Upload weights and configs - David-fully_shared-weighted_sum - Run 20251012_135249
Browse files- weights/David-fully_shared-weighted_sum/20251012_135249/MODEL_SUMMARY.txt +10 -9
- weights/David-fully_shared-weighted_sum/20251012_135249/best_model_acc70.62.safetensors +3 -0
- weights/David-fully_shared-weighted_sum/20251012_135249/best_model_acc70.62_metadata.json +258 -0
- weights/David-fully_shared-weighted_sum/20251012_135249/training_history.json +14 -7
weights/David-fully_shared-weighted_sum/20251012_135249/MODEL_SUMMARY.txt
CHANGED
|
@@ -3,24 +3,24 @@
|
|
| 3 |
║ DAVID MODEL SUMMARY ║
|
| 4 |
╠══════════════════════════════════════════════════════════════╣
|
| 5 |
║ ║
|
| 6 |
-
║ 🎯 VALIDATION ACCURACY: 70.
|
| 7 |
║ ║
|
| 8 |
╚══════════════════════════════════════════════════════════════╝
|
| 9 |
|
| 10 |
MODEL: David-fully_shared-weighted_sum
|
| 11 |
RUN ID: 20251012_135249
|
| 12 |
-
BEST EPOCH:
|
| 13 |
|
| 14 |
═══════════════════════════════════════════════════════════════
|
| 15 |
|
| 16 |
📊 PERFORMANCE BREAKDOWN
|
| 17 |
|
| 18 |
-
Final Training Accuracy: 70.
|
| 19 |
-
Best Validation Accuracy: 70.
|
| 20 |
|
| 21 |
Per-Scale Accuracies:
|
| 22 |
-
• Scale 256:
|
| 23 |
-
• Scale 512: 70.
|
| 24 |
|
| 25 |
═══════════════════════════════════════════════════════════════
|
| 26 |
|
|
@@ -41,17 +41,18 @@ Epoch | Train Acc | Val Acc | Learning Rate
|
|
| 41 |
------|-----------|----------|--------------
|
| 42 |
1 | 63.40% | 68.06% | 9.76e-04
|
| 43 |
2 | 69.48% | 69.62% | 9.05e-04
|
| 44 |
-
3 | 70.34% | 70.14%
|
|
|
|
| 45 |
|
| 46 |
═══════════════════════════════════════════════════════════════
|
| 47 |
|
| 48 |
📁 FILES
|
| 49 |
|
| 50 |
-
Best Model: best_model_acc70.
|
| 51 |
Config: david_config.json
|
| 52 |
Training Cfg: train_config.json
|
| 53 |
History: training_history.json
|
| 54 |
|
| 55 |
═══════════════════════════════════════════════════════════════
|
| 56 |
|
| 57 |
-
Generated: 2025-10-12 13:
|
|
|
|
| 3 |
║ DAVID MODEL SUMMARY ║
|
| 4 |
╠══════════════════════════════════════════════════════════════╣
|
| 5 |
║ ║
|
| 6 |
+
║ 🎯 VALIDATION ACCURACY: 70.62% ║
|
| 7 |
║ ║
|
| 8 |
╚══════════════════════════════════════════════════════════════╝
|
| 9 |
|
| 10 |
MODEL: David-fully_shared-weighted_sum
|
| 11 |
RUN ID: 20251012_135249
|
| 12 |
+
BEST EPOCH: 4/10
|
| 13 |
|
| 14 |
═══════════════════════════════════════════════════════════════
|
| 15 |
|
| 16 |
📊 PERFORMANCE BREAKDOWN
|
| 17 |
|
| 18 |
+
Final Training Accuracy: 70.91%
|
| 19 |
+
Best Validation Accuracy: 70.62%
|
| 20 |
|
| 21 |
Per-Scale Accuracies:
|
| 22 |
+
• Scale 256: 70.00%
|
| 23 |
+
• Scale 512: 70.48%
|
| 24 |
|
| 25 |
═══════════════════════════════════════════════════════════════
|
| 26 |
|
|
|
|
| 41 |
------|-----------|----------|--------------
|
| 42 |
1 | 63.40% | 68.06% | 9.76e-04
|
| 43 |
2 | 69.48% | 69.62% | 9.05e-04
|
| 44 |
+
3 | 70.34% | 70.14% | 7.94e-04
|
| 45 |
+
4 | 70.91% | 70.62% 👑 | 6.55e-04
|
| 46 |
|
| 47 |
═══════════════════════════════════════════════════════════════
|
| 48 |
|
| 49 |
📁 FILES
|
| 50 |
|
| 51 |
+
Best Model: best_model_acc70.62.safetensors
|
| 52 |
Config: david_config.json
|
| 53 |
Training Cfg: train_config.json
|
| 54 |
History: training_history.json
|
| 55 |
|
| 56 |
═══════════════════════════════════════════════════════════════
|
| 57 |
|
| 58 |
+
Generated: 2025-10-12 13:59:36
|
weights/David-fully_shared-weighted_sum/20251012_135249/best_model_acc70.62.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e4cfc3ff5bcf992b2de21faaf4bafd92d879884b82122b9174f4c949f6c882e0
|
| 3 |
+
size 2628344
|
weights/David-fully_shared-weighted_sum/20251012_135249/best_model_acc70.62_metadata.json
ADDED
|
@@ -0,0 +1,258 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3,
|
| 3 |
+
"optimizer_state_dict": {
|
| 4 |
+
"state": {
|
| 5 |
+
"0": {
|
| 6 |
+
"step": "tensor(5008.)",
|
| 7 |
+
"exp_avg": "tensor([[-3.1698e-04, -5.3519e-04, 1.1910e-03, ..., -8.2113e-04,\n 4.8216e-04, 1.6444e-04],\n [ 2.7477e-04, 1.1142e-03, -1.3078e-03, ..., 5.7999e-04,\n -8.9148e-06, 9.1113e-04],\n [-8.2589e-04, -2.9956e-04, -5.1007e-04, ..., 1.1582e-04,\n 1.2160e-04, 2.4747e-04],\n ...,\n [ 3.3656e-04, -9.5074e-04, 1.2575e-03, ..., -6.2508e-04,\n -3.3645e-04, -2.6710e-04],\n [-4.2320e-05, 1.2007e-03, -4.5198e-04, ..., 2.5459e-04,\n -3.1364e-04, 2.5951e-04],\n [-1.3778e-04, -1.1183e-03, 1.4776e-04, ..., -4.8258e-04,\n -1.8598e-04, -4.3040e-04]], device='cuda:0')",
|
| 8 |
+
"exp_avg_sq": "tensor([[6.5189e-06, 3.1642e-05, 1.5296e-05, ..., 5.9374e-06, 3.1292e-06,\n 7.9759e-06],\n [6.9973e-06, 2.9500e-05, 2.1048e-05, ..., 9.4941e-06, 2.7658e-06,\n 1.1527e-05],\n [8.1335e-06, 5.8815e-05, 1.8711e-05, ..., 5.0383e-06, 2.9758e-06,\n 3.7443e-06],\n ...,\n [6.4168e-06, 2.7325e-05, 1.6071e-05, ..., 3.6923e-06, 1.8070e-06,\n 2.8517e-06],\n [6.3439e-06, 3.8155e-05, 1.4116e-05, ..., 5.6530e-06, 3.2216e-06,\n 5.4984e-06],\n [4.8340e-06, 4.6442e-05, 1.1693e-05, ..., 3.3530e-06, 2.3039e-06,\n 2.4168e-06]], device='cuda:0')"
|
| 9 |
+
},
|
| 10 |
+
"1": {
|
| 11 |
+
"step": "tensor(5008.)",
|
| 12 |
+
"exp_avg": "tensor([-2.5104e-03, 3.4168e-02, -6.9777e-03, 3.9801e-02, 7.6481e-03,\n -6.2455e-05, 6.1985e-03, 3.1006e-02, 5.6243e-02, -9.9632e-03,\n 2.1540e-02, 5.7086e-03, -3.6751e-03, 4.2999e-02, -2.3480e-02,\n -8.4001e-04, -1.5057e-02, -1.4524e-03, -1.8966e-02, -5.6840e-04,\n 3.2299e-03, 5.3387e-03, -7.7774e-03, 2.0351e-02, 4.7065e-03,\n 4.0731e-02, 1.0931e-02, -1.3350e-02, -2.7831e-03, -4.9239e-04,\n 6.1323e-03, -2.1082e-02, -6.2548e-03, 3.7704e-03, 3.2727e-02,\n -2.5364e-02, 9.0384e-03, -2.7037e-03, -2.1414e-02, -9.5033e-03,\n 4.7247e-02, -2.9120e-02, 1.7090e-02, 1.0068e-02, -4.4471e-02,\n 8.2953e-03, 9.9657e-03, 3.0187e-03, 1.1810e-02, 1.2307e-02,\n 3.9321e-03, -4.6474e-02, -1.4355e-02, 1.3418e-03, 3.4159e-02,\n 3.8544e-02, -1.4628e-02, 2.3876e-02, 4.0262e-03, 5.9991e-02,\n -2.7855e-03, 5.3585e-02, -2.7273e-02, -3.0340e-02, 5.2893e-03,\n -2.3607e-02, 3.6783e-03, -2.0367e-04, 2.7332e-02, 2.0610e-02,\n -5.4460e-03, -1.5509e-02, 3.7368e-03, -1.2921e-02, 4.1321e-03,\n -1.8554e-02, -2.8406e-03, -1.6064e-02, 1.8157e-02, -3.2822e-02,\n -4.5057e-03, 7.9072e-03, -1.5604e-02, -1.8994e-02, 3.0543e-02,\n -3.2243e-03, 9.4561e-03, 2.0394e-02, -2.1304e-03, 1.1358e-02,\n 1.2420e-02, 1.4215e-02, 1.1900e-02, 9.8848e-03, -7.4076e-03,\n -1.7745e-02, -4.0986e-02, 4.8446e-03, 1.7155e-02, -2.2759e-02,\n -1.7304e-02, -1.3062e-02, 5.0651e-04, -1.5825e-02, 2.3448e-02,\n -1.5464e-02, -1.5430e-02, -3.3077e-03, -3.0272e-03, 3.0234e-02,\n -1.3938e-02, -2.6630e-03, 2.7121e-02, -7.0694e-03, -4.6057e-02,\n -1.2266e-02, -1.5319e-02, 3.4454e-02, 1.2501e-02, 2.3372e-03,\n -9.2317e-04, -1.4815e-02, -1.6036e-02, 8.4539e-03, 2.9866e-02,\n 1.3529e-02, 1.2571e-02, 1.2676e-03, -2.6436e-02, -7.2654e-03,\n 1.7491e-02, -3.7211e-02, 6.9720e-03, 3.2748e-04, -1.3336e-02,\n 1.0386e-02, 1.0123e-02, -2.1221e-02, -2.1771e-02, -7.6480e-03,\n 9.0552e-03, 1.3339e-02, -7.0563e-03, 1.2040e-02, -2.9000e-02,\n 1.7560e-03, -4.0233e-02, 3.4214e-02, 2.0395e-02, -5.5094e-03,\n 3.9331e-02, -1.1777e-02, -2.5983e-02, 1.9328e-03, 6.5185e-03,\n -2.2930e-02, 3.2280e-02, 4.5011e-03, -8.9875e-03, -1.3246e-02,\n -1.3168e-02, -2.8507e-02, 4.5293e-03, 2.2673e-02, -3.7503e-02,\n 4.2065e-03, 3.5675e-03, 8.6966e-03, 8.2690e-03, -2.4359e-02,\n 1.4396e-02, -1.1176e-02, -5.7044e-03, -2.5164e-02, -3.1170e-02,\n -8.0781e-03, -1.1662e-02, 2.4166e-03, -8.4427e-03, -7.0593e-03,\n 3.9129e-03, -4.2690e-02, 3.7000e-03, -5.0837e-03, 2.4719e-02,\n -1.4756e-04, 7.0762e-04, 4.2317e-03, 1.2022e-02, -3.6126e-03,\n 2.0566e-03, -1.5509e-02, -1.7636e-02, -9.1363e-03, 1.6927e-02,\n -6.4621e-04, -3.2787e-02, -2.1185e-02, 1.9380e-03, 4.0818e-02,\n 9.3708e-03, 2.1183e-02, 1.9166e-02, -5.2811e-02, 2.4973e-02,\n -3.9621e-02, 5.8903e-03, 2.2393e-02, -1.0400e-02, 2.0039e-02,\n 2.0438e-02, -2.6781e-02, -1.5929e-03, -9.5791e-03, -4.2876e-02,\n 2.2803e-02, -1.9225e-02, -1.5367e-02, 1.3412e-02, 5.2692e-03,\n -1.2834e-02, 1.2164e-02, 4.1167e-02, 1.0978e-02, -1.2697e-02,\n -1.5837e-02, 1.8167e-03, -1.1672e-02, -2.3085e-02, -1.7108e-02,\n 1.2246e-02, -2.8250e-02, 1.2592e-02, 1.6464e-03, 7.5052e-03,\n 8.9033e-03, -7.4002e-03, 1.5025e-03, -6.7152e-03, 3.5080e-02,\n -5.3596e-02, -2.2000e-02, -1.3317e-02, 2.2882e-02, -5.4364e-03,\n 1.5311e-02, 1.5320e-02, 2.6003e-02, 8.8647e-03, -2.1601e-03,\n 1.3547e-02, 2.3885e-02, 1.3819e-02, -1.7490e-03, -2.4603e-02,\n -1.6598e-02, 1.8353e-03, 1.7605e-02, 2.3116e-02, -6.1384e-03,\n 1.6893e-02, 7.4554e-03, -4.7471e-02, 1.6806e-02, 2.9913e-02,\n -1.1161e-02, 1.1520e-02, -2.0691e-02, 1.4283e-02, 2.7865e-02,\n -1.9033e-02, 2.5173e-02, 3.9828e-03, -1.5492e-02, -1.7457e-02,\n 9.9116e-03, 5.7109e-03, 1.1468e-02, 1.2886e-02, -5.0242e-04,\n 1.1017e-02, -1.5619e-02, 1.5609e-02, -1.1836e-02, 3.0127e-02,\n -1.1025e-02, 4.6805e-03, -5.5822e-03, 6.2969e-03, 2.9904e-03,\n 5.5920e-03, 1.0882e-02, -6.1410e-03, -9.4244e-03, -2.8379e-02,\n 1.6147e-02, 2.4333e-02, -1.2140e-02, -4.6111e-02, -1.8788e-02,\n 1.8420e-02, 1.5638e-02, -1.8587e-02, 1.6414e-02, 1.6508e-02,\n -5.7674e-02, 9.3973e-03, 2.0711e-02, -1.8778e-02, -1.0669e-02,\n -1.0719e-02, -1.8432e-02, -1.1865e-02, 3.6619e-02, 1.2960e-02,\n 6.6173e-03, -3.3672e-03, 1.2410e-02, 7.4468e-03, 1.0811e-02,\n 4.6342e-03, -2.0566e-02, 2.6534e-02, 2.0242e-02, -1.6959e-02,\n 9.7971e-03, -4.0857e-02, -8.4027e-03, 1.3457e-02, -2.4373e-02,\n 2.7009e-02, -1.0271e-02, 1.2585e-02, 3.4747e-03, -2.0042e-02,\n -3.1927e-02, -7.0674e-02, 2.8155e-03, 4.2091e-03, 1.2097e-02,\n -2.5127e-02, 5.6756e-03, -2.0960e-02, -1.6454e-02, 5.6600e-03,\n 1.2758e-02, 8.2530e-03, 4.7455e-02, -8.9722e-03, -1.3702e-02,\n 6.5928e-03, -1.9916e-02, -1.0162e-02, -2.1106e-02, 4.5347e-03,\n 6.5474e-03, -2.9087e-02, 1.1092e-02, -5.1705e-03, -6.2430e-04,\n -1.9594e-02, 1.5242e-02, -1.0947e-02, -5.4965e-03, 3.9349e-02,\n -2.2032e-02, -8.7220e-04, 5.0322e-02, 1.6663e-04, -1.7806e-04,\n 7.8811e-03, 3.1242e-03, -2.9685e-03, -2.1946e-02, -1.5227e-02,\n 7.6305e-03, 9.0282e-03, 2.0346e-02, 3.1065e-03, -6.2133e-03,\n 1.4937e-02, -1.0552e-04, 9.6152e-03, 3.6909e-03, -1.1266e-02,\n -1.2867e-02, 8.6171e-03, 2.5134e-02, -1.5871e-02, -2.1290e-02,\n -2.5301e-02, -8.4784e-02, 1.9918e-02, -1.5291e-02, -1.5164e-02,\n 8.5149e-04, -2.6641e-03, 4.1857e-03, 2.5995e-02, -5.2654e-03,\n -1.2835e-02, 9.9277e-03, 1.4652e-02, -2.9051e-02, -7.6750e-03,\n -6.6661e-03, 1.3328e-03, 1.5302e-02, 2.3645e-02, -1.1087e-02,\n 1.5164e-02, 2.3987e-02, 5.4668e-03, 3.9769e-03, -1.3203e-02,\n -4.4059e-03, -8.0189e-03, -2.7318e-02, 1.0306e-02, -4.3524e-03,\n -7.4530e-04, 4.6738e-03, -1.9953e-02, 2.1375e-02, 2.0218e-02,\n 1.8234e-02, 3.1562e-02, 1.9446e-02, -1.3310e-02, 4.4030e-03,\n 2.5133e-02, -9.7029e-03, 9.9173e-04, -2.6552e-02, -4.3088e-03,\n -6.9973e-03, -2.4709e-02, -2.0177e-02, 7.4533e-03, -8.1761e-02,\n 5.6167e-03, -3.2919e-02, 3.3213e-02, -4.3095e-02, 1.4795e-02,\n 3.1127e-03, -8.7273e-04, -3.4861e-02, 3.0707e-03, 2.8874e-03,\n -1.6061e-03, 2.8353e-02, 2.1192e-02, -1.1797e-02, 1.7467e-02,\n 1.5275e-03, 9.7025e-03, 1.2381e-02, -9.0989e-03, 1.5988e-02,\n -1.2268e-02, 1.8622e-02, 2.4966e-02, 1.2837e-03, 1.6943e-03,\n 3.2358e-03, -5.8723e-03, -4.9841e-03, -7.6963e-03, 1.9622e-03,\n 6.4487e-03, -1.7038e-02, -1.8454e-03, -3.2233e-03, 1.4927e-03,\n -3.9422e-03, -1.3616e-02, -2.2771e-02, 4.3973e-02, -3.4422e-03,\n 2.7356e-02, 1.0631e-02, -4.4992e-03, -1.2811e-02, -1.2279e-02,\n -6.3699e-03, 1.4474e-02, 1.9088e-02, -1.5032e-02, 9.3431e-03,\n 1.8035e-03, 4.1258e-03, 2.7189e-02, 1.3950e-02, 1.9129e-02,\n 4.3516e-03, 5.6194e-03, 1.6647e-02, 1.0471e-02, -3.5357e-02,\n 1.2696e-02, -2.7778e-02, 2.8728e-02, 1.3858e-03, -4.5450e-02,\n 6.6715e-03, 1.0090e-02, 2.5593e-02, 1.2661e-02, -1.7810e-02,\n 2.4147e-02, 1.5395e-02], device='cuda:0')",
|
| 13 |
+
"exp_avg_sq": "tensor([0.0050, 0.0066, 0.0045, 0.0030, 0.0048, 0.0034, 0.0037, 0.0035, 0.0062,\n 0.0043, 0.0034, 0.0046, 0.0053, 0.0035, 0.0044, 0.0042, 0.0012, 0.0025,\n 0.0030, 0.0033, 0.0035, 0.0036, 0.0043, 0.0046, 0.0054, 0.0032, 0.0032,\n 0.0038, 0.0062, 0.0020, 0.0033, 0.0059, 0.0047, 0.0037, 0.0039, 0.0035,\n 0.0025, 0.0044, 0.0038, 0.0037, 0.0056, 0.0037, 0.0046, 0.0064, 0.0045,\n 0.0055, 0.0067, 0.0042, 0.0027, 0.0046, 0.0033, 0.0037, 0.0051, 0.0025,\n 0.0061, 0.0046, 0.0022, 0.0049, 0.0031, 0.0038, 0.0032, 0.0065, 0.0050,\n 0.0031, 0.0024, 0.0032, 0.0035, 0.0049, 0.0089, 0.0057, 0.0068, 0.0043,\n 0.0080, 0.0030, 0.0043, 0.0076, 0.0049, 0.0042, 0.0034, 0.0081, 0.0063,\n 0.0042, 0.0061, 0.0048, 0.0033, 0.0030, 0.0011, 0.0088, 0.0055, 0.0052,\n 0.0032, 0.0037, 0.0027, 0.0045, 0.0063, 0.0040, 0.0066, 0.0052, 0.0044,\n 0.0053, 0.0032, 0.0041, 0.0041, 0.0022, 0.0034, 0.0033, 0.0074, 0.0032,\n 0.0054, 0.0049, 0.0021, 0.0063, 0.0035, 0.0032, 0.0026, 0.0032, 0.0048,\n 0.0061, 0.0027, 0.0044, 0.0038, 0.0027, 0.0048, 0.0052, 0.0062, 0.0045,\n 0.0051, 0.0036, 0.0057, 0.0029, 0.0046, 0.0041, 0.0033, 0.0027, 0.0049,\n 0.0028, 0.0038, 0.0023, 0.0051, 0.0080, 0.0046, 0.0029, 0.0034, 0.0061,\n 0.0056, 0.0037, 0.0040, 0.0039, 0.0049, 0.0055, 0.0035, 0.0043, 0.0027,\n 0.0029, 0.0028, 0.0029, 0.0047, 0.0048, 0.0026, 0.0055, 0.0034, 0.0036,\n 0.0044, 0.0039, 0.0055, 0.0041, 0.0022, 0.0032, 0.0056, 0.0042, 0.0042,\n 