Update README.md
Browse files
README.md
CHANGED
@@ -7,6 +7,9 @@ datasets:
|
|
7 |
model-index:
|
8 |
- name: DeBERTa-finetuned-SNLI2
|
9 |
results: []
|
|
|
|
|
|
|
10 |
---
|
11 |
|
12 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
@@ -16,9 +19,23 @@ should probably proofread and complete it, then remove this comment. -->
|
|
16 |
|
17 |
This model is a fine-tuned version of [gyeoldere/test_trainer](https://huggingface.co/gyeoldere/test_trainer) on the snli dataset.
|
18 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
## Model description
|
20 |
|
21 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
22 |
|
23 |
## Intended uses & limitations
|
24 |
|
@@ -50,4 +67,4 @@ The following hyperparameters were used during training:
|
|
50 |
- Transformers 4.26.0
|
51 |
- Pytorch 1.13.1+cu116
|
52 |
- Datasets 2.9.0
|
53 |
-
- Tokenizers 0.13.2
|
|
|
7 |
model-index:
|
8 |
- name: DeBERTa-finetuned-SNLI2
|
9 |
results: []
|
10 |
+
metrics:
|
11 |
+
- accuracy
|
12 |
+
library_name: transformers
|
13 |
---
|
14 |
|
15 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
19 |
|
20 |
This model is a fine-tuned version of [gyeoldere/test_trainer](https://huggingface.co/gyeoldere/test_trainer) on the snli dataset.
|
21 |
|
22 |
+
Test_trainer model is a fine-tuned version of [microsoft/deberta-base](https://huggingface.co/microsoft/deberta-base) on the snli dataset.
|
23 |
+
|
24 |
+
This model achieves the following results on the evaluation set:
|
25 |
+
- NLI accuracy : 0.86
|
26 |
+
- MLM accuracy : 0.68
|
27 |
+
|
28 |
+
|
29 |
## Model description
|
30 |
|
31 |
+
This model fine-tuned to perform 2 tasks simultaneously; NLI task and MLM task.
|
32 |
+
|
33 |
+
Output vector of DeBERTa processed through two different fc layer to predict.
|
34 |
+
I used layer structure introduced in BERT paper, which is implemented on huggingface transformers; DebertaForTokenClassification and DebertaForMaskedLM.
|
35 |
+
[https://huggingface.co/docs/transformers/index]
|
36 |
+
|
37 |
+
BinaryCrossEntrophyLoss are used for each class, and two losses are added to obtain final loss
|
38 |
+
final_loss = MLM_loss + NLI_loss
|
39 |
|
40 |
## Intended uses & limitations
|
41 |
|
|
|
67 |
- Transformers 4.26.0
|
68 |
- Pytorch 1.13.1+cu116
|
69 |
- Datasets 2.9.0
|
70 |
+
- Tokenizers 0.13.2
|