V6.5-7ds cleanup: remove legacy v5_report.json
Browse files- v5_report.json +0 -127
v5_report.json
DELETED
|
@@ -1,127 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"version": "V5",
|
| 3 |
-
"timestamp": 1786018677.168036,
|
| 4 |
-
"config": {
|
| 5 |
-
"n_multimodal": 10,
|
| 6 |
-
"n_text": 10,
|
| 7 |
-
"epochs": 2,
|
| 8 |
-
"batch_size": 4,
|
| 9 |
-
"max_seq_len": 16,
|
| 10 |
-
"num_layers_hyp": 8,
|
| 11 |
-
"triplet_weight": 0.05,
|
| 12 |
-
"meta_interval": 3,
|
| 13 |
-
"cleanup_interval": 1,
|
| 14 |
-
"triplet_ema_alpha": 0.1
|
| 15 |
-
},
|
| 16 |
-
"v5_improvements": {
|
| 17 |
-
"w8a8_learnable_alpha": true,
|
| 18 |
-
"w8a8_self_contained_trust": true,
|
| 19 |
-
"apply_w8a8_bugfix": true,
|
| 20 |
-
"num_layers_hyp_fixed_8": true,
|
| 21 |
-
"oom_mitigations": [
|
| 22 |
-
"del intermediários explícito",
|
| 23 |
-
"MetaConfigurator a cada 3 steps",
|
| 24 |
-
"gc.collect(0) a cada step",
|
| 25 |
-
"aggressive_cleanup entre epochs",
|
| 26 |
-
"torch.no_grad() em val/eval",
|
| 27 |
-
"alerts limitado a 50 entries"
|
| 28 |
-
],
|
| 29 |
-
"learnable_params": [
|
| 30 |
-
"T (MetaConfigurator)",
|
| 31 |
-
"τ (MetaConfigurator)",
|
| 32 |
-
"λ_ent (MetaConfigurator + treino)",
|
| 33 |
-
"init_gate (residual_gate, MetaConfigurator + treino)",
|
| 34 |
-
"α smoothquant (STE, em SmoothQuantW8A8)",
|
| 35 |
-
"module_logits (treino)",
|
| 36 |
-
"todos os pesos W (AdamW)"
|
| 37 |
-
]
|
| 38 |
-
},
|
| 39 |
-
"xeon_env": {
|
| 40 |
-
"OMP_NUM_THREADS": "2",
|
| 41 |
-
"MKL_NUM_THREADS": "2",
|
| 42 |
-
"ONEDNN_MAX_CPU_ISA": "AMX_INT8",
|
| 43 |
-
"IPEX": true,
|
| 44 |
-
"CPU_ISA": "AMX",
|
| 45 |
-
"torch_threads": 2
|
| 46 |
-
},
|
| 47 |
-
"result": {
|
| 48 |
-
"n_steps": 6,
|
| 49 |
-
"n_epochs": 2,
|
| 50 |
-
"final_train_loss": 9.665669441223145,
|
| 51 |
-
"final_val_loss": 9.436580848693847,
|
| 52 |
-
"final_perplexity": 12538.771572889684,
|
| 53 |
-
"final_overfitting_gap": 0.006072171529135417,
|
| 54 |
-
"final_eval_benchmark": 0.375,
|
| 55 |
-
"avg_tokens_per_sec": 29.505289440522088,
|
| 56 |
-
"avg_tflops_per_cpu": 0.00010685671030126807,
|
| 57 |
-
"avg_time_per_step_ms": 2553.5807878330465,
|
| 58 |
-
"avg_grad_norm": 10.784179131189982,
|
| 59 |
-
"avg_cpu_pct": 121.38333333333333,
|
| 60 |
-
"final_rss_mb": 922.49609375,
|
| 61 |
-
"max_rss_mb": 922.49609375,
|
| 62 |
-
"rss_trend": "stable",
|
| 63 |
-
"n_alerts": 0,
|
| 64 |
-
"alerts": [],
|
| 65 |
-
"alert_types": [],
|
| 66 |
-
"meta_history": [
|
| 67 |
-
{
|
| 68 |
-
"T": 1.0100501775741577,
|
| 69 |
-
"tau": 1.0100501775741577,
|
| 70 |
-
"lambda_ent": 0.009999999776482582,
|
| 71 |
-
"init_gate": 0.047456394881010056,
|
| 72 |
-
"meta_loss": 17.669658660888672,
|
| 73 |
-
"sharpness": 71.65159606933594
|
| 74 |
-
},
|
| 75 |
-
{
|
| 76 |
-
"T": 1.02021062374115,
|
| 77 |
-
"tau": 1.0202070474624634,
|
| 78 |
-
"lambda_ent": 0.009900504723191261,
|
| 79 |
-
"init_gate": 0.02058723010122776,
|
| 80 |
-
"meta_loss": 18.058149337768555,
|
| 81 |
-
"sharpness": 109.38550567626953
|
| 82 |
-
}
|
| 83 |
-
],
|
| 84 |
-
"elapsed_s": 21.375046549000217,
|
| 85 |
-
"n_params": 9657632,
|
| 86 |
-
"hyp_T_params": 1795914,
|
| 87 |
-
"num_layers_hyp": 8,
|
| 88 |
-
"n_w8a8_layers": 80,
|
| 89 |
-
"final_learnable_params": {
|
| 90 |
-
"T": 1.02021062374115,
|
| 91 |
-
"tau": 1.0202070474624634,
|
| 92 |
-
"lambda_ent": 0.009802023880183697,
|
| 93 |
-
"init_gate": 0.05,
|
| 94 |
-
"residual_gate": [
|
| 95 |
-
0.005989676807075739,
|
| 96 |
-
0.006018564570695162,
|
| 97 |
-
0.006039654836058617
|
| 98 |
-
],
|
| 99 |
-
"module_logits": [
|
| 100 |
-
-0.22852112352848053,
|
| 101 |
-
-0.10588397085666656,
|
| 102 |
-
0.12294353544712067,
|
| 103 |
-
-0.22121863067150116,
|
| 104 |
-
0.41890066862106323,
|
| 105 |
-
-0.01949109323322773,
|
| 106 |
-
-0.08710632473230362,
|
| 107 |
-
0.05125897005200386
|
| 108 |
-
]
|
| 109 |
-
},
|
| 110 |
-
"final_bigru_diversity": {
|
| 111 |
-
"mean_cosine_diversity": 0.7214401364326477,
|
| 112 |
-
"mean_parallelism_score": 0.9953877925872803,
|
| 113 |
-
"mean_gate_norm_ratio": 0.09838299453258514,
|
| 114 |
-
"module_level_cosine_diversity": 1.0237387745042466,
|
| 115 |
-
"per_module_parallelism": [
|
| 116 |
-
0.9936217665672302,
|
| 117 |
-
0.9924681186676025,
|
| 118 |
-
0.9992730617523193,
|
| 119 |
-
0.9971094727516174,
|
| 120 |
-
0.9949930310249329,
|
| 121 |
-
0.9948464035987854,
|
| 122 |
-
0.9958065748214722,
|
| 123 |
-
0.9944431781768799
|
| 124 |
-
]
|
| 125 |
-
}
|
| 126 |
-
}
|
| 127 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|