Update README.md
Browse files
README.md
CHANGED
@@ -14,6 +14,7 @@ The **miCSE** language model is trained for sentence similarity computation. Tra
|
|
14 |
|
15 |
```shell
|
16 |
from transformers import AutoTokenizer, AutoModel
|
|
|
17 |
|
18 |
tokenizer = AutoTokenizer.from_pretrained("sap-ai-research/miCSE")
|
19 |
|
@@ -43,6 +44,17 @@ outputs = model(**batch, output_hidden_states=True, return_dict=True)
|
|
43 |
|
44 |
embeddings = outputs.last_hidden_state[:,0]
|
45 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
```
|
47 |
|
48 |
|
|
|
14 |
|
15 |
```shell
|
16 |
from transformers import AutoTokenizer, AutoModel
|
17 |
+
import torch.nn as nn
|
18 |
|
19 |
tokenizer = AutoTokenizer.from_pretrained("sap-ai-research/miCSE")
|
20 |
|
|
|
44 |
|
45 |
embeddings = outputs.last_hidden_state[:,0]
|
46 |
|
47 |
+
# Define similarity metric, e.g., cosine similarity
|
48 |
+
|
49 |
+
sim = nn.CosineSimilarity(dim=-1)
|
50 |
+
|
51 |
+
# Compute similarity between the **first** and the **second** sentence
|
52 |
+
|
53 |
+
cos_sim = sim(embeddings.unsqueeze(1),
|
54 |
+
embeddings.unsqueeze(0))
|
55 |
+
|
56 |
+
print(f"Distance: {cos_sim[0,1].detach().item()}")
|
57 |
+
|
58 |
```
|
59 |
|
60 |
|