V6+: add v6_progressive_report.json
Browse files- v6_progressive_report.json +635 -0
v6_progressive_report.json
ADDED
|
@@ -0,0 +1,635 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"version": "V6_progressive",
|
| 3 |
+
"timestamp": 1786054197.7304018,
|
| 4 |
+
"config": {
|
| 5 |
+
"epochs": 1,
|
| 6 |
+
"batch_size": 8,
|
| 7 |
+
"max_seq_len": 24,
|
| 8 |
+
"num_layers_hyp": 8,
|
| 9 |
+
"stages_punctuation": [
|
| 10 |
+
100,
|
| 11 |
+
200,
|
| 12 |
+
300,
|
| 13 |
+
400,
|
| 14 |
+
500
|
| 15 |
+
],
|
| 16 |
+
"stages_text": [
|
| 17 |
+
100,
|
| 18 |
+
200,
|
| 19 |
+
300,
|
| 20 |
+
400,
|
| 21 |
+
500
|
| 22 |
+
],
|
| 23 |
+
"stages_heavy": [
|
| 24 |
+
10,
|
| 25 |
+
20,
|
| 26 |
+
30,
|
| 27 |
+
40,
|
| 28 |
+
50,
|
| 29 |
+
60,
|
| 30 |
+
70,
|
| 31 |
+
80,
|
| 32 |
+
90,
|
| 33 |
+
100
|
| 34 |
+
]
|
| 35 |
+
},
|
| 36 |
+
"xeon_env": {
|
| 37 |
+
"OMP_NUM_THREADS": "2",
|
| 38 |
+
"ONEDNN_MAX_CPU_ISA": "AMX_INT8",
|
| 39 |
+
"IPEX": true,
|
| 40 |
+
"IPEX_VERSION": "2.8.0+cpu",
|
| 41 |
+
"torch_version": "2.8.0+cpu",
|
| 42 |
+
"torch_threads": 2,
|
| 43 |
+
"onednn_enabled": true
|
| 44 |
+
},
|
| 45 |
+
"evolution": [
|
| 46 |
+
{
|
| 47 |
+
"stage": "punctuation",
|
| 48 |
+
"n_samples": 100,
|
| 49 |
+
"epoch": 0,
|
| 50 |
+
"n_steps": 5,
|
| 51 |
+
"avg_tflops": 0.00366,
|
| 52 |
+
"avg_tokens_per_sec": 62.2,
|
| 53 |
+
"final_rss_mb": 1173,
|
| 54 |
+
"max_rss_mb": 1173,
|
| 55 |
+
"train_loss_mean": 13.0303,
|
| 56 |
+
"val_loss": 8.7272,
|
| 57 |
+
"perplexity": 6168.49,
|
| 58 |
+
"overfitting_gap": 4.3242,
|
| 59 |
+
"eval_acc": 0.125,
|
| 60 |
+
"elapsed_s": 42.8,
|
| 61 |
+
"som": {
|
| 62 |
+
"sigma": 1.844,
|
| 63 |
+
"alpha": 0.0946,
|
| 64 |
+
"step": 11
|
| 65 |
+
}
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"stage": "punctuation",
|
| 69 |
+
"n_samples": 200,
|
| 70 |
+
"epoch": 0,
|
| 71 |
+
"n_steps": 6,
|
| 72 |
+
"avg_tflops": 0.0035833333333333338,
|
| 73 |
+
"avg_tokens_per_sec": 61.333333333333336,
|
| 74 |
+
"final_rss_mb": 1406,
|
| 75 |
+
"max_rss_mb": 1406,
|
| 76 |
+
"train_loss_mean": 13.014000000000001,
|
| 77 |
+
"val_loss": 8.8108,
|
| 78 |
+
"perplexity": 6706.45,
|
| 79 |
+
"overfitting_gap": 4.2318,
|
| 80 |
+
"eval_acc": 0.125,
|
| 81 |
+
"elapsed_s": 87.5,
|
| 82 |
+
"som": {
|
| 83 |
+
"sigma": 1.645,
|
| 84 |
+
"alpha": 0.0874,
|
| 85 |
+
"step": 27
|
| 86 |
+
}
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"stage": "punctuation",
|
| 90 |
+
"n_samples": 300,
|
| 91 |
+
"epoch": 0,
|
