cfahlgren1 HF Staff commited on
Commit
c2cb1ec
·
verified ·
1 Parent(s): 56bf54f

copy the model files to the top level

Browse files
Files changed (5) hide show
  1. best.pt +3 -0
  2. model.onnx +3 -0
  3. model_quint8.onnx +3 -0
  4. report.json +548 -0
  5. vocab.json +9 -0
best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8f2a663ad7a697a00ebf6bd15942824fcefa46975f9935c7ac6dd1eceedf3a36
3
+ size 3780287
model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1ccc402ffaa27e6382114ad0d44a2e5a01e9beabe9d81e3c3ae0ad2f9676627b
3
+ size 3780016
model_quint8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:079424a9d2a72b9e5b2d6002c6466b836023c088780b7d292748f37564dc143b
3
+ size 981294
report.json ADDED
@@ -0,0 +1,548 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "args": {
3
+ "hidden": 512,
4
+ "epochs": 30,
5
+ "batch": 256,
6
+ "seq": 256,
7
+ "burn_in": 32,
8
+ "lr": 0.0015,
9
+ "typo_rate": 0.08,
10
+ "limit_frac": 1.0,
11
+ "eval_positions": 5000,
12
+ "context": 128,
13
+ "horizon": 48,
14
+ "repo": "cfahlgren1/email-autocomplete-gru",
15
+ "run_name": "gru-512",
16
+ "no_push": false
17
+ },
18
+ "device": "cuda",
19
+ "parameters": 944354,
20
+ "data": {
21
+ "train": {
22
+ "emails": 119859,
23
+ "chars": 40136658
24
+ },
25
+ "val": {
26
+ "emails": 8252,
27
+ "chars": 2775980
28
+ },
29
+ "test": {
30
+ "emails": 6922,
31
+ "chars": 2710169
32
+ }
33
+ },
34
+ "history": [
35
+ {
36
+ "epoch": 1,
37
+ "train_loss": 1.7093,
38
+ "val_loss": 1.3716,
39
+ "seconds": 23.9,
40
+ "chars_per_sec": 1678766
41
+ },
42
+ {
43
+ "epoch": 2,
44
+ "train_loss": 1.2792,
45
+ "val_loss": 1.2911,
46
+ "seconds": 23.6,
47
+ "chars_per_sec": 1701572
48
+ },
49
+ {
50
+ "epoch": 3,
51
+ "train_loss": 1.2231,
52
+ "val_loss": 1.2582,
53
+ "seconds": 23.6,
54
+ "chars_per_sec": 1700904
55
+ },
56
+ {
57
+ "epoch": 4,
58
+ "train_loss": 1.195,
59
+ "val_loss": 1.2396,
60
+ "seconds": 23.6,
61
+ "chars_per_sec": 1702926
62
+ },
63
+ {
64
+ "epoch": 5,
65
+ "train_loss": 1.1768,
66
+ "val_loss": 1.2254,
67
+ "seconds": 23.5,
68
+ "chars_per_sec": 1703932
69
+ },
70
+ {
71
+ "epoch": 6,
72
+ "train_loss": 1.1639,
73
+ "val_loss": 1.2171,
74
+ "seconds": 23.6,
75
+ "chars_per_sec": 1701905
76
+ },
77
+ {
78
+ "epoch": 7,
79
+ "train_loss": 1.1538,
80
+ "val_loss": 1.2087,
81
+ "seconds": 23.6,
82
+ "chars_per_sec": 1701450
83
+ },
84
+ {
85
+ "epoch": 8,
86
+ "train_loss": 1.1456,
87
+ "val_loss": 1.2043,
88
+ "seconds": 23.6,
89
+ "chars_per_sec": 1702601
90
+ },
91
+ {
92
+ "epoch": 9,
93
+ "train_loss": 1.1388,
94
+ "val_loss": 1.1969,
95
+ "seconds": 23.5,
96
+ "chars_per_sec": 1703175
97
+ },
98
+ {
99
+ "epoch": 10,
100
+ "train_loss": 1.1327,
101
+ "val_loss": 1.193,
102
+ "seconds": 23.6,
103
