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_acc71.77.safetensors +3 -0
- weights/David-fully_shared-weighted_sum/20251012_135249/best_model_acc71.77_metadata.json +279 -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: 71.
|
| 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:
|
| 19 |
-
Best Validation Accuracy: 71.
|
| 20 |
|
| 21 |
Per-Scale Accuracies:
|
| 22 |
-
• Scale 256: 71.
|
| 23 |
-
• Scale 512: 71.
|
| 24 |
|
| 25 |
═══════════════════════════════════════════════════════════════
|
| 26 |
|
|
@@ -44,17 +44,18 @@ Epoch | Train Acc | Val Acc | Learning Rate
|
|
| 44 |
3 | 70.34% | 70.14% | 7.94e-04
|
| 45 |
4 | 70.91% | 70.62% | 6.55e-04
|
| 46 |
5 | 71.36% | 70.97% | 5.01e-04
|
| 47 |
-
6 | 71.84% | 71.46%
|
|
|
|
| 48 |
|
| 49 |
═══════════════════════════════════════════════════════════════
|
| 50 |
|
| 51 |
📁 FILES
|
| 52 |
|
| 53 |
-
Best Model: best_model_acc71.
|
| 54 |
Config: david_config.json
|
| 55 |
Training Cfg: train_config.json
|
| 56 |
History: training_history.json
|
| 57 |
|
| 58 |
═══════════════════════════════════════════════════════════════
|
| 59 |
|
| 60 |
-
Generated: 2025-10-12 14:
|
|
|
|
| 3 |
║ DAVID MODEL SUMMARY ║
|
| 4 |
╠══════════════════════════════════════════════════════════════╣
|
| 5 |
║ ║
|
| 6 |
+
║ 🎯 VALIDATION ACCURACY: 71.77% ║
|
| 7 |
║ ║
|
| 8 |
╚══════════════════════════════════════════════════════════════╝
|
| 9 |
|
| 10 |
MODEL: David-fully_shared-weighted_sum
|
| 11 |
RUN ID: 20251012_135249
|
| 12 |
+
BEST EPOCH: 7/10
|
| 13 |
|
| 14 |
═══════════════════════════════════════════════════════════════
|
| 15 |
|
| 16 |
📊 PERFORMANCE BREAKDOWN
|
| 17 |
|
| 18 |
+
Final Training Accuracy: 72.29%
|
| 19 |
+
Best Validation Accuracy: 71.77%
|
| 20 |
|
| 21 |
Per-Scale Accuracies:
|
| 22 |
+
• Scale 256: 71.44%
|
| 23 |
+
• Scale 512: 71.58%
|
| 24 |
|
| 25 |
═══════════════════════════════════════════════════════════════
|
| 26 |
|
|
|
|
| 44 |
3 | 70.34% | 70.14% | 7.94e-04
|
| 45 |
4 | 70.91% | 70.62% | 6.55e-04
|
| 46 |
5 | 71.36% | 70.97% | 5.01e-04
|
| 47 |
+
6 | 71.84% | 71.46% | 3.46e-04
|
| 48 |
+
7 | 72.29% | 71.77% 👑 | 2.07e-04
|
| 49 |
|
| 50 |
═══════════════════════════════════════════════════════════════
|
| 51 |
|
| 52 |
📁 FILES
|
| 53 |
|
| 54 |
+
Best Model: best_model_acc71.77.safetensors
|
| 55 |
Config: david_config.json
|
| 56 |
Training Cfg: train_config.json
|
| 57 |
History: training_history.json
|
| 58 |
|
| 59 |
═══════════════════════════════════════════════════════════════
|
| 60 |
|
| 61 |
+
Generated: 2025-10-12 14:04:40
|
weights/David-fully_shared-weighted_sum/20251012_135249/best_model_acc71.77.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:62a05d6efce969bb7d849b33ecd128dea0861262a11275cbbf332545634ab720
|
| 3 |
+
size 2628344
|
weights/David-fully_shared-weighted_sum/20251012_135249/best_model_acc71.77_metadata.json
ADDED
|
@@ -0,0 +1,279 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 6,
|
| 3 |
+
"optimizer_state_dict": {
|
| 4 |
+
"state": {
|
| 5 |
+
"0": {
|
| 6 |
+
"step": "tensor(8764.)",
|
| 7 |
+
"exp_avg": "tensor([[-7.6595e-04, -4.9711e-04, -5.5689e-04, ..., -1.4285e-04,\n 5.4427e-04, 4.2035e-05],\n [-7.0456e-04, -1.6567e-04, 1.8215e-03, ..., -9.5119e-04,\n -4.2188e-04, -8.3417e-04],\n [-1.5256e-03, 4.9532e-03, 2.1280e-03, ..., -3.1872e-03,\n -1.8238e-03, -3.6144e-03],\n ...,\n [-5.8655e-04, -1.5603e-03, 1.1056e-03, ..., -4.1099e-04,\n 7.4459e-04, 9.9583e-06],\n [ 7.7961e-04, -1.3903e-03, -1.4022e-05, ..., -8.0943e-04,\n 3.3024e-04, -6.8289e-04],\n [-4.3288e-04, 3.5151e-03, -1.4447e-04, ..., 1.3471e-04,\n -3.0080e-04, 2.1605e-04]], device='cuda:0')",
|
| 8 |
+
"exp_avg_sq": "tensor([[6.4434e-06, 3.0938e-05, 1.3574e-05, ..., 6.1814e-06, 2.7227e-06,\n 8.0395e-06],\n [5.4389e-06, 2.4205e-05, 1.6791e-05, ..., 8.0591e-06, 2.4355e-06,\n 9.3272e-06],\n [9.9363e-06, 5.9068e-05, 2.0085e-05, ..., 6.1942e-06, 3.2001e-06,\n 4.6118e-06],\n ...,\n [5.7415e-06, 2.4397e-05, 1.3684e-05, ..., 3.3443e-06, 1.7186e-06,\n 2.6019e-06],\n [7.1204e-06, 4.0752e-05, 1.4644e-05, ..., 6.0187e-06, 3.7244e-06,\n 6.0984e-06],\n [3.8211e-06, 4.7766e-05, 1.1008e-05, ..., 2.9526e-06, 1.9265e-06,\n 2.1131e-06]], device='cuda:0')"
|
| 9 |
+
},
|
| 10 |
+
"1": {
|
| 11 |
+
"step": "tensor(8764.)",
|
| 12 |
+
"exp_avg": "tensor([ 3.4129e-03, -2.1748e-02, -6.2969e-02, -7.0336e-02, -9.3040e-03,\n -4.2341e-02, 6.1060e-03, -6.0284e-03, -6.5409e-03, 5.0280e-03,\n -8.0608e-04, 2.8320e-03, -4.8012e-03, 2.5766e-02, -1.7821e-03,\n 1.3754e-02, 1.0800e-02, -5.0385e-03, 3.2383e-03, 8.3145e-03,\n 3.8559e-04, -1.1162e-02, -7.6500e-03, 1.0833e-02, 1.8650e-02,\n 7.2657e-03, -5.8965e-03, 4.0199e-02, 1.0418e-02, -1.1830e-02,\n 2.0008e-02, -1.6611e-02, 1.5422e-02, 3.5734e-04, -1.9141e-02,\n -8.1431e-03, 9.6104e-03, -7.1815e-02, 9.1679e-03, -5.8955e-05,\n 6.9040e-03, -5.0479e-02, 7.9377e-04, -6.8967e-03, 1.5557e-02,\n 2.8870e-02, 1.4851e-03, 4.0326e-02, -2.5615e-02, -3.0537e-03,\n -1.0807e-03, -1.2854e-02, -1.3871e-02, 2.3684e-02, -1.9655e-02,\n 2.9771e-02, 7.0333e-03, -4.0826e-02, 3.9138e-03, 2.7422e-02,\n 7.1039e-03, -4.8494e-02, 2.6766e-02, -1.9137e-02, 2.7289e-02,\n -2.3498e-03, 7.5504e-03, 2.7413e-03, 2.7953e-02, -6.3863e-04,\n 1.6865e-02, -1.3074e-03, -1.3732e-03, -4.8566e-03, 7.4456e-05,\n 1.1931e-02, -4.1991e-03, 4.5222e-02, 1.6806e-02, 9.7158e-03,\n 7.7849e-03, 9.0878e-03, -4.5577e-02, -2.0019e-02, 1.5981e-02,\n 1.1617e-02, -2.3804e-02, 4.1266e-03, 2.8439e-03, 4.0574e-03,\n -2.4182e-03, -2.5976e-02, -5.7536e-03, -1.7754e-02, 1.8575e-02,\n -1.0940e-02, 1.8711e-02, -3.6701e-02, -2.8768e-03, 1.6666e-02,\n -8.4507e-03, -2.2259e-02, -1.5902e-02, 1.0885e-02, -1.3746e-02,\n -1.8516e-02, 9.0371e-03, 2.6806e-03, -5.4553e-02, 6.3336e-03,\n 1.7602e-02, 1.1243e-02, 2.1121e-02, -1.3681e-04, 1.2942e-02,\n -8.2606e-03, -1.3708e-02, -2.2918e-02, 6.7095e-03, 1.4994e-02,\n 1.6369e-02, -3.2365e-02, 1.5279e-02, 3.0412e-02, 4.7654e-03,\n 4.4673e-02, 2.2589e-03, 4.3447e-02, 1.1978e-02, -1.0441e-02,\n -1.4847e-02, 1.5046e-02, -1.0010e-02, -2.2302e-02, -2.0381e-02,\n 9.6476e-03, 1.0261e-02, -2.1461e-02, 1.2084e-02, 3.6214e-02,\n 4.9647e-03, 1.0752e-02, -1.8195e-02, 2.3935e-02, -4.3972e-02,\n -9.0804e-03, -9.5310e-03, -5.5719e-03, 1.2720e-02, 5.0303e-03,\n 1.1237e-03, 2.0417e-02, 3.3178e-03, 2.3727e-02, -4.7525e-03,\n -1.6239e-03, -1.3632e-02, -8.8860e-03, -6.8417e-03, 1.3366e-02,\n 1.9289e-02, 