PowerMachine commited on
Commit
84cfacd
·
verified ·
1 Parent(s): c89c7aa

V6.5-attn-v2 cleanup: remove legacy v6_5_7ds_attn_report.json

Browse files
Files changed (1) hide show
  1. v6_5_7ds_attn_report.json +0 -412
v6_5_7ds_attn_report.json DELETED
@@ -1,412 +0,0 @@
1
- {
2
- "version": "V6.5-attn-100batch-5000meta-train-only",
3
- "timestamp": "2026-08-08T02:07:14.927181",
4
- "user_requirements_checklist": {
5
- "HF_TOKEN_deleted_after_use": true,
6
- "streaming_datasets_active_REAL": true,
7
- "xeon_runtime_active": true,
8
- "V65_ENABLE_STREAMING_forced": true,
9
- "attention_active_and_logic_functional": true,
10
- "streaming_100_per_batch": true,
11
- "meta_minima_5000_atingida": true,
12
- "no_synthetic_data": true,
13
- "streaming_with_pauses": true,
14
- "storage_critical_check": true,
15
- "aggressive_ram_cleanup": true,
16
- "metrics_reasoning_response_verified": true,
17
- "exhausted_7_datasets_in_sequence": true,
18
- "som_grid_864_neurons": true,
19
- "mtp_head_size_increased_K6": true,
20
- "vqvae2_active_in_pipeline": true,
21
- "reasoning_engine_integrated_to_kls": true,
22
- "smoothquant_w8a8_integrated_to_kls": true,
23
- "tool_coordinator_workers_reactivated": true,
24
- "logic_and_bugfixes_verified": true
25
- },
26
- "config": {
27
- "BATCH_SIZE": 16,
28
- "STREAM_BATCH_SIZE": 100,
29
- "MAX_SAMPLES_PER_DATASET": 720,
30
- "TOTAL_SAMPLES_TARGET": 5040,
31
- "META_MINIMA_SAMPLES": 5000,
32
- "SOM_GRID": [
33
- 6,
34
- 6,
35
- 6,
36
- 4
37
- ],
38
- "n_neurons": 864,
39
- "HIDDEN_DIM": 1024,
40
- "VOCAB_SIZE": 16384,
41
- "MTP_K": 6,
42
- "ATTENTION_N_HEADS": 8
43
- },
44
- "xeon_status": {
45
- "version": "V6",
46
- "physical_cores": 2,
47
- "env": {
48
- "MKL_ENABLE_INSTRUCTIONS": "AVX512",
49
- "MKL_NUM_THREADS": "2",
50
- "OMP_NUM_THREADS": "2",
51
- "MKL_DYNAMIC": "FALSE",
52
- "DNNL_PRIMITIVE_CACHE_CAPACITY": "1024",
53
- "ONEDNN_MAX_CPU_ISA": "AMX_INT8",
54
- "KMP_AFFINITY": "granularity=fine,compact,1,0",
55
- "KMP_BLOCKTIME": "1"
56
- },
57
- "avx512": {
58
- "supported": true,
59
- "desc": "AVX512_VNNI (full INT8 acceleration)"
60
- },
61
- "amx": {
62
- "supported": true,
63
- "desc": "AMX (tile + int8 + bf16) — full AMX acceleration"
64
- },
65
- "ipex_available": false,
66
- "init_done": true
67
- },
68
- "fp16_benchmark": {
69
- "best_time_ms": 96.11023900288274,
70
- "avg_time_ms": 100.32488200158696,
71
- "best_tflops": 1.331803992248534,
72
- "avg_tflops": 1.27585497681398,
73
- "matrix_size": 4000.0
74
- },
75
- "training": {
76
- "duration_s": 592.9951963424683,
77
- "n_steps": 357,
78
- "samples_per_dataset_actual": {
79
- "dominguesm/restore-punctuation-ptbr-dataset": 720,
80
- "carolina-c4ai/corpus-carolina": 720,
81
- "CEIA-POSITIVO/ultrachat_br_clustred_balanced_v1": 720,
82
- "dominguesm/Canarim-Instruct-PTBR-Dataset": 720,
83
- "adalbertojunior/punctuation-ptbr": 720,
84
- "iara-project/news-articles-ptbr-dataset": 720,
85
- "manoela/noticias_ptbr": 720
86
- },
87
- "total_samples_processed": 5040,
