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_acc70.97.safetensors +3 -0
- weights/David-fully_shared-weighted_sum/20251012_135249/best_model_acc70.97_metadata.json +265 -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:
|
| 19 |
-
Best Validation Accuracy: 70.
|
| 20 |
|
| 21 |
Per-Scale Accuracies:
|
| 22 |
-
• Scale 256: 70.
|
| 23 |
-
• Scale 512: 70.
|
| 24 |
|
| 25 |
═══════════════════════════════════════════════════════════════
|
| 26 |
|
|
@@ -42,17 +42,18 @@ Epoch | Train Acc | Val Acc | Learning Rate
|
|
| 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%
|
|
|
|
| 46 |
|
| 47 |
═══════════════════════════════════════════════════════════════
|
| 48 |
|
| 49 |
📁 FILES
|
| 50 |
|
| 51 |
-
Best Model: best_model_acc70.
|
| 52 |
Config: david_config.json
|
| 53 |
Training Cfg: train_config.json
|
| 54 |
History: training_history.json
|
| 55 |
|
| 56 |
═══════════════════════════════════════════════════════════════
|
| 57 |
|
| 58 |
-
Generated: 2025-10-12
|
|
|
|
| 3 |
║ DAVID MODEL SUMMARY ║
|
| 4 |
╠══════════════════════════════════════════════════════════════╣
|
| 5 |
║ ║
|
| 6 |
+
║ 🎯 VALIDATION ACCURACY: 70.97% ║
|
| 7 |
║ ║
|
| 8 |
╚══════════════════════════════════════════════════════════════╝
|
| 9 |
|
| 10 |
MODEL: David-fully_shared-weighted_sum
|
| 11 |
RUN ID: 20251012_135249
|
| 12 |
+
BEST EPOCH: 5/10
|
| 13 |
|
| 14 |
═══════════════════════════════════════════════════════════════
|
| 15 |
|
| 16 |
📊 PERFORMANCE BREAKDOWN
|
| 17 |
|
| 18 |
+
Final Training Accuracy: 71.36%
|
| 19 |
+
Best Validation Accuracy: 70.97%
|
| 20 |
|
| 21 |
Per-Scale Accuracies:
|
| 22 |
+
• Scale 256: 70.47%
|
| 23 |
+
• Scale 512: 70.85%
|
| 24 |
|
| 25 |
═══════════════════════════════════════════════════════════════
|
| 26 |
|
|
|
|
| 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 |
+
5 | 71.36% | 70.97% 👑 | 5.01e-04
|
| 47 |
|
| 48 |
═══════════════════════════════════════════════════════════════
|
| 49 |
|
| 50 |
📁 FILES
|
| 51 |
|
| 52 |
+
Best Model: best_model_acc70.97.safetensors
|
| 53 |
Config: david_config.json
|
| 54 |
Training Cfg: train_config.json
|
| 55 |
History: training_history.json
|
| 56 |
|
| 57 |
═══════════════════════════════════════════════════════════════
|
| 58 |
|
| 59 |
+
Generated: 2025-10-12 14:01:18
|
weights/David-fully_shared-weighted_sum/20251012_135249/best_model_acc70.97.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e253e3c4ed2796eddea7ecbfb4448cc5e2e6f5e7a986e9753f5e744d29362bd6
|
| 3 |
+
size 2628344
|
weights/David-fully_shared-weighted_sum/20251012_135249/best_model_acc70.97_metadata.json
ADDED
|
@@ -0,0 +1,265 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4,
|
| 3 |
+
"optimizer_state_dict": {
|
| 4 |
+
"state": {
|
| 5 |
+
"0": {
|
| 6 |
+
"step": "tensor(6260.)",
|
| 7 |
+
"exp_avg": "tensor([[ 2.9444e-04, 1.5816e-03, 1.7542e-03, ..., -1.8502e-05,\n 3.1823e-04, -8.5925e-05],\n [ 6.3246e-04, -8.8365e-04, -4.6243e-04, ..., -4.8646e-04,\n -5.8700e-05, -9.1580e-04],\n [-1.7990e-04, 5.6824e-04, 6.4436e-04, ..., 8.4981e-04,\n 1.9581e-04, 2.8337e-04],\n ...,\n [ 2.9325e-04, 1.6735e-03, 5.3630e-04, ..., -5.0044e-04,\n -2.3893e-04, -4.7796e-04],\n [-2.7429e-04, -1.3650e-03, -6.4345e-04, ..., -9.2895e-04,\n -5.5495e-04, 1.2926e-04],\n [-2.7410e-05, 2.0947e-03, 2.5894e-04, ..., 4.6536e-04,\n -1.1696e-04, -7.2882e-04]], device='cuda:0')",
|
| 8 |
+
"exp_avg_sq": "tensor([[6.4630e-06, 2.9433e-05, 1.4768e-05, ..., 6.2504e-06, 3.0551e-06,\n 8.1732e-06],\n [6.1206e-06, 2.6945e-05, 1.7313e-05, ..., 9.0072e-06, 2.4589e-06,\n 1.0405e-05],\n [9.0886e-06, 5.7423e-05, 1.7951e-05, ..., 5.1766e-06, 2.9185e-06,\n 3.6822e-06],\n ...,\n [5.8958e-06, 2.6014e-05, 1.4828e-05, ..., 3.4494e-06, 1.8928e-06,\n 2.6287e-06],\n [6.3857e-06, 3.9066e-05, 1.4138e-05, ..., 6.0668e-06, 3.3456e-06,\n 5.9265e-06],\n [4.2426e-06, 4.7758e-05, 1.1723e-05, ..., 3.4353e-06, 2.1747e-06,\n 2.4251e-06]], device='cuda:0')"
|
| 9 |
+
},
|
| 10 |
+
"1": {
|
| 11 |
+
"step": "tensor(6260.)",
|
| 12 |
+
"exp_avg": "tensor([ 1.6115e-02, -1.2985e-02, 1.3522e-02, -3.4347e-03, 2.7610e-02,\n -6.4253e-03, 5.1431e-03, -9.6191e-03, 5.5197e-02, -9.4616e-03,\n -7.9327e-03, -5.7669e-03, -1.4809e-02, 7.2589e-03, -1.0305e-02,\n -6.1531e-03, -2.0230e-02, 1.5248e-02, -1.5123e-03, 8.2109e-03,\n -3.7036e-02, 3.0154e-03, -3.9827e-03, 2.2456e-02, 1.5288e-02,\n 1.5090e-02, 7.3689e-04, -5.0872e-03, 5.1057e-03, 4.5780e-03,\n 1.4943e-02, 1.1029e-02, -4.9649e-02, -1.8189e-02, -2.0830e-02,\n 3.8876e-03, -1.3802e-02, 2.7383e-02, -9.3568e-04, 3.2487e-02,\n -1.3262e-03, 2.0812e-03, -1.4976e-02, -1.5549e-02, 7.6433e-03,\n 1.3117e-02, 2.9753e-02, -1.4443e-02, 1.4578e-02, 3.1087e-03,\n -6.7807e-03, 4.4136e-03, -1.4398e-03, -8.2912e-03, 3.8167e-02,\n 3.4669e-02, -1.2520e-02, 2.6787e-02, 6.4173e-03, -9.7896e-03,\n -1.4449e-02, -2.5027e-02, -1.1802e-02, -1.4157e-02, -4.6948e-02,\n -1.8013e-03, 1.2781e-02, 1.1981e-02, -3.2382e-02, 6.2777e-03,\n -3.7741e-02, -1.3979e-02, 1.0531e-02, 4.4233e-03, 1.0183e-02,\n 4.4824e-02, 1.4818e-03, 1.6187e-02, -2.8051e-02, -1.7776e-02,\n 2.5563e-03, -2.9371e-02, 2.1952e-02, -1.0130e-02, -8.0774e-03,\n 1.4117e-02, 3.1835e-03, 4.5456e-02, -9.2051e-02, -1.7981e-02,\n 5.7988e-03, -9.1110e-03, 8.7453e-03, 1.9439e-02, 2.5491e-03,\n -2.6916e-03, 2.9073e-02, -1.6220e-02, 1.3789e-02, 1.4122e-02,\n 2.6584e-02, 1.9143e-02, -2.8918e-02, 4.4204e-03, -4.8768e-03,\n -1.2611e-02, 2.8093e-02, 2.4462e-03, -4.9543e-03, 3.2457e-02,\n 5.6954e-03, -2.2098e-02, -3.4211e-03, 5.7417e-05, -2.6476e-02,\n -1.1746e-02, -8.6627e-03, -1.0888e-02, -1.0258e-02, -3.0225e-02,\n -2.3794e-02, -1.9605e-02, 1.7887e-02, 2.2582e-02, 3.3853e-02,\n 2.1443e-03, -1.0930e-01, -4.1052e-04, -9.5391e-03, -9.1283e-03,\n -2.4861e-03, 1.2341e-02, -6.0640e-04, 6.2284e-04, 5.5166e-03,\n -3.6436e-03, -1.1276e-02, -4.5154e-03, -1.0218e-02, -2.9869e-02,\n -1.2918e-02, -1.2135e-02, 3.9888e-03, -2.4295e-02, 1.9621e-02,\n -1.7784e-02, 1.8933e-02, -3.5194e-03, -4.7470e-03, -4.8891e-02,\n -3.1026e-03, 1.1918e-03, 3.6263e-02, -2.2425e-02, 8.8911e-03,\n -5.4494e-03, -2.1431e-02, -4.6875e-02, 1.8209e-02, 1.0216e-02,\n -3.1670e-02, 3.3724e-02, -1.4010e-02, 