 0.097,
1788
+ "step": 12700
1789
+ },
1790
+ {
1791
+ "epoch": 2.447729672650475,
1792
+ "grad_norm": 0.06078439578413963,
1793
+ "learning_rate": 5.9027825254666025e-05,
1794
+ "loss": 0.0962,
1795
+ "step": 12750
1796
+ },
1797
+ {
1798
+ "epoch": 2.4573293654603052,
1799
+ "grad_norm": 0.05879442021250725,
1800
+ "learning_rate": 5.800232446844876e-05,
1801
+ "loss": 0.0983,
1802
+ "step": 12800
1803
+ },
1804
+ {
1805
+ "epoch": 2.4669290582701353,
1806
+ "grad_norm": 0.07806123793125153,
1807
+ "learning_rate": 5.697682368223148e-05,
1808
+ "loss": 0.0975,
1809
+ "step": 12850
1810
+ },
1811
+ {
1812
+ "epoch": 2.4765287510799654,
1813
+ "grad_norm": 0.0653390884399414,
1814
+ "learning_rate": 5.5951322896014214e-05,
1815
+ "loss": 0.0966,
1816
+ "step": 12900
1817
+ },
1818
+ {
1819
+ "epoch": 2.4861284438897955,
1820
+ "grad_norm": 0.07627742737531662,
1821
+ "learning_rate": 5.4925822109796946e-05,
1822
+ "loss": 0.0975,
1823
+ "step": 12950
1824
+ },
1825
+ {
1826
+ "epoch": 2.4957281366996256,
1827
+ "grad_norm": 0.0665074959397316,
1828
+ "learning_rate": 5.390032132357967e-05,
1829
+ "loss": 0.0953,
1830
+ "step": 13000
1831
+ },
1832
+ {
1833
+ "epoch": 2.5053278295094557,
1834
+ "grad_norm": 0.07060660421848297,
1835
+ "learning_rate": 5.287482053736241e-05,
1836
+ "loss": 0.0971,
1837
+ "step": 13050
1838
+ },
1839
+ {
1840
+ "epoch": 2.5149275223192857,
1841
+ "grad_norm": 0.05964155122637749,
1842
+ "learning_rate": 5.1849319751145135e-05,
1843
+ "loss": 0.096,
1844
+ "step": 13100
1845
+ },
1846
+ {
1847
+ "epoch": 2.524527215129116,
1848
+ "grad_norm": 0.08040162920951843,
1849
+ "learning_rate": 5.082381896492787e-05,
1850
+ "loss": 0.0977,
1851
+ "step": 13150
1852
+ },
1853
+ {
1854
+ "epoch": 2.534126907938946,
1855
+ "grad_norm": 0.06038783863186836,
1856
+ "learning_rate": 4.97983181787106e-05,
1857
+ "loss": 0.0962,
1858
+ "step": 13200
1859
+ },
1860
+ {
1861
+ "epoch": 2.543726600748776,
1862
+ "grad_norm": 0.05058710277080536,
1863
+ "learning_rate": 4.877281739249333e-05,
1864
+ "loss": 0.097,
1865
+ "step": 13250
1866
+ },
1867
+ {
1868
+ "epoch": 2.553326293558606,
1869
+ "grad_norm": 0.06195589527487755,
1870
+ "learning_rate": 4.7747316606276056e-05,
1871
+ "loss": 0.0983,
1872
+ "step": 13300
1873
+ },
1874
+ {
1875
+ "epoch": 2.562925986368436,
1876
+ "grad_norm": 0.06552109122276306,
1877
+ "learning_rate": 4.6721815820058795e-05,
1878
+ "loss": 0.0983,
1879
+ "step": 13350
1880
+ },
1881
+ {
1882
+ "epoch": 2.5725256791782662,
1883
+ "grad_norm": 0.07137109339237213,
1884
+ "learning_rate": 4.569631503384152e-05,
1885
+ "loss": 0.0971,
1886
+ "step": 13400
1887
+ },
1888
+ {
1889
+ "epoch": 2.5821253719880963,
1890
+ "grad_norm": 0.07768446952104568,
1891
+ "learning_rate": 4.467081424762426e-05,
1892
+ "loss": 0.0977,
1893
+ "step": 13450
1894
+ },
1895
+ {
1896
+ "epoch": 2.5917250647979264,
1897
+ "grad_norm": 0.06904645264148712,
1898
+ "learning_rate": 4.3645313461406984e-05,
1899
+ "loss": 0.097,
1900
+ "step": 13500
1901
+ },
1902
+ {
1903
+ "epoch": 2.6013247576077565,
1904
+ "grad_norm": 0.06683080643415451,
1905
+ "learning_rate": 4.261981267518971e-05,
1906
+ "loss": 0.0968,
1907
+ "step": 13550
1908
+ },
1909
+ {
1910
+ "epoch": 2.6109244504175866,
1911
+ "grad_norm": 0.06532671302556992,
1912
+ "learning_rate": 4.159431188897245e-05,
1913
+ "loss": 0.0961,
1914
+ "step": 13600
1915
+ },
1916
+ {
1917
+ "epoch": 2.6205241432274167,
1918
+ "grad_norm": 0.0853889137506485,
1919
+ "learning_rate": 4.056881110275517e-05,
1920
+ "loss": 0.0974,
1921
+ "step": 13650
1922
+ },
1923
+ {
1924
+ "epoch": 2.6301238360372468,
1925
+ "grad_norm": 0.05568193271756172,
1926
+ "learning_rate": 3.9543310316537905e-05,
1927
+ "loss": 0.0974,
1928
+ "step": 13700
1929
+ },
1930
+ {
1931
+ "epoch": 2.639723528847077,
1932
+ "grad_norm": 0.061790887266397476,
1933
+ "learning_rate": 3.851780953032064e-05,
1934
+ "loss": 0.0941,
1935
+ "step": 13750
1936
+ },
1937
+ {
1938
+ "epoch": 2.649323221656907,
1939
+ "grad_norm": 0.08820798248052597,
1940
+ "learning_rate": 3.749230874410337e-05,
1941
+ "loss": 0.0963,
1942
+ "step": 13800
1943
+ },
1944
+ {
1945
+ "epoch": 2.658922914466737,
1946
+ "grad_norm": 0.1133754551410675,
1947
+ "learning_rate": 3.6466807957886094e-05,
1948
+ "loss": 0.0971,
1949
+ "step": 13850
1950
+ },
1951
+ {
1952
+ "epoch": 2.668522607276567,
1953
+ "grad_norm": 0.06903323531150818,
1954
+ "learning_rate": 3.5441307171668826e-05,
1955
+ "loss": 0.098,
1956
+ "step": 13900
1957
+ },
1958
+ {
1959
+ "epoch": 2.678122300086397,
1960
+ "grad_norm": 0.08473759889602661,
1961
+ "learning_rate": 3.441580638545156e-05,
1962
+ "loss": 0.0978,
1963
+ "step": 13950
1964
+ },
1965
+ {
1966
+ "epoch": 2.6877219928962273,
1967
+ "grad_norm": 0.054722901433706284,
1968
+ "learning_rate": 3.339030559923429e-05,
1969
+ "loss": 0.0968,
1970
+ "step": 14000
1971
+ }
1972
+ ],
1973
+ "logging_steps": 50,
1974
+ "max_steps": 15627,
1975
+ "num_input_tokens_seen": 0,
1976
+ "num_train_epochs": 3,
1977
+ "save_steps": 2000,
1978
+ "stateful_callbacks": {
1979
+ "TrainerControl": {
1980
+ "args": {
1981
+ "should_epoch_stop": false,
1982
+ "should_evaluate": false,
1983
+ "should_log": false,
1984
+ "should_save": true,
1985
+ "should_training_stop": false
1986
+ },
1987
+ "attributes": {}
1988
+ }
1989
+ },
1990
+ "total_flos": 3.070877958710231e+17,
1991
+ "train_batch_size": 16,
1992
+ "trial_name": null,
1993
+ "trial_params": null
1994
+ }
checkpoint-14000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:478c5792cae56598ec41be11b09c6fe07ef08f3d28ac262adfe2b7d757dfe6fa
3
+ size 5841
checkpoint-15627/config.json ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "PaTHAttentionForCausalLM"
4
+ ],
5
+ "bos_token_id": 0,
6
+ "dtype": "float32",
7
+ "elementwise_affine": true,
8
+ "eos_token_id": 1,
9
+ "fuse_cross_entropy": true,
10
+ "fuse_linear_cross_entropy": false,
11
+ "fuse_norm": true,
12
+ "fuse_swiglu": true,
13
+ "head_dim": 64,
14
+ "hidden_act": "swish",
15
+ "hidden_ratio": 4,
16
+ "hidden_size": 256,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 768,
19
+ "max_position_embeddings": 2048,
20
+ "model_type": "path_attn",
21
+ "norm_eps": 1e-05,
22
+ "num_heads": 4,
23
+ "num_hidden_layers": 32,
24
+ "num_kv_heads": 4,
25
+ "rope_theta": 10000.0,
26
+ "tie_word_embeddings": true,
27
+ "transformers_version": "4.57.1",
28
+ "use_cache": true,
29
+ "use_forget_gate": true,
30
+ "use_l2warp": false,
31
+ "use_low_rank_w": true,
32
+ "use_w_shortconv": true,
33
+ "vocab_size": 256
34
+ }
checkpoint-15627/generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 0,
4
+ "eos_token_id": 1,
5
+ "transformers_version": "4.57.1"
6
+ }
checkpoint-15627/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d5397c30c8ab71c5b252b1a059a4b4abe1ca51afd6936ac40897e655f5201ed
3
+ size 111892816
checkpoint-15627/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3de14e5dfeedcd9cf8bad996844e3527ea71626dec373aa38e3e323952def66d
3
+ size 224116841
checkpoint-15627/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4a9f217e852f439efa6bd32fde98d6867f11aa6ea13ddc021ba10af6a0b0934
3
+ size 14645
checkpoint-15627/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4fc80450de44a2c06865da7e9ea4bd3e647a4eb1f084d0ca7782e692a133a0b0
3
+ size 1465
checkpoint-15627/special_tokens_map.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<think>",
4
+ "</think>"
5
+ ],
6
+ "bos_token": {
7
+ "content": "<|startoftext|>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false
12
+ },
13
+ "eos_token": {
14
+ "content": "<|endoftext|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false
19
+ },
20
+ "pad_token": "<|pad|>",
21
+ "unk_token": {
22
+ "content": "<|unknown|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false
27
+ }
28
+ }
checkpoint-15627/tokenizer.json ADDED
@@ -0,0 +1,1085 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "1.0",
3
+ "truncation": null,
4
+ "padding": null,
5
+ "added_tokens": [
6
+ {
7
+ "id": 0,
8
+ "content": "<|startoftext|>",
9
+ "single_word": false,
10
+ "lstrip": false,
11
+ "rstrip": false,
12
+ "normalized": false,
13
+ "special": true
14
+ },
15
+ {
16
+ "id": 1,
17
+ "content": "<|endoftext|>",
18
+ "single_word": false,
19
+ "lstrip": false,
20
+ "rstrip": false,
21
+ "normalized": false,
22
+ "special": true
23
+ },
24
+ {
25
+ "id": 2,
26
+ "content": "<|unknown|>",
27
+ "single_word": false,
28
+ "lstrip": false,
29
+ "rstrip": false,
30
+ "normalized": false,
31
+ "special": true
32
+ },
33
+ {
34
+ "id": 3,
35
+ "content": "<|pad|>",
36
+ "single_word": false,
37
+ "lstrip": false,
38
+ "rstrip": false,
39
+ "normalized": false,
40
+ "special": true
41
+ },
42
+ {
43
+ "id": 4,
44
+ "content": "<think>",
45
+ "single_word": false,
46
+ "lstrip": false,
47
+ "rstrip": false,
48
+ "normalized": false,
49
+ "special": true
50
+ },
51
+ {
52
+ "id": 5,
53
+ "content": "</think>",
54
+ "single_word": false,
55
+ "lstrip": false,
56
+ "rstrip": false,
57
+ "normalized": false,
58
+ "special": true
59
+ }
60
+ ],
61
+ "normalizer": null,
62
+ "pre_tokenizer": {
63
+ "type": "ByteLevel",
64
+ "add_prefix_space": false,
65
+ "trim_offsets": true,
66
+ "use_regex": true
67
+ },
68
+ "post_processor": null,
69
+ "decoder": {
70
+ "type": "ByteLevel",
71
+ "add_prefix_space": true,
72
+ "trim_offsets": true,
73
+ "use_regex": true
74
+ },
75
+ "model": {
76
+ "type": "BPE",
77
+ "dropout": null,
78
+ "unk_token": null,
79
+ "continuing_subword_prefix": null,
80
+ "end_of_word_suffix": null,
81
+ "fuse_unk": false,
82
+ "byte_fallback": false,
83
+ "ignore_merges": false,
84
+ "vocab": {
85
+ "<|startoftext|>": 0,
86
+ "<|endoftext|>": 1,
87
+ "<|unknown|>": 2,
88
+ "<|pad|>": 3,
89
+ "<think>": 4,
90
+ "</think>": 5,
91
+ "%": 6,
92
+ "&": 7,
93
+ "'": 8,
94
+ "(": 9,
95
+ ")": 10,
96
+ ",": 11,
97
+ "-": 12,
98
+ ".": 13,
99
+ "0": 14,
100
+ "1": 15,
101
+ "2": 16,
102
+ "3": 17,
103
+ "4": 18,
104
+ "5": 19,
105
+ "6": 20,
106
+ "7": 21,
107
+ "8": 22,
108
+ "9": 23,
109
+ ":": 24,
110
+ ">": 25,
111
+ "A": 26,
112
+ "C": 27,
113
+ "D": 28,
114
+ "E": 29,
115
+ "F": 30,
116
+ "G": 31,
117
+ "H": 32,
118
+ "I": 33,
119
+ "L": 34,
120
+ "M": 35,
121
+ "N": 36,
122
+ "O": 37,
123
+ "P": 38,
124
+ "R": 39,
125
+ "S": 40,
126
+ "T": 41,
127
+ "V": 42,
128
+ "[": 43,
129
+ "]": 44,
130
+ "a": 45,
131
+ "b": 46,
132
+ "c": 47,
133
+ "d": 48,
134
+ "e": 49,
135
+ "f": 50,
136
+ "g": 51,
137
+ "h": 52,
138
+ "i": 53,
139
+ "k": 54,
140
+ "l": 55,
141
+ "m": 56,
142
+ "n": 57,
143
+ "o": 58,
144
+ "p": 59,
145
+ "q": 60,
146
+ "r": 61,
147
+ "s": 62,
148
+ "t": 63,
149
+ "u": 64,
150
+ "v": 65,
151
+ "x": 66,
152
+ "y": 67,
153
+ "|": 68,
154
+ "Ċ": 69,
155
+ "Ġ": 70,
156
+ "00": 71,
157
+ "11": 72,
158
+ "->": 73,
159