0.0040, 0.0022, 0.0038, 0.0048, 0.0027, 0.0035, 0.0043, 0.0030, 0.0045,\n 0.0049, 0.0087, 0.0023, 0.0044, 0.0058, 0.0048, 0.0017, 0.0026, 0.0045,\n 0.0030, 0.0023, 0.0024, 0.0063, 0.0041, 0.0052, 0.0039, 0.0037, 0.0036,\n 0.0045, 0.0084, 0.0022, 0.0039, 0.0028, 0.0059, 0.0042, 0.0028, 0.0022,\n 0.0030, 0.0033, 0.0057, 0.0027, 0.0059, 0.0055, 0.0066, 0.0039, 0.0028,\n 0.0040, 0.0032, 0.0031, 0.0052, 0.0028, 0.0046, 0.0070, 0.0066, 0.0024,\n 0.0044, 0.0038, 0.0057, 0.0037, 0.0070, 0.0041, 0.0035, 0.0054, 0.0028,\n 0.0041, 0.0055, 0.0047, 0.0048, 0.0038, 0.0085, 0.0061, 0.0035, 0.0049,\n 0.0033, 0.0031, 0.0054, 0.0038, 0.0041, 0.0029, 0.0054, 0.0027, 0.0039,\n 0.0018, 0.0083, 0.0049, 0.0048, 0.0073, 0.0042, 0.0047, 0.0044, 0.0040,\n 0.0069, 0.0034, 0.0027, 0.0022, 0.0033, 0.0032, 0.0054, 0.0025, 0.0043,\n 0.0023, 0.0023, 0.0052, 0.0046, 0.0038, 0.0056, 0.0017, 0.0032, 0.0077,\n 0.0045, 0.0045, 0.0051, 0.0043, 0.0032, 0.0058, 0.0050, 0.0032, 0.0041,\n 0.0059, 0.0025, 0.0037, 0.0058, 0.0039, 0.0047, 0.0071, 0.0049, 0.0068,\n 0.0048, 0.0085, 0.0064, 0.0043, 0.0036, 0.0037, 0.0034, 0.0036, 0.0041,\n 0.0045, 0.0032, 0.0063, 0.0043, 0.0042, 0.0054, 0.0029, 0.0044, 0.0022,\n 0.0048, 0.0028, 0.0045, 0.0042, 0.0027, 0.0049, 0.0065, 0.0043, 0.0035,\n 0.0045, 0.0038, 0.0032, 0.0042, 0.0030, 0.0038, 0.0059, 0.0042, 0.0037,\n 0.0036, 0.0025, 0.0028, 0.0044, 0.0035, 0.0052, 0.0048, 0.0033, 0.0036,\n 0.0035, 0.0042, 0.0047, 0.0044, 0.0035, 0.0039, 0.0034, 0.0042, 0.0042,\n 0.0069, 0.0033, 0.0051, 0.0030, 0.0045, 0.0062, 0.0042, 0.0024, 0.0044,\n 0.0048, 0.0047, 0.0044, 0.0025, 0.0033, 0.0042, 0.0030, 0.0084, 0.0030,\n 0.0030, 0.0032, 0.0103, 0.0057, 0.0048, 0.0045, 0.0029, 0.0047, 0.0032,\n 0.0046, 0.0034, 0.0033, 0.0043, 0.0035, 0.0046, 0.0059, 0.0036, 0.0023,\n 0.0050, 0.0058, 0.0018, 0.0050, 0.0078, 0.0030, 0.0042, 0.0027, 0.0044,\n 0.0026, 0.0033, 0.0057, 0.0023, 0.0022, 0.0046, 0.0048, 0.0065, 0.0049,\n 0.0081, 0.0043, 0.0031, 0.0053, 0.0025, 0.0052, 0.0054, 0.0062, 0.0030,\n 0.0061, 0.0033, 0.0039, 0.0029, 0.0065, 0.0030, 0.0046, 0.0039, 0.0053,\n 0.0058, 0.0079, 0.0041, 0.0071, 0.0059, 0.0030, 0.0027, 0.0049, 0.0031,\n 0.0032, 0.0126, 0.0030, 0.0052, 0.0053, 0.0050, 0.0045, 0.0063, 0.0060,\n 0.0062, 0.0039, 0.0030, 0.0028, 0.0036, 0.0057, 0.0026, 0.0031, 0.0038,\n 0.0018, 0.0070, 0.0043, 0.0040, 0.0034, 0.0067, 0.0044, 0.0026, 0.0046,\n 0.0037, 0.0052, 0.0061, 0.0034, 0.0057, 0.0013, 0.0043, 0.0031, 0.0040,\n 0.0039, 0.0052, 0.0045, 0.0027, 0.0059, 0.0042, 0.0031, 0.0031, 0.0032,\n 0.0032, 0.0047, 0.0035, 0.0038, 0.0047, 0.0036, 0.0041, 0.0036, 0.0035,\n 0.0035, 0.0034, 0.0085, 0.0043, 0.0007, 0.0051, 0.0038, 0.0037, 0.0050,\n 0.0047, 0.0037, 0.0057, 0.0046, 0.0043, 0.0037, 0.0044, 0.0058, 0.0036,\n 0.0049, 0.0057, 0.0038, 0.0027, 0.0042, 0.0043, 0.0049, 0.0034],\n device='cuda:0')"
|
| 14 |
+
},
|
| 15 |
+
"2": {
|
| 16 |
+
"step": "tensor(5008.)",
|
| 17 |
+
"exp_avg": "tensor([-1.8522e-04, 3.6255e-03, -1.5313e-04, 1.0535e-02, -1.3510e-04,\n 2.3259e-04, 1.2621e-03, 6.5829e-03, 5.7372e-03, -4.4697e-04,\n 5.3964e-03, 1.3892e-03, -3.1156e-05, 1.0771e-02, -1.9080e-03,\n -5.0768e-04, -1.6288e-02, -2.4194e-04, -4.0788e-03, 1.5116e-03,\n 1.5192e-03, 1.4548e-03, -2.4203e-03, 4.0185e-03, 2.4985e-03,\n 1.1382e-02, 2.5143e-03, -2.9488e-03, -1.2847e-03, -3.8035e-04,\n -8.7077e-05, -2.8631e-03, -5.7437e-04, 9.7467e-04, 1.0646e-02,\n -6.6872e-03, 3.6025e-03, 1.3282e-03, -2.5739e-03, -1.9367e-03,\n 6.9145e-03, -6.5974e-03, 2.4858e-03, -1.2713e-03, -5.3852e-03,\n 1.5849e-03, 1.2842e-03, -2.2610e-04, 3.5451e-03, 1.2807e-03,\n 1.4945e-03, -9.4178e-03, -1.7186e-03, -1.1593e-03, 4.3844e-03,\n 8.6539e-03, -1.9670e-03, 3.4958e-03, -8.4537e-04, 1.3866e-02,\n -1.0347e-03, 6.0470e-03, -4.4387e-03, -8.5358e-03, 3.6776e-03,\n -5.3314e-03, -4.9978e-05, 1.7807e-04, 8.1114e-04, 4.4607e-03,\n -6.8490e-04, -1.7909e-03, 8.3440e-04, -2.4732e-03, -1.3727e-04,\n -1.2413e-03, -2.8047e-03, -2.0074e-03, 2.2137e-03, -5.4175e-03,\n -1.2513e-03, 2.6036e-03, -3.1819e-03, -3.6825e-03, 5.8063e-03,\n -1.4821e-03, 1.0210e-02, 2.7869e-03, -7.1903e-04, 1.7015e-03,\n 3.6882e-03, 4.2374e-03, 1.4004e-03, 1.2421e-03, -4.2434e-04,\n -1.7686e-03, -4.7997e-03, 1.5265e-03, 1.2884e-03, -2.4596e-03,\n -3.3089e-03, -1.8943e-03, 7.1425e-04, -5.4351e-03, 5.3132e-03,\n -3.8989e-03, -2.1099e-03, 1.7109e-03, -3.8276e-04, 3.4299e-03,\n -6.1717e-03, -9.7126e-04, 4.9547e-03, -1.0459e-03, -1.5534e-02,\n -3.3741e-03, -1.7384e-03, 5.7039e-03, 3.3510e-03, 8.1031e-04,\n 1.1959e-03, -3.6146e-03, -3.9584e-03, -5.4033e-04, 3.1327e-03,\n 2.3433e-03, 8.1018e-04, 1.1586e-04, -4.0595e-03, -1.9885e-03,\n 2.3177e-03, -6.0568e-03, 1.3298e-03, -1.6571e-04, -3.4731e-03,\n 2.4547e-03, 1.6228e-03, -7.6658e-03, -3.1885e-03, -2.5370e-04,\n 9.0715e-04, 4.1218e-03, -2.3414e-03, 1.3410e-03, -2.4377e-03,\n 4.0502e-04, -6.1507e-03, 7.0306e-03, 4.1204e-03, -9.8790e-04,\n 5.1375e-03, -1.4875e-03, -5.4780e-03, 6.8532e-04, 2.2716e-03,\n -6.5182e-03, 4.2833e-03, -5.5504e-04, -3.1807e-03, -1.3002e-03,\n -2.8330e-03, -6.4813e-03, 5.6913e-04, 3.6967e-03, -7.4828e-03,\n 4.8717e-05, 5.6363e-04, 1.5913e-03, 1.1905e-03, -3.5574e-03,\n 4.5444e-03, -2.4213e-03, -1.3962e-03, -4.1149e-03, -4.8437e-03,\n -2.7930e-03, -8.8924e-04, 5.2905e-04, -1.7554e-03, -7.5911e-04,\n 1.2830e-03, -5.7228e-03, 2.0183e-03, -1.7293e-03, 3.3561e-03,\n -8.2064e-04, -7.7591e-04, 3.4588e-03, 2.7544e-03, -2.9515e-04,\n 2.0324e-03, -5.2249e-03, -2.6126e-03, -1.5396e-04, 2.6361e-03,\n 4.1819e-05, -5.3868e-03, -3.9165e-03, 3.1560e-04, 5.6261e-03,\n 6.0197e-03, 3.5181e-03, 3.1578e-03, -5.7539e-03, 4.1434e-03,\n -1.0289e-02, 2.7651e-03, 6.9496e-03, -2.9781e-04, 1.7964e-03,\n 5.4778e-03, -3.8229e-03, 1.3869e-04, -2.6039e-03, -9.4778e-03,\n 5.4920e-03, -2.6729e-03, -2.3941e-03, 2.0044e-03, -6.4672e-04,\n -2.8173e-03, 1.5756e-03, 4.0342e-03, 9.3294e-04, -3.3924e-03,\n -2.6743e-03, 6.8801e-04, -1.5608e-03, -6.0295e-03, -2.3657e-03,\n 1.3844e-03, -6.4803e-03, 6.6104e-04, -1.6548e-04, 1.3553e-03,\n 1.8330e-03, -8.2731e-04, 9.7396e-04, -2.0531e-03, 3.5479e-03,\n -6.2968e-03, -3.5432e-03, -3.2174e-03, 2.9831e-03, -1.2154e-03,\n 2.3914e-03, 2.7332e-03, 1.5432e-03, 2.3387e-03, -1.8736e-03,\n 3.0850e-03, 3.1077e-03, 7.4488e-03, -8.7367e-04, -4.2681e-03,\n -1.2819e-03, -3.5317e-04, 3.5018e-03, 4.9122e-03, -5.1903e-04,\n 2.1355e-03, 4.9434e-04, -1.1896e-02, 7.2369e-03, 1.5535e-02,\n -1.4759e-03, 2.4061e-03, -3.0111e-03, 5.5484e-03, 4.5509e-03,\n -6.1841e-03, 5.3608e-03, 2.3555e-03, -1.1969e-03, -2.1683e-03,\n 2.0123e-03, 2.6770e-03, 1.4011e-03, 2.0434e-03, -3.4203e-04,\n 8.9775e-04, -2.7034e-03, 2.0167e-03, -2.1702e-03, 