1.5851e-02, -1.8492e-03,\n 3.7227e-03, 7.0334e-04, 1.9339e-02, 4.7613e-03, -4.6168e-03,\n 3.9991e-02, 1.7090e-02, -1.2667e-03, 2.0163e-02, -2.3835e-02,\n 1.8272e-02, -8.4380e-03, -1.1406e-02, 1.1542e-02, 3.9105e-03,\n 1.9050e-02, 3.2235e-02, 2.7709e-02, -1.7614e-02, -2.8529e-02,\n 1.7852e-03, 2.6043e-03, 2.0261e-02, -1.3606e-04, 3.8334e-02,\n -2.4089e-02, 2.4752e-03, 3.7164e-02, 1.0339e-02, -1.3433e-02,\n 4.3978e-04, 8.2285e-03, -7.7802e-03, -1.2425e-02, -2.4511e-02,\n 7.1996e-03, 1.1630e-02, 8.1733e-03, 5.5850e-03, 3.6392e-03,\n -8.5827e-03, -3.0480e-02, 7.1696e-04, -9.1731e-03, 2.5003e-02,\n -2.2617e-03, -1.8061e-02, 7.2274e-04, -2.9266e-03, -1.1289e-02,\n 5.1229e-03, -1.8274e-02, 1.6052e-02, 2.3634e-02, 1.0965e-03,\n -8.5185e-05, 4.9079e-03, 5.2235e-03, 1.6356e-02, 1.4129e-02,\n 4.0315e-03, 8.7812e-03, 1.1985e-02, 1.8473e-02, 4.1082e-02,\n -5.8017e-03, -5.2495e-03, 3.3375e-02, -5.5583e-03, -3.1185e-02,\n -6.4399e-03, -4.4043e-02, 1.9865e-02, 2.5326e-03, -3.4581e-03,\n 1.5985e-02, -2.8722e-02, 5.2615e-02, -6.9503e-03, -8.7879e-03,\n 1.1712e-02, -3.3142e-02, -6.6450e-04, -3.9057e-03, -1.2059e-02,\n 2.3508e-03, -4.9697e-03, -2.2910e-02, 8.0046e-03, 4.9391e-03,\n 1.7316e-02, 1.6587e-03, -1.5813e-02, 2.4640e-02, -1.7447e-03,\n 2.2804e-03, 1.0307e-02, -1.0328e-02, 1.6188e-02, 1.2374e-03,\n 3.0567e-03, -1.2098e-02, -1.6305e-03, -2.4107e-02, 1.0067e-02,\n 1.4043e-02, 2.2665e-03, 3.1936e-02, -1.2431e-02, -1.2969e-02,\n -1.7437e-02, -1.0231e-02, 6.7890e-03, -3.9469e-02, 9.2027e-03,\n 4.7474e-04, 4.3215e-02, -2.3018e-02, 2.4652e-02, 9.3073e-03,\n 5.7134e-03, -9.1564e-03, 1.1889e-02, 7.1170e-03, -1.8092e-04,\n 5.5591e-02, -1.2029e-02, 1.6430e-02, 2.7878e-04, 6.7954e-03,\n -3.8138e-02, -7.4170e-03, 2.2477e-02, -1.9455e-03, -3.3713e-03,\n -9.1756e-04, 9.3397e-03, 1.6290e-04, 6.8894e-03, -1.6829e-02,\n 5.5790e-03, -1.7707e-02, 1.1850e-02, 8.8881e-03, 1.5580e-02,\n 1.8081e-04, 1.6149e-02, 7.2095e-03, -1.1379e-02, -4.3937e-03,\n 1.0624e-02, -1.6718e-02, 3.7770e-03, -8.9548e-03, -1.9755e-03,\n -4.6450e-03, 7.5443e-03, -1.1950e-02, 6.0382e-03, 9.9748e-04,\n -2.4615e-03, -3.4125e-02, 1.0103e-02, -5.9484e-03, 1.4564e-02,\n -5.9025e-02, -6.4889e-03, 6.5561e-03, 8.4681e-03, 1.9702e-03,\n 1.3516e-02, 8.7004e-04, -3.0895e-02, 5.1054e-03, -1.4017e-02,\n -7.8334e-03, -1.1487e-02, 2.2120e-02, -2.0394e-02, -2.7465e-02,\n 3.6959e-02, -2.1590e-02, 1.1106e-02, -4.8092e-02, 2.7778e-03,\n 2.3124e-02, 3.4452e-02, 2.2668e-02, 3.5217e-02, 1.0120e-02,\n -1.3117e-03, 2.9213e-04, -1.5924e-02, -1.4991e-02, 7.9924e-03,\n 8.3043e-03, 1.6207e-02, -1.8056e-02, -2.5667e-02, -1.7779e-03,\n 2.1305e-03, -7.7397e-03, -2.0501e-02, -4.0559e-02, -1.4185e-03,\n -2.5093e-02, 9.0843e-03, -3.5887e-03, -1.2105e-02, 3.6400e-03,\n 7.8115e-03, 9.3263e-03, 2.5467e-02, -1.2592e-03, 1.3144e-02,\n -2.8729e-02, 1.5775e-02, 1.5583e-02, 5.4556e-03, -3.5615e-02,\n -3.5484e-02, 8.8021e-03, 6.0957e-03, -2.7829e-02, -5.0886e-03,\n 8.3988e-04, -1.5455e-02, 3.1264e-02, 1.5601e-02, -3.0053e-02,\n 8.6765e-03, -1.9134e-02, -6.6015e-04, -3.7888e-03, -1.1400e-02,\n -1.0299e-03, 2.2310e-03, 5.2498e-03, -2.2775e-03, 3.0730e-03,\n 9.8790e-03, 3.1031e-03, -5.0240e-02, 3.9615e-03, 8.3728e-03,\n -1.8132e-02, 1.3670e-02, 1.8238e-02, -2.0456e-02, 6.0309e-02,\n 2.2611e-02, -4.4193e-02, -1.8315e-02, 2.4152e-02, -1.8515e-02,\n 9.9170e-05, 1.5196e-02, 1.9607e-03, 4.2453e-02, -1.1663e-02,\n 4.1528e-03, -7.6586e-04, -7.2375e-03, 9.3575e-03, 1.2261e-02,\n -4.7056e-02, -2.1087e-02, 3.1696e-02, 1.7916e-02, 1.6197e-02,\n -7.1714e-03, 2.8493e-03, 1.3111e-02, 2.4999e-02, -9.9002e-03,\n 1.0854e-03, 7.6700e-03, 1.0273e-02, -3.8370e-03, -1.7352e-02,\n 3.4324e-02, -2.5153e-02, 2.8595e-02, -3.1866e-02, 6.5098e-03,\n -1.5499e-02, -1.6456e-02, 1.3802e-02, 5.2119e-03, 2.8392e-03,\n -1.6928e-02, 2.4582e-02, -9.7368e-03, 1.9703e-02, -1.1195e-02,\n -2.5874e-02, -2.1906e-03, 1.7107e-03, -6.9095e-03, 5.5746e-03,\n 3.3146e-02, -1.4083e-02, -4.6171e-04, 3.4020e-02, 4.4078e-03,\n 1.2891e-02, -5.0650e-03, -4.6103e-03, 1.6155e-02, -1.8039e-02,\n -1.7694e-02, 3.3280e-02, -9.6694e-03, -2.1665e-02, -3.3202e-02,\n -3.4238e-02, 1.3305e-02, 1.4389e-02, 2.4255e-03, 1.9224e-02,\n 1.6362e-02, -9.0868e-03, -2.0100e-03, -8.0016e-03, -6.8188e-03,\n 8.7414e-03, -1.3481e-03, 4.4693e-02, -5.7503e-03, 2.4993e-02,\n -1.7482e-02, 6.7940e-03, 1.6558e-02, 1.4348e-02, -3.4460e-03,\n 4.6056e-03, -3.1238e-03, -9.3381e-04, -1.2510e-02, 1.8821e-02,\n 1.7160e-03, -2.6444e-02, -3.2105e-02, 9.1987e-03, 1.4418e-03,\n 1.7281e-02, 2.2586e-02], device='cuda:0')",
|
| 13 |
+
"exp_avg_sq": "tensor([0.0050, 0.0056, 0.0048, 0.0029, 0.0052, 0.0031, 0.0031, 0.0033, 0.0062,\n 0.0040, 0.0029, 0.0042, 0.0052, 0.0033, 0.0041, 0.0041, 0.0011, 0.0022,\n 0.0029, 0.0029, 0.0033, 0.0037, 0.0043, 0.0044, 0.0052, 0.0031, 0.0032,\n 0.0037, 0.0059, 0.0018, 0.0032, 0.0051, 0.0051, 0.0034, 0.0036, 0.0034,\n 0.0025, 0.0038, 0.0035, 0.0033, 0.0054, 0.0035, 0.0040, 0.0055, 0.0041,\n 0.0051, 0.0078, 0.0044, 0.0027, 0.0045, 0.0032, 0.0036, 0.0046, 0.0022,\n 0.0060, 0.0044, 0.0021, 0.0048, 0.0029, 0.0034, 0.0032, 0.0068, 0.0050,\n 0.0029, 0.0025, 0.0028, 0.0034, 0.0048, 0.0083, 0.0057, 0.0066, 0.0038,\n 0.0076, 0.0027, 0.0037, 0.0075, 0.0045, 0.0039, 0.0037, 0.0076, 0.0057,\n 0.0036, 0.0057, 0.0049, 0.0032, 0.0030, 0.0011, 0.0092, 0.0061, 0.0045,\n 0.0028, 0.0032, 0.0026, 0.0040, 0.0056, 0.0038, 0.0058, 0.0048, 0.0042,\n 0.0044, 0.0029, 0.0038, 0.0044, 0.0019, 0.0031, 0.0032, 0.0069, 0.0032,\n 0.0048, 0.0048, 0.0019, 0.0057, 0.0034, 0.0028, 0.0022, 0.0030, 0.0048,\n 0.0060, 0.0024, 0.0041, 0.0032, 0.0025, 0.0043, 0.0051, 0.0057, 0.0042,\n 0.0054, 0.0033, 0.0054, 0.0026, 0.0043, 0.0034, 0.0036, 0.0027, 0.0047,\n 0.0025, 0.0034, 0.0020, 0.0049, 0.0077, 0.0046, 0.0027, 0.0031, 0.0053,\n 0.0051, 0.0033, 0.0037, 0.0036, 0.0045, 0.0048, 0.0032, 0.0041, 0.0027,\n 0.0028, 0.0023, 0.0026, 0.0047, 0.0042, 0.0025, 0.0051, 0.0035, 0.0039,\n 0.0040, 0.0037, 0.0051, 0.0039, 0.0020, 0.0027, 0.0048, 0.0043, 0.0042,\n 0.0037, 0.0020, 0.0036, 0.0049, 0.0026, 0.0031, 0.0037, 0.0026, 0.0040,\n 0.0048, 0.0077, 0.0021, 0.0045, 0.0062, 0.0044, 0.0016, 0.0025, 0.0045,\n 0.0030, 0.0022, 0.0023, 0.0061, 0.0038, 0.0046, 