| 92 |
+
"n_steps": 7,
|
| 93 |
+
"avg_tflops": 0.003342857142857143,
|
| 94 |
+
"avg_tokens_per_sec": 57.285714285714285,
|
| 95 |
+
"final_rss_mb": 1424,
|
| 96 |
+
"max_rss_mb": 1424,
|
| 97 |
+
"train_loss_mean": 13.063342857142857,
|
| 98 |
+
"val_loss": 8.669,
|
| 99 |
+
"perplexity": 5819.93,
|
| 100 |
+
"overfitting_gap": 4.3253,
|
| 101 |
+
"eval_acc": 0.125,
|
| 102 |
+
"elapsed_s": 138.7,
|
| 103 |
+
"som": {
|
| 104 |
+
"sigma": 1.365,
|
| 105 |
+
"alpha": 0.076,
|
| 106 |
+
"step": 55
|
| 107 |
+
}
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"stage": "punctuation",
|
| 111 |
+
"n_samples": 400,
|
| 112 |
+
"epoch": 0,
|
| 113 |
+
"n_steps": 8,
|
| 114 |
+
"avg_tflops": 0.0033625,
|
| 115 |
+
"avg_tokens_per_sec": 57.5,
|
| 116 |
+
"final_rss_mb": 1450,
|
| 117 |
+
"max_rss_mb": 1450,
|
| 118 |
+
"train_loss_mean": 12.999825,
|
| 119 |
+
"val_loss": 8.8403,
|
| 120 |
+
"perplexity": 6907.28,
|
| 121 |
+
"overfitting_gap": 4.1864,
|
| 122 |
+
"eval_acc": 0.125,
|
| 123 |
+
"elapsed_s": 194.5,
|
| 124 |
+
"som": {
|
| 125 |
+
"sigma": 1.08,
|
| 126 |
+
"alpha": 0.0622,
|
| 127 |
+
"step": 95
|
| 128 |
+
}
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"stage": "punctuation",
|
| 132 |
+
"n_samples": 500,
|
| 133 |
+
"epoch": 0,
|
| 134 |
+
"n_steps": 48,
|
| 135 |
+
"elapsed_s": 238.3462211240003,
|
| 136 |
+
"val_loss": 9.420673779078893,
|
| 137 |
+
"train_loss_mean": 13.594396392504374,
|
| 138 |
+
"overfitting_gap": 4.173722613425481,
|
| 139 |
+
"perplexity": 12340.894455600206,
|
| 140 |
+
"eval_acc": 0.25,
|
| 141 |
+
"final_rss_mb": 1434.72265625,
|
| 142 |
+
"max_rss_mb": 1437.234375,
|
| 143 |
+
"avg_tflops": 0.0031595985181522963,
|
| 144 |
+
"avg_tokens_per_sec": 54.12507616421209,
|
| 145 |
+
"n_alerts_in_stage": 0,
|
| 146 |
+
"som": {
|
| 147 |
+
"sigma": 1.4007433652877808,
|
| 148 |
+
"alpha": 0.0774916559457779,
|
| 149 |
+
"step": 51
|
| 150 |
+
},
|
| 151 |
+
"n_train": 382,
|
| 152 |
+
"n_val": 50
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"stage": "text",
|
| 156 |
+
"n_samples": 100,
|
| 157 |
+
"epoch": 0,
|
| 158 |
+
"n_steps": 10,
|
| 159 |
+
"elapsed_s": 45.20832308199988,
|
| 160 |
+
"val_loss": 9.436400890350342,
|
| 161 |
+
"train_loss_mean": 13.515088558197021,
|
| 162 |
+
"overfitting_gap": 4.07868766784668,
|
| 163 |
+
"perplexity": 12536.515319349752,
|
| 164 |
+
"eval_acc": 0.25,
|
| 165 |
+
"final_rss_mb": 1430.53125,
|
| 166 |
+
"max_rss_mb": 1437.234375,
|
| 167 |
+
"avg_tflops": 0.0031754993304676077,