2.0596e-02, 9.3738e-03, 1.1284e-02, -4.8887e-02,\n -1.4475e-02, -7.8529e-03, -6.7027e-03, 4.8336e-03, -7.8132e-03,\n 9.5977e-03, -4.7805e-03, 5.3266e-03, 4.8743e-02, 5.3873e-04,\n 5.0456e-03, -3.0858e-02, 5.0528e-02, -6.3004e-02, 3.0925e-02,\n 3.1133e-02, -3.2889e-03, 2.4020e-02, -1.4948e-02, -3.1350e-02,\n 2.4013e-03, 8.9293e-03, -2.9007e-03, -1.4674e-02, -7.9688e-03,\n -3.7129e-02, 1.4171e-02, 1.2808e-02, -2.1396e-02, 1.8403e-02,\n 2.4031e-02, -2.8790e-02, 8.8681e-03, -1.2860e-02, -2.4258e-02,\n -1.5860e-02, -1.5131e-04, 2.0620e-02, -3.7955e-03, -3.0998e-02,\n -4.3959e-04, -3.3552e-03, 1.2672e-02, -1.7702e-03, -1.5941e-02,\n 1.3399e-02, -2.6749e-02, -1.5328e-02, -6.8383e-03, 1.2849e-03,\n -4.7108e-03, -2.2116e-02, -3.2365e-02, 3.1105e-02, -2.2760e-03,\n 8.7340e-03, 1.7940e-02, 1.7603e-02, 7.0215e-03, 1.4652e-04,\n -6.4406e-03, -3.9364e-02, 8.9236e-03, -2.0121e-02, -1.2092e-02,\n 2.5855e-03, -1.9984e-02, -7.0807e-03, -4.8816e-03, 5.2982e-03,\n -1.1708e-02, -5.7571e-04, -9.3013e-04, -1.5434e-02, 2.1590e-02,\n 9.2560e-03, 1.8875e-02, -1.0431e-03, -4.1879e-03, -3.1524e-02,\n -3.8941e-03, 1.9654e-02, -1.7718e-02, 1.8732e-02, -3.9222e-02,\n 2.2040e-02, 2.5867e-02, -1.2350e-02, 3.5554e-02, -1.8450e-02,\n -1.4395e-02, 1.6655e-02, 5.3351e-03, 5.8749e-03, -5.3189e-03,\n -7.6923e-04, -5.1199e-02, -3.3351e-02, 1.8172e-02, -2.1025e-03,\n -2.0433e-02, -3.2277e-02, 6.8127e-03, -2.1795e-03, 1.6475e-02,\n 1.9749e-02, 1.5436e-02, 7.2627e-03, 1.2525e-02, -1.3362e-02,\n -3.8278e-02, 3.7212e-02, -9.0869e-04, -4.4911e-03, -6.1861e-04,\n -1.3199e-02, 1.1056e-02, 1.6935e-02, 1.8661e-03, 3.3403e-02,\n -1.0784e-02, 1.6579e-02, 3.3286e-03, 4.3913e-02, 6.1480e-03,\n 3.2718e-03, 5.1104e-02, 1.4425e-02, -1.7729e-02, -1.4164e-02,\n 3.6607e-03, 1.1394e-02, 1.9208e-02, 2.6080e-02, -2.0411e-02,\n 1.4723e-02, 4.8366e-03, 1.2006e-02, -2.3813e-02, -1.2714e-02,\n 5.4300e-03, -2.4237e-02, -3.8326e-03, 3.5854e-02, 5.9531e-03,\n 3.7318e-03, -1.6699e-03, -1.2304e-02, -1.3232e-02, 1.7121e-02,\n -3.3530e-03, -1.6776e-02, -1.6286e-02, 3.1725e-02, -5.1826e-03,\n -4.0769e-04, -4.1575e-02, 1.7686e-02, 2.9299e-02, 1.9493e-02,\n 6.9532e-03, 2.0026e-02, 3.6382e-02, 1.1395e-03, 3.3299e-03,\n 2.1636e-02, 6.5690e-03, 5.2668e-02, 2.6005e-02, 1.1527e-03,\n 7.1510e-03, -4.2978e-02, 1.3060e-02, -2.7612e-02, -1.9152e-02,\n -6.1096e-03, 6.0516e-03, 1.7079e-02, -1.0752e-02, 2.0920e-03,\n -3.0535e-02, 2.6826e-02, -1.2330e-02, -2.1882e-02, 2.4536e-02,\n 1.3725e-02, -2.0055e-02, 9.8593e-04, 2.3281e-05, -4.9044e-03,\n 1.0477e-02, -2.2996e-02, 7.2976e-04, -8.6892e-03, 3.6027e-03,\n 1.3156e-02, 5.9225e-03, 2.4382e-02, -1.7293e-02, 1.2587e-02,\n 1.3080e-02, -1.9795e-02, -3.2899e-02, -1.1350e-03, 1.1024e-03,\n -8.3430e-03, 6.0690e-04, 1.0710e-02, -3.6709e-03, 2.9296e-03,\n -1.0477e-03, 3.4247e-03, -3.3369e-03, 3.7177e-02, -1.9105e-03,\n 1.4298e-02, -7.3553e-03, 2.0909e-02, 9.4052e-03, -9.8668e-03,\n 1.4866e-02, -3.7511e-03, -1.6187e-02, -5.4057e-03, -1.0727e-02,\n -1.4969e-02, -1.1871e-02, -1.8928e-02, 1.7391e-02, -1.0079e-02,\n -9.4460e-03, 3.9699e-03, 1.3333e-02, 2.0546e-03, -4.8634e-03,\n 1.7834e-02, 2.0914e-03, 3.0880e-03, 4.9837e-03, -5.7202e-04,\n -1.3710e-02, 4.3472e-02, 1.6004e-03, -4.0800e-02, 9.3455e-03,\n -1.0549e-02, 9.8545e-03, 3.1032e-02, 1.2305e-02, 1.8383e-02,\n -1.8368e-02, 5.9402e-03, -4.5390e-03, 4.1785e-02, -6.6477e-03,\n 2.0563e-02, 9.4982e-03, 7.5211e-03, -2.5485e-02, -7.2545e-03,\n -1.4194e-02, 1.7963e-03, -1.7765e-02, 6.2256e-04, 1.2401e-03,\n 1.2997e-02, -3.0555e-03, 2.0719e-02, -2.8840e-03, -7.0308e-03,\n 1.3817e-02, 4.0651e-02, 3.0992e-02, -8.9029e-03, 2.3886e-03,\n -9.0107e-03, -8.5236e-02, -2.4490e-02, -1.8329e-02, -1.5477e-02,\n 2.5628e-02, 6.4186e-03, 2.2837e-02, -1.3205e-02, -1.7316e-02,\n -1.8904e-02, -3.5773e-02, 9.9695e-03, 3.5855e-03, -8.8021e-03,\n 4.7696e-03, -1.0839e-04, -1.2240e-02, -1.5413e-02, 4.5439e-04,\n 2.0415e-02, -4.2744e-03, -9.7024e-04, 4.8487e-02, -1.8706e-02,\n 3.6501e-03, 2.0333e-02, -8.3754e-03, -1.3001e-02, -5.0250e-03,\n -1.9993e-02, 3.4180e-02, 4.8345e-03, -7.6883e-03, 3.3489e-02,\n -1.5000e-02, -1.6637e-02, 6.1983e-03, 1.6575e-02, 5.4960e-03,\n -2.0665e-02, 1.6282e-03, -1.9346e-02, 2.4276e-03, 2.8997e-02,\n 2.2135e-02, 3.8659e-03, -1.6921e-02, 5.2914e-02, -1.0792e-02,\n -1.8622e-02, -2.3203e-03, -7.7118e-03, 1.2215e-02, 7.9470e-03,\n 2.0650e-03, -3.6779e-02, -1.5974e-02, -4.2726e-03, 2.3141e-02,\n -2.8717e-02, 1.1501e-03, -8.9447e-03, 6.5250e-04, -2.4045e-02,\n 6.0208e-03, -7.9856e-03, -1.2354e-03, 6.1968e-03, -1.8020e-02,\n -1.9345e-02, 7.8173e-03], device='cuda:0')",
|
| 13 |
+
"exp_avg_sq": "tensor([0.0049, 0.0051, 0.0052, 0.0031, 0.0059, 0.0031, 0.0030, 0.0033, 0.0062,\n 0.0042, 0.0025, 0.0043, 0.0047, 0.0032, 0.0039, 0.0035, 0.0012, 0.0019,\n 0.0027, 0.0029, 0.0032, 0.0035, 0.0041, 0.0043, 0.0052, 0.0034, 0.0035,\n 0.0033, 0.0054, 0.0016, 0.0036, 0.0048, 0.0049, 0.0031, 0.0037, 0.0029,\n 0.0025, 0.0039, 0.0031, 0.0029, 0.0052, 0.0036, 0.0037, 0.0049, 0.0040,\n 0.0057, 0.0076, 0.0045, 0.0031, 0.0045, 0.0033, 0.0038, 0.0043, 0.0020,\n 0.0056, 0.0045, 0.0025, 0.0051, 0.0029, 0.0036, 0.0032, 0.0065, 0.0049,\n 0.0027, 0.0025, 0.0027, 0.0036, 0.0049, 0.0082, 0.0057, 0.0054, 0.0034,\n 0.0069, 0.0026, 0.0035, 0.0074, 0.0042, 0.0039, 0.0039, 0.0074, 0.0059,\n 0.0032, 0.0052, 0.0047, 0.0032, 0.0032, 0.0012, 0.0081, 0.0047, 0.0042,\n 0.0024, 0.0032, 0.0025, 0.0046, 0.0053, 0.0037, 0.0051, 0.0048, 0.0043,\n 0.0044, 0.0028, 0.0033, 0.0042, 0.0018, 0.0030, 0.0031, 0.0069, 0.0028,\n 0.0045, 0.0039, 0.0017, 0.0051, 0.0037, 0.0025, 0.0022, 0.0027, 0.0042,\n 0.0053, 0.0025, 0.0035, 0.0027, 0.0028, 0.0043, 0.0054, 0.0050, 0.0045,\n 0.0048, 0.0034, 0.0049, 0.0027, 0.0043, 0.0031, 0.0036, 0.0029, 0.0043,\n 0.0024, 0.0029, 0.0018, 0.0050, 0.0074, 0.0050, 0.0026, 0.0032, 0.0051,\n 0.0050, 0.0031, 0.0040, 0.0037, 0.0043, 0.0047, 0.0031, 0.0041, 0.0027,\n 0.0025, 0.0020, 0.0024, 0.0044, 0.0041, 0.0024, 0.0042, 0.0030, 0.0039,\n 0.0036, 0.0034, 0.0046, 0.0043, 0.0019, 0.0026, 0.0041, 0.0039, 0.0041,\n 0.0033, 0.0022, 0.0037, 0.0045, 0.0025, 0.0032, 0.0038, 0.0028, 0.0035,\n 0.0053, 0.0073, 0.0020, 0.0045, 0.0067, 0.0043, 0.0016, 0.0022, 0.0041,\n 0.0031, 0.0022, 0.0023, 0.0060, 0.0036, 0.0043, 0.0034, 0.0029, 0.0033,\n 0.0034, 0.0083, 0.0021, 0.0040, 0.0025, 0.0047, 0.0035, 0.0030, 0.0022,\n 0.0029, 0.0034, 0.0049, 0.0023, 0.0049, 0.0051, 0.0070, 0.0037, 0.0024,\n 0.0036, 