0.0036, 0.0032, 0.0033,\n 0.0042, 0.0079, 0.0021, 0.0038, 0.0026, 0.0053, 0.0039, 0.0028, 0.0024,\n 0.0027, 0.0034, 0.0060, 0.0024, 0.0056, 0.0055, 0.0065, 0.0038, 0.0028,\n 0.0036, 0.0029, 0.0030, 0.0049, 0.0026, 0.0042, 0.0067, 0.0063, 0.0022,\n 0.0038, 0.0036, 0.0050, 0.0035, 0.0067, 0.0036, 0.0030, 0.0050, 0.0026,\n 0.0038, 0.0047, 0.0047, 0.0048, 0.0036, 0.0082, 0.0057, 0.0032, 0.0043,\n 0.0028, 0.0029, 0.0046, 0.0035, 0.0037, 0.0027, 0.0055, 0.0025, 0.0035,\n 0.0017, 0.0085, 0.0051, 0.0046, 0.0074, 0.0034, 0.0046, 0.0040, 0.0039,\n 0.0071, 0.0030, 0.0028, 0.0020, 0.0035, 0.0028, 0.0053, 0.0022, 0.0035,\n 0.0021, 0.0023, 0.0049, 0.0042, 0.0036, 0.0054, 0.0016, 0.0029, 0.0075,\n 0.0041, 0.0039, 0.0047, 0.0040, 0.0027, 0.0059, 0.0046, 0.0032, 0.0040,\n 0.0055, 0.0024, 0.0033, 0.0054, 0.0039, 0.0045, 0.0066, 0.0046, 0.0063,\n 0.0046, 0.0077, 0.0064, 0.0041, 0.0033, 0.0034, 0.0030, 0.0032, 0.0038,\n 0.0041, 0.0029, 0.0057, 0.0043, 0.0037, 0.0047, 0.0028, 0.0041, 0.0022,\n 0.0048, 0.0023, 0.0044, 0.0038, 0.0028, 0.0044, 0.0068, 0.0038, 0.0031,\n 0.0042, 0.0033, 0.0034, 0.0037, 0.0027, 0.0038, 0.0060, 0.0038, 0.0035,\n 0.0033, 0.0024, 0.0029, 0.0039, 0.0037, 0.0051, 0.0047, 0.0031, 0.0034,\n 0.0032, 0.0037, 0.0044, 0.0041, 0.0031, 0.0037, 0.0032, 0.0039, 0.0040,\n 0.0066, 0.0033, 0.0049, 0.0027, 0.0041, 0.0055, 0.0043, 0.0025, 0.0041,\n 0.0040, 0.0043, 0.0041, 0.0024, 0.0034, 0.0036, 0.0028, 0.0071, 0.0030,\n 0.0027, 0.0029, 0.0092, 0.0054, 0.0043, 0.0040, 0.0025, 0.0042, 0.0031,\n 0.0041, 0.0032, 0.0028, 0.0043, 0.0033, 0.0042, 0.0055, 0.0035, 0.0022,\n 0.0042, 0.0055, 0.0017, 0.0045, 0.0074, 0.0029, 0.0037, 0.0025, 0.0046,\n 0.0025, 0.0032, 0.0061, 0.0020, 0.0022, 0.0038, 0.0045, 0.0057, 0.0047,\n 0.0070, 0.0039, 0.0029, 0.0048, 0.0024, 0.0049, 0.0051, 0.0062, 0.0029,\n 0.0063, 0.0030, 0.0038, 0.0028, 0.0063, 0.0028, 0.0040, 0.0039, 0.0050,\n 0.0056, 0.0072, 0.0036, 0.0069, 0.0056, 0.0027, 0.0024, 0.0049, 0.0027,\n 0.0031, 0.0121, 0.0027, 0.0047, 0.0047, 0.0042, 0.0046, 0.0052, 0.0055,\n 0.0062, 0.0036, 0.0030, 0.0026, 0.0034, 0.0053, 0.0027, 0.0029, 0.0033,\n 0.0017, 0.0066, 0.0039, 0.0037, 0.0032, 0.0062, 0.0041, 0.0026, 0.0045,\n 0.0036, 0.0048, 0.0053, 0.0033, 0.0055, 0.0013, 0.0042, 0.0029, 0.0038,\n 0.0041, 0.0047, 0.0041, 0.0025, 0.0056, 0.0039, 0.0031, 0.0028, 0.0031,\n 0.0035, 0.0042, 0.0035, 0.0032, 0.0040, 0.0033, 0.0038, 0.0034, 0.0037,\n 0.0033, 0.0032, 0.0083, 0.0042, 0.0006, 0.0045, 0.0040, 0.0034, 0.0050,\n 0.0045, 0.0041, 0.0054, 0.0051, 0.0041, 0.0034, 0.0039, 0.0050, 0.0032,\n 0.0044, 0.0054, 0.0036, 0.0027, 0.0040, 0.0038, 0.0051, 0.0031],\n device='cuda:0')"
|
| 14 |
+
},
|
| 15 |
+
"2": {
|
| 16 |
+
"step": "tensor(6260.)",
|
| 17 |
+
"exp_avg": "tensor([ 2.1815e-03, -1.5750e-03, 1.2153e-03, -1.4064e-03, 3.2024e-03,\n -1.4032e-03, 9.9024e-04, -2.7285e-03, 8.1722e-03, -2.3779e-03,\n -1.9329e-04, -1.9353e-03, -3.4159e-03, 1.2240e-03, -1.6010e-03,\n -8.6661e-04, -2.8030e-02, 5.0582e-03, 4.9008e-04, 2.9720e-03,\n -1.1340e-02, 1.2064e-03, -1.0646e-03, 1.9070e-03, 3.6705e-03,\n 4.6759e-03, -2.5305e-04, -1.7977e-03, 7.0236e-04, 1.6904e-03,\n 3.1620e-03, 1.7427e-03, -1.3331e-02, -2.5837e-03, -8.2051e-03,\n 2.1006e-03, -4.9373e-03, 7.7194e-03, -2.2305e-05, 8.6718e-03,\n -1.4255e-04, 7.2815e-04, -1.5002e-03, -2.5286e-03, 1.4547e-03,\n 1.5040e-03, 5.5302e-03, -3.4842e-03, 5.0568e-03, 3.4672e-04,\n -2.3691e-03, 2.6771e-03, -1.0279e-04, -3.2890e-03, 8.1636e-03,\n 6.8534e-03, -3.9034e-03, 4.7316e-03, 1.3909e-03, -2.9848e-03,\n -4.8885e-03, -1.5679e-03, -2.0981e-03, -5.6768e-03, -2.0175e-02,\n 3.9289e-04, 2.6816e-03, 1.8681e-03, -7.7362e-05, 1.2875e-04,\n 6.9327e-05, -3.0028e-03, 1.5836e-03, 2.1625e-03, 3.5330e-03,\n 5.1825e-03, 2.5320e-03, 2.0467e-03, -8.5242e-03, -1.6905e-03,\n 2.8916e-03, -4.9033e-03, 3.2602e-03, -1.4972e-03, -2.2383e-03,\n 3.2745e-03, 5.1364e-03, 7.6764e-03, -1.0700e-02, -4.1583e-03,\n 1.8435e-03, -2.8657e-03, 2.7881e-03, 2.7471e-03, 5.7782e-04,\n -1.0929e-03, 4.2726e-03, -2.0421e-03, 3.1632e-03, 1.1063e-03,\n 4.4953e-03, 3.2180e-03, -3.9795e-03, 2.3076e-03, -1.5328e-03,\n -5.3663e-03, 2.4667e-03, 5.3655e-04, -1.6288e-04, 5.7450e-03,\n 1.8788e-03, -2.0561e-03, -8.8940e-04, -2.1482e-04, -7.3990e-03,\n -1.9331e-03, -4.2232e-04, -1.7015e-03, -6.2275e-03, -5.8704e-03,\n -4.4072e-03, -5.8091e-03, 3.2677e-03, 3.0064e-03, 4.3835e-03,\n 4.9488e-04, -9.9782e-03, -5.6700e-04, -2.1874e-03, -9.7670e-04,\n -8.0850e-04, 2.0873e-03, 3.9017e-04, -3.0547e-04, 4.7723e-04,\n -1.2464e-03, -1.6741e-03, -2.2002e-03, -1.6638e-03, -2.3424e-03,\n -4.5711e-03, -4.4989e-03, 1.1418e-03, -4.1505e-03, 2.8706e-03,\n -4.1926e-03, 2.3926e-03, -5.0992e-04, -6.6447e-04, -7.0881e-03,\n -5.5679e-04, -6.3291e-04, 7.6182e-03, -7.9149e-03, 2.4448e-03,\n -1.5483e-03, -3.9269e-03, -8.3666e-03, 1.1086e-02, 1.3799e-03,\n -1.0574e-02, 6.9556e-03, -3.2621e-03, 3.0680e-03, -1.2805e-03,\n -1.9559e-03, -7.9585e-06, 5.0499e-03, 5.1782e-04, 8.7272e-05,\n 1.0626e-02, 2.1604e-03, -2.6310e-04, 3.1470e-03, -3.0407e-03,\n 4.0100e-03, -6.2737e-04, -3.0729e-03, 3.3576e-03, 1.8511e-05,\n 4.3836e-03, 4.3905e-03, 1.6452e-02, -3.2269e-03, -3.7450e-03,\n 2.8885e-04, 7.9757e-04, 8.7331e-03, -8.9159e-05, 1.0011e-02,\n -9.3032e-03, 4.5936e-04, 5.9358e-03, 1.9681e-03, -2.1021e-03,\n -9.0191e-04, 1.1002e-03, -9.0266e-04, -8.2858e-04, -2.7078e-03,\n 4.5292e-03, 2.2319e-03, 2.2321e-03, 4.9813e-04, 8.8182e-04,\n 2.4797e-04, -1.0171e-02, -3.9750e-04, -2.3741e-03, 3.0884e-03,\n -8.5472e-04, -3.9259e-03, -2.1774e-03, -1.7506e-03, -4.1809e-03,\n 6.7865e-04, -2.4803e-03, 2.3746e-03, 5.3913e-03, 1.4205e-03,\n -4.1777e-04, 8.8462e-05, 3.3604e-04, 1.9676e-03, 3.7660e-03,\n 1.5429e-04, 1.6968e-03, 9.2055e-04, 4.9649e-03, 5.7365e-03,\n -2.5093e-05, -7.6243e-04, 3.3211e-03, -1.2502e-03, -6.5091e-03,\n -1.4394e-03, -9.2620e-03, 2.8814e-03, 1.5196e-04, -1.0415e-05,\n 1.1853e-03, -6.4035e-03, 9.3163e-03, -4.9424e-04, -1.5375e-03,\n 1.3964e-03, -4.1361e-03, -1.6879e-04, -1.2217e-03, -1.0024e-03,\n -2.6082e-04, -2.6580e-03, -9.2834e-03, 6.5480e-05, 6.7142e-04,\n 2.6442e-03, -2.8667e-03, -3.0574e-03, 6.1347e-03, -3.0899e-04,\n 3.7807e-04, 7.8281e-04, -3.2434e-03, 5.1155e-03, 1.5550e-03,\n 