|
| 168 |
+
"avg_tokens_per_sec": 54.3974628844535,
|
| 169 |
+
"n_alerts_in_stage": 1,
|
| 170 |
+
"som": {
|
| 171 |
+
"sigma": 1.356813669204712,
|
| 172 |
+
"alpha": 0.07557837665081024,
|
| 173 |
+
"step": 56
|
| 174 |
+
},
|
| 175 |
+
"n_train": 76,
|
| 176 |
+
"n_val": 10
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"stage": "text",
|
| 180 |
+
"n_samples": 200,
|
| 181 |
+
"epoch": 0,
|
| 182 |
+
"n_steps": 19,
|
| 183 |
+
"elapsed_s": 95.98724977100028,
|
| 184 |
+
"val_loss": 9.334163665771484,
|
| 185 |
+
"train_loss_mean": 13.43099990643953,
|
| 186 |
+
"overfitting_gap": 4.096836240668045,
|
| 187 |
+
"perplexity": 11318.158544435562,
|
| 188 |
+
"eval_acc": 0.125,
|
| 189 |
+
"final_rss_mb": 1439.046875,
|
| 190 |
+
"max_rss_mb": 1452.44921875,
|
| 191 |
+
"avg_tflops": 0.003095574891463858,
|
| 192 |
+
"avg_tokens_per_sec": 53.028328064441716,
|
| 193 |
+
"n_alerts_in_stage": 1,
|
| 194 |
+
"som": {
|
| 195 |
+
"sigma": 1.230128288269043,
|
| 196 |
+
"alpha": 0.06976763159036636,
|
| 197 |
+
"step": 72
|
| 198 |
+
},
|
| 199 |
+
"n_train": 153,
|
| 200 |
+
"n_val": 20
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"stage": "text",
|
| 204 |
+
"n_samples": 300,
|
| 205 |
+
"epoch": 0,
|
| 206 |
+
"n_steps": 29,
|
| 207 |
+
"elapsed_s": 145.05522435300009,
|
| 208 |
+
"val_loss": 9.164406299591064,
|
| 209 |
+
"train_loss_mean": 13.268132209777832,
|
| 210 |
+
"overfitting_gap": 4.103725910186768,
|
| 211 |
+
"perplexity": 9551.049279049606,
|
| 212 |
+
"eval_acc": 0.125,
|
| 213 |
+
"final_rss_mb": 1430.796875,
|
| 214 |
+
"max_rss_mb": 1452.44921875,
|
| 215 |
+
"avg_tflops": 0.0030906872178963992,
|
| 216 |
+
"avg_tokens_per_sec": 52.94460043177426,
|
| 217 |
+
"n_alerts_in_stage": 1,
|
| 218 |
+
"som": {
|
| 219 |
+
"sigma": 1.0518192052841187,
|
| 220 |
+
"alpha": 0.060653068125247955,
|
| 221 |
+
"step": 100
|
| 222 |
+
},
|
| 223 |
+
"n_train": 229,
|
| 224 |
+
"n_val": 30
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"stage": "text",
|
| 228 |
+
"n_samples": 400,
|
| 229 |
+
"epoch": 0,
|
| 230 |
+
"n_steps": 39,
|
| 231 |
+
"elapsed_s": 190.21534844300004,
|
| 232 |
+
"val_loss": 9.129400253295898,
|
| 233 |
+
"train_loss_mean": 13.21440469301664,
|
| 234 |
+
"overfitting_gap": 4.085004439720741,
|
| 235 |
+
"perplexity": 9222.489152733066,
|
| 236 |
+
"eval_acc": 0.125,
|
| 237 |
+
"final_rss_mb": 1452.8515625,
|
| 238 |
+
"max_rss_mb": 1469.19921875,
|
| 239 |
+
"avg_tflops": 0.003150540588834492,
|
| 240 |
+