0.0025, 0.0031, 0.0045, 0.0024, 0.0041, 0.0065, 0.0055, 0.0019,\n 0.0032, 0.0037, 0.0047, 0.0040, 0.0063, 0.0036, 0.0030, 0.0044, 0.0024,\n 0.0036, 0.0046, 0.0048, 0.0045, 0.0035, 0.0081, 0.0056, 0.0032, 0.0040,\n 0.0028, 0.0028, 0.0049, 0.0033, 0.0035, 0.0024, 0.0068, 0.0025, 0.0032,\n 0.0016, 0.0083, 0.0045, 0.0046, 0.0074, 0.0032, 0.0040, 0.0038, 0.0038,\n 0.0068, 0.0029, 0.0026, 0.0019, 0.0038, 0.0030, 0.0052, 0.0019, 0.0033,\n 0.0022, 0.0024, 0.0046, 0.0037, 0.0031, 0.0056, 0.0016, 0.0027, 0.0076,\n 0.0036, 0.0033, 0.0046, 0.0037, 0.0026, 0.0066, 0.0045, 0.0034, 0.0039,\n 0.0058, 0.0027, 0.0033, 0.0056, 0.0038, 0.0042, 0.0062, 0.0046, 0.0056,\n 0.0047, 0.0065, 0.0061, 0.0042, 0.0028, 0.0031, 0.0027, 0.0029, 0.0034,\n 0.0039, 0.0028, 0.0053, 0.0039, 0.0034, 0.0043, 0.0026, 0.0036, 0.0025,\n 0.0047, 0.0021, 0.0048, 0.0036, 0.0027, 0.0046, 0.0067, 0.0035, 0.0028,\n 0.0040, 0.0028, 0.0033, 0.0037, 0.0024, 0.0038, 0.0054, 0.0039, 0.0035,\n 0.0029, 0.0024, 0.0027, 0.0036, 0.0037, 0.0052, 0.0048, 0.0031, 0.0033,\n 0.0029, 0.0035, 0.0043, 0.0049, 0.0028, 0.0031, 0.0030, 0.0036, 0.0041,\n 0.0062, 0.0031, 0.0042, 0.0024, 0.0038, 0.0049, 0.0043, 0.0025, 0.0037,\n 0.0040, 0.0041, 0.0040, 0.0022, 0.0031, 0.0033, 0.0028, 0.0070, 0.0027,\n 0.0025, 0.0028, 0.0082, 0.0049, 0.0039, 0.0037, 0.0023, 0.0040, 0.0029,\n 0.0043, 0.0030, 0.0026, 0.0037, 0.0035, 0.0043, 0.0053, 0.0031, 0.0022,\n 0.0039, 0.0048, 0.0017, 0.0046, 0.0073, 0.0030, 0.0035, 0.0026, 0.0053,\n 0.0021, 0.0030, 0.0073, 0.0019, 0.0021, 0.0034, 0.0039, 0.0047, 0.0046,\n 0.0067, 0.0038, 0.0028, 0.0047, 0.0022, 0.0050, 0.0049, 0.0056, 0.0027,\n 0.0059, 0.0027, 0.0035, 0.0026, 0.0053, 0.0027, 0.0039, 0.0035, 0.0047,\n 0.0054, 0.0062, 0.0036, 0.0067, 0.0055, 0.0027, 0.0025, 0.0044, 0.0026,\n 0.0029, 0.0109, 0.0024, 0.0042, 0.0045, 0.0049, 0.0046, 0.0047, 0.0050,\n 0.0064, 0.0035, 0.0031, 0.0025, 0.0034, 0.0047, 0.0026, 0.0028, 0.0029,\n 0.0016, 0.0064, 0.0037, 0.0042, 0.0030, 0.0058, 0.0038, 0.0026, 0.0042,\n 0.0038, 0.0045, 0.0053, 0.0030, 0.0051, 0.0013, 0.0040, 0.0029, 0.0039,\n 0.0037, 0.0047, 0.0044, 0.0028, 0.0051, 0.0036, 0.0033, 0.0027, 0.0032,\n 0.0036, 0.0038, 0.0030, 0.0030, 0.0039, 0.0032, 0.0035, 0.0033, 0.0039,\n 0.0027, 0.0029, 0.0076, 0.0044, 0.0006, 0.0043, 0.0037, 0.0036, 0.0050,\n 0.0046, 0.0045, 0.0051, 0.0050, 0.0040, 0.0029, 0.0036, 0.0046, 0.0029,\n 0.0045, 0.0056, 0.0043, 0.0030, 0.0040, 0.0035, 0.0055, 0.0028],\n device='cuda:0')"
|
| 14 |
+
},
|
| 15 |
+
"2": {
|
| 16 |
+
"step": "tensor(8764.)",
|
| 17 |
+
"exp_avg": "tensor([ 2.8681e-03, -2.5426e-03, -1.1098e-02, -2.1047e-02, -1.1849e-03,\n -9.9222e-03, 2.0555e-03, -1.2428e-03, -2.6872e-04, 1.8078e-03,\n -6.7264e-05, 1.4071e-03, -1.2223e-03, 7.1254e-03, -1.1080e-03,\n 2.3781e-03, 2.1188e-02, -1.3111e-03, 1.1534e-03, 2.1766e-03,\n -1.2839e-03, -2.0468e-03, -1.3270e-03, 1.5220e-03, 3.4697e-03,\n 3.0007e-03, 3.5488e-04, 1.1526e-02, 5.8256e-04, -5.3460e-03,\n 5.3838e-03, -4.1705e-03, 4.4176e-03, -4.7307e-04, -7.3420e-03,\n -5.8834e-04, 1.8651e-03, -2.0625e-02, 1.0946e-03, 6.2497e-05,\n 7.8314e-04, -9.4685e-03, 2.8782e-04, -6.5068e-04, 2.8678e-03,\n 4.2983e-03, -3.7640e-04, 1.0684e-02, -9.2082e-03, -9.7055e-04,\n 1.5527e-03, -1.7811e-03, -1.4587e-03, 7.6884e-03, -3.0509e-03,\n 7.3352e-03, 3.4816e-03, -5.9871e-03, 6.3291e-04, 8.9677e-03,\n 2.7173e-03, -3.9758e-03, 4.8630e-03, -7.3777e-03, 1.2686e-02,\n -1.4088e-03, 1.4273e-03, -4.1856e-04, 2.4652e-03, -1.2234e-03,\n 1.9103e-03, -6.7110e-04, 3.9359e-04, -2.2625e-03, -7.8527e-05,\n 1.5347e-03, -9.2910e-04, 8.3121e-03, 3.9311e-03, 1.9366e-03,\n 1.1727e-03, 9.0982e-04, -7.2500e-03, -1.8053e-03, 3.9153e-03,\n 3.1396e-03, -3.0391e-02, 1.4431e-05, -9.4425e-05, 2.4720e-03,\n -1.6301e-03, -6.5771e-03, -1.1544e-03, -2.5443e-03, 3.1572e-03,\n -2.3842e-03, 4.5651e-03, -6.3902e-03, 4.7413e-04, 1.4225e-03,\n -1.8060e-03, -3.2744e-03, -1.6318e-03, 3.9844e-03, -4.3734e-03,\n -3.6517e-03, 1.1291e-03, 6.7261e-04, -6.0756e-03, 1.4136e-03,\n 8.3620e-03, 5.8273e-04, 4.9613e-03, 5.0743e-05, 7.0927e-03,\n -1.5729e-03, -1.0529e-03, -3.4077e-03, 1.1772e-03, 3.4237e-03,\n 1.6531e-03, -8.4811e-03, 2.1894e-03, 4.4242e-03, -4.3291e-04,\n 8.5585e-03, 8.4304e-04, 1.0243e-02, 1.8897e-03, 1.7037e-04,\n -2.2231e-03, 3.6601e-03, -2.8872e-03, -7.3804e-03, -3.3589e-03,\n 2.6865e-03, 1.8622e-03, -6.7265e-03, 3.1440e-03, 4.2371e-03,\n 1.7980e-03, 3.6684e-03, -5.0042e-03, 3.8818e-03, -6.2992e-03,\n -3.2352e-03, -2.7140e-03, -1.3870e-03, 2.6528e-03, 7.2428e-04,\n 6.3131e-04, 4.3145e-03, 5.2718e-04, 7.4506e-03, -7.3047e-04,\n -2.7090e-04, -2.2379e-03, -2.9048e-03, -2.3019e-03, 1.9986e-03,\n 4.5128e-03, 3.7496e-03, 2.5049e-03, 3.5887e-03, -5.7162e-03,\n -1.4182e-03, -3.2813e-03, -8.6704e-04, 9.5800e-04, -1.2890e-03,\n 2.7999e-03, -2.0711e-04, 2.8231e-03, 7.1065e-03, 1.6001e-04,\n 1.6859e-03, -6.1789e-03, 1.0755e-02, -1.3526e-02, 5.3961e-03,\n 3.4315e-03, -8.8927e-05, 1.3732e-02, -2.0392e-03, -4.8629e-03,\n 5.6655e-04, 7.9487e-03, -2.5277e-03, -2.0512e-03, -1.3584e-03,\n -1.5951e-02, 5.0937e-03, 3.0983e-03, -3.1758e-03, 2.4435e-03,\n 4.4139e-03, -5.3191e-03, 5.2663e-03, -2.9857e-03, -3.1467e-03,\n -6.1281e-03, -1.9927e-03, 4.3585e-03, -7.8764e-04, -6.1514e-03,\n -9.2148e-05, -5.9244e-04, 5.6678e-03, -9.4418e-04, -2.3402e-03,\n 3.1200e-03, -5.0208e-03, -3.0276e-03, -5.2710e-04, 1.6247e-03,\n -1.7505e-03, -5.1647e-03, -6.5895e-03, 6.7057e-03, 2.7259e-04,\n 3.3645e-03, 2.3024e-03, 1.7793e-03, 1.3698e-03, -2.9068e-04,\n -7.9918e-04, -1.2903e-02, 1.1487e-03, -6.0069e-03, -3.8158e-03,\n 4.7803e-05, -4.7401e-03, -6.2199e-04, -8.0461e-04, 1.9454e-03,\n -3.7311e-03, 3.0844e-04, -6.6416e-05, -4.3233e-03, 2.6338e-03,\n 4.7086e-03, 4.3283e-03, 1.1313e-03, -9.7041e-04, -1.1100e-02,\n 4.4286e-04, 3.9925e-03, -2.6005e-03, 4.8828e-03, -6.6220e-03,\n 5.9449e-03, 6.1408e-03, -6.7908e-03, 4.7158e-03, -2.5605e-03,\n -1.4791e-03, 3.2489e-03, 1.8531e-03, 2.6310e-03, -6.5045e-04,\n 3.9575e-04, -5.0960e-03, -9.1781e-03, 6.2744e-03, -2.3997e-03,\n -2.6187e-03, -1.1464e-02, 2.0838e-03, -1.7698e-03, 4.1809e-03,\n 6.5008e-03, 5.4419e-03, 4.5196e-04, 1.7683e-03, -1.9098e-03,\n -8.8874e-03, 2.6747e-02, 2.9565e-04, -3.3308e-05, 1.6492e-04,\n -2.2267e-03, 1.8167e-03, 3.2158e-03, 6.3275e-04, 5.4767e-03,\n -1.5388e-03, 6.8913e-03, 8.5729e-04, 3.9159e-03, 3.2994e-03,\n 1.9892e-03, 9.3612e-03, 3.0011e-03, -2.8197e-03, -2.1479e-03,\n 3.3882e-03, 7.1180e-04, 