Update README.md
Browse files
README.md
CHANGED
@@ -3,9 +3,17 @@ license: apache-2.0
|
|
3 |
base_model: bert-base-uncased
|
4 |
tags:
|
5 |
- generated_from_keras_callback
|
|
|
6 |
model-index:
|
7 |
- name: bert-base-uncased-finetuned-glue-sst2
|
8 |
results: []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
---
|
10 |
|
11 |
<!-- This model card has been generated automatically according to the information Keras had access to. You should
|
@@ -13,21 +21,25 @@ probably proofread and complete it, then remove this comment. -->
|
|
13 |
|
14 |
# bert-base-uncased-finetuned-glue-sst2
|
15 |
|
16 |
-
|
|
|
|
|
|
|
17 |
It achieves the following results on the evaluation set:
|
18 |
|
|
|
|
|
|
|
19 |
|
20 |
## Model description
|
21 |
|
22 |
-
|
23 |
|
24 |
-
|
25 |
-
|
26 |
-
More information needed
|
27 |
|
28 |
## Training and evaluation data
|
29 |
|
30 |
-
|
31 |
|
32 |
## Training procedure
|
33 |
|
@@ -39,11 +51,12 @@ The following hyperparameters were used during training:
|
|
39 |
|
40 |
### Training results
|
41 |
|
42 |
-
|
|
|
43 |
|
44 |
### Framework versions
|
45 |
|
46 |
- Transformers 4.35.2
|
47 |
- TensorFlow 2.15.0
|
48 |
- Datasets 2.16.1
|
49 |
-
- Tokenizers 0.15.0
|
|
|
3 |
base_model: bert-base-uncased
|
4 |
tags:
|
5 |
- generated_from_keras_callback
|
6 |
+
- really-cool
|
7 |
model-index:
|
8 |
- name: bert-base-uncased-finetuned-glue-sst2
|
9 |
results: []
|
10 |
+
datasets:
|
11 |
+
- glue
|
12 |
+
language:
|
13 |
+
- en
|
14 |
+
metrics:
|
15 |
+
- accuracy
|
16 |
+
pipeline_tag: text-classification
|
17 |
---
|
18 |
|
19 |
<!-- This model card has been generated automatically according to the information Keras had access to. You should
|
|
|
21 |
|
22 |
# bert-base-uncased-finetuned-glue-sst2
|
23 |
|
24 |
+
Use for **sentiment analysis**. Labels: `positive`, `negative`
|
25 |
+
|
26 |
+
This model is a fine-tuned version of the [bert-base-uncased](https://huggingface.co/bert-base-uncased) model, fine-tuned on a subset of the [glue sst2 dataset](https://huggingface.co/datasets/glue/viewer/sst2).
|
27 |
+
|
28 |
It achieves the following results on the evaluation set:
|
29 |
|
30 |
+
```
|
31 |
+
Evaluation Accuracy: 91.74%
|
32 |
+
```
|
33 |
|
34 |
## Model description
|
35 |
|
36 |
+
The `bert-base-uncased` model is a pretrained English language model which has learned a bidirectional representation through Masked Language Modeling (MLM).
|
37 |
|
38 |
+
The `bert-base-uncased-finetuned-glue-sst2` adds a 2-class classification head to `bert-base-uncased`. It is then fine-tuned for **sentiment analysis** on the [glue sst2 dataset](https://huggingface.co/datasets/glue/viewer/sst2).
|
|
|
|
|
39 |
|
40 |
## Training and evaluation data
|
41 |
|
42 |
+
This model was only trained on 10000 samples, while the entire glue sst2 training set includes 67349 examples. This was done mainly to decrease training time.
|
43 |
|
44 |
## Training procedure
|
45 |
|
|
|
51 |
|
52 |
### Training results
|
53 |
|
54 |
+
- Accuracy (training): `94.08%`
|
55 |
+
- Accuracy (validation): `91.74%`
|
56 |
|
57 |
### Framework versions
|
58 |
|
59 |
- Transformers 4.35.2
|
60 |
- TensorFlow 2.15.0
|
61 |
- Datasets 2.16.1
|
62 |
+
- Tokenizers 0.15.0
|