anudit commited on
Commit
2b59411
·
verified ·
1 Parent(s): a4dc013

Upload run_manifest.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. run_manifest.json +595 -0
run_manifest.json ADDED
@@ -0,0 +1,595 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "format": "lex-training-run-v1",
3
+ "args": {
4
+ "dataset": "./corpus/dataset",
5
+ "total_tokens": 160000000,
6
+ "epochs": 48,
7
+ "batch_size": 64,
8
+ "lr": 0.003,
9
+ "warmup_epochs": 6,
10
+ "out_dir": "./checkpoints_student",
11
+ "workers": 8,
12
+ "precision": "bf16",
13
+ "compile_mode": "none",
14
+ "matmul_precision": "high",
15
+ "pin_memory": true,
16
+ "prefetch_factor": 4,
17
+ "fused_optimizer": true,
18
+ "max_steps": 0,
19
+ "film_rank": 64,
20
+ "erase_rank": 16,
21
+ "sampler": "tempered",
22
+ "sampling_exponent": 0.5,
23
+ "tail_floor": 0.00025,
24
+ "include_excluded_languages": false,
25
+ "construct_boost": 1.0,
26
+ "boundary_boost": 1.0,
27
+ "calibration_fraction": 0.167,
28
+ "teacher_checkpoint": "/workspace/lex/train_large/checkpoints_teacher/best_model.pt",
29
+ "teacher_logits": "./corpus/dataset/teacher_logits.npy",
30
+ "distill_weight": 0.2,
31
+ "distill_temperature": 2.0,
32
+ "full_precision": false,
33
+ "seed": 20260911,
34
+ "lang_loss": 0.15,
35
+ "struct_loss": 0.2,
36
+ "mine_weak_languages": true,
37
+ "real_bench_root": "",
38
+ "dim": 96,
39
+ "embed_dim": 64,
40
+ "head_hidden": 192,
41
+ "resume": ""
42
+ },
43
+ "model": {
44
+ "total_parameters": 434602,
45
+ "quantized_params": 440896,
46
+ "fp16_params": 9700,
47
+ "packed_bytes": 102368,
48
+ "packed_kb": 99.96875,
49
+ "breakdown": {
50
+ "embedding.table": {
51
+ "params": 108544,
52
+ "bits": 3,
53
+ "bytes": 40738
54
+ },
55
+ "embedding.up": {
56
+ "params": 6144,
57
+ "bits": 1,
58
+ "bytes": 960
59
+ },
60
+ "film.down": {
61
+ "params": 12288,
62
+ "bits": 1,
63
+ "bytes": 1664
64
+ },
65
+ "film.up": {
66
+ "params": 49152,
67
+ "bits": 1,
68
+ "bytes": 7680
69
+ },
70
+ "layers.0.dw": {
71
+ "params": 480,
72
+ "bits": 4,
73
+ "bytes": 432
74
+ },
75
+ "layers.0.proj_in": {
76
+ "params": 18432,
77
+ "bits": 1,
78
+ "bytes": 2688
79
+ },
80
+ "layers.0.proj_out": {
81
+ "params": 18432,
82
+ "bits": 1,
83
+ "bytes": 2496
84
+ },
85
+ "layers.0.erase_down": {
86
+ "params": 1536,
87
+ "bits": 1,
88
+ "bytes": 224
89
+ },
90
+ "layers.0.erase_up": {
91
+ "params": 6144,
92
+ "bits": 1,
93
+ "bytes": 1152
94
+ },
95
+ "layers.1.dw": {
96
+ "params": 480,
97
+ "bits": 4,
98
+ "bytes": 432
99
+ },
100
+ "layers.1.proj_in": {
101
+ "params": 18432,
102
+ "bits": 1,
103
+ "bytes": 2688
104
+ },
105
+ "layers.1.proj_out": {
106
+ "params": 18432,
107
+ "bits": 1,
108
+ "bytes": 2496
109
+ },
110
+ "layers.1.erase_down": {
111
+ "params": 1536,
112
+ "bits": 1,
113
+ "bytes": 224
114
+ },
115
+ "layers.1.erase_up": {
116
+ "params": 6144,
117
+ "bits": 1,
118
+ "bytes": 1152
119
+ },
120
+ "layers.2.dw": {
121
+ "params": 480,
122
+ "bits": 4,
123
+ "bytes": 432
124
+ },
125
+ "layers.2.proj_in": {