5.0135e-03, 2.0370e-03, -4.8037e-03,\n 4.2466e-03, 1.2980e-03, 3.4169e-03, -8.7301e-03, -2.1466e-03,\n 1.0303e-03, -3.4258e-03, -9.8762e-04, 5.0254e-03, 1.5524e-03,\n 4.8478e-04, 8.8499e-04, -2.1758e-03, -1.5935e-03, 7.5788e-03,\n -1.3810e-04, -2.7298e-03, -4.1027e-03, 4.7756e-03, -1.5937e-03,\n 1.3753e-03, -6.9684e-03, 3.4761e-03, 5.4077e-03, 3.4081e-03,\n 3.8191e-04, 4.4198e-03, 5.5945e-03, 1.3658e-03, 1.4741e-03,\n 2.9462e-03, 5.3896e-04, 8.5903e-03, 5.9965e-03, 4.3437e-04,\n 2.6523e-03, -7.4729e-03, 5.0929e-03, -4.5410e-03, -2.2749e-03,\n -2.2918e-03, 2.1247e-03, 3.4533e-03, -3.2104e-03, 1.6341e-04,\n -3.7189e-03, 5.1757e-03, -2.0452e-03, -5.8760e-03, 2.2971e-03,\n 1.8635e-03, -2.3802e-03, 8.4757e-04, -6.0137e-04, -1.1488e-03,\n 3.0290e-03, -1.8878e-03, 5.5825e-05, -3.3472e-03, 4.0281e-04,\n 2.4489e-03, 7.0854e-04, 3.7753e-03, -4.5502e-03, 2.2495e-03,\n 3.6308e-03, -2.5660e-03, -5.4018e-03, 7.7914e-04, 1.5507e-04,\n -1.1492e-03, -2.1039e-04, 2.1538e-03, 6.8131e-04, 6.3588e-04,\n -2.9286e-05, -2.1236e-04, -8.5252e-04, 5.9036e-03, 1.5760e-03,\n 2.1250e-03, -6.0997e-04, 5.0176e-03, 1.5197e-03, -1.9402e-03,\n 2.5993e-03, 7.4918e-04, -3.4823e-03, -7.2468e-04, -4.2194e-03,\n -1.0892e-03, -5.4240e-05, -5.3918e-03, 3.8919e-03, -3.2892e-03,\n -1.7569e-03, 1.2285e-03, 4.8447e-03, -3.6477e-04, -2.4990e-03,\n 6.0817e-03, 7.5175e-04, 5.4900e-04, 9.9531e-04, 2.6080e-04,\n -1.8414e-03, 9.5904e-03, -1.7025e-04, -5.3219e-03, 5.9804e-03,\n -2.5852e-03, 1.4591e-03, 5.5317e-03, 4.3993e-03, 2.7757e-03,\n -3.2877e-03, 1.0222e-03, -1.1395e-03, 5.9583e-03, -1.7661e-03,\n 4.6311e-03, 9.7628e-04, 2.0241e-03, -2.2463e-03, -1.2108e-03,\n -1.2408e-03, 5.2611e-04, -2.9247e-03, 7.0759e-04, 2.0615e-04,\n 1.8428e-03, 2.6329e-04, 5.5631e-03, 5.0337e-04, -1.4099e-03,\n 1.7550e-03, 4.8192e-03, 6.5165e-03, -1.9431e-03, 1.4685e-03,\n -6.0536e-04, -1.4339e-02, -4.4233e-03, -4.8754e-03, -3.5004e-03,\n 3.9969e-03, 4.9993e-04, 7.3782e-03, -2.4260e-03, -4.8220e-03,\n -9.0775e-03, -6.5861e-03, 1.5022e-03, -8.0777e-04, -3.6162e-03,\n 1.3262e-03, 5.7841e-04, -7.0322e-03, -1.1192e-03, 9.3897e-05,\n 4.5186e-03, -1.2469e-03, 5.3072e-04, 6.0314e-03, -1.5290e-02,\n 9.1860e-04, 4.9126e-03, -1.0744e-03, -1.2216e-03, 6.1601e-05,\n -2.8087e-03, 1.2703e-02, 8.5163e-04, -1.1605e-03, 8.4059e-03,\n -3.1057e-03, -3.5473e-03, 2.9292e-03, 1.9768e-03, 1.7099e-03,\n -2.4512e-03, 2.6598e-04, -4.1935e-03, 4.8955e-05, 5.6919e-03,\n 6.2311e-03, 1.4660e-03, -3.1074e-03, 6.3021e-03, -2.0057e-03,\n -3.3289e-01, -8.3979e-04, -1.1001e-03, 2.3877e-03, 2.3823e-03,\n -7.5744e-04, -8.1727e-03, -2.6400e-03, -8.0115e-05, 6.6702e-03,\n -5.5432e-03, -2.1764e-04, -8.7235e-04, 8.8147e-04, -3.9210e-03,\n 9.2687e-04, -1.9301e-03, 1.1381e-03, 1.3666e-03, -3.8462e-03,\n -2.5057e-03, 1.6878e-03], device='cuda:0')",
|
| 18 |
+
"exp_avg_sq": "tensor([1.9768e-04, 8.0180e-05, 1.8253e-04, 3.0518e-04, 7.1704e-05, 1.4071e-04,\n 1.0133e-04, 1.6403e-04, 7.1062e-05, 1.7311e-04, 1.3517e-04, 2.2754e-04,\n 7.2690e-05, 1.3572e-04, 5.5349e-05, 7.6881e-05, 2.5367e-03, 1.7233e-04,\n 1.3772e-04, 2.0298e-04, 2.8445e-04, 2.7398e-04, 1.9661e-04, 1.0931e-04,\n 1.4695e-04, 3.1194e-04, 2.0399e-04, 1.5550e-04, 5.8291e-05, 3.8664e-04,\n 2.0227e-04, 9.3472e-05, 2.9498e-04, 6.6138e-05, 4.0015e-04, 2.8698e-04,\n 2.4589e-04, 2.7511e-04, 6.2912e-05, 1.1835e-04, 1.1099e-04, 1.3417e-04,\n 5.8994e-05, 6.3162e-05, 1.2532e-04, 8.8022e-05, 2.7632e-04, 2.0562e-04,\n 3.4739e-04, 1.2112e-04, 2.6233e-04, 3.0998e-04, 8.5372e-05, 2.3476e-04,\n 1.2441e-04, 2.1184e-04, 4.1017e-04, 1.7248e-04, 1.5998e-04, 2.0319e-04,\n 2.5577e-04, 5.5717e-05, 1.4491e-04, 3.1352e-04, 5.7839e-04, 1.7065e-04,\n 2.1822e-04, 1.5947e-04, 5.6698e-05, 1.7689e-04, 5.7506e-05, 1.0465e-04,\n 7.3782e-05, 1.1629e-04, 1.2843e-04, 1.0766e-04, 1.5108e-04, 1.3093e-04,\n 2.3421e-04, 1.4980e-04, 9.8981e-05, 7.3202e-05, 1.1621e-04, 1.4963e-04,\n 1.4048e-04, 2.2269e-04, 2.2599e-03, 1.0878e-04, 7.2461e-05, 1.8170e-04,\n 1.7950e-04, 2.5111e-04, 2.1988e-04, 9.4880e-05, 9.0481e-05, 9.7078e-05,\n 1.3293e-04, 1.0527e-04, 1.2956e-04, 7.0796e-05, 7.0005e-05, 7.6397e-05,\n 9.3242e-05, 2.5020e-04, 2.7548e-04, 1.8844e-04, 7.0208e-05, 2.6035e-04,\n 1.0187e-04, 7.2112e-05, 4.2960e-04, 6.1182e-05, 1.5145e-04, 7.4916e-05,\n 3.5778e-04, 9.7391e-05, 7.5727e-05, 9.8345e-05, 2.0676e-04, 1.1479e-04,\n 9.6535e-05, 2.1777e-04, 1.1175e-04, 1.1864e-04, 1.0773e-04, 1.0396e-04,\n 7.8988e-05, 1.7917e-04, 8.4152e-05, 3.0866e-04, 7.9011e-05, 1.2651e-04,\n 3.5447e-04, 2.5471e-04, 1.3993e-04, 2.0468e-04, 8.2837e-05, 2.3428e-04,\n 1.1722e-04, 1.0315e-04, 2.2730e-04, 1.9924e-04, 1.2231e-04, 1.4177e-04,\n 1.0485e-04, 2.0171e-04, 1.4780e-04, 1.5799e-04, 1.1310e-04, 1.1608e-04,\n 6.4907e-05, 1.9915e-04, 1.3745e-04, 2.5000e-04, 1.1358e-04, 2.3568e-04,\n 1.3468e-04, 1.3055e-04, 5.6706e-04, 5.4034e-05, 2.6510e-04, 1.4077e-04,\n 1.7754e-04, 1.1851e-04, 1.5884e-04, 9.2169e-05, 4.0727e-04, 1.8117e-04,\n 7.0376e-05, 1.1393e-04, 3.2166e-04, 1.3085e-04, 4.3297e-04, 1.1550e-04,\n 1.1813e-04, 1.4456e-04, 1.2249e-04, 1.9817e-04, 1.3220e-04, 7.0056e-05,\n 1.1952e-04, 1.2538e-04, 5.6167e-04, 1.6546e-04, 1.3001e-04, 1.6375e-04,\n 6.7364e-04, 4.3130e-04, 8.7292e-05, 2.1493e-04, 4.0453e-04, 3.9568e-04,\n 1.7269e-04, 1.1536e-04, 1.3206e-04, 1.1835e-04, 6.8386e-05, 1.3570e-04,\n 6.0807e-05, 1.0239e-04, 3.0015e-04, 2.2618e-04, 7.7511e-05, 7.8212e-05,\n 1.2623e-04, 2.6542e-04, 3.7532e-04, 2.1519e-04, 1.7442e-04, 1.1535e-04,\n 1.5827e-04, 1.3184e-04, 1.3372e-04, 1.7949e-04, 2.8878e-04, 1.5077e-04,\n 1.9840e-04, 8.2625e-05, 1.2982e-04, 1.3375e-04, 1.6250e-04, 7.7465e-05,\n 6.9313e-05, 1.2956e-04, 2.7095e-04, 1.0745e-04, 2.0686e-04, 7.5783e-05,\n 4.0564e-04, 1.9968e-04, 8.3593e-05, 1.4072e-04, 7.0149e-05, 1.3673e-04,\n 1.3362e-04, 1.2580e-04, 1.7146e-04, 1.7438e-04, 1.7849e-04, 1.2840e-04,\n 7.5839e-05, 1.6262e-04, 1.4664e-04, 1.1889e-04, 2.9349e-04, 8.7675e-05,\n 1.2428e-04, 5.3416e-05, 2.0033e-04, 1.7415e-04, 1.5217e-04, 1.7423e-04,\n 3.4790e-04, 1.1813e-04, 1.1960e-04, 8.0989e-05, 3.1104e-04, 1.2146e-04,\n 1.6800e-04, 1.5519e-04, 9.9707e-05, 9.5314e-05, 2.4736e-04, 3.2229e-04,\n 6.6895e-04, 2.1765e-04, 1.4487e-04, 1.1941e-04, 1.8930e-04, 1.0600e-04,\n 2.3704e-04, 1.5451e-04, 7.3528e-05, 6.2406e-05, 8.1027e-05, 3.0143e-04,\n 7.6715e-04, 1.9497e-04, 1.3479e-04, 9.0098e-05, 4.0298e-05, 1.5347e-04,\n 1.1401e-04, 1.8521e-04, 1.8632e-04, 8.6596e-05, 7.8526e-04, 