+ "Ġ|": 74,
160
+ ":[": 75,
161
+ "]->": 76,
162
+ "10": 77,
163
+ "Ġ1": 78,
164
+ "000": 79,
165
+ "ĠS": 80,
166
+ "',": 81,
167
+ "Ġ'": 82,
168
+ "te": 83,
169
+ "100": 84,
170
+ "ĊĠ": 85,
171
+ "110": 86,
172
+ "tep": 87,
173
+ "re": 88,
174
+ "111": 89,
175
+ "0000": 90,
176
+ "011": 91,
177
+ "101": 92,
178
+ "001": 93,
179
+ "ĠStep": 94,
180
+ "010": 95,
181
+ "in": 96,
182
+ "im": 97,
183
+ "ĠSim": 98,
184
+ "1111": 99,
185
+ "Ġ[": 100,
186
+ "or": 101,
187
+ "ĠI": 102,
188
+ "ho": 103,
189
+ "od": 104,
190
+ "hood": 105,
191
+ "Ġin": 106,
192
+ "Ġ11": 107,
193
+ "Ġ10": 108,
194
+ "ne": 109,
195
+ "Ġne": 110,
196
+ "']": 111,
197
+ "bor": 112,
198
+ "dx": 113,
199
+ "gh": 114,
200
+ "ire": 115,
201
+ "igh": 116,
202
+ "qu": 117,
203
+ "Ġre": 118,
204
+ "Ġ['": 119,
205
+ "ĠIdx": 120,
206
+ "Ġneigh": 121,
207
+ "borhood": 122,
208
+ "ires": 123,
209
+ "quires": 124,
210
+ "Ġrequires": 125,
211
+ "Ġneighborhood": 126,
212
+ "Step": 127,
213
+ "ut": 128,
214
+ "Ġ3": 129,
215
+ "Ġ2": 130,
216
+ "Ġ4": 131,
217
+ "Ġ5": 132,
218
+ "Ġ6": 133,
219
+ "Ġ9": 134,
220
+ "Ġ8": 135,
221
+ "Ġ7": 136,
222
+ "Ġ0": 137,
223
+ "ca": 138,
224
+ "Ġ12": 139,
225
+ "Ġ13": 140,
226
+ "1010": 141,
227
+ "Ġ14": 142,
228
+ "Out": 143,
229
+ "put": 144,
230
+ "ĠOut": 145,
231
+ "ĠSca": 146,
232
+ "ĠOutput": 147,
233
+ "ĠScan": 148,
234
+ "Ġ011": 149,
235
+ "ing": 150,
236
+ "ti": 151,
237
+ "1100": 152,
238
+ "00000000": 153,
239
+ "Ġ15": 154,
240
+ "Ġ000": 155,
241
+ "on": 156,
242
+ "Ġ110": 157,
243
+ "ul": 158,
244
+ "Ġ100": 159,
245
+ "at": 160,
246
+ "tion": 161,
247
+ "Ġ010": 162,
248
+ "1011": 163,
249
+ "ĠC": 164,
250
+ "Ġ00": 165,
251
+ "00011": 166,
252
+ "50": 167,
253
+ "er": 168,
254
+ "10000": 169,
255
+ "nd": 170,
256
+ "Ġ16": 171,
257
+ "ule": 172,
258
+ "ĠR": 173,
259
+ "ed": 174,
260
+ "li": 175,
261
+ "Ġ001": 176,
262
+ "Ġ101": 177,
263
+ "ate": 178,
264
+ "ĠRule": 179,
265
+ "1000": 180,
266
+ "fi": 181,
267
+ "10100": 182,
268
+ "as": 183,
269
+ "Ġ17": 184,
270
+ "ĠM": 185,
271
+ "Ġ1111": 186,
272
+ "Ver": 187,
273
+ "ch": 188,
274
+ "es": 189,
275
+ "ifi": 190,
276
+ "mp": 191,
277
+ "ting": 192,
278
+ "cation": 193,
279
+ "atch": 194,
280
+ "Verifi": 195,
281
+ "Verification": 196,
282
+ "..": 197,
283
+ "11011": 198,
284
+ "41": 199,
285
+ "and": 200,
286
+ "date": 201,
287
+ "idate": 202,
288
+ "ĠCand": 203,
289
+ "ĠCandidate": 204,
290
+ "58": 205,
291
+ "ating": 206,
292
+ "10011": 207,
293
+ "ction": 208,
294
+ "111000": 209,
295
+ "000000": 210,
296
+ "10110": 211,
297
+ "Ġ00000000": 212,
298
+ "is": 213,
299
+ "match": 214,
300
+ "ĠMis": 215,
301
+ "ĠMismatch": 216,
302
+ "111111": 217,
303
+ "Tas": 218,
304
+ "Task": 219,
305
+ "Ġ18": 220,
306
+ "Gre": 221,
307
+ "Pr": 222,
308
+ "ĠL": 223,
309
+ "ĠGre": 224,
310
+ "ĠPr": 225,
311
+ "edy": 226,
312
+ "ĠGreedy": 227,
313
+ "At": 228,
314
+ "Ex": 229,
315
+ "Tes": 230,
316
+ "ĠTes": 231,
317
+ "temp": 232,
318
+ "Attemp": 233,
319
+ "ĠTesting": 234,
320
+ "Attempt": 235,
321
+ "100000": 236,
322
+ "Ġs": 237,
323
+ "Ġ0000": 238,
324
+ "uction": 239,
325
+ "ĠSteps": 240,
326
+ "xt": 241,
327
+ "101111": 242,
328
+ "al": 243,
329
+ "cul": 244,
330
+ "eli": 245,
331
+ "ior": 246,
332
+ "keli": 247,
333
+ "ver": 248,
334
+ "cal": 249,
335
+ "culating": 250,
336
+ "iors": 251,
337
+ "kelihood": 252,
338
+ "Ġ19": 253,
339
+ "11111111": 254,
340
+ ").": 255
341
+ },
342
+ "merges": [
343
+ [
344
+ "0",
345
+ "0"
346
+ ],
347
+ [
348
+ "1",
349
+ "1"
350
+ ],
351
+ [
352
+ "-",
353
+ ">"
354
+ ],
355
+ [
356
+ "Ġ",
357
+ "|"
358
+ ],
359
+ [
360
+ ":",
361
+ "["
362
+ ],
363
+ [
364
+ "]",
365
+ "->"
366
+ ],
367
+ [
368
+ "1",
369
+ "0"
370
+ ],
371
+ [
372
+ "Ġ",
373
+ "1"
374
+ ],
375
+ [
376
+ "00",
377
+ "0"
378
+ ],
379
+ [
380
+ "Ġ",
381
+ "S"
382
+ ],
383
+ [
384
+ "'",
385
+ ","
386
+ ],
387
+ [
388
+ "Ġ",
389
+ "'"
390
+ ],
391
+ [
392
+ "t",
393
+ "e"
394
+ ],
395
+ [
396
+ "1",
397
+ "00"
398
+ ],
399
+ [
400
+ "Ċ",
401
+ "Ġ"
402
+ ],
403
+ [
404
+ "11",
405
+ "0"
406
+ ],
407
+ [
408
+ "te",
409
+ "p"
410
+ ],
411
+ [
412
+ "r",
413
+ "e"
414
+ ],
415
+ [
416
+ "11",
417
+ "1"
418
+ ],
419
+ [
420
+ "00",
421
+ "00"
422
+ ],
423
+ [
424
+ "0",
425
+ "11"
426
+ ],
427
+ [
428
+ "10",
429
+ "1"
430
+ ],
431
+ [
432
+ "00",
433
+ "1"
434
+ ],
435
+ [
436
+ "ĠS",
437
+ "tep"
438
+ ],
439
+ [
440
+ "0",
441
+ "10"
442
+ ],
443
+ [
444
+ "i",
445
+ "n"
446
+ ],
447
+ [
448
+ "i",
449
+ "m"
450
+ ],
451
+ [
452
+ "ĠS",
453
+ "im"
454
+ ],
455
+ [
456
+ "11",
457
+ "11"
458
+ ],
459
+ [
460
+ "Ġ",
461
+ "["
462
+ ],
463
+ [
464
+ "o",
465
+ "r"
466
+ ],
467
+ [
468
+ "Ġ",
469
+ "I"
470
+ ],
471
+ [
472
+ "h",
473
+ "o"
474
+ ],
475
+ [
476
+ "o",
477
+ "d"
478
+ ],
479
+ [
480
+ "ho",
481
+ "od"
482
+ ],
483
+ [
484
+ "Ġ",
485
+ "in"
486
+ ],
487
+ [
488
+ "Ġ",
489
+ "11"
490
+ ],
491
+ [
492
+ "Ġ",
493
+ "10"
494
+ ],
495
+ [
496
+ "n",
497
+ "e"
498
+ ],
499
+ [
500
+ "Ġ",
501
+ "ne"
502
+ ],
503
+ [
504
+ "'",
505
+ "]"
506
+ ],
507
+ [
508
+ "b",
509
+ "or"
510
+ ],
511
+ [
512
+ "d",
513
+ "x"
514
+ ],
515
+ [
516
+ "g",
517
+ "h"
518
+ ],
519
+ [
520
+ "i",
521
+ "re"
522
+ ],
523
+ [
524
+ "i",
525
+ "gh"
526
+ ],
527
+ [
528
+ "q",
529
+ "u"
530
+ ],
531
+ [
532
+ "Ġ",
533
+ "re"
534
+ ],
535
+ [
536
+ "Ġ[",
537
+ "'"
538
+ ],
539
+ [
540
+ "ĠI",
541
+ "dx"
542
+ ],
543
+ [
544
+ "Ġne",
545
+ "igh"
546
+ ],
547
+ [
548
+ "bor",
549
+ "hood"
550
+ ],
551
+ [
552
+ "ire",
553
+ "s"
554
+ ],
555
+ [
556
+ "qu",
557
+ "ires"
558
+ ],
559
+ [
560
+ "Ġre",
561
+ "quires"
562
+ ],
563
+ [
564
+ "Ġneigh",
565
+ "borhood"
566
+ ],
567
+ [
568
+ "S",
569
+ "tep"
570
+ ],
571
+ [
572
+ "u",
573
+ "t"
574
+ ],
575
+ [
576
+ "Ġ",
577
+ "3"
578
+ ],
579
+ [
580
+ "Ġ",
581
+ "2"
582
+ ],
583
+ [
584
+ "Ġ",
585
+ "4"
586
+ ],
587
+ [
588
+ "Ġ",
589
+ "5"
590
+ ],
591
+ [
592
+ "Ġ",
593
+ "6"
594
+ ],
595
+ [
596
+ "Ġ",
597
+ "9"
598
+ ],
599
+ [
600
+ "Ġ",
601
+ "8"
602
+ ],
603
+ [
604
+ "Ġ",
605
+ "7"
606
+ ],
607
+ [
608
+ "Ġ",
609
+ "0"
610
+ ],
611
+ [
612
+ "c",
613
+ "a"
614
+ ],
615
+ [
616
+ "Ġ1",
617
+ "2"
618
+ ],
619
+ [
620
+ "Ġ1",
621
+ "3"
622
+ ],
623
+ [
624
+ "10",
625
+ "10"
626
+ ],
627
+ [
628
+ "Ġ1",
629
+ "4"
630
+ ],
631
+ [
632
+ "O",
633
+ "ut"
634
+ ],
635
+ [
636
+ "p",
637
+ "ut"
638
+ ],
639
+ [
640
+ "Ġ",
641
+ "Out"
642
+ ],
643
+ [
644
+ "ĠS",
645
+ "ca"
646
+ ],
647
+ [
648
+ "ĠOut",
649
+ "put"
650
+ ],
651
+ [
652
+ "ĠSca",
653
+ "n"
654
+ ],
655
+ [
656
+ "Ġ",
657
+ "011"
658
+ ],
659
+ [
660
+ "in",
661
+ "g"
662
+ ],
663
+ [
664
+ "t",
665
+ "i"
666
+ ],
667
+ [
668
+ "11",
669
+ "00"
670
+ ],
671
+ [
672
+ "0000",
673
+ "0000"
674
+ ],
675
+ [
676
+ "Ġ1",
677
+ "5"
678
+ ],
679
+ [
680
+ "Ġ",
681
+ "000"
682
+ ],
683
+ [
684
+ "o",
685
+ "n"
686
+ ],
687
+ [
688
+ "Ġ",
689
+ "110"
690
+ ],
691
+ [
692
+ "u",
693
+ "l"
694
+ ],
695
+ [
696
+ "Ġ1",
697
+ "00"
698
+ ],
699
+ [
700
+ "a",
701
+ "t"
702
+ ],
703
+ [
704
+ "ti",
705
+ "on"
706
+ ],
707
+ [
708
+ "Ġ",
709
+ "010"
710
+ ],
711
+ [
712
+ "10",
713
+ "11"
714
+ ],
715
+ [
716
+ "Ġ",
717
+ "C"
718
+ ],
719
+ [
720
+ "Ġ",
721
+ "00"
722
+ ],
723
+ [
724
+ "000",
725
+ "11"
726
+ ],
727
+ [
728
+ "5",
729
+ "0"
730
+ ],
731
+ [
732
+ "e",
733
+ "r"
734
+ ],
735
+ [
736
+ "100",
737
+ "00"
738
+ ],
739
+ [
740
+ "n",
741
+ "d"
742
+ ],
743
+ [
744
+ "Ġ1",
745
+ "6"
746
+ ],
747
+ [
748
+ "ul",
749
+ "e"
750
+ ],
751
+ [
752
+ "Ġ",
753
+ "R"
754
+ ],
755
+ [
756
+ "e",
757
+ "d"
758
+ ],
759
+ [
760
+ "l",
761
+ "i"
762
+ ],
763
+ [
764
+ "Ġ",
765
+ "001"
766
+ ],
767
+ [
768
+ "Ġ",
769
+ "101"
770
+ ],
771
+ [
772
+ "a",
773
+ "te"
774
+ ],
775
+ [
776
+ "ĠR",
777
+ "ule"
778
+ ],
779
+ [
780
+ "1",
781
+ "000"
782
+ ],
783
+ [
784
+ "f",
785
+ "i"
786
+ ],
787
+ [
788
+ "10",
789
+ "100"
790
+ ],
791
+ [
792
+ "a",
793
+ "s"
794
+ ],
795
+ [
796
+ "Ġ1",
797
+ "7"
798
+ ],
799
+ [
800
+ "Ġ",
801
+ "M"
802
+ ],
803
+ [
804
+ "Ġ",
805
+ "1111"
806
+ ],
807
+ [
808
+ "V",
809
+ "er"
810
+ ],
811
+ [
812
+ "c",
813
+ "h"
814
+ ],
815
+ [
816
+ "e",
817
+ "s"
818
+ ],
819
+ [
820
+ "i",
821
+ "fi"
822
+ ],
823
+ [
824
+ "m",
825
+ "p"
826
+ ],
827
+ [
828
+ "t",
829
+ "ing"
830
+ ],
831
+ [
832
+ "ca",
833
+ "tion"
834
+ ],
835
+ [
836
+ "at",
837
+ "ch"
838
+ ],
839
+ [
840
+ "Ver",
841
+ "ifi"
842
+ ],
843
+ [
844
+ "Verifi",
845
+ "cation"
846
+ ],
847
+ [
848
+ ".",
849
+ "."
850
+ ],
851
+ [
852
+ "110",
853
+ "11"
854
+ ],
855
+ [
856
+ "4",
857
+ "1"
858
+ ],
859
+ [
860
+ "a",
861
+ "nd"
862
+ ],
863
+ [
864
+ "d",
865
+ "ate"
866
+ ],
867
+ [
868
+ "i",
869
+ "date"
870
+ ],
871
+ [
872
+ "ĠC",
873
+ "and"
874
+ ],
875
+ [
876
+ "ĠCand",
877
+ "idate"
878
+ ],
879
+ [
880
+ "5",
881
+ "8"
882
+ ],
883
+ [
884
+ "at",
885
+ "ing"
886
+ ],
887
+ [
888
+ "100",
889
+ "11"
890
+ ],
891
+ [
892
+ "c",
893
+ "tion"
894
+ ],
895
+ [
896
+ "111",
897
+ "000"
898
+ ],
899
+ [
900
+ "0000",
901
+ "00"
902
+ ],
903
+ [
904
+ "10",
905
+ "110"
906
+ ],
907
+ [
908
+ "Ġ",
909
+ "00000000"
910
+ ],
911
+ [
912
+ "i",
913
+ "s"
914
+ ],
915
+ [
916
+ "m",
917
+ "atch"
918
+ ],
919
+ [
920
+ "ĠM",
921
+ "is"
922
+ ],
923
+ [
924
+ "ĠMis",
925
+ "match"
926
+ ],
927
+ [
928
+ "1111",
929
+ "11"
930
+ ],
931
+ [
932
+ "T",
933
+ "as"
934
+ ],
935
+ [
936
+ "Tas",
937
+ "k"
938
+ ],
939
+ [
940
+ "Ġ1",
941
+ "8"
942
+ ],
943
+ [
944
+ "G",
945
+ "re"
946
+ ],
947
+ [
948
+ "P",
949
+ "r"
950
+ ],
951
+ [
952
+ "Ġ",
953
+ "L"
954
+ ],
955
+ [
956
+ "Ġ",
957
+ "Gre"
958
+ ],
959
+ [
960
+ "Ġ",
961
+ "Pr"
962
+ ],
963
+ [
964
+ "ed",
965
+ "y"
966
+ ],
967
+ [
968
+ "ĠGre",
969
+ "edy"
970
+ ],
971
+ [
972
+ "A",
973
+ "t"
974
+ ],
975
+ [
976
+ "E",
977
+ "x"
978
+ ],
979
+ [
980
+ "T",
981
+ "es"
982
+ ],
983
+ [
984
+ "Ġ",
985
+ "Tes"
986
+ ],
987
+ [
988
+ "te",
989
+ "mp"
990
+ ],
991
+ [
992
+ "At",
993
+ "temp"
994
+ ],
995
+ [
996
+ "ĠTes",
997
+ "ting"
998
+ ],
999
+ [
1000
+ "Attemp",
1001
+ "t"
1002
+ ],
1003
+ [
1004
+ "100",
1005
+ "000"
1006
+ ],
1007
+ [
1008
+ "Ġ",
1009
+ "s"
1010
+ ],
1011
+ [
1012
+ "Ġ",
1013
+ "0000"
1014
+ ],
1015
+ [
1016
+ "u",
1017
+ "ction"
1018
+ ],
1019
+ [
1020
+ "ĠStep",
1021
+ "s"
1022
+ ],
1023
+ [
1024
+ "x",
1025
+ "t"
1026
+ ],
1027
+ [
1028
+ "10",
1029
+ "1111"
1030
+ ],
1031
+ [
1032
+ "a",
1033
+ "l"
1034
+ ],
1035
+ [
1036
+ "c",
1037
+ "ul"
1038
+ ],
1039
+ [
1040
+ "e",
1041
+ "li"
1042
+ ],
1043
+ [
1044
+ "i",
1045
+ "or"
1046
+ ],
1047
+ [
1048
+ "k",
1049
+ "eli"
1050
+ ],
1051
+ [
1052
+ "v",
1053
+ "er"
1054
+ ],
1055
+ [
1056
+ "ca",
1057
+ "l"
1058
+ ],
1059
+ [
1060
+ "cul",
1061
+ "ating"
1062
+ ],
1063
+ [
1064
+ "ior",
1065
+ "s"
1066
+ ],
1067
+ [
1068
+ "keli",
1069
+ "hood"
1070
+ ],
1071
+ [
1072
+ "Ġ1",
1073
+ "9"
1074
+ ],
1075
+ [
1076
+ "1111",
1077
+ "1111"
1078
+ ],
1079
+ [
1080
+ ")",
1081
+ "."
