Spaces:
Runtime error
Runtime error
Create app_.py
Browse files
app_.py
ADDED
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from transformers import AutoTokenizer, AutoModel
|
2 |
+
import torch
|
3 |
+
import torch.nn.functional as F
|
4 |
+
import pandas as pd
|
5 |
+
|
6 |
+
# Function to perform mean pooling
|
7 |
+
def mean_pooling(model_output, attention_mask):
|
8 |
+
token_embeddings = model_output[0]
|
9 |
+
input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
|
10 |
+
return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)
|
11 |
+
|
12 |
+
# Load CSV data
|
13 |
+
df = pd.read_csv('/path/to/your/csvfile.csv')
|
14 |
+
|
15 |
+
# Assuming 'review' is the column with the text you want to encode
|
16 |
+
sentences = df['review'].tolist()
|
17 |
+
|
18 |
+
# Load model and tokenizer from HuggingFace Hub
|
19 |
+
tokenizer = AutoTokenizer.from_pretrained('sentence-transformers/all-MiniLM-L6-v2')
|
20 |
+
model = AutoModel.from_pretrained('sentence-transformers/all-MiniLM-L6-v2')
|
21 |
+
|
22 |
+
# Tokenize sentences
|
23 |
+
encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
|
24 |
+
|
25 |
+
# Compute token embeddings
|
26 |
+
with torch.no_grad():
|
27 |
+
model_output = model(**encoded_input)
|
28 |
+
|
29 |
+
# Perform pooling
|
30 |
+
sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
|
31 |
+
|
32 |
+
# Normalize embeddings
|
33 |
+
sentence_embeddings = F.normalize(sentence_embeddings, p=2, dim=1)
|
34 |
+
|
35 |
+
print("Sentence embeddings:")
|
36 |
+
print(sentence_embeddings)
|