"avg_tokens_per_sec": 53.96991052800877,
|
| 241 |
+
"n_alerts_in_stage": 1,
|
| 242 |
+
"som": {
|
| 243 |
+
"sigma": 0.8698954582214355,
|
| 244 |
+
"alpha": 0.049658533185720444,
|
| 245 |
+
"step": 140
|
| 246 |
+
},
|
| 247 |
+
"n_train": 306,
|
| 248 |
+
"n_val": 40
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"stage": "text",
|
| 252 |
+
"n_samples": 500,
|
| 253 |
+
"epoch": 0,
|
| 254 |
+
"n_steps": 48,
|
| 255 |
+
"elapsed_s": 250.72700701699978,
|
| 256 |
+
"val_loss": 9.049883842468262,
|
| 257 |
+
"train_loss_mean": 13.146486779054007,
|
| 258 |
+
"overfitting_gap": 4.096602936585745,
|
| 259 |
+
"perplexity": 8517.548489696044,
|
| 260 |
+
"eval_acc": 0.125,
|
| 261 |
+
"final_rss_mb": 1453.1875,
|
| 262 |
+
"max_rss_mb": 1469.19921875,
|
| 263 |
+
"avg_tflops": 0.00300811063867445,
|
| 264 |
+
"avg_tokens_per_sec": 51.53003348154613,
|
| 265 |
+
"n_alerts_in_stage": 1,
|
| 266 |
+
"som": {
|
| 267 |
+
"sigma": 0.7221205830574036,
|
| 268 |
+
"alpha": 0.0384812168776989,
|
| 269 |
+
"step": 191
|
| 270 |
+
},
|
| 271 |
+
"n_train": 382,
|
| 272 |
+
"n_val": 50
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"stage": "heavy",
|
| 276 |
+
"n_samples": 10,
|
| 277 |
+
"epoch": 0,
|
| 278 |
+
"n_steps": 1,
|
| 279 |
+
"elapsed_s": 6.019525716999851,
|
| 280 |
+
"val_loss": 9.139408111572266,
|
| 281 |
+
"train_loss_mean": 9.202308654785156,
|
| 282 |
+
"overfitting_gap": 0.06290054321289062,
|
| 283 |
+
"perplexity": 9315.249911170573,
|
| 284 |
+
"eval_acc": 0.125,
|
| 285 |
+
"final_rss_mb": 1453.35546875,
|
| 286 |
+
"max_rss_mb": 1469.19921875,
|
| 287 |
+
"avg_tflops": 0.002980696835029999,
|
| 288 |
+
"avg_tokens_per_sec": 51.060425016520504,
|
| 289 |
+
"n_alerts_in_stage": 1,
|
| 290 |
+
"som": {
|
| 291 |
+
"sigma": 0.7221205830574036,
|
| 292 |
+
"alpha": 0.0384812168776989,
|
| 293 |
+
"step": 191
|
| 294 |
+
},
|
| 295 |
+
"n_train": 6,
|
| 296 |
+
"n_val": 2
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"stage": "heavy",
|
| 300 |
+
"n_samples": 20,
|
| 301 |
+
"epoch": 0,
|
| 302 |
+
"n_steps": 2,
|
| 303 |
+
"elapsed_s": 9.429588502000115,
|
| 304 |
+
"val_loss": 9.128674507141113,
|
| 305 |
+
"train_loss_mean": 9.246623992919922,
|
| 306 |
+
"overfitting_gap": 0.1179494857788086,
|
| 307 |
+
"perplexity": 9215.798394882486,
|
| 308 |
+
"eval_acc": 0.125,
|
| 309 |
+
"final_rss_mb": 1425.8046875,
|
| 310 |
+
"max_rss_mb": 1469.19921875,
|
| 311 |
+
"avg_tflops": 0.0034907268658084137,
|
| 312 |
+
"avg_tokens_per_sec": 59.797425652303986,
