File size: 4,732 Bytes
db6b916
 
 
 
 
 
 
 
 
 
e6e7f5e
 
 
c961f32
e6e7f5e
 
 
c961f32
e6e7f5e
 
 
 
 
 
6b0f25b
 
 
 
 
 
 
e6e7f5e
 
 
 
 
 
 
 
 
 
 
6b0f25b
 
e6e7f5e
 
 
 
 
 
 
 
 
6b0f25b
e6e7f5e
 
 
 
6b0f25b
e6e7f5e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6b0f25b
e6e7f5e
 
 
 
6b0f25b
e6e7f5e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
---
language: en
tags:
- timelms
- twitter
license: mit
datasets:
- twitter-api
---

# Twitter June 2020 (RoBERTa-base, 99M)

This is a RoBERTa-base model trained on 98.66M tweets until the end of June 2020.
More details and performance scores are available in the [TimeLMs paper](https://arxiv.org/abs/2202.03829).

Below, we provide some usage examples using the standard Transformers interface. For another interface more suited to comparing predictions and perplexity scores between models trained at different temporal intervals, check the [TimeLMs repository](https://github.com/cardiffnlp/timelms).

For other models trained until different periods, check this [table](https://github.com/cardiffnlp/timelms#released-models).

## Preprocess Text 
Replace usernames and links for placeholders: "@user" and "http".
If you're interested in retaining verified users which were also retained during training, you may keep the users listed [here](https://github.com/cardiffnlp/timelms/tree/main/data).
```python
def preprocess(text):
    preprocessed_text = []
    for t in text.split():
        if len(t) > 1:
            t = '@user' if t[0] == '@' and t.count('@') == 1 else t
            t = 'http' if t.startswith('http') else t
        preprocessed_text.append(t)
    return ' '.join(preprocessed_text)
```

## Example Masked Language Model 

```python
from transformers import pipeline, AutoTokenizer

MODEL = "cardiffnlp/twitter-roberta-base-jun2020"
fill_mask = pipeline("fill-mask", model=MODEL, tokenizer=MODEL)
tokenizer = AutoTokenizer.from_pretrained(MODEL)

def pprint(candidates, n):
    for i in range(n):
        token = tokenizer.decode(candidates[i]['token'])
        score = candidates[i]['score']
        print("%d) %.5f %s" % (i+1, score, token))

texts = [
    "So glad I'm <mask> vaccinated.",
    "I keep forgetting to bring a <mask>.",
    "Looking forward to watching <mask> Game tonight!",
]

for text in texts:
    t = preprocess(text)
    print(f"{'-'*30}\n{t}")
    candidates = fill_mask(t)
    pprint(candidates, 5)
```

Output: 

```
------------------------------
So glad I'm <mask> vaccinated.
1) 0.52684  not
2) 0.18349  getting
3) 0.07971  fully
4) 0.05598  being
5) 0.02347  self
------------------------------
I keep forgetting to bring a <mask>.
1) 0.13266  mask
2) 0.04859  book
3) 0.04851  laptop
4) 0.03123  pillow
5) 0.02747  blanket
------------------------------
Looking forward to watching <mask> Game tonight!
1) 0.35750  The
2) 0.32703  the
3) 0.13048  End
4) 0.02261  this
5) 0.01066  This
```

## Example Tweet Embeddings
```python
from transformers import AutoTokenizer, AutoModel, TFAutoModel
import numpy as np
from scipy.spatial.distance import cosine
from collections import Counter

def get_embedding(text):  # naive approach for demonstration
  text = preprocess(text)
  encoded_input = tokenizer(text, return_tensors='pt')
  features = model(**encoded_input)
  features = features[0].detach().cpu().numpy() 
  return np.mean(features[0], axis=0) 


MODEL = "cardiffnlp/twitter-roberta-base-jun2020"
tokenizer = AutoTokenizer.from_pretrained(MODEL)
model = AutoModel.from_pretrained(MODEL)

query = "The book was awesome"
tweets = ["I just ordered fried chicken 🐣", 
          "The movie was great",
          "What time is the next game?",
          "Just finished reading 'Embeddings in NLP'"]

sims = Counter()
for tweet in tweets:
    sim = 1 - cosine(get_embedding(query), get_embedding(tweet))
    sims[tweet] = sim

print('Most similar to: ', query)
print(f"{'-'*30}")
for idx, (tweet, sim) in enumerate(sims.most_common()):
    print("%d) %.5f %s" % (idx+1, sim, tweet))
```
Output: 

```
Most similar to:  The book was awesome
------------------------------
1) 0.99078 The movie was great
2) 0.96610 Just finished reading 'Embeddings in NLP'
3) 0.96095 What time is the next game?
4) 0.95855 I just ordered fried chicken 🐣
```

## Example Feature Extraction 

```python
from transformers import AutoTokenizer, AutoModel, TFAutoModel
import numpy as np

MODEL = "cardiffnlp/twitter-roberta-base-jun2020"
tokenizer = AutoTokenizer.from_pretrained(MODEL)

text = "Good night 😊"
text = preprocess(text)

# Pytorch
model = AutoModel.from_pretrained(MODEL)
encoded_input = tokenizer(text, return_tensors='pt')
features = model(**encoded_input)
features = features[0].detach().cpu().numpy() 
features_mean = np.mean(features[0], axis=0) 
#features_max = np.max(features[0], axis=0)

# # Tensorflow
# model = TFAutoModel.from_pretrained(MODEL)
# encoded_input = tokenizer(text, return_tensors='tf')
# features = model(encoded_input)
# features = features[0].numpy()
# features_mean = np.mean(features[0], axis=0) 
# #features_max = np.max(features[0], axis=0)
```