Spaces:
Sleeping
Sleeping
File size: 1,491 Bytes
c1430da a2f7f62 75c4517 a2f7f62 75c4517 a2f7f62 c1430da a2f7f62 c1430da 75c4517 a2f7f62 75c4517 a2f7f62 75c4517 a2f7f62 75c4517 c1430da |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 |
from typing import List, Optional
try:
from transformers import AutoModelForSequenceClassification, AutoTokenizer
import torch
except ImportError:
AutoModelForSequenceClassification = None
AutoTokenizer = None
torch = None
class ThreatModel:
"""
Transformer wrapper. If transformers is not installed,
falls back to dummy mode and returns empty probabilities.
"""
def __init__(self, model_path: str, device: Optional[str] = None):
self.available = AutoModelForSequenceClassification is not None and torch is not None
self.model = None
self.tokenizer = None
self.device = "cpu"
if not self.available:
return
self.device = device or ("cuda" if torch.cuda.is_available() else "cpu")
self.tokenizer = AutoTokenizer.from_pretrained(model_path)
self.model = AutoModelForSequenceClassification.from_pretrained(model_path)
self.model.to(self.device)
def predict_proba(self, text: str) -> List[float]:
if not self.available or self.model is None or self.tokenizer is None:
return []
inputs = self.tokenizer(
text,
return_tensors="pt",
truncation=True,
padding=True
).to(self.device)
with torch.no_grad():
outputs = self.model(**inputs)
logits = outputs.logits
probs = torch.softmax(logits, dim=-1).cpu().tolist()[0]
return probs |