|
| 313 |
+
"n_alerts_in_stage": 2,
|
| 314 |
+
"som": {
|
| 315 |
+
"sigma": 0.7221205830574036,
|
| 316 |
+
"alpha": 0.0384812168776989,
|
| 317 |
+
"step": 191
|
| 318 |
+
},
|
| 319 |
+
"n_train": 15,
|
| 320 |
+
"n_val": 2
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"stage": "heavy",
|
| 324 |
+
"n_samples": 30,
|
| 325 |
+
"epoch": 0,
|
| 326 |
+
"n_steps": 3,
|
| 327 |
+
"elapsed_s": 13.233528376999857,
|
| 328 |
+
"val_loss": 8.997712135314941,
|
| 329 |
+
"train_loss_mean": 9.154280662536621,
|
| 330 |
+
"overfitting_gap": 0.1565685272216797,
|
| 331 |
+
"perplexity": 8084.566358940332,
|
| 332 |
+
"eval_acc": 0.125,
|
| 333 |
+
"final_rss_mb": 1434.00390625,
|
| 334 |
+
"max_rss_mb": 1469.19921875,
|
| 335 |
+
"avg_tflops": 0.003153819574809518,
|
| 336 |
+
"avg_tokens_per_sec": 54.02608075489678,
|
| 337 |
+
"n_alerts_in_stage": 1,
|
| 338 |
+
"som": {
|
| 339 |
+
"sigma": 0.7221205830574036,
|
| 340 |
+
"alpha": 0.0384812168776989,
|
| 341 |
+
"step": 191
|
| 342 |
+
},
|
| 343 |
+
"n_train": 22,
|
| 344 |
+
"n_val": 3
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"stage": "heavy",
|
| 348 |
+
"n_samples": 40,
|
| 349 |
+
"epoch": 0,
|
| 350 |
+
"n_steps": 4,
|
| 351 |
+
"elapsed_s": 17.86659687500014,
|
| 352 |
+
"val_loss": 9.019000053405762,
|
| 353 |
+
"train_loss_mean": 9.115556240081787,
|
| 354 |
+
"overfitting_gap": 0.09655618667602539,
|
| 355 |
+
"perplexity": 8258.514877243768,
|
| 356 |
+
"eval_acc": 0.125,
|
| 357 |
+
"final_rss_mb": 1428.6796875,
|
| 358 |
+
"max_rss_mb": 1469.19921875,
|
| 359 |
+
"avg_tflops": 0.0031227725238592565,
|
| 360 |
+
"avg_tokens_per_sec": 53.49423343704838,
|
| 361 |
+
"n_alerts_in_stage": 1,
|
| 362 |
+
"som": {
|
| 363 |
+
"sigma": 0.7221205830574036,
|
| 364 |
+
"alpha": 0.0384812168776989,
|
| 365 |
+
"step": 191
|
| 366 |
+
},
|
| 367 |
+
"n_train": 30,
|
| 368 |
+
"n_val": 4
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"stage": "heavy",
|
| 372 |
+
"n_samples": 50,
|
| 373 |
+
"epoch": 0,
|
| 374 |
+
"n_steps": 5,
|
| 375 |
+
"elapsed_s": 21.472788195999783,
|
| 376 |
+
"val_loss": 9.079545974731445,
|
| 377 |
+
"train_loss_mean": 9.06125545501709,
|
| 378 |
+
"overfitting_gap": -0.01829051971435547,
|
| 379 |
+
"perplexity": 8773.981513247856,
|
| 380 |
+
"eval_acc": 0.125,
|
| 381 |
+
"final_rss_mb": 1428.89453125,
|
| 382 |
+
"max_rss_mb": 1469.19921875,
|
| 383 |
+
"avg_tflops": 0.003101373347677258,
|
| 384 |
+
"avg_tokens_per_sec": 53.127657736354806,
|
| 385 |
+