7.8879e-05,\n 1.1856e-04, 2.1921e-04, 2.1352e-04, 1.2102e-04, 1.6385e-04, 1.1208e-04,\n 1.2744e-04, 1.2798e-04, 5.8854e-05, 1.8578e-04, 7.1394e-05, 1.1058e-04,\n 2.9132e-04, 1.1677e-04, 1.0281e-04, 3.1256e-04, 1.0929e-04, 1.6368e-04,\n 1.4168e-04, 1.2406e-04, 1.2723e-04, 1.0624e-04, 1.0895e-04, 7.3678e-05,\n 1.2876e-04, 8.9152e-05, 3.6214e-04, 1.8338e-04, 8.3368e-05, 2.1426e-04,\n 8.6668e-05, 3.0719e-04, 6.7128e-05, 2.5823e-04, 1.0011e-04, 1.3461e-04,\n 8.7102e-05, 8.1198e-05, 1.8620e-04, 1.2597e-04, 6.8744e-05, 1.3898e-04,\n 1.1864e-04, 2.4810e-04, 1.0805e-04, 1.3745e-04, 4.7980e-04, 2.7453e-04,\n 1.3568e-04, 3.4512e-04, 1.3602e-04, 9.2495e-05, 1.7073e-04, 2.3372e-04,\n 1.0887e-04, 1.3549e-04, 1.0538e-04, 9.9478e-05, 1.2183e-04, 8.1468e-05,\n 3.0210e-04, 5.6546e-05, 1.9093e-04, 8.8993e-05, 5.8571e-05, 9.5332e-05,\n 1.6907e-04, 9.4349e-05, 4.1315e-05, 8.7081e-05, 4.6658e-04, 9.2075e-05,\n 1.0329e-04, 5.1321e-05, 1.0460e-04, 1.7810e-04, 7.1499e-05, 1.1726e-04,\n 1.2125e-04, 1.0845e-04, 1.5560e-04, 1.7849e-04, 2.7588e-04, 1.1611e-04,\n 1.0169e-04, 1.3284e-04, 9.4087e-05, 1.6947e-04, 7.8862e-05, 1.3332e-04,\n 1.0746e-04, 2.0708e-04, 6.2160e-05, 1.0046e-04, 1.9762e-04, 1.6975e-04,\n 8.9367e-05, 8.4384e-05, 1.8847e-04, 5.9343e-05, 5.0797e-05, 7.5871e-04,\n 1.3622e-04, 8.0442e-05, 2.1418e-04, 1.2661e-04, 2.5228e-04, 2.1861e-04,\n 2.4957e-04, 2.6340e-04, 1.6345e-04, 2.5069e-04, 2.4226e-04, 9.5623e-05,\n 7.1207e-05, 6.2767e-05, 7.5460e-05, 8.4715e-05, 1.4133e-04, 1.3906e-04,\n 9.7759e-05, 4.0805e-04, 1.3234e-04, 6.1827e-05, 1.3996e-04, 4.3982e-04,\n 6.7985e-05, 2.4587e-04, 9.8322e-05, 1.2557e-04, 8.5925e-05, 2.8856e-04,\n 2.0232e-04, 1.1884e-04, 1.0150e-04, 1.3160e-04, 6.5551e-05, 1.2002e-04,\n 1.0719e-04, 9.9095e-05, 2.2404e-04, 2.2727e-04, 6.5201e-05, 4.0749e-04,\n 2.0952e-04, 9.3590e-05, 1.6579e-04, 1.4403e-04, 6.3834e-05, 1.4635e-04,\n 1.3042e-04, 7.8551e-05, 1.2701e-04, 1.4833e-04, 1.1160e-04, 3.0757e-04,\n 1.5378e-04, 7.3294e-05, 8.1727e-05, 2.5441e-04, 2.2201e-04, 2.4366e-04,\n 4.8010e-04, 1.2653e-04, 6.5302e-05, 2.5652e-04, 2.2471e-04, 1.0997e-04,\n 5.6638e-05, 6.3582e-04, 1.4300e-04, 1.4903e-04, 1.8237e-04, 9.3261e-05,\n 2.3131e-04, 6.5392e-05, 9.4362e-04, 2.0567e-04, 1.1215e-04, 7.0293e-05,\n 1.6305e-04, 6.9366e-05, 1.0406e-04, 2.4264e-04, 8.6208e-05, 1.6107e-04,\n 2.5453e-04, 2.3034e-04, 1.4407e-04, 2.0898e-04, 4.0365e-05, 1.2344e-04,\n 8.5382e-05, 4.5204e-05, 2.0379e-04, 1.6478e-04, 1.0425e-04, 2.8313e-04,\n 8.5866e-05, 1.6464e-04, 1.1295e-04, 1.1493e-04, 2.0897e-01, 8.4273e-05,\n 1.6482e-04, 1.3454e-04, 1.2288e-04, 2.6286e-04, 2.3877e-04, 1.4279e-04,\n 1.0611e-04, 1.9302e-04, 1.0848e-04, 1.2726e-04, 8.9845e-05, 1.1576e-04,\n 1.5567e-04, 1.8178e-04, 2.4600e-04, 1.9071e-04, 1.8347e-04, 1.8869e-04,\n 2.1694e-04, 1.4254e-04], device='cuda:0')"
|
| 19 |
+
},
|
| 20 |
+
"3": {
|
| 21 |
+
"step": "tensor(8764.)",
|
| 22 |
+
"exp_avg": "tensor([ 1.2804e-03, -3.4637e-03, -1.1529e-02, -1.7266e-02, -1.3785e-03,\n -9.1008e-03, 1.5672e-03, -7.2443e-04, -2.9534e-04, 1.7846e-03,\n -1.5997e-05, 5.0771e-04, -5.0300e-04, 5.3647e-03, -5.4486e-04,\n 2.6715e-03, 6.5275e-03, -9.0004e-04, 6.3842e-04, 2.3870e-03,\n 8.6821e-04, -1.9584e-03, -1.1307e-03, 2.0918e-03, 4.0097e-03,\n 2.1542e-03, -4.6603e-04, 8.7567e-03, 1.0986e-03, -2.7850e-03,\n 4.4870e-03, -3.7793e-03, 4.6424e-03, 4.1138e-04, -4.9150e-03,\n -7.5625e-04, 2.2446e-03, -1.4657e-02, 1.4496e-03, 3.1602e-04,\n 1.6492e-03, -9.7410e-03, -1.6732e-04, -1.3682e-03, 2.6531e-03,\n 5.6212e-03, -9.1863e-05, 8.6951e-03, -6.2864e-03, -3.3969e-04,\n 4.0106e-04, -1.2044e-03, -1.9468e-03, 6.6910e-03, -4.0932e-03,\n 7.2665e-03, 3.0034e-03, -6.6496e-03, 1.0648e-03, 6.5513e-03,\n 1.5475e-03, -6.4361e-03, 5.7475e-03, -5.7147e-03, 8.4394e-03,\n -9.2223e-04, 1.2380e-03, 7.0921e-04, 4.3208e-03, -4.9811e-04,\n 2.3774e-03, 2.8839e-04, -2.5728e-04, -1.3232e-03, 2.0062e-04,\n 1.8832e-03, -6.6788e-04, 9.3374e-03, 3.9335e-03, 2.3497e-03,\n 1.9820e-03, 2.6162e-03, -9.4130e-03, -2.8745e-03, 3.7062e-03,\n 2.7174e-03, -9.0003e-03, 7.0053e-04, 4.4567e-04, 1.4362e-03,\n -1.8648e-04, -5.6146e-03, -1.1838e-03, -2.6420e-03, 3.8587e-03,\n -2.3089e-03, 3.6410e-03, -7.8821e-03, -2.5644e-04, 3.2619e-03,\n -1.4473e-03, -3.8583e-03, -2.6565e-03, 3.6132e-03, -3.1960e-03,\n -3.5794e-03, 1.7148e-03, 3.9494e-04, -8.7369e-03, 1.2924e-03,\n 5.1762e-03, 1.8680e-03, 5.0546e-03, 4.9872e-04, 5.1648e-03,\n -1.3653e-03, -2.1359e-03, -4.4531e-03, 1.8941e-03, 3.0052e-03,\n 3.2381e-03, -7.3133e-03, 2.6604e-03, 5.6337e-03, 5.3288e-04,\n 9.2514e-03, 6.3910e-04, 9.5321e-03, 2.1489e-03, -1.6780e-03,\n -2.4807e-03, 3.4480e-03, -2.0275e-03, -4.3928e-03, -3.4371e-03,\n 2.3984e-03, 2.2744e-03, -5.5520e-03, 2.2951e-03, 6.0487e-03,\n 2.2344e-03, 2.3807e-03, -2.7566e-03, 3.0324e-03, -7.6594e-03,\n -2.1643e-03, -2.0491e-03, -7.2167e-04, 2.9187e-03, 1.8175e-03,\n 3.9450e-04, 4.6040e-03, 8.7318e-04, 6.5252e-03, -1.0509e-03,\n 2.0637e-04, -2.1269e-03, -1.2693e-03, -1.8272e-03, 2.5797e-03,\n 3.6415e-03, 4.2317e-03, 1.5259e-03, 2.1225e-03, -7.2777e-03,\n -2.7345e-03, -2.3110e-03, -5.8077e-04, 1.2071e-03, -1.6275e-03,\n 3.1597e-03, -1.8182e-04, 2.0184e-03, 9.1216e-03, 7.0238e-05,\n 1.7579e-03, -6.4897e-03, 1.0890e-02, -1.2851e-02, 5.6575e-03,\n 4.7894e-03, -7.8593e-04, 7.7385e-03, -2.7952e-03, -5.0023e-03,\n 3.6711e-04, 3.6896e-03, -1.6313e-03, -2.2163e-03, -1.1046e-03,\n -9.0923e-03, 2.9080e-03, 3.2571e-03, -4.1116e-03, 3.7082e-03,\n 4.8614e-03, -5.4538e-03, 3.6560e-03, -1.9215e-03, -3.3697e-03,\n -4.4402e-03, -1.0113e-03, 4.3602e-03, -1.3771e-03, -6.5165e-03,\n -9.6120e-05, -1.8828e-04, 3.2743e-03, 1.8173e-04, -2.8367e-03,\n 3.6709e-03, -5.1919e-03, -2.8975e-03, -1.3565e-03, 1.5463e-03,\n -1.4713e-03, -4.2399e-03, -6.4759e-03, 6.3291e-03, -5.9926e-04,\n 2.9517e-03, 3.5851e-03, 2.9979e-03, 1.9133e-03, 3.2915e-04,\n -1.2487e-03, -7.7072e-03, 1.5596e-03, -4.7165e-03, -3.5770e-03,\n 5.2127e-04, -4.6731e-03, -1.0197e-03, -6.4986e-04, 1.5605e-03,\n -2.2699e-03, -9.7694e-05, -3.1643e-04, -4.6531e-03, 3.5617e-03,\n 1.8297e-03, 4.0949e-03, 9.3142e-05, -9.2014e-04, -6.8314e-03,\n -3.4301e-04, 3.9986e-03, -2.8270e-03, 4.9924e-03, -8.5687e-03,\n 5.3004e-03, 6.2611e-03, -3.5399e-03, 6.5802e-03, -3.5551e-03,\n -2.1769e-03, 3.4781e-03, 2.1781e-03, 2.0158e-03, -3.5189e-04,\n 4.1647e-04, -7.5686e-03, -7.5854e-03, 5.2776e-03, -6.4812e-04,\n -2.9980e-03, -5.6929e-03, 1.5963e-03, -6.4063e-04, 3.5380e-03,\n 