+ "chars_per_sec": 1701059
104
+ },
105
+ {
106
+ "epoch": 11,
107
+ "train_loss": 1.1276,
108
+ "val_loss": 1.1895,
109
+ "seconds": 23.6,
110
+ "chars_per_sec": 1702827
111
+ },
112
+ {
113
+ "epoch": 12,
114
+ "train_loss": 1.1228,
115
+ "val_loss": 1.1856,
116
+ "seconds": 23.6,
117
+ "chars_per_sec": 1700650
118
+ },
119
+ {
120
+ "epoch": 13,
121
+ "train_loss": 1.1186,
122
+ "val_loss": 1.1824,
123
+ "seconds": 23.5,
124
+ "chars_per_sec": 1703353
125
+ },
126
+ {
127
+ "epoch": 14,
128
+ "train_loss": 1.1141,
129
+ "val_loss": 1.1791,
130
+ "seconds": 23.6,
131
+ "chars_per_sec": 1702586
132
+ },
133
+ {
134
+ "epoch": 15,
135
+ "train_loss": 1.1108,
136
+ "val_loss": 1.1765,
137
+ "seconds": 23.6,
138
+ "chars_per_sec": 1702570
139
+ },
140
+ {
141
+ "epoch": 16,
142
+ "train_loss": 1.1075,
143
+ "val_loss": 1.175,
144
+ "seconds": 23.6,
145
+ "chars_per_sec": 1702686
146
+ },
147
+ {
148
+ "epoch": 17,
149
+ "train_loss": 1.1041,
150
+ "val_loss": 1.1718,
151
+ "seconds": 23.5,
152
+ "chars_per_sec": 1703957
153
+ },
154
+ {
155
+ "epoch": 18,
156
+ "train_loss": 1.1015,
157
+ "val_loss": 1.17,
158
+ "seconds": 23.6,
159
+ "chars_per_sec": 1701778
160
+ },
161
+ {
162
+ "epoch": 19,
163
+ "train_loss": 1.0985,
164
+ "val_loss": 1.1678,
165
+ "seconds": 23.6,
166
+ "chars_per_sec": 1702612
167
+ },
168
+ {
169
+ "epoch": 20,
170
+ "train_loss": 1.0959,
171
+ "val_loss": 1.1659,
172
+ "seconds": 23.6,
173
+ "chars_per_sec": 1702960
174
+ },
175
+ {
176
+ "epoch": 21,
177
+ "train_loss": 1.0935,
178
+ "val_loss": 1.1642,
179
+ "seconds": 23.6,
180
+ "chars_per_sec": 1701776
181
+ },
182
+ {
183
+ "epoch": 22,
184
+ "train_loss": 1.0913,
185
+ "val_loss": 1.1632,
186
+ "seconds": 23.6,
187
+ "chars_per_sec": 1702317
188
+ },
189
+ {
190
+ "epoch": 23,
191
+ "train_loss": 1.0896,
192
+ "val_loss": 1.1624,
193
+ "seconds": 23.6,
194
+ "chars_per_sec": 1701215
195
+ },
196
+ {
197
+ "epoch": 24,
198
+ "train_loss": 1.0874,
199
+ "val_loss": 1.1613,
200
+ "seconds": 23.6,
201
+ "chars_per_sec": 1701278
202
+ },
203
+ {
204
+ "epoch": 25,
205
+ "train_loss": 1.0863,
206
+ "val_loss": 1.1606,
207
+ "seconds": 23.6,
208
+ "chars_per_sec": 1700749
209
+ },
210
+ {
211
+ "epoch": 26,
212
+ "train_loss": 1.085,
213
+ "val_loss": 1.1596,
214
+ "seconds": 23.6,
215
+ "chars_per_sec": 1701575
216
+ },
217
+ {
218
+ "epoch": 27,
219
+ "train_loss": 1.0839,
220
+ "val_loss": 1.1592,
221
+ "seconds": 23.6,
222
+ "chars_per_sec": 1699705
223
+ },
224
+ {
225
+ "epoch": 28,
226
+ "train_loss": 1.083,
227
+ "val_loss": 1.1588,
228
+ "seconds": 23.6,
229
+ "chars_per_sec": 1700990
230
+ },
231
+ {
232
+ "epoch": 29,
233
+ "train_loss": 1.0823,
234
+ "val_loss": 1.1588,
235
+ "seconds": 23.6,
236
+ "chars_per_sec": 1702216