"n_alerts_in_stage": 1,
|
| 386 |
+
"som": {
|
| 387 |
+
"sigma": 0.7221205830574036,
|
| 388 |
+
"alpha": 0.0384812168776989,
|
| 389 |
+
"step": 191
|
| 390 |
+
},
|
| 391 |
+
"n_train": 38,
|
| 392 |
+
"n_val": 5
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"stage": "heavy",
|
| 396 |
+
"n_samples": 60,
|
| 397 |
+
"epoch": 0,
|
| 398 |
+
"n_steps": 6,
|
| 399 |
+
"elapsed_s": 29.922955371999706,
|
| 400 |
+
"val_loss": 8.971521377563477,
|
| 401 |
+
"train_loss_mean": 9.019448280334473,
|
| 402 |
+
"overfitting_gap": 0.047926902770996094,
|
| 403 |
+
"perplexity": 7875.574217657398,
|
| 404 |
+
"eval_acc": 0.125,
|
| 405 |
+
"final_rss_mb": 1429.0546875,
|
| 406 |
+
"max_rss_mb": 1469.19921875,
|
| 407 |
+
"avg_tflops": 0.0031399798215805443,
|
| 408 |
+
"avg_tokens_per_sec": 53.78900072928324,
|
| 409 |
+
"n_alerts_in_stage": 1,
|
| 410 |
+
"som": {
|
| 411 |
+
"sigma": 0.7199104428291321,
|
| 412 |
+
"alpha": 0.03828928992152214,
|
| 413 |
+
"step": 192
|
| 414 |
+
},
|
| 415 |
+
"n_train": 45,
|
| 416 |
+
"n_val": 6
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"stage": "heavy",
|
| 420 |
+
"n_samples": 70,
|
| 421 |
+
"epoch": 0,
|
| 422 |
+
"n_steps": 7,
|
| 423 |
+
"elapsed_s": 31.42283428399969,
|
| 424 |
+
"val_loss": 8.976522445678711,
|
| 425 |
+
"train_loss_mean": 9.019445419311523,
|
| 426 |
+
"overfitting_gap": 0.0429229736328125,
|
| 427 |
+
"perplexity": 7915.059151893913,
|
| 428 |
+
"eval_acc": 0.125,
|
| 429 |
+
"final_rss_mb": 1436.99609375,
|
| 430 |
+
"max_rss_mb": 1469.19921875,
|
| 431 |
+
"avg_tflops": 0.003414139923107274,
|
| 432 |
+
"avg_tokens_per_sec": 58.485463362453466,
|
| 433 |
+
"n_alerts_in_stage": 1,
|
| 434 |
+
"som": {
|
| 435 |
+
"sigma": 0.7155559062957764,
|
| 436 |
+
"alpha": 0.03790830448269844,
|
| 437 |
+
"step": 194
|
| 438 |
+
},
|
| 439 |
+
"n_train": 53,
|
| 440 |
+
"n_val": 7
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"stage": "heavy",
|
| 444 |
+
"n_samples": 80,
|
| 445 |
+
"epoch": 0,
|
| 446 |
+
"n_steps": 8,
|
| 447 |
+
"elapsed_s": 38.12266795300002,
|
| 448 |
+
"val_loss": 9.110323905944824,
|
| 449 |
+
"train_loss_mean": 9.032614469528198,
|
| 450 |
+
"overfitting_gap": -0.07770943641662598,
|
| 451 |
+
"perplexity": 9048.225190773235,
|
| 452 |
+
"eval_acc": 0.125,
|
| 453 |
+
"final_rss_mb": 1450.43359375,
|
| 454 |
+
"max_rss_mb": 1469.19921875,
|
| 455 |
+
"avg_tflops": 0.0031962099789579618,
|
| 456 |
+
"avg_tokens_per_sec": 54.75224385441235,
|
| 457 |
+
"n_alerts_in_stage": 1,
|
