TensorGreed commited on
Commit
8879e60
·
verified ·
1 Parent(s): d1ddca2

BrewSLM export 37: sensitive-data span tagger, run #143 (deberta-v3-base, 30 types): manifest.json

Browse files
Files changed (1) hide show
  1. brewslm/manifest.json +368 -0
brewslm/manifest.json ADDED
@@ -0,0 +1,368 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "token_window": null,
3
+ "span_package": {
4
+ "model_card": "README.md",
5
+ "thresholds": "thresholds.json",
6
+ "entity_types": [
7
+ "address",
8
+ "api_key",
9
+ "bank_account",
10
+ "card_expiry",
11
+ "credit_card",
12
+ "customer_id",
13
+ "cvv",
14
+ "date_of_birth",
15
+ "device_id",
16
+ "diagnosis",
17
+ "email",
18
+ "employee_id",
19
+ "health_plan_id",
20
+ "iban",
21
+ "ip_address",
22
+ "licence_number",
23
+ "license_plate",
24
+ "medical_record_number",
25
+ "medication",
26
+ "national_id",
27
+ "passport",
28
+ "person_name",
29
+ "phone",
30
+ "provider_name",
31
+ "routing_number",
32
+ "tax_id",
33
+ "ticket_id",
34
+ "transaction_id",
35
+ "username",
36
+ "visit_date"
37
+ ],
38
+ "evaluations": [
39
+ "gold_test",
40
+ "test"
41
+ ],
42
+ "benchmarks": [
43
+ "CoNLL-2003 people (news, human-labelled)",
44
+ "Gretel PII (synthetic, independent)",
45
+ "CoNLL-2003 people, large (validation, 2000 rows)"
46
+ ],
47
+ "eval_result_ids": [
48
+ 169,
49
+ 168
50
+ ],
51
+ "window": 512,
52
+ "pipeline_stride": 128
53
+ },
54
+ "seed_selection": {
55
+ "metric_id": "span_f1_tuned",
56
+ "metric_label": "span F1 (tuned thresholds)",
57
+ "higher_is_better": true,
58
+ "task_type": "token_classification",
59
+ "rule": "highest validation span F1 (tuned thresholds); ties go to the lower run id; test examples, the answer key and external benchmarks were not used to pick",
60
+ "picked_experiment_id": 143,
61
+ "picked_seed_value": 45,
62
+ "n_seeds": 3,
63
+ "n_scored": 3,
64
+ "children": [
65
+ {
66
+ "experiment_id": 143,
67
+ "seed_value": 45,
68
+ "value": 0.9467,
69
+ "picked": true,
70
+ "skipped_reason": null
71
+ },
72
+ {
73
+ "experiment_id": 144,
74
+ "seed_value": 46,
75
+ "value": 0.9452,
76
+ "picked": false,
77
+ "skipped_reason": null
78
+ },
79
+ {
80
+ "experiment_id": 145,
81
+ "seed_value": 47,
82
+ "value": 0.9437,
83
+ "picked": false,
84
+ "skipped_reason": null
85
+ }
86
+ ],
87
+ "leader_experiment_id": 142,
88
+ "picked_at": "2026-10-17T15:14:56.348856+00:00",
89
+ "export_id": 37,
90
+ "export_format": "huggingface",
91
+ "text": "Picked seed 45 (run #143) of 3 by validation span F1 (tuned thresholds) (higher is better; values: seed 45 0.9467, seed 46 0.9452, seed 47 0.9437); test examples, the answer key and external benchmarks were not used to pick."