88
- "meta_minima_atingida": true,
89
- "streaming_failures": {
90
- "dominguesm/restore-punctuation-ptbr-dataset": 0,
91
- "carolina-c4ai/corpus-carolina": 0,
92
- "CEIA-POSITIVO/ultrachat_br_clustred_balanced_v1": 0,
93
- "dominguesm/Canarim-Instruct-PTBR-Dataset": 0,
94
- "adalbertojunior/punctuation-ptbr": 0,
95
- "iara-project/news-articles-ptbr-dataset": 0,
96
- "manoela/noticias_ptbr": 0
97
- },
98
- "storage_critical_stopped": false
99
- },
100
- "kohonen_final": {
101
- "t": 10220,
102
- "sigma_t": 5.465144559085541e-05,
103
- "alpha_t": 0.0006036082923599565,
104
- "sigma0": 1.5,
105
- "alpha0": 0.1,
106
- "lambda_ewc": 0.02,
107
- "grid_shape": [
108
- 6,
109
- 6,
110
- 6,
111
- 4
112
- ],
113
- "n_neurons": 864,
114
- "has_ewc_reference": true,
115
- "fisher_w_mean": 0.0,
116
- "fisher_w_max": 0.0,
117
- "fisher_accum_count": 0,
118
- "weights_norm": 1664.1396484375,
119
- "weights_w_mean": 0.02392551675438881
120
- },
121
- "kls_state": {
122
- "som": {
123
- "t": 10220,
124
- "sigma_t": 5.465144559085541e-05,
125
- "alpha_t": 0.0006036082923599565,
126
- "sigma0": 1.5,
127
- "alpha0": 0.1,
128
- "lambda_ewc": 0.02,
129
- "grid_shape": [
130
- 6,
131
- 6,
132
- 6,
133
- 4
134
- ],
135
- "n_neurons": 864,
136
- "has_ewc_reference": true,
137
- "fisher_w_mean": 0.0,
138
- "fisher_w_max": 0.0,
139
- "fisher_accum_count": 0,
140
- "weights_norm": 1664.1396484375,
141
- "weights_w_mean": 0.02392551675438881
142
- },
143
- "kls": {
144
- "time_counter": 5176,
145
- "T_max": 10000,
146
- "buffer_size": 0,
147
- "training_ready": false,
148
- "punishment_count": 0,
149
- "success_count": 0,
150
- "classifier_trained": true,
151
- "histogram_size": 0,
152
- "histogram_max": 0,
153
- "N_start": 10,
154
- "dim_choice": "y",
155
- "som_neuron_count": 864,
156
- "required_new_samples": 10,
157
- "has_classifier": true,
158
- "vocab_size": 16384,
159
- "hidden_dim": 1024,
160
- "seq_len": 8,
161
- "enable_vqvae2": true,
162
- "enable_reasoning": true,
163
- "enable_w8a8": true,
164
- "vqvae2_n_calls": 100,
165
- "w8a8_n_calls": 100,
166
- "tool_coordinator_workers_active": true,
167
- "enable_attention": true,
168
- "attention_active": true,
169
- "attention_n_heads": 8
170
- },
171
- "vqvae2": {
172
- "active": true,
173
- "n_calls": 100,
174
- "latest": {
175
- "vq_loss": 1.3525170743378112e-06,
176
- "recon_loss": 0.21045470237731934,
177
- "total_loss": 0.21045605838298798,
178
- "n_used_top": 64,
179
- "n_used_bot": 128,
180
- "usage_ratio_top": 1.0,
181
- "usage_ratio_bot": 1.0,
182
- "codebook_ppl_top": 17.411011265609076,
183
- "codebook_ppl_bot": 18.660659830381586,
184
- "n_restarted_top": 0,
185
- "n_restarted_bot": 0,
186
- "goose_temp": 0.5,
187
- "active": true
188
- },
189
- "mean_total_loss": 0.14916955351829528,
190
- "mean_recon_loss": 0.14915248230099679,
191
- "n_nan_skipped": 0
192
- },
193
- "reasoning": {
194
- "active": true,
195
- "stats": {
196
- "n_steps": 8,
197
- "n_history": 7,
198
- "tools": [
199
- "som_query",
