Commit
•
183c342
1
Parent(s):
b47e07e
Update README.md
Browse files
README.md
CHANGED
@@ -7,20 +7,13 @@ datasets:
|
|
7 |
---
|
8 |
|
9 |
|
10 |
-
# Twitter-roBERTa-base for Sentiment Analysis
|
11 |
-
|
12 |
-
This is a RoBERTa-base model trained on ~124M tweets from January 2018 to December 2021, and finetuned for sentiment analysis with the TweetEval benchmark.
|
13 |
-
The original Twitter-based RoBERTa model can be found [here](https://huggingface.co/cardiffnlp/twitter-roberta-base-2021-124m) and the original reference paper is [TweetEval](https://github.com/cardiffnlp/tweeteval). This model is suitable for English.
|
14 |
-
|
15 |
-
- Reference Paper: [TimeLMs paper](https://arxiv.org/abs/2202.03829).
|
16 |
-
- Git Repo: [TimeLMs official repository](https://github.com/cardiffnlp/timelms).
|
17 |
|
18 |
<b>Labels</b>:
|
19 |
0 -> Negative;
|
20 |
1 -> Neutral;
|
21 |
2 -> Positive
|
22 |
|
23 |
-
This sentiment analysis model has been integrated into [TweetNLP](https://github.com/cardiffnlp/tweetnlp). You can access the demo [here](https://tweetnlp.org).
|
24 |
|
25 |
## Example Pipeline
|
26 |
```python
|
@@ -31,102 +24,3 @@ sentiment_task("Covid cases are increasing fast!")
|
|
31 |
```
|
32 |
[{'label': 'Negative', 'score': 0.7236}]
|
33 |
```
|
34 |
-
|
35 |
-
## Full classification example
|
36 |
-
|
37 |
-
```python
|
38 |
-
from transformers import AutoModelForSequenceClassification
|
39 |
-
from transformers import TFAutoModelForSequenceClassification
|
40 |
-
from transformers import AutoTokenizer, AutoConfig
|
41 |
-
import numpy as np
|
42 |
-
from scipy.special import softmax
|
43 |
-
# Preprocess text (username and link placeholders)
|
44 |
-
def preprocess(text):
|
45 |
-
new_text = []
|
46 |
-
for t in text.split(" "):
|
47 |
-
t = '@user' if t.startswith('@') and len(t) > 1 else t
|
48 |
-
t = 'http' if t.startswith('http') else t
|
49 |
-
new_text.append(t)
|
50 |
-
return " ".join(new_text)
|
51 |
-
MODEL = f"cardiffnlp/twitter-roberta-base-sentiment-latest"
|
52 |
-
tokenizer = AutoTokenizer.from_pretrained(MODEL)
|
53 |
-
config = AutoConfig.from_pretrained(MODEL)
|
54 |
-
# PT
|
55 |
-
model = AutoModelForSequenceClassification.from_pretrained(MODEL)
|
56 |
-
#model.save_pretrained(MODEL)
|
57 |
-
text = "Covid cases are increasing fast!"
|
58 |
-
text = preprocess(text)
|
59 |
-
encoded_input = tokenizer(text, return_tensors='pt')
|
60 |
-
output = model(**encoded_input)
|
61 |
-
scores = output[0][0].detach().numpy()
|
62 |
-
scores = softmax(scores)
|
63 |
-
# # TF
|
64 |
-
# model = TFAutoModelForSequenceClassification.from_pretrained(MODEL)
|
65 |
-
# model.save_pretrained(MODEL)
|
66 |
-
# text = "Covid cases are increasing fast!"
|
67 |
-
# encoded_input = tokenizer(text, return_tensors='tf')
|
68 |
-
# output = model(encoded_input)
|
69 |
-
# scores = output[0][0].numpy()
|
70 |
-
# scores = softmax(scores)
|
71 |
-
# Print labels and scores
|
72 |
-
ranking = np.argsort(scores)
|
73 |
-
ranking = ranking[::-1]
|
74 |
-
for i in range(scores.shape[0]):
|
75 |
-
l = config.id2label[ranking[i]]
|
76 |
-
s = scores[ranking[i]]
|
77 |
-
print(f"{i+1}) {l} {np.round(float(s), 4)}")
|
78 |
-
```
|
79 |
-
|
80 |
-
Output:
|
81 |
-
|
82 |
-
```
|
83 |
-
1) Negative 0.7236
|
84 |
-
2) Neutral 0.2287
|
85 |
-
3) Positive 0.0477
|
86 |
-
```
|
87 |
-
|
88 |
-
|
89 |
-
### References
|
90 |
-
```
|
91 |
-
@inproceedings{camacho-collados-etal-2022-tweetnlp,
|
92 |
-
title = "{T}weet{NLP}: Cutting-Edge Natural Language Processing for Social Media",
|
93 |
-
author = "Camacho-collados, Jose and
|
94 |
-
Rezaee, Kiamehr and
|
95 |
-
Riahi, Talayeh and
|
96 |
-
Ushio, Asahi and
|
97 |
-
Loureiro, Daniel and
|
98 |
-
Antypas, Dimosthenis and
|
99 |
-
Boisson, Joanne and
|
100 |
-
Espinosa Anke, Luis and
|
101 |
-
Liu, Fangyu and
|
102 |
-
Mart{\'\i}nez C{\'a}mara, Eugenio" and others,
|
103 |
-
booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
|
104 |
-
month = dec,
|
105 |
-
year = "2022",
|
106 |
-
address = "Abu Dhabi, UAE",
|
107 |
-
publisher = "Association for Computational Linguistics",
|
108 |
-
url = "https://aclanthology.org/2022.emnlp-demos.5",
|
109 |
-
pages = "38--49"
|
110 |
-
}
|
111 |
-
|
112 |
-
```
|
113 |
-
|
114 |
-
```
|
115 |
-
@inproceedings{loureiro-etal-2022-timelms,
|
116 |
-
title = "{T}ime{LM}s: Diachronic Language Models from {T}witter",
|
117 |
-
author = "Loureiro, Daniel and
|
118 |
-
Barbieri, Francesco and
|
119 |
-
Neves, Leonardo and
|
120 |
-
Espinosa Anke, Luis and
|
121 |
-
Camacho-collados, Jose",
|
122 |
-
booktitle = "Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics: System Demonstrations",
|
123 |
-
month = may,
|
124 |
-
year = "2022",
|
125 |
-
address = "Dublin, Ireland",
|
126 |
-
publisher = "Association for Computational Linguistics",
|
127 |
-
url = "https://aclanthology.org/2022.acl-demo.25",
|
128 |
-
doi = "10.18653/v1/2022.acl-demo.25",
|
129 |
-
pages = "251--260"
|
130 |
-
}
|
131 |
-
|
132 |
-
```
|
|
|
7 |
---
|
8 |
|
9 |
|
10 |
+
# Twitter-roBERTa-base for Sentiment Analysis
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
|
12 |
<b>Labels</b>:
|
13 |
0 -> Negative;
|
14 |
1 -> Neutral;
|
15 |
2 -> Positive
|
16 |
|
|
|
17 |
|
18 |
## Example Pipeline
|
19 |
```python
|
|
|
24 |
```
|
25 |
[{'label': 'Negative', 'score': 0.7236}]
|
26 |
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|