92
+ },
93
+ "run_id": "20261017T151456Z",
94
+ "export_id": 37,
95
+ "project_id": 31,
96
+ "experiment_id": 143,
97
+ "format": "huggingface",
98
+ "quantization": null,
99
+ "created_at": "2026-10-17T15:14:56.353740+00:00",
100
+ "platform_version": "0.1.0",
101
+ "run_dir": "/home/anuragj/Desktop/GitHub/__SLM__/data/exports/31/37/run-20261017T151456Z",
102
+ "experiment": {
103
+ "name": "deberta-v3-base token classification (Sonnet labels, no-title providers, more news) \u00b7 3 seeds (seed=43) \u00b7 3 seeds (seed=45) \u00b7 3 seeds (seed=45) \u00b7 3 seeds (seed=45) \u00b7 3 seeds (seed=45)",
104
+ "status": "completed",
105
+ "training_mode": "sft",
106
+ "base_model": "microsoft/deberta-v3-base",
107
+ "final_train_loss": 0.01762743145227432,
108
+ "final_eval_loss": 0.09431453794240952
109
+ },
110
+ "model_artifacts": {
111
+ "source": "experiment_model_dir",
112
+ "count": 7,
113
+ "size_bytes": 743903066,
114
+ "path": "model",
115
+ "experiment_id": 143,
116
+ "merged_from_adapter": null
117
+ },
118
+ "artifacts": [
119
+ {
120
+ "path": "model/model.safetensors",
121
+ "size_bytes": 735538204,
122
+ "sha256": "fdefab12891f005e8d0ac933ff871764fcd074d74bd1231baabb3866b5bad7a1"
123
+ },
124
+ {
125
+ "path": "model/config.json",
126
+ "size_bytes": 4005,
127
+ "sha256": "905d4aa2c31c90eb0a613fc7d3acf00a0ee227d63ec304f552e7ea6105f442db"
128
+ },
129
+ {
130
+ "path": "model/tokenizer.json",
131
+ "size_bytes": 8339193,
132
+ "sha256": "fd35ef00ccc8c8242f95106634edc930b1f35220e12bffc5a85e5ae66edd31c1"
133
+ },
134
+ {
135
+ "path": "model/thresholds.json",
136
+ "size_bytes": 11655,
137
+ "sha256": "46ab3b2ac2d4dfd19e8b210708ff3a0c6accd2e7b7b075cb5ad3b02e045d3514"
138
+ },
139
+ {
140
+ "path": "model/tokenizer_config.json",
141
+ "size_bytes": 481,
142
+ "sha256": "71f70d0e7cd97bbf9b5a90b2bb92bf1acb26d34e4758cfe0fac371809b23d1de"
143
+ },
144
+ {
145
+ "path": "model/label_space.json",
146
+ "size_bytes": 1805,
147
+ "sha256": "d93679b76b804b45a8e8b918a4895e58f95bf45dc77623754c75731c5ee7c6c2"
148
+ },
149
+ {
150
+ "path": "model/README.md",
151
+ "size_bytes": 7723,
152
+ "sha256": "9971e96ab583a8fdb51675614b25867950bb964351acbc10e204b5e0875a1229"
153
+ },
154
+ {
155
+ "path": "Dockerfile",
156
+ "size_bytes": 145,
157
+ "sha256": "ce9232b4621fbaa74670e27e7664a45cae1048aa76138b28c9107e86230f16e1"
158
+ },
159
+ {
160
+ "path": "serve.py",
161
+ "size_bytes": 2414,
162
+ "sha256": "779bfcdcaae56f87323f3ce92bef9d9efbab52660228afb9589141577a28fbaf"
163
+ },
164
+ {
165
+ "path": "RELEASE_NOTES.md",
166
+ "size_bytes": 634,
167
+ "sha256": "62b57ee537e9e9b13d9eeb12dba30989911147728abf2c662d2fe9062f48d7a2"
168
+ }
169
+ ],
170
+ "deployment": {
171
+ "contract_version": "slm.deployment-target/v1",
172
+ "captured_at": "2026-10-17T15:14:56.943510+00:00",
173
+ "run_dir": "/home/anuragj/Desktop/GitHub/__SLM__/data/exports/31/37/run-20261017T151456Z",
174
+ "export_format": "huggingface",
175
+ "artifact_profile": "huggingface",
176
+ "selected_target_ids": [
177
+ "exporter.huggingface",
178
+ "runner.vllm",
179
+ "runner.tgi"
180
+ ],
181
+ "unknown_target_ids": [],
182
+ "artifact_validation": {
183
+ "artifact_profile": "huggingface",
184
+ "passed": true,
185
+ "errors": [],
186
+ "warnings": [],
187
+ "checks": [
188
+ {
189
+ "check_id": "model_dir_exists",
190
+ "passed": true,
191
+ "status": "pass",
192
+ "severity": "error",
193
+ "message": "model directory must exist at '/home/anuragj/Desktop/GitHub/__SLM__/data/exports/31/37/run-20261017T151456Z/model'.",
194
+ "details": {}
195
+ },
196
+ {
197
+ "check_id": "model_files_present",
198
+ "passed": true,
199
+ "status": "pass",
200
+ "severity": "error",
201
+ "message": "model directory must contain at least one file.",
202
+ "details": {
203
+ "file_count": 7
204
+ }
205
+ },
206
+ {
207
+ "check_id": "model_bytes_positive",
208
+ "passed": true,
209
+ "status": "pass",
210
+ "severity": "error",
211
+ "message": "model artifact bytes must be > 0.",
212
+ "details": {
213
+ "model_bytes": 743903066
214
+ }
215
+ },
216
+ {
217
+ "check_id": "hf_config_present",
218
+ "passed": true,
219
