Initial FastAPI backend

Dockerfile

@@ -0,0 +1,17 @@
+FROM python:3.11-slim
+
+ENV PIP_NO_CACHE_DIR=1 PYTHONDONTWRITEBYTECODE=1
+WORKDIR /app
+
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+
+COPY server.py .
+
+ENV MODEL_ID="milaadesign/helper-qwen-1_5b"
+ENV MODEL_MAX_LEN=512
+ENV TORCH_NUM_THREADS=2
+ENV OMP_NUM_THREADS=2
+
+EXPOSE 7860
+CMD ["uvicorn", "server:app", "--host", "0.0.0.0", "--port", "7860"]

requirements.txt

@@ -0,0 +1,5 @@
+fastapi==0.115.0
+uvicorn[standard]==0.30.6
+transformers==4.57.1
+torch==2.2.2
+safetensors==0.4.4

server.py

@@ -0,0 +1,78 @@
+import os, torch
+from fastapi import FastAPI
+from fastapi.middleware.cors import CORSMiddleware
+from pydantic import BaseModel
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+# Hugging Face model ID you just uploaded
+MODEL_ID = os.getenv("MODEL_ID", "milaadesign/helper-qwen-1_5b")
+MODEL_MAX_LEN = int(os.getenv("MODEL_MAX_LEN", "512"))
+
+app = FastAPI(title="Patient Helper API")
+
+# CORS so your DreamHost site can call it from the browser
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],  # later you can restrict to your domain
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+class GenerateIn(BaseModel):
+    prompt: str
+    max_new_tokens: int = 128
+    temperature: float = 0.2
+    top_p: float = 0.9
+    repetition_penalty: float = 1.05
+
+device = torch.device("cpu")  # Spaces CPU Basic
+
+torch.set_num_threads(int(os.getenv("TORCH_NUM_THREADS", "2")))
+os.environ.setdefault("OMP_NUM_THREADS", "2")
+
+print(f"Loading tokenizer from {MODEL_ID}...")
+tok = AutoTokenizer.from_pretrained(MODEL_ID)
+if tok.pad_token is None:
+    tok.pad_token = tok.eos_token
+tok.truncation_side = "left"
+tok.model_max_length = MODEL_MAX_LEN
+
+print(f"Loading model from {MODEL_ID}...")
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_ID,
+    torch_dtype=torch.float32,
+    low_cpu_mem_usage=True,
+)
+model.to(device)
+model.eval()
+print("Model loaded.")
+
+SYSTEM = (
+    "You are a supportive, non-clinical assistant. "
+    "Offer gentle, practical tips on how to support a patient. "
+    "Do not diagnose or give unsafe advice. Encourage professional help when needed."
+)
+
+@app.post("/generate")
+def generate(body: GenerateIn):
+    messages = [
+        {"role": "system", "content": SYSTEM},
+        {"role": "user", "content": body.prompt.strip()},
+    ]
+    text = tok.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+    inputs = tok(text, return_tensors="pt", truncation=True).to(device)
+
+    with torch.no_grad():
+        out = model.generate(
+            **inputs,
+            max_new_tokens=body.max_new_tokens,
+            temperature=body.temperature,
+            top_p=body.top_p,
+            repetition_penalty=body.repetition_penalty,
+            do_sample=body.temperature > 0,
+            pad_token_id=tok.eos_token_id,
+        )
+
+    resp = tok.decode(out[0], skip_special_tokens=True)
+    return {"text": resp}