1082
+ ]
1083
+ ]
1084
+ }
1085
+ }
checkpoint-15627/tokenizer_config.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<|startoftext|>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<|endoftext|>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "<|unknown|>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "<|pad|>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ },
35
+ "4": {
36
+ "content": "<think>",
37
+ "lstrip": false,
38
+ "normalized": false,
39
+ "rstrip": false,
40
+ "single_word": false,
41
+ "special": true
42
+ },
43
+ "5": {
44
+ "content": "</think>",
45
+ "lstrip": false,
46
+ "normalized": false,
47
+ "rstrip": false,
48
+ "single_word": false,
49
+ "special": true
50
+ }
51
+ },
52
+ "additional_special_tokens": [
53
+ "<think>",
54
+ "</think>"
55
+ ],
56
+ "bos_token": "<|startoftext|>",
57
+ "clean_up_tokenization_spaces": false,
58
+ "eos_token": "<|endoftext|>",
59
+ "extra_special_tokens": {},
60
+ "model_max_length": 1000000000000000019884624838656,
61
+ "pad_token": "<|pad|>",
62
+ "tokenizer_class": "PreTrainedTokenizerFast",
63
+ "unk_token": "<|unknown|>"
64
+ }
checkpoint-15627/trainer_state.json ADDED
@@ -0,0 +1,2218 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 3.0,
6
+ "eval_steps": 500,
7
+ "global_step": 15627,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.009599692809830085,
14
+ "grad_norm": 7.351896286010742,
15
+ "learning_rate": 1.47e-05,
16
+ "loss": 19.9984,
17
+ "step": 50
18
+ },
19
+ {
20
+ "epoch": 0.01919938561966017,
21
+ "grad_norm": 41.9318733215332,
22
+ "learning_rate": 2.97e-05,
23
+ "loss": 15.992,
24
+ "step": 100
25
+ },
26
+ {
27
+ "epoch": 0.028799078429490255,
28
+ "grad_norm": 41.05648422241211,
29
+ "learning_rate": 4.4699999999999996e-05,
30
+ "loss": 12.7699,
31
+ "step": 150
32
+ },
33
+ {
34
+ "epoch": 0.03839877123932034,
35
+ "grad_norm": 30.36406135559082,
36
+ "learning_rate": 5.97e-05,
37
+ "loss": 10.156,
38
+ "step": 200
39
+ },
40
+ {
41
+ "epoch": 0.047998464049150424,
42
+ "grad_norm": 26.620258331298828,
43
+ "learning_rate": 7.47e-05,
44
+ "loss": 7.7426,
45
+ "step": 250
46
+ },
47
+ {
48
+ "epoch": 0.05759815685898051,
49
+ "grad_norm": 17.737770080566406,
50
+ "learning_rate": 8.969999999999998e-05,
51
+ "loss": 5.743,
52
+ "step": 300
53
+ },
54
+ {
55
+ "epoch": 0.0671978496688106,
56
+ "grad_norm": 16.273975372314453,
57
+ "learning_rate": 0.00010469999999999998,
58
+ "loss": 4.1183,
59
+ "step": 350
60
+ },
61
+ {
62
+ "epoch": 0.07679754247864068,
63
+ "grad_norm": 7.628762722015381,
64
+ "learning_rate": 0.0001197,
65
+ "loss": 3.0495,
66
+ "step": 400
67
+ },
68
+ {
69
+ "epoch": 0.08639723528847076,
70
+ "grad_norm": 11.860695838928223,
71
+ "learning_rate": 0.0001347,
72
+ "loss": 2.3835,
73
+ "step": 450
74
+ },
75
+ {
76
+ "epoch": 0.09599692809830085,
77
+ "grad_norm": 9.723978042602539,
78
+ "learning_rate": 0.00014969999999999998,
79
+ "loss": 2.0257,
80
+ "step": 500
81
+ },
82
+ {
83
+ "epoch": 0.10559662090813093,
84
+ "grad_norm": 8.171798706054688,
85
+ "learning_rate": 0.0001647,
86
+ "loss": 1.7721,
87
+ "step": 550
88
+ },
89
+ {
90
+ "epoch": 0.11519631371796102,
91
+ "grad_norm": 5.052175521850586,
92
+ "learning_rate": 0.00017969999999999998,
93
+ "loss": 1.6081,
94
+ "step": 600
95
+ },
96
+ {
97
+ "epoch": 0.12479600652779112,
98
+ "grad_norm": 6.628243446350098,
99
+ "learning_rate": 0.0001947,
100
+ "loss": 1.4632,
101
+ "step": 650
102
+ },
103
+ {
104
+ "epoch": 0.1343956993376212,
105
+ "grad_norm": 5.753493309020996,
106
+ "learning_rate": 0.00020969999999999997,
107
+ "loss": 1.3574,
108
+ "step": 700
109
+ },
110
+ {
111
+ "epoch": 0.14399539214745127,
112
+ "grad_norm": 3.7359797954559326,
113
+ "learning_rate": 0.0002247,
114
+ "loss": 1.2551,
115
+ "step": 750
116
+ },
117
+ {
118
+ "epoch": 0.15359508495728136,
119
+ "grad_norm": 8.183115005493164,
120
+ "learning_rate": 0.0002397,
121
+ "loss": 1.1362,
122
+ "step": 800
123
+ },
124
+ {
125
+ "epoch": 0.16319477776711144,
126
+ "grad_norm": 5.23441219329834,
127
+ "learning_rate": 0.00025469999999999996,
128
+ "loss": 1.0032,
129
+ "step": 850
130
+ },
131
+ {
132
+ "epoch": 0.17279447057694153,
133
+ "grad_norm": 3.2111752033233643,
134
+ "learning_rate": 0.0002697,
135
+ "loss": 0.9582,
136
+ "step": 900
137
+ },
138
+ {
139
+ "epoch": 0.1823941633867716,
140
+ "grad_norm": 3.7051522731781006,
141
+ "learning_rate": 0.0002847,
142
+ "loss": 0.8247,
143
+ "step": 950
144
+ },
145
+ {
146
+ "epoch": 0.1919938561966017,
147
+ "grad_norm": 2.5947470664978027,
148
+ "learning_rate": 0.00029969999999999997,
149
+ "loss": 0.68,
150
+ "step": 1000
151
+ },
152
+ {
153
+ "epoch": 0.20159354900643178,
154
+ "grad_norm": 2.0631327629089355,
155
+ "learning_rate": 0.00029899500922950706,
156
+ "loss": 0.5732,
157
+ "step": 1050
158
+ },
159
+ {
160
+ "epoch": 0.21119324181626187,
161
+ "grad_norm": 1.6905947923660278,
162
+ "learning_rate": 0.0002979695084432898,
163
+ "loss": 0.5127,
164
+ "step": 1100
165
+ },
166
+ {
167
+ "epoch": 0.22079293462609195,
168
+ "grad_norm": 1.6837217807769775,
169
+ "learning_rate": 0.0002969440076570725,
170
+ "loss": 0.4481,
171
+ "step": 1150
172
+ },
173
+ {
174
+ "epoch": 0.23039262743592204,
175
+ "grad_norm": 1.147798776626587,
176
+ "learning_rate": 0.00029591850687085523,
177
+ "loss": 0.4135,
178
+ "step": 1200
179
+ },
180
+ {
181
+ "epoch": 0.23999232024575212,
182
+ "grad_norm": 2.4957432746887207,
183
+ "learning_rate": 0.00029489300608463796,
184
+ "loss": 0.3848,
185
+ "step": 1250
186
+ },
187
+ {
188
+ "epoch": 0.24959201305558223,
189
+ "grad_norm": 1.402390480041504,
190
+ "learning_rate": 0.0002938675052984207,
191
+ "loss": 0.3413,
192
+ "step": 1300
193
+ },
194
+ {
195
+ "epoch": 0.2591917058654123,
196
+ "grad_norm": 2.157576084136963,
197
+ "learning_rate": 0.0002928420045122034,
198
+ "loss": 0.3,
199
+ "step": 1350
200
+ },
201
+ {
202
+ "epoch": 0.2687913986752424,
203
+ "grad_norm": 1.074357032775879,
204
+ "learning_rate": 0.0002918165037259862,
205
+ "loss": 0.2714,
206
+ "step": 1400
207
+ },
208
+ {
209
+ "epoch": 0.27839109148507246,
210
+ "grad_norm": 0.9258896112442017,
211
+ "learning_rate": 0.0002907910029397689,
212
+ "loss": 0.2665,
213
+ "step": 1450
214
+ },
215
+ {
216
+ "epoch": 0.28799078429490255,
217
+ "grad_norm": 1.9370087385177612,
218
+ "learning_rate": 0.0002897655021535516,
219
+ "loss": 0.2627,
220
+ "step": 1500
221
+ },
222
+ {
223
+ "epoch": 0.29759047710473263,
224
+ "grad_norm": 1.3033190965652466,
225
+ "learning_rate": 0.00028874000136733436,
226
+ "loss": 0.2345,
227
+ "step": 1550
228
+ },
229
+ {
230
+ "epoch": 0.3071901699145627,
231
+ "grad_norm": 1.0974582433700562,
232
+ "learning_rate": 0.0002877145005811171,
233
+ "loss": 0.2414,
234
+ "step": 1600
235
+ },
236
+ {
237
+ "epoch": 0.3167898627243928,
238
+ "grad_norm": 0.8440233469009399,
239
+ "learning_rate": 0.0002866889997948998,
240
+ "loss": 0.2372,
241
+ "step": 1650
242
+ },
243
+ {
244
+ "epoch": 0.3263895555342229,
245
+ "grad_norm": 1.1249887943267822,
246
+ "learning_rate": 0.00028566349900868254,
247
+ "loss": 0.2265,
248
+ "step": 1700
249
+ },
250
+ {
251
+ "epoch": 0.33598924834405297,
252
+ "grad_norm": 0.7356661558151245,
253
+ "learning_rate": 0.00028463799822246526,
254
+ "loss": 0.2182,
255
+ "step": 1750
256
+ },
257
+ {
258
+ "epoch": 0.34558894115388306,
259
+ "grad_norm": 0.7215508818626404,
260
+ "learning_rate": 0.00028361249743624804,
261
+ "loss": 0.2077,
262
+ "step": 1800
263
+ },
264
+ {
265
+ "epoch": 0.35518863396371314,
266
+ "grad_norm": 0.8716081976890564,
267
+ "learning_rate": 0.0002825869966500307,
268
+ "loss": 0.206,
269
+ "step": 1850
270
+ },
271
+ {
272
+ "epoch": 0.3647883267735432,
273
+ "grad_norm": 0.5157195329666138,
274
+ "learning_rate": 0.00028156149586381344,
275
+ "loss": 0.2003,
276
+ "step": 1900
277
+ },
278
+ {
279
+ "epoch": 0.3743880195833733,
280
+ "grad_norm": 0.47487232089042664,
281
+ "learning_rate": 0.0002805359950775962,
282
+ "loss": 0.1955,
283
+ "step": 1950
284
+ },
285
+ {
286
+ "epoch": 0.3839877123932034,
287
+ "grad_norm": 0.5665110349655151,
288
+ "learning_rate": 0.00027951049429137894,
289
+ "loss": 0.1884,
290
+ "step": 2000
291
+ },
292
+ {
293
+ "epoch": 0.3935874052030335,
294
+ "grad_norm": 0.772770345211029,
295
+ "learning_rate": 0.00027848499350516167,
296
+ "loss": 0.1893,
297
+ "step": 2050
298
+ },
299
+ {
300
+ "epoch": 0.40318709801286357,
301
+ "grad_norm": 0.5221526622772217,
302
+ "learning_rate": 0.0002774594927189444,
303
+ "loss": 0.1826,
304
+ "step": 2100
305
+ },
306
+ {
307
+ "epoch": 0.41278679082269365,
308
+ "grad_norm": 0.5089606642723083,
309
+ "learning_rate": 0.0002764339919327271,
310
+ "loss": 0.1707,
311
+ "step": 2150
312
+ },
313
+ {
314
+ "epoch": 0.42238648363252373,
315
+ "grad_norm": 0.42394664883613586,
316
+ "learning_rate": 0.00027540849114650985,
317
+ "loss": 0.1672,
318
+ "step": 2200
319
+ },
320
+ {
321
+ "epoch": 0.4319861764423538,
322
+ "grad_norm": 0.40413081645965576,
323
+ "learning_rate": 0.00027438299036029257,
324
+ "loss": 0.1596,
325
+ "step": 2250
326
+ },
327
+ {
328
+ "epoch": 0.4415858692521839,
329
+ "grad_norm": 0.3184243142604828,
330
+ "learning_rate": 0.0002733574895740753,
331
+ "loss": 0.153,
332
+ "step": 2300
333
+ },
334
+ {
335
+ "epoch": 0.451185562062014,
336
+ "grad_norm": 0.43532413244247437,
337
+ "learning_rate": 0.0002723319887878581,
338
+ "loss": 0.1512,
339
+ "step": 2350
340
+ },
341
+ {
342
+ "epoch": 0.4607852548718441,
343
+ "grad_norm": 0.7987850904464722,
344
+ "learning_rate": 0.0002713064880016408,
345
+ "loss": 0.1395,
346
+ "step": 2400
347
+ },
348
+ {
349
+ "epoch": 0.47038494768167416,
350
+ "grad_norm": 0.37470191717147827,
351
+ "learning_rate": 0.00027028098721542347,
352
+ "loss": 0.1426,
353
+ "step": 2450
354
+ },
355
+ {
356
+ "epoch": 0.47998464049150424,
357
+ "grad_norm": 0.32686635851860046,
358
+ "learning_rate": 0.00026925548642920625,
359
+ "loss": 0.1364,
360
+ "step": 2500
361
+ },
362
+ {
363
+ "epoch": 0.4895843333013344,
364
+ "grad_norm": 0.6824387907981873,
365
+ "learning_rate": 0.000268229985642989,
366
+ "loss": 0.1414,
367
+ "step": 2550
368
+ },
369
+ {
370
+ "epoch": 0.49918402611116447,
371
+ "grad_norm": 0.3873808979988098,
372
+ "learning_rate": 0.0002672044848567717,
373
+ "loss": 0.1413,
374
+ "step": 2600
375
+ },
376
+ {
377
+ "epoch": 0.5087837189209945,
378
+ "grad_norm": 0.30228692293167114,
379
+ "learning_rate": 0.0002661789840705544,
380
+ "loss": 0.1355,
381
+ "step": 2650
382
+ },
383
+ {
384
+ "epoch": 0.5183834117308246,
385
+ "grad_norm": 0.2715003788471222,
386
+ "learning_rate": 0.00026515348328433715,
387
+ "loss": 0.1302,
388
+ "step": 2700
389
+ },
390
+ {
391
+ "epoch": 0.5279831045406547,
392
+ "grad_norm": 0.35757285356521606,
393
+ "learning_rate": 0.00026412798249811993,
394
+ "loss": 0.1317,
395
+ "step": 2750
396
+ },
397
+ {
398
+ "epoch": 0.5375827973504848,
399
+ "grad_norm": 0.42059722542762756,
400
+ "learning_rate": 0.0002631024817119026,
401
+ "loss": 0.1289,
402
+ "step": 2800
403
+ },
404
+ {
405
+ "epoch": 0.5471824901603148,
406
+ "grad_norm": 0.680482029914856,
407
+ "learning_rate": 0.0002620769809256853,
408
+ "loss": 0.1312,
409
+ "step": 2850
410
+ },
411
+ {
412
+ "epoch": 0.5567821829701449,
413
+ "grad_norm": 0.7014706134796143,
414
+ "learning_rate": 0.0002610514801394681,
415
+ "loss": 0.1343,
416
+ "step": 2900
417
+ },
418
+ {
419
+ "epoch": 0.566381875779975,
420
+ "grad_norm": 0.4586208164691925,
421
+ "learning_rate": 0.00026002597935325083,
422
+ "loss": 0.1266,
423
+ "step": 2950
424
+ },
425
+ {
426
+ "epoch": 0.5759815685898051,
427
+ "grad_norm": 0.37353283166885376,
428
+ "learning_rate": 0.00025900047856703356,
429
+ "loss": 0.1328,
430
+ "step": 3000
431
+ },
432
+ {
433
+ "epoch": 0.5855812613996352,
434
+ "grad_norm": 0.30957943201065063,
435
+ "learning_rate": 0.0002579749777808163,
436
+ "loss": 0.1218,
437
+ "step": 3050
438
+ },
439
+ {
440
+ "epoch": 0.5951809542094653,
441
+ "grad_norm": 0.35607343912124634,
442
+ "learning_rate": 0.000256949476994599,
443
+ "loss": 0.1235,
444
+ "step": 3100
445
+ },
446
+ {
447
+ "epoch": 0.6047806470192953,
448
+ "grad_norm": 0.3144589066505432,
449
+ "learning_rate": 0.00025592397620838173,
450
+ "loss": 0.1248,
451
+ "step": 3150
452
+ },
453
+ {
454
+ "epoch": 0.6143803398291254,
455
+ "grad_norm": 0.5591529607772827,
456
+ "learning_rate": 0.00025489847542216446,
457
+ "loss": 0.125,
458
+ "step": 3200
459
+ },
460
+ {
461
+ "epoch": 0.6239800326389555,
462
+ "grad_norm": 0.4985520839691162,
463