1.4293e-03, -3.9785e-03, 3.6727e-04, -5.8771e-03, 2.2628e-03,\n 5.0907e-03, 7.7173e-04, 4.2942e-03, -4.7120e-04, -2.2804e-03,\n -3.9703e-03, -5.2235e-03, 1.2546e-03, -5.8383e-03, 1.5779e-03,\n -9.4083e-04, 5.5843e-03, -3.4919e-03, 5.2849e-03, 1.4270e-03,\n 2.7083e-04, -2.2106e-03, 1.4271e-03, 1.1519e-03, 1.0927e-04,\n 1.2333e-02, -4.0024e-04, 2.4968e-03, -5.3275e-04, 2.5343e-03,\n -4.9150e-03, -9.1251e-04, 4.1958e-03, -1.0247e-03, -2.0711e-04,\n -8.7639e-04, 2.2923e-03, 2.9834e-04, 2.4396e-03, -2.4453e-03,\n 2.4389e-03, -2.3628e-03, 1.1219e-03, 3.1381e-03, 2.5025e-03,\n 2.2930e-04, 8.8084e-04, 2.4307e-03, -1.4101e-03, -2.7235e-03,\n 2.1125e-03, -2.4827e-03, -7.1012e-05, -2.0305e-03, -1.0456e-03,\n -6.9697e-04, 1.1625e-03, -9.9771e-04, 1.3139e-03, 1.7546e-04,\n -1.0306e-03, -6.5988e-03, 2.2450e-03, -1.6262e-03, 3.9411e-03,\n -8.4289e-03, -2.0582e-03, 8.1236e-04, 1.9540e-03, 2.5285e-04,\n 4.6209e-03, 2.2213e-04, -1.3941e-02, 2.9107e-04, -2.0072e-03,\n -3.4023e-03, -1.5814e-03, 2.9058e-03, -4.7598e-03, -3.7612e-03,\n 5.4326e-03, -4.9851e-03, 3.0186e-03, -1.3486e-02, -8.8776e-04,\n 5.7016e-03, 4.6199e-03, 3.1290e-03, 4.7494e-03, 2.5303e-03,\n -5.3701e-04, -5.0677e-05, -1.8334e-03, -6.9192e-03, 3.2641e-03,\n 1.0990e-03, 1.9599e-03, -3.4252e-03, -6.6604e-03, -3.6564e-04,\n 4.6423e-04, -3.4004e-03, -2.2009e-03, -9.4070e-03, -3.2962e-04,\n -7.5894e-03, -1.1021e-04, -1.1937e-03, -2.3560e-03, 1.3160e-03,\n 1.9370e-03, 1.1315e-03, 5.8861e-03, -4.7670e-04, 3.0627e-03,\n -2.2813e-03, 1.7810e-03, 4.9092e-03, 1.9016e-03, -3.9266e-03,\n -4.8917e-03, 1.9392e-03, 6.7212e-04, -3.3254e-03, -5.7917e-03,\n -1.8731e-04, -1.9020e-03, 7.5549e-03, 2.6122e-03, -8.4760e-03,\n 2.2252e-03, -6.3167e-03, 6.1906e-04, 6.4020e-04, -3.4135e-03,\n 5.2990e-04, -6.5507e-04, 8.2542e-04, -2.3868e-04, 1.4448e-03,\n 1.0921e-03, 7.1252e-04, -9.9350e-03, -1.3192e-03, 3.0836e-03,\n -3.9265e-03, 1.5774e-03, 3.5606e-03, -7.4899e-03, 7.9483e-03,\n 5.5697e-03, -5.1107e-03, -4.0319e-03, 2.1731e-03, -6.5773e-03,\n 2.6403e-04, 3.3242e-03, -3.2250e-04, 5.1272e-03, -1.7705e-03,\n 6.6044e-04, 5.8501e-04, -3.1884e-04, 2.1413e-03, 2.8350e-03,\n -7.1141e-03, -7.2957e-03, 8.9510e-03, 1.6507e-03, 2.6872e-03,\n -2.8122e-03, 5.6138e-04, 1.2249e-03, 5.3867e-03, -2.7927e-04,\n 5.0217e-04, 8.1728e-04, 1.5149e-03, -1.2441e-03, -6.0091e-03,\n 5.5786e-03, -3.7537e-03, 9.5912e-03, -7.8214e-03, 2.0287e-03,\n -8.9819e-03, -1.8918e-03, 1.4658e-03, 4.2400e-03, 3.4974e-04,\n -1.9669e-03, 2.2739e-03, -3.4641e-03, 4.6775e-03, -1.2728e-03,\n -7.0636e-03, -2.9515e-04, -4.1867e-04, -1.3701e-03, 4.9584e-04,\n 8.0997e-03, -1.9217e-03, -5.1859e-04, 5.6427e-03, 8.6354e-04,\n 2.9212e-03, -1.7399e-03, -2.9433e-04, 1.0166e-03, -5.5502e-03,\n -3.4874e-03, 6.7845e-03, -3.2977e-03, -2.6393e-03, -5.8886e-03,\n -5.7430e-03, 1.3355e-03, 3.4050e-03, 7.7270e-04, 2.4297e-03,\n 3.8127e-03, -1.5594e-03, -2.1317e-04, -2.2911e-03, 5.0025e-04,\n 5.7662e-02, -3.4714e-05, 8.9560e-03, -5.8375e-04, 2.2579e-03,\n -5.1778e-03, 8.2019e-04, 2.3888e-03, 2.1080e-03, -1.4710e-03,\n -2.4489e-04, -1.4714e-03, -5.4719e-04, -2.5249e-03, 2.7177e-03,\n 1.2439e-04, -5.9094e-03, -7.3606e-03, 7.7940e-04, -3.8409e-04,\n 2.6991e-03, 3.9966e-03], device='cuda:0')",
|
| 18 |
+
"exp_avg_sq": "tensor([1.6259e-04, 7.4263e-05, 1.4621e-04, 2.4085e-04, 5.7840e-05, 1.1287e-04,\n 8.5474e-05, 1.2839e-04, 6.3751e-05, 1.3559e-04, 1.1994e-04, 1.8165e-04,\n 6.7241e-05, 1.1271e-04, 4.9190e-05, 7.2106e-05, 1.7469e-03, 1.5023e-04,\n 1.2310e-04, 1.6174e-04, 2.2928e-04, 2.3640e-04, 1.7522e-04, 9.3834e-05,\n 1.1589e-04, 2.3029e-04, 1.6666e-04, 1.3910e-04, 5.4743e-05, 3.2908e-04,\n 1.3921e-04, 8.1899e-05, 2.5739e-04, 5.8250e-05, 3.2588e-04, 2.6308e-04,\n 1.9834e-04, 2.1101e-04, 5.9280e-05, 1.1603e-04, 9.6014e-05, 1.1483e-04,\n 5.0124e-05, 5.9084e-05, 1.0619e-04, 6.3883e-05, 2.5240e-04, 1.6255e-04,\n 2.5253e-04, 1.0477e-04, 2.1927e-04, 2.4416e-04, 7.3825e-05, 1.9127e-04,\n 1.1328e-04, 1.7742e-04, 2.9624e-04, 1.3831e-04, 1.2970e-04, 1.5211e-04,\n 2.1517e-04, 5.3808e-05, 1.2651e-04, 2.8278e-04, 4.5894e-04, 1.3915e-04,\n 1.6972e-04, 1.2999e-04, 5.3613e-05, 1.4472e-04, 5.7618e-05, 9.1271e-05,\n 6.8656e-05, 9.0335e-05, 1.1363e-04, 9.6759e-05, 1.2954e-04, 1.0649e-04,\n 1.8854e-04, 1.3903e-04, 8.3971e-05, 6.0467e-05, 1.0925e-04, 1.3227e-04,\n 1.1966e-04, 1.6778e-04, 1.7079e-03, 1.1340e-04, 8.0009e-05, 1.6336e-04,\n 1.6000e-04, 1.9787e-04, 1.8697e-04, 6.9820e-05, 7.8413e-05, 8.0780e-05,\n 1.2316e-04, 8.8920e-05, 1.0499e-04, 5.8763e-05, 5.6513e-05, 6.7891e-05,\n 7.8003e-05, 2.0267e-04, 2.2883e-04, 1.4946e-04, 6.3961e-05, 2.3950e-04,\n 9.0071e-05, 7.3528e-05, 3.6890e-04, 5.8970e-05, 1.0507e-04, 6.9140e-05,\n 2.7284e-04, 8.8012e-05, 7.2209e-05, 9.1693e-05, 1.7021e-04, 1.0946e-04,\n 8.9751e-05, 1.6623e-04, 9.4795e-05, 9.8897e-05, 9.8360e-05, 7.8191e-05,\n 6.9924e-05, 1.4241e-04, 8.1097e-05, 2.4518e-04, 6.9664e-05, 1.0992e-04,\n 3.1055e-04, 2.0020e-04, 1.2818e-04, 1.6976e-04, 7.4647e-05, 2.0842e-04,\n 8.9839e-05, 9.4693e-05, 1.7378e-04, 1.7256e-04, 1.0245e-04, 1.2306e-04,\n 9.5464e-05, 1.5460e-04, 1.1230e-04, 1.2459e-04, 9.5148e-05, 9.4619e-05,\n 5.5002e-05, 1.6951e-04, 1.1130e-04, 2.2715e-04, 9.7461e-05, 2.0410e-04,\n 1.2711e-04, 1.0667e-04, 4.5549e-04, 5.4884e-05, 2.4663e-04, 1.1815e-04,\n 1.5817e-04, 1.0538e-04, 1.4186e-04, 7.1983e-05, 3.1837e-04, 1.5958e-04,\n 6.7407e-05, 1.0477e-04, 2.7776e-04, 1.1595e-04, 3.1766e-04, 9.6560e-05,\n 1.1147e-04, 1.2403e-04, 9.5498e-05, 1.5085e-04, 9.6367e-05, 6.1463e-05,\n 9.6672e-05, 1.1447e-04, 4.4657e-04, 1.3616e-04, 1.1060e-04, 1.3438e-04,\n 5.2334e-04, 3.6962e-04, 8.0388e-05, 1.7319e-04, 3.1184e-04, 3.2379e-04,\n 1.4858e-04, 9.6160e-05, 1.1630e-04, 1.0127e-04, 6.1063e-05, 9.9268e-05,\n 6.2536e-05, 8.5728e-05, 2.3744e-04, 1.8103e-04, 6.6551e-05, 7.0454e-05,\n 1.1515e-04, 2.0882e-04, 3.3041e-04, 1.6089e-04, 1.4029e-04, 1.1625e-04,\n 1.3372e-04, 1.1777e-04, 1.2048e-04, 1.4232e-04, 2.4719e-04, 1.3481e-04,\n 1.6501e-04, 7.2444e-05, 1.0719e-04, 1.2630e-04, 1.4090e-04, 6.4571e-05,\n 6.3008e-05, 1.2665e-04, 2.3244e-04, 1.0126e-04, 1.6051e-04, 6.5266e-05,\n 2.8545e-04, 1.7680e-04, 6.7286e-05, 1.1311e-04, 6.5598e-05, 1.1337e-04,\n 1.1643e-04, 