Initial deploy of FastAPI model to Hugging Face Spaces

Dockerfile

@@ -0,0 +1,17 @@
+FROM python:3.10-slim
+
+# Set working directory
+WORKDIR /code
+
+# Install dependencies
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy app code
+COPY app /code/app
+
+# Set environment variables
+ENV TRANSFORMERS_CACHE=/code/cache
+
+# Expose port and run
+CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "7860"]

app/label_classes.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eeaed33f5b1679639ec637bca7f6c14857aa10ac84c201269527543e9a1002dd
+size 573

app/main.py

@@ -0,0 +1,30 @@
+from fastapi import FastAPI
+from pydantic import BaseModel
+from transformers import BertTokenizer, BertForSequenceClassification
+from sklearn.preprocessing import LabelEncoder
+import torch
+import numpy as np
+
+app = FastAPI()
+
+model = BertForSequenceClassification.from_pretrained("JaySenpai/bert-model")
+tokenizer = BertTokenizer.from_pretrained("JaySenpai/bert-model")
+model.eval()
+
+le = LabelEncoder()
+le.classes_ = np.load(r"C:\Users\jayes\Downloads\label_classes.npy", allow_pickle=True)
+
+class TextInput(BaseModel):
+    text: str
+
+@app.post("/predict")
+async def predict(data: TextInput):
+    text = data.text  # ✅ Dot notation
+    inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)
+    with torch.no_grad():
+        outputs = model(**inputs)
+        probs = torch.nn.functional.softmax(outputs.logits, dim=-1)
+        pred_class = torch.argmax(probs, dim=1).item()
+        pred_label = le.classes_[pred_class]
+        confidence = probs[0][pred_class].item()
+    return {"predicted_category": pred_label, "confidence": confidence}