| 458 |
+
"som": {
|
| 459 |
+
"sigma": 0.7091852426528931,
|
| 460 |
+
"alpha": 0.03734392300248146,
|
| 461 |
+
"step": 197
|
| 462 |
+
},
|
| 463 |
+
"n_train": 61,
|
| 464 |
+
"n_val": 8
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"stage": "heavy",
|
| 468 |
+
"n_samples": 90,
|
| 469 |
+
"epoch": 0,
|
| 470 |
+
"n_steps": 9,
|
| 471 |
+
"elapsed_s": 45.12368812299974,
|
| 472 |
+
"val_loss": 9.070122718811035,
|
| 473 |
+
"train_loss_mean": 8.960626072353786,
|
| 474 |
+
"overfitting_gap": -0.10949664645724866,
|
| 475 |
+
"perplexity": 8691.690374177408,
|
| 476 |
+
"eval_acc": 0.125,
|
| 477 |
+
"final_rss_mb": 1439.9609375,
|
| 478 |
+
"max_rss_mb": 1469.19921875,
|
| 479 |
+
"avg_tflops": 0.0029507120752743318,
|
| 480 |
+
"avg_tokens_per_sec": 50.54677513467092,
|
| 481 |
+
"n_alerts_in_stage": 1,
|
| 482 |
+
"som": {
|
| 483 |
+
"sigma": 0.7009830474853516,
|
| 484 |
+
"alpha": 0.036604464054107666,
|
| 485 |
+
"step": 201
|
| 486 |
+
},
|
| 487 |
+
"n_train": 68,
|
| 488 |
+
"n_val": 9
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"stage": "heavy",
|
| 492 |
+
"n_samples": 100,
|
| 493 |
+
"epoch": 0,
|
| 494 |
+
"n_steps": 10,
|
| 495 |
+
"elapsed_s": 43.32768855500035,
|
| 496 |
+
"val_loss": 9.022727489471436,
|
| 497 |
+
"train_loss_mean": 8.872154712677002,
|
| 498 |
+
"overfitting_gap": -0.1505727767944336,
|
| 499 |
+
"perplexity": 8289.355405787683,
|
| 500 |
+
"eval_acc": 0.125,
|
| 501 |
+
"final_rss_mb": 1453.52734375,
|
| 502 |
+
"max_rss_mb": 1469.19921875,
|
| 503 |
+
"avg_tflops": 0.003307840858510533,
|
| 504 |
+
"avg_tokens_per_sec": 56.66452157683456,
|
| 505 |
+
"n_alerts_in_stage": 1,
|
| 506 |
+
"som": {
|
| 507 |
+
"sigma": 0.691180944442749,
|
| 508 |
+
"alpha": 0.03570069745182991,
|
| 509 |
+
"step": 206
|
| 510 |
+
},
|
| 511 |
+
"n_train": 76,
|
| 512 |
+
"n_val": 10
|
| 513 |
+
}
|
| 514 |
+
],
|
| 515 |
+
"final_summary": {
|
| 516 |
+
"n_steps": 248,
|
| 517 |
+
"final_train_loss": 8.874523162841797,
|
| 518 |
+
"final_val_loss": 9.022727489471436,
|
| 519 |
+
"final_perplexity": 8289.355405787683,
|
| 520 |
+
"final_overfitting_gap": -0.1505727767944336,
|
| 521 |
+
"final_eval_acc": 0.125,
|
| 522 |
+
"avg_tokens_per_sec": 53.496948306142976,
|
| 523 |
+
"avg_tflops": 0.0031229310067100195,
|
| 524 |
+
"avg_time_per_step_ms": 3530.4227777419583,
|
| 525 |
+
"avg_grad_norm": 0.9999999535982114,
|
| 526 |
+
"avg_cpu_pct": 132.9350806451613,
|
| 527 |
+
"final_rss_mb": 1453.52734375,
|
| 528 |
+
"max_rss_mb": 1469.19921875,
