Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
This model is a fine-tuned version of fzwd6666/Ged_bert_new on an NLI dataset. It achieves the following results on the evaluation set:
|
2 |
+
|
3 |
+
{'precision': 0.9647540983606557} {'recall': 0.9755491089929549} {'f1': 0.9701215742839481} {'accuracy': 0.9525523560209425}
|
4 |
+
|
5 |
+
Training hyperparameters:
|
6 |
+
|
7 |
+
learning_rate: 2e-5
|
8 |
+
train_batch_size: 8
|
9 |
+
eval_batch_size: 8
|
10 |
+
optimizer: AdamW with betas=(0.9,0.999) and epsilon=1e-08
|
11 |
+
weight_decay= 0.01
|
12 |
+
|
13 |
+
lr_scheduler_type: linear
|
14 |
+
|
15 |
+
num_epochs: 3
|
16 |
+
It achieves the following results on the test set:
|
17 |
+
Incorrect UD Padded:
|
18 |
+
{'precision': 0.6101694915254238} {'recall': 0.6580907244414353} {'f1': 0.6332247557003257} {'accuracy': 0.5230834392206692}
|
19 |
+
|
20 |
+
Incorrect UD Unigram:
|
21 |
+
{'precision': 0.6010863005431503} {'recall': 0.6743398781313473} {'f1': 0.635609444798979} {'accuracy': 0.516306649724693}
|
22 |
+
|
23 |
+
Incorrect UD Bigram:
|
24 |
+
{'precision': 0.6137841352405722} {'recall': 0.6391333784698714} {'f1': 0.6262023217247098} {'accuracy': 0.5226598898771707}
|
25 |
+
|
26 |
+
Incorrect UD All:
|
27 |
+
{'precision': 0.6333333333333333} {'recall': 0.34732566012186866} {'f1': 0.4486226497595103} {'accuracy': 0.46590427784836935}
|
28 |
+
|
29 |
+
Incorrect Sentence:
|
30 |
+
{'precision': 0.6519031141868512} {'recall': 0.6377792823290453} {'f1': 0.6447638603696099} {'accuracy': 0.5603557814485387}
|