126
+ "params": 18432,
127
+ "bits": 1,
128
+ "bytes": 2688
129
+ },
130
+ "layers.2.proj_out": {
131
+ "params": 18432,
132
+ "bits": 1,
133
+ "bytes": 2496
134
+ },
135
+ "layers.2.erase_down": {
136
+ "params": 1536,
137
+ "bits": 1,
138
+ "bytes": 224
139
+ },
140
+ "layers.2.erase_up": {
141
+ "params": 6144,
142
+ "bits": 1,
143
+ "bytes": 1152
144
+ },
145
+ "layers.3.dw": {
146
+ "params": 480,
147
+ "bits": 4,
148
+ "bytes": 432
149
+ },
150
+ "layers.3.proj_in": {
151
+ "params": 18432,
152
+ "bits": 1,
153
+ "bytes": 2688
154
+ },
155
+ "layers.3.proj_out": {
156
+ "params": 18432,
157
+ "bits": 1,
158
+ "bytes": 2496
159
+ },
160
+ "layers.3.erase_down": {
161
+ "params": 1536,
162
+ "bits": 1,
163
+ "bytes": 224
164
+ },
165
+ "layers.3.erase_up": {
166
+ "params": 6144,
167
+ "bits": 1,
168
+ "bytes": 1152
169
+ },
170
+ "global_ctx.summary": {
171
+ "params": 36864,
172
+ "bits": 1,
173
+ "bytes": 4800
174
+ },
175
+ "global_ctx.gate": {
176
+ "params": 9216,
177
+ "bits": 1,
178
+ "bytes": 1344
179
+ },
180
+ "head_hidden": {
181
+ "params": 36864,
182
+ "bits": 1,
183
+ "bytes": 4992
184
+ },
185
+ "head_out": {
186
+ "params": 1728,
187
+ "bits": 1,
188
+ "bytes": 234
189
+ }
190
+ }
191
+ },
192
+ "dataset": {
193
+ "total_tokens": 118139781,
194
+ "files": 82862,
195
+ "languages_target": 193,
196
+ "meta_sha256": "7782709cfebc8802edf50c3a4ecfb557b42777cb5c9cd0957aa20405ef30ba33"
197
+ },
198
+ "training_languages": [
199
+ "1c",
200
+ "abnf",
201
+ "accesslog",
202
+ "actionscript",
203
+ "ada",
204
+ "angelscript",
205
+ "apache",
206
+ "applescript",
207
+ "arcade",
208
+ "arduino",
209
+ "armasm",
210
+ "asciidoc",
211
+ "aspectj",
212
+ "autohotkey",
213
+ "autoit",
214
+ "avrasm",
215
+ "awk",
216
+ "axapta",
217
+ "bash",
218
+ "basic",
219
+ "bnf",
220
+ "brainfuck",
221
+ "c",
222
+ "cal",
223
+ "capnproto",
224
+ "ceylon",
225
+ "clean",
226
+ "clojure",
227
+ "clojure-repl",
228
+ "cmake",
229
+ "coffeescript",
230
+ "coq",
231
+ "cos",
232
+ "cpp",
233
+ "crmsh",
234
+ "crystal",
235
+ "csharp",
236
+ "csp",
237
+ "css",
238
+ "d",
239
+ "dart",
240
+ "delphi",
241
+ "diff",
242
+ "django",
243
+ "dns",
244
+ "dockerfile",
245
+ "dos",
246
+ "dsconfig",
247
+ "dts",
248
+ "dust",
249
+ "ebnf",
250
+ "elixir",
251
+ "elm",
252
+ "erb",
253
+ "erlang",
254
+ "erlang-repl",
255
+ "excel",
256
+ "fix",
257
+ "flix",
258
+ "fortran",
259
+ "freedesktop",
260
+ "fsharp",
261
+ "gams",
262
+ "gauss",
263
+ "gcode",
264
+ "gherkin",
265
+ "glsl",
266
+ "gml",
267
+ "go",
268
+ "golo",
269
+ "gradle",
270
+ "graphql",
271
+ "groovy",
272
+ "haml",
273
+ "handlebars",
274
+ "haskell",
275
+ "haxe",
276
+ "hsp",
277
+ "http",
278
+ "hy",
279
+ "inform7",
280
+ "ini",
281
+ "irpf90",
282
+ "isbl",
283
+ "java",
284
+ "javascript",
285
+ "jboss-cli",
286
+ "json",
287
+ "julia",
288
+ "julia-repl",
289
+ "kotlin",
290
+ "lasso",
291
+ "latex",
292
+ "ldif",
293
+ "leaf",
294
+ "less",
295
+ "lisp",
296
+ "livecodeserver",
297
+ "livescript",
