fixed some issues w/ the uncertainty
Browse files- api/predict.py +8 -8
api/predict.py
CHANGED
|
@@ -175,7 +175,7 @@ def ensemblePredict(text):
|
|
| 175 |
|
| 176 |
finalPred = 1 if isFake else 0
|
| 177 |
agreementCount = sum(1 for p in allPreds if p == finalPred)
|
| 178 |
-
agreement =
|
| 179 |
|
| 180 |
del weightedProbs, allPreds
|
| 181 |
gc.collect()
|
|
@@ -264,23 +264,23 @@ def predict_review(text):
|
|
| 264 |
|
| 265 |
confidence = max(fakeProb, genuineProb)
|
| 266 |
|
|
|
|
|
|
|
|
|
|
| 267 |
if agreement < uncertaintyThreshold:
|
| 268 |
prediction = "uncertain"
|
| 269 |
isFakeOutput = None
|
| 270 |
-
else:
|
| 271 |
-
prediction = "fake" if isFake else "genuine"
|
| 272 |
-
isFakeOutput = isFake
|
| 273 |
|
| 274 |
lengthCat = getLengthCategory(cleaned)
|
| 275 |
tokenCount = len(cleaned.split())
|
| 276 |
|
| 277 |
return {
|
| 278 |
"prediction": prediction,
|
| 279 |
-
"confidence":
|
| 280 |
"is_fake": isFakeOutput,
|
| 281 |
-
"model_agreement":
|
| 282 |
-
"fake_probability":
|
| 283 |
-
"genuine_probability":
|
| 284 |
"length_category": lengthCat,
|
| 285 |
"token_count": tokenCount
|
| 286 |
}
|
|
|
|
| 175 |
|
| 176 |
finalPred = 1 if isFake else 0
|
| 177 |
agreementCount = sum(1 for p in allPreds if p == finalPred)
|
| 178 |
+
agreement = agreementCount / len(allPreds)
|
| 179 |
|
| 180 |
del weightedProbs, allPreds
|
| 181 |
gc.collect()
|
|
|
|
| 264 |
|
| 265 |
confidence = max(fakeProb, genuineProb)
|
| 266 |
|
| 267 |
+
prediction = "fake" if isFake else "genuine"
|
| 268 |
+
isFakeOutput = isFake
|
| 269 |
+
|
| 270 |
if agreement < uncertaintyThreshold:
|
| 271 |
prediction = "uncertain"
|
| 272 |
isFakeOutput = None
|
|
|
|
|
|
|
|
|
|
| 273 |
|
| 274 |
lengthCat = getLengthCategory(cleaned)
|
| 275 |
tokenCount = len(cleaned.split())
|
| 276 |
|
| 277 |
return {
|
| 278 |
"prediction": prediction,
|
| 279 |
+
"confidence": confidence,
|
| 280 |
"is_fake": isFakeOutput,
|
| 281 |
+
"model_agreement": agreement,
|
| 282 |
+
"fake_probability": fakeProb,
|
| 283 |
+
"genuine_probability": genuineProb,
|
| 284 |
"length_category": lengthCat,
|
| 285 |
"token_count": tokenCount
|
| 286 |
}
|