Upload folder using huggingface_hub

README.md

@@ -0,0 +1,13 @@
+# Qwen FastAPI Inference API
+
+This repository contains a secure, OpenAI-compatible REST API for the **Qwen2.5-0.5B-Instruct** model, built with FastAPI.
+
+## Features
+- **OpenAI Compatible**: Implements the `/v1/chat/completions` endpoint.
+- **Secure**: Uses JWT Authentication (HS256) for all inference requests.
+- **Lightweight**: Optimized to run on consumer hardware or free-tier cloud environments like Google Colab.
+
+## Structure
+- `application.py`: The main FastAPI server.
+- `client.py`: A test client to verify the API.
+- `generate_token.py`: Utility to generate JWT tokens for authentication.

application.py

@@ -0,0 +1,66 @@
+import jwt
+import time
+import os
+from datetime import datetime, timedelta
+from fastapi import FastAPI, Depends, HTTPException
+from fastapi.security import HTTPBearer, HTTPAuthorizationCredentials
+from pydantic import BaseModel
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
+from dotenv import load_dotenv
+
+# --- Load Environment Variables ---
+load_dotenv()
+SECRET_KEY = os.getenv("JWT_SECRET_KEY", "default-fallback-secret")
+ALGORITHM = "HS256"
+MODEL_NAME = "Qwen/Qwen2.5-0.5B-Instruct"
+
+security = HTTPBearer()
+
+def verify_token(credentials: HTTPAuthorizationCredentials = Depends(security)):
+    token = credentials.credentials
+    try:
+        payload = jwt.decode(token, SECRET_KEY, algorithms=[ALGORITHM])
+        return payload
+    except Exception:
+        raise HTTPException(status_code=401, detail="Unauthorized")
+
+# --- FastAPI Setup ---
+app = FastAPI()
+tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+model = AutoModelForCausalLM.from_pretrained(MODEL_NAME, torch_dtype="auto", device_map="auto")
+
+class ChatMessage(BaseModel):
+    role: str
+    content: str
+
+class ChatCompletionRequest(BaseModel):
+    messages: list[ChatMessage]
+    max_tokens: int = 100
+
+@app.get("/")
+def read_root():
+    return {"message": "Qwen OpenAI-style API is running with .env auth"}
+
+@app.post("/v1/chat/completions")
+async def chat_generate(request: ChatCompletionRequest, user=Depends(verify_token)):
+    chat_msgs = [msg.dict() for msg in request.messages]
+    text = tokenizer.apply_chat_template(chat_msgs, tokenize=False, add_generation_prompt=True)
+    model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
+
+    generated_ids = model.generate(
+        **model_inputs,
+        max_new_tokens=request.max_tokens
+    )
+    generated_ids = [output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)]
+    response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
+
+    return {
+        "id": f"chatcmpl-{int(time.time())}",
+        "object": "chat.completion",
+        "model": MODEL_NAME,
+        "choices": [{
+            "message": {"role": "assistant", "content": response},
+            "finish_reason": "stop"
+        }]
+    }

client.py

@@ -0,0 +1,43 @@
+import requests
+import argparse
+import sys
+
+BASE_URL = "http://127.0.0.1:8000"
+
+def test_openai_style(token):
+    headers = {"Authorization": f"Bearer {token}"}
+
+    # Send OpenAI-style Chat Messages
+    payload = {
+        "messages": [
+            {"role": "system", "content": "You are a concise AI assistant."},
+            {"role": "user", "content": "What is the main benefit of open-source LLMs?"}
+        ],
+        "max_tokens": 60
+    }
+
+    print(f"🤖 Sending Chat Completion request to {BASE_URL}...")
+    try:
+        response = requests.post(f"{BASE_URL}/v1/chat/completions", json=payload, headers=headers)
+
+        if response.status_code == 200:
+            result = response.json()
+            print("\n✨ Model Response:")
+            print(result['choices'][0]['message']['content'])
+        elif response.status_code == 401:
+            print("❌ Error: 401 Unauthorized. Please check if your token is valid and matches the SECRET_KEY.")
+        else:
+            print(f"❌ Error {response.status_code}: {response.text}")
+    except Exception as e:
+        print(f"❌ Connection failed: {e}")
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Test the Qwen API with a JWT token.")
+    parser.add_argument("--token", type=str, help="The JWT token generated by generate_token.py")
+    args = parser.parse_args()
+
+    if not args.token:
+        print("❌ Error: Please provide a token using --token <YOUR_TOKEN>")
+        sys.exit(1)
+
+    test_openai_style(args.token)

generate_token.py

@@ -0,0 +1,24 @@
+import jwt
+import os
+from datetime import datetime, timedelta
+from dotenv import load_dotenv
+
+load_dotenv()
+SECRET_KEY = os.getenv("JWT_SECRET_KEY")
+ALGORITHM = "HS256"
+
+def create_token(user_id: str):
+    if not SECRET_KEY:
+        print("❌ Error: JWT_SECRET_KEY not found in .env")
+        return
+    
+    payload = {
+        "sub": user_id,
+        "exp": datetime.utcnow() + timedelta(hours=24)
+    }
+    token = jwt.encode(payload, SECRET_KEY, algorithm=ALGORITHM)
+    print(f"✅ Token generated for {user_id}:\n{token}")
+    return token
+
+if __name__ == "__main__":
+    create_token("admin_user")

push_to_hf.py

@@ -0,0 +1,31 @@
+import argparse
+import os
+from huggingface_hub import HfApi, create_repo
+
+def push_api_to_hf(token, repo_name="qwen-api-fastapi"):
+    username = "convaiinnovations"
+    repo_id = f"{username}/{repo_name}"
+    api = HfApi()
+
+    print(f"🚀 Creating/Accessing repo: {repo_id}...")
+    try:
+        create_repo(repo_id=repo_id, token=token, repo_type="model", exist_ok=True)
+
+        print(f"📤 Uploading files from 'api/' folder...")
+        api.upload_folder(
+            folder_path=".",
+            repo_id=repo_id,
+            repo_type="model",
+            token=token
+        )
+        print(f"✅ Successfully pushed to https://huggingface.co/{repo_id}")
+    except Exception as e:
+        print(f"❌ Error: {e}")
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Push API folder to Hugging Face.")
+    parser.add_argument("--token", type=str, required=True, help="Your Hugging Face Write Token")
+    parser.add_argument("--repo", type=str, default="qwen-api-fastapi", help="Name of the repo")
+    args = parser.parse_args()
+
+    push_api_to_hf(args.token, args.repo)