+ "learning_rate": 0.0002538729746359472,
464
+ "loss": 0.1328,
465
+ "step": 3250
466
+ },
467
+ {
468
+ "epoch": 0.6335797254487856,
469
+ "grad_norm": 0.3131736218929291,
470
+ "learning_rate": 0.00025284747384972996,
471
+ "loss": 0.1376,
472
+ "step": 3300
473
+ },
474
+ {
475
+ "epoch": 0.6431794182586157,
476
+ "grad_norm": 1.0179088115692139,
477
+ "learning_rate": 0.0002518219730635127,
478
+ "loss": 0.1393,
479
+ "step": 3350
480
+ },
481
+ {
482
+ "epoch": 0.6527791110684458,
483
+ "grad_norm": 0.40214893221855164,
484
+ "learning_rate": 0.00025079647227729536,
485
+ "loss": 0.1383,
486
+ "step": 3400
487
+ },
488
+ {
489
+ "epoch": 0.6623788038782759,
490
+ "grad_norm": 0.41477128863334656,
491
+ "learning_rate": 0.00024977097149107814,
492
+ "loss": 0.1252,
493
+ "step": 3450
494
+ },
495
+ {
496
+ "epoch": 0.6719784966881059,
497
+ "grad_norm": 0.3146192133426666,
498
+ "learning_rate": 0.00024874547070486086,
499
+ "loss": 0.1283,
500
+ "step": 3500
501
+ },
502
+ {
503
+ "epoch": 0.681578189497936,
504
+ "grad_norm": 0.34901753067970276,
505
+ "learning_rate": 0.0002477199699186436,
506
+ "loss": 0.1185,
507
+ "step": 3550
508
+ },
509
+ {
510
+ "epoch": 0.6911778823077661,
511
+ "grad_norm": 0.4564303159713745,
512
+ "learning_rate": 0.0002466944691324263,
513
+ "loss": 0.1176,
514
+ "step": 3600
515
+ },
516
+ {
517
+ "epoch": 0.7007775751175962,
518
+ "grad_norm": 0.20335577428340912,
519
+ "learning_rate": 0.00024566896834620904,
520
+ "loss": 0.1184,
521
+ "step": 3650
522
+ },
523
+ {
524
+ "epoch": 0.7103772679274263,
525
+ "grad_norm": 0.16178852319717407,
526
+ "learning_rate": 0.00024464346755999176,
527
+ "loss": 0.1136,
528
+ "step": 3700
529
+ },
530
+ {
531
+ "epoch": 0.7199769607372564,
532
+ "grad_norm": 0.270074725151062,
533
+ "learning_rate": 0.00024361796677377452,
534
+ "loss": 0.116,
535
+ "step": 3750
536
+ },
537
+ {
538
+ "epoch": 0.7295766535470865,
539
+ "grad_norm": 0.6328276991844177,
540
+ "learning_rate": 0.00024259246598755724,
541
+ "loss": 0.1152,
542
+ "step": 3800
543
+ },
544
+ {
545
+ "epoch": 0.7391763463569165,
546
+ "grad_norm": 0.290368914604187,
547
+ "learning_rate": 0.00024156696520133997,
548
+ "loss": 0.1145,
549
+ "step": 3850
550
+ },
551
+ {
552
+ "epoch": 0.7487760391667466,
553
+ "grad_norm": 0.3527914881706238,
554
+ "learning_rate": 0.0002405414644151227,
555
+ "loss": 0.1145,
556
+ "step": 3900
557
+ },
558
+ {
559
+ "epoch": 0.7583757319765767,
560
+ "grad_norm": 0.2402893602848053,
561
+ "learning_rate": 0.00023951596362890542,
562
+ "loss": 0.1159,
563
+ "step": 3950
564
+ },
565
+ {
566
+ "epoch": 0.7679754247864068,
567
+ "grad_norm": 0.19898560643196106,
568
+ "learning_rate": 0.00023849046284268817,
569
+ "loss": 0.1141,
570
+ "step": 4000
571
+ },
572
+ {
573
+ "epoch": 0.7775751175962369,
574
+ "grad_norm": 0.19476300477981567,
575
+ "learning_rate": 0.0002374649620564709,
576
+ "loss": 0.1158,
577
+ "step": 4050
578
+ },
579
+ {
580
+ "epoch": 0.787174810406067,
581
+ "grad_norm": 0.3251754343509674,
582
+ "learning_rate": 0.0002364394612702536,
583
+ "loss": 0.1137,
584
+ "step": 4100
585
+ },
586
+ {
587
+ "epoch": 0.796774503215897,
588
+ "grad_norm": 0.787211537361145,
589
+ "learning_rate": 0.00023541396048403634,
590
+ "loss": 0.1199,
591
+ "step": 4150
592
+ },
593
+ {
594
+ "epoch": 0.8063741960257271,
595
+ "grad_norm": 0.27289190888404846,
596
+ "learning_rate": 0.00023438845969781907,
597
+ "loss": 0.1152,
598
+ "step": 4200
599
+ },
600
+ {
601
+ "epoch": 0.8159738888355572,
602
+ "grad_norm": 0.17066220939159393,
603
+ "learning_rate": 0.00023336295891160182,
604
+ "loss": 0.1098,
605
+ "step": 4250
606
+ },
607
+ {
608
+ "epoch": 0.8255735816453873,
609
+ "grad_norm": 0.5239670276641846,
610
+ "learning_rate": 0.00023233745812538455,
611
+ "loss": 0.109,
612
+ "step": 4300
613
+ },
614
+ {
615
+ "epoch": 0.8351732744552174,
616
+ "grad_norm": 0.5953475832939148,
617
+ "learning_rate": 0.00023131195733916727,
618
+ "loss": 0.1174,
619
+ "step": 4350
620
+ },
621
+ {
622
+ "epoch": 0.8447729672650475,
623
+ "grad_norm": 0.3444459140300751,
624
+ "learning_rate": 0.00023028645655295002,
625
+ "loss": 0.1156,
626
+ "step": 4400
627
+ },
628
+ {
629
+ "epoch": 0.8543726600748776,
630
+ "grad_norm": 0.1797035187482834,
631
+ "learning_rate": 0.00022926095576673272,
632
+ "loss": 0.1117,
633
+ "step": 4450
634
+ },
635
+ {
636
+ "epoch": 0.8639723528847076,
637
+ "grad_norm": 0.1553630381822586,
638
+ "learning_rate": 0.00022823545498051545,
639
+ "loss": 0.1099,
640
+ "step": 4500
641
+ },
642
+ {
643
+ "epoch": 0.8735720456945377,
644
+ "grad_norm": 0.2846544086933136,
645
+ "learning_rate": 0.0002272099541942982,
646
+ "loss": 0.1118,
647
+ "step": 4550
648
+ },
649
+ {
650
+ "epoch": 0.8831717385043678,
651
+ "grad_norm": 0.18632030487060547,
652
+ "learning_rate": 0.00022618445340808093,
653
+ "loss": 0.1108,
654
+ "step": 4600
655
+ },
656
+ {
657
+ "epoch": 0.8927714313141979,
658
+ "grad_norm": 0.3297656178474426,
659
+ "learning_rate": 0.00022515895262186365,
660
+ "loss": 0.1159,
661
+ "step": 4650
662
+ },
663
+ {
664
+ "epoch": 0.902371124124028,
665
+ "grad_norm": 0.5725330710411072,
666
+ "learning_rate": 0.0002241334518356464,
667
+ "loss": 0.1256,
668
+ "step": 4700
669
+ },
670
+ {
671
+ "epoch": 0.9119708169338581,
672
+ "grad_norm": 0.4595456123352051,
673
+ "learning_rate": 0.0002231079510494291,
674
+ "loss": 0.1293,
675
+ "step": 4750
676
+ },
677
+ {
678
+ "epoch": 0.9215705097436881,
679
+ "grad_norm": 0.2964916527271271,
680
+ "learning_rate": 0.00022208245026321185,
681
+ "loss": 0.1123,
682
+ "step": 4800
683
+ },
684
+ {
685
+ "epoch": 0.9311702025535182,
686
+ "grad_norm": 0.18435810506343842,
687
+ "learning_rate": 0.00022105694947699458,
688
+ "loss": 0.1141,
689
+ "step": 4850
690
+ },
691
+ {
692
+ "epoch": 0.9407698953633483,
693
+ "grad_norm": 0.18365661799907684,
694
+ "learning_rate": 0.0002200314486907773,
695
+ "loss": 0.109,
696
+ "step": 4900
697
+ },
698
+ {
699
+ "epoch": 0.9503695881731784,
700
+ "grad_norm": 0.19481471180915833,
701
+ "learning_rate": 0.00021900594790456006,
702
+ "loss": 0.1095,
703
+ "step": 4950
704
+ },
705
+ {
706
+ "epoch": 0.9599692809830085,
707
+ "grad_norm": 0.15485212206840515,
708
+ "learning_rate": 0.00021798044711834278,
709
+ "loss": 0.1076,
710
+ "step": 5000
711
+ },
712
+ {
713
+ "epoch": 0.9695689737928387,
714
+ "grad_norm": 0.15106704831123352,
715
+ "learning_rate": 0.00021695494633212548,
716
+ "loss": 0.1093,
717
+ "step": 5050
718
+ },
719
+ {
720
+ "epoch": 0.9791686666026688,
721
+ "grad_norm": 0.1416679173707962,
722
+ "learning_rate": 0.00021592944554590823,
723
+ "loss": 0.1072,
724
+ "step": 5100
725
+ },
726
+ {
727
+ "epoch": 0.9887683594124989,
728
+ "grad_norm": 0.2634902596473694,
729
+ "learning_rate": 0.00021490394475969096,
730
+ "loss": 0.1106,
731
+ "step": 5150
732
+ },
733
+ {
734
+ "epoch": 0.9983680522223289,
735
+ "grad_norm": 0.2642424404621124,
736
+ "learning_rate": 0.00021387844397347368,
737
+ "loss": 0.1088,
738
+ "step": 5200
739
+ },
740
+ {
741
+ "epoch": 1.0078717481040607,
742
+ "grad_norm": 0.16377754509449005,
743
+ "learning_rate": 0.00021285294318725643,
744
+ "loss": 0.1043,
745
+ "step": 5250
746
+ },
747
+ {
748
+ "epoch": 1.0174714409138907,
749
+ "grad_norm": 0.2642650008201599,
750
+ "learning_rate": 0.00021182744240103916,
751
+ "loss": 0.1055,
752
+ "step": 5300
753
+ },
754
+ {
755
+ "epoch": 1.0270711337237208,
756
+ "grad_norm": 0.14765985310077667,
757
+ "learning_rate": 0.0002108019416148219,
758
+ "loss": 0.1065,
759
+ "step": 5350
760
+ },
761
+ {
762
+ "epoch": 1.036670826533551,
763
+ "grad_norm": 0.1779513657093048,
764
+ "learning_rate": 0.0002097764408286046,
765
+ "loss": 0.1062,
766
+ "step": 5400
767
+ },
768
+ {
769
+ "epoch": 1.046270519343381,
770
+ "grad_norm": 0.7090527415275574,
771
+ "learning_rate": 0.00020875094004238733,
772
+ "loss": 0.1098,
773
+ "step": 5450
774
+ },
775
+ {
776
+ "epoch": 1.055870212153211,
777
+ "grad_norm": 0.5224199891090393,
778
+ "learning_rate": 0.0002077254392561701,
779
+ "loss": 0.138,
780
+ "step": 5500
781
+ },
782
+ {
783
+ "epoch": 1.0654699049630412,
784
+ "grad_norm": 0.2672184109687805,
785
+ "learning_rate": 0.0002066999384699528,
786
+ "loss": 0.1349,
787
+ "step": 5550
788
+ },
789
+ {
790
+ "epoch": 1.0750695977728713,
791
+ "grad_norm": 0.19684670865535736,
792
+ "learning_rate": 0.00020567443768373554,
793
+ "loss": 0.109,
794
+ "step": 5600
795
+ },
796
+ {
797
+ "epoch": 1.0846692905827013,
798
+ "grad_norm": 0.11916083097457886,
799
+ "learning_rate": 0.0002046489368975183,
800
+ "loss": 0.1058,
801
+ "step": 5650
802
+ },
803
+ {
804
+ "epoch": 1.0942689833925314,
805
+ "grad_norm": 0.17690648138523102,
806
+ "learning_rate": 0.000203623436111301,
807
+ "loss": 0.1061,
808
+ "step": 5700
809
+ },
810
+ {
811
+ "epoch": 1.1038686762023615,
812
+ "grad_norm": 0.20282995700836182,
813
+ "learning_rate": 0.0002025979353250837,
814
+ "loss": 0.1063,
815
+ "step": 5750
816
+ },
817
+ {
818
+ "epoch": 1.1134683690121916,
819
+ "grad_norm": 0.17138956487178802,
820
+ "learning_rate": 0.00020157243453886647,
821
+ "loss": 0.1069,
822
+ "step": 5800
823
+ },
824
+ {
825
+ "epoch": 1.1230680618220217,
826
+ "grad_norm": 0.1042153611779213,
827
+ "learning_rate": 0.0002005469337526492,
828
+ "loss": 0.1036,
829
+ "step": 5850
830
+ },
831
+ {
832
+ "epoch": 1.1326677546318518,
833
+ "grad_norm": 0.15628878772258759,
834
+ "learning_rate": 0.00019952143296643194,
835
+ "loss": 0.104,
836
+ "step": 5900
837
+ },
838
+ {
839
+ "epoch": 1.1422674474416818,
840
+ "grad_norm": 0.19753845036029816,
841
+ "learning_rate": 0.00019849593218021467,
842
+ "loss": 0.107,
843
+ "step": 5950
844
+ },
845
+ {
846
+ "epoch": 1.151867140251512,
847
+ "grad_norm": 0.3391304612159729,
848
+ "learning_rate": 0.00019747043139399737,
849
+ "loss": 0.1097,
850
+ "step": 6000
851
+ },
852
+ {
853
+ "epoch": 1.161466833061342,
854
+ "grad_norm": 0.16991937160491943,
855
+ "learning_rate": 0.00019644493060778012,
856
+ "loss": 0.1051,
857
+ "step": 6050
858
+ },
859
+ {
860
+ "epoch": 1.171066525871172,
861
+ "grad_norm": 0.2567026913166046,
862
+ "learning_rate": 0.00019541942982156284,
863
+ "loss": 0.1049,
864
+ "step": 6100
865
+ },
866
+ {
867
+ "epoch": 1.1806662186810022,
868
+ "grad_norm": 0.1503693014383316,
869
+ "learning_rate": 0.00019439392903534557,
870
+ "loss": 0.1067,
871
+ "step": 6150
872
+ },
873
+ {
874
+ "epoch": 1.1902659114908323,
875
+ "grad_norm": 0.22622846066951752,
876
+ "learning_rate": 0.00019336842824912832,
877
+ "loss": 0.1059,
878
+ "step": 6200
879
+ },
880
+ {
881
+ "epoch": 1.1998656043006624,
882
+ "grad_norm": 0.15262149274349213,
883
+ "learning_rate": 0.00019234292746291105,
884
+ "loss": 0.1056,
885
+ "step": 6250
886
+ },
887
+ {
888
+ "epoch": 1.2094652971104924,
889
+ "grad_norm": 0.313942015171051,
890
+ "learning_rate": 0.00019131742667669374,
891
+ "loss": 0.1041,
892
+ "step": 6300
893
+ },
894
+ {
895
+ "epoch": 1.2190649899203225,
896
+ "grad_norm": 0.17761795222759247,
897
+ "learning_rate": 0.0001902919258904765,
898
+ "loss": 0.1046,
899
+ "step": 6350
900
+ },
901
+ {
902
+ "epoch": 1.2286646827301526,
903
+ "grad_norm": 0.20730365812778473,
904
+ "learning_rate": 0.00018926642510425922,
905
+ "loss": 0.1041,
906
+ "step": 6400
907
+ },
908
+ {
909
+ "epoch": 1.2382643755399827,
910
+ "grad_norm": 0.15528739988803864,
911
+ "learning_rate": 0.00018824092431804197,
912
+ "loss": 0.1053,
913
+ "step": 6450
914
+ },
915
+ {
916
+ "epoch": 1.2478640683498128,
917
+ "grad_norm": 0.19049355387687683,
918
+ "learning_rate": 0.0001872154235318247,
919
+ "loss": 0.1043,
920
+ "step": 6500
921
+ },
922
+ {
923
+ "epoch": 1.2574637611596429,
924
+ "grad_norm": 0.1925603300333023,
925
+ "learning_rate": 0.00018618992274560742,
926
+ "loss": 0.1043,
927
+ "step": 6550
928
+ },
929
+ {
930
+ "epoch": 1.267063453969473,
931
+ "grad_norm": 0.10907876491546631,
932
+ "learning_rate": 0.00018516442195939018,
933
+ "loss": 0.1013,
934
+ "step": 6600
935
+ },
936
+ {
937
+ "epoch": 1.276663146779303,
938
+ "grad_norm": 0.1089431643486023,
939
+ "learning_rate": 0.00018413892117317287,
940
+ "loss": 0.1028,
941
+ "step": 6650
942
+ },
943
+ {
944
+ "epoch": 1.2862628395891331,
945
+ "grad_norm": 0.13383787870407104,
946
+ "learning_rate": 0.0001831134203869556,
947
+ "loss": 0.1036,
948
+ "step": 6700
949
+ },
950
+ {
951
+ "epoch": 1.2958625323989632,
952
+ "grad_norm": 0.3068512976169586,
953
+ "learning_rate": 0.00018208791960073835,
954
+ "loss": 0.1024,
955
+ "step": 6750
