Upload task1_rnn/tuning/tuning_report.txt with huggingface_hub
Browse files
task1_rnn/tuning/tuning_report.txt
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
======================================================================
|
| 2 |
+
RNN HYPERPARAMETER TUNING REPORT
|
| 3 |
+
======================================================================
|
| 4 |
+
Date: 2026-03-24 08:22:48
|
| 5 |
+
Total trials: 16
|
| 6 |
+
Tune epochs per trial: 10
|
| 7 |
+
Early stopping patience: 5
|
| 8 |
+
|
| 9 |
+
BEST CONFIGURATION:
|
| 10 |
+
model.hidden_size: 256
|
| 11 |
+
model.num_layers: 4
|
| 12 |
+
model.input_size: 128
|
| 13 |
+
model.dropout: 0.1
|
| 14 |
+
training.learning_rate: 0.001
|
| 15 |
+
training.batch_size: 128
|
| 16 |
+
training.epochs: 20
|
| 17 |
+
Val Accuracy: 0.655224
|
| 18 |
+
Val Loss: 1.153426
|
| 19 |
+
Best Epoch: 20
|
| 20 |
+
Parameters: 517,813
|
| 21 |
+
Model: models/rnn_hs256_nl4_emb128_do0.1_lr0.001_bs128_ep20.pt
|
| 22 |
+
|
| 23 |
+
======================================================================
|
| 24 |
+
FULL RANKING (by validation accuracy)
|
| 25 |
+
======================================================================
|
| 26 |
+
Rank | Val Acc | Val Loss | Epochs | Params | Model File
|
| 27 |
+
----------------------------------------------------------------------
|
| 28 |
+
1 | 0.6552 | 1.1534 | 20 | 517,813 | rnn_hs256_nl4_emb128_do0.1_lr0.001_bs128_ep20
|
| 29 |
+
2 | 0.6378 | 1.2029 | 20 | 496,117 | rnn_hs256_nl4_emb64_do0.1_lr0.001_bs128_ep20
|
| 30 |
+
3 | 0.6034 | 1.3251 | 20 | 232,949 | rnn_hs256_nl2_emb64_do0.1_lr0.001_bs128_ep20
|
| 31 |
+
4 | 0.5985 | 1.3473 | 20 | 232,949 | rnn_hs256_nl2_emb64_do0.2_lr0.001_bs128_ep20
|
| 32 |
+
5 | 0.5905 | 1.3540 | 20 | 496,117 | rnn_hs256_nl4_emb64_do0.2_lr0.001_bs128_ep20
|
| 33 |
+
6 | 0.5857 | 1.3732 | 20 | 517,813 | rnn_hs256_nl4_emb128_do0.2_lr0.001_bs128_ep20
|
| 34 |
+
7 | 0.5732 | 1.4206 | 20 | 254,645 | rnn_hs256_nl2_emb128_do0.1_lr0.001_bs128_ep20
|
| 35 |
+
8 | 0.5489 | 1.4997 | 20 | 254,645 | rnn_hs256_nl2_emb128_do0.2_lr0.001_bs128_ep20
|
| 36 |
+
9 | 0.4184 | 1.9296 | 8 | 232,949 | rnn_hs256_nl2_emb64_do0.1_lr0.01_bs128_ep20
|
| 37 |
+
10 | 0.3959 | 1.9900 | 7 | 232,949 | rnn_hs256_nl2_emb64_do0.2_lr0.01_bs128_ep20
|
| 38 |
+
11 | 0.3943 | 1.9947 | 10 | 496,117 | rnn_hs256_nl4_emb64_do0.2_lr0.01_bs128_ep20
|
| 39 |
+
12 | 0.3909 | 2.0104 | 7 | 254,645 | rnn_hs256_nl2_emb128_do0.1_lr0.01_bs128_ep20
|
| 40 |
+
13 | 0.3849 | 2.0276 | 9 | 517,813 | rnn_hs256_nl4_emb128_do0.1_lr0.01_bs128_ep20
|
| 41 |
+
14 | 0.3807 | 2.0429 | 9 | 496,117 | rnn_hs256_nl4_emb64_do0.1_lr0.01_bs128_ep20
|
| 42 |
+
15 | 0.3738 | 2.0487 | 10 | 254,645 | rnn_hs256_nl2_emb128_do0.2_lr0.01_bs128_ep20
|
| 43 |
+
16 | 0.3536 | 2.1031 | 9 | 517,813 | rnn_hs256_nl4_emb128_do0.2_lr0.01_bs128_ep20
|