5.5038e-03, 4.0885e-03, 1.0231e-03, 2.1886e-03, -2.4291e-03,\n -8.9160e-03, 1.2229e-02, -1.9133e-04, -4.6891e-04, 2.2866e-04,\n -2.1697e-03, 2.9469e-03, 3.6015e-03, 5.0194e-05, 6.5058e-03,\n -1.9389e-03, 4.8517e-03, 8.3242e-04, 6.8913e-03, 1.9516e-03,\n 1.6088e-03, 1.0286e-02, 2.5622e-03, -3.3928e-03, -2.9315e-03,\n 1.4821e-03, 2.0834e-03, 4.5568e-03, 4.8393e-03, -4.6469e-03,\n 4.0242e-03, 1.2635e-03, 3.0498e-03, -6.1525e-03, -2.4998e-03,\n 1.3140e-03, -3.8876e-03, -3.9789e-04, 6.6587e-03, 1.6209e-03,\n 6.1306e-04, 1.1635e-04, -2.6020e-03, -1.8090e-03, 5.1812e-03,\n -9.4238e-05, -3.3886e-03, -3.4006e-03, 6.1943e-03, -5.6442e-04,\n 5.3374e-04, -8.1101e-03, 3.0732e-03, 6.3270e-03, 3.5137e-03,\n 1.9637e-03, 4.5855e-03, 7.7430e-03, 7.3972e-04, 9.3472e-04,\n 3.6603e-03, 1.3307e-03, 1.0424e-02, 5.5274e-03, 1.0810e-03,\n 2.5893e-03, -8.4737e-03, 3.8476e-03, -5.0593e-03, -3.2678e-03,\n -1.3961e-03, 1.9634e-03, 3.7079e-03, -1.5899e-03, 9.9236e-05,\n -5.4244e-03, 5.2634e-03, -2.1865e-03, -5.0188e-03, 3.9349e-03,\n 3.1926e-03, -3.1258e-03, 1.7054e-04, 1.4609e-04, -9.3869e-04,\n 2.4700e-03, -2.9859e-03, 7.4101e-04, -2.1405e-03, 9.3972e-04,\n 2.2535e-03, 1.1459e-03, 5.1729e-03, -3.7259e-03, 2.6230e-03,\n 3.1762e-03, -4.1296e-03, -6.4111e-03, -3.5309e-04, 6.5787e-04,\n -1.7023e-03, 2.5462e-05, 2.9233e-03, 5.8682e-04, 6.2334e-04,\n -2.2781e-04, 5.6913e-04, -2.5598e-04, 7.1999e-03, 4.8326e-05,\n 2.1283e-03, -3.5940e-04, 5.2207e-03, 2.4247e-03, -1.5904e-03,\n 2.8506e-03, -2.1438e-04, -2.4489e-03, -5.9956e-04, -2.4288e-03,\n -1.1047e-03, -2.3917e-03, -4.6359e-03, 3.6729e-03, -2.0719e-03,\n -1.7477e-03, 1.4762e-03, 3.2508e-03, 5.7005e-04, -1.0620e-03,\n 5.0835e-03, 4.3334e-04, 7.9480e-04, 1.2404e-03, 1.5024e-04,\n -1.3568e-03, 8.9546e-03, 7.6717e-04, -6.6131e-03, 3.6562e-03,\n -1.5728e-03, 1.6653e-03, 6.4543e-03, 4.0747e-03, 3.6165e-03,\n -2.9014e-03, 1.2150e-03, -5.0409e-04, 7.4697e-03, -1.1703e-03,\n 4.9548e-03, 2.1722e-03, 1.6102e-03, -3.5075e-03, -6.9947e-04,\n -1.8833e-03, 5.4999e-04, -2.7217e-03, 3.9615e-04, 9.2471e-04,\n 3.3871e-03, -3.8144e-04, 5.0513e-03, -2.6286e-04, -1.4826e-03,\n 3.2191e-03, 6.8809e-03, 6.8819e-03, -1.2691e-03, 3.1317e-04,\n -1.1129e-03, -1.7891e-02, -5.0378e-03, -3.9987e-03, -3.4584e-03,\n 5.2274e-03, 2.0806e-03, 5.8934e-03, -1.8187e-03, -3.9361e-03,\n -5.7207e-03, -5.2784e-03, 1.9660e-03, 9.1639e-04, -7.6812e-04,\n 1.4469e-03, 2.1171e-04, -3.1923e-03, -1.8654e-03, 3.5053e-04,\n 4.7562e-03, -1.0611e-03, 4.4022e-04, 7.8939e-03, -6.4566e-03,\n 7.3301e-04, 4.2043e-03, -1.0248e-03, -1.1746e-03, -2.1517e-04,\n -3.4289e-03, 9.6340e-03, 1.5561e-03, -1.0316e-03, 7.4253e-03,\n -3.1056e-03, -3.1547e-03, 2.1971e-03, 3.1388e-03, 1.6758e-03,\n -3.5020e-03, 6.0903e-04, -3.9347e-03, 3.7861e-04, 5.6280e-03,\n 5.4916e-03, 8.9519e-04, -3.1340e-03, 8.7288e-03, -1.9749e-03,\n -1.6694e-02, -2.9915e-04, -1.9367e-03, 2.3445e-03, 1.3861e-03,\n 4.0618e-04, -7.9668e-03, -2.8737e-03, -3.4387e-04, 5.5689e-03,\n -6.3013e-03, 6.1378e-04, -1.8865e-03, 7.0795e-04, -4.9071e-03,\n 1.5080e-03, -1.1484e-03, 6.4996e-04, 1.6229e-03, -3.6942e-03,\n -3.1266e-03, 1.5391e-03], device='cuda:0')",
|
| 23 |
+
"exp_avg_sq": "tensor([2.0811e-04, 1.6199e-04, 1.5988e-04, 1.8633e-04, 1.2830e-04, 1.3844e-04,\n 1.1277e-04, 1.3904e-04, 1.4923e-04, 2.0569e-04, 1.1270e-04, 2.1605e-04,\n 1.0669e-04, 1.2714e-04, 9.4129e-05, 1.0768e-04, 3.0505e-04, 1.0765e-04,\n 1.1965e-04, 1.4245e-04, 1.6749e-04, 1.9722e-04, 1.7134e-04, 1.2481e-04,\n 2.1172e-04, 2.1246e-04, 1.6306e-04, 1.2766e-04, 1.0624e-04, 1.4646e-04,\n 1.6148e-04, 1.5699e-04, 2.6398e-04, 9.6198e-05, 2.1439e-04, 1.6990e-04,\n 1.5518e-04, 1.8118e-04, 8.7381e-05, 1.1744e-04, 1.4259e-04, 1.3721e-04,\n 1.0825e-04, 1.1409e-04, 1.5181e-04, 1.4561e-04, 3.2036e-04, 2.0962e-04,\n 1.8479e-04, 1.4862e-04, 1.5573e-04, 2.0701e-04, 1.2283e-04, 1.3435e-04,\n 1.9378e-04, 2.0828e-04, 1.9261e-04, 1.8780e-04, 1.3667e-04, 1.6203e-04,\n 2.0096e-04, 1.1547e-04, 1.9260e-04, 1.7393e-04, 2.0461e-04, 1.3227e-04,\n 1.8092e-04, 1.7477e-04, 1.5844e-04, 2.0669e-04, 1.3684e-04, 1.1052e-04,\n 1.7376e-04, 1.1686e-04, 1.3921e-04, 2.1414e-04, 1.5726e-04, 1.4833e-04,\n 1.8950e-04, 2.2172e-04, 1.9038e-04, 1.0981e-04, 1.8161e-04, 1.8324e-04,\n 1.3848e-04, 1.6535e-04, 2.3814e-04, 2.0610e-04, 1.0358e-04, 1.9030e-04,\n 1.5239e-04, 1.7803e-04, 1.4947e-04, 1.3803e-04, 1.4794e-04, 1.3462e-04,\n 1.9006e-04, 1.6234e-04, 1.6038e-04, 1.1580e-04, 8.8425e-05, 1.0859e-04,\n 1.3421e-04, 1.2733e-04, 1.8961e-04, 1.6383e-04, 1.6378e-04, 1.5526e-04,\n 1.5468e-04, 1.1799e-04, 1.5392e-04, 1.1527e-04, 1.6643e-04, 9.7351e-05,\n 1.7760e-04, 1.0788e-04, 9.4340e-05, 1.8485e-04, 1.3997e-04, 1.1649e-04,\n 9.1696e-05, 1.2508e-04, 1.4622e-04, 1.7585e-04, 1.5399e-04, 1.3398e-04,\n 1.2727e-04, 1.4896e-04, 1.2428e-04, 1.5994e-04, 1.1731e-04, 1.2125e-04,\n 2.0790e-04, 1.3933e-04, 1.4878e-04, 1.3438e-04, 1.0858e-04, 1.1948e-04,\n 1.6810e-04, 1.6136e-04, 2.2915e-04, 1.3804e-04, 1.0705e-04, 1.8711e-04,\n 1.4464e-04, 1.5769e-04, 1.6297e-04, 1.6272e-04, 1.5118e-04, 1.7192e-04,\n 8.8190e-05, 1.7172e-04, 1.1260e-04, 1.6669e-04, 9.6079e-05, 1.6712e-04,\n 1.6463e-04, 1.5048e-04, 2.2384e-04, 9.6649e-05, 1.6363e-04, 1.3752e-04,\n 1.8239e-04, 1.1326e-04, 1.8395e-04, 1.2057e-04, 1.6184e-04, 1.2413e-04,\n 1.1405e-04, 1.6102e-04, 2.5235e-04, 1.3140e-04, 1.8420e-04, 1.2352e-04,\n 1.4406e-04, 1.0931e-04, 1.2922e-04, 1.7075e-04, 1.1637e-04, 1.0219e-04,\n 1.5858e-04, 2.3406e-04, 1.7868e-04, 1.7634e-04, 2.1081e-04, 1.4693e-04,\n 1.7587e-04, 1.8681e-04, 1.1925e-04, 1.5289e-04, 1.6878e-04, 1.6096e-04,\n 2.3482e-04, 1.2770e-04, 1.3784e-04, 1.2564e-04, 9.4743e-05, 1.5577e-04,\n 1.0160e-04, 2.1947e-04, 1.5920e-04, 2.1226e-04, 8.5431e-05, 1.2926e-04,\n 1.4199e-04, 1.5928e-04, 1.6427e-04, 1.3713e-04, 1.6698e-04, 1.5797e-04,\n 1.1181e-04, 1.8758e-04, 1.9044e-04, 2.3810e-04, 2.0896e-04, 1.1906e-04,\n 1.6551e-04, 8.6136e-05, 1.1838e-04, 1.4741e-04, 1.3959e-04, 1.3456e-04,\n 1.5248e-04, 1.8304e-04, 1.4101e-04, 1.2703e-04, 1.4322e-04, 1.1382e-04,\n 2.5113e-04, 2.4511e-04, 1.2619e-04, 1.2446e-04, 1.2307e-04, 1.1317e-04,\n 1.4444e-04, 1.8872e-04, 1.8568e-04, 2.2521e-04, 1.7407e-04, 2.1511e-04,\n 1.3268e-04, 1.3866e-04, 1.7271e-04, 1.2107e-04, 1.7156e-04, 1.4481e-04,\n 1.2231e-04, 8.3566e-05, 1.4464e-04, 2.4341e-04, 1.2028e-04, 1.4968e-04,\n 1.2731e-04, 2.5819e-04, 1.5446e-04, 1.3139e-04, 3.5766e-04, 1.5389e-04,\n 1.6381e-04, 1.7102e-04, 