237
+ },
238
+ {
239
+ "epoch": 30,
240
+ "train_loss": 1.0821,
241
+ "val_loss": 1.1586,
242
+ "seconds": 23.5,
243
+ "chars_per_sec": 1704907
244
+ }
245
+ ],
246
+ "best_val_loss": 1.1586,
247
+ "test_positions": 5000,
248
+ "held_out_sender_suggestions": {
249
+ "0.05": {
250
+ "coverage": 0.624,
251
+ "precision": 0.2064,
252
+ "avg_chars_when_correct": 6.93,
253
+ "chars_saved_per_position": 0.892
254
+ },
255
+ "0.1": {
256
+ "coverage": 0.4622,
257
+ "precision": 0.3124,
258
+ "avg_chars_when_correct": 6.41,
259
+ "chars_saved_per_position": 0.925
260
+ },
261
+ "0.2": {
262
+ "coverage": 0.2974,
263
+ "precision": 0.4808,
264
+ "avg_chars_when_correct": 6.03,
265
+ "chars_saved_per_position": 0.863
266
+ },
267
+ "0.35": {
268
+ "coverage": 0.1796,
269
+ "precision": 0.6559,
270
+ "avg_chars_when_correct": 5.8,
271
+ "chars_saved_per_position": 0.683
272
+ },
273
+ "0.5": {
274
+ "coverage": 0.1114,
275
+ "precision": 0.7953,
276
+ "avg_chars_when_correct": 5.64,
277
+ "chars_saved_per_position": 0.5
278
+ },
279
+ "0.7": {
280
+ "coverage": 0.0694,
281
+ "precision": 0.8991,
282
+ "avg_chars_when_correct": 5.4,
283
+ "chars_saved_per_position": 0.337
284
+ }
285
+ },
286
+ "exports": {
287
+ "float": {
288
+ "size_mb": 3.78,
289
+ "native_cpu_single_thread_step": {
290
+ "p50_ms": 0.114,
291
+ "p95_ms": 0.125
292
+ },
293
+ "suggestions_1000_positions": {
294
+ "0.05": {
295
+ "coverage": 0.638,
296
+ "precision": 0.2132,
297
+ "avg_chars_when_correct": 6.7,
298
+ "chars_saved_per_position": 0.911
299
+ },
300
+ "0.1": {
301
+ "coverage": 0.496,
302
+ "precision": 0.3085,
303
+ "avg_chars_when_correct": 6.3,
304
+ "chars_saved_per_position": 0.964
305
+ },
306
+ "0.2": {
307
+ "coverage": 0.312,
308
+ "precision": 0.4583,
309
+ "avg_chars_when_correct": 5.94,
310
+ "chars_saved_per_position": 0.849
311
+ },
312
+ "0.35": {
313
+ "coverage": 0.202,
314
+ "precision": 0.604,
315
+ "avg_chars_when_correct": 5.77,
316
+ "chars_saved_per_position": 0.704
317
+ },
318
+ "0.5": {
319
+ "coverage": 0.119,
320
+ "precision": 0.7479,
321
+ "avg_chars_when_correct": 5.58,
322
+ "chars_saved_per_position": 0.497
323
+ },
324
+ "0.7": {
325
+ "coverage": 0.078,
326
+ "precision": 0.8718,
327
+ "avg_chars_when_correct": 5.21,
328
+ "chars_saved_per_position": 0.354
329
+ }
330
+ }
331
+ },
332
+ "quint8": {
333
+ "size_mb": 0.98,
334
+ "native_cpu_single_thread_step": {
335
+ "p50_ms": 0.059,
336
+ "p95_ms": 0.084
337
+ },
338
+ "suggestions_1000_positions": {
339
+ "0.05": {
340
+ "coverage": 0.655,
341
+ "precision": 0.2,
342
+ "avg_chars_when_correct": 6.83,
343
+ "chars_saved_per_position": 0.895
344
+ },
345
+ "0.1": {
346
+ "coverage": 0.492,
347
+ "precision": 0.311,
348
+ "avg_chars_when_correct": 6.35,
