j-hartmann commited on
Commit
5f849fb
·
1 Parent(s): a0a875f

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +12 -14
app.py CHANGED
@@ -10,6 +10,14 @@ from numpy.linalg import norm
10
  # compute dot product of inputs
11
  # summary function - test for single gradio function interfrace
12
  def gr_cosine_similarity(sentence1, sentence2):
 
 
 
 
 
 
 
 
13
  # Create class for data preparation
14
  class SimpleDataset:
15
  def __init__(self, tokenized_texts):
@@ -21,19 +29,10 @@ def gr_cosine_similarity(sentence1, sentence2):
21
  def __getitem__(self, idx):
22
  return {k: v[idx] for k, v in self.tokenized_texts.items()}
23
 
24
- # load tokenizer and model, create trainer
25
- model_name = "j-hartmann/emotion-english-distilroberta-base"
26
- tokenizer = AutoTokenizer.from_pretrained(model_name)
27
- model = AutoModelForSequenceClassification.from_pretrained(model_name)
28
- trainer = Trainer(model=model)
29
-
30
-
31
  # sentences in list
32
  lines_s = [sentence1, sentence2]
33
- print(type(sentence1), type(sentence2))
34
- print(sentence1, sentence2)
35
- print(lines_s)
36
-
37
  # Tokenize texts and create prediction data set
38
  tokenized_texts = tokenizer(lines_s, truncation=True, padding=True)
39
  pred_dataset = SimpleDataset(tokenized_texts)
@@ -49,7 +48,7 @@ def gr_cosine_similarity(sentence1, sentence2):
49
  temp = (np.exp(predictions[0])/np.exp(predictions[0]).sum(-1, keepdims=True)).tolist()
50
 
51
 
52
- # work in progress
53
  # container
54
  anger = []
55
  disgust = []
@@ -59,7 +58,6 @@ def gr_cosine_similarity(sentence1, sentence2):
59
  sadness = []
60
  surprise = []
61
 
62
- print(temp)
63
  # extract scores (as many entries as exist in pred_texts)
64
  for i in range(len(lines_s)):
65
  anger.append(temp[i][0])
@@ -74,7 +72,7 @@ def gr_cosine_similarity(sentence1, sentence2):
74
  # each include all values for both predictions
75
  v1 = temp[0]
76
  v2 = temp[1]
77
- print(type(v1), type(v2))
78
  # compute dot product of all
79
  dot_product = dot(v1, v2)
80
 
 
10
  # compute dot product of inputs
11
  # summary function - test for single gradio function interfrace
12
  def gr_cosine_similarity(sentence1, sentence2):
13
+
14
+ # load tokenizer and model, create trainer
15
+ model_name = "j-hartmann/emotion-english-distilroberta-base"
16
+ tokenizer = AutoTokenizer.from_pretrained(model_name)
17
+ model = AutoModelForSequenceClassification.from_pretrained(model_name)
18
+ trainer = Trainer(model=model)
19
+
20
+
21
  # Create class for data preparation
22
  class SimpleDataset:
23
  def __init__(self, tokenized_texts):
 
29
  def __getitem__(self, idx):
30
  return {k: v[idx] for k, v in self.tokenized_texts.items()}
31
 
32
+
 
 
 
 
 
 
33
  # sentences in list
34
  lines_s = [sentence1, sentence2]
35
+
 
 
 
36
  # Tokenize texts and create prediction data set
37
  tokenized_texts = tokenizer(lines_s, truncation=True, padding=True)
38
  pred_dataset = SimpleDataset(tokenized_texts)
 
48
  temp = (np.exp(predictions[0])/np.exp(predictions[0]).sum(-1, keepdims=True)).tolist()
49
 
50
 
51
+ # work in progress
52
  # container
53
  anger = []
54
  disgust = []
 
58
  sadness = []
59
  surprise = []
60
 
 
61
  # extract scores (as many entries as exist in pred_texts)
62
  for i in range(len(lines_s)):
63
  anger.append(temp[i][0])
 
72
  # each include all values for both predictions
73
  v1 = temp[0]
74
  v2 = temp[1]
75
+
76
  # compute dot product of all
77
  dot_product = dot(v1, v2)
78