Upload 3 files

Dockerfile

@@ -0,0 +1,10 @@
+FROM python:3.10-slim
+
+WORKDIR /app
+
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+
+COPY . .
+
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

app.py

@@ -0,0 +1,76 @@
+from transformers import pipeline, AutoTokenizer
+import torch
+from fastapi import FastAPI, HTTPException
+from pydantic import BaseModel
+import os
+from transformers import pipeline
+
+os.environ['HF_HOME'] = '/tmp/.cache'  #
+MODEL_NAME = "Open-Orca/Mistral-7B-OpenOrca"
+
+device = 0 if torch.cuda.is_available() else -1
+
+
+pipe = pipeline(
+    "text-generation",
+    model="Open-Orca/Mistral-7B-OpenOrca",
+    device_map="auto",
+    model_kwargs={
+        "load_in_4bit": True,
+        "torch_dtype": torch.float16
+    }
+)
+tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+
+SYSTEM_MESSAGE = {
+    "role": "system",
+    "content": (
+        "Você é um chatbot especialista em autismo. "
+        "Seu papel é orientar pais e professores com empatia, linguagem simples e respeitosa. "
+        "Evite termos técnicos, a não ser que a pessoa peça. "
+        "Valide as emoções de quem pergunta e ofereça exemplos práticos e acolhedores. "
+        "Fale sempre em português claro e humano."
+    )
+}
+
+app = FastAPI()
+
+class ChatRequest(BaseModel):
+    message: str
+    chat_history: list = []
+
+@app.post("/chat")
+async def chat(request: ChatRequest):
+    try:
+        full_history = [SYSTEM_MESSAGE] + request.chat_history + [
+            {"role": "user", "content": request.message}
+        ]
+        
+        prompt = tokenizer.apply_chat_template(
+            full_history,
+            tokenize=False,
+            add_generation_prompt=True
+        )
+        
+        outputs = pipe(
+            prompt,
+            max_new_tokens=512,
+            temperature=0.7,
+            top_k=50,
+            top_p=0.9,
+            do_sample=True
+        )
+        
+        resposta = outputs[0]['generated_text'].replace(prompt, "").strip()
+        
+        return {
+            "response": resposta,
+            "updated_history": full_history + [{"role": "assistant", "content": resposta}]
+        }
+        
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
+@app.get("/")
+def health_check():
+    return {"status": "online", "model": MODEL_NAME}

requirements.txt

@@ -0,0 +1,6 @@
+bitsandbytes>=0.43.0  
+accelerate>=0.30.0
+transformers>=4.40.0
+torch>=2.2.0
+fastapi>=0.110.0
+uvicorn>=0.29.0