1.0250e-04, 1.4657e-04, 1.4681e-04, 1.5315e-04, 1.0666e-04,\n 6.8070e-05, 1.3726e-04, 1.2487e-04, 9.6209e-05, 2.5413e-04, 6.7549e-05,\n 9.9600e-05, 4.8470e-05, 1.7470e-04, 1.2367e-04, 1.1960e-04, 1.5134e-04,\n 2.9628e-04, 1.0322e-04, 1.1627e-04, 6.4878e-05, 2.7322e-04, 9.9441e-05,\n 1.5711e-04, 1.2863e-04, 8.4184e-05, 8.7968e-05, 2.0401e-04, 2.7824e-04,\n 5.2535e-04, 1.6667e-04, 1.0751e-04, 1.0063e-04, 1.7346e-04, 8.7089e-05,\n 1.8365e-04, 1.2740e-04, 6.7579e-05, 5.9010e-05, 7.4661e-05, 2.5085e-04,\n 5.7390e-04, 1.6247e-04, 1.1503e-04, 8.4281e-05, 3.7377e-05, 1.2702e-04,\n 1.0636e-04, 1.4726e-04, 1.4834e-04, 7.1180e-05, 6.0822e-04, 6.8215e-05,\n 1.0084e-04, 1.5292e-04, 1.6414e-04, 9.4935e-05, 1.4109e-04, 1.0059e-04,\n 1.1039e-04, 1.0615e-04, 5.6440e-05, 1.6079e-04, 7.0488e-05, 9.3499e-05,\n 2.3090e-04, 1.1779e-04, 9.1287e-05, 2.7044e-04, 9.5098e-05, 1.5380e-04,\n 1.0912e-04, 1.0441e-04, 1.1349e-04, 1.0492e-04, 9.2193e-05, 6.4001e-05,\n 1.1290e-04, 8.5755e-05, 2.6727e-04, 1.5402e-04, 7.0119e-05, 1.6351e-04,\n 7.2068e-05, 2.5983e-04, 5.4843e-05, 2.2198e-04, 8.5080e-05, 1.1943e-04,\n 7.6174e-05, 7.5025e-05, 1.5933e-04, 1.0088e-04, 6.0793e-05, 1.1632e-04,\n 1.0864e-04, 2.0010e-04, 9.2124e-05, 1.2805e-04, 3.8619e-04, 2.3799e-04,\n 1.1882e-04, 2.9447e-04, 1.2016e-04, 7.7832e-05, 1.4414e-04, 1.8275e-04,\n 9.8391e-05, 1.1570e-04, 8.9382e-05, 7.6127e-05, 1.1479e-04, 7.6935e-05,\n 2.6048e-04, 5.3941e-05, 1.5984e-04, 7.9863e-05, 4.9342e-05, 1.0046e-04,\n 1.4874e-04, 8.2659e-05, 3.9218e-05, 7.5203e-05, 3.6718e-04, 8.1971e-05,\n 8.9825e-05, 4.4894e-05, 8.8827e-05, 1.5609e-04, 6.5041e-05, 1.0229e-04,\n 9.6176e-05, 9.2761e-05, 1.3846e-04, 1.5369e-04, 2.2643e-04, 1.0815e-04,\n 9.1364e-05, 1.2183e-04, 8.0192e-05, 1.4305e-04, 7.2903e-05, 1.1667e-04,\n 8.0970e-05, 1.7633e-04, 5.4641e-05, 9.5161e-05, 1.5289e-04, 1.4138e-04,\n 7.4196e-05, 7.9413e-05, 1.4604e-04, 5.3215e-05, 4.8893e-05, 6.0790e-04,\n 1.0696e-04, 6.9853e-05, 1.8139e-04, 1.1000e-04, 2.0111e-04, 1.6617e-04,\n 2.3896e-04, 2.2246e-04, 1.1626e-04, 2.0180e-04, 1.9773e-04, 8.6075e-05,\n 6.7358e-05, 6.2877e-05, 6.6561e-05, 7.4273e-05, 1.1504e-04, 1.1795e-04,\n 8.0788e-05, 3.3990e-04, 1.1366e-04, 5.6727e-05, 1.2605e-04, 3.7521e-04,\n 6.9625e-05, 2.2876e-04, 8.9032e-05, 1.0633e-04, 8.1298e-05, 2.4586e-04,\n 1.6746e-04, 1.1313e-04, 8.8730e-05, 1.1065e-04, 6.3174e-05, 9.5666e-05,\n 9.6353e-05, 8.9300e-05, 1.8200e-04, 1.8485e-04, 6.8437e-05, 3.3329e-04,\n 1.8700e-04, 8.9250e-05, 1.4351e-04, 1.3245e-04, 5.6150e-05, 1.0399e-04,\n 1.1166e-04, 7.1191e-05, 1.1284e-04, 1.2425e-04, 8.8743e-05, 2.4089e-04,\n 1.3345e-04, 6.2514e-05, 7.7861e-05, 2.1905e-04, 1.7370e-04, 2.0547e-04,\n 3.9163e-04, 1.1061e-04, 5.5777e-05, 1.9215e-04, 1.9820e-04, 9.7520e-05,\n 4.8783e-05, 4.8970e-04, 1.2806e-04, 1.1586e-04, 1.5945e-04, 7.5925e-05,\n 2.1954e-04, 6.0643e-05, 7.3751e-04, 1.7903e-04, 8.9310e-05, 5.6981e-05,\n 1.5629e-04, 5.7679e-05, 8.3412e-05, 1.7943e-04, 7.8898e-05, 1.3908e-04,\n 1.8565e-04, 1.8467e-04, 1.1873e-04, 1.7417e-04, 3.8118e-05, 1.1423e-04,\n 7.1163e-05, 3.8170e-05, 1.7007e-04, 1.4438e-04, 8.3679e-05, 2.1934e-04,\n 7.8999e-05, 1.4499e-04, 1.0445e-04, 9.6305e-05, 1.3647e-01, 7.2349e-05,\n 1.4453e-04, 1.0563e-04, 1.0561e-04, 2.1697e-04, 1.9413e-04, 1.2240e-04,\n 9.3112e-05, 1.5906e-04, 9.6666e-05, 1.1159e-04, 7.9673e-05, 1.0566e-04,\n 1.1816e-04, 1.5631e-04, 1.7678e-04, 1.5322e-04, 1.5372e-04, 1.7156e-04,\n 1.7158e-04, 1.2365e-04], device='cuda:0')"
|
| 19 |
+
},
|
| 20 |
+
"3": {
|
| 21 |
+
"step": "tensor(6260.)",
|
| 22 |
+
"exp_avg": "tensor([ 3.0576e-03, -2.4383e-03, 2.0123e-03, -4.6533e-04, 3.9857e-03,\n -1.8040e-03, 8.5587e-04, -2.1510e-03, 7.2788e-03, -1.9565e-03,\n -9.5740e-04, -1.4469e-03, -2.7100e-03, 1.4142e-03, -1.8444e-03,\n -1.3503e-03, -9.6246e-03, 3.8800e-03, -2.7378e-04, 2.1451e-03,\n -7.7998e-03, 8.0722e-04, -2.6682e-04, 3.7307e-03, 3.9518e-03,\n 3.1561e-03, 4.1610e-05, -5.6679e-04, 5.2619e-04, 7.8044e-04,\n 3.1866e-03, 2.8849e-03, -1.2730e-02, -3.2582e-03, -5.8394e-03,\n 1.3447e-03, -3.8257e-03, 5.6271e-03, -4.0581e-04, 7.4219e-03,\n -2.1341e-04, 4.4764e-04, -2.1008e-03, -2.3695e-03, 1.7882e-03,\n 1.8911e-03, 5.9151e-03, -3.4164e-03, 3.6573e-03, 7.3963e-04,\n -1.4932e-03, 1.3483e-03, -2.1760e-04, -2.3611e-03, 6.6687e-03,\n 8.0543e-03, -3.5468e-03, 5.1249e-03, 1.3816e-03, -1.8907e-03,\n -4.1159e-03, -3.3734e-03, -2.1467e-03, -3.6253e-03, -1.2029e-02,\n 1.6520e-05, 2.9876e-03, 2.4029e-03, -4.4460e-03, 1.1810e-03,\n -3.9118e-03, -2.8543e-03, 1.9917e-03, 1.2624e-03, 2.6279e-03,\n 8.2086e-03, 3.3121e-04, 3.0718e-03, -6.3209e-03, -3.1602e-03,\n 1.4771e-03, -5.8154e-03, 4.1496e-03, -1.4599e-03, -1.8345e-03,\n 3.1768e-03, 1.0883e-03, 8.4387e-03, -1.4479e-02, -3.5404e-03,\n 1.5643e-03, -2.7504e-03, 2.0571e-03, 3.2200e-03, 5.3108e-04,\n -6.3322e-04, 5.2096e-03, -2.6863e-03, 2.9344e-03, 2.1072e-03,\n 4.7667e-03, 3.5575e-03, -4.4687e-03, 1.4559e-03, -9.2593e-04,\n -3.9997e-03, 3.8592e-03, 1.1443e-04, -5.2392e-04, 5.7704e-03,\n 8.8708e-04, -3.0065e-03, -1.0977e-03, 4.1595e-04, -6.5232e-03,\n -2.5054e-03, -1.2933e-03, -2.4793e-03, -3.2086e-03, -4.9964e-03,\n -4.3524e-03, -4.0961e-03, 2.6145e-03, 3.6445e-03, 5.5777e-03,\n 8.1194e-05, -1.7024e-02, -8.1321e-04, -9.7074e-04, -1.8292e-03,\n -1.8196e-04, 2.1871e-03, 2.4270e-05, -2.1716e-04, 8.3054e-04,\n -7.6118e-04, -2.3021e-03, -1.5712e-03, -1.8659e-03, -3.5586e-03,\n -3.1864e-03, -3.2469e-03, 1.1370e-03, -3.8721e-03, 3.5752e-03,\n -3.4372e-03, 2.8578e-03, -6.3957e-04, -8.4990e-04, -9.8245e-03,\n -4.3953e-04, -6.3713e-04, 6.8955e-03, -5.7609e-03, 2.3784e-03,\n -1.1265e-03, -4.4155e-03, -8.2714e-03, 6.3412e-03, 9.8921e-04,\n -7.9009e-03, 5.9826e-03, -2.1596e-03, 2.5094e-03, -4.2581e-04,\n 6.3652e-04, -2.0789e-04, 4.6831e-03, 6.1868e-04, -4.3162e-04,\n 9.2014e-03, 3.3861e-03, -4.9341e-04, 4.0128e-03, -4.4349e-03,\n 3.7900e-03, 1.3564e-04, -2.8831e-03, 2.9954e-03, -1.6824e-04,\n 3.7182e-03, 5.6067e-03, 8.4432e-03, -3.8072e-03, -4.3458e-03,\n 4.5262e-04, -4.8785e-05, 5.5765e-03, 1.8578e-04, 7.8424e-03,\n -6.6726e-03, 7.0258e-04, 6.5019e-03, 1.7517e-03, -2.1361e-03,\n 8.9856e-05, 1.3138e-03, -1.6761e-03, -1.7429e-03, -4.0254e-03,\n 2.1805e-03, 2.2232e-03, 