200
- "buffer_stats"
201
- ],
202
- "tool_stats": {
203
- "som_query": {
204
- "n_calls": 129,
205
- "n_success": 129,
206
- "n_failures": 0,
207
- "total_time_s": 0.9299983978271484,
208
- "avg_time_s": 0.007209289905636809,
209
- "cache_size": 0,
210
- "instructions_count": 0,
211
- "error_count": 0,
212
- "success_rate": 1.0,
213
- "circuit_breaker": {
214
- "state": "closed",
215
- "failure_count": 0,
216
- "success_count": 0,
217
- "failure_threshold": 5,
218
- "recovery_timeout_s": 30.0
219
- }
220
- },
221
- "buffer_stats": {
222
- "n_calls": 122,
223
- "n_success": 122,
224
- "n_failures": 0,
225
- "total_time_s": 0.2865762710571289,
226
- "avg_time_s": 0.002348985828337122,
227
- "cache_size": 0,
228
- "instructions_count": 0,
229
- "error_count": 0,
230
- "success_rate": 1.0,
231
- "circuit_breaker": {
232
- "state": "closed",
233
- "failure_count": 0,
234
- "success_count": 0,
235
- "failure_threshold": 5,
236
- "recovery_timeout_s": 30.0
237
- }
238
- }
239
- },
240
- "phases_used": [
241
- "answering",
242
- "monitoring",
243
- "decomposing",
244
- "thinking",
245
- "adjusting",
246
- "planning",
247
- "executing",
248
- "predicting"
249
- ]
250
- },
251
- "n_history": 7
252
- },
253
- "attention": {
254
- "active": true,
255
- "n_calls": 5177,
256
- "n_errors": 0,
257
- "last_norm_in": 110.55284881591797,
258
- "last_norm_out": 115.7743911743164,
259
- "last_attn_activated": true,
260
- "last_attn_diff_norm": 113.8590316772461,
261
- "n_heads": 8,
262
- "logic_functional": true
263
- },
264
- "w8a8": {
265
- "active": true,
266
- "n_calls": 100,
267
- "latest": {
268
- "active": true,
269
- "n_neurons": 864,
270
- "n_features": 4,
271
- "n_calibration_samples": 20,
272
- "smooth_scale_mean": 0.10064595937728882,
273
- "smooth_scale_std": 0.20062804222106934,
274
- "smooth_scale_min": 0.0003319382085464895,
275
- "smooth_scale_max": 0.4015880227088928,
276
- "alpha": 0.5,
277
- "n_bits": 8,
278
- "compression_ratio": 4.0,
279
- "weights_norm_original": 1652.71875,
280
- "weights_norm_dequant": 1664.1396484375,
281
- "absolute_error": 11.421292304992676,
282
- "relative_error": 0.006910608538199664,
283
- "preserves_topology": true,
284
- "w8a8_int8_range": [
285
- -123,
286
- 127
287
- ],
288
- "elapsed_ms": 0.6363391876220703
289
- },
290
- "mean_relative_error": 0.020276215304510577,
291
- "max_relative_error": 0.17164724821171012,
292
- "all_preserve_topology": false
293
- }
294
- },
295
- "attention_eval_summary": {
296
- "active": true,
297
- "n_calls": 5163,
298
- "n_errors": 0,
299
- "last_attn_activated": true,
300
- "logic_functional": true,
301
- "attention_modifies_output": true,
302
- "verdict": "ACTIVE_AND_FUNCTIONAL"
303
- },
304
- "reasoning_eval_summary": {
305
- "n_with_answer": 7,
306
- "n_with_think": 7,
307
- "answer_rate": 1.0,
308
- "think_rate": 1.0,
309
- "avg_reasoning_length": 996.5714285714286,
310
- "reasoning_engine_active": true,
311
- "reasoning_engine_n_history": 7
312
- },
313
- "w8a8_compression_summary": {
