itstalmeez commited on
Commit
0037241
·
verified ·
1 Parent(s): ed1bc35

Upload 8 files

Browse files
INFERENCE.md ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Inference notes
2
+
3
+ Use a Transformers-compatible text-generation runtime. The model is a complete
4
+ GPT-2 checkpoint and does not require `peft`, `bitsandbytes`, an adapter, or a
5
+ separate base model.
6
+
7
+ The expected prompt format is:
8
+
9
+ ```
10
+ You are Vytre, an enterprise workforce operating intelligence model.
11
+ Input: Create marketing department
12
+ Output:
13
+ ```
14
+
15
+ Use deterministic decoding (`do_sample=False`) and validate any response as
16
+ JSON before consuming it programmatically.
README.md CHANGED
@@ -1,3 +1,53 @@
1
  ---
2
  license: apache-2.0
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: apache-2.0
3
+ library_name: transformers
4
+ pipeline_tag: text-generation
5
+ tags:
6
+ - text-generation
7
+ - gpt2
8
+ - pytorch
9
+ - enterprise-automation
10
  ---
11
+
12
+ # Vytre Core
13
+
14
+ Vytre Core is a compact, domain-specific text-generation model trained on
15
+ synthetic enterprise-workforce tasks: department creation, agent definition,
16
+ workflow planning, task decomposition, governance checks, and tool routing.
17
+
18
+ ## Model files
19
+
20
+ This repository is self-contained. It contains a standard Transformers GPT-2
21
+ checkpoint and tokenizer, not a LoRA adapter. Do **not** combine it with the
22
+ legacy `vytre-core-upload` LoRA template or an external Llama base model.
23
+
24
+ ## Load with Transformers
25
+
26
+ ```python
27
+ from transformers import AutoModelForCausalLM, AutoTokenizer
28
+
29
+ model_id = "YOUR_USERNAME/vytre-core"
30
+ tokenizer = AutoTokenizer.from_pretrained(model_id)
31
+ model = AutoModelForCausalLM.from_pretrained(model_id)
32
+
33
+ prompt = (
34
+ "You are Vytre, an enterprise workforce operating intelligence model.\\n"
35
+ "Input: Create marketing department\\n"
36
+ "Output: "
37
+ )
38
+ inputs = tokenizer(prompt, return_tensors="pt")
39
+ tokens = model.generate(
40
+ **inputs,
41
+ max_new_tokens=80,
42
+ do_sample=False,
43
+ pad_token_id=tokenizer.pad_token_id,
44
+ eos_token_id=tokenizer.eos_token_id,
45
+ )
46
+ print(tokenizer.decode(tokens[0], skip_special_tokens=True))
47
+ ```
48
+
49
+ ## Limitations
50
+
51
+ This is a small specialised model, not a general-purpose chat model. Use the
52
+ prompt format above and keep requests close to the listed operational domains.
53
+ Validate generated JSON before taking actions from it.
config.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_function": "gelu_new",
3
+ "add_cross_attention": false,
4
+ "architectures": [
5
+ "GPT2LMHeadModel"
6
+ ],
7
+ "attn_pdrop": 0.1,
8
+ "bos_token_id": 2,
9
+ "dtype": "float32",
10
+ "embd_pdrop": 0.1,
11
+ "eos_token_id": 3,
12
+ "initializer_range": 0.02,
13
+ "layer_norm_epsilon": 1e-05,
14
+ "model_type": "gpt2",
15
+ "n_ctx": 128,
16
+ "n_embd": 256,
17
+ "n_head": 4,