1.4312e-03, 7.9815e-04, 1.1464e-03,\n -1.8315e-03, -6.6392e-03, -1.3692e-04, -2.6575e-03, 4.2446e-03,\n -8.9066e-04, -3.0000e-03, -1.4412e-04, -7.1490e-04, -2.5460e-03,\n 1.0958e-03, -3.2217e-03, 2.9531e-03, 4.8785e-03, 2.2134e-04,\n 1.6402e-04, 6.0318e-04, 4.2801e-04, 2.1467e-03, 3.6204e-03,\n 1.2091e-03, 1.8299e-03, 2.1072e-03, 3.7572e-03, 7.7276e-03,\n -6.5907e-04, -1.0759e-03, 5.5588e-03, -1.4214e-03, -5.4238e-03,\n -1.4473e-03, -8.8114e-03, 4.4587e-03, 1.0498e-04, -2.2716e-04,\n 2.1772e-03, -6.1024e-03, 9.9913e-03, -1.3045e-03, -2.1314e-03,\n 2.1938e-03, -5.8149e-03, 5.6166e-04, -1.0774e-03, -1.6271e-03,\n 2.3696e-04, -1.9319e-03, -5.8765e-03, 1.3348e-04, 1.2950e-03,\n 2.9993e-03, -5.6415e-04, -3.8457e-03, 4.7111e-03, -3.5675e-04,\n 1.6741e-04, 1.9798e-03, -2.7783e-03, 3.9737e-03, 2.8741e-04,\n 7.6907e-04, -2.5956e-03, -3.2669e-04, -5.3119e-03, 2.3118e-03,\n 4.1480e-03, 4.0463e-04, 5.3295e-03, -2.1414e-03, -2.6876e-03,\n -2.6189e-03, -2.3980e-03, 1.1812e-03, -7.0298e-03, 1.5580e-03,\n 1.7193e-04, 7.9450e-03, -3.9554e-03, 5.1264e-03, 1.8944e-03,\n 6.4933e-04, -2.2018e-03, 1.9937e-03, 1.4587e-03, 7.2817e-05,\n 1.2845e-02, -1.8606e-03, 3.0388e-03, 1.7530e-04, 1.8106e-03,\n -7.0242e-03, -9.9191e-04, 4.7677e-03, -3.6177e-04, 4.1683e-05,\n -1.7317e-04, 2.1425e-03, -4.8776e-04, 1.8369e-03, -4.2720e-03,\n 1.9523e-03, -3.2248e-03, 1.7858e-03, 2.1338e-03, 2.5222e-03,\n -7.4965e-04, 2.6664e-03, 1.4283e-03, -1.7565e-03, -1.1585e-03,\n 1.4882e-03, -3.0888e-03, 5.7568e-04, -1.9841e-03, -3.9197e-04,\n -7.5938e-04, 2.2136e-03, -2.1559e-03, 1.2643e-03, -3.1079e-04,\n -1.0388e-03, -5.9337e-03, 2.3226e-03, -9.3170e-04, 3.4965e-03,\n -1.0057e-02, -2.2322e-03, 8.7370e-04, 1.3883e-03, 3.4278e-05,\n 2.4962e-03, 1.7458e-04, -7.7922e-03, 1.0723e-03, -2.7130e-03,\n -2.3845e-03, -1.9572e-03, 3.7346e-03, -4.0683e-03, -4.6727e-03,\n 6.0880e-03, -4.2228e-03, 2.1582e-03, -1.1485e-02, 4.5224e-04,\n 4.7293e-03, 5.9068e-03, 3.8384e-03, 6.0350e-03, 2.9726e-03,\n -1.6562e-04, -2.6643e-04, -2.7856e-03, -4.4869e-03, 1.0651e-03,\n 1.7333e-03, 2.4864e-03, -3.5205e-03, -6.1963e-03, -3.1169e-04,\n 3.4865e-04, -1.6270e-03, -3.8655e-03, -7.9971e-03, -8.1444e-05,\n -5.5927e-03, 8.8965e-04, -8.5329e-04, -2.5021e-03, 6.8270e-04,\n 2.2440e-03, 1.8870e-03, 5.3482e-03, -8.1542e-04, 2.7416e-03,\n -4.4747e-03, 2.5014e-03, 3.3148e-03, 1.4201e-03, -5.9545e-03,\n -5.2510e-03, 1.8699e-03, 1.4257e-03, -4.3990e-03, -2.3976e-03,\n 1.5111e-04, -2.3546e-03, 7.8305e-03, 2.5351e-03, -7.2621e-03,\n 8.6114e-04, -4.7301e-03, 8.7812e-05, -8.7934e-04, -2.4483e-03,\n -2.8825e-04, 8.1513e-04, 1.2060e-03, -2.5598e-04, 1.2580e-03,\n 1.4219e-03, 9.1935e-04, -9.5636e-03, 1.0557e-03, 2.6830e-03,\n -3.4347e-03, 1.9880e-03, 3.3265e-03, -5.8350e-03, 9.2241e-03,\n 4.9799e-03, -8.4782e-03, -3.7250e-03, 4.8651e-03, -5.3890e-03,\n 4.9221e-04, 2.5330e-03, 2.5144e-05, 6.9988e-03, -2.1242e-03,\n 8.8528e-04, 2.7745e-05, -8.5794e-04, 2.1234e-03, 2.4077e-03,\n -9.0757e-03, -5.5194e-03, 7.6537e-03, 2.9082e-03, 3.0573e-03,\n -1.9827e-03, 5.1749e-04, 2.4655e-03, 5.1722e-03, -1.7011e-03,\n 1.7248e-04, 1.2281e-03, 1.9982e-03, -4.0824e-04, -4.0303e-03,\n 6.2369e-03, -4.0103e-03, 6.6201e-03, -7.0165e-03, 1.8238e-03,\n -4.7361e-03, -1.7956e-03, 1.9173e-03, 2.4282e-03, 7.6548e-04,\n -1.9968e-03, 4.0020e-03, -2.0256e-03, 3.8973e-03, -2.4899e-03,\n -5.0275e-03, -1.4823e-04, 2.7833e-04, -1.2021e-03, 1.7869e-03,\n 7.0323e-03, -3.7372e-03, -2.1171e-04, 6.0424e-03, 3.6431e-04,\n 2.2024e-03, -1.0317e-03, -5.7044e-04, 3.3704e-03, -3.7876e-03,\n -3.6016e-03, 6.7456e-03, -2.5092e-03, -3.1183e-03, -7.4504e-03,\n -6.0488e-03, 2.3971e-03, 3.0392e-03, 6.8469e-04, 3.2368e-03,\n 4.1214e-03, -2.4538e-03, -5.4623e-04, -1.4489e-03, -3.0002e-03,\n 3.2442e-03, -2.1593e-05, 8.1980e-03, -9.4128e-04, 4.1015e-03,\n -4.0965e-03, 7.5879e-04, 2.5411e-03, 2.3228e-03, -1.2201e-03,\n 3.3879e-04, -6.3722e-04, -4.8999e-04, -2.0958e-03, 3.5193e-03,\n -4.4927e-04, -5.1589e-03, -5.8711e-03, 1.5655e-03, 1.5057e-04,\n 4.0867e-03, 4.2474e-03], device='cuda:0')",
|
| 23 |
+
"exp_avg_sq": "tensor([1.7416e-04, 1.4355e-04, 1.1869e-04, 1.5052e-04, 9.4175e-05, 1.1325e-04,\n 9.3211e-05, 1.1099e-04, 1.2358e-04, 1.5942e-04, 1.0482e-04, 1.6774e-04,\n 9.4352e-05, 1.0566e-04, 7.7740e-05, 9.9797e-05, 2.1696e-04, 1.0372e-04,\n 1.0360e-04, 1.1441e-04, 1.3902e-04, 1.6908e-04, 1.4741e-04, 1.0402e-04,\n 1.6813e-04, 1.5801e-04, 1.2928e-04, 1.1309e-04, 9.4569e-05, 1.2995e-04,\n 1.1557e-04, 1.3563e-04, 2.2846e-04, 8.8458e-05, 1.7299e-04, 1.6463e-04,\n 1.2656e-04, 1.4124e-04, 7.9128e-05, 1.0941e-04, 1.2078e-04, 1.1017e-04,\n 9.0875e-05, 1.0340e-04, 1.2633e-04, 1.0285e-04, 2.8127e-04, 1.6379e-04,\n 1.3091e-04, 1.2363e-04, 1.2694e-04, 1.6235e-04, 1.0577e-04, 1.1485e-04,\n 1.6901e-04, 1.7013e-04, 1.3450e-04, 1.4376e-04, 1.1162e-04, 1.2378e-04,\n 1.6650e-04, 1.0218e-04, 1.5953e-04, 1.6045e-04, 1.6436e-04, 1.0841e-04,\n 1.3827e-04, 1.3938e-04, 1.3303e-04, 1.7248e-04, 1.3565e-04, 9.6547e-05,\n 1.5081e-04, 9.0405e-05, 1.2369e-04, 1.8280e-04, 1.3555e-04, 1.2031e-04,\n 1.4935e-04, 1.9031e-04, 1.5099e-04, 1.0026e-04, 1.6716e-04, 1.5709e-04,\n 1.1572e-04, 1.2358e-04, 1.8422e-04, 1.9807e-04, 1.1432e-04, 1.6391e-04,\n 1.4277e-04, 1.4379e-04, 1.2730e-04, 9.4009e-05, 1.2552e-04, 1.0894e-04,\n 1.7156e-04, 1.3191e-04, 1.2628e-04, 9.1980e-05, 7.1690e-05, 9.8000e-05,\n 1.1048e-04, 1.0506e-04, 1.5551e-04, 1.3204e-04, 1.3167e-04, 1.4139e-04,\n 1.3585e-04, 1.1498e-04, 1.3672e-04, 1.0623e-04, 1.2196e-04, 8.8364e-05,\n 1.3987e-04, 9.7426e-05, 8.8018e-05, 1.6617e-04, 1.1328e-04, 1.0615e-04,\n 8.5954e-05, 9.3396e-05, 1.2099e-04, 1.3628e-04, 1.4203e-04, 1.0005e-04,\n 1.1729e-04, 1.1855e-04, 1.1349e-04, 1.2744e-04, 1.0108e-04, 1.1025e-04,\n 1.8087e-04, 1.0925e-04, 1.3253e-04, 1.1178e-04, 9.9056e-05, 1.0957e-04,\n 1.3118e-04, 1.3652e-04, 1.7428e-04, 1.2329e-04, 8.3643e-05, 1.5879e-04,\n 1.2056e-04, 1.2919e-04, 1.2268e-04, 1.2949e-04, 1.2449e-04, 1.4266e-04,\n 7.3601e-05, 1.4211e-04, 8.9242e-05, 1.4932e-04, 8.3693e-05, 1.4764e-04,\n 1.4435e-04, 1.2445e-04, 1.8479e-04, 9.2356e-05, 1.5501e-04, 1.1386e-04,\n 1.6943e-04, 1.0151e-04, 1.6567e-04, 8.8777e-05, 1.2874e-04, 1.0957e-04,\n 1.0717e-04, 1.4391e-04, 