4.6517e-03,\n -1.3462e-03, 3.9840e-03, -1.0155e-03, 4.6495e-04, -2.9474e-05,\n 1.4573e-03, 3.7398e-04, -1.3713e-03, -9.1774e-04, -5.5071e-03,\n 2.2785e-03, 3.0812e-03, -1.7073e-03, -4.5601e-03, -3.5555e-03,\n 4.6967e-03, 1.8673e-03, -2.7224e-03, 4.8802e-03, 2.8559e-03,\n -1.2589e-02, 5.4426e-04, 2.6652e-03, -2.4635e-03, -4.5564e-04,\n -2.6494e-03, -2.3336e-03, -2.0879e-03, 6.0169e-03, 3.6055e-03,\n 1.8923e-03, -9.2931e-04, 1.1788e-03, 1.0344e-03, 2.6720e-03,\n 1.2399e-03, -3.7774e-03, 3.3172e-03, 2.3329e-03, -2.4604e-03,\n 1.6091e-03, -8.9713e-03, -2.1939e-03, 2.1609e-03, -2.5709e-03,\n 4.3693e-03, -1.9953e-03, 2.4604e-03, 1.1703e-03, -9.2590e-03,\n -1.3445e-02, -1.0396e-02, -2.4175e-04, 1.3193e-04, 1.0071e-03,\n -4.0485e-03, 1.7197e-04, -4.2295e-03, -2.4214e-03, 1.4804e-03,\n 2.7534e-03, 8.7744e-04, 6.1584e-03, -2.8032e-03, -1.0796e-03,\n 1.5550e-03, -3.7502e-03, -2.2404e-03, -2.0095e-03, 2.9110e-03,\n 6.3046e-04, -3.0893e-03, 1.4013e-03, -1.3491e-03, -5.6905e-04,\n -2.4340e-03, 1.8858e-03, -2.5546e-03, -1.5020e-03, 5.3284e-03,\n -3.8409e-03, 1.7521e-03, 5.5340e-03, 4.3229e-04, 1.0291e-03,\n 3.0174e-03, 1.9975e-03, -7.8852e-04, -4.1374e-03, -3.1997e-03,\n 3.0451e-03, 1.0699e-03, 3.8290e-03, -5.3640e-04, -2.4357e-03,\n 1.3108e-03, 1.3212e-03, 1.4203e-03, 1.8534e-03, -1.4897e-03,\n -1.0373e-03, 1.1560e-03, 2.8497e-03, -1.6156e-04, -1.0524e-02,\n -2.9890e-03, -7.5329e-03, 3.8749e-03, -2.4362e-03, -3.6874e-03,\n 1.1855e-03, -1.2481e-03, 1.0516e-03, 2.8077e-03, -2.2211e-03,\n -2.9537e-03, 1.7688e-03, 1.6699e-03, -3.0042e-03, -8.9280e-04,\n -3.5973e-04, 2.8835e-04, 5.8107e-03, 3.0963e-03, -3.9327e-03,\n 2.9546e-03, 2.9207e-03, 6.1110e-04, -3.8952e-04, -1.8641e-03,\n -1.6460e-03, -1.6752e-03, -3.4937e-03, 2.2742e-03, -1.4443e-03,\n -3.8342e-04, 1.3154e-03, -1.7900e-03, 2.7044e-03, 2.4551e-03,\n 2.2383e-03, 5.2238e-03, 1.8996e-03, -4.6117e-03, 1.8794e-03,\n 2.6167e-03, -2.5035e-03, 9.0428e-06, -1.7697e-03, -1.6883e-03,\n -1.2294e-03, -1.6021e-03, -3.8275e-03, -5.6328e-04, -7.7801e-03,\n 8.6424e-04, -5.8380e-03, 5.4957e-03, -1.2729e-02, 2.7011e-03,\n 2.4458e-03, 7.6065e-05, -1.1920e-02, 3.4239e-03, 8.6423e-04,\n -4.4367e-04, 4.5532e-03, 1.6144e-03, -3.0476e-03, 2.7555e-03,\n -3.4145e-04, 1.2865e-03, 5.9655e-03, -1.6000e-03, 2.7633e-03,\n -2.8360e-03, 2.2849e-03, 6.3621e-03, 6.3192e-04, 3.0164e-03,\n 8.7261e-04, -1.7191e-03, -6.5484e-04, -1.0118e-03, 3.4431e-04,\n 2.1266e-03, -6.1445e-03, -5.2570e-04, -6.6870e-04, 2.6265e-03,\n -1.0713e-03, -2.1326e-03, -5.3897e-03, 4.2042e-03, -1.6089e-03,\n 3.5964e-03, 1.4423e-03, -9.3182e-04, -2.3123e-03, -4.1381e-04,\n 8.8728e-05, 2.6405e-03, 4.5410e-03, -2.4153e-03, 2.3754e-03,\n 1.2695e-01, 1.2544e-04, 6.6326e-03, 1.3619e-03, 3.4678e-03,\n 1.7313e-03, 5.2683e-04, 2.7184e-03, -1.8337e-04, -6.1635e-03,\n 2.1329e-03, -6.7386e-03, 2.8709e-03, -1.4759e-04, -6.9772e-03,\n 1.3039e-03, 1.0317e-03, 6.1746e-03, 8.4434e-04, -2.1455e-03,\n 3.8230e-03, 2.9627e-03], device='cuda:0')",
|
| 18 |
+
"exp_avg_sq": "tensor([1.2523e-04, 6.6986e-05, 1.2116e-04, 1.8424e-04, 4.4554e-05, 9.4944e-05,\n 8.1372e-05, 1.0707e-04, 5.0798e-05, 1.1009e-04, 1.0984e-04, 1.6223e-04,\n 5.3899e-05, 9.2269e-05, 4.2220e-05, 5.8361e-05, 1.4517e-03, 1.2695e-04,\n 9.7182e-05, 1.4598e-04, 1.8377e-04, 1.8327e-04, 1.4754e-04, 8.2306e-05,\n 9.7159e-05, 1.8771e-04, 1.3146e-04, 1.1886e-04, 4.8414e-05, 2.8402e-04,\n 1.0844e-04, 7.7882e-05, 2.0024e-04, 4.9372e-05, 2.9548e-04, 2.0837e-04,\n 1.5225e-04, 1.9960e-04, 5.2442e-05, 1.0319e-04, 8.3050e-05, 9.7335e-05,\n 4.5802e-05, 5.6467e-05, 8.9232e-05, 5.7040e-05, 1.7698e-04, 1.1470e-04,\n 2.0302e-04, 8.7965e-05, 1.7787e-04, 1.9949e-04, 6.4641e-05, 1.5966e-04,\n 8.6019e-05, 1.4765e-04, 2.3529e-04, 1.1355e-04, 1.0628e-04, 1.4521e-04,\n 1.6789e-04, 4.8724e-05, 9.8704e-05, 2.3974e-04, 3.4386e-04, 1.2462e-04,\n 1.3133e-04, 1.0393e-04, 5.1814e-05, 1.1545e-04, 4.9022e-05, 7.7750e-05,\n 6.1831e-05, 7.5455e-05, 1.0759e-04, 8.4392e-05, 1.1330e-04, 8.9470e-05,\n 1.3558e-04, 1.2539e-04, 7.6090e-05, 5.5179e-05, 9.4030e-05, 1.0794e-04,\n 9.6591e-05, 1.3715e-04, 1.3258e-03, 8.8475e-05, 6.0994e-05, 1.4361e-04,\n 1.3464e-04, 1.7668e-04, 1.5458e-04, 6.5137e-05, 7.4552e-05, 6.8471e-05,\n 1.1108e-04, 7.8376e-05, 8.9538e-05, 5.9254e-05, 4.7561e-05, 5.6597e-05,\n 6.0027e-05, 1.8991e-04, 1.9552e-04, 1.1987e-04, 5.5832e-05, 1.8692e-04,\n 7.8943e-05, 6.0437e-05, 3.1178e-04, 5.4642e-05, 8.1673e-05, 6.1334e-05,\n 2.4299e-04, 7.3733e-05, 5.5016e-05, 7.1706e-05, 1.4126e-04, 9.1808e-05,\n 8.0300e-05, 1.4025e-04, 8.6230e-05, 8.7367e-05, 8.4242e-05, 6.8020e-05,\n 5.8353e-05, 1.1828e-04, 7.1540e-05, 2.1147e-04, 6.2599e-05, 1.0091e-04,\n 2.3765e-04, 1.5735e-04, 1.0525e-04, 1.4519e-04, 6.4476e-05, 1.8042e-04,\n 7.2102e-05, 8.7122e-05, 1.4198e-04, 1.4509e-04, 8.9113e-05, 1.1003e-04,\n 8.7329e-05, 1.2871e-04, 9.6270e-05, 1.0591e-04, 8.1956e-05, 8.5205e-05,\n 4.5535e-05, 1.4612e-04, 9.0666e-05, 1.7401e-04, 8.9555e-05, 1.7292e-04,\n 1.0476e-04, 9.1175e-05, 3.5841e-04, 4.7400e-05, 1.8727e-04, 8.8750e-05,\n 1.3041e-04, 8.8197e-05, 1.2690e-04, 6.1981e-05, 2.7223e-04, 1.5020e-04,\n 6.5766e-05, 8.0078e-05, 2.1016e-04, 1.0134e-04, 2.7681e-04, 8.2836e-05,\n 8.4943e-05, 1.0801e-04, 8.6735e-05, 1.3817e-04, 8.2089e-05, 5.5993e-05,\n 7.5534e-05, 1.0496e-04, 3.6425e-04, 1.0551e-04, 9.1130e-05, 1.1942e-04,\n 4.0675e-04, 2.9728e-04, 6.8181e-05, 1.4836e-04, 2.5542e-04, 2.6229e-04,\n 1.2856e-04, 8.1971e-05, 1.0795e-04, 8.7188e-05, 5.5743e-05, 8.2614e-05,\n 5.4218e-05, 8.1069e-05, 1.8615e-04, 1.4801e-04, 5.6193e-05, 6.6997e-05,\n 9.8050e-05, 1.7528e-04, 2.3381e-04, 1.5006e-04, 1.0967e-04, 9.3555e-05,\n 1.1654e-04, 9.8723e-05, 1.0170e-04, 1.1522e-04, 2.0807e-04, 1.0556e-04,\n 1.4668e-04, 6.2732e-05, 9.0497e-05, 1.0994e-04, 1.1244e-04, 5.6724e-05,\n 5.5324e-05, 1.0726e-04, 1.8196e-04, 8.8198e-05, 1.2755e-04, 6.1006e-05,\n 2.3597e-04, 1.4846e-04, 5.8310e-05, 1.0152e-04, 5.5897e-05, 1.0231e-04,\n 1.0070e-04, 9.3514e-05, 1.1760e-04, 1.1127e-04, 1.2740e-04, 9.1303e-05,\n 5.8452e-05, 1.2113e-04, 1.1607e-04, 8.6669e-05, 2.2294e-04, 6.6238e-05,\n 8.3786e-05, 4.4050e-05, 1.4178e-04, 9.7821e-05, 1.0594e-04, 1.3374e-04,\n 2.4270e-04, 8.4597e-05, 8.7168e-05, 5.4871e-05, 2.1885e-04, 9.6127e-05,\n 1.2334e-04, 1.1196e-04, 6.9501e-05, 7.1980e-05, 1.8719e-04, 2.1613e-04,\n 4.5777e-04, 1.2419e-04, 9.6459e-05, 8.0591e-05, 1.4921e-04, 8.1743e-05,\n 1.6421e-04, 9.9901e-05, 5.8463e-05, 5.1760e-05, 6.1055e-05, 2.0938e-04,\n 4.5446e-04, 1.2893e-04, 9.9262e-05, 7.2900e-05, 3.3475e-05, 1.1353e-04,\n 8.8585e-05, 1.3068e-04, 1.2383e-04, 6.3960e-05, 5.0304e-04, 5.7406e-05,\n 9.2829e-05, 1.2557e-04, 1.4560e-04, 