298
+ "llvm",
299
+ "lsl",
300
+ "lua",
301
+ "makefile",
302
+ "markdown",
303
+ "mathematica",
304
+ "matlab",
305
+ "maxima",
306
+ "mel",
307
+ "mercury",
308
+ "mipsasm",
309
+ "mizar",
310
+ "mojolicious",
311
+ "monkey",
312
+ "moonscript",
313
+ "n1ql",
314
+ "nestedtext",
315
+ "nginx",
316
+ "nim",
317
+ "nix",
318
+ "node-repl",
319
+ "nsis",
320
+ "objectivec",
321
+ "ocaml",
322
+ "openscad",
323
+ "oxygene",
324
+ "parser3",
325
+ "perl",
326
+ "pf",
327
+ "pgsql",
328
+ "php",
329
+ "php-template",
330
+ "plaintext",
331
+ "pony",
332
+ "powershell",
333
+ "processing",
334
+ "profile",
335
+ "prolog",
336
+ "properties",
337
+ "protobuf",
338
+ "puppet",
339
+ "purebasic",
340
+ "python",
341
+ "python-repl",
342
+ "q",
343
+ "qml",
344
+ "r",
345
+ "reasonml",
346
+ "rib",
347
+ "roboconf",
348
+ "routeros",
349
+ "rsl",
350
+ "ruby",
351
+ "ruleslanguage",
352
+ "rust",
353
+ "sas",
354
+ "scala",
355
+ "scheme",
356
+ "scilab",
357
+ "scss",
358
+ "shell",
359
+ "smali",
360
+ "smalltalk",
361
+ "sml",
362
+ "sqf",
363
+ "sql",
364
+ "stan",
365
+ "stata",
366
+ "step21",
367
+ "stylus",
368
+ "subunit",
369
+ "swift",
370
+ "taggerscript",
371
+ "tap",
372
+ "tcl",
373
+ "thrift",
374
+ "tp",
375
+ "twig",
376
+ "typescript",
377
+ "vala",
378
+ "vbnet",
379
+ "vbscript",
380
+ "vbscript-html",
381
+ "verilog",
382
+ "vhdl",
383
+ "vim",
384
+ "wasm",
385
+ "wren",
386
+ "x86asm",
387
+ "xl",
388
+ "xml",
389
+ "xquery",
390
+ "yaml",
391
+ "zephir"
392
+ ],
393
+ "excluded_but_evaluated": [],
394
+ "initial_language_probabilities": {
395
+ "1c": 0.003737311954438796,
396
+ "abnf": 0.003737311954438796,
397
+ "accesslog": 0.003737311954438796,
398
+ "actionscript": 0.003737311954438796,
399
+ "ada": 0.003737311954438796,
400
+ "angelscript": 0.003737311954438796,
401
+ "apache": 0.003737311954438796,
402
+ "applescript": 0.003737311954438796,
403
+ "arcade": 0.003737311954438796,
404
+ "arduino": 0.003737311954438796,
405
+ "armasm": 0.003737311954438796,
406
+ "asciidoc": 0.003737311954438796,
407
+ "aspectj": 0.003737311954438796,
408
+ "autohotkey": 0.003737311954438796,
409
+ "autoit": 0.003737311954438796,
410
+ "avrasm": 0.003737311954438796,
411
+ "awk": 0.0050465064595763606,
412
+ "axapta": 0.003737311954438796,
413
+ "bash": 0.015526979905295444,
414
+ "basic": 0.003737311954438796,
415
+ "bnf": 0.003737311954438796,
416
+ "brainfuck": 0.003737311954438796,
417
+ "c": 0.014767791012061991,
418
+ "cal": 0.003737311954438796,
419
+ "capnproto": 0.003737311954438796,
420
+ "ceylon": 0.003737311954438796,
421
+ "clean": 0.003737311954438796,
422
+ "clojure": 0.003737311954438796,
423
+ "clojure-repl": 0.003737311954438796,
424
+ "cmake": 0.0067521662708805415,
425
+ "coffeescript": 0.003737311954438796,
426
+ "coq": 0.003737311954438796,
427
+ "cos": 0.003737311954438796,
428
+ "cpp": 0.01660079170957557,
429
+ "crmsh": 0.003737311954438796,
430
+ "crystal": 0.003737311954438796,
431
+ "csharp": 0.017279633647396073,
432