18
+ "n_inner": null,
19
+ "n_layer": 2,
20
+ "n_positions": 128,
21
+ "pad_token_id": 0,
22
+ "reorder_and_upcast_attn": false,
23
+ "resid_pdrop": 0.1,
24
+ "scale_attn_by_inverse_layer_idx": false,
25
+ "scale_attn_weights": true,
26
+ "summary_activation": null,
27
+ "summary_first_dropout": 0.1,
28
+ "summary_proj_to_labels": true,
29
+ "summary_type": "cls_index",
30
+ "summary_use_proj": true,
31
+ "tie_word_embeddings": true,
32
+ "transformers_version": "5.12.1",
33
+ "use_cache": false,
34
+ "vocab_size": 349
35
+ }
generation_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 2,
4
+ "eos_token_id": 3,
5
+ "output_attentions": false,
6
+ "output_hidden_states": false,
7
+ "pad_token_id": 0,
8
+ "transformers_version": "5.12.1",
9
+ "use_cache": true
10
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8925df25c979e21c07cd9d2efc9666ea3b7f74bdb38ac3bc52240d6173749e30
3
+ size 6811304
special_tokens_map.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
+ "pad_token": "<pad>",
5
+ "unk_token": "<unk>"
6
+ }
tokenizer.json ADDED
@@ -0,0 +1,478 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "1.0",
3
+ "truncation": {
4
+ "direction": "Right",
5
+ "max_length": 128,
6
+ "strategy": "LongestFirst",
7
+ "stride": 0
8
+ },
9
+ "padding": {
10
+ "strategy": "BatchLongest",
11
+ "direction": "Right",
12
+ "pad_to_multiple_of": null,
13
+ "pad_id": 0,
14
+ "pad_type_id": 0,
15
+ "pad_token": "<pad>"
16
+ },
17
+ "added_tokens": [
18
+ {
19
+ "id": 0,
20
+ "content": "<pad>",
21
+ "single_word": false,
22
+ "lstrip": false,
23
+ "rstrip": false,
24
+ "normalized": false,
25
+ "special": true
26
+ },
27
+ {
28
+ "id": 1,
29
+ "content": "<unk>",
30
+ "single_word": false,
31
+ "lstrip": false,
32
+ "rstrip": false,
33
+ "normalized": false,
34
+ "special": true
35
+ },
36
+ {
37
+ "id": 2,
38
+ "content": "<s>",
39
+ "single_word": false,
40
+ "lstrip": false,
41
+ "rstrip": false,
42
+ "normalized": false,
43
+ "special": true
44
+ },
45
+ {
46
+ "id": 3,
47
+ "content": "</s>",
48
+ "single_word": false,
49
+ "lstrip": false,
50
+ "rstrip": false,
51
+ "normalized": false,
52
+ "special": true
53
+ }
54
+ ],
55
+ "normalizer": null,
56
+ "pre_tokenizer": {
57
+ "type": "WhitespaceSplit"
58
+ },
59
+ "post_processor": {
60
+ "type": "TemplateProcessing",
61
+ "single": [
62
+ {
63
+ "Sequence": {
64
+ "id": "A",
65
+ "type_id": 0
66
+ }
67
+ },
68
+ {
69
+ "SpecialToken": {
70
+ "id": "</s>",
71
+ "type_id": 0
72
+ }
73
+ }
74
+ ],
75
+ "pair": [
76
+ {
77
+ "Sequence": {
78
+ "id": "A",
79
+ "type_id": 0
80
+ }
81
+ },
82
+ {
83
+ "SpecialToken": {
84
+ "id": "</s>",
85
+ "type_id": 0
86
+ }
87
+ },
88
+ {
89
+ "Sequence": {
90
+ "id": "B",
91
+ "type_id": 0
92
+ }
93
+ },
94
+ {
95
+ "SpecialToken": {
96
+ "id": "</s>",
97
+ "type_id": 0
98
+ }
99
+ }
100
+ ],
101
+ "special_tokens": {
102
+ "</s>": {
103
+ "id": "</s>",
104
+ "ids": [
105
+ 3
106
+ ],
107
+ "tokens": [
108
+ "</s>"
109
+ ]
110
+ },
111
+ "<s>": {
112