956
+ },
957
+ {
958
+ "epoch": 1.3054622252087933,
959
+ "grad_norm": 0.25988587737083435,
960
+ "learning_rate": 0.00018106241881452108,
961
+ "loss": 0.1053,
962
+ "step": 6800
963
+ },
964
+ {
965
+ "epoch": 1.3150619180186234,
966
+ "grad_norm": 0.16754411160945892,
967
+ "learning_rate": 0.0001800369180283038,
968
+ "loss": 0.1069,
969
+ "step": 6850
970
+ },
971
+ {
972
+ "epoch": 1.3246616108284535,
973
+ "grad_norm": 0.11950711905956268,
974
+ "learning_rate": 0.00017901141724208653,
975
+ "loss": 0.1021,
976
+ "step": 6900
977
+ },
978
+ {
979
+ "epoch": 1.3342613036382835,
980
+ "grad_norm": 0.19078785181045532,
981
+ "learning_rate": 0.00017798591645586925,
982
+ "loss": 0.1047,
983
+ "step": 6950
984
+ },
985
+ {
986
+ "epoch": 1.3438609964481136,
987
+ "grad_norm": 0.1755492091178894,
988
+ "learning_rate": 0.000176960415669652,
989
+ "loss": 0.1059,
990
+ "step": 7000
991
+ },
992
+ {
993
+ "epoch": 1.3534606892579437,
994
+ "grad_norm": 0.15602967143058777,
995
+ "learning_rate": 0.00017593491488343473,
996
+ "loss": 0.1034,
997
+ "step": 7050
998
+ },
999
+ {
1000
+ "epoch": 1.3630603820677738,
1001
+ "grad_norm": 0.2121271789073944,
1002
+ "learning_rate": 0.00017490941409721746,
1003
+ "loss": 0.1033,
1004
+ "step": 7100
1005
+ },
1006
+ {
1007
+ "epoch": 1.3726600748776039,
1008
+ "grad_norm": 0.24380344152450562,
1009
+ "learning_rate": 0.0001738839133110002,
1010
+ "loss": 0.104,
1011
+ "step": 7150
1012
+ },
1013
+ {
1014
+ "epoch": 1.382259767687434,
1015
+ "grad_norm": 0.2051132768392563,
1016
+ "learning_rate": 0.0001728584125247829,
1017
+ "loss": 0.102,
1018
+ "step": 7200
1019
+ },
1020
+ {
1021
+ "epoch": 1.391859460497264,
1022
+ "grad_norm": 0.196175217628479,
1023
+ "learning_rate": 0.00017183291173856563,
1024
+ "loss": 0.1019,
1025
+ "step": 7250
1026
+ },
1027
+ {
1028
+ "epoch": 1.4014591533070941,
1029
+ "grad_norm": 0.11703308671712875,
1030
+ "learning_rate": 0.00017080741095234838,
1031
+ "loss": 0.1066,
1032
+ "step": 7300
1033
+ },
1034
+ {
1035
+ "epoch": 1.4110588461169242,
1036
+ "grad_norm": 0.13491560518741608,
1037
+ "learning_rate": 0.0001697819101661311,
1038
+ "loss": 0.1026,
1039
+ "step": 7350
1040
+ },
1041
+ {
1042
+ "epoch": 1.4206585389267543,
1043
+ "grad_norm": 0.11467725783586502,
1044
+ "learning_rate": 0.00016875640937991386,
1045
+ "loss": 0.1024,
1046
+ "step": 7400
1047
+ },
1048
+ {
1049
+ "epoch": 1.4302582317365844,
1050
+ "grad_norm": 0.23132862150669098,
1051
+ "learning_rate": 0.00016773090859369659,
1052
+ "loss": 0.1041,
1053
+ "step": 7450
1054
+ },
1055
+ {
1056
+ "epoch": 1.4398579245464145,
1057
+ "grad_norm": 0.09554282575845718,
1058
+ "learning_rate": 0.00016670540780747928,
1059
+ "loss": 0.1077,
1060
+ "step": 7500
1061
+ },
1062
+ {
1063
+ "epoch": 1.4494576173562446,
1064
+ "grad_norm": 0.1522522270679474,
1065
+ "learning_rate": 0.00016567990702126204,
1066
+ "loss": 0.1044,
1067
+ "step": 7550
1068
+ },
1069
+ {
1070
+ "epoch": 1.4590573101660747,
1071
+ "grad_norm": 0.13647988438606262,
1072
+ "learning_rate": 0.00016465440623504476,
1073
+ "loss": 0.1038,
1074
+ "step": 7600
1075
+ },
1076
+ {
1077
+ "epoch": 1.4686570029759047,
1078
+ "grad_norm": 0.21043138206005096,
1079
+ "learning_rate": 0.0001636289054488275,
1080
+ "loss": 0.1013,
1081
+ "step": 7650
1082
+ },
1083
+ {
1084
+ "epoch": 1.4782566957857348,
1085
+ "grad_norm": 0.15425056219100952,
1086
+ "learning_rate": 0.00016260340466261024,
1087
+ "loss": 0.1032,
1088
+ "step": 7700
1089
+ },
1090
+ {
1091
+ "epoch": 1.487856388595565,
1092
+ "grad_norm": 0.1538008749485016,
1093
+ "learning_rate": 0.00016157790387639296,
1094
+ "loss": 0.1022,
1095
+ "step": 7750
1096
+ },
1097
+ {
1098
+ "epoch": 1.497456081405395,
1099
+ "grad_norm": 0.09889411181211472,
1100
+ "learning_rate": 0.00016055240309017566,
1101
+ "loss": 0.1034,
1102
+ "step": 7800
1103
+ },
1104
+ {
1105
+ "epoch": 1.507055774215225,
1106
+ "grad_norm": 0.0999131053686142,
1107
+ "learning_rate": 0.00015952690230395841,
1108
+ "loss": 0.1042,
1109
+ "step": 7850
1110
+ },
1111
+ {
1112
+ "epoch": 1.5166554670250552,
1113
+ "grad_norm": 0.10318354517221451,
1114
+ "learning_rate": 0.00015850140151774114,
1115
+ "loss": 0.1026,
1116
+ "step": 7900
1117
+ },
1118
+ {
1119
+ "epoch": 1.5262551598348852,
1120
+ "grad_norm": 0.1949579417705536,
1121
+ "learning_rate": 0.0001574759007315239,
1122
+ "loss": 0.101,
1123
+ "step": 7950
1124
+ },
1125
+ {
1126
+ "epoch": 1.5358548526447153,
1127
+ "grad_norm": 0.17605358362197876,
1128
+ "learning_rate": 0.00015645039994530662,
1129
+ "loss": 0.1052,
1130
+ "step": 8000
1131
+ },
1132
+ {
1133
+ "epoch": 1.5454545454545454,
1134
+ "grad_norm": 0.23278145492076874,
1135
+ "learning_rate": 0.00015542489915908934,
1136
+ "loss": 0.1066,
1137
+ "step": 8050
1138
+ },
1139
+ {
1140
+ "epoch": 1.5550542382643755,
1141
+ "grad_norm": 0.24276228249073029,
1142
+ "learning_rate": 0.0001543993983728721,
1143
+ "loss": 0.1064,
1144
+ "step": 8100
1145
+ },
1146
+ {
1147
+ "epoch": 1.5646539310742056,
1148
+ "grad_norm": 0.16037604212760925,
1149
+ "learning_rate": 0.0001533738975866548,
1150
+ "loss": 0.1023,
1151
+ "step": 8150
1152
+ },
1153
+ {
1154
+ "epoch": 1.5742536238840357,
1155
+ "grad_norm": 0.2868128716945648,
1156
+ "learning_rate": 0.00015234839680043752,
1157
+ "loss": 0.1031,
1158
+ "step": 8200
1159
+ },
1160
+ {
1161
+ "epoch": 1.5838533166938658,
1162
+ "grad_norm": 0.10882575064897537,
1163
+ "learning_rate": 0.00015132289601422027,
1164
+ "loss": 0.1033,
1165
+ "step": 8250
1166
+ },
1167
+ {
1168
+ "epoch": 1.5934530095036958,
1169
+ "grad_norm": 0.45313918590545654,
1170
+ "learning_rate": 0.000150297395228003,
1171
+ "loss": 0.1042,
1172
+ "step": 8300
1173
+ },
1174
+ {
1175
+ "epoch": 1.603052702313526,
1176
+ "grad_norm": 1.096095323562622,
1177
+ "learning_rate": 0.00014927189444178572,
1178
+ "loss": 0.1123,
1179
+ "step": 8350
1180
+ },
1181
+ {
1182
+ "epoch": 1.612652395123356,
1183
+ "grad_norm": 0.15017154812812805,
1184
+ "learning_rate": 0.00014824639365556847,
1185
+ "loss": 0.1158,
1186
+ "step": 8400
1187
+ },
1188
+ {
1189
+ "epoch": 1.622252087933186,
1190
+ "grad_norm": 0.11559096723794937,
1191
+ "learning_rate": 0.00014722089286935117,
1192
+ "loss": 0.1024,
1193
+ "step": 8450
1194
+ },
1195
+ {
1196
+ "epoch": 1.6318517807430162,
1197
+ "grad_norm": 0.13219603896141052,
1198
+ "learning_rate": 0.00014619539208313392,
1199
+ "loss": 0.1017,
1200
+ "step": 8500
1201
+ },
1202
+ {
1203
+ "epoch": 1.6414514735528463,
1204
+ "grad_norm": 0.11071953922510147,
1205
+ "learning_rate": 0.00014516989129691665,
1206
+ "loss": 0.1005,
1207
+ "step": 8550
1208
+ },
1209
+ {
1210
+ "epoch": 1.6510511663626763,
1211
+ "grad_norm": 0.09291780740022659,
1212
+ "learning_rate": 0.00014414439051069937,
1213
+ "loss": 0.1002,
1214
+ "step": 8600
1215
+ },
1216
+ {
1217
+ "epoch": 1.6606508591725064,
1218
+ "grad_norm": 0.24729683995246887,
1219
+ "learning_rate": 0.0001431188897244821,
1220
+ "loss": 0.1001,
1221
+ "step": 8650
1222
+ },
1223
+ {
1224
+ "epoch": 1.6702505519823365,
1225
+ "grad_norm": 0.10102404654026031,
1226
+ "learning_rate": 0.00014209338893826485,
1227
+ "loss": 0.1017,
1228
+ "step": 8700
1229
+ },
1230
+ {
1231
+ "epoch": 1.6798502447921666,
1232
+ "grad_norm": 0.08600980043411255,
1233
+ "learning_rate": 0.00014106788815204758,
1234
+ "loss": 0.0992,
1235
+ "step": 8750
1236
+ },
1237
+ {
1238
+ "epoch": 1.6894499376019967,
1239
+ "grad_norm": 0.11300817131996155,
1240
+ "learning_rate": 0.0001400423873658303,
1241
+ "loss": 0.102,
1242
+ "step": 8800
1243
+ },
1244
+ {
1245
+ "epoch": 1.6990496304118268,
1246
+ "grad_norm": 0.07726065069437027,
1247
+ "learning_rate": 0.00013901688657961303,
1248
+ "loss": 0.101,
1249
+ "step": 8850
1250
+ },
1251
+ {
1252
+ "epoch": 1.7086493232216569,
1253
+ "grad_norm": 0.07588201761245728,
1254
+ "learning_rate": 0.00013799138579339575,
1255
+ "loss": 0.0981,
1256
+ "step": 8900
1257
+ },
1258
+ {
1259
+ "epoch": 1.718249016031487,
1260
+ "grad_norm": 0.12571977078914642,
1261
+ "learning_rate": 0.0001369658850071785,
1262
+ "loss": 0.0998,
1263
+ "step": 8950
1264
+ },
1265
+ {
1266
+ "epoch": 1.727848708841317,
1267
+ "grad_norm": 0.11262161284685135,
1268
+ "learning_rate": 0.00013594038422096123,
1269
+ "loss": 0.1018,
1270
+ "step": 9000
1271
+ },
1272
+ {
1273
+ "epoch": 1.7374484016511471,
1274
+ "grad_norm": 0.07787909358739853,
1275
+ "learning_rate": 0.00013491488343474395,
1276
+ "loss": 0.1008,
1277
+ "step": 9050
1278
+ },
1279
+ {
1280
+ "epoch": 1.7470480944609772,
1281
+ "grad_norm": 0.19743576645851135,
1282
+ "learning_rate": 0.00013388938264852668,
1283
+ "loss": 0.1018,
1284
+ "step": 9100
1285
+ },
1286
+ {
1287
+ "epoch": 1.7566477872708073,
1288
+ "grad_norm": 0.10666341334581375,
1289
+ "learning_rate": 0.00013286388186230943,
1290
+ "loss": 0.1004,
1291
+ "step": 9150
1292
+ },
1293
+ {
1294
+ "epoch": 1.7662474800806374,
1295
+ "grad_norm": 0.12897364795207977,
1296
+ "learning_rate": 0.00013183838107609213,
1297
+ "loss": 0.1005,
1298
+ "step": 9200
1299
+ },
1300
+ {
1301
+ "epoch": 1.7758471728904675,
1302
+ "grad_norm": 0.10816533863544464,
1303
+ "learning_rate": 0.00013081288028987488,
1304
+ "loss": 0.1001,
1305
+ "step": 9250
1306
+ },
1307
+ {
1308
+ "epoch": 1.7854468657002975,
1309
+ "grad_norm": 0.16208283603191376,
1310
+ "learning_rate": 0.0001297873795036576,
1311
+ "loss": 0.1002,
1312
+ "step": 9300
1313
+ },
1314
+ {
1315
+ "epoch": 1.7950465585101276,
1316
+ "grad_norm": 0.26765286922454834,
1317
+ "learning_rate": 0.00012876187871744033,
1318
+ "loss": 0.1014,
1319
+ "step": 9350
1320
+ },
1321
+ {
1322
+ "epoch": 1.8046462513199577,
1323
+ "grad_norm": 0.0876406878232956,
1324
+ "learning_rate": 0.00012773637793122306,
1325
+ "loss": 0.1006,
1326
+ "step": 9400
1327
+ },
1328
+ {
1329
+ "epoch": 1.8142459441297878,
1330
+ "grad_norm": 0.3883680999279022,
1331
+ "learning_rate": 0.0001267108771450058,
1332
+ "loss": 0.0999,
1333
+ "step": 9450
1334
+ },
1335
+ {
1336
+ "epoch": 1.8238456369396179,
1337
+ "grad_norm": 0.17567026615142822,
1338
+ "learning_rate": 0.00012568537635878854,
1339
+ "loss": 0.1001,
1340
+ "step": 9500
1341
+ },
1342
+ {
1343
+ "epoch": 1.833445329749448,
1344
+ "grad_norm": 0.11428207159042358,
1345
+ "learning_rate": 0.00012465987557257126,
1346
+ "loss": 0.1019,
1347
+ "step": 9550
1348
+ },
1349
+ {
1350
+ "epoch": 1.843045022559278,
1351
+ "grad_norm": 0.10391507297754288,
1352
+ "learning_rate": 0.00012363437478635399,
1353
+ "loss": 0.0986,
1354
+ "step": 9600
1355
+ },
1356
+ {
1357
+ "epoch": 1.8526447153691081,
1358
+ "grad_norm": 0.08824704587459564,
1359
+ "learning_rate": 0.00012260887400013674,
1360
+ "loss": 0.0994,
1361
+ "step": 9650
1362
+ },
1363
+ {
1364
+ "epoch": 1.8622444081789382,
1365
+ "grad_norm": 0.10438057780265808,
1366
+ "learning_rate": 0.00012158337321391946,
1367
+ "loss": 0.1006,
1368
+ "step": 9700
1369
+ },
1370
+ {
1371
+ "epoch": 1.8718441009887683,
1372
+ "grad_norm": 0.10627109557390213,
1373
+ "learning_rate": 0.00012055787242770217,
1374
+ "loss": 0.1003,
1375
+ "step": 9750
1376
+ },
1377
+ {
1378
+ "epoch": 1.8814437937985984,
1379
+ "grad_norm": 0.1276552826166153,
1380
+ "learning_rate": 0.00011953237164148491,
1381
+ "loss": 0.1006,
1382
+ "step": 9800
1383
+ },
1384
+ {
1385
+ "epoch": 1.8910434866084285,
1386
+ "grad_norm": 0.1135965958237648,
1387
+ "learning_rate": 0.00011850687085526765,
1388
+ "loss": 0.0994,
1389
+ "step": 9850
1390
+ },
1391
+ {
1392
+ "epoch": 1.9006431794182586,
1393
+ "grad_norm": 0.06545350700616837,
1394
+ "learning_rate": 0.00011748137006905036,
1395
+ "loss": 0.0999,
1396
+ "step": 9900
1397
+ },
1398
+ {
1399
+ "epoch": 1.9102428722280886,
1400
+ "grad_norm": 0.07519547641277313,
1401
+ "learning_rate": 0.0001164558692828331,
1402
+ "loss": 0.0994,
1403
+ "step": 9950
1404
+ },
1405
+ {
1406
+ "epoch": 1.9198425650379187,
1407
+ "grad_norm": 0.06651521474123001,
1408
+ "learning_rate": 0.00011543036849661584,
1409
+ "loss": 0.1022,
1410
+ "step": 10000
1411
+ },
1412
+ {
1413
+ "epoch": 1.9294422578477488,
1414
+ "grad_norm": 0.08063454180955887,
1415
+ "learning_rate": 0.00011440486771039857,
1416
+ "loss": 0.0995,
1417
+ "step": 10050
1418
+ },
1419
+ {
1420
+ "epoch": 1.939041950657579,
1421
+ "grad_norm": 0.09275515377521515,
1422
+ "learning_rate": 0.00011337936692418129,
1423
+ "loss": 0.1018,
1424
+ "step": 10100
1425
+ },
1426
+ {
1427
+ "epoch": 1.948641643467409,
1428
+ "grad_norm": 0.1343824565410614,
1429
+ "learning_rate": 0.00011235386613796403,
1430