2.1645e-04, 1.1636e-04, 1.3423e-04, 9.7566e-05,\n 1.3441e-04, 9.0030e-05, 9.8250e-05, 1.3286e-04, 8.7371e-05, 8.9611e-05,\n 1.1649e-04, 1.9932e-04, 1.4603e-04, 1.3986e-04, 1.7035e-04, 1.2128e-04,\n 1.4264e-04, 1.6597e-04, 1.0574e-04, 1.2117e-04, 1.3542e-04, 1.3551e-04,\n 1.9606e-04, 1.0761e-04, 1.1816e-04, 1.0578e-04, 8.4411e-05, 1.1974e-04,\n 1.0079e-04, 1.7223e-04, 1.3004e-04, 1.6398e-04, 7.2652e-05, 1.1771e-04,\n 1.2851e-04, 1.2319e-04, 1.4718e-04, 1.0254e-04, 1.3105e-04, 1.5839e-04,\n 9.8774e-05, 1.7199e-04, 1.6610e-04, 1.8015e-04, 1.7580e-04, 1.0864e-04,\n 1.3589e-04, 7.6144e-05, 9.7863e-05, 1.3182e-04, 1.1990e-04, 1.0731e-04,\n 1.3122e-04, 1.6999e-04, 1.2504e-04, 1.2003e-04, 1.1162e-04, 9.5869e-05,\n 1.7928e-04, 2.1152e-04, 1.0548e-04, 1.0133e-04, 1.1382e-04, 9.8902e-05,\n 1.2327e-04, 1.5176e-04, 1.5353e-04, 1.9001e-04, 1.4776e-04, 1.7834e-04,\n 1.1126e-04, 1.1541e-04, 1.5252e-04, 1.0108e-04, 1.4369e-04, 1.1135e-04,\n 1.0365e-04, 7.0733e-05, 1.3134e-04, 1.6245e-04, 9.7318e-05, 1.3038e-04,\n 1.1039e-04, 2.1971e-04, 1.4225e-04, 1.0137e-04, 2.9740e-04, 1.2739e-04,\n 1.5214e-04, 1.4237e-04, 1.1008e-04, 1.6156e-04, 1.2584e-04, 1.4110e-04,\n 1.5925e-04, 1.3816e-04, 8.7970e-05, 1.3165e-04, 1.0495e-04, 9.8535e-05,\n 9.9411e-05, 8.4219e-05, 1.0483e-04, 9.1056e-05, 1.0736e-04, 2.1972e-04,\n 1.3209e-04, 1.3868e-04, 1.9200e-04, 1.3247e-04, 7.1706e-05, 1.6130e-04,\n 1.3027e-04, 1.0684e-04, 1.8548e-04, 1.0111e-04, 2.0014e-04, 9.0562e-05,\n 1.2350e-04, 1.1274e-04, 1.4594e-04, 1.3987e-04, 1.5562e-04, 1.2580e-04,\n 1.8317e-04, 1.2765e-04, 1.2024e-04, 1.8020e-04, 1.3704e-04, 1.8026e-04,\n 1.7685e-04, 1.1979e-04, 1.0788e-04, 1.6956e-04, 1.0082e-04, 1.3510e-04,\n 1.2016e-04, 9.4847e-05, 1.4658e-04, 1.2566e-04, 1.0430e-04, 9.7173e-05,\n 9.3602e-05, 9.4493e-05, 1.1996e-04, 1.4768e-04, 7.4651e-05, 1.4605e-04,\n 1.0331e-04, 1.4355e-04, 8.6561e-05, 2.1594e-04, 1.2740e-04, 1.1103e-04,\n 9.8297e-05, 1.0576e-04, 1.2482e-04, 1.3195e-04, 6.7172e-05, 1.1822e-04,\n 1.5096e-04, 1.5610e-04, 9.9747e-05, 1.0317e-04, 1.5707e-04, 1.3838e-04,\n 1.1008e-04, 1.8349e-04, 1.5064e-04, 1.1627e-04, 1.1991e-04, 1.5396e-04,\n 1.0306e-04, 1.0736e-04, 1.1201e-04, 1.1041e-04, 9.7897e-05, 1.0705e-04,\n 1.5869e-04, 6.2404e-05, 1.4382e-04, 1.3687e-04, 7.1303e-05, 1.3452e-04,\n 1.1810e-04, 1.1170e-04, 8.6409e-05, 1.0593e-04, 1.7626e-04, 9.2237e-05,\n 1.0227e-04, 7.3095e-05, 1.0426e-04, 1.0541e-04, 8.1770e-05, 1.2126e-04,\n 8.3670e-05, 1.7919e-04, 1.0469e-04, 1.0476e-04, 1.3256e-04, 1.8783e-04,\n 1.4144e-04, 1.3311e-04, 9.8134e-05, 1.0491e-04, 1.0741e-04, 1.1016e-04,\n 1.1172e-04, 1.1363e-04, 7.3233e-05, 1.2262e-04, 1.3712e-04, 1.4283e-04,\n 1.2043e-04, 8.0162e-05, 1.0117e-04, 8.9632e-05, 1.0806e-04, 1.7444e-04,\n 1.2672e-04, 1.2469e-04, 1.3272e-04, 9.8364e-05, 1.2665e-04, 1.5219e-04,\n 1.3452e-04, 1.3645e-04, 1.6823e-04, 1.0821e-04, 1.0265e-04, 1.1450e-04,\n 1.1098e-04, 1.1749e-04, 9.7300e-05, 1.2355e-04, 1.1986e-04, 9.4835e-05,\n 1.2218e-04, 1.5840e-04, 1.2678e-04, 9.4902e-05, 1.6429e-04, 1.9343e-04,\n 1.2084e-04, 1.4107e-04, 9.9896e-05, 8.9186e-05, 1.5231e-04, 1.4361e-04,\n 1.3716e-04, 1.1134e-04, 1.2957e-04, 1.3993e-04, 1.4469e-04, 1.0671e-04,\n 1.5213e-04, 1.1272e-04, 1.1875e-04, 1.1468e-04, 1.1745e-04, 1.5468e-04,\n 1.3974e-04, 2.3939e-04, 1.0475e-04, 1.6334e-04, 8.8701e-05, 1.3925e-04,\n 1.3895e-04, 1.1479e-04, 1.4186e-04, 1.8474e-04, 1.1379e-04, 1.3482e-04,\n 1.0408e-04, 8.6881e-05, 1.2510e-04, 1.2698e-04, 1.2623e-04, 1.5005e-04,\n 1.3857e-04, 1.3473e-04, 7.9182e-05, 1.5633e-04, 1.5468e-04, 1.5601e-04,\n 9.4698e-05, 1.7785e-04, 1.3596e-04, 1.3193e-04, 1.5887e-04, 1.3276e-04,\n 1.4721e-04, 1.0718e-04, 1.3753e-04, 1.5356e-04, 9.4866e-05, 8.7387e-05,\n 1.2072e-04, 9.9527e-05, 1.0040e-04, 1.2088e-04, 1.2591e-04, 1.5325e-04,\n 1.1700e-04, 1.2619e-04, 1.1293e-04, 1.3636e-04, 7.3986e-05, 1.1392e-04,\n 8.9719e-05, 7.6911e-05, 1.3034e-04, 1.1580e-04, 9.4912e-05, 1.4476e-04,\n 8.8999e-05, 1.1772e-04, 1.6945e-04, 1.2451e-04, 3.7486e-04, 1.1383e-04,\n 1.4500e-04, 1.0511e-04, 1.5546e-04, 1.7330e-04, 1.6907e-04, 1.6187e-04,\n 1.2337e-04, 1.4496e-04, 1.1924e-04, 1.0859e-04, 1.2662e-04, 1.0081e-04,\n 1.4077e-04, 1.7626e-04, 1.3205e-04, 1.0946e-04, 1.2627e-04, 1.2556e-04,\n 1.7272e-04, 1.0765e-04], device='cuda:0')"
|
| 24 |
+
},
|
| 25 |
+
"4": {
|
| 26 |
+
"step": "tensor(6260.)",
|
| 27 |
+
"exp_avg": "tensor([[ 4.2159e-05, 3.2645e-04, -2.3258e-05, ..., -1.1176e-05,\n 7.0849e-05, 9.4611e-05],\n [-1.5888e-04, 1.5549e-04, 5.3501e-05, ..., -2.9499e-04,\n -4.3635e-04, -9.6584e-05],\n [ 1.6640e-04, -7.6085e-05, -6.4825e-05, ..., 4.6750e-05,\n 1.6101e-04, -7.1049e-04],\n ...,\n [-1.0713e-04, -2.5067e-04, 8.8786e-05, ..., 1.1621e-04,\n -6.6531e-05, -2.2607e-04],\n [ 1.7656e-04, -1.2060e-04, -2.8125e-05, ..., 9.2466e-05,\n 9.4009e-04, 4.5998e-04],\n [ 3.8876e-04, -8.5232e-05, 4.3521e-05, ..., -1.2697e-04,\n -6.9682e-04, 9.3170e-05]], device='cuda:0')",
|
| 28 |
+
"exp_avg_sq": "tensor([[3.4817e-07, 6.3495e-07, 1.5190e-07, ..., 2.8201e-07, 3.6795e-07,\n 4.8263e-07],\n [7.3953e-07, 1.3551e-06, 5.0576e-07, ..., 9.3177e-07, 9.8023e-07,\n 1.2683e-06],\n [7.6249e-07, 5.0674e-07, 4.5204e-07, ..., 9.4227e-07, 6.2711e-07,\n 9.7581e-07],\n ...,\n [6.0291e-07, 1.4065e-06, 6.2417e-07, ..., 8.2718e-07, 1.1139e-06,\n 1.0584e-06],\n [5.6556e-07, 1.0090e-06, 3.3843e-07, ..., 1.0553e-06, 8.8484e-07,\n 8.4628e-07],\n [5.4614e-07, 7.8776e-07, 4.7127e-07, ..., 1.5093e-06, 6.4674e-07,\n 1.0058e-06]], device='cuda:0')"
|
| 29 |
+
},
|
| 30 |
+
"5": {
|
| 31 |
+
"step": "tensor(6260.)",
|
| 32 |
+
"exp_avg": "tensor([[ 1.1149e-04, 1.2096e-04, -4.6538e-05, ..., -1.5702e-04,\n -2.4097e-05, -4.0246e-05],\n [ 5.2053e-05, 1.9731e-04, 7.0488e-05, ..., -1.4215e-04,\n -2.6759e-04, -1.8175e-04],\n [ 9.1463e-06, -1.2508e-04, 4.2805e-05, ..., -2.2176e-04,\n 1.9160e-04, -2.9522e-04],\n ...,\n [-1.1556e-04, -1.5739e-05, -8.6264e-05, ..., -1.1041e-04,\n -1.8824e-04, 2.3869e-04],\n [-1.1284e-04, 1.7561e-04, -2.4730e-05, ..., 2.4603e-04,\n 3.6132e-04, 2.6090e-04],\n [-6.3783e-05, 6.6978e-06, 1.0567e-04, ..., 3.0647e-05,\n -3.0833e-04, 1.1474e-04]], device='cuda:0')",
|
| 33 |
+