+ "id": "<s>",
113
+ "ids": [
114
+ 2
115
+ ],
116
+ "tokens": [
117
+ "<s>"
118
+ ]
119
+ }
120
+ }
121
+ },
122
+ "decoder": null,
123
+ "model": {
124
+ "type": "WordLevel",
125
+ "vocab": {
126
+ "<pad>": 0,
127
+ "<unk>": 1,
128
+ "<s>": 2,
129
+ "</s>": 3,
130
+ "\"30d\"}}": 4,
131
+ "\"Account": 5,
132
+ "\"Accounting": 6,
133
+ "\"Affects": 7,
134
+ "\"Analytics": 8,
135
+ "\"Analytics\":": 9,
136
+ "\"Analyze": 10,
137
+ "\"Assign": 11,
138
+ "\"BDR": 12,
139
+ "\"Backend": 13,
140
+ "\"Budget": 14,
141
+ "\"Bug": 15,
142
+ "\"Build": 16,
143
+ "\"CMO\":": 17,
144
+ "\"CS": 18,
145
+ "\"CSM": 19,
146
+ "\"CTO\":": 20,
147
+ "\"Campaign": 21,
148
+ "\"Close": 22,
149
+ "\"Closing\"],": 23,
150
+ "\"Cloud": 24,
151
+ "\"Complete": 25,
152
+ "\"Compliance": 26,
153
+ "\"Content": 27,
154
+ "\"Contract": 28,
155
+ "\"Customer": 29,
156
+ "\"Data": 30,
157
+ "\"Database": 31,
158
+ "\"Demo": 32,
159
+ "\"Deploy": 33,
160
+ "\"Design": 34,
161
+ "\"Design\",": 35,
162
+ "\"DevOps": 36,
163
+ "\"Engineering": 37,
164
+ "\"Engineering\",": 38,
165
+ "\"Escalation": 39,
166
+ "\"Escalation\"],": 40,
167
+ "\"Finance": 41,
168
+ "\"Finance\",": 42,
169
+ "\"Finance\":": 43,
170
+ "\"Frontend": 44,
171
+ "\"Graphic": 45,
172
+ "\"HR": 46,
173
+ "\"HR\",": 47,
174
+ "\"HR\":": 48,
175
+ "\"Interview\",": 49,
176
+ "\"Legal": 50,
177
+ "\"Legal\",": 51,
178
+ "\"Logistics": 52,
179
+ "\"Low-risk": 53,
180
+ "\"Marketing": 54,
181
+ "\"Marketing\",": 55,
182
+ "\"Minor": 56,
183
+ "\"Monitoring\"],": 57,
184
+ "\"Negotiate\",": 58,
185
+ "\"Offer\",": 59,
186
+ "\"Onboard\"]}": 60,
187
+ "\"Onboarding": 61,
188
+ "\"Operations": 62,
189
+ "\"Operations\",": 63,
190
+ "\"Pricing": 64,
191
+ "\"Process": 65,
192
+ "\"Product": 66,
193
+ "\"Product\",": 67,
194
+ "\"Product\":": 68,
195
+ "\"QA": 69,
196
+ "\"Recruiting": 70,
197
+ "\"Regression": 71,
198
+ "\"Renewals": 72,
199
+ "\"Reporting\"],": 73,
200
+ "\"Resolve": 74,
201
+ "\"Responsive": 75,
202
+ "\"Run": 76,
203
+ "\"SEO": 77,
204
+ "\"Sales": 78,
205
+ "\"Sales\",": 79,
206
+ "\"Schedule": 80,
207
+ "\"Screening": 81,
208
+ "\"Security": 82,
209
+ "\"Security\",": 83,
210
+ "\"Send": 84,
211
+ "\"Social": 85,
212
+ "\"Support": 86,
213
+ "\"Support\",": 87,
214
+ "\"Support\":": 88,
215
+ "\"System": 89,
216
+ "\"Tier": 90,
217
+ "\"UI": 91,
218
+ "\"UX": 92,
219
+ "\"Welcome\"}}": 93,
220
+ "\"access_db\"]}": 94,
221
+ "\"active\"}}": 95,
222
+ "\"agent.deploy\",": 96,
223
+ "\"analytics.report\",": 97,
224
+ "\"conversion\",": 98,
225
+ "\"customer@example.com\",": 99,
226
+ "\"customer_onboarding\"}}": 100,
227
+ "\"customers\",": 101,
228
+ "\"database.query\",": 102,
229
+ "\"deploy_code\"]}": 103,
230
+ "\"email.send\",": 104,
231
+ "\"filter\":": 105,
232
+ "\"manager\":": 106,
233
+ "\"parameters\":": 107,
234
+ "\"period\":": 108,
235
+ "\"permissions\":": 109,
236
+ "\"publish_content\"]}": 110,
237
+ "\"reason\":": 111,
238
+ "\"skills\":": 112,
239
+ "\"subject\":": 113,