+ "csp": 0.003737311954438796,
433
+ "css": 0.01244343818911347,
434
+ "d": 0.003737311954438796,
435
+ "dart": 0.008949353594248092,
436
+ "delphi": 0.003737311954438796,
437
+ "diff": 0.003737311954438796,
438
+ "django": 0.003737311954438796,
439
+ "dns": 0.003737311954438796,
440
+ "dockerfile": 0.006312042929887236,
441
+ "dos": 0.003737311954438796,
442
+ "dsconfig": 0.003737311954438796,
443
+ "dts": 0.003737311954438796,
444
+ "dust": 0.003737311954438796,
445
+ "ebnf": 0.003737311954438796,
446
+ "elixir": 0.003737311954438796,
447
+ "elm": 0.003737311954438796,
448
+ "erb": 0.003737311954438796,
449
+ "erlang": 0.003737311954438796,
450
+ "erlang-repl": 0.003737311954438796,
451
+ "excel": 0.003737311954438796,
452
+ "fix": 0.003737311954438796,
453
+ "flix": 0.003737311954438796,
454
+ "fortran": 0.003737311954438796,
455
+ "freedesktop": 0.003737311954438796,
456
+ "fsharp": 0.003737311954438796,
457
+ "gams": 0.003737311954438796,
458
+ "gauss": 0.003737311954438796,
459
+ "gcode": 0.003737311954438796,
460
+ "gherkin": 0.005323735450506969,
461
+ "glsl": 0.005323735450506969,
462
+ "gml": 0.003737311954438796,
463
+ "go": 0.014373166824822708,
464
+ "golo": 0.003737311954438796,
465
+ "gradle": 0.003737311954438796,
466
+ "graphql": 0.003737311954438796,
467
+ "groovy": 0.006312042929887236,
468
+ "haml": 0.003737311954438796,
469
+ "handlebars": 0.003737311954438796,
470
+ "haskell": 0.003737311954438796,
471
+ "haxe": 0.003737311954438796,
472
+ "hsp": 0.003737311954438796,
473
+ "http": 0.003737311954438796,
474
+ "hy": 0.003737311954438796,
475
+ "inform7": 0.003737311954438796,
476
+ "ini": 0.003737311954438796,
477
+ "irpf90": 0.003737311954438796,
478
+ "isbl": 0.003737311954438796,
479
+ "java": 0.02170441524240044,
480
+ "javascript": 0.03144152265863584,
481
+ "jboss-cli": 0.003737311954438796,
482
+ "json": 0.009265030661350164,
483
+ "julia": 0.003737311954438796,
484
+ "julia-repl": 0.003737311954438796,
485
+ "kotlin": 0.010432652910897988,
486
+ "lasso": 0.003737311954438796,
487
+ "latex": 0.003737311954438796,
488
+ "ldif": 0.003737311954438796,
489
+ "leaf": 0.003737311954438796,
490
+ "less": 0.003737311954438796,
491
+ "lisp": 0.003737311954438796,
492
+ "livecodeserver": 0.003737311954438796,
493
+ "livescript": 0.003737311954438796,
494
+ "llvm": 0.003737311954438796,
495
+ "lsl": 0.003737311954438796,
496
+ "lua": 0.007165306186793014,
497
+ "makefile": 0.007555890162738713,
498
+ "markdown": 0.01221021255201604,
499
+ "mathematica": 0.003737311954438796,
500
+ "matlab": 0.003737311954438796,
501
+ "maxima": 0.003737311954438796,
502
+ "mel": 0.003737311954438796,
503
+ "mercury": 0.003737311954438796,
504
+ "mipsasm": 0.003737311954438796,
505
+ "mizar": 0.003737311954438796,
506
+ "mojolicious": 0.003737311954438796,
507
+ "monkey": 0.003737311954438796,
508
+ "moonscript": 0.003737311954438796,
509
+ "n1ql": 0.003737311954438796,
510
+ "nestedtext": 0.003737311954438796,
511
+ "nginx": 0.003737311954438796,
512
+ "nim": 0.003737311954438796,
