File size: 4,271 Bytes
e479458
 
 
 
 
 
 
 
 
 
 
 
 
 
 
dae37dc
e479458
 
 
 
 
 
 
dae37dc
e479458
dae37dc
e479458
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5dfc1b2
a0fa569
ee65246
 
 
3e7469b
5dfc1b2
3e7469b
 
5dfc1b2
3e7469b
ee65246
3e7469b
 
 
ee65246
3e7469b
5dfc1b2
 
3e7469b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5dfc1b2
3e7469b
 
 
 
 
5dfc1b2
3e7469b
 
5dfc1b2
3e7469b
 
5dfc1b2
3e7469b
 
5dfc1b2
 
 
 
 
 
3e7469b
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
---
library_name: transformers
license: apache-2.0
base_model: bert-base-uncased
tags:
- generated_from_trainer
datasets:
- financial_phrasebank
metrics:
- f1
model-index:
- name: FIN_BERT_sentiment
  results:
  - task:
      type: text-classification
      name: Text Classification
    dataset:
      name: financial_phrasebank
      type: financial_phrasebank
      config: sentences_66agree
      split: train
      args: sentences_66agree
    metrics:
    - type: f1
      value: 0.8890693407692588
      name: F1
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# FIN_BERT_sentiment

This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the financial_phrasebank dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4905
- F1: 0.8891
- Acc: 0.8886

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 5

### Training results

| Training Loss | Epoch | Step | Validation Loss | F1     | Acc    |
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
| 0.5295        | 1.0   | 211  | 0.3757          | 0.8731 | 0.8720 |
| 0.2174        | 2.0   | 422  | 0.3117          | 0.8911 | 0.8910 |
| 0.1129        | 3.0   | 633  | 0.4066          | 0.8886 | 0.8874 |
| 0.0459        | 4.0   | 844  | 0.4923          | 0.8896 | 0.8886 |
| 0.0275        | 5.0   | 1055 | 0.4905          | 0.8891 | 0.8886 |


### Framework versions

- Transformers 4.46.2
- Pytorch 2.5.1
- Datasets 3.1.0
- Tokenizers 0.20.3

## Code to use model as pipeline classifier


```python
import matplotlib.pyplot as plt
import plotly.graph_objects as go
from IPython.display import display, HTML
import numpy as np
from transformers import pipeline
%matplotlib inline

# Pipelines
classifier = pipeline("text-classification", model="Sharpaxis/Finance_DistilBERT_sentiment", top_k=None)
pipe = pipeline("text-classification", model="Sharpaxis/News_classification_distilbert")

def finance_text_predictor(text):
    text = str(text)
    out = classifier(text)[0]
    type_news = pipe(text)[0]
    
    # Display news type and text in HTML
    if type_news['label'] == 'LABEL_1':
        display(HTML(f"""
        <div style="border: 2px solid red; padding: 10px; margin: 10px; background-color: #ffe6e6; color: black; font-weight: bold;">
            IMPORTANT TECH/FIN News<br>
            <div style="margin-top: 10px; font-weight: normal; font-size: 14px; color: darkred;">{text}</div>
        </div>
        """))
    elif type_news['label'] == 'LABEL_0':
        display(HTML(f"""
        <div style="border: 2px solid green; padding: 10px; margin: 10px; background-color: #e6ffe6; color: black; font-weight: bold;">
            NON IMPORTANT NEWS<br>
            <div style="margin-top: 10px; font-weight: normal; font-size: 14px; color: darkgreen;">{text}</div>
        </div>
        """))
    
    # Sentiment analysis scores
    scores = [sample['score'] for sample in out]
    labels = [sample['label'] for sample in out]
    label_map = {'LABEL_0': "Negative", 'LABEL_1': "Neutral", 'LABEL_2': "Positive"}
    sentiments = [label_map[label] for label in labels]
    
    print("SCORES")
    for i in range(len(scores)):
        print(f"{sentiments[i]} : {scores[i]:.4f}")
    
    print(f"Sentiment of text is {sentiments[np.argmax(scores)]}")
    
    # Bar chart for sentiment scores
    fig = go.Figure(
        data=[go.Bar(x=sentiments, y=scores, marker=dict(color=["red", "blue", "green"]), width=0.3)]
    )
    fig.update_layout(
        title="Sentiment Analysis Scores",
        xaxis_title="Sentiments",
        yaxis_title="Scores",
        template="plotly_dark"
    )
    fig.show()