"exp_avg_sq": "tensor([[1.1632e-07, 1.0374e-07, 1.1000e-07, ..., 1.4496e-07, 1.1411e-07,\n 1.5935e-07],\n [2.4185e-07, 4.8743e-07, 2.1979e-07, ..., 2.1995e-07, 2.8930e-07,\n 3.7548e-07],\n [2.6801e-07, 2.0033e-07, 2.1496e-07, ..., 4.2774e-07, 3.7957e-07,\n 3.0880e-07],\n ...,\n [2.4389e-07, 6.2915e-07, 1.2548e-07, ..., 5.0794e-07, 3.7525e-07,\n 3.3738e-07],\n [2.8802e-07, 3.7794e-07, 1.0434e-07, ..., 4.5410e-07, 2.9091e-07,\n 2.9497e-07],\n [2.5284e-07, 1.7640e-07, 4.9333e-07, ..., 2.6480e-07, 2.4195e-07,\n 3.3802e-07]], device='cuda:0')"
|
| 34 |
+
},
|
| 35 |
+
"6": {
|
| 36 |
+
"step": "tensor(6260.)",
|
| 37 |
+
"exp_avg": "tensor([ 0.0012, -0.0012], device='cuda:0')",
|
| 38 |
+
"exp_avg_sq": "tensor([9.0897e-06, 9.0897e-06], device='cuda:0')"
|
| 39 |
+
}
|
| 40 |
+
},
|
| 41 |
+
"param_groups": [
|
| 42 |
+
{
|
| 43 |
+
"lr": 0.0005005000000000001,
|
| 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": 0.0005005000000000001,
|
| 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": 0.0005005000000000001,
|
| 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": 0.0002505,
|
| 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": 5,
|
| 137 |
+
"base_lrs": [
|
| 138 |
+
0.001,
|
| 139 |
+
0.001,
|
| 140 |
+
0.001,
|
| 141 |
+
0.0005
|
| 142 |
+
],
|
| 143 |
+
"last_epoch": 5,
|
| 144 |
+
"_step_count": 0,
|
| 145 |
+
"_is_initial": false,
|
| 146 |
+
"_get_lr_called_within_step": false,
|
| 147 |
+
"_last_lr": [
|
| 148 |
+
0.0005005000000000001,
|
| 149 |
+
0.0005005000000000001,
|
| 150 |
+
0.0005005000000000001,
|
| 151 |
+
0.0002505
|
| 152 |
+
]
|
| 153 |
+
},
|
| 154 |
+
"metrics": {
|
| 155 |
+
"best_val_acc": 70.966,
|
| 156 |
+
"best_epoch": 4,
|
| 157 |
+
"scale_accuracies": {
|
| 158 |
+
"256": 70.468,
|
| 159 |
+
"512": 70.854
|
| 160 |
+
},
|
| 161 |
+
"training_history": {
|
| 162 |
+
"epochs": [
|
| 163 |
+
1,
|
| 164 |
+
2,
|
| 165 |
+
3,
|
| 166 |
+
4,
|
| 167 |
+
5
|
| 168 |
+
],
|
| 169 |
+
"train_loss": [
|
| 170 |
+
5.603825211905824,
|
| 171 |
+
4.145695094292918,
|
| 172 |
+
3.767625377201044,
|
| 173 |
+
3.572572173021091,
|
| 174 |
+
3.45402966711087
|
| 175 |
+
],
|
| 176 |
+
"train_acc": [
|
| 177 |
+
63.39829233815732,
|
| 178 |
+
69.48282308239285,
|
| 179 |
+
70.34266414917025,
|
| 180 |
+
70.91120829681064,
|
| 181 |
+
71.35642738222262
|
| 182 |
+
],
|
| 183 |
+
"val_acc": [
|
| 184 |
+
68.056,
|
| 185 |
+
69.624,
|
| 186 |
+
70.142,
|
| 187 |
+
70.622,
|
| 188 |
+
70.966
|
| 189 |
+
],
|
| 190 |
+
"scale_accs": {
|
| 191 |
+
"256": [
|
| 192 |
+
66.924,
|
| 193 |
+
68.828,
|
| 194 |
+
69.54,
|
| 195 |
+
69.996,
|
| 196 |
+
70.468
|
| 197 |
+
],
|
| 198 |
+
"512": [
|
| 199 |
+
67.882,
|
| 200 |
+
69.572,
|
| 201 |
+
70.028,
|
| 202 |
+
70.482,
|
| 203 |
+
70.854
|
| 204 |
+
]
|
| 205 |
+
},
|
| 206 |
+
"lr": [
|
| 207 |
+
0.0009755527298894294,
|
| 208 |
+
0.0009046039886902864,
|
| 209 |
+
0.0007940987335200904,
|
| 210 |
+
0.0006548539886902864,
|
| 211 |
+
0.0005005000000000001
|
| 212 |
+
]
|
| 213 |
+
}
|
| 214 |
+
},
|
| 215 |
+
"train_config": {
|
| 216 |
+
"name": "david_training",
|
| 217 |
+
"run_id": "20251012_135249",
|
| 218 |
+
"dataset_name": "AbstractPhil/imagenet-clip-features-orderly",
|
| 219 |
+
"model_variant": "clip_vit_laion_b32",
|
| 220 |
+
"num_classes": 1000,
|
| 221 |
+
"preset": "small_fast",
|
| 222 |
+
"custom_config_path": null,
|
| 223 |
+
"num_classes_override": null,
|
| 224 |
+
"use_belly_override": null,
|
| 225 |
+
"belly_expand_override": null,
|
| 226 |
+
"progressive_training_override": false,
|
| 227 |
+
"scale_warmup_epochs_override": null,
|
| 228 |
+
"num_epochs": 10,
|
| 229 |
+
"batch_size": 1024,
|
| 230 |
+
"learning_rate": 0.001,
|
| 231 |
+
"weight_decay": 1e-05,
|
| 232 |
+
"warmup_epochs": 3,
|
| 233 |
+
"use_rose_loss": true,
|
| 234 |
+
"rose_initial_weight": 0.1,
|
| 235 |
+
"rose_max_weight": 0.5,
|
| 236 |
+
"rose_weight_schedule": "adaptive",
|
| 237 |
+
"use_cayley_loss": false,
|
| 238 |
+
"cayley_weight": 0.001,
|
| 239 |
+
"scale_loss_balance": null,
|
| 240 |
+
"use_mixed_precision": false,
|
| 241 |
+
"gradient_clip": 10.0,
|
| 242 |
+
"scheduler_type": "cosine_restarts",
|
| 243 |
+
"min_lr": 1e-06,
|
| 244 |
+
"freeze_strategy": "never",
|
| 245 |
+
"freeze_threshold": 90.0,
|
| 246 |
+
"unfreeze_on_plateau": true,
|
| 247 |
+
"patience": 10,
|
| 248 |
+
"track_gradients": true,
|
| 249 |
+
"gradient_scale_threshold": 1e-05,
|
| 250 |
+
"gradient_scale_multiplier": 10.0,
|
| 251 |
+
"log_interval": 50,
|
| 252 |
+
"val_interval": 1,
|
| 253 |
+
"save_interval": 5,
|
| 254 |
+
"log_fusion_weights": true,
|
| 255 |
+
"log_loss_components": true,
|
| 256 |
+
"save_format": "safetensors",
|
| 257 |
+
"hf_repo": "AbstractPhil/gated-david",
|
| 258 |
+
"upload_to_hub": true,
|
| 259 |
+
"base_dir": "./david_training",
|
| 260 |
+
"num_workers": 10,
|
| 261 |
+
"pin_memory": true,
|
| 262 |
+
"prefetch_factor": 4,
|
| 263 |
+
"persistent_workers": true
|
| 264 |
+
}
|
| 265 |
+
}
|
weights/David-fully_shared-weighted_sum/20251012_135249/training_history.json
CHANGED
|
@@ -3,44 +3,51 @@
|
|
| 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 |
}
|
|
|
|
| 3 |
1,
|
| 4 |
2,
|
| 5 |
3,
|
| 6 |
+
4,
|
| 7 |
+
5
|
| 8 |
],
|
| 9 |
"train_loss": [
|
| 10 |
5.603825211905824,
|
| 11 |
4.145695094292918,
|
| 12 |
3.767625377201044,
|
| 13 |
+
3.572572173021091,
|
| 14 |
+
3.45402966711087
|
| 15 |
],
|
| 16 |
"train_acc": [
|
| 17 |
63.39829233815732,
|
| 18 |
69.48282308239285,
|
| 19 |
70.34266414917025,
|
| 20 |
+
70.91120829681064,
|
| 21 |
+
71.35642738222262
|
| 22 |
],
|
| 23 |
"val_acc": [
|
| 24 |
68.056,
|
| 25 |
69.624,
|
| 26 |
70.142,
|
| 27 |
+
70.622,
|
| 28 |
+
70.966
|
| 29 |
],
|
| 30 |
"scale_accs": {
|
| 31 |
"256": [
|
| 32 |
66.924,
|
| 33 |
68.828,
|
| 34 |
69.54,
|
| 35 |
+
69.996,
|
| 36 |
+
70.468
|
| 37 |
],
|
| 38 |
"512": [
|
| 39 |
67.882,
|
| 40 |
69.572,
|
| 41 |
70.028,
|
| 42 |
+
70.482,
|
| 43 |
+
70.854
|
| 44 |
]
|
| 45 |
},
|
| 46 |
"lr": [
|
| 47 |
0.0009755527298894294,
|
| 48 |
0.0009046039886902864,
|
| 49 |
0.0007940987335200904,
|
| 50 |
+
0.0006548539886902864,
|
| 51 |
+
0.0005005000000000001
|
| 52 |
]
|
| 53 |
}
|