Update README.md
Browse files
README.md
CHANGED
@@ -9,6 +9,33 @@ This is an embedding model for clinical papers
|
|
9 |
|
10 |
## How to Use
|
11 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
```python
|
13 |
from transformers import AutoTokenizer, AutoModel
|
14 |
import torch
|
@@ -36,5 +63,5 @@ model.to(DEVICE)
|
|
36 |
with torch.no_grad():
|
37 |
output = model(**dummy_input)
|
38 |
|
39 |
-
embeddings = output.last_hidden_state[:, 0]
|
40 |
```
|
|
|
9 |
|
10 |
## How to Use
|
11 |
|
12 |
+
### Simple finetuned model
|
13 |
+
|
14 |
+
```python
|
15 |
+
from transformers import AutoTokenizer, AutoModel
|
16 |
+
import torch
|
17 |
+
|
18 |
+
DEVICE = "cuda:0" if torch.cuda.is_available() else "cpu"
|
19 |
+
|
20 |
+
PATH = "josh-oo/aspect-based-embeddings-v3"
|
21 |
+
|
22 |
+
tokenizer = AutoTokenizer.from_pretrained(PATH)
|
23 |
+
model = AutoModel.from_pretrained(PATH)
|
24 |
+
|
25 |
+
dummy_text = "This is a title of a medical paper"
|
26 |
+
dummy_input = tokenizer([dummy_text], return_tensors="pt")
|
27 |
+
|
28 |
+
dummy_input.to(DEVICE)
|
29 |
+
model.to(DEVICE)
|
30 |
+
|
31 |
+
with torch.no_grad():
|
32 |
+
output = model(**dummy_input)
|
33 |
+
|
34 |
+
embeddings = output.last_hidden_state[:, 0] #cls pooling
|
35 |
+
```
|
36 |
+
|
37 |
+
### Aspect guided model
|
38 |
+
|
39 |
```python
|
40 |
from transformers import AutoTokenizer, AutoModel
|
41 |
import torch
|
|
|
63 |
with torch.no_grad():
|
64 |
output = model(**dummy_input)
|
65 |
|
66 |
+
embeddings = output.last_hidden_state[:, 0] #cls pooling
|
67 |
```
|