Skip to content

Commit 0adadef

Browse files
committed
update GPT exercise 2b
1 parent 8318a5b commit 0adadef

File tree

1 file changed

+4
-18
lines changed

1 file changed

+4
-18
lines changed

007_GPT/ex2b.ipynb

Lines changed: 4 additions & 18 deletions
Original file line numberDiff line numberDiff line change
@@ -1631,10 +1631,10 @@
16311631
"----\n",
16321632
"## A. Model Performance Log for Different Hyperparameter Combinations (`ndigit = 2`)\n",
16331633
"\n",
1634-
"| S/o. | Layers | Heads | Embed. Dim. | Learn. rate | Batch Size | Epochs | Eval_iters | Eval_interval | Parameters (M) | Train loss | Val. loss | Val. Accuracy | Val. MAE | Test loss (avg) | Test Accuracy | Test MAE \n",
1635-
"| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- \n",
1636-
"| A. | **6** | **8** | **512** | **6e-4** | **64** | 100 | **32** | **562** | 18.90 | 0.0331 | 0.4072 | 69.87% | - | 0.2665 | 65.61% | - \n",
1637-
"| B. | **6** | **8** | **512** | **3e-5** | **64** | 100 | **32** | **562** | 18.90 | 0.0050 | 0.6533 | 63.41% | 0.2662 | 0.6437 | 66.87% | 0.2377 \n",
1634+
"| S/o. | Layers | Heads | Embed. Dim. | Learn. rate | Batch Size | Epochs | Eval_iters | Eval_interval | Parameters (M) | Train loss | Val. loss | Val. Accuracy | Val. MAE | Test loss (avg) | Test Accuracy | Test MAE |\n",
1635+
"| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |\n",
1636+
"| A. | **6** | **8** | **512** | **6e-4** | **64** | 100 | **32** | **562** | 18.90 | 0.0331 | 0.4072 | 69.87% | - | 0.2665 | 65.61% | - |\n",
1637+
"| B. | **6** | **8** | **512** | **3e-5** | **64** | 100 | **32** | **562** | 18.90 | 0.0050 | 0.6533 | 63.41% | 0.2662 | 0.6437 | 66.87% | 0.2377 |\n",
16381638
"\n",
16391639
"----\n"
16401640
]
@@ -1680,20 +1680,6 @@
16801680
"# print(f\"No. of test samples: {test_total} \\nAvg. Test Loss: {avg_test_loss:.4f} \\\n",
16811681
"# \\nTest Accuracy: {test_accuracy:.2f}% \\nAvg. Test MAE: {avg_test_mae:.4f}\")"
16821682
]
1683-
},
1684-
{
1685-
"cell_type": "code",
1686-
"execution_count": null,
1687-
"metadata": {},
1688-
"outputs": [],
1689-
"source": []
1690-
},
1691-
{
1692-
"cell_type": "code",
1693-
"execution_count": null,
1694-
"metadata": {},
1695-
"outputs": [],
1696-
"source": []
16971683
}
16981684
],
16991685
"metadata": {

0 commit comments

Comments
 (0)