Jiahuita
commited on
Commit
•
50fd5a6
1
Parent(s):
cec65b9
error changes
Browse files- app.py +50 -18
- requirements.txt +0 -1
app.py
CHANGED
@@ -4,16 +4,29 @@ from pydantic import BaseModel
|
|
4 |
from tensorflow.keras.models import load_model
|
5 |
from tensorflow.keras.preprocessing.text import tokenizer_from_json
|
6 |
from tensorflow.keras.preprocessing.sequence import pad_sequences
|
|
|
7 |
import json
|
8 |
from typing import Union, List
|
9 |
|
10 |
app = FastAPI()
|
11 |
|
12 |
-
#
|
13 |
-
model =
|
14 |
-
|
15 |
-
|
16 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
17 |
|
18 |
class PredictionInput(BaseModel):
|
19 |
text: Union[str, List[str]]
|
@@ -22,35 +35,54 @@ class PredictionOutput(BaseModel):
|
|
22 |
label: str
|
23 |
score: float
|
24 |
|
25 |
-
@app.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
26 |
async def predict(input_data: PredictionInput):
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
try:
|
28 |
-
#
|
29 |
texts = input_data.text if isinstance(input_data.text, list) else [input_data.text]
|
30 |
|
31 |
# Preprocess
|
32 |
sequences = tokenizer.texts_to_sequences(texts)
|
33 |
-
padded = pad_sequences(sequences, maxlen=41) #
|
34 |
|
35 |
-
#
|
36 |
-
predictions = model.predict(padded)
|
37 |
|
38 |
-
#
|
39 |
results = []
|
40 |
for pred in predictions:
|
41 |
-
|
42 |
-
|
43 |
results.append({
|
44 |
"label": label,
|
45 |
-
"score": score
|
46 |
})
|
47 |
|
48 |
# Return single result if input was single string
|
49 |
return results[0] if isinstance(input_data.text, str) else results
|
50 |
-
|
51 |
except Exception as e:
|
52 |
raise HTTPException(status_code=500, detail=str(e))
|
53 |
|
54 |
-
@app.
|
55 |
-
async def
|
56 |
-
|
|
|
|
|
|
|
|
|
|
4 |
from tensorflow.keras.models import load_model
|
5 |
from tensorflow.keras.preprocessing.text import tokenizer_from_json
|
6 |
from tensorflow.keras.preprocessing.sequence import pad_sequences
|
7 |
+
import numpy as np
|
8 |
import json
|
9 |
from typing import Union, List
|
10 |
|
11 |
app = FastAPI()
|
12 |
|
13 |
+
# Global variables for model and tokenizer
|
14 |
+
model = None
|
15 |
+
tokenizer = None
|
16 |
+
|
17 |
+
def load_model_and_tokenizer():
|
18 |
+
global model, tokenizer
|
19 |
+
try:
|
20 |
+
model = load_model('news_classifier.h5')
|
21 |
+
with open('tokenizer.json', 'r') as f:
|
22 |
+
tokenizer_data = json.load(f)
|
23 |
+
tokenizer = tokenizer_from_json(tokenizer_data)
|
24 |
+
except Exception as e:
|
25 |
+
print(f"Error loading model or tokenizer: {str(e)}")
|
26 |
+
raise e
|
27 |
+
|
28 |
+
# Load on startup
|
29 |
+
load_model_and_tokenizer()
|
30 |
|
31 |
class PredictionInput(BaseModel):
|
32 |
text: Union[str, List[str]]
|
|
|
35 |
label: str
|
36 |
score: float
|
37 |
|
38 |
+
@app.get("/")
|
39 |
+
def read_root():
|
40 |
+
return {
|
41 |
+
"message": "News Source Classifier API",
|
42 |
+
"model_type": "LSTM",
|
43 |
+
"version": "1.0",
|
44 |
+
"status": "ready" if model and tokenizer else "not_loaded"
|
45 |
+
}
|
46 |
+
|
47 |
+
@app.post("/predict", response_model=Union[PredictionOutput, List[PredictionOutput]])
|
48 |
async def predict(input_data: PredictionInput):
|
49 |
+
if not model or not tokenizer:
|
50 |
+
try:
|
51 |
+
load_model_and_tokenizer()
|
52 |
+
except Exception as e:
|
53 |
+
raise HTTPException(status_code=500, detail="Model not loaded")
|
54 |
+
|
55 |
try:
|
56 |
+
# Handle both single string and list inputs
|
57 |
texts = input_data.text if isinstance(input_data.text, list) else [input_data.text]
|
58 |
|
59 |
# Preprocess
|
60 |
sequences = tokenizer.texts_to_sequences(texts)
|
61 |
+
padded = pad_sequences(sequences, maxlen=41) # Match your model's input length
|
62 |
|
63 |
+
# Get predictions
|
64 |
+
predictions = model.predict(padded, verbose=0)
|
65 |
|
66 |
+
# Process results
|
67 |
results = []
|
68 |
for pred in predictions:
|
69 |
+
label = "foxnews" if pred[1] > 0.5 else "nbc"
|
70 |
+
score = float(pred[1] if label == "foxnews" else 1 - pred[1])
|
71 |
results.append({
|
72 |
"label": label,
|
73 |
+
"score": score
|
74 |
})
|
75 |
|
76 |
# Return single result if input was single string
|
77 |
return results[0] if isinstance(input_data.text, str) else results
|
78 |
+
|
79 |
except Exception as e:
|
80 |
raise HTTPException(status_code=500, detail=str(e))
|
81 |
|
82 |
+
@app.post("/reload")
|
83 |
+
async def reload_model():
|
84 |
+
try:
|
85 |
+
load_model_and_tokenizer()
|
86 |
+
return {"message": "Model reloaded successfully"}
|
87 |
+
except Exception as e:
|
88 |
+
raise HTTPException(status_code=500, detail=str(e))
|
requirements.txt
CHANGED
@@ -3,5 +3,4 @@ fastapi>=0.68.0
|
|
3 |
uvicorn>=0.15.0
|
4 |
pydantic>=1.8.2
|
5 |
numpy>=1.19.2
|
6 |
-
scikit-learn>=0.24.2
|
7 |
python-multipart
|
|
|
3 |
uvicorn>=0.15.0
|
4 |
pydantic>=1.8.2
|
5 |
numpy>=1.19.2
|
|
|
6 |
python-multipart
|