Update app.py
Browse files
app.py
CHANGED
@@ -1,36 +1,36 @@
|
|
1 |
-
import torch.nn.functional as F
|
2 |
-
import torch
|
3 |
-
from pinecone_text.sparse import SpladeEncoder
|
4 |
-
import re
|
5 |
-
from fastapi import FastAPI, Depends
|
6 |
-
from fastapi_health import health
|
7 |
-
from fastapi import FastAPI, Query
|
8 |
-
|
9 |
-
|
10 |
-
def get_session():
|
11 |
-
return True
|
12 |
-
|
13 |
-
def is_database_online(session: bool = Depends(get_session)):
|
14 |
-
return session
|
15 |
-
|
16 |
-
app = FastAPI()
|
17 |
-
app.add_api_route("/healthz", health([is_database_online]))
|
18 |
-
|
19 |
-
class Load_EmbeddingModels:
|
20 |
-
def __init__(self
|
21 |
-
self.device = "cuda" if torch.cuda.is_available() else "cpu"
|
22 |
-
self.sparse_model = SpladeEncoder(device=self.device)
|
23 |
-
|
24 |
-
def get_single_sparse_text_embedding(self, df_chunk):
|
25 |
-
return self.sparse_model.encode_documents(df_chunk)
|
26 |
-
|
27 |
-
|
28 |
-
model = Load_EmbeddingModels()
|
29 |
-
|
30 |
-
@app.post("/embed-text-sparse/")
|
31 |
-
async def embed_text(text: str = Query(...)):
|
32 |
-
try:
|
33 |
-
embeddings = model.get_single_sparse_text_embedding(text)
|
34 |
-
return embeddings
|
35 |
-
except Exception as e:
|
36 |
print(f'Error: {e}')
|
|
|
1 |
+
import torch.nn.functional as F
|
2 |
+
import torch
|
3 |
+
from pinecone_text.sparse import SpladeEncoder
|
4 |
+
import re
|
5 |
+
from fastapi import FastAPI, Depends
|
6 |
+
from fastapi_health import health
|
7 |
+
from fastapi import FastAPI, Query
|
8 |
+
|
9 |
+
|
10 |
+
def get_session():
|
11 |
+
return True
|
12 |
+
|
13 |
+
def is_database_online(session: bool = Depends(get_session)):
|
14 |
+
return session
|
15 |
+
|
16 |
+
app = FastAPI()
|
17 |
+
app.add_api_route("/healthz", health([is_database_online]))
|
18 |
+
|
19 |
+
class Load_EmbeddingModels:
|
20 |
+
def __init__(self):
|
21 |
+
self.device = "cuda" if torch.cuda.is_available() else "cpu"
|
22 |
+
self.sparse_model = SpladeEncoder(device=self.device)
|
23 |
+
|
24 |
+
def get_single_sparse_text_embedding(self, df_chunk):
|
25 |
+
return self.sparse_model.encode_documents(df_chunk)
|
26 |
+
|
27 |
+
|
28 |
+
model = Load_EmbeddingModels()
|
29 |
+
|
30 |
+
@app.post("/embed-text-sparse/")
|
31 |
+
async def embed_text(text: str = Query(...)):
|
32 |
+
try:
|
33 |
+
embeddings = model.get_single_sparse_text_embedding(text)
|
34 |
+
return embeddings
|
35 |
+
except Exception as e:
|
36 |
print(f'Error: {e}')
|