8.3618e-05, 1.0956e-04, 7.9975e-05,\n 1.0283e-04, 9.2704e-05, 5.1257e-05, 1.3949e-04, 6.6230e-05, 7.6203e-05,\n 1.9613e-04, 9.7039e-05, 8.2208e-05, 2.4278e-04, 8.3813e-05, 1.3782e-04,\n 9.3250e-05, 9.2586e-05, 1.0282e-04, 8.2992e-05, 8.1455e-05, 5.7956e-05,\n 9.6528e-05, 8.1171e-05, 2.0499e-04, 1.2683e-04, 6.4962e-05, 1.4277e-04,\n 6.2362e-05, 1.9602e-04, 5.1536e-05, 1.7418e-04, 7.5165e-05, 1.0726e-04,\n 6.3585e-05, 6.9655e-05, 1.2161e-04, 9.2993e-05, 5.2467e-05, 9.4874e-05,\n 8.0111e-05, 1.7686e-04, 7.7687e-05, 1.1123e-04, 3.2704e-04, 1.8123e-04,\n 1.0602e-04, 2.0768e-04, 9.9920e-05, 6.7729e-05, 1.3056e-04, 1.4894e-04,\n 8.6595e-05, 9.9493e-05, 8.0257e-05, 6.6612e-05, 1.0458e-04, 6.3755e-05,\n 2.2046e-04, 4.9061e-05, 1.4207e-04, 6.6854e-05, 4.0077e-05, 8.5543e-05,\n 1.3267e-04, 7.1772e-05, 3.7755e-05, 5.9806e-05, 2.7584e-04, 6.8443e-05,\n 8.4962e-05, 3.9981e-05, 7.3324e-05, 1.2425e-04, 5.0420e-05, 9.6223e-05,\n 7.4396e-05, 9.0699e-05, 1.1028e-04, 1.2445e-04, 1.9289e-04, 1.0415e-04,\n 7.8881e-05, 1.1117e-04, 7.4217e-05, 1.2787e-04, 6.3342e-05, 9.5199e-05,\n 7.1017e-05, 1.5413e-04, 5.0901e-05, 7.4639e-05, 1.2584e-04, 1.2741e-04,\n 6.5531e-05, 6.5949e-05, 1.1776e-04, 5.2917e-05, 4.0147e-05, 4.7319e-04,\n 9.5597e-05, 5.9757e-05, 1.4823e-04, 1.0383e-04, 1.6301e-04, 1.2839e-04,\n 1.9682e-04, 1.9227e-04, 8.4731e-05, 1.7100e-04, 1.5789e-04, 8.5146e-05,\n 5.7628e-05, 6.0284e-05, 5.7087e-05, 6.9985e-05, 9.6312e-05, 9.7002e-05,\n 7.1006e-05, 2.6565e-04, 1.0438e-04, 5.1345e-05, 9.7686e-05, 2.8777e-04,\n 5.7980e-05, 2.0136e-04, 7.4364e-05, 8.9109e-05, 6.4790e-05, 2.0653e-04,\n 1.4892e-04, 8.7637e-05, 7.1935e-05, 9.1308e-05, 5.8890e-05, 8.7834e-05,\n 8.3344e-05, 8.1662e-05, 1.5072e-04, 1.5744e-04, 5.5135e-05, 2.7939e-04,\n 1.5539e-04, 7.4746e-05, 1.2828e-04, 1.1838e-04, 5.0376e-05, 9.6000e-05,\n 8.7741e-05, 6.7925e-05, 9.8187e-05, 1.0400e-04, 7.7922e-05, 1.9234e-04,\n 1.1549e-04, 5.7317e-05, 6.8711e-05, 1.6703e-04, 1.3772e-04, 1.7837e-04,\n 3.3044e-04, 9.9899e-05, 4.9899e-05, 1.6240e-04, 1.6448e-04, 8.5568e-05,\n 3.9680e-05, 3.6811e-04, 1.0637e-04, 9.2970e-05, 1.4092e-04, 7.2128e-05,\n 1.8498e-04, 5.1157e-05, 5.6825e-04, 1.4610e-04, 7.3791e-05, 4.8595e-05,\n 1.2591e-04, 5.0822e-05, 7.4429e-05, 1.5391e-04, 6.8650e-05, 1.1462e-04,\n 1.3953e-04, 1.5881e-04, 9.6444e-05, 1.3288e-04, 3.3301e-05, 8.9038e-05,\n 6.5211e-05, 3.6194e-05, 1.4765e-04, 1.2478e-04, 6.7140e-05, 1.7360e-04,\n 6.5815e-05, 1.1666e-04, 9.1772e-05, 8.2484e-05, 8.1899e-02, 6.5510e-05,\n 1.1032e-04, 9.0318e-05, 8.7481e-05, 1.8969e-04, 1.4189e-04, 1.0422e-04,\n 7.0140e-05, 1.3374e-04, 8.1858e-05, 1.0193e-04, 7.7059e-05, 9.4111e-05,\n 1.0622e-04, 1.3230e-04, 1.4502e-04, 1.2911e-04, 1.3381e-04, 1.5797e-04,\n 1.3248e-04, 1.0332e-04], device='cuda:0')"
|
| 19 |
+
},
|
| 20 |
+
"3": {
|
| 21 |
+
"step": "tensor(5008.)",
|
| 22 |
+
"exp_avg": "tensor([-5.3964e-04, 5.1920e-03, -7.9238e-05, 8.2080e-03, 1.1108e-03,\n 3.0399e-04, 5.4545e-04, 5.9778e-03, 8.2515e-03, -1.0820e-03,\n 4.8625e-03, 1.0962e-03, 1.1282e-04, 8.6880e-03, -3.3768e-03,\n 2.2461e-04, -5.3551e-03, -2.2456e-05, -3.8214e-03, 6.6616e-04,\n 9.2035e-04, 1.4369e-03, -1.6637e-03, 3.2231e-03, 9.1798e-04,\n 9.8198e-03, 2.3023e-03, -2.4209e-03, -7.5709e-04, 1.7604e-04,\n 8.8143e-04, -3.3961e-03, -1.5548e-03, 7.0570e-04, 8.3684e-03,\n -5.0051e-03, 2.4990e-03, -7.4374e-04, -3.3238e-03, -2.0285e-03,\n 7.4764e-03, -5.3989e-03, 2.7939e-03, 1.2212e-03, -6.9228e-03,\n 1.6297e-03, 1.5626e-03, 5.7429e-04, 2.9722e-03, 1.7010e-03,\n 9.0802e-04, -8.9562e-03, -1.9832e-03, -6.0188e-05, 4.9709e-03,\n 7.6623e-03, -1.5739e-03, 4.1249e-03, 8.8980e-04, 1.1258e-02,\n -4.1084e-04, 7.0437e-03, -5.3664e-03, -8.2905e-03, 1.9395e-03,\n -4.9557e-03, 6.2988e-04, 1.8835e-04, 2.9324e-03, 4.6302e-03,\n -6.4482e-04, -2.4608e-03, 1.0183e-03, -2.5275e-03, 2.3567e-04,\n -3.1159e-03, 1.8288e-04, -2.8354e-03, 2.4892e-03, -4.9638e-03,\n -3.8529e-04, 1.3058e-03, -2.1435e-03, -3.4796e-03, 5.8854e-03,\n -8.1148e-04, 3.5425e-03, 3.3004e-03, -4.9591e-04, 1.7259e-03,\n 3.2970e-03, 3.2061e-03, 2.4125e-03, 1.6018e-03, -1.1534e-03,\n -2.8140e-03, -6.8754e-03, 1.2752e-03, 2.8058e-03, -3.0074e-03,\n -2.4535e-03, -2.1798e-03, 6.2365e-04, -4.1338e-03, 5.5330e-03,\n -3.1626e-03, -2.2267e-03, -2.9264e-04, -2.4661e-04, 5.2080e-03,\n -3.6884e-03, -6.1901e-04, 5.1321e-03, -1.0389e-03, -1.1335e-02,\n -2.4950e-03, -1.8596e-03, 6.5044e-03, 3.5547e-03, 6.4808e-04,\n -4.0406e-05, -2.8028e-03, -2.6312e-03, 1.2551e-03, 4.3578e-03,\n 2.2895e-03, 1.5159e-03, 9.6477e-05, -3.9099e-03, -1.9560e-03,\n 2.4197e-03, -6.1737e-03, 6.2626e-04, 5.7437e-05, -1.4652e-03,\n 2.1790e-03, 1.9018e-03, -4.8499e-03, -3.2068e-03, 4.0980e-05,\n 1.1575e-03, 3.3596e-03, -1.1029e-03, 1.9664e-03, -3.6771e-03,\n 1.0035e-03, -6.8754e-03, 7.0768e-03, 3.6226e-03, -1.2421e-03,\n 6.2435e-03, -1.7167e-03, -4.4202e-03, 6.2273e-04, 1.5522e-03,\n -5.7137e-03, 5.3323e-03, 1.3951e-04, -1.8772e-03, -1.5273e-03,\n -2.5821e-03, -5.2467e-03, 1.1413e-03, 3.3479e-03, -8.2672e-03,\n 5.4231e-04, 8.9307e-04, 2.3667e-03, 8.7038e-04, -3.9045e-03,\n 3.9032e-03, -2.0981e-03, -1.5370e-03, -4.0267e-03, -5.6557e-03,\n -1.4134e-03, -1.4256e-03, 1.2149e-03, -1.9334e-03, -1.1134e-03,\n 1.6291e-03, -6.6504e-03, 1.1090e-03, -1.7354e-03, 4.6700e-03,\n 4.7648e-04, 2.1795e-05, 1.8247e-03, 2.5655e-03, 3.4343e-04,\n 1.1088e-03, -3.5564e-03, -3.3327e-03, -1.2563e-03, 2.8004e-03,\n 4.6647e-04, -6.0435e-03, -4.1074e-03, 3.3701e-04, 6.6762e-03,\n 3.0045e-03, 3.6024e-03, 2.9930e-03, -6.8770e-03, 4.7142e-03,\n -8.6219e-03, 5.2213e-04, 4.7256e-03, -1.6607e-03, 2.5546e-03,\n 4.4569e-03, -4.6768e-03, -5.6664e-04, -2.2962e-03, -7.4776e-03,\n 4.7323e-03, -3.6221e-03, -2.7448e-03, 2.4937e-03, 8.4287e-04,\n -2.7587e-03, 6.4463e-04, 5.7417e-03, 1.5572e-03, -2.6696e-03,\n -2.8353e-03, 5.3177e-04, -2.2475e-03, -4.7813e-03, -3.4979e-03,\n 2.3578e-03, -5.2859e-03, 1.5669e-03, -2.5137e-04, 1.1949e-03,\n 2.1902e-03, -1.3697e-03, 3.7363e-04, -1.7370e-03, 5.1196e-03,\n -7.3674e-03, -3.5217e-03, -2.7973e-03, 4.4318e-03, -9.9293e-04,\n 2.5264e-03, 2.5463e-03, 3.3606e-03, 2.0512e-03, -1.2739e-03,\n 3.0510e-03, 5.1831e-03, 4.1393e-03, -5.1023e-04, -4.1901e-03,\n -2.0656e-03, 1.1089e-03, 3.3400e-03, 3.9137e-03, -1.1080e-03,\n 2.7010e-03, 1.3710e-03, -1.0401e-02, 4.6836e-03, 8.5274e-03,\n -2.2581e-03, 1.7976e-03, -3.2544e-03, 3.1592e-03, 4.9433e-03,\n -3.9480e-03, 4.9504e-03, 1.4423e-03, -2.4687e-03, -3.2660e-03,\n 1.8635e-03, 8.7358e-04, 2.5336e-03, 3.0099e-03, -4.8321e-04,\n 1.1848e-03, -3.2318e-03, 2.8226e-03, -2.2348e-03, 5.2193e-03,\n -1.7592e-03, 1.0780e-03, -5.5960e-04, 1.3754e-03, 4.1106e-04,\n 1.2537e-03, 