240
+ "\"task.create\",": 114,
241
+ "\"update_deals\"]}": 115,
242
+ "\"workers\":": 116,
243
+ "\"workflow.create\",": 117,
244
+ "2": 118,
245
+ "30%": 119,
246
+ "Add": 120,
247
+ "Agent\",": 121,
248
+ "Agent\"]}": 122,
249
+ "Agent\"}}": 123,
250
+ "Analyst": 124,
251
+ "Build": 125,
252
+ "Change": 126,
253
+ "Counsel\",": 127,
254
+ "Create": 128,
255
+ "Delete": 129,
256
+ "Deploy": 130,
257
+ "Design": 131,
258
+ "FAQ": 132,
259
+ "Fire": 133,
260
+ "Get": 134,
261
+ "Improve": 135,
262
+ "Increase": 136,
263
+ "Input:": 137,
264
+ "Launch": 138,
265
+ "Manager": 139,
266
+ "Manager\",": 140,
267
+ "Manager\"]}": 141,
268
+ "Media": 142,
269
+ "Open": 143,
270
+ "Output:": 144,
271
+ "Q4": 145,
272
+ "Send": 146,
273
+ "Success\",": 147,
274
+ "Success\":": 148,
275
+ "Support": 149,
276
+ "Team\":": 150,
277
+ "UI": 151,
278
+ "Update": 152,
279
+ "Vytre,": 153,
280
+ "You": 154,
281
+ "[\"API": 155,
282
+ "[\"Account": 156,
283
+ "[\"Accounting": 157,
284
+ "[\"Add": 158,
285
+ "[\"Allocate": 159,
286
+ "[\"Analytics": 160,
287
+ "[\"Analyze": 161,
288
+ "[\"BDR": 162,
289
+ "[\"Backend": 163,
290
+ "[\"Budget": 164,
291
+ "[\"CI/CD\",": 165,
292
+ "[\"CSM": 166,
293
+ "[\"Code": 167,
294
+ "[\"Compliance": 168,
295
+ "[\"Content": 169,
296
+ "[\"Contract": 170,
297
+ "[\"Create": 171,
298
+ "[\"Define": 172,
299
+ "[\"DevOps": 173,
300
+ "[\"Escalation": 174,
301
+ "[\"Estimate": 175,
302
+ "[\"Find": 176,
303
+ "[\"Frontend": 177,
304
+ "[\"Graphic": 178,
305
+ "[\"Hire": 179,
306
+ "[\"Improve": 180,
307
+ "[\"Invoicing\",": 181,
308
+ "[\"Lead": 182,
309
+ "[\"Logistics": 183,
310
+ "[\"Onboarding": 184,
311
+ "[\"Plan": 185,
312
+ "[\"Prepare": 186,
313
+ "[\"Process": 187,
314
+ "[\"Product": 188,
315
+ "[\"QA": 189,
316
+ "[\"Qualify": 190,
317
+ "[\"React": 191,
318
+ "[\"Receive": 192,
319
+ "[\"Recruiting": 193,
320
+ "[\"Renewals": 194,
321
+ "[\"Review": 195,
322
+ "[\"SEO": 196,
323
+ "[\"Sales": 197,
324
+ "[\"Security": 198,
325
+ "[\"Social": 199,
326
+ "[\"Support": 200,
327
+ "[\"Test": 201,
328
+ "[\"Ticket": 202,
329
+ "[\"Tier": 203,
330
+ "[\"Track": 204,
331
+ "[\"UI": 205,
332
+ "[\"UX": 206,
333
+ "[\"deploy_code\",": 207,
334
+ "[\"deploy_code\"]}": 208,
335
+ "[\"manage_servers\",": 209,
336
+ "[\"read_analytics\",": 210,
337
+ "[\"read_customer_records\"]}": 211,
338
+ "[\"read_financials\"]}": 212,
339
+ "[\"read_leads\",": 213,
340
+ "[\"read_test_cases\"]}": 214,
341
+ "account\",": 215,
342
+ "active": 216,
343
+ "ads\"],": 217,
344
+ "affect": 218,
345
+ "agent": 219,
346
+ "agent\",": 220,
347
+ "an": 221,
348
+ "analysis\",": 222,
349
+ "and": 223,
350
+ "architecture\"],": 224,
351
+ "are": 225,
352
+ "backend": 226,
353
+ "budget\"]}": 227,
354
+ "button": 228,
355
+ "by": 229,
356
+ "call\",": 230,
357
+ "campaign": 231,
358
+ "campaign\"]}": 232,
359
+ "change\"}": 233,
360
+ "changes": 234,
361
+ "churn\"],": 235,
362
+ "color": 236,
363
+ "compliance": 237,
364
+ "content": 238,
365
