V6.5-7ds cleanup: remove legacy v4_report.json
Browse files- v4_report.json +0 -122
v4_report.json
DELETED
|
@@ -1,122 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"version": "V4",
|
| 3 |
-
"timestamp": 1786016531.119913,
|
| 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_sweep": [
|
| 11 |
-
2,
|
| 12 |
-
8
|
| 13 |
-
],
|
| 14 |
-
"triplet_weight": 0.05
|
| 15 |
-
},
|
| 16 |
-
"xeon_env": {
|
| 17 |
-
"OMP_NUM_THREADS": "2",
|
| 18 |
-
"MKL_NUM_THREADS": "2",
|
| 19 |
-
"ONEDNN_MAX_CPU_ISA": "AMX_INT8",
|
| 20 |
-
"IPEX": true,
|
| 21 |
-
"CPU_ISA": "AMX",
|
| 22 |
-
"torch_threads": 2
|
| 23 |
-
},
|
| 24 |
-
"results": {
|
| 25 |
-
"num_layers_hyp_2": {
|
| 26 |
-
"n_steps": 6,
|
| 27 |
-
"n_epochs": 2,
|
| 28 |
-
"final_train_loss": 1.492246150970459,
|
| 29 |
-
"final_val_loss": 5.687218344211578,
|
| 30 |
-
"final_perplexity": 295.0716901317322,
|
| 31 |
-
"final_overfitting_gap": -1.3897555629412333,
|
| 32 |
-
"final_eval_benchmark": 0.375,
|
| 33 |
-
"avg_tokens_per_sec": 30.952014369483933,
|
| 34 |
-
"avg_tflops_per_cpu": 0.00010286999861443428,
|
| 35 |
-
"avg_time_per_step_ms": 2092.294035499966,
|
| 36 |
-
"avg_grad_norm": 19.630864143371582,
|
| 37 |
-
"avg_cpu_pct": 120.28333333333335,
|
| 38 |
-
"n_alerts": 2,
|
| 39 |
-
"alerts": [
|
| 40 |
-
{
|
| 41 |
-
"step": 5,
|
| 42 |
-
"type": "loss_spike",
|
| 43 |
-
"delta_pct": 0.5995766717862026,
|
| 44 |
-
"from": 8.140497207641602,
|
| 45 |
-
"to": 3.2596449851989746
|
| 46 |
-
},
|
| 47 |
-
{
|
| 48 |
-
"step": 6,
|
| 49 |
-
"type": "loss_spike",
|
| 50 |
-
"delta_pct": 0.5422059280239778,
|
| 51 |
-
"from": 3.2596449851989746,
|
| 52 |
-
"to": 1.492246150970459
|
| 53 |
-
}
|
| 54 |
-
],
|
| 55 |
-
"alert_types": [
|
| 56 |
-
"loss_spike"
|
| 57 |
-
],
|
| 58 |
-
"elapsed_s": 18.622390889000144,
|
| 59 |
-
"n_params": 8862751,
|
| 60 |
-
"hyp_T_params": 1001034,
|
| 61 |
-
"num_layers_hyp": 2,
|
| 62 |
-
"final_bigru_diversity": {
|
| 63 |
-
"mean_cosine_diversity": 0.6990604996681213,
|
| 64 |
-
"mean_parallelism_score": 0.9938665628433228,
|
| 65 |
-
"mean_gate_norm_ratio": 0.11281532049179077,
|
| 66 |
-
"module_level_cosine_diversity": 0.8235123827388244,
|
| 67 |
-
"per_module_parallelism": [
|
| 68 |
-
0.9952576756477356,
|
| 69 |
-
0.9962905645370483,
|
| 70 |
-
0.9829346537590027,
|
| 71 |
-
0.9921810626983643,
|
| 72 |
-
0.9968430399894714,
|
| 73 |
-
0.9968783855438232,
|
| 74 |
-
0.9929060935974121,
|
| 75 |
-
0.9943419694900513
|
| 76 |
-
]
|
| 77 |
-
}
|
| 78 |
-
},
|
| 79 |
-
"num_layers_hyp_8": {
|
| 80 |
-
"n_steps": 6,
|
| 81 |
-
"n_epochs": 2,
|
| 82 |
-
"final_train_loss": 4.706010818481445,
|
| 83 |
-
"final_val_loss": 5.576698219776153,
|
| 84 |
-
"final_perplexity": 264.19784088397313,
|
| 85 |
-
"final_overfitting_gap": -1.148307160536448,
|
| 86 |
-
"final_eval_benchmark": 0.25,
|
| 87 |
-
"avg_tokens_per_sec": 31.42855771366708,
|
| 88 |
-
"avg_tflops_per_cpu": 0.0001138220299728001,
|
| 89 |
-
"avg_time_per_step_ms": 2074.0577321666933,
|
| 90 |
-
"avg_grad_norm": 15.662032763163248,
|
| 91 |
-
"avg_cpu_pct": 120.39999999999999,
|
| 92 |
-
"n_alerts": 0,
|
| 93 |
-
"alerts": [],
|
| 94 |
-
"alert_types": [],
|
| 95 |
-
"elapsed_s": 18.50429308999992,
|
| 96 |
-
"n_params": 9657631,
|
| 97 |
-
"hyp_T_params": 1795914,
|
| 98 |
-
"num_layers_hyp": 8,
|
| 99 |
-
"final_bigru_diversity": {
|
| 100 |
-
"mean_cosine_diversity": 0.6962708830833435,
|
| 101 |
-
"mean_parallelism_score": 0.9929798245429993,
|
| 102 |
-
"mean_gate_norm_ratio": 0.124363474547863,
|
| 103 |
-
"module_level_cosine_diversity": 0.7712274699338845,
|
| 104 |
-
"per_module_parallelism": [
|
| 105 |
-
0.9903037548065186,
|
| 106 |
-
0.9907274842262268,
|
| 107 |
-
0.9923153519630432,
|
| 108 |
-
0.9945146441459656,
|
| 109 |
-
0.9895800948143005,
|
| 110 |
-
0.9969634413719177,
|
| 111 |
-
0.9970486164093018,
|
| 112 |
-
0.997769296169281
|
| 113 |
-
]
|
| 114 |
-
}
|
| 115 |
-
}
|
| 116 |
-
},
|
| 117 |
-
"analysis": {
|
| 118 |
-
"best_num_layers_hyp": 8,
|
| 119 |
-
"best_val_loss": 5.576698219776153,
|
| 120 |
-
"best_eval_acc": 0.375
|
| 121 |
-
}
|
| 122 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|