1.5410e-03, -1.4493e-03, -2.8969e-03, -5.6393e-03,\n 2.9271e-03, 3.2870e-03, -2.2273e-03, -5.3732e-03, -4.0142e-03,\n 3.8634e-03, 2.2634e-03, -3.2369e-03, 3.6193e-03, 2.8434e-03,\n -1.1309e-02, 1.7537e-03, 4.0685e-03, -2.7759e-03, -2.3507e-03,\n -1.8386e-03, -2.3843e-03, -2.1136e-03, 5.8385e-03, 3.2306e-03,\n 1.4916e-03, -2.6341e-04, 2.0013e-03, 1.7156e-03, 2.0737e-03,\n 5.1980e-04, -3.7068e-03, 5.1941e-03, 3.3687e-03, -2.8455e-03,\n 2.1087e-03, -7.8757e-03, -1.6158e-03, 1.8677e-03, -4.4176e-03,\n 3.7034e-03, -2.1691e-03, 1.7772e-03, 6.8871e-04, -6.0094e-03,\n -7.5394e-03, -1.0820e-02, 4.2749e-04, 6.8343e-04, 1.4281e-03,\n -4.5942e-03, 7.5448e-04, -4.0341e-03, -2.5095e-03, 1.0617e-03,\n 2.3168e-03, 8.5658e-04, 8.4911e-03, -1.6406e-03, -1.5598e-03,\n 1.5236e-03, -2.9797e-03, -1.5023e-03, -3.0125e-03, 1.2939e-03,\n 1.5192e-03, -3.5616e-03, 1.5143e-03, -1.3709e-03, -9.6589e-05,\n -2.6987e-03, 2.2411e-03, -1.8494e-03, -1.0688e-03, 6.3826e-03,\n -4.3016e-03, -6.3558e-04, 8.4495e-03, -1.6208e-04, -1.1405e-04,\n 2.3365e-03, 1.1899e-03, -5.4177e-04, -5.0964e-03, -2.7411e-03,\n 1.5369e-03, 1.3951e-03, 3.8729e-03, 4.6911e-04, -1.3028e-03,\n 2.3521e-03, 6.3279e-04, 1.5088e-03, 9.6679e-04, -1.7749e-03,\n -1.1800e-03, 1.8554e-03, 3.4925e-03, -1.8911e-03, -6.3263e-03,\n -4.0906e-03, -1.0375e-02, 3.8335e-03, -2.1418e-03, -4.0777e-03,\n 7.1894e-04, -3.9432e-04, 1.2084e-03, 3.8937e-03, -2.7151e-03,\n -2.9979e-03, 1.4944e-03, 1.8224e-03, -4.4652e-03, -1.3527e-03,\n -7.0123e-04, 9.5354e-05, 2.6658e-03, 3.4359e-03, -2.8820e-03,\n 2.1304e-03, 3.6645e-03, 7.9296e-04, 1.0635e-03, -1.4484e-03,\n -9.0268e-04, -8.2662e-04, -3.6925e-03, 1.7863e-03, -1.0300e-03,\n -1.3142e-04, 1.1266e-03, -2.7553e-03, 2.9889e-03, 3.5096e-03,\n 2.8093e-03, 5.0705e-03, 2.5980e-03, -2.6944e-03, 1.5217e-03,\n 3.8053e-03, -1.6345e-03, 5.8273e-04, -4.0953e-03, -4.8256e-04,\n -1.4093e-03, -2.9648e-03, -3.2487e-03, 1.5495e-03, -1.1704e-02,\n 9.7374e-04, -5.3015e-03, 6.6485e-03, -8.7173e-03, 2.7846e-03,\n 1.0280e-03, 2.2158e-04, -7.6407e-03, 1.9642e-03, 6.4721e-04,\n 4.8783e-04, 3.5163e-03, 2.7594e-03, -2.9732e-03, 3.0057e-03,\n 1.4152e-04, 1.2253e-03, 3.1639e-03, -1.4867e-03, 3.0506e-03,\n -1.9196e-03, 1.8687e-03, 5.6044e-03, 1.5031e-04, 5.2659e-04,\n 5.1592e-04, -8.4125e-04, -5.4459e-04, -1.1591e-03, 6.1041e-04,\n 1.5060e-03, -3.5486e-03, -4.7872e-04, -1.0735e-03, 9.6822e-04,\n -9.6781e-04, -2.7638e-03, -3.8716e-03, 5.9930e-03, -9.4325e-04,\n 4.7475e-03, 1.2431e-03, -6.8248e-04, -2.1609e-03, -2.0041e-03,\n -2.1217e-04, 2.6483e-03, 3.8704e-03, -2.3065e-03, 2.1744e-03,\n 7.0854e-03, 5.3041e-04, 4.9846e-03, 2.2578e-03, 3.3766e-03,\n 7.9863e-04, 1.2011e-03, 2.4197e-03, 8.2233e-04, -7.0744e-03,\n 2.9096e-03, -4.3992e-03, 4.5149e-03, 7.3246e-04, -9.9619e-03,\n 1.2951e-03, 1.9099e-03, 5.8394e-03, 1.8489e-03, -1.5571e-03,\n 4.1167e-03, 3.0760e-03], device='cuda:0')",
|
| 23 |
+
"exp_avg_sq": "tensor([1.3854e-04, 1.3112e-04, 9.0897e-05, 1.2028e-04, 6.9633e-05, 9.5087e-05,\n 8.8874e-05, 9.3115e-05, 9.6932e-05, 1.2965e-04, 9.7748e-05, 1.4495e-04,\n 7.3863e-05, 8.6330e-05, 6.5287e-05, 7.7820e-05, 1.8193e-04, 9.3712e-05,\n 8.5449e-05, 1.0419e-04, 1.1173e-04, 1.3041e-04, 1.2279e-04, 8.6422e-05,\n 1.3657e-04, 1.3103e-04, 1.0462e-04, 9.1728e-05, 7.8435e-05, 1.1320e-04,\n 9.3317e-05, 1.2315e-04, 1.6848e-04, 7.6684e-05, 1.5697e-04, 1.3339e-04,\n 9.5848e-05, 1.3281e-04, 6.6556e-05, 9.5286e-05, 1.0139e-04, 9.4564e-05,\n 8.0919e-05, 9.7228e-05, 1.0644e-04, 8.8391e-05, 1.8929e-04, 1.1962e-04,\n 1.0498e-04, 1.0193e-04, 1.0363e-04, 1.3614e-04, 9.0660e-05, 9.8889e-05,\n 1.3357e-04, 1.4165e-04, 1.0996e-04, 1.1747e-04, 9.3784e-05, 1.1671e-04,\n 1.3036e-04, 8.2134e-05, 1.2632e-04, 1.3762e-04, 1.2460e-04, 9.9765e-05,\n 1.0898e-04, 1.1249e-04, 1.1821e-04, 1.3803e-04, 1.0645e-04, 8.2570e-05,\n 1.2592e-04, 7.5925e-05, 1.1948e-04, 1.4978e-04, 1.1912e-04, 1.0140e-04,\n 1.0864e-04, 1.6708e-04, 1.3053e-04, 9.0424e-05, 1.3856e-04, 1.2253e-04,\n 9.4405e-05, 9.9471e-05, 1.4768e-04, 1.4584e-04, 7.7769e-05, 1.5028e-04,\n 1.2489e-04, 1.3265e-04, 1.0510e-04, 8.4669e-05, 1.1634e-04, 9.2667e-05,\n 1.5178e-04, 1.1207e-04, 1.0643e-04, 8.9648e-05, 6.1620e-05, 8.1117e-05,\n 8.1269e-05, 9.8977e-05, 1.3288e-04, 1.0531e-04, 1.1421e-04, 1.1214e-04,\n 1.1662e-04, 9.2517e-05, 1.1637e-04, 9.3102e-05, 9.8305e-05, 7.8700e-05,\n 1.3013e-04, 8.2180e-05, 6.8687e-05, 1.2851e-04, 9.9181e-05, 8.7724e-05,\n 7.8987e-05, 8.0986e-05, 1.0464e-04, 1.1376e-04, 1.2181e-04, 8.5134e-05,\n 8.8810e-05, 1.0001e-04, 9.6391e-05, 1.1506e-04, 9.0210e-05, 1.0450e-04,\n 1.3824e-04, 8.7078e-05, 1.0704e-04, 9.8977e-05, 8.6453e-05, 9.7262e-05,\n 1.0356e-04, 1.1586e-04, 1.4150e-04, 1.0684e-04, 7.2580e-05, 1.4149e-04,\n 1.0854e-04, 1.1204e-04, 1.0605e-04, 1.0952e-04, 1.0661e-04, 1.2876e-04,\n 6.1340e-05, 1.1922e-04, 7.1335e-05, 1.1616e-04, 7.7629e-05, 1.2707e-04,\n 1.1608e-04, 1.1006e-04, 1.4890e-04, 7.7412e-05, 1.2091e-04, 8.6633e-05,\n 1.4280e-04, 8.3843e-05, 1.4835e-04, 7.5311e-05, 1.1250e-04, 1.0451e-04,\n 1.0406e-04, 1.0633e-04, 1.6868e-04, 1.0075e-04, 1.1682e-04, 8.0041e-05,\n 1.0827e-04, 7.4310e-05, 8.9498e-05, 1.2029e-04, 7.5654e-05, 7.8913e-05,\n 9.4131e-05, 1.7754e-04, 1.2469e-04, 1.0790e-04, 1.3197e-04, 1.0605e-04,\n 1.1549e-04, 1.3629e-04, 8.4059e-05, 1.0088e-04, 1.1281e-04, 1.1201e-04,\n 1.6530e-04, 9.1472e-05, 1.0708e-04, 9.0065e-05, 7.4927e-05, 1.0255e-04,\n 8.4244e-05, 1.4751e-04, 1.0680e-04, 1.3152e-04, 6.1270e-05, 1.0468e-04,\n 1.0777e-04, 1.0087e-04, 1.0852e-04, 9.5905e-05, 9.9333e-05, 1.2101e-04,\n 8.9669e-05, 1.4260e-04, 1.3264e-04, 1.4552e-04, 1.4398e-04, 8.4717e-05,\n 1.2116e-04, 6.5953e-05, 8.3157e-05, 1.1357e-04, 9.8613e-05, 9.1962e-05,\n 1.0922e-04, 1.4024e-04, 1.0255e-04, 1.0864e-04, 9.2608e-05, 8.8273e-05,\n 1.5127e-04, 1.7476e-04, 9.2935e-05, 9.1212e-05, 9.7338e-05, 8.6986e-05,\n 1.0687e-04, 1.3784e-04, 1.2105e-04, 1.4491e-04, 1.2124e-04, 1.4883e-04,\n 9.3174e-05, 1.0133e-04, 1.3710e-04, 9.3716e-05, 1.2418e-04, 1.0757e-04,\n 8.8832e-05, 6.2666e-05, 1.0978e-04, 1.2446e-04, 8.6885e-05, 1.1525e-04,\n 9.3031e-05, 1.6713e-04, 1.0444e-04, 8.2183e-05, 2.3794e-04, 1.2176e-04,\n 1.1940e-04, 1.2292e-04, 9.1186e-05, 1.2601e-04, 1.1783e-04, 1.1165e-04,\n 1.3905e-04, 1.0272e-04, 8.1296e-05, 1.0314e-04, 9.3496e-05, 9.4301e-05,\n 9.0150e-05, 6.7371e-05, 8.7926e-05, 7.6902e-05, 8.7602e-05, 1.7743e-04,\n 1.0890e-04, 1.1388e-04, 1.5793e-04, 1.1296e-04, 6.5005e-05, 1.4170e-04,\n 1.0831e-04, 9.7242e-05, 1.4707e-04, 8.6157e-05, 1.6355e-04, 7.4700e-05,\n 1.0787e-04, 9.2319e-05, 1.2743e-04, 1.2023e-04, 1.2227e-04, 9.9926e-05,\n 1.6205e-04, 1.0556e-04, 1.0429e-04, 1.5327e-04, 1.1656e-04, 1.3935e-04,\n 