Upload 3 files

Dockerfile

@@ -0,0 +1,11 @@
+FROM python:3.9
+
+WORKDIR /code
+
+COPY ./requirements.txt /code/requirements.txt
+
+RUN pip install --no-cache-dir --upgrade -r /code/requirements.txt
+
+COPY ./main.py /code/main.py
+
+CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860"]

main.py

@@ -0,0 +1,49 @@
+import os
+
+# Force Hugging Face to store files in /tmp (which is writable)
+os.environ["HF_HOME"] = "/tmp/huggingface"
+os.environ["TRANSFORMERS_CACHE"] = "/tmp/huggingface"
+os.environ["HUGGINGFACE_HUB_CACHE"] = "/tmp/huggingface"
+
+token = os.getenv('HF_TOKEN')
+
+from datasets import load_dataset
+from sentence_transformers import SentenceTransformer, util
+
+dataset = load_dataset("AhmedElSherbeny/arabic-blog-embeddings-v2", split="train")
+model = SentenceTransformer("aubmindlab/bert-base-arabertv02", token = token)
+print("Dataset and Model loaded successfully")
+
+import torch
+import numpy as np
+# Convert all stored embeddings to tensor
+stored_embeddings = torch.from_numpy(np.array([item["embedding"] for item in dataset])).type(torch.float32)
+
+
+def recommendation(query, top_k=10):
+    query_embedding = model.encode(query, convert_to_tensor=True)
+    cosine_score = util.cos_sim(query_embedding, stored_embeddings)[0]
+    top_indices = torch.topk(cosine_score, k=top_k).indices
+    recomm = []
+    for idx in top_indices:
+        recomm.append(dataset[int(idx)]['id'])
+    return recomm[1:]
+
+from fastapi import FastAPI
+from pydantic import BaseModel
+
+app = FastAPI()
+
+class RecommendationRequest(BaseModel):
+    query: str
+    top_k: int = 5 # Number of recommendations to return
+
+@app.post("/recommend")
+async def recommend(request: RecommendationRequest):
+
+    recommendations = recommendation(request.query, request.top_k)
+
+    return {"recommendations": recommendations}
+
+    
+

requirements.txt

@@ -0,0 +1,10 @@
+arabert==1.0.1
+datasets==3.6.0
+sentence-transformers==4.1.0
+tensorflow-datasets==4.9.8
+torchao==0.10.0
+torchdata==0.11.0
+torchsummary==1.5.1
+torchtune==0.6.1
+transformers==4.51.3
+vega-datasets==0.9.0