+ "status": "pass",
220
+ "severity": "error",
221
+ "message": "HuggingFace export requires model/config.json.",
222
+ "details": {}
223
+ },
224
+ {
225
+ "check_id": "hf_weights_present",
226
+ "passed": true,
227
+ "status": "pass",
228
+ "severity": "error",
229
+ "message": "HuggingFace export requires at least one weights file (.safetensors/.bin/.pt).",
230
+ "details": {
231
+ "weights_count": 1
232
+ }
233
+ },
234
+ {
235
+ "check_id": "hf_tokenizer_present",
236
+ "passed": true,
237
+ "status": "pass",
238
+ "severity": "warning",
239
+ "message": "Tokenizer metadata is recommended for serving compatibility.",
240
+ "details": {
241
+ "tokenizer_like_files": 5
242
+ }
243
+ }
244
+ ],
245
+ "file_count": 7,
246
+ "model_bytes": 743903066,
247
+ "captured_at": "2026-10-17T15:14:56.931966+00:00"
248
+ },
249
+ "target_reports": [
250
+ {
251
+ "target_id": "exporter.huggingface",
252
+ "kind": "exporter",
253
+ "compatible": true,
254
+ "passed": true,
255
+ "local_ready": true,
256
+ "smoke_supported": false,
257
+ "smoke_executed": false,
258
+ "smoke_passed": null,
259
+ "checks": [
260
+ {
261
+ "check_id": "artifact_profile_compatible",
262
+ "passed": true,
263
+ "status": "pass",
264
+ "severity": "error",
265
+ "message": "Target 'exporter.huggingface' is compatible with artifact profile 'huggingface'.",
266
+ "details": {}
267
+ },
268
+ {
269
+ "check_id": "artifact_validation_passed",
270
+ "passed": true,
271
+ "status": "pass",
272
+ "severity": "error",
273
+ "message": "Artifact validation passed for exporter target.",
274
+ "details": {}
275
+ }
276
+ ],
277
+ "errors": [],
278
+ "warnings": [],
279
+ "launch_example": "python serve.py --model ./model"
280
+ },
281
+ {
282
+ "target_id": "runner.vllm",
283
+ "kind": "runner",
284
+ "compatible": true,
285
+ "passed": false,
286
+ "local_ready": false,
287
+ "smoke_supported": true,
288
+ "smoke_executed": false,
289
+ "smoke_passed": null,
290
+ "checks": [
291
+ {
292
+ "check_id": "artifact_profile_compatible",
293
+ "passed": true,
294
+ "status": "pass",
295
+ "severity": "error",
296
+ "message": "Target 'runner.vllm' is compatible with artifact profile 'huggingface'.",
297
+ "details": {}
298
+ },
299
+ {
300
+ "check_id": "vllm_module_installed",
301
+ "passed": false,
302
+ "status": "fail",
303
+ "severity": "error",
304
+ "message": "Python module 'vllm' must be importable for local vLLM smoke tests.",
305
+ "details": {}
306
+ }
307
+ ],
308
+ "errors": [
309
+ "Python module 'vllm' must be importable for local vLLM smoke tests."
310
+ ],
311
+ "warnings": [],
312
+ "launch_example": "python -m vllm.entrypoints.openai.api_server --model ./model"
313
+ },
314
+ {
315
+ "target_id": "runner.tgi",
316
+ "kind": "runner",
317
+ "compatible": true,
318
+ "passed": true,
319
+ "local_ready": true,
320
+ "smoke_supported": true,
321
+ "smoke_executed": true,
322
+ "smoke_passed": true,
323
+ "checks": [
324
+ {
325
+ "check_id": "artifact_profile_compatible",
326
+ "passed": true,
327
+ "status": "pass",
328
+ "severity": "error",
329
+ "message": "Target 'runner.tgi' is compatible with artifact profile 'huggingface'.",
330
+ "details": {}
331
+ },
332
+ {
333
+ "check_id": "tgi_runtime_present",
334
+ "passed": true,
335
+ "status": "pass",
336
+ "severity": "error",
337
+ "message": "TGI runtime requires `text-generation-launcher` or `docker` binary.",
338
+ "details": {
339
+ "launcher": false,
340
+ "docker": true
341
+ }
342
+ },
343
+ {
344
+ "check_id": "tgi_smoke_runtime",
345
+ "passed": true,
346
+ "status": "pass",
347
+ "severity": "error",
348
+ "message": "TGI runtime smoke test succeeded.",
349
+ "details": {
350
+ "output": "Docker version 29.2.1, build a5c7197"
351
+ }
352
+ }
353
+ ],
354
+ "errors": [],
355
+ "warnings": [],
356
+ "launch_example": "text-generation-launcher --model-id ./model"
357
+ }
358
+ ],
359
+ "summary": {
360
+ "deployable_artifact": true,
361
+ "selected_target_count": 3,
362
+ "compatible_runner_count": 2,
363
+ "runner_local_ready_count": 1,
364
+ "runner_smoke_success_count": 1,
365
+ "local_smoke_passed": true
366
+ }
367
+ }
368
+ }