1.3123e-04, 1.8391e-04, 1.4864e-04, 1.6171e-04,\n 1.9716e-04, 1.8075e-04, 1.1445e-04, 1.6375e-04, 1.1136e-04, 1.1922e-04,\n 1.3131e-04, 1.0672e-04, 1.2211e-04, 1.0189e-04, 1.1956e-04, 2.7604e-04,\n 1.7143e-04, 1.6266e-04, 2.3903e-04, 1.4331e-04, 7.8775e-05, 1.9395e-04,\n 1.4316e-04, 1.3149e-04, 2.4827e-04, 1.2548e-04, 2.5972e-04, 1.0715e-04,\n 1.6143e-04, 1.6152e-04, 1.8626e-04, 1.8076e-04, 1.8591e-04, 1.4456e-04,\n 2.1388e-04, 1.5856e-04, 1.3125e-04, 2.1669e-04, 1.4388e-04, 2.1310e-04,\n 2.2238e-04, 1.1820e-04, 1.2292e-04, 1.9509e-04, 1.1183e-04, 1.4582e-04,\n 1.5021e-04, 1.1572e-04, 1.6947e-04, 1.3762e-04, 1.2009e-04, 1.1161e-04,\n 1.0771e-04, 1.0221e-04, 1.6575e-04, 1.7704e-04, 8.7361e-05, 2.0265e-04,\n 1.2690e-04, 1.7051e-04, 1.1365e-04, 2.6059e-04, 1.4409e-04, 1.2414e-04,\n 1.1335e-04, 1.1154e-04, 1.4607e-04, 1.6318e-04, 7.3859e-05, 1.4483e-04,\n 1.6814e-04, 1.9932e-04, 1.2423e-04, 1.1306e-04, 1.9313e-04, 1.5970e-04,\n 1.3024e-04, 2.2330e-04, 1.7998e-04, 1.4522e-04, 1.4680e-04, 1.9191e-04,\n 1.1679e-04, 1.2284e-04, 1.3364e-04, 1.6138e-04, 1.0895e-04, 1.1338e-04,\n 1.8360e-04, 7.1230e-05, 1.7824e-04, 1.5792e-04, 8.4801e-05, 1.3265e-04,\n 1.3279e-04, 1.2727e-04, 9.4857e-05, 1.2851e-04, 2.2151e-04, 1.0622e-04,\n 1.2396e-04, 8.5136e-05, 1.2759e-04, 1.1785e-04, 9.4623e-05, 1.3759e-04,\n 1.1001e-04, 2.1568e-04, 1.1927e-04, 1.2080e-04, 1.6187e-04, 2.0637e-04,\n 1.5654e-04, 1.4947e-04, 1.1636e-04, 1.2524e-04, 1.2478e-04, 1.2650e-04,\n 1.4276e-04, 1.3377e-04, 8.3176e-05, 1.2996e-04, 1.7703e-04, 1.7965e-04,\n 1.4305e-04, 8.6871e-05, 1.2554e-04, 1.0437e-04, 1.1914e-04, 2.0765e-04,\n 1.6326e-04, 1.5105e-04, 1.5966e-04, 1.1654e-04, 1.5634e-04, 2.0461e-04,\n 1.3963e-04, 1.5798e-04, 2.4313e-04, 1.3119e-04, 1.2345e-04, 1.2504e-04,\n 1.1810e-04, 1.2028e-04, 1.1941e-04, 1.4706e-04, 1.4241e-04, 1.1106e-04,\n 1.4870e-04, 1.8341e-04, 1.5711e-04, 1.1041e-04, 1.8669e-04, 2.2229e-04,\n 1.3051e-04, 1.4665e-04, 1.1351e-04, 1.0196e-04, 1.6289e-04, 1.5941e-04,\n 1.6579e-04, 1.2131e-04, 1.5280e-04, 1.6764e-04, 1.5254e-04, 1.3181e-04,\n 1.7992e-04, 1.3109e-04, 1.4379e-04, 1.4258e-04, 1.2104e-04, 1.8586e-04,\n 1.5899e-04, 2.6683e-04, 1.1901e-04, 1.7940e-04, 1.0469e-04, 2.0514e-04,\n 1.6810e-04, 1.2812e-04, 1.6077e-04, 2.2635e-04, 1.3917e-04, 1.7051e-04,\n 1.2242e-04, 1.0582e-04, 1.3639e-04, 1.5161e-04, 1.5401e-04, 1.7317e-04,\n 1.6714e-04, 1.5746e-04, 9.5169e-05, 2.1095e-04, 1.7495e-04, 1.8143e-04,\n 1.0963e-04, 2.2253e-04, 1.5626e-04, 1.7260e-04, 1.8655e-04, 1.6421e-04,\n 1.5853e-04, 1.2386e-04, 1.6779e-04, 1.7795e-04, 1.1686e-04, 1.0949e-04,\n 1.3296e-04, 1.2812e-04, 1.3082e-04, 1.6115e-04, 1.4588e-04, 1.7207e-04,\n 1.5204e-04, 1.5453e-04, 1.4421e-04, 1.6905e-04, 8.1649e-05, 1.2123e-04,\n 1.0467e-04, 9.4631e-05, 1.5379e-04, 1.3523e-04, 1.1629e-04, 1.9365e-04,\n 9.3331e-05, 1.3464e-04, 1.8830e-04, 1.5379e-04, 5.5830e-04, 1.3540e-04,\n 1.6515e-04, 1.3888e-04, 1.8580e-04, 2.1674e-04, 2.2447e-04, 1.9341e-04,\n 1.4553e-04, 1.7603e-04, 1.2972e-04, 1.2806e-04, 1.4265e-04, 1.1037e-04,\n 1.8264e-04, 2.1676e-04, 1.9029e-04, 1.3763e-04, 1.5466e-04, 1.4290e-04,\n 2.2236e-04, 1.2509e-04], device='cuda:0')"
|
| 24 |
+
},
|
| 25 |
+
"4": {
|
| 26 |
+
"step": "tensor(8764.)",
|
| 27 |
+
"exp_avg": "tensor([[-6.6012e-05, 3.9669e-05, 1.8250e-04, ..., -4.8578e-05,\n -1.0647e-04, 4.3771e-05],\n [-2.0590e-05, 3.8642e-04, -7.4833e-05, ..., 1.7502e-04,\n -3.4472e-04, -9.1284e-05],\n [-1.6215e-04, 8.8511e-05, 4.7994e-04, ..., 1.7969e-04,\n 1.4575e-04, 9.5616e-04],\n ...,\n [ 2.6037e-05, 1.9928e-05, -4.4082e-04, ..., -5.8299e-05,\n -2.3242e-04, 1.7487e-05],\n [ 7.8950e-05, -3.6032e-05, -1.5033e-05, ..., 1.0474e-04,\n -8.7855e-05, 1.1673e-06],\n [-9.7676e-05, -2.7155e-05, 8.3371e-05, ..., 1.9306e-04,\n 2.0368e-04, -1.1948e-04]], device='cuda:0')",
|
| 28 |
+
"exp_avg_sq": "tensor([[2.9796e-07, 4.8296e-07, 1.1026e-07, ..., 2.2397e-07, 2.6441e-07,\n 3.8608e-07],\n [5.9741e-07, 1.1644e-06, 3.7930e-07, ..., 7.4063e-07, 7.6337e-07,\n 9.9018e-07],\n [7.0944e-07, 3.7636e-07, 3.2120e-07, ..., 7.9120e-07, 4.1355e-07,\n 7.9955e-07],\n ...,\n [4.5589e-07, 1.1186e-06, 4.7918e-07, ..., 6.0954e-07, 9.6010e-07,\n 8.2623e-07],\n [4.4050e-07, 8.1701e-07, 2.5761e-07, ..., 8.6424e-07, 5.5936e-07,\n 6.6472e-07],\n [4.0324e-07, 5.7054e-07, 3.6762e-07, ..., 1.2668e-06, 4.6971e-07,\n 8.1087e-07]], device='cuda:0')"
|
| 29 |
+
},
|
| 30 |
+
"5": {
|
| 31 |
+
"step": "tensor(8764.)",
|
| 32 |
+
"exp_avg": "tensor([[ 1.0938e-05, -1.5379e-05, 9.7786e-05, ..., -1.1924e-04,\n -3.9572e-05, 2.2988e-05],\n [ 1.0972e-04, 2.4915e-04, 1.1523e-04, ..., 4.1636e-05,\n -2.3503e-04, -1.5411e-04],\n [-8.8508e-05, -3.8149e-05, 5.3010e-04, ..., 2.6580e-05,\n 1.0924e-04, 4.9948e-04],\n ...,\n [-1.4922e-04, -2.2232e-04, -1.5833e-04, ..., 1.2746e-05,\n 2.1103e-04, 1.1412e-04],\n [ 1.6357e-04, 1.2076e-04, -1.0500e-04, ..., 1.6955e-04,\n 8.8701e-05, -3.8843e-04],\n [-8.3054e-05, 7.5673e-05, -1.1961e-04, ..., 1.3030e-04,\n -2.0520e-04, 3.5107e-04]], device='cuda:0')",
|
| 33 |
+
"exp_avg_sq": "tensor([[1.0261e-07, 8.7625e-08, 9.8434e-08, ..., 1.1874e-07, 9.1794e-08,\n 1.2444e-07],\n [2.0114e-07, 4.4795e-07, 1.7925e-07, ..., 1.8601e-07, 2.5437e-07,\n 2.8999e-07],\n [2.6550e-07, 1.5233e-07, 2.0797e-07, ..., 3.8242e-07, 3.0396e-07,\n 2.7454e-07],\n ...,\n [2.1556e-07, 5.3854e-07, 8.8819e-08, ..., 4.3871e-07, 3.1187e-07,\n 2.9636e-07],\n [2.3131e-07, 3.0611e-07, 7.8379e-08, ..., 3.6081e-07, 2.3129e-07,\n 2.5500e-07],\n [2.5510e-07, 1.5700e-07, 4.1064e-07, ..., 2.1228e-07, 1.7715e-07,\n 2.7903e-07]], device='cuda:0')"
|
| 34 |
+
},
|
| 35 |
+
"6": {
|
| 36 |
+
"step": "tensor(8764.)",
|
| 37 |
+
"exp_avg": "tensor([ 0.0004, -0.0004], device='cuda:0')",
|
| 38 |
+
"exp_avg_sq": "tensor([7.8745e-06, 7.8744e-06], device='cuda:0')"
|
| 39 |
+
}
|
| 40 |
+
},
|
| 41 |
+
"param_groups": [
|
| 42 |
+
{
|
| 43 |
+
"lr": 0.00020690126647990973,
|
| 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": 0.00020690126647990973,
|
| 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": 0.00020690126647990973,
|
| 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": 0.00010384757955302797,
|
| 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": 7,
|
| 137 |
+
"base_lrs": [
|
| 138 |
+
0.001,
|
| 139 |
+
0.001,
|
| 140 |
+
0.001,
|
| 141 |
+
0.0005
|
| 142 |
+
],
|
| 143 |
+
"last_epoch": 7,
|
| 144 |
+
"_step_count": 0,
|
| 145 |
+
"_is_initial": false,
