Upload 4 files

Dockerfile

@@ -3,11 +3,10 @@ FROM python:3.10-slim
 WORKDIR /app
 
 # Install system dependencies
-RUN apt-get update && apt-get install -y \
+RUN sudo apt-get update && sudapt-get install -y \
     git \
     && rm -rf /var/lib/apt/lists/*
 
-RUN mkdir cacheddir_joblib
 
 # Copy requirements first to leverage Docker cache
 COPY requirements.txt .

app/config.py

@@ -1,10 +1,8 @@
 from pydantic_settings import BaseSettings
 
 class Settings(BaseSettings):
-    MONGODB_URL: str
-    COLLECTION_NAME: str = "documents"
-    DATABASE_NAME: str = "ragbot"
     HUGGINGFACE_TOKEN: str
+    MODEL_NAME: str = "deepseek-ai/deepseek-coder-33b-instruct"
     
     class Config:
         env_file = ".env"

app/main.py

@@ -1,44 +1,43 @@
-from contextlib import asynccontextmanager
-from fastapi import FastAPI, UploadFile, File
-from .utils.dspy_patch import patch_dspy_cache
-
-# Apply patch before importing DSPy
-patch_dspy_cache()
-
-from .database.mongodb import db
-from .rag.document_processor import DocumentProcessor
-from .rag.retriever import RAGRetriever
-
-@asynccontextmanager
-async def lifespan(app: FastAPI):
-    # Startup
-    await db.connect()
-    yield
-    # Shutdown
-    await db.close()
+from fastapi import FastAPI
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
+from .config import settings
 
 app = FastAPI(
-    title="RAG Chatbot",
-    description="A RAG-based chatbot using DeepSeek model",
+    title="Deepseek Chat API",
+    description="A simple chat API using DeepSeek model",
     version="1.0.0"
 )
-document_processor = DocumentProcessor()
-rag_retriever = RAGRetriever()
 
-@app.post("/upload-pdf")
-async def upload_pdf(file: UploadFile = File(...)):
-    content = await file.read()
-    await document_processor.process_pdf(content)
-    return {"message": "PDF processed successfully"}
+# Initialize model and tokenizer
+tokenizer = AutoTokenizer.from_pretrained(settings.MODEL_NAME, token=settings.HUGGINGFACE_TOKEN)
+model = AutoModelForCausalLM.from_pretrained(
+    settings.MODEL_NAME,
+    token=settings.HUGGINGFACE_TOKEN,
+    torch_dtype=torch.float16,
+    device_map="auto"
+)
 
 @app.post("/chat")
-async def chat(query: str):
-    response = await rag_retriever.get_response(query)
-    return {
-        "reasoning": response["reasoning"],
-        "answer": response["answer"],
-        "context_used": response["context_used"]
-    }
+async def chat(message: str):
+    # Prepare the prompt
+    prompt = f"### Instruction: {message}\n\n### Response:"
+    
+    # Generate response
+    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+    outputs = model.generate(
+        **inputs,
+        max_new_tokens=512,
+        temperature=0.7,
+        do_sample=True,
+        pad_token_id=tokenizer.eos_token_id
+    )
+    
+    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    # Extract only the response part
+    response = response.split("### Response:")[-1].strip()
+    
+    return {"response": response}
 
 if __name__ == "__main__":
     import uvicorn

requirements.txt

@@ -1,15 +1,8 @@
 fastapi
 uvicorn
-motor
-pymongo
-python-multipart
-PyPDF2
-dspy-ai
 pydantic
 pydantic-settings
 python-dotenv
-sentence-transformers
-numpy
 transformers
-cloudpickle
-accelerate>=0.26.0
+accelerate>=0.26.0
+torch