+ "loss": 0.1013,
1431
+ "step": 10150
1432
+ },
1433
+ {
1434
+ "epoch": 1.958241336277239,
1435
+ "grad_norm": 0.07984330505132675,
1436
+ "learning_rate": 0.00011132836535174676,
1437
+ "loss": 0.0975,
1438
+ "step": 10200
1439
+ },
1440
+ {
1441
+ "epoch": 1.9678410290870691,
1442
+ "grad_norm": 0.17874419689178467,
1443
+ "learning_rate": 0.0001103028645655295,
1444
+ "loss": 0.0989,
1445
+ "step": 10250
1446
+ },
1447
+ {
1448
+ "epoch": 1.9774407218968992,
1449
+ "grad_norm": 0.10690494626760483,
1450
+ "learning_rate": 0.00010927736377931222,
1451
+ "loss": 0.0994,
1452
+ "step": 10300
1453
+ },
1454
+ {
1455
+ "epoch": 1.9870404147067293,
1456
+ "grad_norm": 0.08402542024850845,
1457
+ "learning_rate": 0.00010825186299309494,
1458
+ "loss": 0.098,
1459
+ "step": 10350
1460
+ },
1461
+ {
1462
+ "epoch": 1.9966401075165594,
1463
+ "grad_norm": 0.08502990752458572,
1464
+ "learning_rate": 0.00010722636220687768,
1465
+ "loss": 0.0992,
1466
+ "step": 10400
1467
+ },
1468
+ {
1469
+ "epoch": 2.0061438033982912,
1470
+ "grad_norm": 0.18513023853302002,
1471
+ "learning_rate": 0.00010620086142066042,
1472
+ "loss": 0.103,
1473
+ "step": 10450
1474
+ },
1475
+ {
1476
+ "epoch": 2.0157434962081213,
1477
+ "grad_norm": 0.10088857263326645,
1478
+ "learning_rate": 0.00010517536063444313,
1479
+ "loss": 0.099,
1480
+ "step": 10500
1481
+ },
1482
+ {
1483
+ "epoch": 2.0253431890179514,
1484
+ "grad_norm": 0.08932794630527496,
1485
+ "learning_rate": 0.00010414985984822587,
1486
+ "loss": 0.0986,
1487
+ "step": 10550
1488
+ },
1489
+ {
1490
+ "epoch": 2.0349428818277815,
1491
+ "grad_norm": 0.12779156863689423,
1492
+ "learning_rate": 0.00010312435906200861,
1493
+ "loss": 0.0987,
1494
+ "step": 10600
1495
+ },
1496
+ {
1497
+ "epoch": 2.0445425746376116,
1498
+ "grad_norm": 0.08535610139369965,
1499
+ "learning_rate": 0.00010209885827579132,
1500
+ "loss": 0.0976,
1501
+ "step": 10650
1502
+ },
1503
+ {
1504
+ "epoch": 2.0541422674474417,
1505
+ "grad_norm": 0.13694776594638824,
1506
+ "learning_rate": 0.00010107335748957406,
1507
+ "loss": 0.0992,
1508
+ "step": 10700
1509
+ },
1510
+ {
1511
+ "epoch": 2.0637419602572717,
1512
+ "grad_norm": 0.09219267219305038,
1513
+ "learning_rate": 0.0001000478567033568,
1514
+ "loss": 0.0985,
1515
+ "step": 10750
1516
+ },
1517
+ {
1518
+ "epoch": 2.073341653067102,
1519
+ "grad_norm": 0.20255500078201294,
1520
+ "learning_rate": 9.902235591713954e-05,
1521
+ "loss": 0.0992,
1522
+ "step": 10800
1523
+ },
1524
+ {
1525
+ "epoch": 2.082941345876932,
1526
+ "grad_norm": 0.09075580537319183,
1527
+ "learning_rate": 9.799685513092225e-05,
1528
+ "loss": 0.0993,
1529
+ "step": 10850
1530
+ },
1531
+ {
1532
+ "epoch": 2.092541038686762,
1533
+ "grad_norm": 0.0955098420381546,
1534
+ "learning_rate": 9.697135434470499e-05,
1535
+ "loss": 0.0977,
1536
+ "step": 10900
1537
+ },
1538
+ {
1539
+ "epoch": 2.102140731496592,
1540
+ "grad_norm": 0.14620476961135864,
1541
+ "learning_rate": 9.594585355848773e-05,
1542
+ "loss": 0.0988,
1543
+ "step": 10950
1544
+ },
1545
+ {
1546
+ "epoch": 2.111740424306422,
1547
+ "grad_norm": 0.06722934544086456,
1548
+ "learning_rate": 9.492035277227045e-05,
1549
+ "loss": 0.0994,
1550
+ "step": 11000
1551
+ },
1552
+ {
1553
+ "epoch": 2.1213401171162523,
1554
+ "grad_norm": 0.11390715092420578,
1555
+ "learning_rate": 9.389485198605318e-05,
1556
+ "loss": 0.0996,
1557
+ "step": 11050
1558
+ },
1559
+ {
1560
+ "epoch": 2.1309398099260823,
1561
+ "grad_norm": 0.08618480712175369,
1562
+ "learning_rate": 9.286935119983592e-05,
1563
+ "loss": 0.0985,
1564
+ "step": 11100
1565
+ },
1566
+ {
1567
+ "epoch": 2.1405395027359124,
1568
+ "grad_norm": 0.09730342775583267,
1569
+ "learning_rate": 9.184385041361864e-05,
1570
+ "loss": 0.0987,
1571
+ "step": 11150
1572
+ },
1573
+ {
1574
+ "epoch": 2.1501391955457425,
1575
+ "grad_norm": 0.23508231341838837,
1576
+ "learning_rate": 9.081834962740137e-05,
1577
+ "loss": 0.0965,
1578
+ "step": 11200
1579
+ },
1580
+ {
1581
+ "epoch": 2.1597388883555726,
1582
+ "grad_norm": 0.0964941680431366,
1583
+ "learning_rate": 8.97928488411841e-05,
1584
+ "loss": 0.1017,
1585
+ "step": 11250
1586
+ },
1587
+ {
1588
+ "epoch": 2.1693385811654027,
1589
+ "grad_norm": 0.06535065919160843,
1590
+ "learning_rate": 8.876734805496683e-05,
1591
+ "loss": 0.0986,
1592
+ "step": 11300
1593
+ },
1594
+ {
1595
+ "epoch": 2.1789382739752328,
1596
+ "grad_norm": 0.06535002589225769,
1597
+ "learning_rate": 8.774184726874957e-05,
1598
+ "loss": 0.0988,
1599
+ "step": 11350
1600
+ },
1601
+ {
1602
+ "epoch": 2.188537966785063,
1603
+ "grad_norm": 0.06418288499116898,
1604
+ "learning_rate": 8.67163464825323e-05,
1605
+ "loss": 0.0987,
1606
+ "step": 11400
1607
+ },
1608
+ {
1609
+ "epoch": 2.198137659594893,
1610
+ "grad_norm": 0.1190769225358963,
1611
+ "learning_rate": 8.569084569631502e-05,
1612
+ "loss": 0.0972,
1613
+ "step": 11450
1614
+ },
1615
+ {
1616
+ "epoch": 2.207737352404723,
1617
+ "grad_norm": 0.07921498268842697,
1618
+ "learning_rate": 8.466534491009776e-05,
1619
+ "loss": 0.0975,
1620
+ "step": 11500
1621
+ },
1622
+ {
1623
+ "epoch": 2.217337045214553,
1624
+ "grad_norm": 0.08557629585266113,
1625
+ "learning_rate": 8.36398441238805e-05,
1626
+ "loss": 0.0964,
1627
+ "step": 11550
1628
+ },
1629
+ {
1630
+ "epoch": 2.226936738024383,
1631
+ "grad_norm": 0.09319253265857697,
1632
+ "learning_rate": 8.261434333766321e-05,
1633
+ "loss": 0.0976,
1634
+ "step": 11600
1635
+ },
1636
+ {
1637
+ "epoch": 2.2365364308342133,
1638
+ "grad_norm": 0.08666611462831497,
1639
+ "learning_rate": 8.158884255144595e-05,
1640
+ "loss": 0.0987,
1641
+ "step": 11650
1642
+ },
1643
+ {
1644
+ "epoch": 2.2461361236440434,
1645
+ "grad_norm": 0.05421123653650284,
1646
+ "learning_rate": 8.056334176522869e-05,
1647
+ "loss": 0.0961,
1648
+ "step": 11700
1649
+ },
1650
+ {
1651
+ "epoch": 2.2557358164538734,
1652
+ "grad_norm": 0.1075129583477974,
1653
+ "learning_rate": 7.95378409790114e-05,
1654
+ "loss": 0.0989,
1655
+ "step": 11750
1656
+ },
1657
+ {
1658
+ "epoch": 2.2653355092637035,
1659
+ "grad_norm": 0.10412245243787766,
1660
+ "learning_rate": 7.851234019279414e-05,
1661
+ "loss": 0.0988,
1662
+ "step": 11800
1663
+ },
1664
+ {
1665
+ "epoch": 2.2749352020735336,
1666
+ "grad_norm": 0.10957926511764526,
1667
+ "learning_rate": 7.748683940657688e-05,
1668
+ "loss": 0.0972,
1669
+ "step": 11850
1670
+ },
1671
+ {
1672
+ "epoch": 2.2845348948833637,
1673
+ "grad_norm": 0.06667158007621765,
1674
+ "learning_rate": 7.64613386203596e-05,
1675
+ "loss": 0.0959,
1676
+ "step": 11900
1677
+ },
1678
+ {
1679
+ "epoch": 2.294134587693194,
1680
+ "grad_norm": 0.07828564196825027,
1681
+ "learning_rate": 7.543583783414233e-05,
1682
+ "loss": 0.0968,
1683
+ "step": 11950
1684
+ },
1685
+ {
1686
+ "epoch": 2.303734280503024,
1687
+ "grad_norm": 0.07054735720157623,
1688
+ "learning_rate": 7.441033704792507e-05,
1689
+ "loss": 0.0975,
1690
+ "step": 12000
1691
+ },
1692
+ {
1693
+ "epoch": 2.313333973312854,
1694
+ "grad_norm": 0.06447632610797882,
1695
+ "learning_rate": 7.338483626170779e-05,
1696
+ "loss": 0.0962,
1697
+ "step": 12050
1698
+ },
1699
+ {
1700
+ "epoch": 2.322933666122684,
1701
+ "grad_norm": 0.07028626650571823,
1702
+ "learning_rate": 7.235933547549053e-05,
1703
+ "loss": 0.0961,
1704
+ "step": 12100
1705
+ },
1706
+ {
1707
+ "epoch": 2.332533358932514,
1708
+ "grad_norm": 0.06260377168655396,
1709
+ "learning_rate": 7.133383468927325e-05,
1710
+ "loss": 0.0964,
1711
+ "step": 12150
1712
+ },
1713
+ {
1714
+ "epoch": 2.342133051742344,
1715
+ "grad_norm": 0.07249750196933746,
1716
+ "learning_rate": 7.030833390305598e-05,
1717
+ "loss": 0.0976,
1718
+ "step": 12200
1719
+ },
1720
+ {
1721
+ "epoch": 2.3517327445521743,
1722
+ "grad_norm": 0.08855904638767242,
1723
+ "learning_rate": 6.928283311683872e-05,
1724
+ "loss": 0.0967,
1725
+ "step": 12250
1726
+ },
1727
+ {
1728
+ "epoch": 2.3613324373620044,
1729
+ "grad_norm": 0.051147934049367905,
1730
+ "learning_rate": 6.825733233062144e-05,
1731
+ "loss": 0.0981,
1732
+ "step": 12300
1733
+ },
1734
+ {
1735
+ "epoch": 2.3709321301718345,
1736
+ "grad_norm": 0.08245785534381866,
1737
+ "learning_rate": 6.723183154440417e-05,
1738
+ "loss": 0.0965,
1739
+ "step": 12350
1740
+ },
1741
+ {
1742
+ "epoch": 2.3805318229816645,
1743
+ "grad_norm": 0.08681339770555496,
1744
+ "learning_rate": 6.620633075818691e-05,
1745
+ "loss": 0.0977,
1746
+ "step": 12400
1747
+ },
1748
+ {
1749
+ "epoch": 2.3901315157914946,
1750
+ "grad_norm": 0.05284340679645538,
1751
+ "learning_rate": 6.518082997196963e-05,
1752
+ "loss": 0.0945,
1753
+ "step": 12450
1754
+ },
1755
+ {
1756
+ "epoch": 2.3997312086013247,
1757
+ "grad_norm": 0.05780034512281418,
1758
+ "learning_rate": 6.415532918575237e-05,
1759
+ "loss": 0.0988,
1760
+ "step": 12500
1761
+ },
1762
+ {
1763
+ "epoch": 2.409330901411155,
1764
+ "grad_norm": 0.05405628681182861,
1765
+ "learning_rate": 6.31298283995351e-05,
1766
+ "loss": 0.0991,
1767
+ "step": 12550
1768
+ },
1769
+ {
1770
+ "epoch": 2.418930594220985,
1771
+ "grad_norm": 0.06091487780213356,
1772
+ "learning_rate": 6.210432761331784e-05,
1773
+ "loss": 0.0973,
1774
+ "step": 12600
1775
+ },
1776
+ {
1777
+ "epoch": 2.428530287030815,
1778
+ "grad_norm": 0.09573518484830856,
1779
+ "learning_rate": 6.107882682710056e-05,
1780
+ "loss": 0.0968,
1781
+ "step": 12650
1782
+ },
1783
+ {
1784
+ "epoch": 2.438129979840645,
1785
+ "grad_norm": 0.06391169875860214,
1786
+ "learning_rate": 6.005332604088329e-05,
1787
+ "loss": 0.097,
1788
+ "step": 12700
1789
+ },
1790
+ {
1791
+ "epoch": 2.447729672650475,
1792
+ "grad_norm": 0.06078439578413963,
1793
+ "learning_rate": 5.9027825254666025e-05,
1794
+ "loss": 0.0962,
1795
+ "step": 12750
1796
+ },
1797
+ {
1798
+ "epoch": 2.4573293654603052,
1799
+ "grad_norm": 0.05879442021250725,
1800
+ "learning_rate": 5.800232446844876e-05,
1801
+ "loss": 0.0983,
1802
+ "step": 12800
1803
+ },
1804
+ {
1805
+ "epoch": 2.4669290582701353,
1806
+ "grad_norm": 0.07806123793125153,
1807
+ "learning_rate": 5.697682368223148e-05,
1808
+ "loss": 0.0975,
1809
+ "step": 12850
1810
+ },
1811
+ {
1812
+ "epoch": 2.4765287510799654,
1813
+ "grad_norm": 0.0653390884399414,
1814
+ "learning_rate": 5.5951322896014214e-05,
1815
+ "loss": 0.0966,
1816
+ "step": 12900
1817
+ },
1818
+ {
1819
+ "epoch": 2.4861284438897955,
1820
+ "grad_norm": 0.07627742737531662,
1821
+ "learning_rate": 5.4925822109796946e-05,
1822
+ "loss": 0.0975,
1823
+ "step": 12950
1824
+ },
1825
+ {
1826
+ "epoch": 2.4957281366996256,
1827
+ "grad_norm": 0.0665074959397316,
1828
+ "learning_rate": 5.390032132357967e-05,
1829
+ "loss": 0.0953,
1830
+ "step": 13000
1831
+ },
1832
+ {
1833
+ "epoch": 2.5053278295094557,
1834
+ "grad_norm": 0.07060660421848297,
1835
+ "learning_rate": 5.287482053736241e-05,
1836
+ "loss": 0.0971,
1837
+ "step": 13050
1838
+ },
1839
+ {
1840
+ "epoch": 2.5149275223192857,
1841
+ "grad_norm": 0.05964155122637749,
1842
+ "learning_rate": 5.1849319751145135e-05,
1843
+ "loss": 0.096,
1844
+ "step": 13100
1845
+ },
1846
+ {
1847
+ "epoch": 2.524527215129116,
1848
+ "grad_norm": 0.08040162920951843,
1849
+ "learning_rate": 5.082381896492787e-05,
1850
+ "loss": 0.0977,
1851
+ "step": 13150
1852
+ },
1853
+ {
1854
+ "epoch": 2.534126907938946,
1855
+ "grad_norm": 0.06038783863186836,
1856
+ "learning_rate": 4.97983181787106e-05,
1857
+ "loss": 0.0962,
1858
+ "step": 13200
1859
+ },
1860
+ {
1861
+ "epoch": 2.543726600748776,
1862
+ "grad_norm": 0.05058710277080536,
1863
+ "learning_rate": 4.877281739249333e-05,
1864
+ "loss": 0.097,
1865
+ "step": 13250
1866
+ },
1867
+ {
1868
+ "epoch": 2.553326293558606,
1869
+ "grad_norm": 0.06195589527487755,
1870
+ "learning_rate": 4.7747316606276056e-05,
1871
+ "loss": 0.0983,
1872
+ "step": 13300
1873
+ },
1874
+ {
1875
+ "epoch": 2.562925986368436,
1876
+ "grad_norm": 0.06552109122276306,
1877
+ "learning_rate": 4.6721815820058795e-05,
1878
+ "loss": 0.0983,
1879
+ "step": 13350
1880
+ },
1881
+ {
1882
+ "epoch": 2.5725256791782662,
1883
+ "grad_norm": 0.07137109339237213,
1884
+ "learning_rate": 4.569631503384152e-05,
1885
+ "loss": 0.0971,
1886
+ "step": 13400
1887
+ },
1888
+ {
1889
+ "epoch": 2.5821253719880963,
1890
+ "grad_norm": 0.07768446952104568,
1891
+ "learning_rate": 4.467081424762426e-05,
1892
+ "loss": 0.0977,
1893
+ "step": 13450
1894
+ },
1895
+ {
1896
+ "epoch": 2.5917250647979264,
1897
+ "grad_norm": 0.06904645264148712,
1898
+ "learning_rate": 4.3645313461406984e-05,
1899
+ "loss": 0.097,
1900
+ "step": 13500
1901
+ },
1902
+ {
1903
+ "epoch": 2.6013247576077565,