314
- "active": true,
315
- "n_calls": 100,
316
- "latest": {
317
- "active": true,
318
- "n_neurons": 864,
319
- "n_features": 4,
320
- "n_calibration_samples": 20,
321
- "smooth_scale_mean": 0.10064595937728882,
322
- "smooth_scale_std": 0.20062804222106934,
323
- "smooth_scale_min": 0.0003319382085464895,
324
- "smooth_scale_max": 0.4015880227088928,
325
- "alpha": 0.5,
326
- "n_bits": 8,
327
- "compression_ratio": 4.0,
328
- "weights_norm_original": 1652.71875,
329
- "weights_norm_dequant": 1664.1396484375,
330
- "absolute_error": 11.421292304992676,
331
- "relative_error": 0.006910608538199664,
332
- "preserves_topology": true,
333
- "w8a8_int8_range": [
334
- -123,
335
- 127
336
- ],
337
- "elapsed_ms": 0.6363391876220703
338
- },
339
- "mean_relative_error": 0.020276215304510577,
340
- "max_relative_error": 0.17164724821171012,
341
- "all_preserve_topology": false
342
- },
343
- "attention_metrics_final": {
344
- "active": true,
345
- "n_calls": 5177,
346
- "n_errors": 0,
347
- "last_norm_in": 110.55284881591797,
348
- "last_norm_out": 115.7743911743164,
349
- "last_attn_activated": true,
350
- "last_attn_diff_norm": 113.8590316772461,
351
- "n_heads": 8,
352
- "logic_functional": true
353
- },
354
- "datasets_used": [
355
- "dominguesm/restore-punctuation-ptbr-dataset",
356
- "carolina-c4ai/corpus-carolina",
357
- "CEIA-POSITIVO/ultrachat_br_clustred_balanced_v1",
358
- "dominguesm/Canarim-Instruct-PTBR-Dataset",
359
- "adalbertojunior/punctuation-ptbr",
360
- "iara-project/news-articles-ptbr-dataset",
361
- "manoela/noticias_ptbr"
362
- ],
363
- "verification": {
364
- "checks": {
365
- "kls_instantiation": {
366
- "status": "PASS",
367
- "details": "vqvae2=True, reasoning=True, w8a8=True, attention=True, workers=True"
368
- },
369
- "find_bmu_no_premature_return": {
370
- "status": "PASS",
371
- "details": "bmu=(0, 0, 1, 1)"
372
- },
373
- "activate_hypothesis_detach": {
374
- "status": "PASS",
375
- "details": "classifier_trained=True"
376
- },
377
- "pgvector_lookup_removed": {
378
- "status": "PASS"
379
- },
380
- "kohonen_refactored_removed": {
381
- "status": "PASS",
382
- "details": "old_folder=False, new_file=True"
383
- },
384
- "trainer_removed": {
385
- "status": "PASS"
386
- },
387
- "vqvae2_valid_output": {
388
- "status": "PASS",
389
- "details": "n_calls=1, total_loss=0.1358"
390
- },
391
- "reasoning_engine_tags": {
392
- "status": "PASS",
393
- "details": "reasoning length=934 chars"
394
- },
395
- "smoothquant_w8a8_som_weights": {
396
- "status": "PASS",
397
- "details": "relative_error=0.005070, compression_ratio=4.0x"
398
- },
399
- "tool_coordinator_workers_reactivated": {
400
- "status": "PASS",
401
- "details": "n_workers=4, n_tools=2, elapsed_ms=3.08"
402
- },
403
- "attention_integrated_and_functional": {
404
- "status": "PASS",
405
- "details": "n_calls=15, n_errors=0, logic_functional=True, n_heads=8"
406
- }
407
- },
408
- "n_pass": 11,
409
- "n_fail": 0,
410
- "all_pass": true
411
- }
412
- }