Create app.py

app.py

@@ -0,0 +1,103 @@
+from fastapi import FastAPI, HTTPException
+from fastapi.middleware.cors import CORSMiddleware
+from pydantic import BaseModel, Field
+from llama_cpp import Llama
+import re
+
+# ==================== MODEL CONFIG ====================
+MODEL_REPO = "bartowski/Phi-3.5-mini-instruct-GGUF"
+MODEL_FILE = "Phi-3.5-mini-instruct-Q4_K_M.gguf"
+
+print("🚀 Loading Phi-3.5 Mini (Fast Summarizer)...")
+llm = Llama.from_pretrained(
+    repo_id=MODEL_REPO,
+    filename=MODEL_FILE,
+    n_threads=4,
+    n_ctx=2048,      # Enough for summarization
+    n_batch=256,
+    n_gpu_layers=0,
+    verbose=False,
+)
+print("✅ Model loaded")
+
+# ==================== FASTAPI ====================
+app = FastAPI(
+    title="AI Summarizer",
+    description="Fast & Accurate AI Text Summarizer",
+    version="1.0"
+)
+
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+# ==================== REQUEST ====================
+class SummarizeRequest(BaseModel):
+    text: str = Field(..., min_length=1, max_length=2000)
+    length: str = "short"  # short | medium | long
+
+# ==================== PROMPTS ====================
+LENGTH_INSTRUCTIONS = {
+    "short": "Summarize in 2–3 concise sentences.",
+    "medium": "Summarize in 4–5 clear sentences.",
+    "long": "Summarize in a detailed paragraph.",
+}
+
+# ==================== CLEAN OUTPUT ====================
+def clean_output(text: str) -> str:
+    text = re.sub(r"<\|.*?\|>", "", text)
+    text = re.sub(r"\s+", " ", text)
+    return text.strip()
+
+# ==================== ENDPOINT ====================
+@app.post("/api/summarize")
+async def summarize(req: SummarizeRequest):
+    text = req.text.strip()
+    if not text:
+        raise HTTPException(status_code=400, detail="Text cannot be empty")
+
+    length_instruction = LENGTH_INSTRUCTIONS.get(req.length, LENGTH_INSTRUCTIONS["short"])
+
+    prompt = (
+        f"<|user|>\n"
+        f"You are an expert text summarizer.\n"
+        f"{length_instruction}\n\n"
+        f"Text:\n{text}\n"
+        f"<|end|>\n"
+        f"<|assistant|>\n"
+    )
+
+    try:
+        output = llm(
+            prompt,
+            max_tokens=140 if req.length == "short" else 220,
+            temperature=0.3,     # Low = factual
+            top_p=0.9,
+            top_k=40,
+            repeat_penalty=1.05,
+            stop=["<|end|>", "<|user|>"],
+            echo=False,
+        )
+
+        summary = clean_output(output["choices"][0]["text"])
+
+        if not summary:
+            raise RuntimeError("Empty summary")
+
+        return {
+            "original_length": len(text),
+            "summary": summary,
+            "length": req.length,
+            "success": True
+        }
+
+    except Exception as e:
+        print("❌ Summarization error:", e)
+        raise HTTPException(status_code=500, detail="Summarization failed")
+
+@app.get("/")
+def health():
+    return {"status": "ok", "model": MODEL_FILE}