1904
+ "grad_norm": 0.06683080643415451,
1905
+ "learning_rate": 4.261981267518971e-05,
1906
+ "loss": 0.0968,
1907
+ "step": 13550
1908
+ },
1909
+ {
1910
+ "epoch": 2.6109244504175866,
1911
+ "grad_norm": 0.06532671302556992,
1912
+ "learning_rate": 4.159431188897245e-05,
1913
+ "loss": 0.0961,
1914
+ "step": 13600
1915
+ },
1916
+ {
1917
+ "epoch": 2.6205241432274167,
1918
+ "grad_norm": 0.0853889137506485,
1919
+ "learning_rate": 4.056881110275517e-05,
1920
+ "loss": 0.0974,
1921
+ "step": 13650
1922
+ },
1923
+ {
1924
+ "epoch": 2.6301238360372468,
1925
+ "grad_norm": 0.05568193271756172,
1926
+ "learning_rate": 3.9543310316537905e-05,
1927
+ "loss": 0.0974,
1928
+ "step": 13700
1929
+ },
1930
+ {
1931
+ "epoch": 2.639723528847077,
1932
+ "grad_norm": 0.061790887266397476,
1933
+ "learning_rate": 3.851780953032064e-05,
1934
+ "loss": 0.0941,
1935
+ "step": 13750
1936
+ },
1937
+ {
1938
+ "epoch": 2.649323221656907,
1939
+ "grad_norm": 0.08820798248052597,
1940
+ "learning_rate": 3.749230874410337e-05,
1941
+ "loss": 0.0963,
1942
+ "step": 13800
1943
+ },
1944
+ {
1945
+ "epoch": 2.658922914466737,
1946
+ "grad_norm": 0.1133754551410675,
1947
+ "learning_rate": 3.6466807957886094e-05,
1948
+ "loss": 0.0971,
1949
+ "step": 13850
1950
+ },
1951
+ {
1952
+ "epoch": 2.668522607276567,
1953
+ "grad_norm": 0.06903323531150818,
1954
+ "learning_rate": 3.5441307171668826e-05,
1955
+ "loss": 0.098,
1956
+ "step": 13900
1957
+ },
1958
+ {
1959
+ "epoch": 2.678122300086397,
1960
+ "grad_norm": 0.08473759889602661,
1961
+ "learning_rate": 3.441580638545156e-05,
1962
+ "loss": 0.0978,
1963
+ "step": 13950
1964
+ },
1965
+ {
1966
+ "epoch": 2.6877219928962273,
1967
+ "grad_norm": 0.054722901433706284,
1968
+ "learning_rate": 3.339030559923429e-05,
1969
+ "loss": 0.0968,
1970
+ "step": 14000
1971
+ },
1972
+ {
1973
+ "epoch": 2.6973216857060573,
1974
+ "grad_norm": 0.061523571610450745,
1975
+ "learning_rate": 3.236480481301702e-05,
1976
+ "loss": 0.0961,
1977
+ "step": 14050
1978
+ },
1979
+ {
1980
+ "epoch": 2.7069213785158874,
1981
+ "grad_norm": 0.0766662210226059,
1982
+ "learning_rate": 3.1339304026799754e-05,
1983
+ "loss": 0.0962,
1984
+ "step": 14100
1985
+ },
1986
+ {
1987
+ "epoch": 2.7165210713257175,
1988
+ "grad_norm": 0.061326321214437485,
1989
+ "learning_rate": 3.0313803240582483e-05,
1990
+ "loss": 0.0963,
1991
+ "step": 14150
1992
+ },
1993
+ {
1994
+ "epoch": 2.7261207641355476,
1995
+ "grad_norm": 0.06567474454641342,
1996
+ "learning_rate": 2.9288302454365215e-05,
1997
+ "loss": 0.0968,
1998
+ "step": 14200
1999
+ },
2000
+ {
2001
+ "epoch": 2.7357204569453777,
2002
+ "grad_norm": 0.043411437422037125,
2003
+ "learning_rate": 2.8262801668147943e-05,
2004
+ "loss": 0.0969,
2005
+ "step": 14250
2006
+ },
2007
+ {
2008
+ "epoch": 2.7453201497552078,
2009
+ "grad_norm": 0.04852946475148201,
2010
+ "learning_rate": 2.7237300881930675e-05,
2011
+ "loss": 0.0969,
2012
+ "step": 14300
2013
+ },
2014
+ {
2015
+ "epoch": 2.754919842565038,
2016
+ "grad_norm": 0.057755179703235626,
2017
+ "learning_rate": 2.6211800095713407e-05,
2018
+ "loss": 0.0956,
2019
+ "step": 14350
2020
+ },
2021
+ {
2022
+ "epoch": 2.764519535374868,
2023
+ "grad_norm": 0.07642246782779694,
2024
+ "learning_rate": 2.5186299309496136e-05,
2025
+ "loss": 0.0956,
2026
+ "step": 14400
2027
+ },
2028
+ {
2029
+ "epoch": 2.774119228184698,
2030
+ "grad_norm": 0.05348903685808182,
2031
+ "learning_rate": 2.4160798523278864e-05,
2032
+ "loss": 0.0965,
2033
+ "step": 14450
2034
+ },
2035
+ {
2036
+ "epoch": 2.783718920994528,
2037
+ "grad_norm": 0.05108194053173065,
2038
+ "learning_rate": 2.3135297737061596e-05,
2039
+ "loss": 0.0968,
2040
+ "step": 14500
2041
+ },
2042
+ {
2043
+ "epoch": 2.793318613804358,
2044
+ "grad_norm": 0.054020293056964874,
2045
+ "learning_rate": 2.2109796950844325e-05,
2046
+ "loss": 0.0943,
2047
+ "step": 14550
2048
+ },
2049
+ {
2050
+ "epoch": 2.8029183066141883,
2051
+ "grad_norm": 0.05261579900979996,
2052
+ "learning_rate": 2.1084296164627057e-05,
2053
+ "loss": 0.0962,
2054
+ "step": 14600
2055
+ },
2056
+ {
2057
+ "epoch": 2.8125179994240184,
2058
+ "grad_norm": 0.05358601361513138,
2059
+ "learning_rate": 2.005879537840979e-05,
2060
+ "loss": 0.0951,
2061
+ "step": 14650
2062
+ },
2063
+ {
2064
+ "epoch": 2.8221176922338485,
2065
+ "grad_norm": 0.053367454558610916,
2066
+ "learning_rate": 1.9033294592192517e-05,
2067
+ "loss": 0.0953,
2068
+ "step": 14700
2069
+ },
2070
+ {
2071
+ "epoch": 2.8317173850436785,
2072
+ "grad_norm": 0.05833417549729347,
2073
+ "learning_rate": 1.800779380597525e-05,
2074
+ "loss": 0.0967,
2075
+ "step": 14750
2076
+ },
2077
+ {
2078
+ "epoch": 2.8413170778535086,
2079
+ "grad_norm": 0.06067025661468506,
2080
+ "learning_rate": 1.698229301975798e-05,
2081
+ "loss": 0.0949,
2082
+ "step": 14800
2083
+ },
2084
+ {
2085
+ "epoch": 2.8509167706633387,
2086
+ "grad_norm": 0.048758141696453094,
2087
+ "learning_rate": 1.5956792233540713e-05,
2088
+ "loss": 0.0975,
2089
+ "step": 14850
2090
+ },
2091
+ {
2092
+ "epoch": 2.860516463473169,
2093
+ "grad_norm": 0.053699832409620285,
2094
+ "learning_rate": 1.493129144732344e-05,
2095
+ "loss": 0.0959,
2096
+ "step": 14900
2097
+ },
2098
+ {
2099
+ "epoch": 2.870116156282999,
2100
+ "grad_norm": 0.05183803662657738,
2101
+ "learning_rate": 1.3905790661106172e-05,
2102
+ "loss": 0.0967,
2103
+ "step": 14950
2104
+ },
2105
+ {
2106
+ "epoch": 2.879715849092829,
2107
+ "grad_norm": 0.06660551577806473,
2108
+ "learning_rate": 1.2880289874888902e-05,
2109
+ "loss": 0.0965,
2110
+ "step": 15000
2111
+ },
2112
+ {
2113
+ "epoch": 2.889315541902659,
2114
+ "grad_norm": 0.05176248028874397,
2115
+ "learning_rate": 1.1854789088671634e-05,
2116
+ "loss": 0.0961,
2117
+ "step": 15050
2118
+ },
2119
+ {
2120
+ "epoch": 2.898915234712489,
2121
+ "grad_norm": 0.061364296823740005,
2122
+ "learning_rate": 1.0829288302454365e-05,
2123
+ "loss": 0.0968,
2124
+ "step": 15100
2125
+ },
2126
+ {
2127
+ "epoch": 2.908514927522319,
2128
+ "grad_norm": 0.04964745044708252,
2129
+ "learning_rate": 9.803787516237096e-06,
2130
+ "loss": 0.0953,
2131
+ "step": 15150
2132
+ },
2133
+ {
2134
+ "epoch": 2.9181146203321493,
2135
+ "grad_norm": 0.04729118198156357,
2136
+ "learning_rate": 8.778286730019825e-06,
2137
+ "loss": 0.0962,
2138
+ "step": 15200
2139
+ },
2140
+ {
2141
+ "epoch": 2.9277143131419794,
2142
+ "grad_norm": 0.05218329653143883,
2143
+ "learning_rate": 7.752785943802555e-06,
2144
+ "loss": 0.0952,
2145
+ "step": 15250
2146
+ },
2147
+ {
2148
+ "epoch": 2.9373140059518095,
2149
+ "grad_norm": 0.06349076330661774,
2150
+ "learning_rate": 6.727285157585286e-06,
2151
+ "loss": 0.0949,
2152
+ "step": 15300
2153
+ },
2154
+ {
2155
+ "epoch": 2.9469136987616396,
2156
+ "grad_norm": 0.05340540409088135,
2157
+ "learning_rate": 5.7017843713680175e-06,
2158
+ "loss": 0.098,
2159
+ "step": 15350
2160
+ },
2161
+ {
2162
+ "epoch": 2.9565133915714696,
2163
+ "grad_norm": 0.05193692818284035,
2164
+ "learning_rate": 4.676283585150749e-06,
2165
+ "loss": 0.0962,
2166
+ "step": 15400
2167
+ },
2168
+ {
2169
+ "epoch": 2.9661130843812997,
2170
+ "grad_norm": 0.05778556317090988,
2171
+ "learning_rate": 3.650782798933479e-06,
2172
+ "loss": 0.096,
2173
+ "step": 15450
2174
+ },
2175
+ {
2176
+ "epoch": 2.97571277719113,
2177
+ "grad_norm": 0.0513499453663826,
2178
+ "learning_rate": 2.6252820127162096e-06,
2179
+ "loss": 0.0957,
2180
+ "step": 15500
2181
+ },
2182
+ {
2183
+ "epoch": 2.98531247000096,
2184
+ "grad_norm": 0.053084537386894226,
2185
+ "learning_rate": 1.5997812264989403e-06,
2186
+ "loss": 0.0949,
2187
+ "step": 15550
2188
+ },
2189
+ {
2190
+ "epoch": 2.99491216281079,
2191
+ "grad_norm": 0.06831531971693039,
2192
+ "learning_rate": 5.742804402816709e-07,
2193
+ "loss": 0.0958,
2194
+ "step": 15600
2195
+ }
2196
+ ],
2197
+ "logging_steps": 50,
2198
+ "max_steps": 15627,
2199
+ "num_input_tokens_seen": 0,
2200
+ "num_train_epochs": 3,
2201
+ "save_steps": 2000,
2202
+ "stateful_callbacks": {
2203
+ "TrainerControl": {
2204
+ "args": {
2205
+ "should_epoch_stop": false,
2206
+ "should_evaluate": false,
2207
+ "should_log": false,
2208
+ "should_save": true,
2209
+ "should_training_stop": true
2210
+ },
2211
+ "attributes": {}
2212
+ }
2213
+ },
2214
+ "total_flos": 3.4276447157118566e+17,
2215
+ "train_batch_size": 16,
2216
+ "trial_name": null,
2217
+ "trial_params": null
2218
+ }
checkpoint-15627/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:478c5792cae56598ec41be11b09c6fe07ef08f3d28ac262adfe2b7d757dfe6fa
3
+ size 5841
config.json ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "PaTHAttentionForCausalLM"
4
+ ],
5
+ "bos_token_id": 0,
6
+ "dtype": "float32",
7
+ "elementwise_affine": true,
8
+ "eos_token_id": 1,
9
+ "fuse_cross_entropy": true,
10
+ "fuse_linear_cross_entropy": false,
11
+ "fuse_norm": true,
12
+ "fuse_swiglu": true,
13
+ "head_dim": 64,
14
+ "hidden_act": "swish",
15
+ "hidden_ratio": 4,
16
+ "hidden_size": 256,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 768,
19
+ "max_position_embeddings": 2048,
20
+ "model_type": "path_attn",
21
+ "norm_eps": 1e-05,
22
+ "num_heads": 4,
23
+ "num_hidden_layers": 32,
24
+ "num_kv_heads": 4,
25
+ "rope_theta": 10000.0,
26
+ "tie_word_embeddings": true,
27
+ "transformers_version": "4.57.1",
28
+ "use_cache": true,
29
+ "use_forget_gate": true,
30
+ "use_l2warp": false,
31
+ "use_low_rank_w": true,
32
+ "use_w_shortconv": true,
33
+ "vocab_size": 256
34
+ }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 0,
4
+ "eos_token_id": 1,
5
+ "transformers_version": "4.57.1"
6
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d5397c30c8ab71c5b252b1a059a4b4abe1ca51afd6936ac40897e655f5201ed
3
+ size 111892816
special_tokens_map.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<think>",
4
+ "</think>"
5
+ ],
6
+ "bos_token": {
7
+ "content": "<|startoftext|>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false
12
+ },
13
+ "eos_token": {
14
+ "content": "<|endoftext|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false
19
+ },
20
+ "pad_token": "<|pad|>",
21
+ "unk_token": {
22
+ "content": "<|unknown|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false
27
+ }
28
+ }
tokenizer.json ADDED
@@ -0,0 +1,1085 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "1.0",
3
+ "truncation": null,
4
+ "padding": null,
5
+ "added_tokens": [
6
+ {
7
+ "id": 0,
8
+ "content": "<|startoftext|>",
9
+ "single_word": false,
10
+ "lstrip": false,
11
+ "rstrip": false,
12
+ "normalized": false,
13
+ "special": true
14
+ },
15
+ {
16
+ "id": 1,
17
+ "content": "<|endoftext|>",
18
+ "single_word": false,
19
+ "lstrip": false,
20
+ "rstrip": false,
21
+ "normalized": false,
22
+ "special": true
23
+ },
24
+ {
25
+ "id": 2,
26
+ "content": "<|unknown|>",
27
+ "single_word": false,
28
+ "lstrip": false,
29
+ "rstrip": false,
30
+ "normalized": false,
31
+ "special": true
32
+ },
33
+ {
34
+ "id": 3,
35
+ "content": "<|pad|>",
36
+ "single_word": false,
37
+ "lstrip": false,
38
+ "rstrip": false,
39
+ "normalized": false,
40
+ "special": true
41
+ },
42
+ {
43
+ "id": 4,
44
+ "content": "<think>",
45
+ "single_word": false,
46
+ "lstrip": false,
47
+ "rstrip": false,
48
+ "normalized": false,
49
+ "special": true
50
+ },
51
+ {
52
+ "id": 5,
53
+ "content": "</think>",
54
+ "single_word": false,
55
+ "lstrip": false,
56
+ "rstrip": false,
57
+ "normalized": false,
58
+ "special": true
59
+ }
60
+ ],
61
+ "normalizer": null,
62
+ "pre_tokenizer": {
63
+ "type": "ByteLevel",
64
+ "add_prefix_space": false,
65
+ "trim_offsets": true,
66
+ "use_regex": true
67
+ },
68
+ "post_processor": null,
69
+ "decoder": {
70
+ "type": "ByteLevel",
71
+ "add_prefix_space": true,
72
+ "trim_offsets": true,
73
+ "use_regex": true
74
+ },
75
+ "model": {
76
+ "type": "BPE",
77
+ "dropout": null,
78
+ "unk_token": null,
79
+ "continuing_subword_prefix": null,
80
+ "end_of_word_suffix": null,
81
+ "fuse_unk": false,
82
+ "byte_fallback": false,
83
+ "ignore_merges": false,
84
+ "vocab": {
85
+ "<|startoftext|>": 0,
86
+ "<|endoftext|>": 1,
87
+ "<|unknown|>": 2,
88
+ "<|pad|>": 3,
89
+ "<think>": 4,
90
+ "</think>": 5,
91
+ "%": 6,
92
+ "&": 7,
93
+ "'": 8,
94
+ "(": 9,
95
+ ")": 10,
96
+ ",": 11,
97
+ "-": 12,
98
+ ".": 13,
99
+ "0": 14,
100
+ "1": 15,
101
+ "2": 16,
102
+ "3": 17,
103
+ "4": 18,
104
+ "5": 19,
105
+ "6": 20,
106
+ "7": 21,
107
+ "8": 22,
108
+ "9": 23,
109
+ ":": 24,
110
+ ">": 25,
111
+ "A": 26,
112
+ "C": 27,
113
+ "D": 28,
114
+ "E": 29,
115
+ "F": 30,
116
+ "G": 31,
117
+ "H": 32,
118
+ "I": 33,
119
+ "L": 34,
120
+ "M": 35,
121