|
| 529 |
+
"rss_trend": "growing",
|
| 530 |
+
"n_alerts": 15,
|
| 531 |
+
"alerts": [
|
| 532 |
+
[
|
| 533 |
+
49,
|
| 534 |
+
"cpu_underutil",
|
| 535 |
+
"cpu=0.0%"
|
| 536 |
+
],
|
| 537 |
+
[
|
| 538 |
+
59,
|
| 539 |
+
"cpu_underutil",
|
| 540 |
+
"cpu=0.0%"
|
| 541 |
+
],
|
| 542 |
+
[
|
| 543 |
+
78,
|
| 544 |
+
"cpu_underutil",
|
| 545 |
+
"cpu=0.0%"
|
| 546 |
+
],
|
| 547 |
+
[
|
| 548 |
+
107,
|
| 549 |
+
"cpu_underutil",
|
| 550 |
+
"cpu=0.0%"
|
| 551 |
+
],
|
| 552 |
+
[
|
| 553 |
+
146,
|
| 554 |
+
"cpu_underutil",
|
| 555 |
+
"cpu=0.0%"
|
| 556 |
+
],
|
| 557 |
+
[
|
| 558 |
+
194,
|
| 559 |
+
"cpu_underutil",
|
| 560 |
+
"cpu=0.0%"
|
| 561 |
+
],
|
| 562 |
+
[
|
| 563 |
+
195,
|
| 564 |
+
"cpu_underutil",
|
| 565 |
+
"cpu=0.0%"
|
| 566 |
+
],
|
| 567 |
+
[
|
| 568 |
+
197,
|
| 569 |
+
"cpu_underutil",
|
| 570 |
+
"cpu=0.0%"
|
| 571 |
+
],
|
| 572 |
+
[
|
| 573 |
+
200,
|
| 574 |
+
"cpu_underutil",
|
| 575 |
+
"cpu=0.0%"
|
| 576 |
+
],
|
| 577 |
+
[
|
| 578 |
+
204,
|
| 579 |
+
"cpu_underutil",
|
| 580 |
+
"cpu=0.0%"
|
| 581 |
+
],
|
| 582 |
+
[
|
| 583 |
+
209,
|
| 584 |
+
"cpu_underutil",
|
| 585 |
+
"cpu=0.0%"
|
| 586 |
+
],
|
| 587 |
+
[
|
| 588 |
+
215,
|
| 589 |
+
"cpu_underutil",
|
| 590 |
+
"cpu=0.0%"
|
| 591 |
+
],
|
| 592 |
+
[
|
| 593 |
+
222,
|
| 594 |
+
"cpu_underutil",
|
| 595 |
+
"cpu=0.0%"
|
| 596 |
+
],
|
| 597 |
+
[
|
| 598 |
+
230,
|
| 599 |
+
"cpu_underutil",
|
| 600 |
+
"cpu=0.0%"
|
| 601 |
+
],
|
| 602 |
+
[
|
| 603 |
+
239,
|
| 604 |
+
"cpu_underutil",
|
| 605 |
+
"cpu=0.0%"
|
| 606 |
+
]
|
| 607 |
+
],
|
| 608 |
+
"alert_types": [
|
| 609 |
+
"cpu_underutil"
|
| 610 |
+
],
|
| 611 |
+
"mtp_alpha_mean": 0.3333333432674408,
|
| 612 |
+
"som_quant_loss_final": 0.0,
|
| 613 |
+
"som_entropy_final": null,
|
| 614 |
+
"som_sigma_final": 0.0,
|
| 615 |
+
"ewc_lambda_final": 10.0,
|
| 616 |
+
"rlhf_reward_final": 0.0,
|
| 617 |
+
"reasoning_steps_final": null,
|
| 618 |
+
"final_learnable_params": {
|
| 619 |
+
"T": 1.4550622701644897,
|
| 620 |
+
"tau": 1.4946794509887695,
|
| 621 |
+
"lambda_ent": 0.006898071616888046,
|
| 622 |
+
"init_gate": 0.0012438705889508128,
|
| 623 |
+
"mtp_alphas": [
|
| 624 |
+
0.3333333432674408,
|
| 625 |
+
0.3333333432674408,
|
| 626 |
+
0.3333333432674408
|
| 627 |
+
],
|
| 628 |
+
"rep_penalty_lambda": 1.0,
|
| 629 |
+
"ewc_lambda": 10.0,
|
| 630 |
+
"rlhf_eta": 0.009999999776482582
|
| 631 |
+
},
|
| 632 |
+
"n_w8a8_layers": 81
|
| 633 |
+
},
|
| 634 |
+
"n_params": 9729312
|
| 635 |
+
}
|