+ "content\",": 239,
366
+ "conversion": 240,
367
+ "creation\",": 241,
368
+ "customer": 242,
369
+ "customers": 243,
370
+ "customers\"}": 244,
371
+ "cycle": 245,
372
+ "data": 246,
373
+ "deal\"]}": 247,
374
+ "demo\",": 248,
375
+ "department": 249,
376
+ "design": 250,
377
+ "design\",": 251,
378
+ "design\"],": 252,
379
+ "development\",": 253,
380
+ "devops": 254,
381
+ "email": 255,
382
+ "email\",": 256,
383
+ "employee": 257,
384
+ "engineering": 258,
385
+ "enterprise": 259,
386
+ "experience\"}": 260,
387
+ "false,": 261,
388
+ "features\"]}": 262,
389
+ "finance": 263,
390
+ "for": 264,
391
+ "frontend": 265,
392
+ "hiring": 266,
393
+ "hours": 267,
394
+ "hr": 268,
395
+ "implementation\",": 269,
396
+ "implications\"}": 270,
397
+ "infrastructure\",": 271,
398
+ "infrastructure\"],": 272,
399
+ "intelligence": 273,
400
+ "issue\",": 274,
401
+ "launch": 275,
402
+ "lead\",": 276,
403
+ "legal": 277,
404
+ "management\"],": 278,
405
+ "marketing": 279,
406
+ "meeting\"]}": 280,
407
+ "model.": 281,
408
+ "new": 282,
409
+ "objectives\"],": 283,
410
+ "office": 284,
411
+ "onboarding": 285,
412
+ "operating": 286,
413
+ "operations": 287,
414
+ "optimization\",": 288,
415
+ "package\",": 289,
416
+ "page": 290,
417
+ "performance\"]}": 291,
418
+ "planning\",": 292,
419
+ "pricing": 293,
420
+ "privacy": 294,
421
+ "process": 295,
422
+ "product": 296,
423
+ "production\"]}": 297,
424
+ "proposal\",": 298,
425
+ "qa": 299,
426
+ "qualification\",": 300,
427
+ "quarterly": 301,
428
+ "release": 302,
429
+ "report": 303,
430
+ "report\"}}": 304,
431
+ "reporting\",": 305,
432
+ "requested": 306,
433
+ "response": 307,
434
+ "response\",": 308,
435
+ "resume\",": 309,
436
+ "retention": 310,
437
+ "review\",": 311,
438
+ "risk\"}": 312,
439
+ "sales": 313,
440
+ "scheduling\",": 314,
441
+ "security": 315,
442
+ "software": 316,
443
+ "space\"],": 317,
444
+ "staff\"],": 318,
445
+ "staging\",": 319,
446
+ "strategy\"],": 320,
447
+ "success": 321,
448
+ "support": 322,
449
+ "task": 323,
450
+ "testing\"],": 324,
451
+ "tests\",": 325,
452
+ "ticket\",": 326,
453
+ "ticket\"]}": 327,
454
+ "time\"],": 328,
455
+ "to": 329,
456
+ "tracking\",": 330,
457
+ "true,": 331,
458
+ "visual": 332,
459
+ "welcome": 333,
460
+ "workflow": 334,
461
+ "workforce": 335,
462
+ "{\"CEO\":": 336,
463
+ "{\"CMO\":": 337,
464
+ "{\"Customer": 338,
465
+ "{\"Operations\":": 339,
466
+ "{\"department\":": 340,
467
+ "{\"metric\":": 341,
468
+ "{\"name\":": 342,
469
+ "{\"requires_approval\":": 343,
470
+ "{\"table\":": 344,
471
+ "{\"title\":": 345,
472
+ "{\"to\":": 346,
473
+ "{\"tool\":": 347,
474
+ "{\"workflow\":": 348
475
+ },
476
+ "unk_token": "<unk>"
477
+ }
478
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "backend": "tokenizers",
3
+ "bos_token": "<s>",
4
+ "eos_token": "</s>",
5
+ "model_max_length": 1000000000000000019884624838656,
6
+ "pad_token": "<pad>",
7
+ "tokenizer_class": "TokenizersBackend",
8
+ "unk_token": "<unk>"
9
+ }