349
+ "chars_saved_per_position": 0.971
350
+ },
351
+ "0.2": {
352
+ "coverage": 0.318,
353
+ "precision": 0.4686,
354
+ "avg_chars_when_correct": 5.97,
355
+ "chars_saved_per_position": 0.89
356
+ },
357
+ "0.35": {
358
+ "coverage": 0.198,
359
+ "precision": 0.6414,
360
+ "avg_chars_when_correct": 5.66,
361
+ "chars_saved_per_position": 0.719
362
+ },
363
+ "0.5": {
364
+ "coverage": 0.123,
365
+ "precision": 0.7642,
366
+ "avg_chars_when_correct": 5.53,
367
+ "chars_saved_per_position": 0.52
368
+ },
369
+ "0.7": {
370
+ "coverage": 0.081,
371
+ "precision": 0.8889,
372
+ "avg_chars_when_correct": 5.42,
373
+ "chars_saved_per_position": 0.39
374
+ }
375
+ }
376
+ }
377
+ },
378
+ "prompt_samples": {
379
+ "Hi John,\n\nThanks for ": {
380
+ "continuation": "the info. I will be out of the office tomorrow a",
381
+ "suggestion_at_0.2": "the"
382
+ },
383
+ "Please let me know if ": {
384
+ "continuation": "you have any questions.\n\nThanks,\n\nKayThe follow",
385
+ "suggestion_at_0.2": "you have any"
386
+ },
387
+ "I will send you ": {
388
+ "continuation": "a copy of the attached document. I will be out o",
389
+ "suggestion_at_0.2": "a"
390
+ },
391
+ "Let me kn": {
392
+ "continuation": "ow if you have any questions.\n\nThanks,\n\nKayThe ",
393
+ "suggestion_at_0.2": "ow if you have"
394
+ },
395
+ "Thnaks for the ": {
396
+ "continuation": "info. I will be out of the office tomorrow and w",
397
+ "suggestion_at_0.2": ""
398
+ },
399
+ "I look forward to ": {
400
+ "continuation": "seeing you there. I will be out of the office to",
401
+ "suggestion_at_0.2": ""
402
+ },
403
+ "Can we schedule a ": {
404
+ "continuation": "meeting to discuss the attached spreadsheet for ",
405
+ "suggestion_at_0.2": "meeting"
406
+ },
407
+ "Sorry for the dela": {
408
+ "continuation": "y. I will be out of the office tomorrow and we w",
409
+ "suggestion_at_0.2": "y."
410
+ },
411
+ "Plese find attached ": {
412
+ "continuation": "the following contract for the Enron Corp. Gas P",
413
+ "suggestion_at_0.2": "the"
414
+ },
415
+ "We need to discuss ": {
416
+ "continuation": "the attached spreadsheet for the following dates",
417
+ "suggestion_at_0.2": "the"
418
+ }
419
+ },
420
+ "held_out_examples": [
421
+ {
422
+ "context": "ou trading a book yet? Playing any poker? Bingham was here f",
423
+ "truth": "or Q-School last week. Didn't go so well.",
424
+ "suggestion_at_0.2": "or"
425
+ },
426
+ {
427
+ "context": "Please use this model. I had a few change",
428
+ "truth": "s after talking to the guys in the west. Thanks!",
429
+ "suggestion_at_0.2": "s to"
430
+ },
431
+ {
432
+ "context": "not open the attachment. Please, send it as a MS Word docume",
433
+ "truth": "nt or a .pdf file.",
434
+ "suggestion_at_0.2": "nt."