+ "N": 36,
122
+ "O": 37,
123
+ "P": 38,
124
+ "R": 39,
125
+ "S": 40,
126
+ "T": 41,
127
+ "V": 42,
128
+ "[": 43,
129
+ "]": 44,
130
+ "a": 45,
131
+ "b": 46,
132
+ "c": 47,
133
+ "d": 48,
134
+ "e": 49,
135
+ "f": 50,
136
+ "g": 51,
137
+ "h": 52,
138
+ "i": 53,
139
+ "k": 54,
140
+ "l": 55,
141
+ "m": 56,
142
+ "n": 57,
143
+ "o": 58,
144
+ "p": 59,
145
+ "q": 60,
146
+ "r": 61,
147
+ "s": 62,
148
+ "t": 63,
149
+ "u": 64,
150
+ "v": 65,
151
+ "x": 66,
152
+ "y": 67,
153
+ "|": 68,
154
+ "Ċ": 69,
155
+ "Ġ": 70,
156
+ "00": 71,
157
+ "11": 72,
158
+ "->": 73,
159
+ "Ġ|": 74,
160
+ ":[": 75,
161
+ "]->": 76,
162
+ "10": 77,
163
+ "Ġ1": 78,
164
+ "000": 79,
165
+ "ĠS": 80,
166
+ "',": 81,
167
+ "Ġ'": 82,
168
+ "te": 83,
169
+ "100": 84,
170
+ "ĊĠ": 85,
171
+ "110": 86,
172
+ "tep": 87,
173
+ "re": 88,
174
+ "111": 89,
175
+ "0000": 90,
176
+ "011": 91,
177
+ "101": 92,
178
+ "001": 93,
179
+ "ĠStep": 94,
180
+ "010": 95,
181
+ "in": 96,
182
+ "im": 97,
183
+ "ĠSim": 98,
184
+ "1111": 99,
185
+ "Ġ[": 100,
186
+ "or": 101,
187
+ "ĠI": 102,
188
+ "ho": 103,
189
+ "od": 104,
190
+ "hood": 105,
191
+ "Ġin": 106,
192
+ "Ġ11": 107,
193
+ "Ġ10": 108,
194
+ "ne": 109,
195
+ "Ġne": 110,
196
+ "']": 111,
197
+ "bor": 112,
198
+ "dx": 113,
199
+ "gh": 114,
200
+ "ire": 115,
201
+ "igh": 116,
202
+ "qu": 117,
203
+ "Ġre": 118,
204
+ "Ġ['": 119,
205
+ "ĠIdx": 120,
206
+ "Ġneigh": 121,
207
+ "borhood": 122,
208
+ "ires": 123,
209
+ "quires": 124,
210
+ "Ġrequires": 125,
211
+ "Ġneighborhood": 126,
212
+ "Step": 127,
213
+ "ut": 128,
214
+ "Ġ3": 129,
215
+ "Ġ2": 130,
216
+ "Ġ4": 131,
217
+ "Ġ5": 132,
218
+ "Ġ6": 133,
219
+ "Ġ9": 134,
220
+ "Ġ8": 135,
221
+ "Ġ7": 136,
222
+ "Ġ0": 137,
223
+ "ca": 138,
224
+ "Ġ12": 139,
225
+ "Ġ13": 140,
226
+ "1010": 141,
227
+ "Ġ14": 142,
228
+ "Out": 143,
229
+ "put": 144,
230
+ "ĠOut": 145,
231
+ "ĠSca": 146,
232
+ "ĠOutput": 147,
233
+ "ĠScan": 148,
234
+ "Ġ011": 149,
235
+ "ing": 150,
236
+ "ti": 151,
237
+ "1100": 152,
238
+ "00000000": 153,
239
+ "Ġ15": 154,
240
+ "Ġ000": 155,
241
+ "on": 156,
242
+ "Ġ110": 157,
243
+ "ul": 158,
244
+ "Ġ100": 159,
245
+ "at": 160,
246
+ "tion": 161,
247
+ "Ġ010": 162,
248
+ "1011": 163,
249
+ "ĠC": 164,
250
+ "Ġ00": 165,
251
+ "00011": 166,
252
+ "50": 167,
253
+ "er": 168,
254
+ "10000": 169,
255
+ "nd": 170,
256
+ "Ġ16": 171,
257
+ "ule": 172,
258
+ "ĠR": 173,
259
+ "ed": 174,
260
+ "li": 175,
261
+ "Ġ001": 176,
262
+ "Ġ101": 177,
263
+ "ate": 178,
264
+ "ĠRule": 179,
265
+ "1000": 180,
266
+ "fi": 181,
267
+ "10100": 182,
268
+ "as": 183,
269
+ "Ġ17": 184,
270
+ "ĠM": 185,
271
+ "Ġ1111": 186,
272
+ "Ver": 187,
273
+ "ch": 188,
274
+ "es": 189,
275
+ "ifi": 190,
276
+ "mp": 191,
277
+ "ting": 192,
278
+ "cation": 193,
279
+ "atch": 194,
280
+ "Verifi": 195,
281
+ "Verification": 196,
282
+ "..": 197,
283
+ "11011": 198,
284
+ "41": 199,
285
+ "and": 200,
286
+ "date": 201,
287
+ "idate": 202,
288
+ "ĠCand": 203,
289
+ "ĠCandidate": 204,
290
+ "58": 205,
291
+ "ating": 206,
292
+ "10011": 207,
293
+ "ction": 208,
294
+ "111000": 209,
295
+ "000000": 210,
296
+ "10110": 211,
297
+ "Ġ00000000": 212,
298
+ "is": 213,
299
+ "match": 214,
300
+ "ĠMis": 215,
301
+ "ĠMismatch": 216,
302
+ "111111": 217,
303
+ "Tas": 218,
304
+ "Task": 219,
305
+ "Ġ18": 220,
306
+ "Gre": 221,
307
+ "Pr": 222,
308
+ "ĠL": 223,
309
+ "ĠGre": 224,
310
+ "ĠPr": 225,
311
+ "edy": 226,
312
+ "ĠGreedy": 227,
313
+ "At": 228,
314
+ "Ex": 229,
315
+ "Tes": 230,
316
+ "ĠTes": 231,
317
+ "temp": 232,
318
+ "Attemp": 233,
319
+ "ĠTesting": 234,
320
+ "Attempt": 235,
321
+ "100000": 236,
322
+ "Ġs": 237,
323
+ "Ġ0000": 238,
324
+ "uction": 239,
325
+ "ĠSteps": 240,
326
+ "xt": 241,
327
+ "101111": 242,
328
+ "al": 243,
329
+ "cul": 244,
330
+ "eli": 245,
331
+ "ior": 246,
332
+ "keli": 247,
333
+ "ver": 248,
334
+ "cal": 249,
335
+ "culating": 250,
336
+ "iors": 251,
337
+ "kelihood": 252,
338
+ "Ġ19": 253,
339
+ "11111111": 254,
340
+ ").": 255
341
+ },
342
+ "merges": [
343
+ [
344
+ "0",
345
+ "0"
346
+ ],
347
+ [
348
+ "1",
349
+ "1"
350
+ ],
351
+ [
352
+ "-",
353
+ ">"
354
+ ],
355
+ [
356
+ "Ġ",
357
+ "|"
358
+ ],
359
+ [
360
+ ":",
361
+ "["
362
+ ],
363
+ [
364
+ "]",
365
+ "->"
366
+ ],
367
+ [
368
+ "1",
369
+ "0"
370
+ ],
371
+ [
372
+ "Ġ",
373
+ "1"
374
+ ],
375
+ [
376
+ "00",
377
+ "0"
378
+ ],
379
+ [
380
+ "Ġ",
381
+ "S"
382
+ ],
383
+ [
384
+ "'",
385
+ ","
386
+ ],
387
+ [
388
+ "Ġ",
389
+ "'"
390
+ ],
391
+ [
392
+ "t",
393
+ "e"
394
+ ],
395
+ [
396
+ "1",
397
+ "00"
398
+ ],
399
+ [
400
+ "Ċ",
401
+ "Ġ"
402
+ ],
403
+ [
404
+ "11",
405
+ "0"
406
+ ],
407
+ [
408
+ "te",
409
+ "p"
410
+ ],
411
+ [
412
+ "r",
413
+ "e"
414
+ ],
415
+ [
416
+ "11",
417
+ "1"
418
+ ],
419
+ [
420
+ "00",
421
+ "00"
422
+ ],
423
+ [
424
+ "0",
425
+ "11"
426
+ ],
427
+ [
428
+ "10",
429
+ "1"
430
+ ],
431
+ [
432
+ "00",
433
+ "1"
434
+ ],
435
+ [
436
+ "ĠS",
437
+ "tep"
438
+ ],
439
+ [
440
+ "0",
441
+ "10"
442
+ ],
443
+ [
444
+ "i",
445
+ "n"
446
+ ],
447
+ [
448
+ "i",
449
+ "m"
450
+ ],
451
+ [
452
+ "ĠS",
453
+ "im"
454
+ ],
455
+ [
456
+ "11",
457
+ "11"
458
+ ],
459
+ [
460
+ "Ġ",
461
+ "["
462
+ ],
463
+ [
464
+ "o",
465
+ "r"
466
+ ],
467
+ [
468
+ "Ġ",
469
+ "I"
470
+ ],
471
+ [
472
+ "h",
473
+ "o"
474
+ ],
475
+ [
476
+ "o",
477
+ "d"
478
+ ],
479
+ [
480
+ "ho",
481
+ "od"
482
+ ],
483
+ [
484
+ "Ġ",
485
+ "in"
486
+ ],
487
+ [
488
+ "Ġ",
489
+ "11"
490
+ ],
491
+ [
492
+ "Ġ",
493
+ "10"
494
+ ],
495
+ [
496
+ "n",
497
+ "e"
498
+ ],
499
+ [
500
+ "Ġ",
501
+ "ne"
502
+ ],
503
+ [
504
+ "'",
505
+ "]"
506
+ ],
507
+ [
508
+ "b",
509
+ "or"
510
+ ],
511
+ [
512
+ "d",
513
+ "x"
514
+ ],
515
+ [
516
+ "g",
517
+ "h"
518
+ ],
519
+ [
520
+ "i",
521
+ "re"
522
+ ],
523
+ [
524
+ "i",
525
+ "gh"
526
+ ],
527
+ [
528
+ "q",
529
+ "u"
530
+ ],
531
+ [
532
+ "Ġ",
533
+ "re"
534
+ ],
535
+ [
536
+ "Ġ[",
537
+ "'"
538
+ ],
539
+ [
540
+ "ĠI",
541
+ "dx"
542
+ ],
543
+ [
544
+ "Ġne",
545
+ "igh"
546
+ ],
547
+ [
548
+ "bor",
549
+ "hood"
550
+ ],
551
+ [
552
+ "ire",
553
+ "s"
554
+ ],
555
+ [
556
+ "qu",
557
+ "ires"
558
+ ],
559
+ [
560
+ "Ġre",
561
+ "quires"
562
+ ],
563
+ [
564
+ "Ġneigh",
565
+ "borhood"
566
+ ],
567
+ [
568
+ "S",
569
+ "tep"
570
+ ],
571
+ [
572
+ "u",
573
+ "t"
574
+ ],
575
+ [
576
+ "Ġ",
577
+ "3"
578
+ ],
579
+ [
580
+ "Ġ",
581
+ "2"
582
+ ],
583
+ [
584
+ "Ġ",
585
+ "4"
586
+ ],
587
+ [
588
+ "Ġ",
589
+ "5"
590
+ ],
591
+ [
592
+ "Ġ",
593
+ "6"
594
+ ],
595
+ [
596
+ "Ġ",
597
+ "9"
598
+ ],
599
+ [
600
+ "Ġ",
601
+ "8"
602
+ ],
603
+ [
604
+ "Ġ",
605
+ "7"
606
+ ],
607
+ [
608
+ "Ġ",
609
+ "0"
610
+ ],
611
+ [
612
+ "c",
613
+ "a"
614
+ ],
615
+ [
616
+ "Ġ1",
617
+ "2"
618
+ ],
619
+ [
620
+ "Ġ1",
621
+ "3"
622
+ ],
623
+ [
624
+ "10",
625
+ "10"
626
+ ],
627
+ [
628
+ "Ġ1",
629
+ "4"
630
+ ],
631
+ [
632
+ "O",
633
+ "ut"
634
+ ],
635
+ [
636
+ "p",
637
+ "ut"
638
+ ],
639
+ [
640
+ "Ġ",
641
+ "Out"
642
+ ],
643
+ [
644
+ "ĠS",
645
+ "ca"
646
+ ],
647
+ [
648
+ "ĠOut",
649
+ "put"
650
+ ],
651
+ [
652
+ "ĠSca",
653
+ "n"
654
+ ],
655
+ [
656
+ "Ġ",
657
+ "011"
658
+ ],
659
+ [
660
+ "in",
661
+ "g"
662
+ ],
663
+ [
664
+ "t",
665
+ "i"
666
+ ],
667
+ [
668
+ "11",
669
+ "00"
670
+ ],
671
+ [
672
+ "0000",
673
+ "0000"
674
+ ],
675
+ [
676
+ "Ġ1",
677
+ "5"
678
+ ],
679
+ [
680
+ "Ġ",
681
+ "000"
682
+ ],
683
+ [
684
+ "o",
685
+ "n"
686
+ ],
687
+ [
688
+ "Ġ",
689
+ "110"
690
+ ],
691
+ [
692
+ "u",
693
+ "l"
694
+ ],
695
+ [
696
+ "Ġ1",
697
+ "00"
698
+ ],
699
+ [
700
+ "a",
701
+ "t"
702
+ ],
703
+ [
704
+ "ti",
705
+ "on"
706
+ ],
707
+ [
708
+ "Ġ",
709
+ "010"
710
+ ],
711
+ [
712
+ "10",
713
+ "11"
714
+ ],
715
+ [
716
+ "Ġ",
717
+ "C"
718
+ ],
719
+ [
720
+ "Ġ",
721
+ "00"
722
+ ],
723
+ [
724
+ "000",
725
+ "11"
726
+ ],
727
+ [
728
+ "5",
729
+ "0"
730
+ ],
731
+ [
732
+ "e",
733
+ "r"
734
+ ],
735
+ [
736
+ "100",
737
+ "00"
738
+ ],
739
+ [
740
+ "n",
741
+ "d"
742
+ ],
743
+ [
744
+ "Ġ1",
745
+ "6"
746
+ ],
747
+ [
748
+ "ul",
749
+ "e"
750
+ ],
751
+ [
752
+ "Ġ",
753
+ "R"
754
+ ],
755
+ [
756
+ "e",
757
+ "d"
758
+ ],
759
+ [
760
+ "l",
761
+ "i"
762
+ ],
763
+ [
764
+ "Ġ",
765
+ "001"
766
+ ],
767
+ [
768
+ "Ġ",
769
+ "101"
770
+ ],
771
+ [
772
+ "a",
773
+ "te"
774
+ ],
775
+ [
776
+ "ĠR",
777
+ "ule"
778
+ ],
779
+ [
780
+ "1",
781
+ "000"
782
+ ],
783
+ [
784
+ "f",
785
+ "i"
786
+ ],
787
+ [
788
+ "10",
789
+ "100"
790
+ ],
791
+ [
792
+ "a",
793
+ "s"
794
+ ],
795
+ [
796
+ "Ġ1",
797
+ "7"
798
+ ],
799
+ [
800
+ "Ġ",
801
+ "M"
802
+ ],
803
+ [
804
+ "Ġ",
805
+ "1111"
806
+ ],
807
+ [
808
+ "V",
809
+ "er"
810
+ ],
811
+ [
812
+ "c",
813
+ "h"
814
+ ],
815
+ [
816
+ "e",
817
+ "s"
818
+ ],
819
+ [
820
+ "i",
821
+ "fi"
822
+ ],
823
+ [
824
+ "m",
825
+ "p"
826
+ ],
827
+ [
828
+ "t",
829
+ "ing"
830
+ ],
831
+ [
832
+ "ca",
833
+ "tion"
834
+ ],
835
+ [
836
+ "at",
837
+ "ch"
838
+ ],
839
+ [
840
+ "Ver",
841
+ "ifi"
842
+ ],
843
+ [
844
+ "Verifi",
845
+ "cation"
846
+ ],
847
+ [
848
+ ".",
849
+ "."
850
+ ],
851
+ [
852
+ "110",
853
+ "11"
854
+ ],
855
+ [
856
+ "4",
857
+ "1"
858
+ ],
859
+ [
860
+ "a",
861
+ "nd"
862
+ ],
863
+ [
864
+ "d",
865
+ "ate"
866
+ ],
867
+ [
868
+ "i",
869
+ "date"
870
+ ],
871
+ [
872
+ "ĠC",
873
+ "and"
874
+ ],
875
+ [
876
+ "ĠCand",
877
+ "idate"
878
+ ],
879
+ [
880
+ "5",
881
+ "8"
882
+ ],
883
+ [
884
+ "at",
885
+ "ing"
886
+ ],
887
+ [
888
+ "100",
889
+ "11"
890
+ ],
891
+ [
892
+ "c",
893
+ "tion"
894
+ ],
895
+ [
896
+ "111",
897
+ "000"
898
+ ],
899
+ [
900
+ "0000",
901
+ "00"
902
+ ],
903
+ [
904
+ "10",
905
+ "110"
906
+ ],
907
+ [
908
+ "Ġ",
909
+ "00000000"
910
+ ],
911
+ [
912
+ "i",
913
+ "s"
914
+ ],
915
+ [
916
+ "m",
917
+ "atch"
918
+ ],
919
+ [
920
+ "ĠM",
921
+ "is"
922
+ ],
923
+ [
924
+ "ĠMis",
925
+ "match"
926
+ ],
927
+ [
928
+ "1111",
929
+ "11"
930
+ ],
931
+ [
932
+ "T",
933
+ "as"
934
+ ],
935
+ [
936
+ "Tas",
937
+ "k"
938
+ ],
939
+ [
940
+ "Ġ1",
941
+ "8"
942
+ ],
943
+ [
944
+ "G",
945
+ "re"
946
+ ],
947
+ [
948
+ "P",
949
+ "r"
950
+ ],
951
+ [
952
+ "Ġ",
953
+ "L"
954
+ ],
955
+ [
956
+ "Ġ",
957
+ "Gre"
958
+ ],
959
+ [
960
+ "Ġ",
961
+ "Pr"
962
+ ],
963
+ [
964
+ "ed",
965
+ "y"
966
+ ],
967
+ [
968
+ "ĠGre",
969
+ "edy"
970
+ ],
971
+ [
972
+ "A",
973
+ "t"
974
+ ],
975
+ [
976
+ "E",
977
+ "x"
978
+ ],
979
+ [
980
+ "T",
981
+ "es"
982
+ ],
983
+ [
984
+ "Ġ",
985
+ "Tes"
986
+ ],
987
+ [
988
+ "te",
989
+ "mp"
990
+ ],
991
+ [
992
+ "At",
993
+ "temp"
994
+ ],
995
+ [
996
+ "ĠTes",
997
+ "ting"
998
+ ],
999
+ [
1000
+ "Attemp",
1001
+ "t"
1002
+ ],
1003
+ [
1004
+ "100",
1005
+ "000"
1006
+ ],
1007
+ [
1008
+ "Ġ",
1009
+ "s"
1010
+ ],
1011
+ [
1012
+ "Ġ",
1013
+ "0000"
1014
+ ],
1015
+ [
1016
+ "u",
1017
+ "ction"
1018
+ ],
1019
+ [
1020
+ "ĠStep",
1021
+ "s"
1022
+ ],
1023
+ [
1024
+ "x",
1025
+ "t"
1026
+ ],
1027
+ [
1028
+ "10",
1029
+ "1111"
1030
+ ],
1031
+ [
1032
+ "a",
1033
+ "l"
1034
+ ],
1035
+ [
1036
+ "c",
1037
+ "ul"
1038
+ ],
1039
+ [
1040
+ "e",
1041
+ "li"
1042
+ ],
1043
+ [
1044
+ "i",
1045
+ "or"
1046
+ ],
1047
+ [
1048
+ "k",
1049
+ "eli"
1050
+ ],
1051
+ [
1052
+ "v",
1053
+ "er"
1054
+ ],
1055
+ [
1056
+ "ca",
1057
+ "l"
1058
+ ],
1059
+ [
1060
+ "cul",
1061
+ "ating"
1062
+ ],
1063
+ [
1064
+ "ior",
1065
+ "s"
1066
+ ],
1067
+ [
1068
+ "keli",
1069
+ "hood"
1070
+ ],
1071
+ [
1072
+ "Ġ1",
1073
+ "9"
1074
+ ],
1075
+ [
1076
+ "1111",
1077
+ "1111"
1078
+ ],
1079
+ [
1080
+ ")",
1081
+ "."
1082
+ ]
1083
+ ]
1084
+ }
1085
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<|startoftext|>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<|endoftext|>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "<|unknown|>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "<|pad|>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ },
35
+ "4": {
36
+ "content": "<think>",
37
+ "lstrip": false,
38
+ "normalized": false,
39
+ "rstrip": false,
40
+ "single_word": false,
41
+ "special": true
42
+ },
43
+ "5": {
44
+ "content": "</think>",
45
+ "lstrip": false,
46
+ "normalized": false,
47
+ "rstrip": false,
48
+ "single_word": false,
49
+ "special": true
50
+ }
51
+ },
52
+ "additional_special_tokens": [
53
+ "<think>",
54
+ "</think>"
55
+ ],
56
+ "bos_token": "<|startoftext|>",
57
+ "clean_up_tokenization_spaces": false,
58
+ "eos_token": "<|endoftext|>",
59
+ "extra_special_tokens": {},
60
+ "model_max_length": 1000000000000000019884624838656,
61
+ "pad_token": "<|pad|>",
62
+ "tokenizer_class": "PreTrainedTokenizerFast",
63
+ "unk_token": "<|unknown|>"
64
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:478c5792cae56598ec41be11b09c6fe07ef08f3d28ac262adfe2b7d757dfe6fa
3
+ size 5841