1.4777e-04, 1.0085e-04, 9.4279e-05, 1.5449e-04, 8.9503e-05, 1.2114e-04,\n 1.0170e-04, 8.1910e-05, 1.2901e-04, 9.8399e-05, 9.2381e-05, 8.6548e-05,\n 7.8436e-05, 8.3059e-05, 9.2269e-05, 1.2143e-04, 7.1706e-05, 1.2080e-04,\n 8.5964e-05, 1.1090e-04, 7.7520e-05, 1.6363e-04, 1.1304e-04, 9.9559e-05,\n 8.3627e-05, 9.7476e-05, 9.3955e-05, 1.2029e-04, 5.7565e-05, 9.4592e-05,\n 1.1278e-04, 1.3364e-04, 8.5296e-05, 9.0660e-05, 1.3480e-04, 1.0774e-04,\n 1.0003e-04, 1.3573e-04, 1.2295e-04, 9.7754e-05, 1.0403e-04, 1.2579e-04,\n 9.1009e-05, 9.6402e-05, 9.4061e-05, 9.2652e-05, 8.8274e-05, 8.7655e-05,\n 1.3650e-04, 5.2599e-05, 1.2332e-04, 1.1072e-04, 5.7206e-05, 1.1303e-04,\n 1.0695e-04, 9.5614e-05, 7.9728e-05, 8.1891e-05, 1.3405e-04, 7.6508e-05,\n 9.7079e-05, 6.2229e-05, 8.7777e-05, 8.5165e-05, 6.1486e-05, 1.1138e-04,\n 6.9365e-05, 1.7148e-04, 8.1756e-05, 8.7423e-05, 1.1709e-04, 1.6986e-04,\n 1.1976e-04, 1.2052e-04, 8.8718e-05, 9.6394e-05, 9.2979e-05, 8.9997e-05,\n 9.9164e-05, 9.8883e-05, 6.6943e-05, 9.3284e-05, 1.1222e-04, 1.2680e-04,\n 1.0226e-04, 6.5115e-05, 8.0855e-05, 8.5276e-05, 8.8865e-05, 1.3958e-04,\n 1.1188e-04, 1.0242e-04, 1.0931e-04, 9.1211e-05, 1.0370e-04, 1.2046e-04,\n 1.1005e-04, 1.1490e-04, 1.2616e-04, 9.5147e-05, 8.3996e-05, 1.1140e-04,\n 9.3521e-05, 1.0499e-04, 7.7854e-05, 1.1276e-04, 1.0141e-04, 8.1519e-05,\n 1.0480e-04, 1.2605e-04, 1.0957e-04, 8.0392e-05, 1.2528e-04, 1.5315e-04,\n 9.4738e-05, 1.2412e-04, 7.8682e-05, 7.7901e-05, 1.2034e-04, 1.2122e-04,\n 1.2583e-04, 8.5852e-05, 1.0613e-04, 1.1352e-04, 1.2689e-04, 9.6879e-05,\n 1.2389e-04, 9.7920e-05, 1.0203e-04, 1.0051e-04, 8.9728e-05, 1.3428e-04,\n 1.1388e-04, 1.9116e-04, 9.2480e-05, 1.4214e-04, 7.8615e-05, 1.3025e-04,\n 1.1047e-04, 1.1000e-04, 1.2374e-04, 1.4933e-04, 9.8642e-05, 1.1048e-04,\n 9.0698e-05, 7.6811e-05, 1.0695e-04, 9.6879e-05, 1.0350e-04, 1.3360e-04,\n 1.1894e-04, 1.1410e-04, 6.9174e-05, 1.3090e-04, 1.2925e-04, 1.3253e-04,\n 7.6813e-05, 1.4133e-04, 1.0971e-04, 1.0414e-04, 1.4161e-04, 1.2224e-04,\n 1.2178e-04, 8.7643e-05, 1.1019e-04, 1.2337e-04, 7.7850e-05, 7.2689e-05,\n 9.3432e-05, 8.5854e-05, 8.7800e-05, 1.0402e-04, 1.0670e-04, 1.2624e-04,\n 9.3457e-05, 1.0912e-04, 9.1249e-05, 1.0191e-04, 6.4175e-05, 8.7674e-05,\n 8.4831e-05, 7.1827e-05, 1.1152e-04, 9.8758e-05, 7.5863e-05, 1.1259e-04,\n 7.3176e-05, 9.3988e-05, 1.3702e-04, 1.0302e-04, 2.3208e-04, 1.0289e-04,\n 1.0858e-04, 8.7788e-05, 1.2672e-04, 1.4726e-04, 1.1936e-04, 1.3589e-04,\n 8.8409e-05, 1.2093e-04, 1.0248e-04, 9.8048e-05, 1.1695e-04, 9.0827e-05,\n 1.2963e-04, 1.4888e-04, 1.1285e-04, 9.0159e-05, 1.0786e-04, 1.1430e-04,\n 1.2962e-04, 9.0194e-05], device='cuda:0')"
|
| 24 |
+
},
|
| 25 |
+
"4": {
|
| 26 |
+
"step": "tensor(5008.)",
|
| 27 |
+
"exp_avg": "tensor([[-3.3437e-05, -2.1454e-04, 2.6428e-05, ..., -4.3370e-04,\n 7.4132e-05, -5.1018e-05],\n [-1.6346e-04, -1.3189e-04, -1.4464e-04, ..., 6.0752e-05,\n -3.9591e-04, 6.4749e-05],\n [ 3.7050e-04, -3.6265e-05, -2.5236e-05, ..., -2.3121e-04,\n 2.6310e-04, 7.8634e-05],\n ...,\n [-2.3939e-05, 1.5839e-04, 2.2052e-05, ..., -3.4691e-04,\n 9.2907e-04, -7.2713e-04],\n [-2.3775e-04, -5.5225e-05, 9.6148e-05, ..., 2.9180e-04,\n -3.2707e-04, 1.7867e-04],\n [-2.3452e-04, 1.6892e-04, -4.5868e-05, ..., 5.1117e-04,\n 2.9470e-04, -2.5631e-04]], device='cuda:0')",
|
| 28 |
+
"exp_avg_sq": "tensor([[4.7784e-07, 6.8901e-07, 2.3416e-07, ..., 3.8408e-07, 4.9863e-07,\n 6.2447e-07],\n [8.9839e-07, 1.5923e-06, 7.3401e-07, ..., 1.1550e-06, 1.1921e-06,\n 1.4621e-06],\n [9.6123e-07, 6.4338e-07, 6.2835e-07, ..., 1.1700e-06, 8.2420e-07,\n 1.0937e-06],\n ...,\n [8.3337e-07, 1.6742e-06, 8.6301e-07, ..., 1.0382e-06, 1.4441e-06,\n 1.1824e-06],\n [7.5329e-07, 1.2938e-06, 4.9955e-07, ..., 1.3572e-06, 1.0167e-06,\n 1.0448e-06],\n [7.7859e-07, 9.2573e-07, 6.5775e-07, ..., 1.8798e-06, 8.1790e-07,\n 1.3136e-06]], device='cuda:0')"
|
| 29 |
+
},
|
| 30 |
+
"5": {
|
| 31 |
+
"step": "tensor(5008.)",
|
| 32 |
+
"exp_avg": "tensor([[ 2.6256e-05, -3.7534e-05, 2.0893e-05, ..., -2.3458e-04,\n -4.2165e-05, -9.5247e-05],\n [-5.0779e-05, -1.3112e-04, 6.3805e-05, ..., 7.4720e-05,\n -7.0062e-05, 1.4386e-04],\n [ 1.6429e-04, -4.1389e-05, 4.7424e-05, ..., -3.4594e-04,\n 1.5033e-04, 8.5005e-05],\n ...,\n [-1.0397e-04, -5.2253e-05, 6.7666e-05, ..., -6.7078e-05,\n 2.6366e-04, -1.0518e-04],\n [ 1.2352e-04, -1.3457e-04, 5.8187e-05, ..., 1.0859e-04,\n 1.7479e-04, -4.0825e-04],\n [-1.2678e-04, -4.4832e-05, -9.7806e-05, ..., 4.1502e-04,\n -9.8700e-05, 1.5175e-04]], device='cuda:0')",
|
| 33 |
+
"exp_avg_sq": "tensor([[1.3910e-07, 1.2612e-07, 1.3261e-07, ..., 1.6240e-07, 1.4769e-07,\n 1.8049e-07],\n [2.7921e-07, 5.4683e-07, 2.7471e-07, ..., 2.6795e-07, 3.3320e-07,\n 4.1752e-07],\n [3.2733e-07, 2.3743e-07, 2.6710e-07, ..., 4.7971e-07, 4.2869e-07,\n 3.5024e-07],\n ...,\n [2.8947e-07, 6.9914e-07, 1.7282e-07, ..., 5.9552e-07, 4.5774e-07,\n 3.7736e-07],\n [3.3837e-07, 4.5033e-07, 1.4194e-07, ..., 4.7474e-07, 3.5595e-07,\n 3.4965e-07],\n [3.0937e-07, 2.1560e-07, 5.7631e-07, ..., 3.3268e-07, 2.9232e-07,\n 4.2349e-07]], device='cuda:0')"
|
| 34 |
+
},
|
| 35 |
+
"6": {
|
| 36 |
+
"step": "tensor(5008.)",
|
| 37 |
+
"exp_avg": "tensor([ 0.0008, -0.0008], device='cuda:0')",
|
| 38 |
+
"exp_avg_sq": "tensor([1.0308e-05, 1.0308e-05], device='cuda:0')"
|
| 39 |
+
}
|
| 40 |
+
},
|
| 41 |
+
"param_groups": [
|
| 42 |
+
{
|
| 43 |
+
"lr": 0.0006548539886902864,
|
| 44 |
+
"name": "shared",
|
| 45 |
+
"betas": [
|
| 46 |
+
0.9,
|
| 47 |
+
0.999
|
| 48 |
+
],
|
| 49 |
+
"eps": 1e-08,
|
| 50 |
+
"weight_decay": 1e-05,
|
| 51 |
+
"amsgrad": false,
|
| 52 |
+
"maximize": false,
|
| 53 |
+
"foreach": null,
|
| 54 |
+
"capturable": false,
|
| 55 |
+
"differentiable": false,
|
| 56 |
+
"fused": null,
|
| 57 |
+
"decoupled_weight_decay": true,
|
| 58 |
+
"initial_lr": 0.001,
|
| 59 |
+
"params": [
|
| 60 |
+
0,
|
| 61 |
+
1,
|
| 62 |
+
2,
|
| 63 |
+
3
|
| 64 |
+
]
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"lr": 0.0006548539886902864,
|
| 68 |
+
"name": "scale_256",
|
| 69 |
+
"betas": [
|
| 70 |
+
0.9,
|
| 71 |
+
0.999
|
| 72 |
+
],
|
| 73 |
+
"eps": 1e-08,
|
| 74 |
+
"weight_decay": 1e-05,
|
| 75 |
+
"amsgrad": false,
|
| 76 |
+
"maximize": false,
|
| 77 |
+
"foreach": null,
|
| 78 |
+
"capturable": false,
|
| 79 |
+
"differentiable": false,
|
| 80 |
+
"fused": null,
|
| 81 |
+
"decoupled_weight_decay": true,
|
| 82 |
+
"initial_lr": 0.001,
|
| 83 |
+
"params": [
|
| 84 |
+
4
|
| 85 |
+
]
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"lr": 0.0006548539886902864,
|
| 89 |
+
"name": "scale_512",
|
| 90 |
+
"betas": [
|
| 91 |
+
0.9,
|
| 92 |
+
0.999
|