435
+ },
436
+ {
437
+ "context": "n I do to help. I would also like to visit on status of e*p ",
438
+ "truth": "group and mariner.",
439
+ "suggestion_at_0.2": ""
440
+ },
441
+ {
442
+ "context": " complete hourly schedule for June 1, including IBM and inco",
443
+ "truth": "rporating all known changes confirmed by that ti",
444
+ "suggestion_at_0.2": "me"
445
+ },
446
+ {
447
+ "context": " delivery rights at SoCal/Topock for at least a pro rata por",
448
+ "truth": "tion of its released capacity.",
449
+ "suggestion_at_0.2": "tion of"
450
+ },
451
+ {
452
+ "context": "i talked to chris a li",
453
+ "truth": "ttle bit and he is going to call back tonight at",
454
+ "suggestion_at_0.2": "ttle"
455
+ },
456
+ {
457
+ "context": " I might be missing something about the way you guys do thes",
458
+ "truth": "e deals in Canada so I thought I would run this ",
459
+ "suggestion_at_0.2": "e"
460
+ },
461
+ {
462
+ "context": "Traci,\n\nThanks. This is the best solution given all the unce",
463
+ "truth": "rtainties we face.",
464
+ "suggestion_at_0.2": "rtainty"
465
+ },
466
+ {
467
+ "context": "\nAquila,s strategy comes as little surprise to Stuebi. As mu",
468
+ "truth": "ch success as the company has had, he said he th",
469
+ "suggestion_at_0.2": "ch"
470
+ },
471
+ {
472
+ "context": "Jim,\n\nWhat about 2 p.m., my time, on Friday, Au",
473
+ "truth": "gust 17. If this works for you, I shall try to i",
474
+ "suggestion_at_0.2": "gust"
475
+ },
476
+ {
477
+ "context": "'s and CDD's for Philedelphia, Allentown Washington DC, Balt",
478
+ "truth": "imore, Pittsburg, Newark, New York City, Cinncin",
479
+ "suggestion_at_0.2": "imore"
480
+ },
481
+ {
482
+ "context": " release.\n\nThe company will offer derivatives to hedge volat",
483
+ "truth": "ile electricity prices.",
484
+ "suggestion_at_0.2": "ility"
485
+ },
486
+ {
487
+ "context": "day, I need access to West deals in Sitara. At a mininum I s",
488
+ "truth": "till need access to all Chicago/Midwest deals al",
489
+ "suggestion_at_0.2": "ent"
490
+ },
491
+ {
492
+ "context": "on, attached for your review is a draft of the First Amendme",
493
+ "truth": "nt to Master Agreement between Enron North Ameri",
494
+ "suggestion_at_0.2": "nt to"
495
+ },
496
+ {
497
+ "context": "I have provided all the info that Tami Scudder ",
498
+ "truth": "requested today.",
499
+ "suggestion_at_0.2": "is"
500
+ },
501
+ {
502
+ "context": "ontana Power Company (MPC) System Border for Thursday and Fr",
503
+ "truth": "iday (July 12, 13) @ $67.85/MWh. Thanks, John",
504
+ "suggestion_at_0.2": "iday"
505
+ },
506
+ {
507
+ "context": "e EnBank contracts physical or financial? It seems to me we ",
508
+ "truth": "should use the same customer profiles - that is ",
509
+ "suggestion_at_0.2": ""
510
+ },
511
+ {
512
+ "context": "I have an appointment at 1:00 today and should be ",
513
+ "truth": "back to the office between 2 - 2:30.",
514
+ "suggestion_at_0.2": ""
515
+ },
516
+ {
517
+ "context": "I'm glad everythi",
518
+ "truth": "ng worked out for you. Hope you have a great tim",
519
+ "suggestion_at_0.2": "ng is"
520
+ },
521
+ {
522
+ "context": "o's Line 1100 from Eunice Station to Pecos River Station wil",
523
+ "truth": "l be out of service for 16 hours Saturday beginn",
524
+ "suggestion_at_0.2": "l be"
525
+ },
526
+ {
527
+ "context": "rty\" in the second line of paragraph 1. Otherwise they look ",
528
+ "truth": "great to me.",
529
+ "suggestion_at_0.2": ""
530
+ },
531
+ {
532
+ "context": "on Houston in the next day or so.\n\nPlease call me if you hav",
533
+ "truth": "e any questions.",
534
+ "suggestion_at_0.2": "e any questions."
535
+ },
536
+ {
537
+ "context": "s not guarantee any continuity of support for you.\n\nPlease, ",
538
+ "truth": "let me know what you decide to do over the next ",
539
+ "suggestion_at_0.2": ""
540
+ },
541
+ {
542
+ "context": " tend to agree with you, but Nora Dobin is the lawyer who kn",
543
+ "truth": "ows the most about the DealBench system - you sh",
544
+ "suggestion_at_0.2": "ows"
545
+ }
546
+ ],
547
+ "total_seconds": 831.5
548
+ }
vocab.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "pad": 0,
3
+ "eos": 1,
4
+ "newline": 2,
5
+ "first_printable_id": 3,
6
+ "first_printable_codepoint": 32,
7
+ "size": 98,
8
+ "hidden": 512
9
+ }