513
+ "nix": 0.003737311954438796,
514
+ "node-repl": 0.003737311954438796,
515
+ "nsis": 0.003737311954438796,
516
+ "objectivec": 0.003737311954438796,
517
+ "ocaml": 0.003737311954438796,
518
+ "openscad": 0.003737311954438796,
519
+ "oxygene": 0.003737311954438796,
520
+ "parser3": 0.003737311954438796,
521
+ "perl": 0.0067521662708805415,
522
+ "pf": 0.003737311954438796,
523
+ "pgsql": 0.003737311954438796,
524
+ "php": 0.015892980582043238,
525
+ "php-template": 0.003737311954438796,
526
+ "plaintext": 0.003737311954438796,
527
+ "pony": 0.003737311954438796,
528
+ "powershell": 0.007555890162738713,
529
+ "processing": 0.003737311954438796,
530
+ "profile": 0.003737311954438796,
531
+ "prolog": 0.003737311954438796,
532
+ "properties": 0.003737311954438796,
533
+ "protobuf": 0.003737311954438796,
534
+ "puppet": 0.003737311954438796,
535
+ "purebasic": 0.003737311954438796,
536
+ "python": 0.030038665018278524,
537
+ "python-repl": 0.003737311954438796,
538
+ "q": 0.003737311954438796,
539
+ "qml": 0.003737311954438796,
540
+ "r": 0.0067521662708805415,
541
+ "reasonml": 0.003737311954438796,
542
+ "rib": 0.003737311954438796,
543
+ "roboconf": 0.003737311954438796,
544
+ "routeros": 0.003737311954438796,
545
+ "rsl": 0.003737311954438796,
546
+ "ruby": 0.01267237220756891,
547
+ "ruleslanguage": 0.003737311954438796,
548
+ "rust": 0.011729857789575086,
549
+ "sas": 0.003737311954438796,
550
+ "scala": 0.007927252963810923,
551
+ "scheme": 0.003737311954438796,
552
+ "scilab": 0.003737311954438796,
553
+ "scss": 0.003737311954438796,
554
+ "shell": 0.003737311954438796,
555
+ "smali": 0.003737311954438796,
556
+ "smalltalk": 0.003737311954438796,
557
+ "sml": 0.003737311954438796,
558
+ "sqf": 0.003737311954438796,
559
+ "sql": 0.006312042929887236,
560
+ "stan": 0.003737311954438796,
561
+ "stata": 0.003737311954438796,
562
+ "step21": 0.003737311954438796,
563
+ "stylus": 0.003737311954438796,
564
+ "subunit": 0.003737311954438796,
565
+ "swift": 0.009866129938203402,
566
+ "taggerscript": 0.003737311954438796,
567
+ "tap": 0.003737311954438796,
568
+ "tcl": 0.0050465064595763606,
569
+ "thrift": 0.003737311954438796,
570
+ "tp": 0.003737311954438796,
571
+ "twig": 0.003737311954438796,
572
+ "typescript": 0.024795729996039233,
573
+ "vala": 0.003737311954438796,
574
+ "vbnet": 0.003737311954438796,
575
+ "vbscript": 0.003737311954438796,
576
+ "vbscript-html": 0.003737311954438796,
577
+ "verilog": 0.003737311954438796,
578
+ "vhdl": 0.003737311954438796,
579
+ "vim": 0.00583883743118971,
580
+ "wasm": 0.003737311954438796,
581
+ "wren": 0.003737311954438796,
582
+ "x86asm": 0.003737311954438796,
583
+ "xl": 0.003737311954438796,
584
+ "xml": 0.013967397673597948,
585
+ "xquery": 0.003737311954438796,
586
+ "yaml": 0.009570300755808881,
587
+ "zephir": 0.003737311954438796
588
+ },
589
+ "benchmark": {
590
+ "verification_sha256": null,
591
+ "popularity_sha256": null,
592
+ "npm_version": null,
593
+ "npm_dist_sha256": null
594
+ }
595
+ }