| 93 |
+
],
|
| 94 |
+
"eps": 1e-08,
|
| 95 |
+
"weight_decay": 1e-05,
|
| 96 |
+
"amsgrad": false,
|
| 97 |
+
"maximize": false,
|
| 98 |
+
"foreach": null,
|
| 99 |
+
"capturable": false,
|
| 100 |
+
"differentiable": false,
|
| 101 |
+
"fused": null,
|
| 102 |
+
"decoupled_weight_decay": true,
|
| 103 |
+
"initial_lr": 0.001,
|
| 104 |
+
"params": [
|
| 105 |
+
5
|
| 106 |
+
]
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"lr": 0.0003275997400965494,
|
| 110 |
+
"name": "fusion",
|
| 111 |
+
"betas": [
|
| 112 |
+
0.9,
|
| 113 |
+
0.999
|
| 114 |
+
],
|
| 115 |
+
"eps": 1e-08,
|
| 116 |
+
"weight_decay": 1e-05,
|
| 117 |
+
"amsgrad": false,
|
| 118 |
+
"maximize": false,
|
| 119 |
+
"foreach": null,
|
| 120 |
+
"capturable": false,
|
| 121 |
+
"differentiable": false,
|
| 122 |
+
"fused": null,
|
| 123 |
+
"decoupled_weight_decay": true,
|
| 124 |
+
"initial_lr": 0.0005,
|
| 125 |
+
"params": [
|
| 126 |
+
6
|
| 127 |
+
]
|
| 128 |
+
}
|
| 129 |
+
]
|
| 130 |
+
},
|
| 131 |
+
"scheduler_state_dict": {
|
| 132 |
+
"T_0": 10,
|
| 133 |
+
"T_i": 10,
|
| 134 |
+
"T_mult": 2,
|
| 135 |
+
"eta_min": 1e-06,
|
| 136 |
+
"T_cur": 4,
|
| 137 |
+
"base_lrs": [
|
| 138 |
+
0.001,
|
| 139 |
+
0.001,
|
| 140 |
+
0.001,
|
| 141 |
+
0.0005
|
| 142 |
+
],
|
| 143 |
+
"last_epoch": 4,
|
| 144 |
+
"_step_count": 0,
|
| 145 |
+
"_is_initial": false,
|
| 146 |
+
"_get_lr_called_within_step": false,
|
| 147 |
+
"_last_lr": [
|
| 148 |
+
0.0006548539886902864,
|
| 149 |
+
0.0006548539886902864,
|
| 150 |
+
0.0006548539886902864,
|
| 151 |
+
0.0003275997400965494
|
| 152 |
+
]
|
| 153 |
+
},
|
| 154 |
+
"metrics": {
|
| 155 |
+
"best_val_acc": 70.622,
|
| 156 |
+
"best_epoch": 3,
|
| 157 |
+
"scale_accuracies": {
|
| 158 |
+
"256": 69.996,
|
| 159 |
+
"512": 70.482
|
| 160 |
+
},
|
| 161 |
+
"training_history": {
|
| 162 |
+
"epochs": [
|
| 163 |
+
1,
|
| 164 |
+
2,
|
| 165 |
+
3,
|
| 166 |
+
4
|
| 167 |
+
],
|
| 168 |
+
"train_loss": [
|
| 169 |
+
5.603825211905824,
|
| 170 |
+
4.145695094292918,
|
| 171 |
+
3.767625377201044,
|
| 172 |
+
3.572572173021091
|
| 173 |
+
],
|
| 174 |
+
"train_acc": [
|
| 175 |
+
63.39829233815732,
|
| 176 |
+
69.48282308239285,
|
| 177 |
+
70.34266414917025,
|
| 178 |
+
70.91120829681064
|
| 179 |
+
],
|
| 180 |
+
"val_acc": [
|
| 181 |
+
68.056,
|
| 182 |
+
69.624,
|
| 183 |
+
70.142,
|
| 184 |
+
70.622
|
| 185 |
+
],
|
| 186 |
+
"scale_accs": {
|
| 187 |
+
"256": [
|
| 188 |
+
66.924,
|
| 189 |
+
68.828,
|
| 190 |
+
69.54,
|
| 191 |
+
69.996
|
| 192 |
+
],
|
| 193 |
+
"512": [
|
| 194 |
+
67.882,
|
| 195 |
+
69.572,
|
| 196 |
+
70.028,
|
| 197 |
+
70.482
|
| 198 |
+
]
|
| 199 |
+
},
|
| 200 |
+
"lr": [
|
| 201 |
+
0.0009755527298894294,
|
| 202 |
+
0.0009046039886902864,
|
| 203 |
+
0.0007940987335200904,
|
| 204 |
+
0.0006548539886902864
|
| 205 |
+
]
|
| 206 |
+
}
|
| 207 |
+
},
|
| 208 |
+
"train_config": {
|
| 209 |
+
"name": "david_training",
|
| 210 |
+
"run_id": "20251012_135249",
|
| 211 |
+
"dataset_name": "AbstractPhil/imagenet-clip-features-orderly",
|
| 212 |
+
"model_variant": "clip_vit_laion_b32",
|
| 213 |
+
"num_classes": 1000,
|
| 214 |
+
"preset": "small_fast",
|
| 215 |
+
"custom_config_path": null,
|
| 216 |
+
"num_classes_override": null,
|
| 217 |
+
"use_belly_override": null,
|
| 218 |
+
"belly_expand_override": null,
|
| 219 |
+
"progressive_training_override": false,
|
| 220 |
+
"scale_warmup_epochs_override": null,
|
| 221 |
+
"num_epochs": 10,
|
| 222 |
+
"batch_size": 1024,
|
| 223 |
+
"learning_rate": 0.001,
|
| 224 |
+
"weight_decay": 1e-05,
|
| 225 |
+
"warmup_epochs": 3,
|
| 226 |
+
"use_rose_loss": true,
|
| 227 |
+
"rose_initial_weight": 0.1,
|
| 228 |
+
"rose_max_weight": 0.5,
|
| 229 |
+
"rose_weight_schedule": "adaptive",
|
| 230 |
+
"use_cayley_loss": false,
|
| 231 |
+
"cayley_weight": 0.001,
|
| 232 |
+
"scale_loss_balance": null,
|
| 233 |
+
"use_mixed_precision": false,
|
| 234 |
+
"gradient_clip": 10.0,
|
| 235 |
+
"scheduler_type": "cosine_restarts",
|
| 236 |
+
"min_lr": 1e-06,
|
| 237 |
+
"freeze_strategy": "never",
|
| 238 |
+
"freeze_threshold": 90.0,
|
| 239 |
+
"unfreeze_on_plateau": true,
|
| 240 |
+
"patience": 10,
|
| 241 |
+
"track_gradients": true,
|
| 242 |
+
"gradient_scale_threshold": 1e-05,
|
| 243 |
+
"gradient_scale_multiplier": 10.0,
|
| 244 |
+
"log_interval": 50,
|
| 245 |
+
"val_interval": 1,
|
| 246 |
+
"save_interval": 5,
|
| 247 |
+
"log_fusion_weights": true,
|
| 248 |
+
"log_loss_components": true,
|
| 249 |
+
"save_format": "safetensors",
|
| 250 |
+
"hf_repo": "AbstractPhil/gated-david",
|
| 251 |
+
"upload_to_hub": true,
|
| 252 |
+
"base_dir": "./david_training",
|
| 253 |
+
"num_workers": 10,
|
| 254 |
+
"pin_memory": true,
|
| 255 |
+
"prefetch_factor": 4,
|
| 256 |
+
"persistent_workers": true
|
| 257 |
+
}
|
| 258 |
+
}
|
weights/David-fully_shared-weighted_sum/20251012_135249/training_history.json
CHANGED
|
@@ -2,38 +2,45 @@
|
|
| 2 |
"epochs": [
|
| 3 |
1,
|
| 4 |
2,
|
| 5 |
-
3
|
|
|
|
| 6 |
],
|
| 7 |
"train_loss": [
|
| 8 |
5.603825211905824,
|
| 9 |
4.145695094292918,
|
| 10 |
-
3.767625377201044
|
|
|
|
| 11 |
],
|
| 12 |
"train_acc": [
|
| 13 |
63.39829233815732,
|
| 14 |
69.48282308239285,
|
| 15 |
-
70.34266414917025
|
|
|
|
| 16 |
],
|
| 17 |
"val_acc": [
|
| 18 |
68.056,
|
| 19 |
69.624,
|
| 20 |
-
70.142
|
|
|
|
| 21 |
],
|
| 22 |
"scale_accs": {
|
| 23 |
"256": [
|
| 24 |
66.924,
|
| 25 |
68.828,
|
| 26 |
-
69.54
|
|
|
|
| 27 |
],
|
| 28 |
"512": [
|
| 29 |
67.882,
|
| 30 |
69.572,
|
| 31 |
-
70.028
|
|
|
|
| 32 |
]
|
| 33 |
},
|
| 34 |
"lr": [
|
| 35 |
0.0009755527298894294,
|
| 36 |
0.0009046039886902864,
|
| 37 |
-
0.0007940987335200904
|
|
|
|
| 38 |
]
|
| 39 |
}
|
|
|
|
| 2 |
"epochs": [
|
| 3 |
1,
|
| 4 |
2,
|
| 5 |
+
3,
|
| 6 |
+
4
|
| 7 |
],
|
| 8 |
"train_loss": [
|
| 9 |
5.603825211905824,
|
| 10 |
4.145695094292918,
|
| 11 |
+
3.767625377201044,
|
| 12 |
+
3.572572173021091
|
| 13 |
],
|
| 14 |
"train_acc": [
|
| 15 |
63.39829233815732,
|
| 16 |
69.48282308239285,
|
| 17 |
+
70.34266414917025,
|
| 18 |
+
70.91120829681064
|
| 19 |
],
|
| 20 |
"val_acc": [
|
| 21 |
68.056,
|
| 22 |
69.624,
|
| 23 |
+
70.142,
|
| 24 |
+
70.622
|
| 25 |
],
|
| 26 |
"scale_accs": {
|
| 27 |
"256": [
|
| 28 |
66.924,
|
| 29 |
68.828,
|
| 30 |
+
69.54,
|
| 31 |
+
69.996
|
| 32 |
],
|
| 33 |
"512": [
|
| 34 |
67.882,
|
| 35 |
69.572,
|
| 36 |
+
70.028,
|
| 37 |
+
70.482
|
| 38 |
]
|
| 39 |
},
|
| 40 |
"lr": [
|
| 41 |
0.0009755527298894294,
|
| 42 |
0.0009046039886902864,
|
| 43 |
+
0.0007940987335200904,
|
| 44 |
+
0.0006548539886902864
|
| 45 |
]
|
| 46 |
}
|