seongyeon1
commited on
Commit
•
8e2429b
1
Parent(s):
df14000
Update README.md
Browse files
README.md
CHANGED
@@ -6,4 +6,95 @@ language:
|
|
6 |
metrics:
|
7 |
- accuracy
|
8 |
pipeline_tag: text-classification
|
9 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
metrics:
|
7 |
- accuracy
|
8 |
pipeline_tag: text-classification
|
9 |
+
---
|
10 |
+
# Model Card for Model ID
|
11 |
+
|
12 |
+
|
13 |
+
## Model Details
|
14 |
+
|
15 |
+
{'eval_loss': 0.2575262784957886,
|
16 |
+
'eval_accuracy': 0.9041,
|
17 |
+
'eval_runtime': 163.2129,
|
18 |
+
'eval_samples_per_second': 306.348,
|
19 |
+
'eval_steps_per_second': 9.576,
|
20 |
+
'epoch': 2.0}
|
21 |
+
|
22 |
+
|
23 |
+
### Model Description
|
24 |
+
|
25 |
+
- **Finetuned from model klue/bert :** [More Information Needed]
|
26 |
+
-
|
27 |
+
|
28 |
+
## Uses
|
29 |
+
|
30 |
+
- use to sentimental analysis task
|
31 |
+
|
32 |
+
## How to Get Started with the Model
|
33 |
+
|
34 |
+
```python
|
35 |
+
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
36 |
+
|
37 |
+
tokenizer = AutoTokenizer.from_pretrained("seongyeon1/klue-base-finetuned-nsmc")
|
38 |
+
model = AutoModelForSequenceClassification.from_pretrained("seongyeon1/klue-base-finetuned-nsmc")
|
39 |
+
```
|
40 |
+
|
41 |
+
```python
|
42 |
+
from transformers import pipeline
|
43 |
+
|
44 |
+
pipe = pipeline("text-classification", model="seongyeon1/klue-base-finetuned-nsmc")
|
45 |
+
pipe("진짜 별로더라") # [{'label': 'LABEL_0', 'score': 0.999700665473938}]
|
46 |
+
pipe("굿굿") # [{'label': 'LABEL_1', 'score': 0.9875587224960327}]
|
47 |
+
|
48 |
+
```
|
49 |
+
|
50 |
+
## Training Details
|
51 |
+
|
52 |
+
```
|
53 |
+
### Training Data
|
54 |
+
|
55 |
+
- nsmc datasets
|
56 |
+
|
57 |
+
|
58 |
+
#### Preprocessing
|
59 |
+
|
60 |
+
- bert's default is 512, but it costs a lot of time.
|
61 |
+
- maxlen = 55
|
62 |
+
![image/png](https://cdn-uploads.huggingface.co/production/uploads/634330a304d4ff28aeb8de56/t7axSlo4JI4bPLynUB3OP.png)
|
63 |
+
|
64 |
+
```python
|
65 |
+
def tokenize_function_with_max(examples, maxlen=maxlen):
|
66 |
+
encodings = tokenizer(examples['document'],max_length=maxlen, truncation=True, padding='max_length')
|
67 |
+
return encodings
|
68 |
+
|
69 |
+
|
70 |
+
#### Training Hyperparameters
|
71 |
+
|
72 |
+
- learning rate=2e-5, weight decay=0.01, batch size=32, epochs=2
|
73 |
+
|
74 |
+
#### Speeds, Sizes, Times [optional]
|
75 |
+
|
76 |
+
- about 40 minutes
|
77 |
+
|
78 |
+
|
79 |
+
[More Information Needed]
|
80 |
+
|
81 |
+
## Evaluation
|
82 |
+
|
83 |
+
### Testing Data, Factors & Metrics
|
84 |
+
|
85 |
+
#### Testing Data
|
86 |
+
|
87 |
+
- nsmc test datasets
|
88 |
+
|
89 |
+
[More Information Needed]
|
90 |
+
|
91 |
+
#### Factors
|
92 |
+
|
93 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
94 |
+
|
95 |
+
[More Information Needed]
|
96 |
+
|
97 |
+
#### Metrics
|
98 |
+
|
99 |
+
- accuracy
|
100 |
+
- label ratio is about almost balanced
|