|
| 146 |
+
"_get_lr_called_within_step": false,
|
| 147 |
+
"_last_lr": [
|
| 148 |
+
0.00020690126647990973,
|
| 149 |
+
0.00020690126647990973,
|
| 150 |
+
0.00020690126647990973,
|
| 151 |
+
0.00010384757955302797
|
| 152 |
+
]
|
| 153 |
+
},
|
| 154 |
+
"metrics": {
|
| 155 |
+
"best_val_acc": 71.766,
|
| 156 |
+
"best_epoch": 6,
|
| 157 |
+
"scale_accuracies": {
|
| 158 |
+
"256": 71.436,
|
| 159 |
+
"512": 71.578
|
| 160 |
+
},
|
| 161 |
+
"training_history": {
|
| 162 |
+
"epochs": [
|
| 163 |
+
1,
|
| 164 |
+
2,
|
| 165 |
+
3,
|
| 166 |
+
4,
|
| 167 |
+
5,
|
| 168 |
+
6,
|
| 169 |
+
7
|
| 170 |
+
],
|
| 171 |
+
"train_loss": [
|
| 172 |
+
5.603825211905824,
|
| 173 |
+
4.145695094292918,
|
| 174 |
+
3.767625377201044,
|
| 175 |
+
3.572572173021091,
|
| 176 |
+
3.45402966711087,
|
| 177 |
+
3.3684399169854844,
|
| 178 |
+
3.3046513376906277
|
| 179 |
+
],
|
| 180 |
+
"train_acc": [
|
| 181 |
+
63.39829233815732,
|
| 182 |
+
69.48282308239285,
|
| 183 |
+
70.34266414917025,
|
| 184 |
+
70.91120829681064,
|
| 185 |
+
71.35642738222262,
|
| 186 |
+
71.84090754757186,
|
| 187 |
+
72.29198067074785
|
| 188 |
+
],
|
| 189 |
+
"val_acc": [
|
| 190 |
+
68.056,
|
| 191 |
+
69.624,
|
| 192 |
+
70.142,
|
| 193 |
+
70.622,
|
| 194 |
+
70.966,
|
| 195 |
+
71.462,
|
| 196 |
+
71.766
|
| 197 |
+
],
|
| 198 |
+
"scale_accs": {
|
| 199 |
+
"256": [
|
| 200 |
+
66.924,
|
| 201 |
+
68.828,
|
| 202 |
+
69.54,
|
| 203 |
+
69.996,
|
| 204 |
+
70.468,
|
| 205 |
+
71.106,
|
| 206 |
+
71.436
|
| 207 |
+
],
|
| 208 |
+
"512": [
|
| 209 |
+
67.882,
|
| 210 |
+
69.572,
|
| 211 |
+
70.028,
|
| 212 |
+
70.482,
|
| 213 |
+
70.854,
|
| 214 |
+
71.28,
|
| 215 |
+
71.578
|
| 216 |
+
]
|
| 217 |
+
},
|
| 218 |
+
"lr": [
|
| 219 |
+
0.0009755527298894294,
|
| 220 |
+
0.0009046039886902864,
|
| 221 |
+
0.0007940987335200904,
|
| 222 |
+
0.0006548539886902864,
|
| 223 |
+
0.0005005000000000001,
|
| 224 |
+
0.0003461460113097139,
|
| 225 |
+
0.00020690126647990973
|
| 226 |
+
]
|
| 227 |
+
}
|
| 228 |
+
},
|
| 229 |
+
"train_config": {
|
| 230 |
+
"name": "david_training",
|
| 231 |
+
"run_id": "20251012_135249",
|
| 232 |
+
"dataset_name": "AbstractPhil/imagenet-clip-features-orderly",
|
| 233 |
+
"model_variant": "clip_vit_laion_b32",
|
| 234 |
+
"num_classes": 1000,
|
| 235 |
+
"preset": "small_fast",
|
| 236 |
+
"custom_config_path": null,
|
| 237 |
+
"num_classes_override": null,
|
| 238 |
+
"use_belly_override": null,
|
| 239 |
+
"belly_expand_override": null,
|
| 240 |
+
"progressive_training_override": false,
|
| 241 |
+
"scale_warmup_epochs_override": null,
|
| 242 |
+
"num_epochs": 10,
|
| 243 |
+
"batch_size": 1024,
|
| 244 |
+
"learning_rate": 0.001,
|
| 245 |
+
"weight_decay": 1e-05,
|
| 246 |
+
"warmup_epochs": 3,
|
| 247 |
+
"use_rose_loss": true,
|
| 248 |
+
"rose_initial_weight": 0.1,
|
| 249 |
+
"rose_max_weight": 0.5,
|
| 250 |
+
"rose_weight_schedule": "adaptive",
|
| 251 |
+
"use_cayley_loss": false,
|
| 252 |
+
"cayley_weight": 0.001,
|
| 253 |
+
"scale_loss_balance": null,
|
| 254 |
+
"use_mixed_precision": false,
|
| 255 |
+
"gradient_clip": 10.0,
|
| 256 |
+
"scheduler_type": "cosine_restarts",
|
| 257 |
+
"min_lr": 1e-06,
|
| 258 |
+
"freeze_strategy": "never",
|
| 259 |
+
"freeze_threshold": 90.0,
|
| 260 |
+
"unfreeze_on_plateau": true,
|
| 261 |
+
"patience": 10,
|
| 262 |
+
"track_gradients": true,
|
| 263 |
+
"gradient_scale_threshold": 1e-05,
|
| 264 |
+
"gradient_scale_multiplier": 10.0,
|
| 265 |
+
"log_interval": 50,
|
| 266 |
+
"val_interval": 1,
|
| 267 |
+
"save_interval": 5,
|
| 268 |
+
"log_fusion_weights": true,
|
| 269 |
+
"log_loss_components": true,
|
| 270 |
+
"save_format": "safetensors",
|
| 271 |
+
"hf_repo": "AbstractPhil/gated-david",
|
| 272 |
+
"upload_to_hub": true,
|
| 273 |
+
"base_dir": "./david_training",
|
| 274 |
+
"num_workers": 10,
|
| 275 |
+
"pin_memory": true,
|
| 276 |
+
"prefetch_factor": 4,
|
| 277 |
+
"persistent_workers": true
|
| 278 |
+
}
|
| 279 |
+
}
|
weights/David-fully_shared-weighted_sum/20251012_135249/training_history.json
CHANGED
|
@@ -5,7 +5,8 @@
|
|
| 5 |
3,
|
| 6 |
4,
|
| 7 |
5,
|
| 8 |
-
6
|
|
|
|
| 9 |
],
|
| 10 |
"train_loss": [
|
| 11 |
5.603825211905824,
|
|
@@ -13,7 +14,8 @@
|
|
| 13 |
3.767625377201044,
|
| 14 |
3.572572173021091,
|
| 15 |
3.45402966711087,
|
| 16 |
-
3.3684399169854844
|
|
|
|
| 17 |
],
|
| 18 |
"train_acc": [
|
| 19 |
63.39829233815732,
|
|
@@ -21,7 +23,8 @@
|
|
| 21 |
70.34266414917025,
|
| 22 |
70.91120829681064,
|
| 23 |
71.35642738222262,
|
| 24 |
-
71.84090754757186
|
|
|
|
| 25 |
],
|
| 26 |
"val_acc": [
|
| 27 |
68.056,
|
|
@@ -29,7 +32,8 @@
|
|
| 29 |
70.142,
|
| 30 |
70.622,
|
| 31 |
70.966,
|
| 32 |
-
71.462
|
|
|
|
| 33 |
],
|
| 34 |
"scale_accs": {
|
| 35 |
"256": [
|
|
@@ -38,7 +42,8 @@
|
|
| 38 |
69.54,
|
| 39 |
69.996,
|
| 40 |
70.468,
|
| 41 |
-
71.106
|
|
|
|
| 42 |
],
|
| 43 |
"512": [
|
| 44 |
67.882,
|
|
@@ -46,7 +51,8 @@
|
|
| 46 |
70.028,
|
| 47 |
70.482,
|
| 48 |
70.854,
|
| 49 |
-
71.28
|
|
|
|
| 50 |
]
|
| 51 |
},
|
| 52 |
"lr": [
|
|
@@ -55,6 +61,7 @@
|
|
| 55 |
0.0007940987335200904,
|
| 56 |
0.0006548539886902864,
|
| 57 |
0.0005005000000000001,
|
| 58 |
-
0.0003461460113097139
|
|
|
|
| 59 |
]
|
| 60 |
}
|
|
|
|
| 5 |
3,
|
| 6 |
4,
|
| 7 |
5,
|
| 8 |
+
6,
|
| 9 |
+
7
|
| 10 |
],
|
| 11 |
"train_loss": [
|
| 12 |
5.603825211905824,
|
|
|
|
| 14 |
3.767625377201044,
|
| 15 |
3.572572173021091,
|
| 16 |
3.45402966711087,
|
| 17 |
+
3.3684399169854844,
|
| 18 |
+
3.3046513376906277
|
| 19 |
],
|
| 20 |
"train_acc": [
|
| 21 |
63.39829233815732,
|
|
|
|
| 23 |
70.34266414917025,
|
| 24 |
70.91120829681064,
|
| 25 |
71.35642738222262,
|
| 26 |
+
71.84090754757186,
|
| 27 |
+
72.29198067074785
|
| 28 |
],
|
| 29 |
"val_acc": [
|
| 30 |
68.056,
|
|
|
|
| 32 |
70.142,
|
| 33 |
70.622,
|
| 34 |
70.966,
|
| 35 |
+
71.462,
|
| 36 |
+
71.766
|
| 37 |
],
|
| 38 |
"scale_accs": {
|
| 39 |
"256": [
|
|
|
|
| 42 |
69.54,
|
| 43 |
69.996,
|
| 44 |
70.468,
|
| 45 |
+
71.106,
|
| 46 |
+
71.436
|
| 47 |
],
|
| 48 |
"512": [
|
| 49 |
67.882,
|
|
|
|
| 51 |
70.028,
|
| 52 |
70.482,
|
| 53 |
70.854,
|
| 54 |
+
71.28,
|
| 55 |
+
71.578
|
| 56 |
]
|
| 57 |
},
|
| 58 |
"lr": [
|
|
|
|
| 61 |
0.0007940987335200904,
|
| 62 |
0.0006548539886902864,
|
| 63 |
0.0005005000000000001,
|
| 64 |
+
0.0003461460113097139,
|
| 65 |
+
0.00020690126647990973
|
| 66 |
]
|
| 67 |
}
|